From 3dc8316f17bd2a6287d4f0521fc0196a0ea184ab Mon Sep 17 00:00:00 2001 From: Maschell Date: Wed, 14 Oct 2020 13:54:25 +0200 Subject: [PATCH] First commit --- .gitignore | 6 + .gitmodules | 3 + Makefile | 161 + payload | 1 + source/ApplicationState.cpp | 411 ++ source/ApplicationState.h | 80 + source/Input.cpp | 5 + source/Input.h | 60 + source/InstallerService.cpp | 620 ++ source/InstallerService.h | 82 + source/ScreenUtils.cpp | 48 + source/ScreenUtils.h | 55 + source/StringTools.cpp | 234 + source/StringTools.h | 95 + source/VPADInput.h | 59 + source/WiiUScreen.cpp | 94 + source/WiiUScreen.h | 53 + source/common/common.cpp | 117 + source/common/common.h | 51 + source/common/fst_structs.h | 62 + source/fs/CFile.cpp | 173 + source/fs/CFile.hpp | 71 + source/fs/FSUtils.cpp | 174 + source/fs/FSUtils.h | 17 + source/main.cpp | 96 + source/utils/TinySHA1.hpp | 196 + source/utils/logger.h | 23 + source/utils/pugiconfig.hpp | 74 + source/utils/pugixml.cpp | 12865 ++++++++++++++++++++++++++++++++++ source/utils/pugixml.hpp | 1477 ++++ source/utils/utils.cpp | 64 + source/utils/utils.h | 45 + 32 files changed, 17572 insertions(+) create mode 100644 .gitignore create mode 100644 .gitmodules create mode 100644 Makefile create mode 160000 payload create mode 100644 source/ApplicationState.cpp create mode 100644 source/ApplicationState.h create mode 100644 source/Input.cpp create mode 100644 source/Input.h create mode 100644 source/InstallerService.cpp create mode 100644 source/InstallerService.h create mode 100644 source/ScreenUtils.cpp create mode 100644 source/ScreenUtils.h create mode 100644 source/StringTools.cpp create mode 100644 source/StringTools.h create mode 100644 source/VPADInput.h create mode 100644 source/WiiUScreen.cpp create mode 100644 source/WiiUScreen.h create mode 100644 source/common/common.cpp create mode 100644 source/common/common.h create mode 100644 source/common/fst_structs.h create mode 100644 source/fs/CFile.cpp create mode 100644 source/fs/CFile.hpp create mode 100644 source/fs/FSUtils.cpp create mode 100644 source/fs/FSUtils.h create mode 100644 source/main.cpp create mode 100644 source/utils/TinySHA1.hpp create mode 100644 source/utils/logger.h create mode 100644 source/utils/pugiconfig.hpp create mode 100644 source/utils/pugixml.cpp create mode 100644 source/utils/pugixml.hpp create mode 100644 source/utils/utils.cpp create mode 100644 source/utils/utils.h diff --git a/.gitignore b/.gitignore new file mode 100644 index 0000000..d1cd803 --- /dev/null +++ b/.gitignore @@ -0,0 +1,6 @@ +.idea/ +cmake-build-debug/ +build/ +*.rpx +*.elf +CMakeLists.txt diff --git a/.gitmodules b/.gitmodules new file mode 100644 index 0000000..b06d16c --- /dev/null +++ b/.gitmodules @@ -0,0 +1,3 @@ +[submodule "payload"] + path = payload + url = https://github.com/wiiu-env/PayloadFromRPX diff --git a/Makefile b/Makefile new file mode 100644 index 0000000..bd73d33 --- /dev/null +++ b/Makefile @@ -0,0 +1,161 @@ +#------------------------------------------------------------------------------- +.SUFFIXES: +#------------------------------------------------------------------------------- + +ifeq ($(strip $(DEVKITPRO)),) +$(error "Please set DEVKITPRO in your environment. export DEVKITPRO=/devkitpro") +endif + +TOPDIR ?= $(CURDIR) + +include $(DEVKITPRO)/wut/share/wut_rules + +#------------------------------------------------------------------------------- +# TARGET is the name of the output +# BUILD is the directory where object files & intermediate files will be placed +# SOURCES is a list of directories containing source code +# DATA is a list of directories containing data files +# INCLUDES is a list of directories containing header files +#------------------------------------------------------------------------------- +TARGET := $(notdir $(CURDIR)) +BUILD := build +SOURCES := source \ + source/common \ + source/utils \ + source/fs +DATA := data +INCLUDES := include source + +#------------------------------------------------------------------------------- +# options for code generation +#------------------------------------------------------------------------------- +CFLAGS := -g -Wall -O2 -ffunction-sections -fstack-protector-all \ + $(MACHDEP) + +CFLAGS += $(INCLUDE) -D__WIIU__ -D__WUT__ + +CXXFLAGS := $(CFLAGS) -std=c++17 + +ASFLAGS := -g $(ARCH) +LDFLAGS = -g $(ARCH) $(RPXSPECS) -Wl,-Map,$(notdir $*.map) + +LIBS := -lwut -liosuhax + +#------------------------------------------------------------------------------- +# list of directories containing libraries, this must be the top level +# containing include and lib +#------------------------------------------------------------------------------- +LIBDIRS := $(PORTLIBS) $(WUT_ROOT) $(WUT_ROOT)/usr + + +#------------------------------------------------------------------------------- +# no real need to edit anything past this point unless you need to add additional +# rules for different file extensions +#------------------------------------------------------------------------------- +ifneq ($(BUILD),$(notdir $(CURDIR))) +#------------------------------------------------------------------------------- + +export OUTPUT := $(CURDIR)/$(TARGET) +export TOPDIR := $(CURDIR) + +export VPATH := $(foreach dir,$(SOURCES),$(CURDIR)/$(dir)) \ + $(foreach dir,$(DATA),$(CURDIR)/$(dir)) + +export DEPSDIR := $(CURDIR)/$(BUILD) + +CFILES := $(foreach dir,$(SOURCES),$(notdir $(wildcard $(dir)/*.c))) +CPPFILES := $(foreach dir,$(SOURCES),$(notdir $(wildcard $(dir)/*.cpp))) +SFILES := $(foreach dir,$(SOURCES),$(notdir $(wildcard $(dir)/*.s))) +BINFILES := $(foreach dir,$(DATA),$(notdir $(wildcard $(dir)/*.*))) + +#------------------------------------------------------------------------------- +# use CXX for linking C++ projects, CC for standard C +#------------------------------------------------------------------------------- +ifeq ($(strip $(CPPFILES)),) +#------------------------------------------------------------------------------- + export LD := $(CC) +#------------------------------------------------------------------------------- +else +#------------------------------------------------------------------------------- + export LD := $(CXX) +#------------------------------------------------------------------------------- +endif +#------------------------------------------------------------------------------- + +export OFILES_BIN := $(addsuffix .o,$(BINFILES)) +export OFILES_SRC := $(CPPFILES:.cpp=.o) $(CFILES:.c=.o) $(SFILES:.s=.o) +export OFILES := $(OFILES_BIN) $(OFILES_SRC) root.rpx.o +export HFILES_BIN := $(addsuffix .h,$(subst .,_,$(BINFILES))) + +export INCLUDE := $(foreach dir,$(INCLUDES),-I$(CURDIR)/$(dir)) \ + $(foreach dir,$(LIBDIRS),-I$(dir)/include) \ + -I$(CURDIR)/$(BUILD) + +export LIBPATHS := $(foreach dir,$(LIBDIRS),-L$(dir)/lib) + +.PHONY: $(BUILD) clean all + +#------------------------------------------------------------------------------- +all: $(BUILD) + +$(BUILD): + @[ -d $@ ] || mkdir -p $@ + @$(MAKE) --no-print-directory -C payload + @$(MAKE) --no-print-directory -C $(BUILD) -f $(CURDIR)/Makefile + +#------------------------------------------------------------------------------- +clean: + @echo clean ... + make clean -C payload + @rm -fr $(BUILD) $(TARGET).rpx $(TARGET).elf + +#------------------------------------------------------------------------------- +else +.PHONY: all + +DEPENDS := $(OFILES:.o=.d) + +#------------------------------------------------------------------------------- +# main targets +#------------------------------------------------------------------------------- +#------------------------------------------------------------------------------- +# main targets +#------------------------------------------------------------------------------- + +safe_payload := ../payload/root.rpx + +all : $(OUTPUT).rpx + +$(safe_payload): + make -C ../payload + +$(OUTPUT).rpx : $(OUTPUT).elf +$(OUTPUT).elf : $(OFILES) +$(OFILES) : safe_payload.h + +$(OFILES_SRC) : $(HFILES_BIN) + +#------------------------------------------------------------------------------- +# you need a rule like this for each extension you use as binary data +#------------------------------------------------------------------------------- +safe_payload.h: $(safe_payload) + @bin2s -a 32 -H `(echo $( $@ + @printf '#include "' >> $@ + @(printf $(> $@ + @echo '.h"' >> $@ + @printf '#define RPX_HASH "' >> $@ + @sha1sum $(<) | cut -f1 -d' ' | tr -d '\n' >> $@ + @printf '"' >> $@ + +#------------------------------------------------------------------------------- +%.bin.o %_bin.h : %.bin +#------------------------------------------------------------------------------- + @echo $(notdir $<) + @$(bin2o) + +-include $(DEPENDS) + +#------------------------------------------------------------------------------- +endif +#------------------------------------------------------------------------------- diff --git a/payload b/payload new file mode 160000 index 0000000..c4c1449 --- /dev/null +++ b/payload @@ -0,0 +1 @@ +Subproject commit c4c14496afc12b3f4f5ae6e9c335e379a212d1c2 diff --git a/source/ApplicationState.cpp b/source/ApplicationState.cpp new file mode 100644 index 0000000..6c09623 --- /dev/null +++ b/source/ApplicationState.cpp @@ -0,0 +1,411 @@ +#include "ApplicationState.h" +#include "WiiUScreen.h" +#include "ScreenUtils.h" +#include "../build/safe_payload.h" +#include +#include + +extern "C" void OSShutdown(); + +void ApplicationState::render() { + WiiUScreen::clearScreen(); + WiiUScreen::drawLine("Aroma Installer"); + WiiUScreen::drawLine("=================="); + WiiUScreen::drawLine(""); + + if (this->state == STATE_ERROR) { + WiiUScreen::drawLine("The installation failed:"); + WiiUScreen::drawLine(); + WiiUScreen::drawLinef("Error: %s", ErrorMessage().c_str()); + WiiUScreen::drawLinef("Description: %s", ErrorDescription().c_str()); + WiiUScreen::drawLine(); + WiiUScreen::drawLine(); + WiiUScreen::drawLine("Press A to return to the Wii U Menu."); + } else if (this->state == STATE_WELCOME_SCREEN) { + WiiUScreen::drawLine("Welcome to the Aroma Installer!"); + WiiUScreen::drawLine("Do you want to check if an installation is possible?"); + WiiUScreen::drawLine(""); + if (this->selectedOption == 0) { + WiiUScreen::drawLine("> Check Exit"); + } else if (this->selectedOption == 1) { + WiiUScreen::drawLine(" Check > Exit"); + } + } else if (this->state == STATE_GET_APP_INFORMATION) { + WiiUScreen::drawLine("Getting app information"); + } else if (this->state == STATE_CHECK_PATCH_POSSIBLE) { + WiiUScreen::drawLine("Check if console can be patched."); + } else if (this->state == STATE_CHECK_PATCH_POSSIBLE_DONE) { + WiiUScreen::drawLinef("Compatible title:"); + WiiUScreen::drawLinef("%s", appInfo->appName); + WiiUScreen::drawLine(); + + if (this->fstAlreadyPatched) { + WiiUScreen::drawLine("[ X ] title.fst is already patched!"); + } else if (this->fstPatchPossible) { + WiiUScreen::drawLine("[ X ] title.fst can be patched!"); + } else { + WiiUScreen::drawLine("[ ] title.fst can NOT be patched!"); + } + if (this->cosAlreadyPatched) { + WiiUScreen::drawLine("[ X ] cos.xml is already patched!"); + } else if (this->cosPatchPossible) { + WiiUScreen::drawLine("[ X ] cos.xml can be patched!"); + } else { + WiiUScreen::drawLine("[ ] cos.xml can NOT be patched!"); + } + if (this->rpxAlreadyPatched) { + WiiUScreen::drawLine("[ X ] safe.rpx is already patched!"); + } else { + WiiUScreen::drawLine("[ X ] safe.rpx can to be patched!"); + } + + WiiUScreen::drawLine(); + WiiUScreen::drawLinef("System is booting into: "); + if (this->coldbootTitle == nullptr) { + WiiUScreen::drawLinef("%ll016X (Unknown title)", this->coldbootTitleId); + } else { + WiiUScreen::drawLinef("%ll016X (%s)", this->coldbootTitle->tid, this->coldbootTitle->name); + } + + WiiUScreen::drawLine(); + + if (!this->fstPatchPossible || !this->cosPatchPossible) { + WiiUScreen::drawLine("A safe installation of Aroma can not be provided."); + WiiUScreen::drawLine(); + WiiUScreen::drawLine("Press A to return to the Wii U Menu"); + } else { + WiiUScreen::drawLine("Do you want to install Aroma?"); + WiiUScreen::drawLine(""); + if (this->selectedOption == 0) { + WiiUScreen::drawLine("> Install Exit"); + } else if (this->selectedOption == 1) { + WiiUScreen::drawLine(" Install > Exit"); + } + } + } else if (this->state == STATE_INSTALL_CHOOSE_COLDBOOT) { + WiiUScreen::drawLine("Select your installation type:"); + WiiUScreen::drawLine(); + WiiUScreen::drawLine("[Coldboot] Aroma will launch directly after booting the console."); + WiiUScreen::drawLine("[No Coldboot] Aroma will need to be launched manually."); + WiiUScreen::drawLine(""); + if (this->selectedOption == 0) { + WiiUScreen::drawLine("> Back Coldboot No Coldboot"); + } else if (this->selectedOption == 1) { + WiiUScreen::drawLine(" Back > Coldboot No Coldboot"); + } else if (this->selectedOption == 2) { + WiiUScreen::drawLine(" Back Coldboot > No Coldboot"); + } + } else if (this->state == STATE_INSTALL_NO_COLDBOOT_ALLOWED) { + WiiUScreen::drawLine("Note: To install Aroma as coldboot you need to run this installer"); + WiiUScreen::drawLine("from an already running Aroma instance (and not the browser)"); + WiiUScreen::drawLine("After the installation has finished, reboot the console, open the"); + WiiUScreen::drawLine("Health & Safety app and run the Aroma installer."); + WiiUScreen::drawLine(); + if (this->selectedOption == 0) { + WiiUScreen::drawLine("> Back Install without Coldboot"); + } else if (this->selectedOption == 1) { + WiiUScreen::drawLine("> Back > Install without Coldboot"); + } + } else if (this->state == STATE_INSTALL_CONFIRM_DIALOG) { + WiiUScreen::drawLine("Are you REALLY sure you want to install Aroma?"); + WiiUScreen::drawLine("Installing could permanently damage your console"); + WiiUScreen::drawLine(); + WiiUScreen::drawLine("After the installation you can NOT longer use:"); + WiiUScreen::drawLinef("- %s", appInfo->appName); + WiiUScreen::drawLine(); + WiiUScreen::drawLine("Selected installation type:"); + if (this->installColdboot) { + WiiUScreen::drawLine("- Coldboot"); + } else { + WiiUScreen::drawLine("- No Coldboot"); + } + + WiiUScreen::drawLine(); + WiiUScreen::drawLine(); + if (this->selectedOption == 0) { + WiiUScreen::drawLine("> Back Install"); + } else if (this->selectedOption == 1) { + WiiUScreen::drawLine(" Back > Install"); + } + } else if (this->state == STATE_INSTALL_STARTED) { + WiiUScreen::drawLine("Installing..."); + } else if (this->state == STATE_INSTALL_FST) { + WiiUScreen::drawLine("... patching title.fst"); + } else if (this->state == STATE_INSTALL_COS) { + WiiUScreen::drawLine("... patching cos.xml"); + } else if (this->state == STATE_INSTALL_SYSTEM_XML) { + WiiUScreen::drawLine("... patching system.xml"); + } else if (this->state == STATE_INSTALL_RPX) { + WiiUScreen::drawLine("... install safe.rpx"); + } else if (this->state == STATE_INSTALL_SUCCESS) { + WiiUScreen::drawLine("Aroma was successfully installed"); + WiiUScreen::drawLine(); + WiiUScreen::drawLine("Press A to shutdown the console"); + } + printFooter(); + WiiUScreen::flipBuffers(); +} + +void ApplicationState::update(Input *input) { + if (this->state == STATE_ERROR) { + OSEnableHomeButtonMenu(true); + if (entrySelected(input)) { + SYSLaunchMenu(); + } + } else if (this->state == STATE_WELCOME_SCREEN) { + proccessMenuNavigation(input, 2); + if (entrySelected(input)) { + if (this->selectedOption == 0) { + this->state = STATE_GET_APP_INFORMATION; + } else { + SYSLaunchMenu(); + } + this->selectedOption = 0; + return; + } + } else if (this->state == STATE_GET_APP_INFORMATION) { + getAppInformation(); + } else if (this->state == STATE_CHECK_PATCH_POSSIBLE) { + checkPatchPossible(); + } else if (this->state == STATE_CHECK_PATCH_POSSIBLE_DONE) { + if (this->fstPatchPossible && this->cosPatchPossible) { + proccessMenuNavigation(input, 2); + if (entrySelected(input)) { + if (this->selectedOption == 0) { + if (systemXMLPatchPossible) { + this->state = STATE_INSTALL_CHOOSE_COLDBOOT; + this->installColdboot = false; + } else { + this->state = STATE_INSTALL_CONFIRM_DIALOG; + } + } else { + SYSLaunchMenu(); + } + this->selectedOption = 0; + return; + } + } else { + if (entrySelected(input)) { + SYSLaunchMenu(); + } + } + } else if (this->state == STATE_INSTALL_CHOOSE_COLDBOOT) { + + if (!InstallerService::isColdBootAllowed()) { + this->installColdboot = false; + this->state = STATE_INSTALL_NO_COLDBOOT_ALLOWED; + return; + } + + proccessMenuNavigation(input, 3); + if (entrySelected(input)) { + if (this->selectedOption == 0) { // Back + this->state = STATE_CHECK_PATCH_POSSIBLE_DONE; + } else { + if (selectedOption == 1) { // Install with coldboot + this->installColdboot = true; + } + this->state = STATE_INSTALL_CONFIRM_DIALOG; + } + this->selectedOption = 0; + return; + } + } else if (this->state == STATE_INSTALL_NO_COLDBOOT_ALLOWED) { + proccessMenuNavigation(input, 2); + if (entrySelected(input)) { + if (this->selectedOption == 0) { + this->state = STATE_CHECK_PATCH_POSSIBLE_DONE; + } else { + this->state = STATE_INSTALL_CONFIRM_DIALOG; + } + this->selectedOption = 0; + } + } else if (this->state == STATE_INSTALL_CONFIRM_DIALOG) { + proccessMenuNavigation(input, 2); + if (entrySelected(input)) { + if (this->selectedOption == 0) { + this->state = STATE_CHECK_PATCH_POSSIBLE_DONE; + } else { + this->state = STATE_INSTALL_STARTED; + OSEnableHomeButtonMenu(false); + } + this->selectedOption = 0; + return; + } + } else if (this->state == STATE_INSTALL_STARTED) { + this->state = STATE_INSTALL_FST; + } else if (this->state == STATE_INSTALL_FST) { + auto result = InstallerService::patchFST(this->appInfo->path, this->appInfo->fstHash); + if (result != InstallerService::SUCCESS) { + setError(ERROR_INSTALLER_ERROR); + this->installerError = result; + } else { + this->state = STATE_INSTALL_COS; + } + } else if (this->state == STATE_INSTALL_COS) { + auto result = InstallerService::patchCOS(this->appInfo->path, this->appInfo->cosHash); + if (result != InstallerService::SUCCESS) { + setError(ERROR_INSTALLER_ERROR); + this->installerError = result; + } else { + this->state = STATE_INSTALL_RPX; + } + } else if (this->state == STATE_INSTALL_RPX) { + auto result = InstallerService::copyRPX(this->appInfo->path, root_rpx, root_rpx_size, RPX_HASH); + if (result != InstallerService::SUCCESS) { + setError(ERROR_INSTALLER_ERROR); + this->installerError = result; + } else { + if (this->installColdboot) { + this->state = STATE_INSTALL_SYSTEM_XML; + } else { + this->state = STATE_INSTALL_SUCCESS; + } + } + } else if (this->state == STATE_INSTALL_SYSTEM_XML) { + auto result = InstallerService::patchSystemXML("storage_slc_installer:/config", this->appInfo->titleId); + if (result != InstallerService::SUCCESS) { + setError(ERROR_INSTALLER_ERROR); + this->installerError = result; + } else { + auto fsaFd = IOSUHAX_FSA_Open(); + if (fsaFd >= 0) { + if (IOSUHAX_FSA_FlushVolume(fsaFd, "/vol/storage_mlc01") == 0) { + DEBUG_FUNCTION_LINE("Flushed mlc"); + } + if (IOSUHAX_FSA_FlushVolume(fsaFd, "/vol/system") == 0) { + DEBUG_FUNCTION_LINE("Flushed slc"); + } + IOSUHAX_FSA_Close(fsaFd); + } else { + DEBUG_FUNCTION_LINE("Failed to open fsa"); + } + this->state = STATE_INSTALL_SUCCESS; + } + } else if (this->state == STATE_INSTALL_SUCCESS) { + if (entrySelected(input)) { + OSShutdown(); + } + } +} + +ApplicationState::ApplicationState() { + this->state = STATE_WELCOME_SCREEN; + this->selectedOption = 0; + DEBUG_FUNCTION_LINE("State has changed to \"STATE_WELCOME_SCREEN\""); +} + +void ApplicationState::checkPatchPossible() { + DEBUG_FUNCTION_LINE("Check patch possible"); + if (!this->appInfo) { + this->state = STATE_ERROR; + this->error = ERROR_NO_APP_INSTALLED; + DEBUG_FUNCTION_LINE("ERROR"); + return; + } + + this->coldbootTitleId = InstallerService::getColdbootTitleId("storage_slc_installer:/config"); + + this->coldbootTitle = nullptr; + for (int i = 0; GameList[i].tid != 0; i++) { + if (GameList[i].tid == this->coldbootTitleId) { + this->coldbootTitle = &GameList[i]; + break; + } + } + + DEBUG_FUNCTION_LINE("CHECK FST"); + + InstallerService::eResults result; + + this->fstAlreadyPatched = ((result = InstallerService::checkFSTAlreadyValid(this->appInfo->path, this->appInfo->fstHash)) == InstallerService::SUCCESS); + this->rpxAlreadyPatched = ((result = InstallerService::checkRPXAlreadyValid(this->appInfo->path, RPX_HASH)) == InstallerService::SUCCESS); + this->cosAlreadyPatched = ((result = InstallerService::checkCOSAlreadyValid(this->appInfo->path, this->appInfo->cosHash)) == InstallerService::SUCCESS); + this->tmdValid = ((result = InstallerService::checkTMDValid(this->appInfo->path, this->appInfo->tmdHash)) == InstallerService::SUCCESS); + + this->fstPatchPossible = ((result = InstallerService::checkFST(this->appInfo->path, this->appInfo->fstHash)) == InstallerService::SUCCESS); + if (result != InstallerService::SUCCESS) { + DEBUG_FUNCTION_LINE("ERROR: %s", InstallerService::ErrorMessage(result).c_str()); + } + this->cosPatchPossible = ((result = InstallerService::checkCOS(this->appInfo->path, this->appInfo->cosHash)) == InstallerService::SUCCESS); + if (result != InstallerService::SUCCESS) { + DEBUG_FUNCTION_LINE("ERROR: %s", InstallerService::ErrorMessage(result).c_str()); + } + this->systemXMLPatchPossible = ((result = InstallerService::checkSystemXML("storage_slc_installer:/config", this->appInfo->titleId)) == InstallerService::SUCCESS); + if (result != InstallerService::SUCCESS) { + DEBUG_FUNCTION_LINE("ERROR: %s", InstallerService::ErrorMessage(result).c_str()); + } + this->state = STATE_CHECK_PATCH_POSSIBLE_DONE; + +} + +void ApplicationState::getAppInformation() { + DEBUG_FUNCTION_LINE("About to call getInstalledAppInformation"); + this->appInfo = InstallerService::getInstalledAppInformation(); + if (!this->appInfo) { + DEBUG_FUNCTION_LINE("ERROR =("); + this->state = STATE_ERROR; + this->error = ERROR_NO_APP_INSTALLED; + } else { + DEBUG_FUNCTION_LINE("WORKED!"); + this->state = STATE_CHECK_PATCH_POSSIBLE; + } +} + +std::string ApplicationState::ErrorMessage() { + if (this->error == ERROR_NONE) { + return "NONE"; + } else if (this->error == ERROR_NO_APP_INSTALLED) { + return "ERROR_NO_APP_INSTALLED"; + } else if (this->error == ERROR_IOSUHAX_FAILED) { + return "ERROR_IOSUHAX_FAILED"; + } else if (this->error == ERROR_INSTALLER_ERROR) { + return InstallerService::ErrorMessage(this->installerError); + } + return "UNKNOWN_ERROR"; +} + +std::string ApplicationState::ErrorDescription() { + if (this->error == ERROR_NONE) { + return "-"; + } else if (this->error == ERROR_NO_APP_INSTALLED) { + return "No compatible application is installed. A safe installation is not possible."; + } else if (this->error == ERROR_INSTALLER_ERROR) { + return InstallerService::ErrorDescription(this->installerError); + } else if (this->error == ERROR_IOSUHAX_FAILED) { + return "Failed to init IOSUHAX."; + } + return "UNKNOWN_ERROR"; +} + +void ApplicationState::setError(eErrorState err) { + this->state = STATE_ERROR; + this->error = err; + OSEnableHomeButtonMenu(true); +} + +void ApplicationState::handleError() { + +} + +void ApplicationState::printFooter() { + ScreenUtils::printTextOnScreen(CONSOLE_SCREEN_TV, 0, 27, "By Maschell"); + ScreenUtils::printTextOnScreen(CONSOLE_SCREEN_DRC, 0, 17, "By Maschell"); +} + +void ApplicationState::proccessMenuNavigation(Input *input, int maxOptionValue) { + if (input->data.buttons_d & Input::BUTTON_LEFT) { + this->selectedOption--; + } else if (input->data.buttons_d & Input::BUTTON_RIGHT) { + this->selectedOption++; + } + if (this->selectedOption < 0) { + this->selectedOption = maxOptionValue; + } else if (this->selectedOption >= maxOptionValue) { + this->selectedOption = 0; + } +} + +bool ApplicationState::entrySelected(Input *input) { + return input->data.buttons_d & Input::BUTTON_A; +} diff --git a/source/ApplicationState.h b/source/ApplicationState.h new file mode 100644 index 0000000..2e6aa08 --- /dev/null +++ b/source/ApplicationState.h @@ -0,0 +1,80 @@ +#pragma once + +#include +#include +#include "common/common.h" +#include "InstallerService.h" +#include "Input.h" + +class ApplicationState { +public: + + enum eErrorState { + ERROR_NONE, + ERROR_IOSUHAX_FAILED, + ERROR_NO_APP_INSTALLED, + ERROR_INSTALLER_ERROR + }; + + enum eGameState { + STATE_ERROR, + STATE_WELCOME_SCREEN, + STATE_GET_APP_INFORMATION, + STATE_CHECK_PATCH_POSSIBLE, + STATE_CHECK_PATCH_POSSIBLE_DONE, + STATE_INSTALL_CHOOSE_COLDBOOT, + STATE_INSTALL_NO_COLDBOOT_ALLOWED, + STATE_INSTALL_CONFIRM_DIALOG, + STATE_INSTALL_STARTED, + STATE_INSTALL_FST, + STATE_INSTALL_COS, + STATE_INSTALL_RPX, + STATE_INSTALL_SYSTEM_XML, + STATE_INSTALL_SUCCESS + }; + + ApplicationState(); + + void setError(eErrorState error); + + void render(); + + void update(Input *input); + + void checkPatchPossible(); + + void getAppInformation(); + + std::optional appInfo; + + std::string ErrorMessage(); + + std::string ErrorDescription(); + + void handleError(); + + int selectedOption; + + static void printFooter(); + + void proccessMenuNavigation(Input *input, int maxOptionValue); + + static bool entrySelected(Input *input); + + bool installColdboot = false; + InstallerService::eResults installerError = InstallerService::eResults::SUCCESS; + +private: + bool fstPatchPossible = false; + bool cosPatchPossible = false; + bool systemXMLPatchPossible = false; + + eGameState state; + eErrorState error = ERROR_NONE; + uint64_t coldbootTitleId; + _gList_t *coldbootTitle; + bool fstAlreadyPatched; + bool rpxAlreadyPatched; + bool cosAlreadyPatched; + bool tmdValid; +}; diff --git a/source/Input.cpp b/source/Input.cpp new file mode 100644 index 0000000..e0b715d --- /dev/null +++ b/source/Input.cpp @@ -0,0 +1,5 @@ +// +// Created by Maschell on 15.10.2020. +// + +#include "Input.h" diff --git a/source/Input.h b/source/Input.h new file mode 100644 index 0000000..ac91686 --- /dev/null +++ b/source/Input.h @@ -0,0 +1,60 @@ +#pragma once + +#include +#include + +class Input { +public: + //!Constructor + Input() = default; + + //!Destructor + virtual ~Input() = default; + + enum eButtons { + BUTTON_NONE = 0x0000, + VPAD_TOUCH = 0x80000000, + BUTTON_Z = 0x20000, + BUTTON_C = 0x10000, + BUTTON_A = 0x8000, + BUTTON_B = 0x4000, + BUTTON_X = 0x2000, + BUTTON_Y = 0x1000, + BUTTON_1 = BUTTON_Y, + BUTTON_2 = BUTTON_X, + BUTTON_LEFT = 0x0800, + BUTTON_RIGHT = 0x0400, + BUTTON_UP = 0x0200, + BUTTON_DOWN = 0x0100, + BUTTON_ZL = 0x0080, + BUTTON_ZR = 0x0040, + BUTTON_L = 0x0020, + BUTTON_R = 0x0010, + BUTTON_PLUS = 0x0008, + BUTTON_MINUS = 0x0004, + BUTTON_HOME = 0x0002, + BUTTON_SYNC = 0x0001, + STICK_R_LEFT = 0x04000000, + STICK_R_RIGHT = 0x02000000, + STICK_R_UP = 0x01000000, + STICK_R_DOWN = 0x00800000, + STICK_L_LEFT = 0x40000000, + STICK_L_RIGHT = 0x20000000, + STICK_L_UP = 0x10000000, + STICK_L_DOWN = 0x08000000 + }; + + typedef struct { + uint32_t buttons_h; + uint32_t buttons_d; + uint32_t buttons_r; + bool validPointer; + bool touched; + float pointerAngle; + int32_t x; + int32_t y; + } PadData; + + PadData data{}; + PadData lastData{}; +}; diff --git a/source/InstallerService.cpp b/source/InstallerService.cpp new file mode 100644 index 0000000..c373c0e --- /dev/null +++ b/source/InstallerService.cpp @@ -0,0 +1,620 @@ +#include "common/common.h" +#include "utils/logger.h" +#include "WiiUScreen.h" +#include "StringTools.h" +#include "fs/FSUtils.h" +#include "common/fst_structs.h" +#include "InstallerService.h" +#include "utils/utils.h" +#include "utils/pugixml.hpp" +#include +#include +#include +#include +#include +#include + +InstallerService::eResults InstallerService::checkCOS(const std::string &path, char *hash) { + std::string cosFilePath = path + "/code/cos.xml"; + DEBUG_FUNCTION_LINE("Loading %s", cosFilePath.c_str()); + + pugi::xml_document doc; + pugi::xml_parse_result result = doc.load_file(cosFilePath.c_str()); + if (!result) { + DEBUG_FUNCTION_LINE("failed to open %s : %s", cosFilePath.c_str(), result.description()); + return COS_XML_PARSING_FAILED; + } + + patchCOSXMLData(&doc); + + std::stringstream ss; + doc.save(ss, " ", pugi::format_default, pugi::encoding_utf8); + + std::string newHash = Utils::calculateSHA1(ss.str().c_str(), ss.str().size()); + + if (std::string(hash) == newHash) { + DEBUG_FUNCTION_LINE("Success! cos.xml is compatible"); + return SUCCESS; + } else { + DEBUG_FUNCTION_LINE("Hash mismatch! cos.xml is NOT compatible. Expected hash: %s actual hash: %s",hash, newHash.c_str()); + } + + return COS_XML_HASH_MISMATCH; +} + +InstallerService::eResults InstallerService::checkSystemXML(const std::string &path, uint64_t titleId) { + std::string inputFile = std::string(path + "/system.xml"); + + systemXMLInformation *data = nullptr; + for (int i = 0; systemXMLHashInformation[i].titleId != 0; i++) { + if (systemXMLHashInformation[i].titleId == titleId) { + data = &systemXMLHashInformation[i]; + break; + } + } + + if (data == nullptr) { + DEBUG_FUNCTION_LINE("system xml information were not found."); + return SYSTEM_XML_INFORMATION_NOT_FOUND; + } + + DEBUG_FUNCTION_LINE("Setting coldboot to %016llX", data->titleId); + + pugi::xml_document doc; + pugi::xml_parse_result resultSystem = doc.load_file(inputFile.c_str()); + + if (!resultSystem) { + DEBUG_FUNCTION_LINE("Error while parsing %s: %s", inputFile.c_str(), resultSystem.description()); + return SYSTEM_XML_PARSING_FAILED; + } + + char tmp[18]; + snprintf(tmp, 17, "%016llX", data->titleId); + + doc.child("system").child("default_title_id").first_child().set_value(tmp); + if (!doc.child("system").child("log").attribute("length")) { + doc.child("system").child("log").append_attribute("length") = "0"; + } + if (!doc.child("system").child("standby").attribute("length")) { + doc.child("system").child("standby").append_attribute("length") = "0"; + } + if (!doc.child("system").child("ramdisk").attribute("length")) { + doc.child("system").child("ramdisk").append_attribute("length") = "0"; + } + + std::stringstream ss; + doc.save(ss, " ", pugi::format_default, pugi::encoding_utf8); + + std::string newHash = Utils::calculateSHA1(ss.str().c_str(), ss.str().size()); + + if (std::string(data->hash) == newHash || std::string(data->hash2) == newHash) { + DEBUG_FUNCTION_LINE("Success! system.xml is compatible"); + return SUCCESS; + } else { + DEBUG_FUNCTION_LINE("Hash mismatch! system.xml is NOT compatible. Expected hash: %s actual hash: %s", data->hash, newHash.c_str()); + } + return SYSTEM_XML_HASH_MISMATCH; +} + +InstallerService::eResults InstallerService::checkFST(const std::string &path, const char *fstHash) { + std::string fstFilePath = path + "/code/title.fst"; + + uint8_t *fstData = nullptr; + uint32_t fstDataSize = 0; + + DEBUG_FUNCTION_LINE("Trying to load FST from %s", fstFilePath.c_str()); + if (FSUtils::LoadFileToMem(fstFilePath.c_str(), &fstData, &fstDataSize) < 0) { + DEBUG_FUNCTION_LINE("Failed to load title.fst"); + return FAILED_TO_LOAD_FILE; + } + InstallerService::eResults res = patchFSTData(fstData, fstDataSize); + if (res != SUCCESS) { + free(fstData); + fstData = nullptr; + return res; + } + + std::string newHash = Utils::calculateSHA1((const char *) fstData, fstDataSize); + + free(fstData); + fstData = nullptr; + + if (std::string(fstHash) == newHash) { + DEBUG_FUNCTION_LINE("title.fst is compatible"); + return SUCCESS; + } else { + DEBUG_FUNCTION_LINE("title.fst is NOT compatible. expected hash: %s actual hash: %s", fstHash, newHash.c_str()); + return FST_HASH_MISMATCH; + } +} + +bool InstallerService::patchCOSXMLData(pugi::xml_document *doc) { + pugi::xml_node appEntry = doc->child("app"); + appEntry.child("argstr").first_child().set_value("safe.rpx"); + appEntry.child("avail_size").first_child().set_value("00000000"); + appEntry.child("codegen_size").first_child().set_value("02000000"); + appEntry.child("codegen_core").first_child().set_value("80000001"); + appEntry.child("max_size").first_child().set_value("40000000"); + appEntry.child("max_codesize").first_child().set_value("00800000"); + for (pugi::xml_node permission: appEntry.child("permissions").children()) { + auto mask = permission.child("mask"); + mask.first_child().set_value("FFFFFFFFFFFFFFFF"); + } + return true; +} + +std::optional InstallerService::getInstalledAppInformation() { + auto mcpHandle = (int32_t) MCP_Open(); + auto titleCount = (uint32_t) MCP_TitleCount(mcpHandle); + auto *titleList = (MCPTitleListType *) memalign(32, sizeof(MCPTitleListType) * titleCount); + + MCP_TitleListByAppType(mcpHandle, MCP_APP_TYPE_SYSTEM_APPS, &titleCount, titleList, sizeof(MCPTitleListType) * titleCount); + + MCP_Close(mcpHandle); + + DEBUG_FUNCTION_LINE("%d titles found on the WiiU", titleCount); + bool success = false; + + for (uint32_t i = 0; i < titleCount; ++i) { + for (int j = 0; supportedApps[j].titleId != 0; j++) { + if (titleList[i].titleId == supportedApps[j].titleId) { + DEBUG_FUNCTION_LINE("%s is on the Wii U (%s) %d", supportedApps[j].appName, titleList[i].path, sizeof(supportedApps[j].path)); + supportedApps[j].onTheWiiU = true; + strncpy(supportedApps[j].path, titleList[i].path, 255); + success = true; + break; + } + } + } + + free(titleList); + + if (success) { + success = false; + for (int j = 0; supportedApps[j].titleId != 0; j++) { + if (supportedApps[j].onTheWiiU) { + std::string path(supportedApps[j].path); + if (!StringTools::replace(path, "/vol/storage_mlc01", "storage_mlc_installer:")) { + DEBUG_FUNCTION_LINE("Title is not on MLC. This is not expected. %s", path.c_str()); + return {}; + } + strncpy(supportedApps[j].path, path.c_str(), 255); + return supportedApps[j]; + } + } + } + return {}; +} + +InstallerService::eResults InstallerService::patchFSTData(uint8_t *fstData, uint32_t size) { + auto *fstHeader = (FSTHeader *) fstData; + if (strncmp(FSTHEADER_MAGIC, fstHeader->magic, 3) != 0) { + DEBUG_FUNCTION_LINE("FST magic is wrong %s", fstHeader->magic); + return InstallerService::FST_HEADER_MISMATCH; + } + + auto numberOfSections = fstHeader->numberOfSections; + DEBUG_FUNCTION_LINE("Found %d sections", numberOfSections); + auto *sections = (FSTSectionEntry *) (fstData + sizeof(FSTHeader)); + + auto usableSectionIndex = -1; + + for (uint32_t i = 0; i < numberOfSections; i++) { + if (sections[i].hashMode == 2) { + usableSectionIndex = i; + } + } + + if (usableSectionIndex < 0) { + DEBUG_FUNCTION_LINE("Failed to find suitable section"); + return InstallerService::FST_NO_USABLE_SECTION_FOUND; + } else { + DEBUG_FUNCTION_LINE("Section %d can be used as a base", usableSectionIndex); + } + + auto *rootEntry = (FSTNodeEntry *) (fstData + sizeof(FSTHeader) + numberOfSections * sizeof(FSTSectionEntry)); + auto numberOfNodeEntries = rootEntry->directory.lastEntryNumber; + + char *stringTableOffset = (char *) ((uint32_t) rootEntry + (sizeof(FSTNodeEntry) * numberOfNodeEntries)); + + FSTNodeEntry *nodeEntries = rootEntry; + for (uint32_t i = 0; i < numberOfNodeEntries; i++) { + if (sections[nodeEntries[i].sectionNumber].hashMode != 2) { + auto nameOffset = (*((uint32_t *) nodeEntries[i].nameOffset) & 0xFFFFFF00) >> 8; + DEBUG_FUNCTION_LINE("Updating sectionNumber for %s (entry %d)", stringTableOffset + nameOffset, i); + nodeEntries[i].sectionNumber = usableSectionIndex; + } + } + return InstallerService::SUCCESS; +} + +std::string InstallerService::ErrorDescription(InstallerService::eResults error) { + if (error == SUCCESS) { + return "Success"; + } else if (error == NO_COMPATIBLE_APP_INSTALLED) { + return "No compatible application was found on the console."; + } else if (error == FAILED_TO_COPY_FILES) { + return "Unable to copy files."; + } else if (error == FAILED_TO_CHECK_HASH_COPIED_FILES) { + return "The copy of a file has a different hash."; + } else if (error == SYSTEM_XML_INFORMATION_NOT_FOUND) { + return "Expected hashes for the target system.xml were not found."; + } else if (error == SYSTEM_XML_PARSING_FAILED) { + return "Failed to parse the system.xml"; + } else if (error == SYSTEM_XML_HASH_MISMATCH_RESTORE_FAILED) { + return "DO NOT REBOOT BEFORE FIXING THIS: Failed to restore the system.xml after an error."; + } else if (error == SYSTEM_XML_HASH_MISMATCH) { + return "The patched system.xml had an unexpected hash but was successfully restored"; + } else if (error == RPX_HASH_MISMATCH) { + return "The installed safe.rpx had an unexpected hash but was successfully restored"; + } else if (error == RPX_HASH_MISMATCH_RESTORE_FAILED) { + return "DO NOT REBOOT BEFORE FIXING THIS: Failed to restore the safe.rpx after an error."; + } else if (error == COS_XML_PARSING_FAILED) { + return "Failed to parse the cos.xml"; + } else if (error == COS_XML_HASH_MISMATCH) { + return "The patched cos.xml had an unexpected hash but was successfully restored"; + } else if (error == COS_XML_HASH_MISMATCH_RESTORE_FAILED) { + return "DO NOT REBOOT BEFORE FIXING THIS: Failed to restore the cos.xml after an error"; + } else if (error == MALLOC_FAILED) { + return "Failed to allocate memory"; + } else if (error == FST_HASH_MISMATCH) { + return "The patched title.fst had an unexpected hash but was successfully restored"; + } else if (error == FST_HASH_MISMATCH_RESTORE_FAILED) { + return "DO NOT REBOOT BEFORE FIXING THIS: Failed to restore the title.fst after an error"; + } else if (error == FST_HEADER_MISMATCH) { + return "Unexpected header in title.fst found. The file is probably broken."; + } else if (error == FST_NO_USABLE_SECTION_FOUND) { + return "Unable to patch title.fst to allow FailST"; + } else if (error == FAILED_TO_LOAD_FILE) { + return "Failed to load file."; + } else { + return "UNKNOWN ERROR"; + } +} + +std::string InstallerService::ErrorMessage(InstallerService::eResults error) { + if (error == SUCCESS) { + return "Success"; + } else if (error == NO_COMPATIBLE_APP_INSTALLED) { + return "NO_COMPATIBLE_APP_INSTALLED"; + } else if (error == FAILED_TO_COPY_FILES) { + return "FAILED_TO_COPY_FILES"; + } else if (error == FAILED_TO_CHECK_HASH_COPIED_FILES) { + return "FAILED_TO_CHECK_HASH_COPIED_FILES"; + } else if (error == SYSTEM_XML_INFORMATION_NOT_FOUND) { + return "SYSTEM_XML_INFORMATION_NOT_FOUND"; + } else if (error == SYSTEM_XML_PARSING_FAILED) { + return "SYSTEM_XML_PARSING_FAILED"; + } else if (error == SYSTEM_XML_HASH_MISMATCH_RESTORE_FAILED) { + return "SYSTEM_XML_HASH_MISMATCH_RESTORE_FAILED"; + } else if (error == SYSTEM_XML_HASH_MISMATCH) { + return "SYSTEM_XML_HASH_MISMATCH"; + } else if (error == RPX_HASH_MISMATCH) { + return "RPX_HASH_MISMATCH"; + } else if (error == RPX_HASH_MISMATCH_RESTORE_FAILED) { + return "RPX_HASH_MISMATCH_RESTORE_FAILED"; + } else if (error == COS_XML_PARSING_FAILED) { + return "COS_XML_PARSING_FAILED"; + } else if (error == COS_XML_HASH_MISMATCH) { + return "COS_XML_HASH_MISMATCH"; + } else if (error == COS_XML_HASH_MISMATCH_RESTORE_FAILED) { + return "COS_XML_HASH_MISMATCH_RESTORE_FAILED"; + } else if (error == MALLOC_FAILED) { + return "MALLOC_FAILED"; + } else if (error == FST_HASH_MISMATCH) { + return "FST_HASH_MISMATCH"; + } else if (error == FST_HASH_MISMATCH_RESTORE_FAILED) { + return "FST_HASH_MISMATCH_RESTORE_FAILED"; + } else if (error == FST_HEADER_MISMATCH) { + return "FST_HEADER_MISMATCH"; + } else if (error == FST_NO_USABLE_SECTION_FOUND) { + return "FST_NO_USABLE_SECTION_FOUND"; + } else if (error == FAILED_TO_LOAD_FILE) { + return "FAILED_TO_LOAD_FILE"; + } else { + return "UNKNOWN ERROR"; + } + +} + +InstallerService::eResults InstallerService::patchFST(const std::string &path, const char *fstHash) { + std::string fstFilePath = path + "/code/title.fst"; + std::string fstBackupFilePath = path + "/code/backup.fst"; + std::string fstTargetFilePath = path + "/code/title.fst"; + + if (!FSUtils::copyFile(fstFilePath, fstBackupFilePath)) { + DEBUG_FUNCTION_LINE("Failed to copy files"); + return FAILED_TO_COPY_FILES; + } + + std::string srcHash = Utils::hashFile(fstFilePath); + std::string dstHash = Utils::hashFile(fstBackupFilePath); + + if (srcHash != dstHash) { + ::remove(fstBackupFilePath.c_str()); + DEBUG_FUNCTION_LINE("Hashes do not match. %s %s", srcHash.c_str(), dstHash.c_str()); + return FAILED_TO_CHECK_HASH_COPIED_FILES; + } + + uint8_t *fstData = nullptr; + uint32_t fstDataSize = 0; + + DEBUG_FUNCTION_LINE("Trying to load FST from %s", fstFilePath.c_str()); + if (FSUtils::LoadFileToMem(fstFilePath.c_str(), &fstData, &fstDataSize) < 0) { + DEBUG_FUNCTION_LINE("Failed to load title.fst"); + return FAILED_TO_LOAD_FILE; + } + InstallerService::eResults res = patchFSTData(fstData, fstDataSize); + if (res != SUCCESS) { + free(fstData); + return res; + } + FSUtils::saveBufferToFile(fstTargetFilePath.c_str(), fstData, fstDataSize); + free(fstData); + fstData = nullptr; + + std::string newHash = Utils::hashFile(fstTargetFilePath); + + if (std::string(fstHash) == newHash) { + ::remove(fstBackupFilePath.c_str()); + DEBUG_FUNCTION_LINE("Successfully patched the title.fst"); + return SUCCESS; + } else { + DEBUG_FUNCTION_LINE("Hash mismatch! Expected %s but got %s while patching FST", fstHash, newHash.c_str()); + } + + FSUtils::copyFile(fstBackupFilePath, fstTargetFilePath); + + std::string srcHash2 = Utils::hashFile(fstTargetFilePath); + + if (srcHash != srcHash2) { + DEBUG_FUNCTION_LINE("Something went wrong. Failed to restore the title.fst. DO NOT RESTART THE SYSTEM until you manually restored the title.fst"); + return FST_HASH_MISMATCH_RESTORE_FAILED; + } + + ::remove(fstBackupFilePath.c_str()); + + return FST_HASH_MISMATCH; +} + +InstallerService::eResults InstallerService::patchCOS(const std::string &path, char *hash) { + std::string cosFilePath = path + "/code/cos.xml"; + std::string cosBackupFilePath = path + "/code/cback.xml"; + std::string cosTargetFilePath = path + "/code/cos.xml"; + + if (!FSUtils::copyFile(cosFilePath, cosBackupFilePath)) { + DEBUG_FUNCTION_LINE("Failed to copy files"); + return FAILED_TO_COPY_FILES; + } + + std::string srcHash = Utils::hashFile(cosFilePath); + std::string dstHash = Utils::hashFile(cosBackupFilePath); + + if (srcHash != dstHash) { + ::remove(cosBackupFilePath.c_str()); + DEBUG_FUNCTION_LINE("Hashes do not match. %s %s", srcHash.c_str(), dstHash.c_str()); + return FAILED_TO_CHECK_HASH_COPIED_FILES; + } + + pugi::xml_document doc; + pugi::xml_parse_result result = doc.load_file(cosFilePath.c_str()); + + if (!result) { + ::remove(cosBackupFilePath.c_str()); + DEBUG_FUNCTION_LINE("failed to open %s : %s", cosFilePath.c_str(), result.description()); + return COS_XML_PARSING_FAILED; + } + + patchCOSXMLData(&doc); + + doc.save_file(cosTargetFilePath.c_str(), " ", pugi::format_default, pugi::encoding_utf8); + + std::string newHash = Utils::hashFile(cosTargetFilePath); + + if (std::string(hash) == newHash) { + ::remove(cosBackupFilePath.c_str()); + DEBUG_FUNCTION_LINE("Successfully patched the cos.xml"); + return SUCCESS; + } else { + DEBUG_FUNCTION_LINE("Hash mismatch! Expected %s but got %s while patching cos.xml", hash, newHash.c_str()); + } + + FSUtils::copyFile(cosBackupFilePath, cosTargetFilePath); + + std::string srcHash2 = Utils::hashFile(cosTargetFilePath); + + if (srcHash != srcHash2) { + DEBUG_FUNCTION_LINE("Something went wrong. Failed to restore the title.fst. DO NOT RESTART THE SYSTEM until you manually restored the cos.xml"); + return COS_XML_HASH_MISMATCH_RESTORE_FAILED; + } + ::remove(cosBackupFilePath.c_str()); + + return COS_XML_HASH_MISMATCH; +} + +InstallerService::eResults InstallerService::copyRPX(const std::string &path, const uint8_t *rpx_data, size_t size, const std::string &rpx_hash) { + std::string rpxSourceFilePath = path + "/code/safe.rpx"; + std::string rpxBackupFilePath = path + "/code/sbac.rpx"; + std::string rpxTargetFilePath = path + "/code/safe.rpx"; + + if (!FSUtils::copyFile(rpxSourceFilePath, rpxBackupFilePath)) { + DEBUG_CONSOLE_LOG("Failed to copy files"); + return FAILED_TO_COPY_FILES; + } + + std::string srcHash = Utils::hashFile(rpxSourceFilePath); + std::string dstHash = Utils::hashFile(rpxBackupFilePath); + + if (srcHash != dstHash) { + ::remove(rpxBackupFilePath.c_str()); + DEBUG_CONSOLE_LOG("Hashes do not match. %s %s", srcHash.c_str(), dstHash.c_str()); + return FAILED_TO_CHECK_HASH_COPIED_FILES; + } + + if (FSUtils::saveBufferToFile(rpxTargetFilePath.c_str(), (void *) rpx_data, size) < 0) { + DEBUG_CONSOLE_LOG("Cannot save rpx to file %s", rpxTargetFilePath.c_str()); + return SUCCESS; + } + + std::string newHash = Utils::hashFile(rpxTargetFilePath); + + if (StringTools::strCompareIC(rpx_hash, newHash)) { + ::remove(rpxBackupFilePath.c_str()); + DEBUG_CONSOLE_LOG("Successfully patched the safe.rpx"); + return SUCCESS; + } else { + DEBUG_CONSOLE_LOG("Hash mismatch! Expected %s but got %s while patching safe.rpx", rpx_hash.c_str(), newHash.c_str()); + } + + FSUtils::copyFile(rpxBackupFilePath, rpxTargetFilePath); + + std::string srcHash2 = Utils::hashFile(rpxTargetFilePath); + + if (srcHash != srcHash2) { + DEBUG_CONSOLE_LOG("Something went wrong. Failed to restore the safe.rpx. DO NOT RESTART THE SYSTEM until you manually restored the safe.rpx"); + return RPX_HASH_MISMATCH_RESTORE_FAILED; + } + ::remove(rpxBackupFilePath.c_str()); + + return RPX_HASH_MISMATCH; +} + +InstallerService::eResults InstallerService::patchSystemXML(const std::string &path, uint64_t titleId) { + std::string inputFile = std::string(path + "/system.xml"); + std::string backupFile = std::string(path + "/sbackup.xml"); + std::string finalFile = std::string(path + "/system.xml"); + + if (!FSUtils::copyFile(inputFile, backupFile)) { + DEBUG_CONSOLE_LOG("Failed to copy files"); + return FAILED_TO_COPY_FILES; + } + + std::string srcHash = Utils::hashFile(inputFile); + std::string dstHash = Utils::hashFile(backupFile); + + if (srcHash != dstHash) { + ::remove(backupFile.c_str()); + DEBUG_CONSOLE_LOG("Hashes do not match. %s %s", srcHash.c_str(), dstHash.c_str()); + return FAILED_TO_CHECK_HASH_COPIED_FILES; + } + + systemXMLInformation *data = nullptr; + for (int i = 0; systemXMLHashInformation[i].titleId != 0; i++) { + if (systemXMLHashInformation[i].titleId == titleId) { + data = &systemXMLHashInformation[i]; + break; + } + } + + if (data == nullptr) { + ::remove(backupFile.c_str()); + DEBUG_FUNCTION_LINE("Data was NULL"); + return SYSTEM_XML_INFORMATION_NOT_FOUND; + } + + pugi::xml_document doc; + pugi::xml_parse_result resultSystem = doc.load_file(inputFile.c_str()); + if (!resultSystem) { + ::remove(backupFile.c_str()); + DEBUG_FUNCTION_LINE("Error while parsing %s: %s", inputFile.c_str(), resultSystem.description()); + return SYSTEM_XML_PARSING_FAILED; + } + + char tmp[18]; + snprintf(tmp, 17, "%016llX", data->titleId); + + doc.child("system").child("default_title_id").first_child().set_value(tmp); + if (!doc.child("system").child("log").attribute("length")) { + doc.child("system").child("log").append_attribute("length") = "0"; + } + if (!doc.child("system").child("standby").attribute("length")) { + doc.child("system").child("standby").append_attribute("length") = "0"; + } + if (!doc.child("system").child("ramdisk").attribute("length")) { + doc.child("system").child("ramdisk").append_attribute("length") = "0"; + } + doc.save_file(finalFile.c_str(), " ", pugi::format_default, pugi::encoding_utf8); + + std::string newHash = Utils::hashFile(finalFile); + + if (std::string(data->hash) == newHash || std::string(data->hash2) == newHash) { + ::remove(backupFile.c_str()); + DEBUG_FUNCTION_LINE("Successfully patched the system.xml"); + return SUCCESS; + } else { + DEBUG_FUNCTION_LINE("Hash mismatch! Expected %s but got %s while setting default titleID to %s", data->hash, newHash.c_str(), data->titleId); + } + + FSUtils::copyFile(backupFile, finalFile); + + std::string srcHash2 = Utils::hashFile(finalFile); + + if (srcHash != srcHash2) { + DEBUG_FUNCTION_LINE("Something went wrong. Failed to restore the system.xml. DO NOT RESTART THE SYSTEM until you manually restored the system.xml"); + return SYSTEM_XML_HASH_MISMATCH_RESTORE_FAILED; + } + + ::remove(backupFile.c_str()); + + return SYSTEM_XML_HASH_MISMATCH; +} + +uint64_t InstallerService::getColdbootTitleId(const std::string &path) { + std::string inputFile = std::string(path + "/system.xml"); + + pugi::xml_document doc; + pugi::xml_parse_result resultSystem = doc.load_file(inputFile.c_str()); + + if (!resultSystem) { + DEBUG_FUNCTION_LINE("Error while parsing %s: %s", inputFile.c_str(), resultSystem.description()); + return SYSTEM_XML_PARSING_FAILED; + } + + DEBUG_FUNCTION_LINE("%s", doc.child("system").child("default_title_id").first_child().value()); + + uint64_t result = strtoull(doc.child("system").child("default_title_id").first_child().value(), nullptr, 16); + + return result; +} + +InstallerService::eResults InstallerService::checkFSTAlreadyValid(const std::string &path, const std::string &hash) { + std::string filePath = path + "/code/title.fst"; + return checkFileHash(filePath, hash); +} + +InstallerService::eResults InstallerService::checkTMDValid(const std::string &path, const std::string &hash) { + std::string filePath = path + "/code/title.fst"; + return checkFileHash(filePath, hash); +} + +InstallerService::eResults InstallerService::checkCOSAlreadyValid(const std::string &path, const std::string &hash) { + std::string filePath = path + "/code/cos.xml"; + return checkFileHash(filePath, hash); +} + +InstallerService::eResults InstallerService::checkRPXAlreadyValid(const std::string &path, const std::string &hash) { + std::string filePath = path + "/code/safe.rpx"; + return checkFileHash(filePath, hash); +} + +InstallerService::eResults InstallerService::checkFileHash(const std::string &filePath, const std::string &hash) { + uint8_t *fileData = nullptr; + uint32_t fileDataSize = 0; + + if (FSUtils::LoadFileToMem(filePath.c_str(), &fileData, &fileDataSize) < 0) { + return FAILED_TO_LOAD_FILE; + } + + std::string newHash = Utils::calculateSHA1((const char *) fileData, fileDataSize); + + free(fileData); + fileData = nullptr; + + if (StringTools::strCompareIC(hash, newHash)) { + return SUCCESS; + } else { + DEBUG_FUNCTION_LINE("expected %s actual %s", hash.c_str(), newHash.c_str()); + return FST_HASH_MISMATCH; + } +} + diff --git a/source/InstallerService.h b/source/InstallerService.h new file mode 100644 index 0000000..8548fc7 --- /dev/null +++ b/source/InstallerService.h @@ -0,0 +1,82 @@ +#pragma once + +#include "utils/pugixml.hpp" +#include "common/common.h" +#include +#include + +class InstallerService { +public: + enum eResults { + SUCCESS = 0, + NO_COMPATIBLE_APP_INSTALLED = -1, + FAILED_TO_COPY_FILES = -2, + FAILED_TO_CHECK_HASH_COPIED_FILES = -3, + SYSTEM_XML_INFORMATION_NOT_FOUND = -4, + SYSTEM_XML_PARSING_FAILED = -5, + SYSTEM_XML_HASH_MISMATCH_RESTORE_FAILED = -6, + SYSTEM_XML_HASH_MISMATCH = -7, + RPX_HASH_MISMATCH = -8, + RPX_HASH_MISMATCH_RESTORE_FAILED = -9, + COS_XML_PARSING_FAILED = -10, + COS_XML_HASH_MISMATCH = -11, + COS_XML_HASH_MISMATCH_RESTORE_FAILED = -12, + MALLOC_FAILED = -13, + FST_HASH_MISMATCH = -14, + FST_HASH_MISMATCH_RESTORE_FAILED = -15, + FST_HEADER_MISMATCH = -16, + FST_NO_USABLE_SECTION_FOUND = -17, + FAILED_TO_LOAD_FILE = -18, + }; + + static bool isColdBootAllowed(){ + if (OSIsAddressValid(0x00FFFFF8)) { + uint64_t bootedFrom = *((uint64_t *) 0x00FFFFF8); + if ( + bootedFrom == 0x000500101004E000L || // H&S JAP + bootedFrom == 0x000500101004E100L || // H&S USA + bootedFrom == 0x000500101004E200L // H&S EUR + ) { + return true; + } + } + return false; + } + + static eResults patchCOS(const std::string &path, char *hash); + + static eResults checkCOS(const std::string &path, char *hash); + + static eResults checkSystemXML(const std::string &path, uint64_t titleId); + + static eResults checkFST(const std::string &path, const char *fstHash); + + static std::optional getInstalledAppInformation(); + + static std::string ErrorMessage(eResults error); + + static std::string ErrorDescription(eResults error); + + static eResults patchFST(const std::string &path, const char *hash); + + static eResults copyRPX(const std::string &path, const uint8_t *rpx_data, size_t size, const std::string &rpx_hash); + + static eResults patchSystemXML(const std::string &path, uint64_t id); + + static uint64_t getColdbootTitleId(const std::string &path); + + static eResults checkFSTAlreadyValid(const std::string &path, const std::string &hash); + + static eResults checkTMDValid(const std::string &path, const std::string &hash); + + static eResults checkCOSAlreadyValid(const std::string &path, const std::string &hash); + + static eResults checkRPXAlreadyValid(const std::string &path, const std::string &hash); + +private: + static eResults patchFSTData(uint8_t *fstData, uint32_t size); + + static bool patchCOSXMLData(pugi::xml_document *doc); + + static eResults checkFileHash(const std::string &filePath, const std::string &hash); +}; \ No newline at end of file diff --git a/source/ScreenUtils.cpp b/source/ScreenUtils.cpp new file mode 100644 index 0000000..b593368 --- /dev/null +++ b/source/ScreenUtils.cpp @@ -0,0 +1,48 @@ +/**************************************************************************** + * Copyright (C) 2018 Maschell + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + ****************************************************************************/ +#include +#include "ScreenUtils.h" + +void ScreenUtils::printTextOnScreen(ConsoleScreenID screen, int x, int y, const char *msg) { + if (!msg) { + return; + } + if (screen != CONSOLE_SCREEN_DRC) { // Draw TV if it's not DRC exclusive. + OSScreenPutFontEx(SCREEN_TV, x, y, msg); + } + if (screen != CONSOLE_SCREEN_TV) { // Draw DRC if it's not TV exclusive. + OSScreenPutFontEx(SCREEN_DRC, x, y, msg); + } +} + +void ScreenUtils::OSScreenClear(ConsoleScreenID screen, uint32_t color) { + if (screen != CONSOLE_SCREEN_DRC) { // Clear TV if it's not DRC exclusive. + OSScreenClearBufferEx(SCREEN_TV, color); + } + if (screen != CONSOLE_SCREEN_TV) { // Clear DRC if it's not TV exclusive. + OSScreenClearBufferEx(SCREEN_DRC, color); + } +} + +void ScreenUtils::flipBuffers(ConsoleScreenID screen) { + if (screen != CONSOLE_SCREEN_DRC) { // Flip TV buffer if it's not DRC exclusive. + OSScreenFlipBuffersEx(SCREEN_TV); + } + if (screen != CONSOLE_SCREEN_TV) { // Flip DRC buffer if it's not TV exclusive. + OSScreenFlipBuffersEx(SCREEN_DRC); + } +} diff --git a/source/ScreenUtils.h b/source/ScreenUtils.h new file mode 100644 index 0000000..3c1ad44 --- /dev/null +++ b/source/ScreenUtils.h @@ -0,0 +1,55 @@ +/**************************************************************************** + * Copyright (C) 2018-2020 Maschell + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + ****************************************************************************/ +#pragma once + +//! Defines the ID of a display usable with OSScreen. +typedef enum ConsoleScreenID { + //! Represents the TV connected to the system. + CONSOLE_SCREEN_TV = 0, + //! Represents the screen in the DRC (gamepad). + CONSOLE_SCREEN_DRC = 1, + //! Represents both screens + CONSOLE_SCREEN_BOTH = 2, +} ConsoleScreenID; + +class ScreenUtils { +public: + /** + Clears the screen for the given screens + \param screen defines on which screens should be printed + \param x defines the x position (character position) where the text should be printed + \param y defines on which line the text should be printed + \param msg C string that contains the text to be printed. + **/ + static void printTextOnScreen(ConsoleScreenID screen, int x, int y, const char *msg); + + /** + Clears the screen for the given screens + \param screen defines which screens should be cleared + **/ + static void OSScreenClear(ConsoleScreenID screen, uint32_t i); + + /** + Flips the buffer for the given screens + \param screen defines which screens should be flipped. + **/ + static void flipBuffers(ConsoleScreenID screen); + +private: + ScreenUtils() = default; + ~ScreenUtils() = default; +}; \ No newline at end of file diff --git a/source/StringTools.cpp b/source/StringTools.cpp new file mode 100644 index 0000000..f313e3e --- /dev/null +++ b/source/StringTools.cpp @@ -0,0 +1,234 @@ +/*************************************************************************** + * Copyright (C) 2010 + * by Dimok + * + * This software is provided 'as-is', without any express or implied + * warranty. In no event will the authors be held liable for any + * damages arising from the use of this software. + * + * Permission is granted to anyone to use this software for any + * purpose, including commercial applications, and to alter it and + * redistribute it freely, subject to the following restrictions: + * + * 1. The origin of this software must not be misrepresented; you + * must not claim that you wrote the original software. If you use + * this software in a product, an acknowledgment in the product + * documentation would be appreciated but is not required. + * + * 2. Altered source versions must be plainly marked as such, and + * must not be misrepresented as being the original software. + * + * 3. This notice may not be removed or altered from any source + * distribution. + * + * for WiiXplorer 2010 + ***************************************************************************/ +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include "StringTools.h" + + +BOOL StringTools::EndsWith(const std::string &a, const std::string &b) { + if (b.size() > a.size()) + return false; + return std::equal(a.begin() + a.size() - b.size(), a.end(), b.begin()); +} + +const char *StringTools::byte_to_binary(int32_t x) { + static char b[9]; + b[0] = '\0'; + + int32_t z; + for (z = 128; z > 0; z >>= 1) { + strcat(b, ((x & z) == z) ? "1" : "0"); + } + + return b; +} + +std::string StringTools::removeCharFromString(std::string &input, char toBeRemoved) { + std::string output = input; + size_t position; + while (1) { + position = output.find(toBeRemoved); + if (position == std::string::npos) + break; + output.erase(position, 1); + } + return output; +} + +const char *StringTools::fmt(const char *format, ...) { + static char strChar[512]; + strChar[0] = 0; + + va_list va; + va_start(va, format); + if ((vsprintf(strChar, format, va) >= 0)) { + va_end(va); + return (const char *) strChar; + } + va_end(va); + + return NULL; +} + +const wchar_t *StringTools::wfmt(const char *format, ...) { + static char tmp[512]; + static wchar_t strWChar[512]; + strWChar[0] = 0; + tmp[0] = 0; + + if (!format) + return (const wchar_t *) strWChar; + + if (strcmp(format, "") == 0) + return (const wchar_t *) strWChar; + + va_list va; + va_start(va, format); + if ((vsprintf(tmp, format, va) >= 0)) { + int32_t bt; + int32_t strlength = strlen(tmp); + bt = mbstowcs(strWChar, tmp, (strlength < 512) ? strlength : 512); + + if (bt > 0) { + strWChar[bt] = 0; + return (const wchar_t *) strWChar; + } + } + va_end(va); + + return NULL; +} + +int32_t StringTools::strprintf(std::string &str, const char *format, ...) { + static char tmp[512]; + tmp[0] = 0; + int32_t result = 0; + + va_list va; + va_start(va, format); + if ((vsprintf(tmp, format, va) >= 0)) { + str = tmp; + result = str.size(); + } + va_end(va); + + return result; +} + +std::string StringTools::strfmt(const char *format, ...) { + std::string str; + static char tmp[512]; + tmp[0] = 0; + + va_list va; + va_start(va, format); + if ((vsprintf(tmp, format, va) >= 0)) { + str = tmp; + } + va_end(va); + + return str; +} + +BOOL StringTools::char2wchar_t(const char *strChar, wchar_t *dest) { + if (!strChar || !dest) + return false; + + int32_t bt; + bt = mbstowcs(dest, strChar, strlen(strChar)); + if (bt > 0) { + dest[bt] = 0; + return true; + } + + return false; +} + +int32_t StringTools::strtokcmp(const char *string, const char *compare, const char *separator) { + if (!string || !compare) + return -1; + + char TokCopy[512]; + strncpy(TokCopy, compare, sizeof(TokCopy)); + TokCopy[511] = '\0'; + + char *strTok = strtok(TokCopy, separator); + + while (strTok != NULL) { + if (strcasecmp(string, strTok) == 0) { + return 0; + } + strTok = strtok(NULL, separator); + } + + return -1; +} + +int32_t StringTools::strextcmp(const char *string, const char *extension, char seperator) { + if (!string || !extension) + return -1; + + char *ptr = strrchr(string, seperator); + if (!ptr) + return -1; + + return strcasecmp(ptr + 1, extension); +} + + +std::vector StringTools::stringSplit(const std::string &inValue, const std::string &splitter) { + std::string value = inValue; + std::vector result; + while (true) { + uint32_t index = value.find(splitter); + if (index == std::string::npos) { + result.push_back(value); + break; + } + std::string first = value.substr(0, index); + result.push_back(first); + if (index + splitter.size() == value.length()) { + result.push_back(""); + break; + } + if (index + splitter.size() > value.length()) { + break; + } + value = value.substr(index + splitter.size(), value.length()); + } + return result; +} + +bool StringTools::findStringIC(const std::string &strHaystack, const std::string &strNeedle) { + auto it = std::search( + strHaystack.begin(), strHaystack.end(), + strNeedle.begin(), strNeedle.end(), + [](char ch1, char ch2) { return std::toupper(ch1) == std::toupper(ch2); } + ); + return (it != strHaystack.end()); +} + +bool StringTools::replace(std::string &str, const std::string &from, const std::string &to) { + size_t start_pos = str.find(from); + if (start_pos == std::string::npos) { + return false; + } + str.replace(start_pos, from.length(), to); + return true; +} + + +bool StringTools::strCompareIC(const std::string &str1, const std::string &str2) { + return str1.size() == str2.size() && std::equal(str1.begin(), str1.end(), str2.begin(), [](auto a, auto b) { return std::tolower(a) == std::tolower(b); }); +} \ No newline at end of file diff --git a/source/StringTools.h b/source/StringTools.h new file mode 100644 index 0000000..450d760 --- /dev/null +++ b/source/StringTools.h @@ -0,0 +1,95 @@ +/*************************************************************************** + * Copyright (C) 2010 + * by Dimok + * + * This software is provided 'as-is', without any express or implied + * warranty. In no event will the authors be held liable for any + * damages arising from the use of this software. + * + * Permission is granted to anyone to use this software for any + * purpose, including commercial applications, and to alter it and + * redistribute it freely, subject to the following restrictions: + * + * 1. The origin of this software must not be misrepresented; you + * must not claim that you wrote the original software. If you use + * this software in a product, an acknowledgment in the product + * documentation would be appreciated but is not required. + * + * 2. Altered source versions must be plainly marked as such, and + * must not be misrepresented as being the original software. + * + * 3. This notice may not be removed or altered from any source + * distribution. + * + * for WiiXplorer 2010 + ***************************************************************************/ +#pragma once + +#include +#include +#include +#include +#include + +class StringTools { +public: + static BOOL EndsWith(const std::string &a, const std::string &b); + + static const char *byte_to_binary(int32_t x); + + static std::string removeCharFromString(std::string &input, char toBeRemoved); + + static const char *fmt(const char *format, ...); + + static const wchar_t *wfmt(const char *format, ...); + + static int32_t strprintf(std::string &str, const char *format, ...); + + static std::string strfmt(const char *format, ...); + + static BOOL char2wchar_t(const char *src, wchar_t *dest); + + static int32_t strtokcmp(const char *string, const char *compare, const char *separator); + + static int32_t strextcmp(const char *string, const char *extension, char seperator); + + static const char *FullpathToFilename(const char *path) { + if (!path) + return path; + + const char *ptr = path; + const char *Filename = ptr; + + while (*ptr != '\0') { + if (ptr[0] == '/' && ptr[1] != '\0') + Filename = ptr + 1; + + ++ptr; + } + + return Filename; + } + + static void RemoveDoubleSlashs(std::string &str) { + uint32_t length = str.size(); + + //! clear path of double slashes + for (uint32_t i = 1; i < length; ++i) { + if (str[i - 1] == '/' && str[i] == '/') { + str.erase(i, 1); + i--; + length--; + } + } + } + + static std::vector stringSplit(const std::string &value, const std::string &splitter); + + // https://stackoverflow.com/a/19839371 + static bool findStringIC(const std::string &strHaystack, const std::string &strNeedle); + + // https://stackoverflow.com/a/3418285 + static bool replace(std::string& str, const std::string& from, const std::string& to); + + static bool strCompareIC(const std::string &str1, const std::string &str2); +}; diff --git a/source/VPADInput.h b/source/VPADInput.h new file mode 100644 index 0000000..e7cd0ec --- /dev/null +++ b/source/VPADInput.h @@ -0,0 +1,59 @@ +#pragma once +/**************************************************************************** + * Copyright (C) 2015 Dimok + * + * This program is free software: you can redistribute it and/or modify + * it under the terms of the GNU General Public License as published by + * the Free Software Foundation, either version 3 of the License, or + * (at your option) any later version. + * + * This program is distributed in the hope that it will be useful, + * but WITHOUT ANY WARRANTY; without even the implied warranty of + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the + * GNU General Public License for more details. + * + * You should have received a copy of the GNU General Public License + * along with this program. If not, see . + ****************************************************************************/ + +#include +#include "Input.h" + +class VPadInput : public Input { +public: + //!Constructor + VPadInput() { + memset(&vpad, 0, sizeof(vpad)); + } + + //!Destructor + ~VPadInput() override {} + + bool update(int32_t width, int32_t height) { + lastData = data; + + VPADReadError vpadError = VPAD_READ_NO_SAMPLES; + VPADRead(VPAD_CHAN_0, &vpad, 1, &vpadError); + + if (vpadError == VPAD_READ_SUCCESS) { + data.buttons_r = vpad.release; + data.buttons_h = vpad.hold; + data.buttons_d = vpad.trigger; + data.validPointer = !vpad.tpNormal.validity; + data.touched = vpad.tpNormal.touched; + + VPADGetTPCalibratedPoint(VPAD_CHAN_0, &tpCalib, &vpad.tpFiltered1); + + //! calculate the screen offsets + data.x = -(width >> 1) + (int32_t) (((float) tpCalib.x / 1280.0f) * (float) width); + data.y = -(height >> 1) + (int32_t) (float) height - (((float) tpCalib.y / 720.0f) * (float) height); + + return true; + } + return false; + } + +private: + VPADStatus vpad{}; + VPADTouchData tpCalib{}; +}; \ No newline at end of file diff --git a/source/WiiUScreen.cpp b/source/WiiUScreen.cpp new file mode 100644 index 0000000..ead7b52 --- /dev/null +++ b/source/WiiUScreen.cpp @@ -0,0 +1,94 @@ +#include "WiiUScreen.h" +#include "ScreenUtils.h" +#include +#include +#include + +void *WiiUScreen::sBufferTV = nullptr; +void *WiiUScreen::sBufferDRC = nullptr; +uint32_t WiiUScreen::sBufferSizeTV = 0; +uint32_t WiiUScreen::sBufferSizeDRC = 0; +bool WiiUScreen::sConsoleHasForeground = true; +uint32_t WiiUScreen::consoleColor = 0x041F60FF; +uint32_t WiiUScreen::consoleCursorY = 0; + + +uint32_t WiiUScreen::ProcCallbackAcquired(void *context) { + MEMHeapHandle heap = MEMGetBaseHeapHandle(MEM_BASE_HEAP_MEM1); + if (sBufferSizeTV) { + sBufferTV = MEMAllocFromFrmHeapEx(heap, sBufferSizeTV, 4); + } + + if (sBufferSizeDRC) { + sBufferDRC = MEMAllocFromFrmHeapEx(heap, sBufferSizeDRC, 4); + } + + sConsoleHasForeground = true; + OSScreenSetBufferEx(SCREEN_TV, sBufferTV); + OSScreenSetBufferEx(SCREEN_DRC, sBufferDRC); + return 0; +} + +uint32_t WiiUScreen::ProcCallbackReleased(void *context) { + MEMHeapHandle heap = MEMGetBaseHeapHandle(MEM_BASE_HEAP_MEM1); + MEMFreeByStateToFrmHeap(heap, CONSOLE_FRAME_HEAP_TAG); + sConsoleHasForeground = FALSE; + return 0; +} + +bool WiiUScreen::Init() { + OSScreenInit(); + sBufferSizeTV = OSScreenGetBufferSizeEx(SCREEN_TV); + sBufferSizeDRC = OSScreenGetBufferSizeEx(SCREEN_DRC); + + WiiUScreen::ProcCallbackAcquired(nullptr); + OSScreenEnableEx(SCREEN_TV, 1); + OSScreenEnableEx(SCREEN_DRC, 1); + + ScreenUtils::OSScreenClear(CONSOLE_SCREEN_BOTH, WiiUScreen::consoleColor); + + ProcUIRegisterCallback(PROCUI_CALLBACK_ACQUIRE, WiiUScreen::ProcCallbackAcquired, nullptr, 100); + ProcUIRegisterCallback(PROCUI_CALLBACK_RELEASE, WiiUScreen::ProcCallbackReleased, nullptr, 100); + return true; +} + +void WiiUScreen::DeInit() { + if (sConsoleHasForeground) { + OSScreenShutdown(); + WiiUScreen::ProcCallbackReleased(nullptr); + } +} + +void WiiUScreen::drawLinef(const char *fmt, ...) { + char *buf = (char *) MEMAllocFromDefaultHeapEx(PRINTF_BUFFER_LENGTH, 4); + va_list va; + + if (!buf) { + return; + } + + va_start(va, fmt); + vsnprintf(buf, PRINTF_BUFFER_LENGTH, fmt, va); + + WiiUScreen::drawLine(buf); + + MEMFreeToDefaultHeap(buf); + va_end(va); +} + +void WiiUScreen::drawLine() { + WiiUScreen::drawLine(""); +} + +void WiiUScreen::drawLine(const char *msg) { + ScreenUtils::printTextOnScreen(CONSOLE_SCREEN_BOTH, 0, consoleCursorY++, msg); +} + +void WiiUScreen::flipBuffers() { + ScreenUtils::flipBuffers(CONSOLE_SCREEN_BOTH); +} + +void WiiUScreen::clearScreen() { + ScreenUtils::OSScreenClear(CONSOLE_SCREEN_BOTH, WiiUScreen::consoleColor); + consoleCursorY = 0; +} diff --git a/source/WiiUScreen.h b/source/WiiUScreen.h new file mode 100644 index 0000000..69127ef --- /dev/null +++ b/source/WiiUScreen.h @@ -0,0 +1,53 @@ +#pragma once + +#include +#include + +#include +#include +#include +#include +#include +#include +#include +#include "utils/logger.h" + +#define NUM_LINES (16) +#define LINE_LENGTH (128) +#define CONSOLE_FRAME_HEAP_TAG (0x000DECAF) +#define PRINTF_BUFFER_LENGTH 2048 + +#define DEBUG_CONSOLE_LOG(FMT, ARGS...)do { \ + DEBUG_FUNCTION_LINE(FMT, ## ARGS); \ + WiiUScreen::drawLinef(FMT, ## ARGS); \ + } while (0) + + +class WiiUScreen { + +public: + static uint32_t ProcCallbackAcquired(void *context); + + static uint32_t ProcCallbackReleased(void *context); + + static bool Init(); + + static void DeInit(); + + static void drawLinef(const char *fmt, ...); + + static void drawLine(const char *fmt); + + static void drawLine(); + + static void flipBuffers(); + + static void clearScreen(); + +private: + static void *sBufferTV, *sBufferDRC; + static uint32_t sBufferSizeTV, sBufferSizeDRC; + static bool sConsoleHasForeground; + static uint32_t consoleColor; + static uint32_t consoleCursorY; +}; \ No newline at end of file diff --git a/source/common/common.cpp b/source/common/common.cpp new file mode 100644 index 0000000..3baa01c --- /dev/null +++ b/source/common/common.cpp @@ -0,0 +1,117 @@ +#include "common.h" + +systemXMLInformation systemXMLHashInformation[] = { + {WII_U_MENU_JAP, 0x0005001010040000L, "2645065A42D18D390C78543E3C4FE7E1D1957A63", "5E5C707E6DAF82393E93971BE98BE3B12204932A"}, + {WII_U_MENU_USA, 0x0005001010040100L, "124562D41A02C7112DDD5F9A8F0EE5DF97E23471", "DC0F9941E99C629625419F444B5A5B177A67309F"}, + {WII_U_MENU_EUR, 0x0005001010040200L, "F06041A4E5B3F899E748F1BAEB524DE058809F1D", "A0273C466DE15F33EC161BCD908B5BFE359FE6E0"}, + {HEALTH_SAFETY_JPN, 0x000500101004E000L, "066D672824128713F0A7D156142A68B998080148", "2849DE91560F6667FE7415F89FC916BE3A27DE75"}, + {HEALTH_SAFETY_USA, 0x000500101004E100L, "0EBCA1DFC0AB7A6A7FE8FB5EAF23179621B726A1", "83CF5B1CE0B64C51D15B1EFCAD659063790EB590"}, + {HEALTH_SAFETY_EUR, 0x000500101004E200L, "DE46EC3E9B823ABA6CB0638D0C4CDEEF9C793BDD", "ED59630448EC6946F3E51618DA3681EC3A84D391"}, + {MAX_SYSTEM_XML_DEFAULT_TITLE_ID, 0, {'\0'}, {'\0'}}, +}; + +appInformation supportedApps[] = { + {0x000500101004E000L, "Health and Safety Information [JPN]", false, {'\0'}, "9D34DDD91604D781FDB0727AC75021833304964C", "F6EBF7BC8AE3AF3BB8A42E0CF3FDA051278AEB03", "2CF358E1F51932D305911A6836ED37DB0F94ABE4"}, + {0x000500101004E100L, "Health and Safety Information [USA]", false, {'\0'}, "045734666A36C7EF0258A740855886EBDB20D59B", "F6EBF7BC8AE3AF3BB8A42E0CF3FDA051278AEB03", "C53C219FB8F777F9AB8F430D6BE4BC034D5638BA"}, + {0x000500101004E200L, "Health and Safety Information [PAL]", false, {'\0'}, "130A76F8B36B36D43B88BBC74393D9AFD9CFD2A4", "F6EBF7BC8AE3AF3BB8A42E0CF3FDA051278AEB03", "4A29A60E5FBDAF410B7C22ECAEBDDBF29D1A874E"}, + {0, nullptr, false, {'\0'}, {'\0'}, {'\0'}}, +}; + +gList_t GameList[] = { + {0x0005001010040000L, "Wii U Menu [JPN]", MCP_REGION_JAPAN}, + {0x0005001010040100L, "Wii U Menu [USA]", MCP_REGION_USA}, + {0x0005001010040200L, "Wii U Menu [PAL]", MCP_REGION_EUROPE}, + + {0x000500101004E000L, "Health and Safety Information [JPN]", MCP_REGION_JAPAN}, + {0x000500101004E100L, "Health and Safety Information [USA]", MCP_REGION_USA}, + {0x000500101004E200L, "Health and Safety Information [PAL]", MCP_REGION_EUROPE}, + + {0x0005000010179A00L, "Kawashima: Motto Nou wo Kitaeru Otona no DS Training [JPN]", MCP_REGION_JAPAN}, + {0x0005000010179B00L, "Brain Age: Train Your Brain in Minutes a Day! [USA]", MCP_REGION_USA}, + {0x0005000010179C00L, "Dr. Kawashima's Brain Training [PAL]", MCP_REGION_EUROPE}, + + {0x0005000010179D00L, "Catch! Touch! Yoshi! [JPN]", MCP_REGION_JAPAN}, + {0x0005000010179E00L, "Yoshi Touch & Go [USA]", MCP_REGION_USA}, + {0x0005000010179F00L, "Yoshi Touch & Go [PAL]", MCP_REGION_EUROPE}, + + {0x0005000010195600L, "Mario Kart DS [JPN]", MCP_REGION_JAPAN}, + {0x0005000010195700L, "Mario Kart DS [USA]", MCP_REGION_USA}, + {0x0005000010195800L, "Mario Kart DS [PAL]", MCP_REGION_EUROPE}, + + {0x0005000010195900L, "New Super Mario Bros. [JPN]", MCP_REGION_JAPAN}, + {0x0005000010195A00L, "New Super Mario Bros. [USA]", MCP_REGION_USA}, + {0x0005000010195B00L, "New Super Mario Bros. [PAL]", MCP_REGION_EUROPE}, + + {0x0005000010198800L, "Yoshi's Island DS [JPN]", MCP_REGION_JAPAN}, + {0x0005000010198900L, "Yoshi's Island DS [USA]", MCP_REGION_USA}, + {0x0005000010198A00L, "Yoshi's Island DS [PAL]", MCP_REGION_EUROPE}, + + {0x0005000010198B00L, "Yawaraka Atama Juku [JPN]", MCP_REGION_JAPAN}, + {0x0005000010198C00L, "Big Brain Academy [USA]", MCP_REGION_USA}, + {0x0005000010198D00L, "Big Brain Academy [PAL]", MCP_REGION_EUROPE}, + + {0x00050000101A1E00L, "Sawaru: Made in Wario [JPN]", MCP_REGION_JAPAN}, + {0x00050000101A1F00L, "WarioWare: Touched! [USA]", MCP_REGION_USA}, + {0x00050000101A2000L, "WarioWare: Touched! [PAL]", MCP_REGION_EUROPE}, + + {0x00050000101A2100L, "Mario & Luigi RPG 2x2 [JPN]", MCP_REGION_JAPAN}, + {0x00050000101A2200L, "Mario & Luigi: Partners in Time [USA]", MCP_REGION_USA}, + {0x00050000101A2300L, "Mario & Luigi: Partners in Time [PAL]", MCP_REGION_EUROPE}, + + {0x00050000101A5200L, "Donkey Kong: Jungle Climber [JPN]", MCP_REGION_JAPAN}, + {0x00050000101A5300L, "DK: Jungle Climber [USA]", MCP_REGION_USA}, + {0x00050000101A5400L, "Donkey Kong: Jungle Climber [PAL]", MCP_REGION_EUROPE}, + + {0x00050000101A5500L, "Hoshi no Kirby: Sanjou! Dorocche Dan [JPN]", MCP_REGION_JAPAN}, + {0x00050000101A5600L, "Kirby: Squeak Squad [USA]", MCP_REGION_USA}, + {0x00050000101A5700L, "Kirby: Mouse Attack [PAL]", MCP_REGION_EUROPE}, + + {0x00050000101ABD00L, "Kaitou Wario the Seven [JPN]", MCP_REGION_JAPAN}, + {0x00050000101ABE00L, "Wario: Master of Disguise [USA]", MCP_REGION_USA}, + {0x00050000101ABF00L, "Wario: Master of Disguise [PAL]", MCP_REGION_EUROPE}, + + {0x00050000101AC000L, "Star Fox Command [JPN]", MCP_REGION_JAPAN}, + {0x00050000101AC100L, "Star Fox Command [USA]", MCP_REGION_USA}, + {0x00050000101AC200L, "Star Fox Command [PAL]", MCP_REGION_EUROPE}, + + {0x00050000101B8800L, "Touch! Kirby's Magic Paintbrush [JPN]", MCP_REGION_JAPAN}, + {0x00050000101B8900L, "Kirby: Canvas Curse [USA]", MCP_REGION_USA}, + {0x00050000101B8A00L, "Kirby: Power Paintbrush [PAL]", MCP_REGION_EUROPE}, + + {0x00050000101B8B00L, "Zelda no Densetsu: Daichi no Kiteki [JPN]", MCP_REGION_JAPAN}, + {0x00050000101B8C00L, "The Legend of Zelda: Spirit Tracks [USA]", MCP_REGION_USA}, + {0x00050000101B8D00L, "The Legend of Zelda: Spirit Tracks [PAL]", MCP_REGION_EUROPE}, + + {0x00050000101C3300L, "Super Mario 64 DS [JPN]", MCP_REGION_JAPAN}, + {0x00050000101C3400L, "Super Mario 64 DS [USA]", MCP_REGION_USA}, + {0x00050000101C3500L, "Super Mario 64 DS [PAL]", MCP_REGION_EUROPE}, + + {0x00050000101C3600L, "Zelda no Densetsu: Mugen no Sunadokei [JPN]", MCP_REGION_JAPAN}, + {0x00050000101C3700L, "The Legend of Zelda: Phantom Hourglass [USA]", MCP_REGION_USA}, + {0x00050000101C3800L, "The Legend of Zelda: Phantom Hourglass [PAL]", MCP_REGION_EUROPE}, + + {0x00050000101C8600L, "Atsumete! Kirby [JPN]", MCP_REGION_JAPAN}, + {0x00050000101C8700L, "Kirby Mass Attack [USA]", MCP_REGION_USA}, + {0x00050000101C8800L, "Kirby Mass Attack [PAL]", MCP_REGION_EUROPE}, + + {0x00050000101CC200L, "Pokemon Ranger [JPN]", MCP_REGION_JAPAN}, + {0x00050000101CC300L, "Pokemon Ranger [USA]", MCP_REGION_USA}, + {0x00050000101CC400L, "Pokemon Ranger [PAL]", MCP_REGION_EUROPE}, + + {0x00050000101D1F00L, "Oideyo Doubutsu no Mori [JPN]", MCP_REGION_JAPAN}, + {0x00050000101D2000L, "Animal Crossing: Wild World [USA]", MCP_REGION_USA}, + {0x00050000101D2100L, "Animal Crossing: Wild World [PAL]", MCP_REGION_EUROPE}, + + {0x00050000101E0C00L, "Pokemon Fushigi no Dungeon: Sora no Tankentai [JPN]", MCP_REGION_JAPAN}, + {0x00050000101E0D00L, "Pokemon Mystery Dungeon: Explorers of Sky [USA]", MCP_REGION_USA}, + {0x00050000101E0E00L, "Pokemon Mystery Dungeon: Explorers of Sky [PAL]", MCP_REGION_EUROPE}, + + {0x00050000101E0F00L, "Pokemon Ranger: Batonnage [JPN]", MCP_REGION_JAPAN}, + {0x00050000101E1000L, "Pokemon Ranger: Shadows of Almia [USA]", MCP_REGION_USA}, + {0x00050000101E1100L, "Pokemon Ranger: Shadows of Almia [PAL]", MCP_REGION_EUROPE}, + + {0x00050000101E6F00L, "Pokemon Ranger: Hikari no Kiseki [JPN]", MCP_REGION_JAPAN}, + {0x00050000101E7000L, "Pokemon Ranger: Guardian Signs [USA]", MCP_REGION_USA}, + {0x00050000101E7100L, "Pokemon Ranger: Guardian Signs [PAL]", MCP_REGION_EUROPE}, + {0, "", MCP_REGION_JAPAN}, +}; diff --git a/source/common/common.h b/source/common/common.h new file mode 100644 index 0000000..9a18234 --- /dev/null +++ b/source/common/common.h @@ -0,0 +1,51 @@ +#pragma once + +#include +#include + +#ifdef __cplusplus +extern "C" { +#endif + +#define VERSION "v0.1" + +enum SYSTEM_XML_DEFAULT_TITLE_ID { + WII_U_MENU_EUR, + WII_U_MENU_USA, + WII_U_MENU_JAP, + HEALTH_SAFETY_EUR, + HEALTH_SAFETY_USA, + HEALTH_SAFETY_JPN, + MAX_SYSTEM_XML_DEFAULT_TITLE_ID +}; + +typedef struct systemXMLInformation { + SYSTEM_XML_DEFAULT_TITLE_ID type; + uint64_t titleId; + char hash[41]; + char hash2[41]; +} systemXMLInformation; + +typedef struct compatApps { + uint64_t titleId; + const char *appName; + bool onTheWiiU; + char path[255]; + char fstHash[41]; + char cosHash[41]; + char tmdHash[41]; +} appInformation; + +typedef struct _gList_t { + uint64_t tid; + char name[64]; + MCPRegion region; +} gList_t; + +extern systemXMLInformation systemXMLHashInformation[]; +extern appInformation supportedApps[]; +extern _gList_t GameList[]; + +#ifdef __cplusplus +} +#endif \ No newline at end of file diff --git a/source/common/fst_structs.h b/source/common/fst_structs.h new file mode 100644 index 0000000..2da9f1e --- /dev/null +++ b/source/common/fst_structs.h @@ -0,0 +1,62 @@ +#pragma once + +#include + +#define FSTHEADER_MAGIC "FST" + +typedef struct WUT_PACKED FSTHeader { + char magic[3]; + uint8_t version; + uint32_t blockSize; + uint32_t numberOfSections; + uint8_t hashDisabled; + WUT_PADDING_BYTES(32 - 13); +} FSTHeader; +WUT_CHECK_OFFSET(FSTHeader, 0x00, magic); +WUT_CHECK_OFFSET(FSTHeader, 0x03, version); +WUT_CHECK_OFFSET(FSTHeader, 0x04, blockSize); +WUT_CHECK_OFFSET(FSTHeader, 0x08, numberOfSections); +WUT_CHECK_OFFSET(FSTHeader, 0x0C, hashDisabled); +WUT_CHECK_SIZE(FSTHeader, 0x20); + +typedef struct WUT_PACKED FSTSectionEntry { + uint32_t addressInVolumeBlocks; + uint32_t sizeInVolumeBlocks; + uint64_t ownerID; + uint32_t groupID; + uint8_t hashMode; + WUT_PADDING_BYTES(32 - 21); +} FSTSectionEntry; +WUT_CHECK_OFFSET(FSTSectionEntry, 0x00, addressInVolumeBlocks); +WUT_CHECK_OFFSET(FSTSectionEntry, 0x04, sizeInVolumeBlocks); +WUT_CHECK_OFFSET(FSTSectionEntry, 0x08, ownerID); +WUT_CHECK_OFFSET(FSTSectionEntry, 0x10, groupID); +WUT_CHECK_OFFSET(FSTSectionEntry, 0x14, hashMode); +WUT_CHECK_SIZE(FSTSectionEntry, 0x20); + +typedef struct WUT_PACKED FSTNodeEntry { + uint8_t type; + char nameOffset[3]; // 24 bit int + union { + struct { + uint32_t parentEntryNumber; + uint32_t lastEntryNumber; + } directory; + struct { + uint32_t addressInBlocks; + uint32_t size; + } file; + }; + uint16_t permission; + uint16_t sectionNumber; +} FSTNodeEntry; +WUT_CHECK_OFFSET(FSTNodeEntry, 0x00, type); +WUT_CHECK_OFFSET(FSTNodeEntry, 0x01, nameOffset); +WUT_CHECK_OFFSET(FSTNodeEntry, 0x04, directory.parentEntryNumber); +WUT_CHECK_OFFSET(FSTNodeEntry, 0x08, directory.lastEntryNumber); +WUT_CHECK_OFFSET(FSTNodeEntry, 0x04, file.addressInBlocks); +WUT_CHECK_OFFSET(FSTNodeEntry, 0x08, file.size); +WUT_CHECK_OFFSET(FSTNodeEntry, 0x0C, permission); +WUT_CHECK_OFFSET(FSTNodeEntry, 0x0E, sectionNumber); +WUT_CHECK_SIZE(FSTNodeEntry, 0x10); + diff --git a/source/fs/CFile.cpp b/source/fs/CFile.cpp new file mode 100644 index 0000000..17a6c5e --- /dev/null +++ b/source/fs/CFile.cpp @@ -0,0 +1,173 @@ +#include +#include +#include +#include "CFile.hpp" + +CFile::CFile() { + iFd = -1; + mem_file = NULL; + filesize = 0; + pos = 0; +} + +CFile::CFile(const std::string &filepath, eOpenTypes mode) { + iFd = -1; + this->open(filepath, mode); +} + +CFile::CFile(const uint8_t *mem, int32_t size) { + iFd = -1; + this->open(mem, size); +} + +CFile::~CFile() { + this->close(); +} + +int32_t CFile::open(const std::string &filepath, eOpenTypes mode) { + this->close(); + int32_t openMode = 0; + + // This depend on the devoptab implementation. + // see https://github.com/devkitPro/wut/blob/master/libraries/wutdevoptab/devoptab_fs_open.c#L21 fpr reference + + switch (mode) { + default: + case ReadOnly: // file must exist + openMode = O_RDONLY; + break; + case WriteOnly: // file will be created / zerod + openMode = O_TRUNC | O_CREAT | O_WRONLY; + break; + case ReadWrite: // file must exist + openMode = O_RDWR; + break; + case Append: // append to file, file will be created if missing. write only + openMode = O_CREAT | O_APPEND | O_WRONLY; + break; + } + + //! Using fopen works only on the first launch as expected + //! on the second launch it causes issues because we don't overwrite + //! the .data sections which is needed for a normal application to re-init + //! this will be added with launching as RPX + iFd = ::open(filepath.c_str(), openMode); + if (iFd < 0) + return iFd; + + + filesize = ::lseek(iFd, 0, SEEK_END); + ::lseek(iFd, 0, SEEK_SET); + + return 0; +} + +int32_t CFile::open(const uint8_t *mem, int32_t size) { + this->close(); + + mem_file = mem; + filesize = size; + + return 0; +} + +void CFile::close() { + if (iFd >= 0) + ::close(iFd); + + iFd = -1; + mem_file = NULL; + filesize = 0; + pos = 0; +} + +int32_t CFile::read(uint8_t *ptr, size_t size) { + if (iFd >= 0) { + int32_t ret = ::read(iFd, ptr, size); + if (ret > 0) + pos += ret; + return ret; + } + + int32_t readsize = size; + + if (readsize > (int64_t) (filesize - pos)) + readsize = filesize - pos; + + if (readsize <= 0) + return readsize; + + if (mem_file != NULL) { + memcpy(ptr, mem_file + pos, readsize); + pos += readsize; + return readsize; + } + + return -1; +} + +int32_t CFile::write(const uint8_t *ptr, size_t size) { + if (iFd >= 0) { + size_t done = 0; + while (done < size) { + int32_t ret = ::write(iFd, ptr, size - done); + if (ret <= 0) + return ret; + + ptr += ret; + done += ret; + pos += ret; + } + return done; + } + + return -1; +} + +int32_t CFile::seek(long int offset, int32_t origin) { + int32_t ret = 0; + int64_t newPos = pos; + + if (origin == SEEK_SET) { + newPos = offset; + } else if (origin == SEEK_CUR) { + newPos += offset; + } else if (origin == SEEK_END) { + newPos = filesize + offset; + } + + if (newPos < 0) { + pos = 0; + } else { + pos = newPos; + } + + if (iFd >= 0) + ret = ::lseek(iFd, pos, SEEK_SET); + + if (mem_file != NULL) { + if (pos > filesize) { + pos = filesize; + } + } + + return ret; +} + +int32_t CFile::fwrite(const char *format, ...) { + char tmp[512]; + tmp[0] = 0; + int32_t result = -1; + + va_list va; + va_start(va, format); + if ((vsprintf(tmp, format, va) >= 0)) { + result = this->write((uint8_t *) tmp, strlen(tmp)); + } + va_end(va); + + + return result; +} + + diff --git a/source/fs/CFile.hpp b/source/fs/CFile.hpp new file mode 100644 index 0000000..8192fab --- /dev/null +++ b/source/fs/CFile.hpp @@ -0,0 +1,71 @@ +#ifndef CFILE_HPP_ +#define CFILE_HPP_ + +#include +#include +#include +#include +#include +#include + +class CFile { +public: + enum eOpenTypes { + ReadOnly, + WriteOnly, + ReadWrite, + Append + }; + + CFile(); + + CFile(const std::string &filepath, eOpenTypes mode); + + CFile(const uint8_t *memory, int32_t memsize); + + virtual ~CFile(); + + int32_t open(const std::string &filepath, eOpenTypes mode); + + int32_t open(const uint8_t *memory, int32_t memsize); + + BOOL isOpen() const { + if (iFd >= 0) + return true; + + if (mem_file) + return true; + + return false; + } + + void close(); + + int32_t read(uint8_t *ptr, size_t size); + + int32_t write(const uint8_t *ptr, size_t size); + + int32_t fwrite(const char *format, ...); + + int32_t seek(long int offset, int32_t origin); + + uint64_t tell() { + return pos; + }; + + uint64_t size() { + return filesize; + }; + + void rewind() { + this->seek(0, SEEK_SET); + }; + +protected: + int32_t iFd; + const uint8_t *mem_file; + uint64_t filesize; + uint64_t pos; +}; + +#endif diff --git a/source/fs/FSUtils.cpp b/source/fs/FSUtils.cpp new file mode 100644 index 0000000..8c8bac6 --- /dev/null +++ b/source/fs/FSUtils.cpp @@ -0,0 +1,174 @@ +#include +#include +#include +#include +#include +#include "FSUtils.h" +#include "CFile.hpp" +#include "utils/logger.h" + +int32_t FSUtils::LoadFileToMem(const char *filepath, uint8_t **inbuffer, uint32_t *size) { + //! always initialze input + *inbuffer = nullptr; + if (size) + *size = 0; + + int32_t iFd = open(filepath, O_RDONLY); + if (iFd < 0) + return -1; + + uint32_t filesize = lseek(iFd, 0, SEEK_END); + lseek(iFd, 0, SEEK_SET); + + auto *buffer = (uint8_t *) malloc(filesize); + if (buffer == nullptr) { + close(iFd); + return -2; + } + + uint32_t blocksize = 0x4000; + uint32_t done = 0; + int32_t readBytes = 0; + + while (done < filesize) { + if (done + blocksize > filesize) { + blocksize = filesize - done; + } + readBytes = read(iFd, buffer + done, blocksize); + if (readBytes <= 0) + break; + done += readBytes; + } + + close(iFd); + + if (done != filesize) { + free(buffer); + buffer = nullptr; + return -3; + } + + *inbuffer = buffer; + + //! sign is optional input + if (size) { + *size = filesize; + } + + return filesize; +} + +int32_t FSUtils::CheckFile(const char *filepath) { + if (!filepath) + return 0; + + struct stat filestat; + + char dirnoslash[strlen(filepath) + 2]; + snprintf(dirnoslash, sizeof(dirnoslash), "%s", filepath); + + while (dirnoslash[strlen(dirnoslash) - 1] == '/') + dirnoslash[strlen(dirnoslash) - 1] = '\0'; + + char *notRoot = strrchr(dirnoslash, '/'); + if (!notRoot) { + strcat(dirnoslash, "/"); + } + + if (stat(dirnoslash, &filestat) == 0) + return 1; + + return 0; +} + +int32_t FSUtils::CreateSubfolder(const char *fullpath) { + if (!fullpath) + return 0; + + int32_t result = 0; + + char dirnoslash[strlen(fullpath) + 1]; + strcpy(dirnoslash, fullpath); + + int32_t pos = strlen(dirnoslash) - 1; + while (dirnoslash[pos] == '/') { + dirnoslash[pos] = '\0'; + pos--; + } + + if (CheckFile(dirnoslash)) { + return 1; + } else { + char parentpath[strlen(dirnoslash) + 2]; + strcpy(parentpath, dirnoslash); + char *ptr = strrchr(parentpath, '/'); + + if (!ptr) { + //!Device root directory (must be with '/') + strcat(parentpath, "/"); + struct stat filestat; + if (stat(parentpath, &filestat) == 0) + return 1; + + return 0; + } + + ptr++; + ptr[0] = '\0'; + + result = CreateSubfolder(parentpath); + } + + if (!result) + return 0; + + if (mkdir(dirnoslash, 0777) == -1) { + return 0; + } + + return 1; +} + + +bool FSUtils::copyFile(const std::string &in, const std::string &out) { + // Using C++ buffers is **really** slow. Copying in 1023 byte chunks. + // Let's do it the old way. + size_t size; + + int source = open(in.c_str(), O_RDONLY, 0); + int dest = open(out.c_str(), 0x602, 0644); + if (source < 0) { + return false; + } + if (dest < 0) { + close(source); + return false; + } + + auto bufferSize = 1024 * 1024; + char *buf = (char *) malloc(bufferSize); + if (buf == NULL) { + return false; + } + + while ((size = read(source, buf, bufferSize)) > 0) { + write(dest, buf, size); + } + + free(buf); + + close(source); + close(dest); + return true; +} + +int32_t FSUtils::saveBufferToFile(const char *path, void *buffer, uint32_t size) { + CFile file(path, CFile::WriteOnly); + if (!file.isOpen()) { + return -1; + } + int32_t written = file.write((const uint8_t *) buffer, size); + file.close(); + return written; +} + diff --git a/source/fs/FSUtils.h b/source/fs/FSUtils.h new file mode 100644 index 0000000..99e2cd4 --- /dev/null +++ b/source/fs/FSUtils.h @@ -0,0 +1,17 @@ +#pragma once + +#include +#include + +class FSUtils { +public: + static int32_t LoadFileToMem(const char *filepath, uint8_t **inbuffer, uint32_t *size); + + static int32_t CreateSubfolder(const char *fullpath); + + static int32_t CheckFile(const char *filepath); + + static bool copyFile(const std::string &in, const std::string &out); + + static int32_t saveBufferToFile(const char *path, void *buffer, uint32_t size); +}; \ No newline at end of file diff --git a/source/main.cpp b/source/main.cpp new file mode 100644 index 0000000..a79fd7b --- /dev/null +++ b/source/main.cpp @@ -0,0 +1,96 @@ +#include + +#include +#include +#include + +#include +#include +#include +#include +#include "WiiUScreen.h" +#include "InstallerService.h" + +#include "ApplicationState.h" +#include "VPADInput.h" +#include "../build/safe_payload.h" + +constexpr bool strings_equal(char const *a, char const *b) { + return std::string_view(a) == b; +} + +static_assert(strings_equal(RPX_HASH, "116cff322148216f38b1eaaf296d1cffc6a3a98f"), "Built with an untested root.rpx! Remove this check if you really know what you're doing."); + +void initIOSUHax(); + +void deInitIOSUHax(); + +int sFSAFd = -1; +bool sIosuhaxMount = false; + +int main_loop() { + DEBUG_FUNCTION_LINE("Creating state"); + ApplicationState state; + VPadInput input; + + if (sFSAFd < 0 || !sIosuhaxMount) { + state.setError(ApplicationState::eErrorState::ERROR_IOSUHAX_FAILED); + } + + DEBUG_FUNCTION_LINE("Entering main loop"); + while (WHBProcIsRunning()) { + input.update(1280, 720); + state.update(&input); + state.render(); + } + + return 0; +} + +int main(int argc, char **argv) { + WHBLogUdpInit(); + DEBUG_FUNCTION_LINE("Hello from Aroma Installer!"); + WHBProcInit(); + WiiUScreen::Init(); + + initIOSUHax(); + + main_loop(); + + deInitIOSUHax(); + + WiiUScreen::DeInit(); + WHBProcShutdown(); + + return 0; +} + +void initIOSUHax() { + sIosuhaxMount = false; + int res = IOSUHAX_Open(nullptr); + if (res < 0) { + DEBUG_FUNCTION_LINE("IOSUHAX_open failed"); + OSFatal("IOSUHAX_open failed, please start this installer with an CFW"); + } else { + sIosuhaxMount = true; + sFSAFd = IOSUHAX_FSA_Open(); + if (sFSAFd < 0) { + DEBUG_FUNCTION_LINE("IOSUHAX_FSA_Open failed"); + } else { + mount_fs("storage_slc_installer", sFSAFd, nullptr, "/vol/system"); + mount_fs("storage_mlc_installer", sFSAFd, nullptr, "/vol/storage_mlc01"); + } + DEBUG_FUNCTION_LINE("IOSUHAX done"); + } +} + +void deInitIOSUHax() { + if (sIosuhaxMount) { + unmount_fs("storage_slc_installer"); + unmount_fs("storage_mlc_installer"); + if (sFSAFd >= 0) { + IOSUHAX_FSA_Close(sFSAFd); + } + IOSUHAX_Close(); + } +} \ No newline at end of file diff --git a/source/utils/TinySHA1.hpp b/source/utils/TinySHA1.hpp new file mode 100644 index 0000000..70af046 --- /dev/null +++ b/source/utils/TinySHA1.hpp @@ -0,0 +1,196 @@ +/* + * + * TinySHA1 - a header only implementation of the SHA1 algorithm in C++. Based + * on the implementation in boost::uuid::details. + * + * SHA1 Wikipedia Page: http://en.wikipedia.org/wiki/SHA-1 + * + * Copyright (c) 2012-22 SAURAV MOHAPATRA + * + * Permission to use, copy, modify, and distribute this software for any + * purpose with or without fee is hereby granted, provided that the above + * copyright notice and this permission notice appear in all copies. + * + * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES + * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF + * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR + * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES + * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN + * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF + * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE. + */ +#ifndef _TINY_SHA1_HPP_ +#define _TINY_SHA1_HPP_ +#include +#include +#include +#include +namespace sha1 +{ + class SHA1 + { + public: + typedef uint32_t digest32_t[5]; + typedef uint8_t digest8_t[20]; + inline static uint32_t LeftRotate(uint32_t value, size_t count) { + return (value << count) ^ (value >> (32-count)); + } + SHA1(){ reset(); } + virtual ~SHA1() {} + SHA1(const SHA1& s) { *this = s; } + const SHA1& operator = (const SHA1& s) { + memcpy(m_digest, s.m_digest, 5 * sizeof(uint32_t)); + memcpy(m_block, s.m_block, 64); + m_blockByteIndex = s.m_blockByteIndex; + m_byteCount = s.m_byteCount; + return *this; + } + SHA1& reset() { + m_digest[0] = 0x67452301; + m_digest[1] = 0xEFCDAB89; + m_digest[2] = 0x98BADCFE; + m_digest[3] = 0x10325476; + m_digest[4] = 0xC3D2E1F0; + m_blockByteIndex = 0; + m_byteCount = 0; + return *this; + } + SHA1& processByte(uint8_t octet) { + this->m_block[this->m_blockByteIndex++] = octet; + ++this->m_byteCount; + if(m_blockByteIndex == 64) { + this->m_blockByteIndex = 0; + processBlock(); + } + return *this; + } + SHA1& processBlock(const void* const start, const void* const end) { + const uint8_t* begin = static_cast(start); + const uint8_t* finish = static_cast(end); + while(begin != finish) { + processByte(*begin); + begin++; + } + return *this; + } + SHA1& processBytes(const void* const data, size_t len) { + const uint8_t* block = static_cast(data); + processBlock(block, block + len); + return *this; + } + const uint32_t* getDigest(digest32_t digest) { + size_t bitCount = this->m_byteCount * 8; + processByte(0x80); + if (this->m_blockByteIndex > 56) { + while (m_blockByteIndex != 0) { + processByte(0); + } + while (m_blockByteIndex < 56) { + processByte(0); + } + } else { + while (m_blockByteIndex < 56) { + processByte(0); + } + } + processByte(0); + processByte(0); + processByte(0); + processByte(0); + processByte( static_cast((bitCount>>24) & 0xFF)); + processByte( static_cast((bitCount>>16) & 0xFF)); + processByte( static_cast((bitCount>>8 ) & 0xFF)); + processByte( static_cast((bitCount) & 0xFF)); + + memcpy(digest, m_digest, 5 * sizeof(uint32_t)); + return digest; + } + const uint8_t* getDigestBytes(digest8_t digest) { + digest32_t d32; + getDigest(d32); + size_t di = 0; + digest[di++] = ((d32[0] >> 24) & 0xFF); + digest[di++] = ((d32[0] >> 16) & 0xFF); + digest[di++] = ((d32[0] >> 8) & 0xFF); + digest[di++] = ((d32[0]) & 0xFF); + + digest[di++] = ((d32[1] >> 24) & 0xFF); + digest[di++] = ((d32[1] >> 16) & 0xFF); + digest[di++] = ((d32[1] >> 8) & 0xFF); + digest[di++] = ((d32[1]) & 0xFF); + + digest[di++] = ((d32[2] >> 24) & 0xFF); + digest[di++] = ((d32[2] >> 16) & 0xFF); + digest[di++] = ((d32[2] >> 8) & 0xFF); + digest[di++] = ((d32[2]) & 0xFF); + + digest[di++] = ((d32[3] >> 24) & 0xFF); + digest[di++] = ((d32[3] >> 16) & 0xFF); + digest[di++] = ((d32[3] >> 8) & 0xFF); + digest[di++] = ((d32[3]) & 0xFF); + + digest[di++] = ((d32[4] >> 24) & 0xFF); + digest[di++] = ((d32[4] >> 16) & 0xFF); + digest[di++] = ((d32[4] >> 8) & 0xFF); + digest[di++] = ((d32[4]) & 0xFF); + return digest; + } + + protected: + void processBlock() { + uint32_t w[80]; + for (size_t i = 0; i < 16; i++) { + w[i] = (m_block[i*4 + 0] << 24); + w[i] |= (m_block[i*4 + 1] << 16); + w[i] |= (m_block[i*4 + 2] << 8); + w[i] |= (m_block[i*4 + 3]); + } + for (size_t i = 16; i < 80; i++) { + w[i] = LeftRotate((w[i-3] ^ w[i-8] ^ w[i-14] ^ w[i-16]), 1); + } + + uint32_t a = m_digest[0]; + uint32_t b = m_digest[1]; + uint32_t c = m_digest[2]; + uint32_t d = m_digest[3]; + uint32_t e = m_digest[4]; + + for (std::size_t i=0; i<80; ++i) { + uint32_t f = 0; + uint32_t k = 0; + + if (i<20) { + f = (b & c) | (~b & d); + k = 0x5A827999; + } else if (i<40) { + f = b ^ c ^ d; + k = 0x6ED9EBA1; + } else if (i<60) { + f = (b & c) | (b & d) | (c & d); + k = 0x8F1BBCDC; + } else { + f = b ^ c ^ d; + k = 0xCA62C1D6; + } + uint32_t temp = LeftRotate(a, 5) + f + e + k + w[i]; + e = d; + d = c; + c = LeftRotate(b, 30); + b = a; + a = temp; + } + + m_digest[0] += a; + m_digest[1] += b; + m_digest[2] += c; + m_digest[3] += d; + m_digest[4] += e; + } + private: + digest32_t m_digest; + uint8_t m_block[64]; + size_t m_blockByteIndex; + size_t m_byteCount; + }; +} +#endif diff --git a/source/utils/logger.h b/source/utils/logger.h new file mode 100644 index 0000000..7a56ddf --- /dev/null +++ b/source/utils/logger.h @@ -0,0 +1,23 @@ +#pragma once + +#ifdef __cplusplus +extern "C" { +#endif + +#include +#include + +#define __FILENAME_X__ (strrchr(__FILE__, '\\') ? strrchr(__FILE__, '\\') + 1 : __FILE__) +#define __FILENAME__ (strrchr(__FILE__, '/') ? strrchr(__FILE__, '/') + 1 : __FILENAME_X__) + +#define DEBUG_FUNCTION_LINE(FMT, ARGS...)do { \ + WHBLogPrintf("[%23s]%30s@L%04d: " FMT "",__FILENAME__,__FUNCTION__, __LINE__, ## ARGS); \ + } while (0); + +#define DEBUG_FUNCTION_LINE_WRITE(FMT, ARGS...)do { \ + WHBLogWritef("[%23s]%30s@L%04d: " FMT "",__FILENAME__,__FUNCTION__, __LINE__, ## ARGS); \ + } while (0); + +#ifdef __cplusplus +} +#endif \ No newline at end of file diff --git a/source/utils/pugiconfig.hpp b/source/utils/pugiconfig.hpp new file mode 100644 index 0000000..ac3ce7f --- /dev/null +++ b/source/utils/pugiconfig.hpp @@ -0,0 +1,74 @@ +/** + * pugixml parser - version 1.10 + * -------------------------------------------------------- + * Copyright (C) 2006-2019, by Arseny Kapoulkine (arseny.kapoulkine@gmail.com) + * Report bugs and download new versions at https://pugixml.org/ + * + * This library is distributed under the MIT License. See notice at the end + * of this file. + * + * This work is based on the pugxml parser, which is: + * Copyright (C) 2003, by Kristen Wegner (kristen@tima.net) + */ + +#ifndef HEADER_PUGICONFIG_HPP +#define HEADER_PUGICONFIG_HPP + +// Uncomment this to enable wchar_t mode +// #define PUGIXML_WCHAR_MODE + +// Uncomment this to enable compact mode +// #define PUGIXML_COMPACT + +// Uncomment this to disable XPath +// #define PUGIXML_NO_XPATH + +// Uncomment this to disable STL +// #define PUGIXML_NO_STL + +// Uncomment this to disable exceptions +#define PUGIXML_NO_EXCEPTIONS + +// Set this to control attributes for public classes/functions, i.e.: +// #define PUGIXML_API __declspec(dllexport) // to export all public symbols from DLL +// #define PUGIXML_CLASS __declspec(dllimport) // to import all classes from DLL +// #define PUGIXML_FUNCTION __fastcall // to set calling conventions to all public functions to fastcall +// In absence of PUGIXML_CLASS/PUGIXML_FUNCTION definitions PUGIXML_API is used instead + +// Tune these constants to adjust memory-related behavior +// #define PUGIXML_MEMORY_PAGE_SIZE 32768 +// #define PUGIXML_MEMORY_OUTPUT_STACK 10240 +// #define PUGIXML_MEMORY_XPATH_PAGE_SIZE 4096 + +// Uncomment this to switch to header-only version +// #define PUGIXML_HEADER_ONLY + +// Uncomment this to enable long long support +// #define PUGIXML_HAS_LONG_LONG + +#endif + +/** + * Copyright (c) 2006-2019 Arseny Kapoulkine + * + * Permission is hereby granted, free of charge, to any person + * obtaining a copy of this software and associated documentation + * files (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use, + * copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following + * conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES + * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ diff --git a/source/utils/pugixml.cpp b/source/utils/pugixml.cpp new file mode 100644 index 0000000..f92fdef --- /dev/null +++ b/source/utils/pugixml.cpp @@ -0,0 +1,12865 @@ +/** + * pugixml parser - version 1.10 + * -------------------------------------------------------- + * Copyright (C) 2006-2019, by Arseny Kapoulkine (arseny.kapoulkine@gmail.com) + * Report bugs and download new versions at https://pugixml.org/ + * + * This library is distributed under the MIT License. See notice at the end + * of this file. + * + * This work is based on the pugxml parser, which is: + * Copyright (C) 2003, by Kristen Wegner (kristen@tima.net) + */ + +#ifndef SOURCE_PUGIXML_CPP +#define SOURCE_PUGIXML_CPP + +#include "pugixml.hpp" + +#include +#include +#include +#include +#include + +#ifdef PUGIXML_WCHAR_MODE +# include +#endif + +#ifndef PUGIXML_NO_XPATH +# include +# include +#endif + +#ifndef PUGIXML_NO_STL +# include +# include +# include +#endif + +// For placement new +#include + +#ifdef _MSC_VER +# pragma warning(push) +# pragma warning(disable: 4127) // conditional expression is constant +# pragma warning(disable: 4324) // structure was padded due to __declspec(align()) +# pragma warning(disable: 4702) // unreachable code +# pragma warning(disable: 4996) // this function or variable may be unsafe +#endif + +#if defined(_MSC_VER) && defined(__c2__) +# pragma clang diagnostic push +# pragma clang diagnostic ignored "-Wdeprecated" // this function or variable may be unsafe +#endif + +#ifdef __INTEL_COMPILER +# pragma warning(disable: 177) // function was declared but never referenced +# pragma warning(disable: 279) // controlling expression is constant +# pragma warning(disable: 1478 1786) // function was declared "deprecated" +# pragma warning(disable: 1684) // conversion from pointer to same-sized integral type +#endif + +#if defined(__BORLANDC__) && defined(PUGIXML_HEADER_ONLY) +# pragma warn -8080 // symbol is declared but never used; disabling this inside push/pop bracket does not make the warning go away +#endif + +#ifdef __BORLANDC__ +# pragma option push +# pragma warn -8008 // condition is always false +# pragma warn -8066 // unreachable code +#endif + +#ifdef __SNC__ +// Using diag_push/diag_pop does not disable the warnings inside templates due to a compiler bug +# pragma diag_suppress=178 // function was declared but never referenced +# pragma diag_suppress=237 // controlling expression is constant +#endif + +#ifdef __TI_COMPILER_VERSION__ +# pragma diag_suppress 179 // function was declared but never referenced +#endif + +// Inlining controls +#if defined(_MSC_VER) && _MSC_VER >= 1300 +# define PUGI__NO_INLINE __declspec(noinline) +#elif defined(__GNUC__) +# define PUGI__NO_INLINE __attribute__((noinline)) +#else +# define PUGI__NO_INLINE +#endif + +// Branch weight controls +#if defined(__GNUC__) && !defined(__c2__) +# define PUGI__UNLIKELY(cond) __builtin_expect(cond, 0) +#else +# define PUGI__UNLIKELY(cond) (cond) +#endif + +// Simple static assertion +#define PUGI__STATIC_ASSERT(cond) { static const char condition_failed[(cond) ? 1 : -1] = {0}; (void)condition_failed[0]; } + +// Digital Mars C++ bug workaround for passing char loaded from memory via stack +#ifdef __DMC__ +# define PUGI__DMC_VOLATILE volatile +#else +# define PUGI__DMC_VOLATILE +#endif + +// Integer sanitizer workaround; we only apply this for clang since gcc8 has no_sanitize but not unsigned-integer-overflow and produces "attribute directive ignored" warnings +#if defined(__clang__) && defined(__has_attribute) +# if __has_attribute(no_sanitize) +# define PUGI__UNSIGNED_OVERFLOW __attribute__((no_sanitize("unsigned-integer-overflow"))) +# else +# define PUGI__UNSIGNED_OVERFLOW +# endif +#else +# define PUGI__UNSIGNED_OVERFLOW +#endif + +// Borland C++ bug workaround for not defining ::memcpy depending on header include order (can't always use std::memcpy because some compilers don't have it at all) +#if defined(__BORLANDC__) && !defined(__MEM_H_USING_LIST) +using std::memcpy; +using std::memmove; +using std::memset; +#endif + +// Some MinGW/GCC versions have headers that erroneously omit LLONG_MIN/LLONG_MAX/ULLONG_MAX definitions from limits.h in some configurations +#if defined(PUGIXML_HAS_LONG_LONG) && defined(__GNUC__) && !defined(LLONG_MAX) && !defined(LLONG_MIN) && !defined(ULLONG_MAX) +# define LLONG_MIN (-LLONG_MAX - 1LL) +# define LLONG_MAX __LONG_LONG_MAX__ +# define ULLONG_MAX (LLONG_MAX * 2ULL + 1ULL) +#endif + +// In some environments MSVC is a compiler but the CRT lacks certain MSVC-specific features +#if defined(_MSC_VER) && !defined(__S3E__) +# define PUGI__MSVC_CRT_VERSION _MSC_VER +#endif + +// Not all platforms have snprintf; we define a wrapper that uses snprintf if possible. This only works with buffers with a known size. +#if __cplusplus >= 201103 +# define PUGI__SNPRINTF(buf, ...) snprintf(buf, sizeof(buf), __VA_ARGS__) +#elif defined(PUGI__MSVC_CRT_VERSION) && PUGI__MSVC_CRT_VERSION >= 1400 +# define PUGI__SNPRINTF(buf, ...) _snprintf_s(buf, _countof(buf), _TRUNCATE, __VA_ARGS__) +#else +# define PUGI__SNPRINTF sprintf +#endif + +// We put implementation details into an anonymous namespace in source mode, but have to keep it in non-anonymous namespace in header-only mode to prevent binary bloat. +#ifdef PUGIXML_HEADER_ONLY +# define PUGI__NS_BEGIN namespace pugi { namespace impl { +# define PUGI__NS_END } } +# define PUGI__FN inline +# define PUGI__FN_NO_INLINE inline +#else +# if defined(_MSC_VER) && _MSC_VER < 1300 // MSVC6 seems to have an amusing bug with anonymous namespaces inside namespaces +# define PUGI__NS_BEGIN namespace pugi { namespace impl { +# define PUGI__NS_END } } +# else +# define PUGI__NS_BEGIN namespace pugi { namespace impl { namespace { +# define PUGI__NS_END } } } +# endif +# define PUGI__FN +# define PUGI__FN_NO_INLINE PUGI__NO_INLINE +#endif + +// uintptr_t +#if (defined(_MSC_VER) && _MSC_VER < 1600) || (defined(__BORLANDC__) && __BORLANDC__ < 0x561) +namespace pugi +{ +# ifndef _UINTPTR_T_DEFINED + typedef size_t uintptr_t; +# endif + + typedef unsigned __int8 uint8_t; + typedef unsigned __int16 uint16_t; + typedef unsigned __int32 uint32_t; +} +#else +# include +#endif + +// Memory allocation +PUGI__NS_BEGIN + PUGI__FN void* default_allocate(size_t size) + { + return malloc(size); + } + + PUGI__FN void default_deallocate(void* ptr) + { + free(ptr); + } + + template + struct xml_memory_management_function_storage + { + static allocation_function allocate; + static deallocation_function deallocate; + }; + + // Global allocation functions are stored in class statics so that in header mode linker deduplicates them + // Without a template<> we'll get multiple definitions of the same static + template allocation_function xml_memory_management_function_storage::allocate = default_allocate; + template deallocation_function xml_memory_management_function_storage::deallocate = default_deallocate; + + typedef xml_memory_management_function_storage xml_memory; +PUGI__NS_END + +// String utilities +PUGI__NS_BEGIN + // Get string length + PUGI__FN size_t strlength(const char_t* s) + { + assert(s); + + #ifdef PUGIXML_WCHAR_MODE + return wcslen(s); + #else + return strlen(s); + #endif + } + + // Compare two strings + PUGI__FN bool strequal(const char_t* src, const char_t* dst) + { + assert(src && dst); + + #ifdef PUGIXML_WCHAR_MODE + return wcscmp(src, dst) == 0; + #else + return strcmp(src, dst) == 0; + #endif + } + + // Compare lhs with [rhs_begin, rhs_end) + PUGI__FN bool strequalrange(const char_t* lhs, const char_t* rhs, size_t count) + { + for (size_t i = 0; i < count; ++i) + if (lhs[i] != rhs[i]) + return false; + + return lhs[count] == 0; + } + + // Get length of wide string, even if CRT lacks wide character support + PUGI__FN size_t strlength_wide(const wchar_t* s) + { + assert(s); + + #ifdef PUGIXML_WCHAR_MODE + return wcslen(s); + #else + const wchar_t* end = s; + while (*end) end++; + return static_cast(end - s); + #endif + } +PUGI__NS_END + +// auto_ptr-like object for exception recovery +PUGI__NS_BEGIN + template struct auto_deleter + { + typedef void (*D)(T*); + + T* data; + D deleter; + + auto_deleter(T* data_, D deleter_): data(data_), deleter(deleter_) + { + } + + ~auto_deleter() + { + if (data) deleter(data); + } + + T* release() + { + T* result = data; + data = 0; + return result; + } + }; +PUGI__NS_END + +#ifdef PUGIXML_COMPACT +PUGI__NS_BEGIN + class compact_hash_table + { + public: + compact_hash_table(): _items(0), _capacity(0), _count(0) + { + } + + void clear() + { + if (_items) + { + xml_memory::deallocate(_items); + _items = 0; + _capacity = 0; + _count = 0; + } + } + + void* find(const void* key) + { + if (_capacity == 0) return 0; + + item_t* item = get_item(key); + assert(item); + assert(item->key == key || (item->key == 0 && item->value == 0)); + + return item->value; + } + + void insert(const void* key, void* value) + { + assert(_capacity != 0 && _count < _capacity - _capacity / 4); + + item_t* item = get_item(key); + assert(item); + + if (item->key == 0) + { + _count++; + item->key = key; + } + + item->value = value; + } + + bool reserve(size_t extra = 16) + { + if (_count + extra >= _capacity - _capacity / 4) + return rehash(_count + extra); + + return true; + } + + private: + struct item_t + { + const void* key; + void* value; + }; + + item_t* _items; + size_t _capacity; + + size_t _count; + + bool rehash(size_t count); + + item_t* get_item(const void* key) + { + assert(key); + assert(_capacity > 0); + + size_t hashmod = _capacity - 1; + size_t bucket = hash(key) & hashmod; + + for (size_t probe = 0; probe <= hashmod; ++probe) + { + item_t& probe_item = _items[bucket]; + + if (probe_item.key == key || probe_item.key == 0) + return &probe_item; + + // hash collision, quadratic probing + bucket = (bucket + probe + 1) & hashmod; + } + + assert(false && "Hash table is full"); // unreachable + return 0; + } + + static PUGI__UNSIGNED_OVERFLOW unsigned int hash(const void* key) + { + unsigned int h = static_cast(reinterpret_cast(key)); + + // MurmurHash3 32-bit finalizer + h ^= h >> 16; + h *= 0x85ebca6bu; + h ^= h >> 13; + h *= 0xc2b2ae35u; + h ^= h >> 16; + + return h; + } + }; + + PUGI__FN_NO_INLINE bool compact_hash_table::rehash(size_t count) + { + size_t capacity = 32; + while (count >= capacity - capacity / 4) + capacity *= 2; + + compact_hash_table rt; + rt._capacity = capacity; + rt._items = static_cast(xml_memory::allocate(sizeof(item_t) * capacity)); + + if (!rt._items) + return false; + + memset(rt._items, 0, sizeof(item_t) * capacity); + + for (size_t i = 0; i < _capacity; ++i) + if (_items[i].key) + rt.insert(_items[i].key, _items[i].value); + + if (_items) + xml_memory::deallocate(_items); + + _capacity = capacity; + _items = rt._items; + + assert(_count == rt._count); + + return true; + } + +PUGI__NS_END +#endif + +PUGI__NS_BEGIN +#ifdef PUGIXML_COMPACT + static const uintptr_t xml_memory_block_alignment = 4; +#else + static const uintptr_t xml_memory_block_alignment = sizeof(void*); +#endif + + // extra metadata bits + static const uintptr_t xml_memory_page_contents_shared_mask = 64; + static const uintptr_t xml_memory_page_name_allocated_mask = 32; + static const uintptr_t xml_memory_page_value_allocated_mask = 16; + static const uintptr_t xml_memory_page_type_mask = 15; + + // combined masks for string uniqueness + static const uintptr_t xml_memory_page_name_allocated_or_shared_mask = xml_memory_page_name_allocated_mask | xml_memory_page_contents_shared_mask; + static const uintptr_t xml_memory_page_value_allocated_or_shared_mask = xml_memory_page_value_allocated_mask | xml_memory_page_contents_shared_mask; + +#ifdef PUGIXML_COMPACT + #define PUGI__GETHEADER_IMPL(object, page, flags) // unused + #define PUGI__GETPAGE_IMPL(header) (header).get_page() +#else + #define PUGI__GETHEADER_IMPL(object, page, flags) (((reinterpret_cast(object) - reinterpret_cast(page)) << 8) | (flags)) + // this macro casts pointers through void* to avoid 'cast increases required alignment of target type' warnings + #define PUGI__GETPAGE_IMPL(header) static_cast(const_cast(static_cast(reinterpret_cast(&header) - (header >> 8)))) +#endif + + #define PUGI__GETPAGE(n) PUGI__GETPAGE_IMPL((n)->header) + #define PUGI__NODETYPE(n) static_cast((n)->header & impl::xml_memory_page_type_mask) + + struct xml_allocator; + + struct xml_memory_page + { + static xml_memory_page* construct(void* memory) + { + xml_memory_page* result = static_cast(memory); + + result->allocator = 0; + result->prev = 0; + result->next = 0; + result->busy_size = 0; + result->freed_size = 0; + + #ifdef PUGIXML_COMPACT + result->compact_string_base = 0; + result->compact_shared_parent = 0; + result->compact_page_marker = 0; + #endif + + return result; + } + + xml_allocator* allocator; + + xml_memory_page* prev; + xml_memory_page* next; + + size_t busy_size; + size_t freed_size; + + #ifdef PUGIXML_COMPACT + char_t* compact_string_base; + void* compact_shared_parent; + uint32_t* compact_page_marker; + #endif + }; + + static const size_t xml_memory_page_size = + #ifdef PUGIXML_MEMORY_PAGE_SIZE + (PUGIXML_MEMORY_PAGE_SIZE) + #else + 32768 + #endif + - sizeof(xml_memory_page); + + struct xml_memory_string_header + { + uint16_t page_offset; // offset from page->data + uint16_t full_size; // 0 if string occupies whole page + }; + + struct xml_allocator + { + xml_allocator(xml_memory_page* root): _root(root), _busy_size(root->busy_size) + { + #ifdef PUGIXML_COMPACT + _hash = 0; + #endif + } + + xml_memory_page* allocate_page(size_t data_size) + { + size_t size = sizeof(xml_memory_page) + data_size; + + // allocate block with some alignment, leaving memory for worst-case padding + void* memory = xml_memory::allocate(size); + if (!memory) return 0; + + // prepare page structure + xml_memory_page* page = xml_memory_page::construct(memory); + assert(page); + + page->allocator = _root->allocator; + + return page; + } + + static void deallocate_page(xml_memory_page* page) + { + xml_memory::deallocate(page); + } + + void* allocate_memory_oob(size_t size, xml_memory_page*& out_page); + + void* allocate_memory(size_t size, xml_memory_page*& out_page) + { + if (PUGI__UNLIKELY(_busy_size + size > xml_memory_page_size)) + return allocate_memory_oob(size, out_page); + + void* buf = reinterpret_cast(_root) + sizeof(xml_memory_page) + _busy_size; + + _busy_size += size; + + out_page = _root; + + return buf; + } + + #ifdef PUGIXML_COMPACT + void* allocate_object(size_t size, xml_memory_page*& out_page) + { + void* result = allocate_memory(size + sizeof(uint32_t), out_page); + if (!result) return 0; + + // adjust for marker + ptrdiff_t offset = static_cast(result) - reinterpret_cast(out_page->compact_page_marker); + + if (PUGI__UNLIKELY(static_cast(offset) >= 256 * xml_memory_block_alignment)) + { + // insert new marker + uint32_t* marker = static_cast(result); + + *marker = static_cast(reinterpret_cast(marker) - reinterpret_cast(out_page)); + out_page->compact_page_marker = marker; + + // since we don't reuse the page space until we reallocate it, we can just pretend that we freed the marker block + // this will make sure deallocate_memory correctly tracks the size + out_page->freed_size += sizeof(uint32_t); + + return marker + 1; + } + else + { + // roll back uint32_t part + _busy_size -= sizeof(uint32_t); + + return result; + } + } + #else + void* allocate_object(size_t size, xml_memory_page*& out_page) + { + return allocate_memory(size, out_page); + } + #endif + + void deallocate_memory(void* ptr, size_t size, xml_memory_page* page) + { + if (page == _root) page->busy_size = _busy_size; + + assert(ptr >= reinterpret_cast(page) + sizeof(xml_memory_page) && ptr < reinterpret_cast(page) + sizeof(xml_memory_page) + page->busy_size); + (void)!ptr; + + page->freed_size += size; + assert(page->freed_size <= page->busy_size); + + if (page->freed_size == page->busy_size) + { + if (page->next == 0) + { + assert(_root == page); + + // top page freed, just reset sizes + page->busy_size = 0; + page->freed_size = 0; + + #ifdef PUGIXML_COMPACT + // reset compact state to maximize efficiency + page->compact_string_base = 0; + page->compact_shared_parent = 0; + page->compact_page_marker = 0; + #endif + + _busy_size = 0; + } + else + { + assert(_root != page); + assert(page->prev); + + // remove from the list + page->prev->next = page->next; + page->next->prev = page->prev; + + // deallocate + deallocate_page(page); + } + } + } + + char_t* allocate_string(size_t length) + { + static const size_t max_encoded_offset = (1 << 16) * xml_memory_block_alignment; + + PUGI__STATIC_ASSERT(xml_memory_page_size <= max_encoded_offset); + + // allocate memory for string and header block + size_t size = sizeof(xml_memory_string_header) + length * sizeof(char_t); + + // round size up to block alignment boundary + size_t full_size = (size + (xml_memory_block_alignment - 1)) & ~(xml_memory_block_alignment - 1); + + xml_memory_page* page; + xml_memory_string_header* header = static_cast(allocate_memory(full_size, page)); + + if (!header) return 0; + + // setup header + ptrdiff_t page_offset = reinterpret_cast(header) - reinterpret_cast(page) - sizeof(xml_memory_page); + + assert(page_offset % xml_memory_block_alignment == 0); + assert(page_offset >= 0 && static_cast(page_offset) < max_encoded_offset); + header->page_offset = static_cast(static_cast(page_offset) / xml_memory_block_alignment); + + // full_size == 0 for large strings that occupy the whole page + assert(full_size % xml_memory_block_alignment == 0); + assert(full_size < max_encoded_offset || (page->busy_size == full_size && page_offset == 0)); + header->full_size = static_cast(full_size < max_encoded_offset ? full_size / xml_memory_block_alignment : 0); + + // round-trip through void* to avoid 'cast increases required alignment of target type' warning + // header is guaranteed a pointer-sized alignment, which should be enough for char_t + return static_cast(static_cast(header + 1)); + } + + void deallocate_string(char_t* string) + { + // this function casts pointers through void* to avoid 'cast increases required alignment of target type' warnings + // we're guaranteed the proper (pointer-sized) alignment on the input string if it was allocated via allocate_string + + // get header + xml_memory_string_header* header = static_cast(static_cast(string)) - 1; + assert(header); + + // deallocate + size_t page_offset = sizeof(xml_memory_page) + header->page_offset * xml_memory_block_alignment; + xml_memory_page* page = reinterpret_cast(static_cast(reinterpret_cast(header) - page_offset)); + + // if full_size == 0 then this string occupies the whole page + size_t full_size = header->full_size == 0 ? page->busy_size : header->full_size * xml_memory_block_alignment; + + deallocate_memory(header, full_size, page); + } + + bool reserve() + { + #ifdef PUGIXML_COMPACT + return _hash->reserve(); + #else + return true; + #endif + } + + xml_memory_page* _root; + size_t _busy_size; + + #ifdef PUGIXML_COMPACT + compact_hash_table* _hash; + #endif + }; + + PUGI__FN_NO_INLINE void* xml_allocator::allocate_memory_oob(size_t size, xml_memory_page*& out_page) + { + const size_t large_allocation_threshold = xml_memory_page_size / 4; + + xml_memory_page* page = allocate_page(size <= large_allocation_threshold ? xml_memory_page_size : size); + out_page = page; + + if (!page) return 0; + + if (size <= large_allocation_threshold) + { + _root->busy_size = _busy_size; + + // insert page at the end of linked list + page->prev = _root; + _root->next = page; + _root = page; + + _busy_size = size; + } + else + { + // insert page before the end of linked list, so that it is deleted as soon as possible + // the last page is not deleted even if it's empty (see deallocate_memory) + assert(_root->prev); + + page->prev = _root->prev; + page->next = _root; + + _root->prev->next = page; + _root->prev = page; + + page->busy_size = size; + } + + return reinterpret_cast(page) + sizeof(xml_memory_page); + } +PUGI__NS_END + +#ifdef PUGIXML_COMPACT +PUGI__NS_BEGIN + static const uintptr_t compact_alignment_log2 = 2; + static const uintptr_t compact_alignment = 1 << compact_alignment_log2; + + class compact_header + { + public: + compact_header(xml_memory_page* page, unsigned int flags) + { + PUGI__STATIC_ASSERT(xml_memory_block_alignment == compact_alignment); + + ptrdiff_t offset = (reinterpret_cast(this) - reinterpret_cast(page->compact_page_marker)); + assert(offset % compact_alignment == 0 && static_cast(offset) < 256 * compact_alignment); + + _page = static_cast(offset >> compact_alignment_log2); + _flags = static_cast(flags); + } + + void operator&=(uintptr_t mod) + { + _flags &= static_cast(mod); + } + + void operator|=(uintptr_t mod) + { + _flags |= static_cast(mod); + } + + uintptr_t operator&(uintptr_t mod) const + { + return _flags & mod; + } + + xml_memory_page* get_page() const + { + // round-trip through void* to silence 'cast increases required alignment of target type' warnings + const char* page_marker = reinterpret_cast(this) - (_page << compact_alignment_log2); + const char* page = page_marker - *reinterpret_cast(static_cast(page_marker)); + + return const_cast(reinterpret_cast(static_cast(page))); + } + + private: + unsigned char _page; + unsigned char _flags; + }; + + PUGI__FN xml_memory_page* compact_get_page(const void* object, int header_offset) + { + const compact_header* header = reinterpret_cast(static_cast(object) - header_offset); + + return header->get_page(); + } + + template PUGI__FN_NO_INLINE T* compact_get_value(const void* object) + { + return static_cast(compact_get_page(object, header_offset)->allocator->_hash->find(object)); + } + + template PUGI__FN_NO_INLINE void compact_set_value(const void* object, T* value) + { + compact_get_page(object, header_offset)->allocator->_hash->insert(object, value); + } + + template class compact_pointer + { + public: + compact_pointer(): _data(0) + { + } + + void operator=(const compact_pointer& rhs) + { + *this = rhs + 0; + } + + void operator=(T* value) + { + if (value) + { + // value is guaranteed to be compact-aligned; 'this' is not + // our decoding is based on 'this' aligned to compact alignment downwards (see operator T*) + // so for negative offsets (e.g. -3) we need to adjust the diff by compact_alignment - 1 to + // compensate for arithmetic shift rounding for negative values + ptrdiff_t diff = reinterpret_cast(value) - reinterpret_cast(this); + ptrdiff_t offset = ((diff + int(compact_alignment - 1)) >> compact_alignment_log2) - start; + + if (static_cast(offset) <= 253) + _data = static_cast(offset + 1); + else + { + compact_set_value(this, value); + + _data = 255; + } + } + else + _data = 0; + } + + operator T*() const + { + if (_data) + { + if (_data < 255) + { + uintptr_t base = reinterpret_cast(this) & ~(compact_alignment - 1); + + return reinterpret_cast(base + (_data - 1 + start) * compact_alignment); + } + else + return compact_get_value(this); + } + else + return 0; + } + + T* operator->() const + { + return *this; + } + + private: + unsigned char _data; + }; + + template class compact_pointer_parent + { + public: + compact_pointer_parent(): _data(0) + { + } + + void operator=(const compact_pointer_parent& rhs) + { + *this = rhs + 0; + } + + void operator=(T* value) + { + if (value) + { + // value is guaranteed to be compact-aligned; 'this' is not + // our decoding is based on 'this' aligned to compact alignment downwards (see operator T*) + // so for negative offsets (e.g. -3) we need to adjust the diff by compact_alignment - 1 to + // compensate for arithmetic shift behavior for negative values + ptrdiff_t diff = reinterpret_cast(value) - reinterpret_cast(this); + ptrdiff_t offset = ((diff + int(compact_alignment - 1)) >> compact_alignment_log2) + 65533; + + if (static_cast(offset) <= 65533) + { + _data = static_cast(offset + 1); + } + else + { + xml_memory_page* page = compact_get_page(this, header_offset); + + if (PUGI__UNLIKELY(page->compact_shared_parent == 0)) + page->compact_shared_parent = value; + + if (page->compact_shared_parent == value) + { + _data = 65534; + } + else + { + compact_set_value(this, value); + + _data = 65535; + } + } + } + else + { + _data = 0; + } + } + + operator T*() const + { + if (_data) + { + if (_data < 65534) + { + uintptr_t base = reinterpret_cast(this) & ~(compact_alignment - 1); + + return reinterpret_cast(base + (_data - 1 - 65533) * compact_alignment); + } + else if (_data == 65534) + return static_cast(compact_get_page(this, header_offset)->compact_shared_parent); + else + return compact_get_value(this); + } + else + return 0; + } + + T* operator->() const + { + return *this; + } + + private: + uint16_t _data; + }; + + template class compact_string + { + public: + compact_string(): _data(0) + { + } + + void operator=(const compact_string& rhs) + { + *this = rhs + 0; + } + + void operator=(char_t* value) + { + if (value) + { + xml_memory_page* page = compact_get_page(this, header_offset); + + if (PUGI__UNLIKELY(page->compact_string_base == 0)) + page->compact_string_base = value; + + ptrdiff_t offset = value - page->compact_string_base; + + if (static_cast(offset) < (65535 << 7)) + { + // round-trip through void* to silence 'cast increases required alignment of target type' warnings + uint16_t* base = reinterpret_cast(static_cast(reinterpret_cast(this) - base_offset)); + + if (*base == 0) + { + *base = static_cast((offset >> 7) + 1); + _data = static_cast((offset & 127) + 1); + } + else + { + ptrdiff_t remainder = offset - ((*base - 1) << 7); + + if (static_cast(remainder) <= 253) + { + _data = static_cast(remainder + 1); + } + else + { + compact_set_value(this, value); + + _data = 255; + } + } + } + else + { + compact_set_value(this, value); + + _data = 255; + } + } + else + { + _data = 0; + } + } + + operator char_t*() const + { + if (_data) + { + if (_data < 255) + { + xml_memory_page* page = compact_get_page(this, header_offset); + + // round-trip through void* to silence 'cast increases required alignment of target type' warnings + const uint16_t* base = reinterpret_cast(static_cast(reinterpret_cast(this) - base_offset)); + assert(*base); + + ptrdiff_t offset = ((*base - 1) << 7) + (_data - 1); + + return page->compact_string_base + offset; + } + else + { + return compact_get_value(this); + } + } + else + return 0; + } + + private: + unsigned char _data; + }; +PUGI__NS_END +#endif + +#ifdef PUGIXML_COMPACT +namespace pugi +{ + struct xml_attribute_struct + { + xml_attribute_struct(impl::xml_memory_page* page): header(page, 0), namevalue_base(0) + { + PUGI__STATIC_ASSERT(sizeof(xml_attribute_struct) == 8); + } + + impl::compact_header header; + + uint16_t namevalue_base; + + impl::compact_string<4, 2> name; + impl::compact_string<5, 3> value; + + impl::compact_pointer prev_attribute_c; + impl::compact_pointer next_attribute; + }; + + struct xml_node_struct + { + xml_node_struct(impl::xml_memory_page* page, xml_node_type type): header(page, type), namevalue_base(0) + { + PUGI__STATIC_ASSERT(sizeof(xml_node_struct) == 12); + } + + impl::compact_header header; + + uint16_t namevalue_base; + + impl::compact_string<4, 2> name; + impl::compact_string<5, 3> value; + + impl::compact_pointer_parent parent; + + impl::compact_pointer first_child; + + impl::compact_pointer prev_sibling_c; + impl::compact_pointer next_sibling; + + impl::compact_pointer first_attribute; + }; +} +#else +namespace pugi +{ + struct xml_attribute_struct + { + xml_attribute_struct(impl::xml_memory_page* page): name(0), value(0), prev_attribute_c(0), next_attribute(0) + { + header = PUGI__GETHEADER_IMPL(this, page, 0); + } + + uintptr_t header; + + char_t* name; + char_t* value; + + xml_attribute_struct* prev_attribute_c; + xml_attribute_struct* next_attribute; + }; + + struct xml_node_struct + { + xml_node_struct(impl::xml_memory_page* page, xml_node_type type): name(0), value(0), parent(0), first_child(0), prev_sibling_c(0), next_sibling(0), first_attribute(0) + { + header = PUGI__GETHEADER_IMPL(this, page, type); + } + + uintptr_t header; + + char_t* name; + char_t* value; + + xml_node_struct* parent; + + xml_node_struct* first_child; + + xml_node_struct* prev_sibling_c; + xml_node_struct* next_sibling; + + xml_attribute_struct* first_attribute; + }; +} +#endif + +PUGI__NS_BEGIN + struct xml_extra_buffer + { + char_t* buffer; + xml_extra_buffer* next; + }; + + struct xml_document_struct: public xml_node_struct, public xml_allocator + { + xml_document_struct(xml_memory_page* page): xml_node_struct(page, node_document), xml_allocator(page), buffer(0), extra_buffers(0) + { + } + + const char_t* buffer; + + xml_extra_buffer* extra_buffers; + + #ifdef PUGIXML_COMPACT + compact_hash_table hash; + #endif + }; + + template inline xml_allocator& get_allocator(const Object* object) + { + assert(object); + + return *PUGI__GETPAGE(object)->allocator; + } + + template inline xml_document_struct& get_document(const Object* object) + { + assert(object); + + return *static_cast(PUGI__GETPAGE(object)->allocator); + } +PUGI__NS_END + +// Low-level DOM operations +PUGI__NS_BEGIN + inline xml_attribute_struct* allocate_attribute(xml_allocator& alloc) + { + xml_memory_page* page; + void* memory = alloc.allocate_object(sizeof(xml_attribute_struct), page); + if (!memory) return 0; + + return new (memory) xml_attribute_struct(page); + } + + inline xml_node_struct* allocate_node(xml_allocator& alloc, xml_node_type type) + { + xml_memory_page* page; + void* memory = alloc.allocate_object(sizeof(xml_node_struct), page); + if (!memory) return 0; + + return new (memory) xml_node_struct(page, type); + } + + inline void destroy_attribute(xml_attribute_struct* a, xml_allocator& alloc) + { + if (a->header & impl::xml_memory_page_name_allocated_mask) + alloc.deallocate_string(a->name); + + if (a->header & impl::xml_memory_page_value_allocated_mask) + alloc.deallocate_string(a->value); + + alloc.deallocate_memory(a, sizeof(xml_attribute_struct), PUGI__GETPAGE(a)); + } + + inline void destroy_node(xml_node_struct* n, xml_allocator& alloc) + { + if (n->header & impl::xml_memory_page_name_allocated_mask) + alloc.deallocate_string(n->name); + + if (n->header & impl::xml_memory_page_value_allocated_mask) + alloc.deallocate_string(n->value); + + for (xml_attribute_struct* attr = n->first_attribute; attr; ) + { + xml_attribute_struct* next = attr->next_attribute; + + destroy_attribute(attr, alloc); + + attr = next; + } + + for (xml_node_struct* child = n->first_child; child; ) + { + xml_node_struct* next = child->next_sibling; + + destroy_node(child, alloc); + + child = next; + } + + alloc.deallocate_memory(n, sizeof(xml_node_struct), PUGI__GETPAGE(n)); + } + + inline void append_node(xml_node_struct* child, xml_node_struct* node) + { + child->parent = node; + + xml_node_struct* head = node->first_child; + + if (head) + { + xml_node_struct* tail = head->prev_sibling_c; + + tail->next_sibling = child; + child->prev_sibling_c = tail; + head->prev_sibling_c = child; + } + else + { + node->first_child = child; + child->prev_sibling_c = child; + } + } + + inline void prepend_node(xml_node_struct* child, xml_node_struct* node) + { + child->parent = node; + + xml_node_struct* head = node->first_child; + + if (head) + { + child->prev_sibling_c = head->prev_sibling_c; + head->prev_sibling_c = child; + } + else + child->prev_sibling_c = child; + + child->next_sibling = head; + node->first_child = child; + } + + inline void insert_node_after(xml_node_struct* child, xml_node_struct* node) + { + xml_node_struct* parent = node->parent; + + child->parent = parent; + + if (node->next_sibling) + node->next_sibling->prev_sibling_c = child; + else + parent->first_child->prev_sibling_c = child; + + child->next_sibling = node->next_sibling; + child->prev_sibling_c = node; + + node->next_sibling = child; + } + + inline void insert_node_before(xml_node_struct* child, xml_node_struct* node) + { + xml_node_struct* parent = node->parent; + + child->parent = parent; + + if (node->prev_sibling_c->next_sibling) + node->prev_sibling_c->next_sibling = child; + else + parent->first_child = child; + + child->prev_sibling_c = node->prev_sibling_c; + child->next_sibling = node; + + node->prev_sibling_c = child; + } + + inline void remove_node(xml_node_struct* node) + { + xml_node_struct* parent = node->parent; + + if (node->next_sibling) + node->next_sibling->prev_sibling_c = node->prev_sibling_c; + else + parent->first_child->prev_sibling_c = node->prev_sibling_c; + + if (node->prev_sibling_c->next_sibling) + node->prev_sibling_c->next_sibling = node->next_sibling; + else + parent->first_child = node->next_sibling; + + node->parent = 0; + node->prev_sibling_c = 0; + node->next_sibling = 0; + } + + inline void append_attribute(xml_attribute_struct* attr, xml_node_struct* node) + { + xml_attribute_struct* head = node->first_attribute; + + if (head) + { + xml_attribute_struct* tail = head->prev_attribute_c; + + tail->next_attribute = attr; + attr->prev_attribute_c = tail; + head->prev_attribute_c = attr; + } + else + { + node->first_attribute = attr; + attr->prev_attribute_c = attr; + } + } + + inline void prepend_attribute(xml_attribute_struct* attr, xml_node_struct* node) + { + xml_attribute_struct* head = node->first_attribute; + + if (head) + { + attr->prev_attribute_c = head->prev_attribute_c; + head->prev_attribute_c = attr; + } + else + attr->prev_attribute_c = attr; + + attr->next_attribute = head; + node->first_attribute = attr; + } + + inline void insert_attribute_after(xml_attribute_struct* attr, xml_attribute_struct* place, xml_node_struct* node) + { + if (place->next_attribute) + place->next_attribute->prev_attribute_c = attr; + else + node->first_attribute->prev_attribute_c = attr; + + attr->next_attribute = place->next_attribute; + attr->prev_attribute_c = place; + place->next_attribute = attr; + } + + inline void insert_attribute_before(xml_attribute_struct* attr, xml_attribute_struct* place, xml_node_struct* node) + { + if (place->prev_attribute_c->next_attribute) + place->prev_attribute_c->next_attribute = attr; + else + node->first_attribute = attr; + + attr->prev_attribute_c = place->prev_attribute_c; + attr->next_attribute = place; + place->prev_attribute_c = attr; + } + + inline void remove_attribute(xml_attribute_struct* attr, xml_node_struct* node) + { + if (attr->next_attribute) + attr->next_attribute->prev_attribute_c = attr->prev_attribute_c; + else + node->first_attribute->prev_attribute_c = attr->prev_attribute_c; + + if (attr->prev_attribute_c->next_attribute) + attr->prev_attribute_c->next_attribute = attr->next_attribute; + else + node->first_attribute = attr->next_attribute; + + attr->prev_attribute_c = 0; + attr->next_attribute = 0; + } + + PUGI__FN_NO_INLINE xml_node_struct* append_new_node(xml_node_struct* node, xml_allocator& alloc, xml_node_type type = node_element) + { + if (!alloc.reserve()) return 0; + + xml_node_struct* child = allocate_node(alloc, type); + if (!child) return 0; + + append_node(child, node); + + return child; + } + + PUGI__FN_NO_INLINE xml_attribute_struct* append_new_attribute(xml_node_struct* node, xml_allocator& alloc) + { + if (!alloc.reserve()) return 0; + + xml_attribute_struct* attr = allocate_attribute(alloc); + if (!attr) return 0; + + append_attribute(attr, node); + + return attr; + } +PUGI__NS_END + +// Helper classes for code generation +PUGI__NS_BEGIN + struct opt_false + { + enum { value = 0 }; + }; + + struct opt_true + { + enum { value = 1 }; + }; +PUGI__NS_END + +// Unicode utilities +PUGI__NS_BEGIN + inline uint16_t endian_swap(uint16_t value) + { + return static_cast(((value & 0xff) << 8) | (value >> 8)); + } + + inline uint32_t endian_swap(uint32_t value) + { + return ((value & 0xff) << 24) | ((value & 0xff00) << 8) | ((value & 0xff0000) >> 8) | (value >> 24); + } + + struct utf8_counter + { + typedef size_t value_type; + + static value_type low(value_type result, uint32_t ch) + { + // U+0000..U+007F + if (ch < 0x80) return result + 1; + // U+0080..U+07FF + else if (ch < 0x800) return result + 2; + // U+0800..U+FFFF + else return result + 3; + } + + static value_type high(value_type result, uint32_t) + { + // U+10000..U+10FFFF + return result + 4; + } + }; + + struct utf8_writer + { + typedef uint8_t* value_type; + + static value_type low(value_type result, uint32_t ch) + { + // U+0000..U+007F + if (ch < 0x80) + { + *result = static_cast(ch); + return result + 1; + } + // U+0080..U+07FF + else if (ch < 0x800) + { + result[0] = static_cast(0xC0 | (ch >> 6)); + result[1] = static_cast(0x80 | (ch & 0x3F)); + return result + 2; + } + // U+0800..U+FFFF + else + { + result[0] = static_cast(0xE0 | (ch >> 12)); + result[1] = static_cast(0x80 | ((ch >> 6) & 0x3F)); + result[2] = static_cast(0x80 | (ch & 0x3F)); + return result + 3; + } + } + + static value_type high(value_type result, uint32_t ch) + { + // U+10000..U+10FFFF + result[0] = static_cast(0xF0 | (ch >> 18)); + result[1] = static_cast(0x80 | ((ch >> 12) & 0x3F)); + result[2] = static_cast(0x80 | ((ch >> 6) & 0x3F)); + result[3] = static_cast(0x80 | (ch & 0x3F)); + return result + 4; + } + + static value_type any(value_type result, uint32_t ch) + { + return (ch < 0x10000) ? low(result, ch) : high(result, ch); + } + }; + + struct utf16_counter + { + typedef size_t value_type; + + static value_type low(value_type result, uint32_t) + { + return result + 1; + } + + static value_type high(value_type result, uint32_t) + { + return result + 2; + } + }; + + struct utf16_writer + { + typedef uint16_t* value_type; + + static value_type low(value_type result, uint32_t ch) + { + *result = static_cast(ch); + + return result + 1; + } + + static value_type high(value_type result, uint32_t ch) + { + uint32_t msh = static_cast(ch - 0x10000) >> 10; + uint32_t lsh = static_cast(ch - 0x10000) & 0x3ff; + + result[0] = static_cast(0xD800 + msh); + result[1] = static_cast(0xDC00 + lsh); + + return result + 2; + } + + static value_type any(value_type result, uint32_t ch) + { + return (ch < 0x10000) ? low(result, ch) : high(result, ch); + } + }; + + struct utf32_counter + { + typedef size_t value_type; + + static value_type low(value_type result, uint32_t) + { + return result + 1; + } + + static value_type high(value_type result, uint32_t) + { + return result + 1; + } + }; + + struct utf32_writer + { + typedef uint32_t* value_type; + + static value_type low(value_type result, uint32_t ch) + { + *result = ch; + + return result + 1; + } + + static value_type high(value_type result, uint32_t ch) + { + *result = ch; + + return result + 1; + } + + static value_type any(value_type result, uint32_t ch) + { + *result = ch; + + return result + 1; + } + }; + + struct latin1_writer + { + typedef uint8_t* value_type; + + static value_type low(value_type result, uint32_t ch) + { + *result = static_cast(ch > 255 ? '?' : ch); + + return result + 1; + } + + static value_type high(value_type result, uint32_t ch) + { + (void)ch; + + *result = '?'; + + return result + 1; + } + }; + + struct utf8_decoder + { + typedef uint8_t type; + + template static inline typename Traits::value_type process(const uint8_t* data, size_t size, typename Traits::value_type result, Traits) + { + const uint8_t utf8_byte_mask = 0x3f; + + while (size) + { + uint8_t lead = *data; + + // 0xxxxxxx -> U+0000..U+007F + if (lead < 0x80) + { + result = Traits::low(result, lead); + data += 1; + size -= 1; + + // process aligned single-byte (ascii) blocks + if ((reinterpret_cast(data) & 3) == 0) + { + // round-trip through void* to silence 'cast increases required alignment of target type' warnings + while (size >= 4 && (*static_cast(static_cast(data)) & 0x80808080) == 0) + { + result = Traits::low(result, data[0]); + result = Traits::low(result, data[1]); + result = Traits::low(result, data[2]); + result = Traits::low(result, data[3]); + data += 4; + size -= 4; + } + } + } + // 110xxxxx -> U+0080..U+07FF + else if (static_cast(lead - 0xC0) < 0x20 && size >= 2 && (data[1] & 0xc0) == 0x80) + { + result = Traits::low(result, ((lead & ~0xC0) << 6) | (data[1] & utf8_byte_mask)); + data += 2; + size -= 2; + } + // 1110xxxx -> U+0800-U+FFFF + else if (static_cast(lead - 0xE0) < 0x10 && size >= 3 && (data[1] & 0xc0) == 0x80 && (data[2] & 0xc0) == 0x80) + { + result = Traits::low(result, ((lead & ~0xE0) << 12) | ((data[1] & utf8_byte_mask) << 6) | (data[2] & utf8_byte_mask)); + data += 3; + size -= 3; + } + // 11110xxx -> U+10000..U+10FFFF + else if (static_cast(lead - 0xF0) < 0x08 && size >= 4 && (data[1] & 0xc0) == 0x80 && (data[2] & 0xc0) == 0x80 && (data[3] & 0xc0) == 0x80) + { + result = Traits::high(result, ((lead & ~0xF0) << 18) | ((data[1] & utf8_byte_mask) << 12) | ((data[2] & utf8_byte_mask) << 6) | (data[3] & utf8_byte_mask)); + data += 4; + size -= 4; + } + // 10xxxxxx or 11111xxx -> invalid + else + { + data += 1; + size -= 1; + } + } + + return result; + } + }; + + template struct utf16_decoder + { + typedef uint16_t type; + + template static inline typename Traits::value_type process(const uint16_t* data, size_t size, typename Traits::value_type result, Traits) + { + while (size) + { + uint16_t lead = opt_swap::value ? endian_swap(*data) : *data; + + // U+0000..U+D7FF + if (lead < 0xD800) + { + result = Traits::low(result, lead); + data += 1; + size -= 1; + } + // U+E000..U+FFFF + else if (static_cast(lead - 0xE000) < 0x2000) + { + result = Traits::low(result, lead); + data += 1; + size -= 1; + } + // surrogate pair lead + else if (static_cast(lead - 0xD800) < 0x400 && size >= 2) + { + uint16_t next = opt_swap::value ? endian_swap(data[1]) : data[1]; + + if (static_cast(next - 0xDC00) < 0x400) + { + result = Traits::high(result, 0x10000 + ((lead & 0x3ff) << 10) + (next & 0x3ff)); + data += 2; + size -= 2; + } + else + { + data += 1; + size -= 1; + } + } + else + { + data += 1; + size -= 1; + } + } + + return result; + } + }; + + template struct utf32_decoder + { + typedef uint32_t type; + + template static inline typename Traits::value_type process(const uint32_t* data, size_t size, typename Traits::value_type result, Traits) + { + while (size) + { + uint32_t lead = opt_swap::value ? endian_swap(*data) : *data; + + // U+0000..U+FFFF + if (lead < 0x10000) + { + result = Traits::low(result, lead); + data += 1; + size -= 1; + } + // U+10000..U+10FFFF + else + { + result = Traits::high(result, lead); + data += 1; + size -= 1; + } + } + + return result; + } + }; + + struct latin1_decoder + { + typedef uint8_t type; + + template static inline typename Traits::value_type process(const uint8_t* data, size_t size, typename Traits::value_type result, Traits) + { + while (size) + { + result = Traits::low(result, *data); + data += 1; + size -= 1; + } + + return result; + } + }; + + template struct wchar_selector; + + template <> struct wchar_selector<2> + { + typedef uint16_t type; + typedef utf16_counter counter; + typedef utf16_writer writer; + typedef utf16_decoder decoder; + }; + + template <> struct wchar_selector<4> + { + typedef uint32_t type; + typedef utf32_counter counter; + typedef utf32_writer writer; + typedef utf32_decoder decoder; + }; + + typedef wchar_selector::counter wchar_counter; + typedef wchar_selector::writer wchar_writer; + + struct wchar_decoder + { + typedef wchar_t type; + + template static inline typename Traits::value_type process(const wchar_t* data, size_t size, typename Traits::value_type result, Traits traits) + { + typedef wchar_selector::decoder decoder; + + return decoder::process(reinterpret_cast(data), size, result, traits); + } + }; + +#ifdef PUGIXML_WCHAR_MODE + PUGI__FN void convert_wchar_endian_swap(wchar_t* result, const wchar_t* data, size_t length) + { + for (size_t i = 0; i < length; ++i) + result[i] = static_cast(endian_swap(static_cast::type>(data[i]))); + } +#endif +PUGI__NS_END + +PUGI__NS_BEGIN + enum chartype_t + { + ct_parse_pcdata = 1, // \0, &, \r, < + ct_parse_attr = 2, // \0, &, \r, ', " + ct_parse_attr_ws = 4, // \0, &, \r, ', ", \n, tab + ct_space = 8, // \r, \n, space, tab + ct_parse_cdata = 16, // \0, ], >, \r + ct_parse_comment = 32, // \0, -, >, \r + ct_symbol = 64, // Any symbol > 127, a-z, A-Z, 0-9, _, :, -, . + ct_start_symbol = 128 // Any symbol > 127, a-z, A-Z, _, : + }; + + static const unsigned char chartype_table[256] = + { + 55, 0, 0, 0, 0, 0, 0, 0, 0, 12, 12, 0, 0, 63, 0, 0, // 0-15 + 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, // 16-31 + 8, 0, 6, 0, 0, 0, 7, 6, 0, 0, 0, 0, 0, 96, 64, 0, // 32-47 + 64, 64, 64, 64, 64, 64, 64, 64, 64, 64, 192, 0, 1, 0, 48, 0, // 48-63 + 0, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, // 64-79 + 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 0, 0, 16, 0, 192, // 80-95 + 0, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, // 96-111 + 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 0, 0, 0, 0, 0, // 112-127 + + 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, // 128+ + 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, + 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, + 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, + 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, + 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, + 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, + 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192, 192 + }; + + enum chartypex_t + { + ctx_special_pcdata = 1, // Any symbol >= 0 and < 32 (except \t, \r, \n), &, <, > + ctx_special_attr = 2, // Any symbol >= 0 and < 32, &, <, ", ' + ctx_start_symbol = 4, // Any symbol > 127, a-z, A-Z, _ + ctx_digit = 8, // 0-9 + ctx_symbol = 16 // Any symbol > 127, a-z, A-Z, 0-9, _, -, . + }; + + static const unsigned char chartypex_table[256] = + { + 3, 3, 3, 3, 3, 3, 3, 3, 3, 2, 2, 3, 3, 2, 3, 3, // 0-15 + 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, // 16-31 + 0, 0, 2, 0, 0, 0, 3, 2, 0, 0, 0, 0, 0, 16, 16, 0, // 32-47 + 24, 24, 24, 24, 24, 24, 24, 24, 24, 24, 0, 0, 3, 0, 1, 0, // 48-63 + + 0, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, // 64-79 + 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 0, 0, 0, 0, 20, // 80-95 + 0, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, // 96-111 + 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 0, 0, 0, 0, 0, // 112-127 + + 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, // 128+ + 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, + 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, + 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, + 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, + 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, + 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, + 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20 + }; + +#ifdef PUGIXML_WCHAR_MODE + #define PUGI__IS_CHARTYPE_IMPL(c, ct, table) ((static_cast(c) < 128 ? table[static_cast(c)] : table[128]) & (ct)) +#else + #define PUGI__IS_CHARTYPE_IMPL(c, ct, table) (table[static_cast(c)] & (ct)) +#endif + + #define PUGI__IS_CHARTYPE(c, ct) PUGI__IS_CHARTYPE_IMPL(c, ct, chartype_table) + #define PUGI__IS_CHARTYPEX(c, ct) PUGI__IS_CHARTYPE_IMPL(c, ct, chartypex_table) + + PUGI__FN bool is_little_endian() + { + unsigned int ui = 1; + + return *reinterpret_cast(&ui) == 1; + } + + PUGI__FN xml_encoding get_wchar_encoding() + { + PUGI__STATIC_ASSERT(sizeof(wchar_t) == 2 || sizeof(wchar_t) == 4); + + if (sizeof(wchar_t) == 2) + return is_little_endian() ? encoding_utf16_le : encoding_utf16_be; + else + return is_little_endian() ? encoding_utf32_le : encoding_utf32_be; + } + + PUGI__FN bool parse_declaration_encoding(const uint8_t* data, size_t size, const uint8_t*& out_encoding, size_t& out_length) + { + #define PUGI__SCANCHAR(ch) { if (offset >= size || data[offset] != ch) return false; offset++; } + #define PUGI__SCANCHARTYPE(ct) { while (offset < size && PUGI__IS_CHARTYPE(data[offset], ct)) offset++; } + + // check if we have a non-empty XML declaration + if (size < 6 || !((data[0] == '<') & (data[1] == '?') & (data[2] == 'x') & (data[3] == 'm') & (data[4] == 'l') && PUGI__IS_CHARTYPE(data[5], ct_space))) + return false; + + // scan XML declaration until the encoding field + for (size_t i = 6; i + 1 < size; ++i) + { + // declaration can not contain ? in quoted values + if (data[i] == '?') + return false; + + if (data[i] == 'e' && data[i + 1] == 'n') + { + size_t offset = i; + + // encoding follows the version field which can't contain 'en' so this has to be the encoding if XML is well formed + PUGI__SCANCHAR('e'); PUGI__SCANCHAR('n'); PUGI__SCANCHAR('c'); PUGI__SCANCHAR('o'); + PUGI__SCANCHAR('d'); PUGI__SCANCHAR('i'); PUGI__SCANCHAR('n'); PUGI__SCANCHAR('g'); + + // S? = S? + PUGI__SCANCHARTYPE(ct_space); + PUGI__SCANCHAR('='); + PUGI__SCANCHARTYPE(ct_space); + + // the only two valid delimiters are ' and " + uint8_t delimiter = (offset < size && data[offset] == '"') ? '"' : '\''; + + PUGI__SCANCHAR(delimiter); + + size_t start = offset; + + out_encoding = data + offset; + + PUGI__SCANCHARTYPE(ct_symbol); + + out_length = offset - start; + + PUGI__SCANCHAR(delimiter); + + return true; + } + } + + return false; + + #undef PUGI__SCANCHAR + #undef PUGI__SCANCHARTYPE + } + + PUGI__FN xml_encoding guess_buffer_encoding(const uint8_t* data, size_t size) + { + // skip encoding autodetection if input buffer is too small + if (size < 4) return encoding_utf8; + + uint8_t d0 = data[0], d1 = data[1], d2 = data[2], d3 = data[3]; + + // look for BOM in first few bytes + if (d0 == 0 && d1 == 0 && d2 == 0xfe && d3 == 0xff) return encoding_utf32_be; + if (d0 == 0xff && d1 == 0xfe && d2 == 0 && d3 == 0) return encoding_utf32_le; + if (d0 == 0xfe && d1 == 0xff) return encoding_utf16_be; + if (d0 == 0xff && d1 == 0xfe) return encoding_utf16_le; + if (d0 == 0xef && d1 == 0xbb && d2 == 0xbf) return encoding_utf8; + + // look for <, (contents); + + return guess_buffer_encoding(data, size); + } + + PUGI__FN bool get_mutable_buffer(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, bool is_mutable) + { + size_t length = size / sizeof(char_t); + + if (is_mutable) + { + out_buffer = static_cast(const_cast(contents)); + out_length = length; + } + else + { + char_t* buffer = static_cast(xml_memory::allocate((length + 1) * sizeof(char_t))); + if (!buffer) return false; + + if (contents) + memcpy(buffer, contents, length * sizeof(char_t)); + else + assert(length == 0); + + buffer[length] = 0; + + out_buffer = buffer; + out_length = length + 1; + } + + return true; + } + +#ifdef PUGIXML_WCHAR_MODE + PUGI__FN bool need_endian_swap_utf(xml_encoding le, xml_encoding re) + { + return (le == encoding_utf16_be && re == encoding_utf16_le) || (le == encoding_utf16_le && re == encoding_utf16_be) || + (le == encoding_utf32_be && re == encoding_utf32_le) || (le == encoding_utf32_le && re == encoding_utf32_be); + } + + PUGI__FN bool convert_buffer_endian_swap(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, bool is_mutable) + { + const char_t* data = static_cast(contents); + size_t length = size / sizeof(char_t); + + if (is_mutable) + { + char_t* buffer = const_cast(data); + + convert_wchar_endian_swap(buffer, data, length); + + out_buffer = buffer; + out_length = length; + } + else + { + char_t* buffer = static_cast(xml_memory::allocate((length + 1) * sizeof(char_t))); + if (!buffer) return false; + + convert_wchar_endian_swap(buffer, data, length); + buffer[length] = 0; + + out_buffer = buffer; + out_length = length + 1; + } + + return true; + } + + template PUGI__FN bool convert_buffer_generic(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, D) + { + const typename D::type* data = static_cast(contents); + size_t data_length = size / sizeof(typename D::type); + + // first pass: get length in wchar_t units + size_t length = D::process(data, data_length, 0, wchar_counter()); + + // allocate buffer of suitable length + char_t* buffer = static_cast(xml_memory::allocate((length + 1) * sizeof(char_t))); + if (!buffer) return false; + + // second pass: convert utf16 input to wchar_t + wchar_writer::value_type obegin = reinterpret_cast(buffer); + wchar_writer::value_type oend = D::process(data, data_length, obegin, wchar_writer()); + + assert(oend == obegin + length); + *oend = 0; + + out_buffer = buffer; + out_length = length + 1; + + return true; + } + + PUGI__FN bool convert_buffer(char_t*& out_buffer, size_t& out_length, xml_encoding encoding, const void* contents, size_t size, bool is_mutable) + { + // get native encoding + xml_encoding wchar_encoding = get_wchar_encoding(); + + // fast path: no conversion required + if (encoding == wchar_encoding) + return get_mutable_buffer(out_buffer, out_length, contents, size, is_mutable); + + // only endian-swapping is required + if (need_endian_swap_utf(encoding, wchar_encoding)) + return convert_buffer_endian_swap(out_buffer, out_length, contents, size, is_mutable); + + // source encoding is utf8 + if (encoding == encoding_utf8) + return convert_buffer_generic(out_buffer, out_length, contents, size, utf8_decoder()); + + // source encoding is utf16 + if (encoding == encoding_utf16_be || encoding == encoding_utf16_le) + { + xml_encoding native_encoding = is_little_endian() ? encoding_utf16_le : encoding_utf16_be; + + return (native_encoding == encoding) ? + convert_buffer_generic(out_buffer, out_length, contents, size, utf16_decoder()) : + convert_buffer_generic(out_buffer, out_length, contents, size, utf16_decoder()); + } + + // source encoding is utf32 + if (encoding == encoding_utf32_be || encoding == encoding_utf32_le) + { + xml_encoding native_encoding = is_little_endian() ? encoding_utf32_le : encoding_utf32_be; + + return (native_encoding == encoding) ? + convert_buffer_generic(out_buffer, out_length, contents, size, utf32_decoder()) : + convert_buffer_generic(out_buffer, out_length, contents, size, utf32_decoder()); + } + + // source encoding is latin1 + if (encoding == encoding_latin1) + return convert_buffer_generic(out_buffer, out_length, contents, size, latin1_decoder()); + + assert(false && "Invalid encoding"); // unreachable + return false; + } +#else + template PUGI__FN bool convert_buffer_generic(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, D) + { + const typename D::type* data = static_cast(contents); + size_t data_length = size / sizeof(typename D::type); + + // first pass: get length in utf8 units + size_t length = D::process(data, data_length, 0, utf8_counter()); + + // allocate buffer of suitable length + char_t* buffer = static_cast(xml_memory::allocate((length + 1) * sizeof(char_t))); + if (!buffer) return false; + + // second pass: convert utf16 input to utf8 + uint8_t* obegin = reinterpret_cast(buffer); + uint8_t* oend = D::process(data, data_length, obegin, utf8_writer()); + + assert(oend == obegin + length); + *oend = 0; + + out_buffer = buffer; + out_length = length + 1; + + return true; + } + + PUGI__FN size_t get_latin1_7bit_prefix_length(const uint8_t* data, size_t size) + { + for (size_t i = 0; i < size; ++i) + if (data[i] > 127) + return i; + + return size; + } + + PUGI__FN bool convert_buffer_latin1(char_t*& out_buffer, size_t& out_length, const void* contents, size_t size, bool is_mutable) + { + const uint8_t* data = static_cast(contents); + size_t data_length = size; + + // get size of prefix that does not need utf8 conversion + size_t prefix_length = get_latin1_7bit_prefix_length(data, data_length); + assert(prefix_length <= data_length); + + const uint8_t* postfix = data + prefix_length; + size_t postfix_length = data_length - prefix_length; + + // if no conversion is needed, just return the original buffer + if (postfix_length == 0) return get_mutable_buffer(out_buffer, out_length, contents, size, is_mutable); + + // first pass: get length in utf8 units + size_t length = prefix_length + latin1_decoder::process(postfix, postfix_length, 0, utf8_counter()); + + // allocate buffer of suitable length + char_t* buffer = static_cast(xml_memory::allocate((length + 1) * sizeof(char_t))); + if (!buffer) return false; + + // second pass: convert latin1 input to utf8 + memcpy(buffer, data, prefix_length); + + uint8_t* obegin = reinterpret_cast(buffer); + uint8_t* oend = latin1_decoder::process(postfix, postfix_length, obegin + prefix_length, utf8_writer()); + + assert(oend == obegin + length); + *oend = 0; + + out_buffer = buffer; + out_length = length + 1; + + return true; + } + + PUGI__FN bool convert_buffer(char_t*& out_buffer, size_t& out_length, xml_encoding encoding, const void* contents, size_t size, bool is_mutable) + { + // fast path: no conversion required + if (encoding == encoding_utf8) + return get_mutable_buffer(out_buffer, out_length, contents, size, is_mutable); + + // source encoding is utf16 + if (encoding == encoding_utf16_be || encoding == encoding_utf16_le) + { + xml_encoding native_encoding = is_little_endian() ? encoding_utf16_le : encoding_utf16_be; + + return (native_encoding == encoding) ? + convert_buffer_generic(out_buffer, out_length, contents, size, utf16_decoder()) : + convert_buffer_generic(out_buffer, out_length, contents, size, utf16_decoder()); + } + + // source encoding is utf32 + if (encoding == encoding_utf32_be || encoding == encoding_utf32_le) + { + xml_encoding native_encoding = is_little_endian() ? encoding_utf32_le : encoding_utf32_be; + + return (native_encoding == encoding) ? + convert_buffer_generic(out_buffer, out_length, contents, size, utf32_decoder()) : + convert_buffer_generic(out_buffer, out_length, contents, size, utf32_decoder()); + } + + // source encoding is latin1 + if (encoding == encoding_latin1) + return convert_buffer_latin1(out_buffer, out_length, contents, size, is_mutable); + + assert(false && "Invalid encoding"); // unreachable + return false; + } +#endif + + PUGI__FN size_t as_utf8_begin(const wchar_t* str, size_t length) + { + // get length in utf8 characters + return wchar_decoder::process(str, length, 0, utf8_counter()); + } + + PUGI__FN void as_utf8_end(char* buffer, size_t size, const wchar_t* str, size_t length) + { + // convert to utf8 + uint8_t* begin = reinterpret_cast(buffer); + uint8_t* end = wchar_decoder::process(str, length, begin, utf8_writer()); + + assert(begin + size == end); + (void)!end; + (void)!size; + } + +#ifndef PUGIXML_NO_STL + PUGI__FN std::string as_utf8_impl(const wchar_t* str, size_t length) + { + // first pass: get length in utf8 characters + size_t size = as_utf8_begin(str, length); + + // allocate resulting string + std::string result; + result.resize(size); + + // second pass: convert to utf8 + if (size > 0) as_utf8_end(&result[0], size, str, length); + + return result; + } + + PUGI__FN std::basic_string as_wide_impl(const char* str, size_t size) + { + const uint8_t* data = reinterpret_cast(str); + + // first pass: get length in wchar_t units + size_t length = utf8_decoder::process(data, size, 0, wchar_counter()); + + // allocate resulting string + std::basic_string result; + result.resize(length); + + // second pass: convert to wchar_t + if (length > 0) + { + wchar_writer::value_type begin = reinterpret_cast(&result[0]); + wchar_writer::value_type end = utf8_decoder::process(data, size, begin, wchar_writer()); + + assert(begin + length == end); + (void)!end; + } + + return result; + } +#endif + + template + inline bool strcpy_insitu_allow(size_t length, const Header& header, uintptr_t header_mask, char_t* target) + { + // never reuse shared memory + if (header & xml_memory_page_contents_shared_mask) return false; + + size_t target_length = strlength(target); + + // always reuse document buffer memory if possible + if ((header & header_mask) == 0) return target_length >= length; + + // reuse heap memory if waste is not too great + const size_t reuse_threshold = 32; + + return target_length >= length && (target_length < reuse_threshold || target_length - length < target_length / 2); + } + + template + PUGI__FN bool strcpy_insitu(String& dest, Header& header, uintptr_t header_mask, const char_t* source, size_t source_length) + { + if (source_length == 0) + { + // empty string and null pointer are equivalent, so just deallocate old memory + xml_allocator* alloc = PUGI__GETPAGE_IMPL(header)->allocator; + + if (header & header_mask) alloc->deallocate_string(dest); + + // mark the string as not allocated + dest = 0; + header &= ~header_mask; + + return true; + } + else if (dest && strcpy_insitu_allow(source_length, header, header_mask, dest)) + { + // we can reuse old buffer, so just copy the new data (including zero terminator) + memcpy(dest, source, source_length * sizeof(char_t)); + dest[source_length] = 0; + + return true; + } + else + { + xml_allocator* alloc = PUGI__GETPAGE_IMPL(header)->allocator; + + if (!alloc->reserve()) return false; + + // allocate new buffer + char_t* buf = alloc->allocate_string(source_length + 1); + if (!buf) return false; + + // copy the string (including zero terminator) + memcpy(buf, source, source_length * sizeof(char_t)); + buf[source_length] = 0; + + // deallocate old buffer (*after* the above to protect against overlapping memory and/or allocation failures) + if (header & header_mask) alloc->deallocate_string(dest); + + // the string is now allocated, so set the flag + dest = buf; + header |= header_mask; + + return true; + } + } + + struct gap + { + char_t* end; + size_t size; + + gap(): end(0), size(0) + { + } + + // Push new gap, move s count bytes further (skipping the gap). + // Collapse previous gap. + void push(char_t*& s, size_t count) + { + if (end) // there was a gap already; collapse it + { + // Move [old_gap_end, new_gap_start) to [old_gap_start, ...) + assert(s >= end); + memmove(end - size, end, reinterpret_cast(s) - reinterpret_cast(end)); + } + + s += count; // end of current gap + + // "merge" two gaps + end = s; + size += count; + } + + // Collapse all gaps, return past-the-end pointer + char_t* flush(char_t* s) + { + if (end) + { + // Move [old_gap_end, current_pos) to [old_gap_start, ...) + assert(s >= end); + memmove(end - size, end, reinterpret_cast(s) - reinterpret_cast(end)); + + return s - size; + } + else return s; + } + }; + + PUGI__FN char_t* strconv_escape(char_t* s, gap& g) + { + char_t* stre = s + 1; + + switch (*stre) + { + case '#': // &#... + { + unsigned int ucsc = 0; + + if (stre[1] == 'x') // &#x... (hex code) + { + stre += 2; + + char_t ch = *stre; + + if (ch == ';') return stre; + + for (;;) + { + if (static_cast(ch - '0') <= 9) + ucsc = 16 * ucsc + (ch - '0'); + else if (static_cast((ch | ' ') - 'a') <= 5) + ucsc = 16 * ucsc + ((ch | ' ') - 'a' + 10); + else if (ch == ';') + break; + else // cancel + return stre; + + ch = *++stre; + } + + ++stre; + } + else // &#... (dec code) + { + char_t ch = *++stre; + + if (ch == ';') return stre; + + for (;;) + { + if (static_cast(ch - '0') <= 9) + ucsc = 10 * ucsc + (ch - '0'); + else if (ch == ';') + break; + else // cancel + return stre; + + ch = *++stre; + } + + ++stre; + } + + #ifdef PUGIXML_WCHAR_MODE + s = reinterpret_cast(wchar_writer::any(reinterpret_cast(s), ucsc)); + #else + s = reinterpret_cast(utf8_writer::any(reinterpret_cast(s), ucsc)); + #endif + + g.push(s, stre - s); + return stre; + } + + case 'a': // &a + { + ++stre; + + if (*stre == 'm') // &am + { + if (*++stre == 'p' && *++stre == ';') // & + { + *s++ = '&'; + ++stre; + + g.push(s, stre - s); + return stre; + } + } + else if (*stre == 'p') // &ap + { + if (*++stre == 'o' && *++stre == 's' && *++stre == ';') // ' + { + *s++ = '\''; + ++stre; + + g.push(s, stre - s); + return stre; + } + } + break; + } + + case 'g': // &g + { + if (*++stre == 't' && *++stre == ';') // > + { + *s++ = '>'; + ++stre; + + g.push(s, stre - s); + return stre; + } + break; + } + + case 'l': // &l + { + if (*++stre == 't' && *++stre == ';') // < + { + *s++ = '<'; + ++stre; + + g.push(s, stre - s); + return stre; + } + break; + } + + case 'q': // &q + { + if (*++stre == 'u' && *++stre == 'o' && *++stre == 't' && *++stre == ';') // " + { + *s++ = '"'; + ++stre; + + g.push(s, stre - s); + return stre; + } + break; + } + + default: + break; + } + + return stre; + } + + // Parser utilities + #define PUGI__ENDSWITH(c, e) ((c) == (e) || ((c) == 0 && endch == (e))) + #define PUGI__SKIPWS() { while (PUGI__IS_CHARTYPE(*s, ct_space)) ++s; } + #define PUGI__OPTSET(OPT) ( optmsk & (OPT) ) + #define PUGI__PUSHNODE(TYPE) { cursor = append_new_node(cursor, *alloc, TYPE); if (!cursor) PUGI__THROW_ERROR(status_out_of_memory, s); } + #define PUGI__POPNODE() { cursor = cursor->parent; } + #define PUGI__SCANFOR(X) { while (*s != 0 && !(X)) ++s; } + #define PUGI__SCANWHILE(X) { while (X) ++s; } + #define PUGI__SCANWHILE_UNROLL(X) { for (;;) { char_t ss = s[0]; if (PUGI__UNLIKELY(!(X))) { break; } ss = s[1]; if (PUGI__UNLIKELY(!(X))) { s += 1; break; } ss = s[2]; if (PUGI__UNLIKELY(!(X))) { s += 2; break; } ss = s[3]; if (PUGI__UNLIKELY(!(X))) { s += 3; break; } s += 4; } } + #define PUGI__ENDSEG() { ch = *s; *s = 0; ++s; } + #define PUGI__THROW_ERROR(err, m) return error_offset = m, error_status = err, static_cast(0) + #define PUGI__CHECK_ERROR(err, m) { if (*s == 0) PUGI__THROW_ERROR(err, m); } + + PUGI__FN char_t* strconv_comment(char_t* s, char_t endch) + { + gap g; + + while (true) + { + PUGI__SCANWHILE_UNROLL(!PUGI__IS_CHARTYPE(ss, ct_parse_comment)); + + if (*s == '\r') // Either a single 0x0d or 0x0d 0x0a pair + { + *s++ = '\n'; // replace first one with 0x0a + + if (*s == '\n') g.push(s, 1); + } + else if (s[0] == '-' && s[1] == '-' && PUGI__ENDSWITH(s[2], '>')) // comment ends here + { + *g.flush(s) = 0; + + return s + (s[2] == '>' ? 3 : 2); + } + else if (*s == 0) + { + return 0; + } + else ++s; + } + } + + PUGI__FN char_t* strconv_cdata(char_t* s, char_t endch) + { + gap g; + + while (true) + { + PUGI__SCANWHILE_UNROLL(!PUGI__IS_CHARTYPE(ss, ct_parse_cdata)); + + if (*s == '\r') // Either a single 0x0d or 0x0d 0x0a pair + { + *s++ = '\n'; // replace first one with 0x0a + + if (*s == '\n') g.push(s, 1); + } + else if (s[0] == ']' && s[1] == ']' && PUGI__ENDSWITH(s[2], '>')) // CDATA ends here + { + *g.flush(s) = 0; + + return s + 1; + } + else if (*s == 0) + { + return 0; + } + else ++s; + } + } + + typedef char_t* (*strconv_pcdata_t)(char_t*); + + template struct strconv_pcdata_impl + { + static char_t* parse(char_t* s) + { + gap g; + + char_t* begin = s; + + while (true) + { + PUGI__SCANWHILE_UNROLL(!PUGI__IS_CHARTYPE(ss, ct_parse_pcdata)); + + if (*s == '<') // PCDATA ends here + { + char_t* end = g.flush(s); + + if (opt_trim::value) + while (end > begin && PUGI__IS_CHARTYPE(end[-1], ct_space)) + --end; + + *end = 0; + + return s + 1; + } + else if (opt_eol::value && *s == '\r') // Either a single 0x0d or 0x0d 0x0a pair + { + *s++ = '\n'; // replace first one with 0x0a + + if (*s == '\n') g.push(s, 1); + } + else if (opt_escape::value && *s == '&') + { + s = strconv_escape(s, g); + } + else if (*s == 0) + { + char_t* end = g.flush(s); + + if (opt_trim::value) + while (end > begin && PUGI__IS_CHARTYPE(end[-1], ct_space)) + --end; + + *end = 0; + + return s; + } + else ++s; + } + } + }; + + PUGI__FN strconv_pcdata_t get_strconv_pcdata(unsigned int optmask) + { + PUGI__STATIC_ASSERT(parse_escapes == 0x10 && parse_eol == 0x20 && parse_trim_pcdata == 0x0800); + + switch (((optmask >> 4) & 3) | ((optmask >> 9) & 4)) // get bitmask for flags (trim eol escapes); this simultaneously checks 3 options from assertion above + { + case 0: return strconv_pcdata_impl::parse; + case 1: return strconv_pcdata_impl::parse; + case 2: return strconv_pcdata_impl::parse; + case 3: return strconv_pcdata_impl::parse; + case 4: return strconv_pcdata_impl::parse; + case 5: return strconv_pcdata_impl::parse; + case 6: return strconv_pcdata_impl::parse; + case 7: return strconv_pcdata_impl::parse; + default: assert(false); return 0; // unreachable + } + } + + typedef char_t* (*strconv_attribute_t)(char_t*, char_t); + + template struct strconv_attribute_impl + { + static char_t* parse_wnorm(char_t* s, char_t end_quote) + { + gap g; + + // trim leading whitespaces + if (PUGI__IS_CHARTYPE(*s, ct_space)) + { + char_t* str = s; + + do ++str; + while (PUGI__IS_CHARTYPE(*str, ct_space)); + + g.push(s, str - s); + } + + while (true) + { + PUGI__SCANWHILE_UNROLL(!PUGI__IS_CHARTYPE(ss, ct_parse_attr_ws | ct_space)); + + if (*s == end_quote) + { + char_t* str = g.flush(s); + + do *str-- = 0; + while (PUGI__IS_CHARTYPE(*str, ct_space)); + + return s + 1; + } + else if (PUGI__IS_CHARTYPE(*s, ct_space)) + { + *s++ = ' '; + + if (PUGI__IS_CHARTYPE(*s, ct_space)) + { + char_t* str = s + 1; + while (PUGI__IS_CHARTYPE(*str, ct_space)) ++str; + + g.push(s, str - s); + } + } + else if (opt_escape::value && *s == '&') + { + s = strconv_escape(s, g); + } + else if (!*s) + { + return 0; + } + else ++s; + } + } + + static char_t* parse_wconv(char_t* s, char_t end_quote) + { + gap g; + + while (true) + { + PUGI__SCANWHILE_UNROLL(!PUGI__IS_CHARTYPE(ss, ct_parse_attr_ws)); + + if (*s == end_quote) + { + *g.flush(s) = 0; + + return s + 1; + } + else if (PUGI__IS_CHARTYPE(*s, ct_space)) + { + if (*s == '\r') + { + *s++ = ' '; + + if (*s == '\n') g.push(s, 1); + } + else *s++ = ' '; + } + else if (opt_escape::value && *s == '&') + { + s = strconv_escape(s, g); + } + else if (!*s) + { + return 0; + } + else ++s; + } + } + + static char_t* parse_eol(char_t* s, char_t end_quote) + { + gap g; + + while (true) + { + PUGI__SCANWHILE_UNROLL(!PUGI__IS_CHARTYPE(ss, ct_parse_attr)); + + if (*s == end_quote) + { + *g.flush(s) = 0; + + return s + 1; + } + else if (*s == '\r') + { + *s++ = '\n'; + + if (*s == '\n') g.push(s, 1); + } + else if (opt_escape::value && *s == '&') + { + s = strconv_escape(s, g); + } + else if (!*s) + { + return 0; + } + else ++s; + } + } + + static char_t* parse_simple(char_t* s, char_t end_quote) + { + gap g; + + while (true) + { + PUGI__SCANWHILE_UNROLL(!PUGI__IS_CHARTYPE(ss, ct_parse_attr)); + + if (*s == end_quote) + { + *g.flush(s) = 0; + + return s + 1; + } + else if (opt_escape::value && *s == '&') + { + s = strconv_escape(s, g); + } + else if (!*s) + { + return 0; + } + else ++s; + } + } + }; + + PUGI__FN strconv_attribute_t get_strconv_attribute(unsigned int optmask) + { + PUGI__STATIC_ASSERT(parse_escapes == 0x10 && parse_eol == 0x20 && parse_wconv_attribute == 0x40 && parse_wnorm_attribute == 0x80); + + switch ((optmask >> 4) & 15) // get bitmask for flags (wnorm wconv eol escapes); this simultaneously checks 4 options from assertion above + { + case 0: return strconv_attribute_impl::parse_simple; + case 1: return strconv_attribute_impl::parse_simple; + case 2: return strconv_attribute_impl::parse_eol; + case 3: return strconv_attribute_impl::parse_eol; + case 4: return strconv_attribute_impl::parse_wconv; + case 5: return strconv_attribute_impl::parse_wconv; + case 6: return strconv_attribute_impl::parse_wconv; + case 7: return strconv_attribute_impl::parse_wconv; + case 8: return strconv_attribute_impl::parse_wnorm; + case 9: return strconv_attribute_impl::parse_wnorm; + case 10: return strconv_attribute_impl::parse_wnorm; + case 11: return strconv_attribute_impl::parse_wnorm; + case 12: return strconv_attribute_impl::parse_wnorm; + case 13: return strconv_attribute_impl::parse_wnorm; + case 14: return strconv_attribute_impl::parse_wnorm; + case 15: return strconv_attribute_impl::parse_wnorm; + default: assert(false); return 0; // unreachable + } + } + + inline xml_parse_result make_parse_result(xml_parse_status status, ptrdiff_t offset = 0) + { + xml_parse_result result; + result.status = status; + result.offset = offset; + + return result; + } + + struct xml_parser + { + xml_allocator* alloc; + char_t* error_offset; + xml_parse_status error_status; + + xml_parser(xml_allocator* alloc_): alloc(alloc_), error_offset(0), error_status(status_ok) + { + } + + // DOCTYPE consists of nested sections of the following possible types: + // , , "...", '...' + // + // + // First group can not contain nested groups + // Second group can contain nested groups of the same type + // Third group can contain all other groups + char_t* parse_doctype_primitive(char_t* s) + { + if (*s == '"' || *s == '\'') + { + // quoted string + char_t ch = *s++; + PUGI__SCANFOR(*s == ch); + if (!*s) PUGI__THROW_ERROR(status_bad_doctype, s); + + s++; + } + else if (s[0] == '<' && s[1] == '?') + { + // + s += 2; + PUGI__SCANFOR(s[0] == '?' && s[1] == '>'); // no need for ENDSWITH because ?> can't terminate proper doctype + if (!*s) PUGI__THROW_ERROR(status_bad_doctype, s); + + s += 2; + } + else if (s[0] == '<' && s[1] == '!' && s[2] == '-' && s[3] == '-') + { + s += 4; + PUGI__SCANFOR(s[0] == '-' && s[1] == '-' && s[2] == '>'); // no need for ENDSWITH because --> can't terminate proper doctype + if (!*s) PUGI__THROW_ERROR(status_bad_doctype, s); + + s += 3; + } + else PUGI__THROW_ERROR(status_bad_doctype, s); + + return s; + } + + char_t* parse_doctype_ignore(char_t* s) + { + size_t depth = 0; + + assert(s[0] == '<' && s[1] == '!' && s[2] == '['); + s += 3; + + while (*s) + { + if (s[0] == '<' && s[1] == '!' && s[2] == '[') + { + // nested ignore section + s += 3; + depth++; + } + else if (s[0] == ']' && s[1] == ']' && s[2] == '>') + { + // ignore section end + s += 3; + + if (depth == 0) + return s; + + depth--; + } + else s++; + } + + PUGI__THROW_ERROR(status_bad_doctype, s); + } + + char_t* parse_doctype_group(char_t* s, char_t endch) + { + size_t depth = 0; + + assert((s[0] == '<' || s[0] == 0) && s[1] == '!'); + s += 2; + + while (*s) + { + if (s[0] == '<' && s[1] == '!' && s[2] != '-') + { + if (s[2] == '[') + { + // ignore + s = parse_doctype_ignore(s); + if (!s) return s; + } + else + { + // some control group + s += 2; + depth++; + } + } + else if (s[0] == '<' || s[0] == '"' || s[0] == '\'') + { + // unknown tag (forbidden), or some primitive group + s = parse_doctype_primitive(s); + if (!s) return s; + } + else if (*s == '>') + { + if (depth == 0) + return s; + + depth--; + s++; + } + else s++; + } + + if (depth != 0 || endch != '>') PUGI__THROW_ERROR(status_bad_doctype, s); + + return s; + } + + char_t* parse_exclamation(char_t* s, xml_node_struct* cursor, unsigned int optmsk, char_t endch) + { + // parse node contents, starting with exclamation mark + ++s; + + if (*s == '-') // 'value = s; // Save the offset. + } + + if (PUGI__OPTSET(parse_eol) && PUGI__OPTSET(parse_comments)) + { + s = strconv_comment(s, endch); + + if (!s) PUGI__THROW_ERROR(status_bad_comment, cursor->value); + } + else + { + // Scan for terminating '-->'. + PUGI__SCANFOR(s[0] == '-' && s[1] == '-' && PUGI__ENDSWITH(s[2], '>')); + PUGI__CHECK_ERROR(status_bad_comment, s); + + if (PUGI__OPTSET(parse_comments)) + *s = 0; // Zero-terminate this segment at the first terminating '-'. + + s += (s[2] == '>' ? 3 : 2); // Step over the '\0->'. + } + } + else PUGI__THROW_ERROR(status_bad_comment, s); + } + else if (*s == '[') + { + // 'value = s; // Save the offset. + + if (PUGI__OPTSET(parse_eol)) + { + s = strconv_cdata(s, endch); + + if (!s) PUGI__THROW_ERROR(status_bad_cdata, cursor->value); + } + else + { + // Scan for terminating ']]>'. + PUGI__SCANFOR(s[0] == ']' && s[1] == ']' && PUGI__ENDSWITH(s[2], '>')); + PUGI__CHECK_ERROR(status_bad_cdata, s); + + *s++ = 0; // Zero-terminate this segment. + } + } + else // Flagged for discard, but we still have to scan for the terminator. + { + // Scan for terminating ']]>'. + PUGI__SCANFOR(s[0] == ']' && s[1] == ']' && PUGI__ENDSWITH(s[2], '>')); + PUGI__CHECK_ERROR(status_bad_cdata, s); + + ++s; + } + + s += (s[1] == '>' ? 2 : 1); // Step over the last ']>'. + } + else PUGI__THROW_ERROR(status_bad_cdata, s); + } + else if (s[0] == 'D' && s[1] == 'O' && s[2] == 'C' && s[3] == 'T' && s[4] == 'Y' && s[5] == 'P' && PUGI__ENDSWITH(s[6], 'E')) + { + s -= 2; + + if (cursor->parent) PUGI__THROW_ERROR(status_bad_doctype, s); + + char_t* mark = s + 9; + + s = parse_doctype_group(s, endch); + if (!s) return s; + + assert((*s == 0 && endch == '>') || *s == '>'); + if (*s) *s++ = 0; + + if (PUGI__OPTSET(parse_doctype)) + { + while (PUGI__IS_CHARTYPE(*mark, ct_space)) ++mark; + + PUGI__PUSHNODE(node_doctype); + + cursor->value = mark; + } + } + else if (*s == 0 && endch == '-') PUGI__THROW_ERROR(status_bad_comment, s); + else if (*s == 0 && endch == '[') PUGI__THROW_ERROR(status_bad_cdata, s); + else PUGI__THROW_ERROR(status_unrecognized_tag, s); + + return s; + } + + char_t* parse_question(char_t* s, xml_node_struct*& ref_cursor, unsigned int optmsk, char_t endch) + { + // load into registers + xml_node_struct* cursor = ref_cursor; + char_t ch = 0; + + // parse node contents, starting with question mark + ++s; + + // read PI target + char_t* target = s; + + if (!PUGI__IS_CHARTYPE(*s, ct_start_symbol)) PUGI__THROW_ERROR(status_bad_pi, s); + + PUGI__SCANWHILE(PUGI__IS_CHARTYPE(*s, ct_symbol)); + PUGI__CHECK_ERROR(status_bad_pi, s); + + // determine node type; stricmp / strcasecmp is not portable + bool declaration = (target[0] | ' ') == 'x' && (target[1] | ' ') == 'm' && (target[2] | ' ') == 'l' && target + 3 == s; + + if (declaration ? PUGI__OPTSET(parse_declaration) : PUGI__OPTSET(parse_pi)) + { + if (declaration) + { + // disallow non top-level declarations + if (cursor->parent) PUGI__THROW_ERROR(status_bad_pi, s); + + PUGI__PUSHNODE(node_declaration); + } + else + { + PUGI__PUSHNODE(node_pi); + } + + cursor->name = target; + + PUGI__ENDSEG(); + + // parse value/attributes + if (ch == '?') + { + // empty node + if (!PUGI__ENDSWITH(*s, '>')) PUGI__THROW_ERROR(status_bad_pi, s); + s += (*s == '>'); + + PUGI__POPNODE(); + } + else if (PUGI__IS_CHARTYPE(ch, ct_space)) + { + PUGI__SKIPWS(); + + // scan for tag end + char_t* value = s; + + PUGI__SCANFOR(s[0] == '?' && PUGI__ENDSWITH(s[1], '>')); + PUGI__CHECK_ERROR(status_bad_pi, s); + + if (declaration) + { + // replace ending ? with / so that 'element' terminates properly + *s = '/'; + + // we exit from this function with cursor at node_declaration, which is a signal to parse() to go to LOC_ATTRIBUTES + s = value; + } + else + { + // store value and step over > + cursor->value = value; + + PUGI__POPNODE(); + + PUGI__ENDSEG(); + + s += (*s == '>'); + } + } + else PUGI__THROW_ERROR(status_bad_pi, s); + } + else + { + // scan for tag end + PUGI__SCANFOR(s[0] == '?' && PUGI__ENDSWITH(s[1], '>')); + PUGI__CHECK_ERROR(status_bad_pi, s); + + s += (s[1] == '>' ? 2 : 1); + } + + // store from registers + ref_cursor = cursor; + + return s; + } + + char_t* parse_tree(char_t* s, xml_node_struct* root, unsigned int optmsk, char_t endch) + { + strconv_attribute_t strconv_attribute = get_strconv_attribute(optmsk); + strconv_pcdata_t strconv_pcdata = get_strconv_pcdata(optmsk); + + char_t ch = 0; + xml_node_struct* cursor = root; + char_t* mark = s; + + while (*s != 0) + { + if (*s == '<') + { + ++s; + + LOC_TAG: + if (PUGI__IS_CHARTYPE(*s, ct_start_symbol)) // '<#...' + { + PUGI__PUSHNODE(node_element); // Append a new node to the tree. + + cursor->name = s; + + PUGI__SCANWHILE_UNROLL(PUGI__IS_CHARTYPE(ss, ct_symbol)); // Scan for a terminator. + PUGI__ENDSEG(); // Save char in 'ch', terminate & step over. + + if (ch == '>') + { + // end of tag + } + else if (PUGI__IS_CHARTYPE(ch, ct_space)) + { + LOC_ATTRIBUTES: + while (true) + { + PUGI__SKIPWS(); // Eat any whitespace. + + if (PUGI__IS_CHARTYPE(*s, ct_start_symbol)) // <... #... + { + xml_attribute_struct* a = append_new_attribute(cursor, *alloc); // Make space for this attribute. + if (!a) PUGI__THROW_ERROR(status_out_of_memory, s); + + a->name = s; // Save the offset. + + PUGI__SCANWHILE_UNROLL(PUGI__IS_CHARTYPE(ss, ct_symbol)); // Scan for a terminator. + PUGI__ENDSEG(); // Save char in 'ch', terminate & step over. + + if (PUGI__IS_CHARTYPE(ch, ct_space)) + { + PUGI__SKIPWS(); // Eat any whitespace. + + ch = *s; + ++s; + } + + if (ch == '=') // '<... #=...' + { + PUGI__SKIPWS(); // Eat any whitespace. + + if (*s == '"' || *s == '\'') // '<... #="...' + { + ch = *s; // Save quote char to avoid breaking on "''" -or- '""'. + ++s; // Step over the quote. + a->value = s; // Save the offset. + + s = strconv_attribute(s, ch); + + if (!s) PUGI__THROW_ERROR(status_bad_attribute, a->value); + + // After this line the loop continues from the start; + // Whitespaces, / and > are ok, symbols and EOF are wrong, + // everything else will be detected + if (PUGI__IS_CHARTYPE(*s, ct_start_symbol)) PUGI__THROW_ERROR(status_bad_attribute, s); + } + else PUGI__THROW_ERROR(status_bad_attribute, s); + } + else PUGI__THROW_ERROR(status_bad_attribute, s); + } + else if (*s == '/') + { + ++s; + + if (*s == '>') + { + PUGI__POPNODE(); + s++; + break; + } + else if (*s == 0 && endch == '>') + { + PUGI__POPNODE(); + break; + } + else PUGI__THROW_ERROR(status_bad_start_element, s); + } + else if (*s == '>') + { + ++s; + + break; + } + else if (*s == 0 && endch == '>') + { + break; + } + else PUGI__THROW_ERROR(status_bad_start_element, s); + } + + // !!! + } + else if (ch == '/') // '<#.../' + { + if (!PUGI__ENDSWITH(*s, '>')) PUGI__THROW_ERROR(status_bad_start_element, s); + + PUGI__POPNODE(); // Pop. + + s += (*s == '>'); + } + else if (ch == 0) + { + // we stepped over null terminator, backtrack & handle closing tag + --s; + + if (endch != '>') PUGI__THROW_ERROR(status_bad_start_element, s); + } + else PUGI__THROW_ERROR(status_bad_start_element, s); + } + else if (*s == '/') + { + ++s; + + mark = s; + + char_t* name = cursor->name; + if (!name) PUGI__THROW_ERROR(status_end_element_mismatch, mark); + + while (PUGI__IS_CHARTYPE(*s, ct_symbol)) + { + if (*s++ != *name++) PUGI__THROW_ERROR(status_end_element_mismatch, mark); + } + + if (*name) + { + if (*s == 0 && name[0] == endch && name[1] == 0) PUGI__THROW_ERROR(status_bad_end_element, s); + else PUGI__THROW_ERROR(status_end_element_mismatch, mark); + } + + PUGI__POPNODE(); // Pop. + + PUGI__SKIPWS(); + + if (*s == 0) + { + if (endch != '>') PUGI__THROW_ERROR(status_bad_end_element, s); + } + else + { + if (*s != '>') PUGI__THROW_ERROR(status_bad_end_element, s); + ++s; + } + } + else if (*s == '?') // 'first_child) continue; + } + } + + if (!PUGI__OPTSET(parse_trim_pcdata)) + s = mark; + + if (cursor->parent || PUGI__OPTSET(parse_fragment)) + { + if (PUGI__OPTSET(parse_embed_pcdata) && cursor->parent && !cursor->first_child && !cursor->value) + { + cursor->value = s; // Save the offset. + } + else + { + PUGI__PUSHNODE(node_pcdata); // Append a new node on the tree. + + cursor->value = s; // Save the offset. + + PUGI__POPNODE(); // Pop since this is a standalone. + } + + s = strconv_pcdata(s); + + if (!*s) break; + } + else + { + PUGI__SCANFOR(*s == '<'); // '...<' + if (!*s) break; + + ++s; + } + + // We're after '<' + goto LOC_TAG; + } + } + + // check that last tag is closed + if (cursor != root) PUGI__THROW_ERROR(status_end_element_mismatch, s); + + return s; + } + + #ifdef PUGIXML_WCHAR_MODE + static char_t* parse_skip_bom(char_t* s) + { + unsigned int bom = 0xfeff; + return (s[0] == static_cast(bom)) ? s + 1 : s; + } + #else + static char_t* parse_skip_bom(char_t* s) + { + return (s[0] == '\xef' && s[1] == '\xbb' && s[2] == '\xbf') ? s + 3 : s; + } + #endif + + static bool has_element_node_siblings(xml_node_struct* node) + { + while (node) + { + if (PUGI__NODETYPE(node) == node_element) return true; + + node = node->next_sibling; + } + + return false; + } + + static xml_parse_result parse(char_t* buffer, size_t length, xml_document_struct* xmldoc, xml_node_struct* root, unsigned int optmsk) + { + // early-out for empty documents + if (length == 0) + return make_parse_result(PUGI__OPTSET(parse_fragment) ? status_ok : status_no_document_element); + + // get last child of the root before parsing + xml_node_struct* last_root_child = root->first_child ? root->first_child->prev_sibling_c + 0 : 0; + + // create parser on stack + xml_parser parser(static_cast(xmldoc)); + + // save last character and make buffer zero-terminated (speeds up parsing) + char_t endch = buffer[length - 1]; + buffer[length - 1] = 0; + + // skip BOM to make sure it does not end up as part of parse output + char_t* buffer_data = parse_skip_bom(buffer); + + // perform actual parsing + parser.parse_tree(buffer_data, root, optmsk, endch); + + xml_parse_result result = make_parse_result(parser.error_status, parser.error_offset ? parser.error_offset - buffer : 0); + assert(result.offset >= 0 && static_cast(result.offset) <= length); + + if (result) + { + // since we removed last character, we have to handle the only possible false positive (stray <) + if (endch == '<') + return make_parse_result(status_unrecognized_tag, length - 1); + + // check if there are any element nodes parsed + xml_node_struct* first_root_child_parsed = last_root_child ? last_root_child->next_sibling + 0 : root->first_child+ 0; + + if (!PUGI__OPTSET(parse_fragment) && !has_element_node_siblings(first_root_child_parsed)) + return make_parse_result(status_no_document_element, length - 1); + } + else + { + // roll back offset if it occurs on a null terminator in the source buffer + if (result.offset > 0 && static_cast(result.offset) == length - 1 && endch == 0) + result.offset--; + } + + return result; + } + }; + + // Output facilities + PUGI__FN xml_encoding get_write_native_encoding() + { + #ifdef PUGIXML_WCHAR_MODE + return get_wchar_encoding(); + #else + return encoding_utf8; + #endif + } + + PUGI__FN xml_encoding get_write_encoding(xml_encoding encoding) + { + // replace wchar encoding with utf implementation + if (encoding == encoding_wchar) return get_wchar_encoding(); + + // replace utf16 encoding with utf16 with specific endianness + if (encoding == encoding_utf16) return is_little_endian() ? encoding_utf16_le : encoding_utf16_be; + + // replace utf32 encoding with utf32 with specific endianness + if (encoding == encoding_utf32) return is_little_endian() ? encoding_utf32_le : encoding_utf32_be; + + // only do autodetection if no explicit encoding is requested + if (encoding != encoding_auto) return encoding; + + // assume utf8 encoding + return encoding_utf8; + } + + template PUGI__FN size_t convert_buffer_output_generic(typename T::value_type dest, const char_t* data, size_t length, D, T) + { + PUGI__STATIC_ASSERT(sizeof(char_t) == sizeof(typename D::type)); + + typename T::value_type end = D::process(reinterpret_cast(data), length, dest, T()); + + return static_cast(end - dest) * sizeof(*dest); + } + + template PUGI__FN size_t convert_buffer_output_generic(typename T::value_type dest, const char_t* data, size_t length, D, T, bool opt_swap) + { + PUGI__STATIC_ASSERT(sizeof(char_t) == sizeof(typename D::type)); + + typename T::value_type end = D::process(reinterpret_cast(data), length, dest, T()); + + if (opt_swap) + { + for (typename T::value_type i = dest; i != end; ++i) + *i = endian_swap(*i); + } + + return static_cast(end - dest) * sizeof(*dest); + } + +#ifdef PUGIXML_WCHAR_MODE + PUGI__FN size_t get_valid_length(const char_t* data, size_t length) + { + if (length < 1) return 0; + + // discard last character if it's the lead of a surrogate pair + return (sizeof(wchar_t) == 2 && static_cast(static_cast(data[length - 1]) - 0xD800) < 0x400) ? length - 1 : length; + } + + PUGI__FN size_t convert_buffer_output(char_t* r_char, uint8_t* r_u8, uint16_t* r_u16, uint32_t* r_u32, const char_t* data, size_t length, xml_encoding encoding) + { + // only endian-swapping is required + if (need_endian_swap_utf(encoding, get_wchar_encoding())) + { + convert_wchar_endian_swap(r_char, data, length); + + return length * sizeof(char_t); + } + + // convert to utf8 + if (encoding == encoding_utf8) + return convert_buffer_output_generic(r_u8, data, length, wchar_decoder(), utf8_writer()); + + // convert to utf16 + if (encoding == encoding_utf16_be || encoding == encoding_utf16_le) + { + xml_encoding native_encoding = is_little_endian() ? encoding_utf16_le : encoding_utf16_be; + + return convert_buffer_output_generic(r_u16, data, length, wchar_decoder(), utf16_writer(), native_encoding != encoding); + } + + // convert to utf32 + if (encoding == encoding_utf32_be || encoding == encoding_utf32_le) + { + xml_encoding native_encoding = is_little_endian() ? encoding_utf32_le : encoding_utf32_be; + + return convert_buffer_output_generic(r_u32, data, length, wchar_decoder(), utf32_writer(), native_encoding != encoding); + } + + // convert to latin1 + if (encoding == encoding_latin1) + return convert_buffer_output_generic(r_u8, data, length, wchar_decoder(), latin1_writer()); + + assert(false && "Invalid encoding"); // unreachable + return 0; + } +#else + PUGI__FN size_t get_valid_length(const char_t* data, size_t length) + { + if (length < 5) return 0; + + for (size_t i = 1; i <= 4; ++i) + { + uint8_t ch = static_cast(data[length - i]); + + // either a standalone character or a leading one + if ((ch & 0xc0) != 0x80) return length - i; + } + + // there are four non-leading characters at the end, sequence tail is broken so might as well process the whole chunk + return length; + } + + PUGI__FN size_t convert_buffer_output(char_t* /* r_char */, uint8_t* r_u8, uint16_t* r_u16, uint32_t* r_u32, const char_t* data, size_t length, xml_encoding encoding) + { + if (encoding == encoding_utf16_be || encoding == encoding_utf16_le) + { + xml_encoding native_encoding = is_little_endian() ? encoding_utf16_le : encoding_utf16_be; + + return convert_buffer_output_generic(r_u16, data, length, utf8_decoder(), utf16_writer(), native_encoding != encoding); + } + + if (encoding == encoding_utf32_be || encoding == encoding_utf32_le) + { + xml_encoding native_encoding = is_little_endian() ? encoding_utf32_le : encoding_utf32_be; + + return convert_buffer_output_generic(r_u32, data, length, utf8_decoder(), utf32_writer(), native_encoding != encoding); + } + + if (encoding == encoding_latin1) + return convert_buffer_output_generic(r_u8, data, length, utf8_decoder(), latin1_writer()); + + assert(false && "Invalid encoding"); // unreachable + return 0; + } +#endif + + class xml_buffered_writer + { + xml_buffered_writer(const xml_buffered_writer&); + xml_buffered_writer& operator=(const xml_buffered_writer&); + + public: + xml_buffered_writer(xml_writer& writer_, xml_encoding user_encoding): writer(writer_), bufsize(0), encoding(get_write_encoding(user_encoding)) + { + PUGI__STATIC_ASSERT(bufcapacity >= 8); + } + + size_t flush() + { + flush(buffer, bufsize); + bufsize = 0; + return 0; + } + + void flush(const char_t* data, size_t size) + { + if (size == 0) return; + + // fast path, just write data + if (encoding == get_write_native_encoding()) + writer.write(data, size * sizeof(char_t)); + else + { + // convert chunk + size_t result = convert_buffer_output(scratch.data_char, scratch.data_u8, scratch.data_u16, scratch.data_u32, data, size, encoding); + assert(result <= sizeof(scratch)); + + // write data + writer.write(scratch.data_u8, result); + } + } + + void write_direct(const char_t* data, size_t length) + { + // flush the remaining buffer contents + flush(); + + // handle large chunks + if (length > bufcapacity) + { + if (encoding == get_write_native_encoding()) + { + // fast path, can just write data chunk + writer.write(data, length * sizeof(char_t)); + return; + } + + // need to convert in suitable chunks + while (length > bufcapacity) + { + // get chunk size by selecting such number of characters that are guaranteed to fit into scratch buffer + // and form a complete codepoint sequence (i.e. discard start of last codepoint if necessary) + size_t chunk_size = get_valid_length(data, bufcapacity); + assert(chunk_size); + + // convert chunk and write + flush(data, chunk_size); + + // iterate + data += chunk_size; + length -= chunk_size; + } + + // small tail is copied below + bufsize = 0; + } + + memcpy(buffer + bufsize, data, length * sizeof(char_t)); + bufsize += length; + } + + void write_buffer(const char_t* data, size_t length) + { + size_t offset = bufsize; + + if (offset + length <= bufcapacity) + { + memcpy(buffer + offset, data, length * sizeof(char_t)); + bufsize = offset + length; + } + else + { + write_direct(data, length); + } + } + + void write_string(const char_t* data) + { + // write the part of the string that fits in the buffer + size_t offset = bufsize; + + while (*data && offset < bufcapacity) + buffer[offset++] = *data++; + + // write the rest + if (offset < bufcapacity) + { + bufsize = offset; + } + else + { + // backtrack a bit if we have split the codepoint + size_t length = offset - bufsize; + size_t extra = length - get_valid_length(data - length, length); + + bufsize = offset - extra; + + write_direct(data - extra, strlength(data) + extra); + } + } + + void write(char_t d0) + { + size_t offset = bufsize; + if (offset > bufcapacity - 1) offset = flush(); + + buffer[offset + 0] = d0; + bufsize = offset + 1; + } + + void write(char_t d0, char_t d1) + { + size_t offset = bufsize; + if (offset > bufcapacity - 2) offset = flush(); + + buffer[offset + 0] = d0; + buffer[offset + 1] = d1; + bufsize = offset + 2; + } + + void write(char_t d0, char_t d1, char_t d2) + { + size_t offset = bufsize; + if (offset > bufcapacity - 3) offset = flush(); + + buffer[offset + 0] = d0; + buffer[offset + 1] = d1; + buffer[offset + 2] = d2; + bufsize = offset + 3; + } + + void write(char_t d0, char_t d1, char_t d2, char_t d3) + { + size_t offset = bufsize; + if (offset > bufcapacity - 4) offset = flush(); + + buffer[offset + 0] = d0; + buffer[offset + 1] = d1; + buffer[offset + 2] = d2; + buffer[offset + 3] = d3; + bufsize = offset + 4; + } + + void write(char_t d0, char_t d1, char_t d2, char_t d3, char_t d4) + { + size_t offset = bufsize; + if (offset > bufcapacity - 5) offset = flush(); + + buffer[offset + 0] = d0; + buffer[offset + 1] = d1; + buffer[offset + 2] = d2; + buffer[offset + 3] = d3; + buffer[offset + 4] = d4; + bufsize = offset + 5; + } + + void write(char_t d0, char_t d1, char_t d2, char_t d3, char_t d4, char_t d5) + { + size_t offset = bufsize; + if (offset > bufcapacity - 6) offset = flush(); + + buffer[offset + 0] = d0; + buffer[offset + 1] = d1; + buffer[offset + 2] = d2; + buffer[offset + 3] = d3; + buffer[offset + 4] = d4; + buffer[offset + 5] = d5; + bufsize = offset + 6; + } + + // utf8 maximum expansion: x4 (-> utf32) + // utf16 maximum expansion: x2 (-> utf32) + // utf32 maximum expansion: x1 + enum + { + bufcapacitybytes = + #ifdef PUGIXML_MEMORY_OUTPUT_STACK + PUGIXML_MEMORY_OUTPUT_STACK + #else + 10240 + #endif + , + bufcapacity = bufcapacitybytes / (sizeof(char_t) + 4) + }; + + char_t buffer[bufcapacity]; + + union + { + uint8_t data_u8[4 * bufcapacity]; + uint16_t data_u16[2 * bufcapacity]; + uint32_t data_u32[bufcapacity]; + char_t data_char[bufcapacity]; + } scratch; + + xml_writer& writer; + size_t bufsize; + xml_encoding encoding; + }; + + PUGI__FN void text_output_escaped(xml_buffered_writer& writer, const char_t* s, chartypex_t type, unsigned int flags) + { + while (*s) + { + const char_t* prev = s; + + // While *s is a usual symbol + PUGI__SCANWHILE_UNROLL(!PUGI__IS_CHARTYPEX(ss, type)); + + writer.write_buffer(prev, static_cast(s - prev)); + + switch (*s) + { + case 0: break; + case '&': + writer.write('&', 'a', 'm', 'p', ';'); + ++s; + break; + case '<': + writer.write('&', 'l', 't', ';'); + ++s; + break; + case '>': + writer.write('&', 'g', 't', ';'); + ++s; + break; + case '"': + if (flags & format_attribute_single_quote) + writer.write('"'); + else + writer.write('&', 'q', 'u', 'o', 't', ';'); + ++s; + break; + case '\'': + if (flags & format_attribute_single_quote) + writer.write('&', 'a', 'p', 'o', 's', ';'); + else + writer.write('\''); + ++s; + break; + default: // s is not a usual symbol + { + unsigned int ch = static_cast(*s++); + assert(ch < 32); + + if (!(flags & format_skip_control_chars)) + writer.write('&', '#', static_cast((ch / 10) + '0'), static_cast((ch % 10) + '0'), ';'); + } + } + } + } + + PUGI__FN void text_output(xml_buffered_writer& writer, const char_t* s, chartypex_t type, unsigned int flags) + { + if (flags & format_no_escapes) + writer.write_string(s); + else + text_output_escaped(writer, s, type, flags); + } + + PUGI__FN void text_output_cdata(xml_buffered_writer& writer, const char_t* s) + { + do + { + writer.write('<', '!', '[', 'C', 'D'); + writer.write('A', 'T', 'A', '['); + + const char_t* prev = s; + + // look for ]]> sequence - we can't output it as is since it terminates CDATA + while (*s && !(s[0] == ']' && s[1] == ']' && s[2] == '>')) ++s; + + // skip ]] if we stopped at ]]>, > will go to the next CDATA section + if (*s) s += 2; + + writer.write_buffer(prev, static_cast(s - prev)); + + writer.write(']', ']', '>'); + } + while (*s); + } + + PUGI__FN void text_output_indent(xml_buffered_writer& writer, const char_t* indent, size_t indent_length, unsigned int depth) + { + switch (indent_length) + { + case 1: + { + for (unsigned int i = 0; i < depth; ++i) + writer.write(indent[0]); + break; + } + + case 2: + { + for (unsigned int i = 0; i < depth; ++i) + writer.write(indent[0], indent[1]); + break; + } + + case 3: + { + for (unsigned int i = 0; i < depth; ++i) + writer.write(indent[0], indent[1], indent[2]); + break; + } + + case 4: + { + for (unsigned int i = 0; i < depth; ++i) + writer.write(indent[0], indent[1], indent[2], indent[3]); + break; + } + + default: + { + for (unsigned int i = 0; i < depth; ++i) + writer.write_buffer(indent, indent_length); + } + } + } + + PUGI__FN void node_output_comment(xml_buffered_writer& writer, const char_t* s) + { + writer.write('<', '!', '-', '-'); + + while (*s) + { + const char_t* prev = s; + + // look for -\0 or -- sequence - we can't output it since -- is illegal in comment body + while (*s && !(s[0] == '-' && (s[1] == '-' || s[1] == 0))) ++s; + + writer.write_buffer(prev, static_cast(s - prev)); + + if (*s) + { + assert(*s == '-'); + + writer.write('-', ' '); + ++s; + } + } + + writer.write('-', '-', '>'); + } + + PUGI__FN void node_output_pi_value(xml_buffered_writer& writer, const char_t* s) + { + while (*s) + { + const char_t* prev = s; + + // look for ?> sequence - we can't output it since ?> terminates PI + while (*s && !(s[0] == '?' && s[1] == '>')) ++s; + + writer.write_buffer(prev, static_cast(s - prev)); + + if (*s) + { + assert(s[0] == '?' && s[1] == '>'); + + writer.write('?', ' ', '>'); + s += 2; + } + } + } + + PUGI__FN void node_output_attributes(xml_buffered_writer& writer, xml_node_struct* node, const char_t* indent, size_t indent_length, unsigned int flags, unsigned int depth) + { + const char_t* default_name = PUGIXML_TEXT(":anonymous"); + const char_t enquotation_char = (flags & format_attribute_single_quote) ? '\'' : '"'; + + for (xml_attribute_struct* a = node->first_attribute; a; a = a->next_attribute) + { + if ((flags & (format_indent_attributes | format_raw)) == format_indent_attributes) + { + writer.write('\n'); + + text_output_indent(writer, indent, indent_length, depth + 1); + } + else + { + writer.write(' '); + } + + writer.write_string(a->name ? a->name + 0 : default_name); + writer.write('=', enquotation_char); + + if (a->value) + text_output(writer, a->value, ctx_special_attr, flags); + + writer.write(enquotation_char); + } + } + + PUGI__FN bool node_output_start(xml_buffered_writer& writer, xml_node_struct* node, const char_t* indent, size_t indent_length, unsigned int flags, unsigned int depth) + { + const char_t* default_name = PUGIXML_TEXT(":anonymous"); + const char_t* name = node->name ? node->name + 0 : default_name; + + writer.write('<'); + writer.write_string(name); + + if (node->first_attribute) + node_output_attributes(writer, node, indent, indent_length, flags, depth); + + // element nodes can have value if parse_embed_pcdata was used + if (!node->value) + { + if (!node->first_child) + { + if (flags & format_no_empty_element_tags) + { + writer.write('>', '<', '/'); + writer.write_string(name); + writer.write('>'); + + return false; + } + else + { + if ((flags & format_raw) == 0) + writer.write(' '); + + writer.write('/', '>'); + + return false; + } + } + else + { + writer.write('>'); + + return true; + } + } + else + { + writer.write('>'); + + text_output(writer, node->value, ctx_special_pcdata, flags); + + if (!node->first_child) + { + writer.write('<', '/'); + writer.write_string(name); + writer.write('>'); + + return false; + } + else + { + return true; + } + } + } + + PUGI__FN void node_output_end(xml_buffered_writer& writer, xml_node_struct* node) + { + const char_t* default_name = PUGIXML_TEXT(":anonymous"); + const char_t* name = node->name ? node->name + 0 : default_name; + + writer.write('<', '/'); + writer.write_string(name); + writer.write('>'); + } + + PUGI__FN void node_output_simple(xml_buffered_writer& writer, xml_node_struct* node, unsigned int flags) + { + const char_t* default_name = PUGIXML_TEXT(":anonymous"); + + switch (PUGI__NODETYPE(node)) + { + case node_pcdata: + text_output(writer, node->value ? node->value + 0 : PUGIXML_TEXT(""), ctx_special_pcdata, flags); + break; + + case node_cdata: + text_output_cdata(writer, node->value ? node->value + 0 : PUGIXML_TEXT("")); + break; + + case node_comment: + node_output_comment(writer, node->value ? node->value + 0 : PUGIXML_TEXT("")); + break; + + case node_pi: + writer.write('<', '?'); + writer.write_string(node->name ? node->name + 0 : default_name); + + if (node->value) + { + writer.write(' '); + node_output_pi_value(writer, node->value); + } + + writer.write('?', '>'); + break; + + case node_declaration: + writer.write('<', '?'); + writer.write_string(node->name ? node->name + 0 : default_name); + node_output_attributes(writer, node, PUGIXML_TEXT(""), 0, flags | format_raw, 0); + writer.write('?', '>'); + break; + + case node_doctype: + writer.write('<', '!', 'D', 'O', 'C'); + writer.write('T', 'Y', 'P', 'E'); + + if (node->value) + { + writer.write(' '); + writer.write_string(node->value); + } + + writer.write('>'); + break; + + default: + assert(false && "Invalid node type"); // unreachable + } + } + + enum indent_flags_t + { + indent_newline = 1, + indent_indent = 2 + }; + + PUGI__FN void node_output(xml_buffered_writer& writer, xml_node_struct* root, const char_t* indent, unsigned int flags, unsigned int depth) + { + size_t indent_length = ((flags & (format_indent | format_indent_attributes)) && (flags & format_raw) == 0) ? strlength(indent) : 0; + unsigned int indent_flags = indent_indent; + + xml_node_struct* node = root; + + do + { + assert(node); + + // begin writing current node + if (PUGI__NODETYPE(node) == node_pcdata || PUGI__NODETYPE(node) == node_cdata) + { + node_output_simple(writer, node, flags); + + indent_flags = 0; + } + else + { + if ((indent_flags & indent_newline) && (flags & format_raw) == 0) + writer.write('\n'); + + if ((indent_flags & indent_indent) && indent_length) + text_output_indent(writer, indent, indent_length, depth); + + if (PUGI__NODETYPE(node) == node_element) + { + indent_flags = indent_newline | indent_indent; + + if (node_output_start(writer, node, indent, indent_length, flags, depth)) + { + // element nodes can have value if parse_embed_pcdata was used + if (node->value) + indent_flags = 0; + + node = node->first_child; + depth++; + continue; + } + } + else if (PUGI__NODETYPE(node) == node_document) + { + indent_flags = indent_indent; + + if (node->first_child) + { + node = node->first_child; + continue; + } + } + else + { + node_output_simple(writer, node, flags); + + indent_flags = indent_newline | indent_indent; + } + } + + // continue to the next node + while (node != root) + { + if (node->next_sibling) + { + node = node->next_sibling; + break; + } + + node = node->parent; + + // write closing node + if (PUGI__NODETYPE(node) == node_element) + { + depth--; + + if ((indent_flags & indent_newline) && (flags & format_raw) == 0) + writer.write('\n'); + + if ((indent_flags & indent_indent) && indent_length) + text_output_indent(writer, indent, indent_length, depth); + + node_output_end(writer, node); + + indent_flags = indent_newline | indent_indent; + } + } + } + while (node != root); + + if ((indent_flags & indent_newline) && (flags & format_raw) == 0) + writer.write('\n'); + } + + PUGI__FN bool has_declaration(xml_node_struct* node) + { + for (xml_node_struct* child = node->first_child; child; child = child->next_sibling) + { + xml_node_type type = PUGI__NODETYPE(child); + + if (type == node_declaration) return true; + if (type == node_element) return false; + } + + return false; + } + + PUGI__FN bool is_attribute_of(xml_attribute_struct* attr, xml_node_struct* node) + { + for (xml_attribute_struct* a = node->first_attribute; a; a = a->next_attribute) + if (a == attr) + return true; + + return false; + } + + PUGI__FN bool allow_insert_attribute(xml_node_type parent) + { + return parent == node_element || parent == node_declaration; + } + + PUGI__FN bool allow_insert_child(xml_node_type parent, xml_node_type child) + { + if (parent != node_document && parent != node_element) return false; + if (child == node_document || child == node_null) return false; + if (parent != node_document && (child == node_declaration || child == node_doctype)) return false; + + return true; + } + + PUGI__FN bool allow_move(xml_node parent, xml_node child) + { + // check that child can be a child of parent + if (!allow_insert_child(parent.type(), child.type())) + return false; + + // check that node is not moved between documents + if (parent.root() != child.root()) + return false; + + // check that new parent is not in the child subtree + xml_node cur = parent; + + while (cur) + { + if (cur == child) + return false; + + cur = cur.parent(); + } + + return true; + } + + template + PUGI__FN void node_copy_string(String& dest, Header& header, uintptr_t header_mask, char_t* source, Header& source_header, xml_allocator* alloc) + { + assert(!dest && (header & header_mask) == 0); + + if (source) + { + if (alloc && (source_header & header_mask) == 0) + { + dest = source; + + // since strcpy_insitu can reuse document buffer memory we need to mark both source and dest as shared + header |= xml_memory_page_contents_shared_mask; + source_header |= xml_memory_page_contents_shared_mask; + } + else + strcpy_insitu(dest, header, header_mask, source, strlength(source)); + } + } + + PUGI__FN void node_copy_contents(xml_node_struct* dn, xml_node_struct* sn, xml_allocator* shared_alloc) + { + node_copy_string(dn->name, dn->header, xml_memory_page_name_allocated_mask, sn->name, sn->header, shared_alloc); + node_copy_string(dn->value, dn->header, xml_memory_page_value_allocated_mask, sn->value, sn->header, shared_alloc); + + for (xml_attribute_struct* sa = sn->first_attribute; sa; sa = sa->next_attribute) + { + xml_attribute_struct* da = append_new_attribute(dn, get_allocator(dn)); + + if (da) + { + node_copy_string(da->name, da->header, xml_memory_page_name_allocated_mask, sa->name, sa->header, shared_alloc); + node_copy_string(da->value, da->header, xml_memory_page_value_allocated_mask, sa->value, sa->header, shared_alloc); + } + } + } + + PUGI__FN void node_copy_tree(xml_node_struct* dn, xml_node_struct* sn) + { + xml_allocator& alloc = get_allocator(dn); + xml_allocator* shared_alloc = (&alloc == &get_allocator(sn)) ? &alloc : 0; + + node_copy_contents(dn, sn, shared_alloc); + + xml_node_struct* dit = dn; + xml_node_struct* sit = sn->first_child; + + while (sit && sit != sn) + { + // when a tree is copied into one of the descendants, we need to skip that subtree to avoid an infinite loop + if (sit != dn) + { + xml_node_struct* copy = append_new_node(dit, alloc, PUGI__NODETYPE(sit)); + + if (copy) + { + node_copy_contents(copy, sit, shared_alloc); + + if (sit->first_child) + { + dit = copy; + sit = sit->first_child; + continue; + } + } + } + + // continue to the next node + do + { + if (sit->next_sibling) + { + sit = sit->next_sibling; + break; + } + + sit = sit->parent; + dit = dit->parent; + } + while (sit != sn); + } + } + + PUGI__FN void node_copy_attribute(xml_attribute_struct* da, xml_attribute_struct* sa) + { + xml_allocator& alloc = get_allocator(da); + xml_allocator* shared_alloc = (&alloc == &get_allocator(sa)) ? &alloc : 0; + + node_copy_string(da->name, da->header, xml_memory_page_name_allocated_mask, sa->name, sa->header, shared_alloc); + node_copy_string(da->value, da->header, xml_memory_page_value_allocated_mask, sa->value, sa->header, shared_alloc); + } + + inline bool is_text_node(xml_node_struct* node) + { + xml_node_type type = PUGI__NODETYPE(node); + + return type == node_pcdata || type == node_cdata; + } + + // get value with conversion functions + template PUGI__FN PUGI__UNSIGNED_OVERFLOW U string_to_integer(const char_t* value, U minv, U maxv) + { + U result = 0; + const char_t* s = value; + + while (PUGI__IS_CHARTYPE(*s, ct_space)) + s++; + + bool negative = (*s == '-'); + + s += (*s == '+' || *s == '-'); + + bool overflow = false; + + if (s[0] == '0' && (s[1] | ' ') == 'x') + { + s += 2; + + // since overflow detection relies on length of the sequence skip leading zeros + while (*s == '0') + s++; + + const char_t* start = s; + + for (;;) + { + if (static_cast(*s - '0') < 10) + result = result * 16 + (*s - '0'); + else if (static_cast((*s | ' ') - 'a') < 6) + result = result * 16 + ((*s | ' ') - 'a' + 10); + else + break; + + s++; + } + + size_t digits = static_cast(s - start); + + overflow = digits > sizeof(U) * 2; + } + else + { + // since overflow detection relies on length of the sequence skip leading zeros + while (*s == '0') + s++; + + const char_t* start = s; + + for (;;) + { + if (static_cast(*s - '0') < 10) + result = result * 10 + (*s - '0'); + else + break; + + s++; + } + + size_t digits = static_cast(s - start); + + PUGI__STATIC_ASSERT(sizeof(U) == 8 || sizeof(U) == 4 || sizeof(U) == 2); + + const size_t max_digits10 = sizeof(U) == 8 ? 20 : sizeof(U) == 4 ? 10 : 5; + const char_t max_lead = sizeof(U) == 8 ? '1' : sizeof(U) == 4 ? '4' : '6'; + const size_t high_bit = sizeof(U) * 8 - 1; + + overflow = digits >= max_digits10 && !(digits == max_digits10 && (*start < max_lead || (*start == max_lead && result >> high_bit))); + } + + if (negative) + { + // Workaround for crayc++ CC-3059: Expected no overflow in routine. + #ifdef _CRAYC + return (overflow || result > ~minv + 1) ? minv : ~result + 1; + #else + return (overflow || result > 0 - minv) ? minv : 0 - result; + #endif + } + else + return (overflow || result > maxv) ? maxv : result; + } + + PUGI__FN int get_value_int(const char_t* value) + { + return string_to_integer(value, static_cast(INT_MIN), INT_MAX); + } + + PUGI__FN unsigned int get_value_uint(const char_t* value) + { + return string_to_integer(value, 0, UINT_MAX); + } + + PUGI__FN double get_value_double(const char_t* value) + { + #ifdef PUGIXML_WCHAR_MODE + return wcstod(value, 0); + #else + return strtod(value, 0); + #endif + } + + PUGI__FN float get_value_float(const char_t* value) + { + #ifdef PUGIXML_WCHAR_MODE + return static_cast(wcstod(value, 0)); + #else + return static_cast(strtod(value, 0)); + #endif + } + + PUGI__FN bool get_value_bool(const char_t* value) + { + // only look at first char + char_t first = *value; + + // 1*, t* (true), T* (True), y* (yes), Y* (YES) + return (first == '1' || first == 't' || first == 'T' || first == 'y' || first == 'Y'); + } + +#ifdef PUGIXML_HAS_LONG_LONG + PUGI__FN long long get_value_llong(const char_t* value) + { + return string_to_integer(value, static_cast(LLONG_MIN), LLONG_MAX); + } + + PUGI__FN unsigned long long get_value_ullong(const char_t* value) + { + return string_to_integer(value, 0, ULLONG_MAX); + } +#endif + + template PUGI__FN PUGI__UNSIGNED_OVERFLOW char_t* integer_to_string(char_t* begin, char_t* end, U value, bool negative) + { + char_t* result = end - 1; + U rest = negative ? 0 - value : value; + + do + { + *result-- = static_cast('0' + (rest % 10)); + rest /= 10; + } + while (rest); + + assert(result >= begin); + (void)begin; + + *result = '-'; + + return result + !negative; + } + + // set value with conversion functions + template + PUGI__FN bool set_value_ascii(String& dest, Header& header, uintptr_t header_mask, char* buf) + { + #ifdef PUGIXML_WCHAR_MODE + char_t wbuf[128]; + assert(strlen(buf) < sizeof(wbuf) / sizeof(wbuf[0])); + + size_t offset = 0; + for (; buf[offset]; ++offset) wbuf[offset] = buf[offset]; + + return strcpy_insitu(dest, header, header_mask, wbuf, offset); + #else + return strcpy_insitu(dest, header, header_mask, buf, strlen(buf)); + #endif + } + + template + PUGI__FN bool set_value_integer(String& dest, Header& header, uintptr_t header_mask, U value, bool negative) + { + char_t buf[64]; + char_t* end = buf + sizeof(buf) / sizeof(buf[0]); + char_t* begin = integer_to_string(buf, end, value, negative); + + return strcpy_insitu(dest, header, header_mask, begin, end - begin); + } + + template + PUGI__FN bool set_value_convert(String& dest, Header& header, uintptr_t header_mask, float value) + { + char buf[128]; + PUGI__SNPRINTF(buf, "%.9g", double(value)); + + return set_value_ascii(dest, header, header_mask, buf); + } + + template + PUGI__FN bool set_value_convert(String& dest, Header& header, uintptr_t header_mask, double value) + { + char buf[128]; + PUGI__SNPRINTF(buf, "%.17g", value); + + return set_value_ascii(dest, header, header_mask, buf); + } + + template + PUGI__FN bool set_value_bool(String& dest, Header& header, uintptr_t header_mask, bool value) + { + return strcpy_insitu(dest, header, header_mask, value ? PUGIXML_TEXT("true") : PUGIXML_TEXT("false"), value ? 4 : 5); + } + + PUGI__FN xml_parse_result load_buffer_impl(xml_document_struct* doc, xml_node_struct* root, void* contents, size_t size, unsigned int options, xml_encoding encoding, bool is_mutable, bool own, char_t** out_buffer) + { + // check input buffer + if (!contents && size) return make_parse_result(status_io_error); + + // get actual encoding + xml_encoding buffer_encoding = impl::get_buffer_encoding(encoding, contents, size); + + // get private buffer + char_t* buffer = 0; + size_t length = 0; + + // coverity[var_deref_model] + if (!impl::convert_buffer(buffer, length, buffer_encoding, contents, size, is_mutable)) return impl::make_parse_result(status_out_of_memory); + + // delete original buffer if we performed a conversion + if (own && buffer != contents && contents) impl::xml_memory::deallocate(contents); + + // grab onto buffer if it's our buffer, user is responsible for deallocating contents himself + if (own || buffer != contents) *out_buffer = buffer; + + // store buffer for offset_debug + doc->buffer = buffer; + + // parse + xml_parse_result res = impl::xml_parser::parse(buffer, length, doc, root, options); + + // remember encoding + res.encoding = buffer_encoding; + + return res; + } + + // we need to get length of entire file to load it in memory; the only (relatively) sane way to do it is via seek/tell trick + PUGI__FN xml_parse_status get_file_size(FILE* file, size_t& out_result) + { + #if defined(PUGI__MSVC_CRT_VERSION) && PUGI__MSVC_CRT_VERSION >= 1400 && !defined(_WIN32_WCE) + // there are 64-bit versions of fseek/ftell, let's use them + typedef __int64 length_type; + + _fseeki64(file, 0, SEEK_END); + length_type length = _ftelli64(file); + _fseeki64(file, 0, SEEK_SET); + #elif defined(__MINGW32__) && !defined(__NO_MINGW_LFS) && (!defined(__STRICT_ANSI__) || defined(__MINGW64_VERSION_MAJOR)) + // there are 64-bit versions of fseek/ftell, let's use them + typedef off64_t length_type; + + fseeko64(file, 0, SEEK_END); + length_type length = ftello64(file); + fseeko64(file, 0, SEEK_SET); + #else + // if this is a 32-bit OS, long is enough; if this is a unix system, long is 64-bit, which is enough; otherwise we can't do anything anyway. + typedef long length_type; + + fseek(file, 0, SEEK_END); + length_type length = ftell(file); + fseek(file, 0, SEEK_SET); + #endif + + // check for I/O errors + if (length < 0) return status_io_error; + + // check for overflow + size_t result = static_cast(length); + + if (static_cast(result) != length) return status_out_of_memory; + + // finalize + out_result = result; + + return status_ok; + } + + // This function assumes that buffer has extra sizeof(char_t) writable bytes after size + PUGI__FN size_t zero_terminate_buffer(void* buffer, size_t size, xml_encoding encoding) + { + // We only need to zero-terminate if encoding conversion does not do it for us + #ifdef PUGIXML_WCHAR_MODE + xml_encoding wchar_encoding = get_wchar_encoding(); + + if (encoding == wchar_encoding || need_endian_swap_utf(encoding, wchar_encoding)) + { + size_t length = size / sizeof(char_t); + + static_cast(buffer)[length] = 0; + return (length + 1) * sizeof(char_t); + } + #else + if (encoding == encoding_utf8) + { + static_cast(buffer)[size] = 0; + return size + 1; + } + #endif + + return size; + } + + PUGI__FN xml_parse_result load_file_impl(xml_document_struct* doc, FILE* file, unsigned int options, xml_encoding encoding, char_t** out_buffer) + { + if (!file) return make_parse_result(status_file_not_found); + + // get file size (can result in I/O errors) + size_t size = 0; + xml_parse_status size_status = get_file_size(file, size); + if (size_status != status_ok) return make_parse_result(size_status); + + size_t max_suffix_size = sizeof(char_t); + + // allocate buffer for the whole file + char* contents = static_cast(xml_memory::allocate(size + max_suffix_size)); + if (!contents) return make_parse_result(status_out_of_memory); + + // read file in memory + size_t read_size = fread(contents, 1, size, file); + + if (read_size != size) + { + xml_memory::deallocate(contents); + return make_parse_result(status_io_error); + } + + xml_encoding real_encoding = get_buffer_encoding(encoding, contents, size); + + return load_buffer_impl(doc, doc, contents, zero_terminate_buffer(contents, size, real_encoding), options, real_encoding, true, true, out_buffer); + } + + PUGI__FN void close_file(FILE* file) + { + fclose(file); + } + +#ifndef PUGIXML_NO_STL + template struct xml_stream_chunk + { + static xml_stream_chunk* create() + { + void* memory = xml_memory::allocate(sizeof(xml_stream_chunk)); + if (!memory) return 0; + + return new (memory) xml_stream_chunk(); + } + + static void destroy(xml_stream_chunk* chunk) + { + // free chunk chain + while (chunk) + { + xml_stream_chunk* next_ = chunk->next; + + xml_memory::deallocate(chunk); + + chunk = next_; + } + } + + xml_stream_chunk(): next(0), size(0) + { + } + + xml_stream_chunk* next; + size_t size; + + T data[xml_memory_page_size / sizeof(T)]; + }; + + template PUGI__FN xml_parse_status load_stream_data_noseek(std::basic_istream& stream, void** out_buffer, size_t* out_size) + { + auto_deleter > chunks(0, xml_stream_chunk::destroy); + + // read file to a chunk list + size_t total = 0; + xml_stream_chunk* last = 0; + + while (!stream.eof()) + { + // allocate new chunk + xml_stream_chunk* chunk = xml_stream_chunk::create(); + if (!chunk) return status_out_of_memory; + + // append chunk to list + if (last) last = last->next = chunk; + else chunks.data = last = chunk; + + // read data to chunk + stream.read(chunk->data, static_cast(sizeof(chunk->data) / sizeof(T))); + chunk->size = static_cast(stream.gcount()) * sizeof(T); + + // read may set failbit | eofbit in case gcount() is less than read length, so check for other I/O errors + if (stream.bad() || (!stream.eof() && stream.fail())) return status_io_error; + + // guard against huge files (chunk size is small enough to make this overflow check work) + if (total + chunk->size < total) return status_out_of_memory; + total += chunk->size; + } + + size_t max_suffix_size = sizeof(char_t); + + // copy chunk list to a contiguous buffer + char* buffer = static_cast(xml_memory::allocate(total + max_suffix_size)); + if (!buffer) return status_out_of_memory; + + char* write = buffer; + + for (xml_stream_chunk* chunk = chunks.data; chunk; chunk = chunk->next) + { + assert(write + chunk->size <= buffer + total); + memcpy(write, chunk->data, chunk->size); + write += chunk->size; + } + + assert(write == buffer + total); + + // return buffer + *out_buffer = buffer; + *out_size = total; + + return status_ok; + } + + template PUGI__FN xml_parse_status load_stream_data_seek(std::basic_istream& stream, void** out_buffer, size_t* out_size) + { + // get length of remaining data in stream + typename std::basic_istream::pos_type pos = stream.tellg(); + stream.seekg(0, std::ios::end); + std::streamoff length = stream.tellg() - pos; + stream.seekg(pos); + + if (stream.fail() || pos < 0) return status_io_error; + + // guard against huge files + size_t read_length = static_cast(length); + + if (static_cast(read_length) != length || length < 0) return status_out_of_memory; + + size_t max_suffix_size = sizeof(char_t); + + // read stream data into memory (guard against stream exceptions with buffer holder) + auto_deleter buffer(xml_memory::allocate(read_length * sizeof(T) + max_suffix_size), xml_memory::deallocate); + if (!buffer.data) return status_out_of_memory; + + stream.read(static_cast(buffer.data), static_cast(read_length)); + + // read may set failbit | eofbit in case gcount() is less than read_length (i.e. line ending conversion), so check for other I/O errors + if (stream.bad() || (!stream.eof() && stream.fail())) return status_io_error; + + // return buffer + size_t actual_length = static_cast(stream.gcount()); + assert(actual_length <= read_length); + + *out_buffer = buffer.release(); + *out_size = actual_length * sizeof(T); + + return status_ok; + } + + template PUGI__FN xml_parse_result load_stream_impl(xml_document_struct* doc, std::basic_istream& stream, unsigned int options, xml_encoding encoding, char_t** out_buffer) + { + void* buffer = 0; + size_t size = 0; + xml_parse_status status = status_ok; + + // if stream has an error bit set, bail out (otherwise tellg() can fail and we'll clear error bits) + if (stream.fail()) return make_parse_result(status_io_error); + + // load stream to memory (using seek-based implementation if possible, since it's faster and takes less memory) + if (stream.tellg() < 0) + { + stream.clear(); // clear error flags that could be set by a failing tellg + status = load_stream_data_noseek(stream, &buffer, &size); + } + else + status = load_stream_data_seek(stream, &buffer, &size); + + if (status != status_ok) return make_parse_result(status); + + xml_encoding real_encoding = get_buffer_encoding(encoding, buffer, size); + + return load_buffer_impl(doc, doc, buffer, zero_terminate_buffer(buffer, size, real_encoding), options, real_encoding, true, true, out_buffer); + } +#endif + +#if defined(PUGI__MSVC_CRT_VERSION) || defined(__BORLANDC__) || (defined(__MINGW32__) && (!defined(__STRICT_ANSI__) || defined(__MINGW64_VERSION_MAJOR))) + PUGI__FN FILE* open_file_wide(const wchar_t* path, const wchar_t* mode) + { + return _wfopen(path, mode); + } +#else + PUGI__FN char* convert_path_heap(const wchar_t* str) + { + assert(str); + + // first pass: get length in utf8 characters + size_t length = strlength_wide(str); + size_t size = as_utf8_begin(str, length); + + // allocate resulting string + char* result = static_cast(xml_memory::allocate(size + 1)); + if (!result) return 0; + + // second pass: convert to utf8 + as_utf8_end(result, size, str, length); + + // zero-terminate + result[size] = 0; + + return result; + } + + PUGI__FN FILE* open_file_wide(const wchar_t* path, const wchar_t* mode) + { + // there is no standard function to open wide paths, so our best bet is to try utf8 path + char* path_utf8 = convert_path_heap(path); + if (!path_utf8) return 0; + + // convert mode to ASCII (we mirror _wfopen interface) + char mode_ascii[4] = {0}; + for (size_t i = 0; mode[i]; ++i) mode_ascii[i] = static_cast(mode[i]); + + // try to open the utf8 path + FILE* result = fopen(path_utf8, mode_ascii); + + // free dummy buffer + xml_memory::deallocate(path_utf8); + + return result; + } +#endif + + PUGI__FN bool save_file_impl(const xml_document& doc, FILE* file, const char_t* indent, unsigned int flags, xml_encoding encoding) + { + if (!file) return false; + + xml_writer_file writer(file); + doc.save(writer, indent, flags, encoding); + + return ferror(file) == 0; + } + + struct name_null_sentry + { + xml_node_struct* node; + char_t* name; + + name_null_sentry(xml_node_struct* node_): node(node_), name(node_->name) + { + node->name = 0; + } + + ~name_null_sentry() + { + node->name = name; + } + }; +PUGI__NS_END + +namespace pugi +{ + PUGI__FN xml_writer_file::xml_writer_file(void* file_): file(file_) + { + } + + PUGI__FN void xml_writer_file::write(const void* data, size_t size) + { + size_t result = fwrite(data, 1, size, static_cast(file)); + (void)!result; // unfortunately we can't do proper error handling here + } + +#ifndef PUGIXML_NO_STL + PUGI__FN xml_writer_stream::xml_writer_stream(std::basic_ostream >& stream): narrow_stream(&stream), wide_stream(0) + { + } + + PUGI__FN xml_writer_stream::xml_writer_stream(std::basic_ostream >& stream): narrow_stream(0), wide_stream(&stream) + { + } + + PUGI__FN void xml_writer_stream::write(const void* data, size_t size) + { + if (narrow_stream) + { + assert(!wide_stream); + narrow_stream->write(reinterpret_cast(data), static_cast(size)); + } + else + { + assert(wide_stream); + assert(size % sizeof(wchar_t) == 0); + + wide_stream->write(reinterpret_cast(data), static_cast(size / sizeof(wchar_t))); + } + } +#endif + + PUGI__FN xml_tree_walker::xml_tree_walker(): _depth(0) + { + } + + PUGI__FN xml_tree_walker::~xml_tree_walker() + { + } + + PUGI__FN int xml_tree_walker::depth() const + { + return _depth; + } + + PUGI__FN bool xml_tree_walker::begin(xml_node&) + { + return true; + } + + PUGI__FN bool xml_tree_walker::end(xml_node&) + { + return true; + } + + PUGI__FN xml_attribute::xml_attribute(): _attr(0) + { + } + + PUGI__FN xml_attribute::xml_attribute(xml_attribute_struct* attr): _attr(attr) + { + } + + PUGI__FN static void unspecified_bool_xml_attribute(xml_attribute***) + { + } + + PUGI__FN xml_attribute::operator xml_attribute::unspecified_bool_type() const + { + return _attr ? unspecified_bool_xml_attribute : 0; + } + + PUGI__FN bool xml_attribute::operator!() const + { + return !_attr; + } + + PUGI__FN bool xml_attribute::operator==(const xml_attribute& r) const + { + return (_attr == r._attr); + } + + PUGI__FN bool xml_attribute::operator!=(const xml_attribute& r) const + { + return (_attr != r._attr); + } + + PUGI__FN bool xml_attribute::operator<(const xml_attribute& r) const + { + return (_attr < r._attr); + } + + PUGI__FN bool xml_attribute::operator>(const xml_attribute& r) const + { + return (_attr > r._attr); + } + + PUGI__FN bool xml_attribute::operator<=(const xml_attribute& r) const + { + return (_attr <= r._attr); + } + + PUGI__FN bool xml_attribute::operator>=(const xml_attribute& r) const + { + return (_attr >= r._attr); + } + + PUGI__FN xml_attribute xml_attribute::next_attribute() const + { + return _attr ? xml_attribute(_attr->next_attribute) : xml_attribute(); + } + + PUGI__FN xml_attribute xml_attribute::previous_attribute() const + { + return _attr && _attr->prev_attribute_c->next_attribute ? xml_attribute(_attr->prev_attribute_c) : xml_attribute(); + } + + PUGI__FN const char_t* xml_attribute::as_string(const char_t* def) const + { + return (_attr && _attr->value) ? _attr->value + 0 : def; + } + + PUGI__FN int xml_attribute::as_int(int def) const + { + return (_attr && _attr->value) ? impl::get_value_int(_attr->value) : def; + } + + PUGI__FN unsigned int xml_attribute::as_uint(unsigned int def) const + { + return (_attr && _attr->value) ? impl::get_value_uint(_attr->value) : def; + } + + PUGI__FN double xml_attribute::as_double(double def) const + { + return (_attr && _attr->value) ? impl::get_value_double(_attr->value) : def; + } + + PUGI__FN float xml_attribute::as_float(float def) const + { + return (_attr && _attr->value) ? impl::get_value_float(_attr->value) : def; + } + + PUGI__FN bool xml_attribute::as_bool(bool def) const + { + return (_attr && _attr->value) ? impl::get_value_bool(_attr->value) : def; + } + +#ifdef PUGIXML_HAS_LONG_LONG + PUGI__FN long long xml_attribute::as_llong(long long def) const + { + return (_attr && _attr->value) ? impl::get_value_llong(_attr->value) : def; + } + + PUGI__FN unsigned long long xml_attribute::as_ullong(unsigned long long def) const + { + return (_attr && _attr->value) ? impl::get_value_ullong(_attr->value) : def; + } +#endif + + PUGI__FN bool xml_attribute::empty() const + { + return !_attr; + } + + PUGI__FN const char_t* xml_attribute::name() const + { + return (_attr && _attr->name) ? _attr->name + 0 : PUGIXML_TEXT(""); + } + + PUGI__FN const char_t* xml_attribute::value() const + { + return (_attr && _attr->value) ? _attr->value + 0 : PUGIXML_TEXT(""); + } + + PUGI__FN size_t xml_attribute::hash_value() const + { + return static_cast(reinterpret_cast(_attr) / sizeof(xml_attribute_struct)); + } + + PUGI__FN xml_attribute_struct* xml_attribute::internal_object() const + { + return _attr; + } + + PUGI__FN xml_attribute& xml_attribute::operator=(const char_t* rhs) + { + set_value(rhs); + return *this; + } + + PUGI__FN xml_attribute& xml_attribute::operator=(int rhs) + { + set_value(rhs); + return *this; + } + + PUGI__FN xml_attribute& xml_attribute::operator=(unsigned int rhs) + { + set_value(rhs); + return *this; + } + + PUGI__FN xml_attribute& xml_attribute::operator=(long rhs) + { + set_value(rhs); + return *this; + } + + PUGI__FN xml_attribute& xml_attribute::operator=(unsigned long rhs) + { + set_value(rhs); + return *this; + } + + PUGI__FN xml_attribute& xml_attribute::operator=(double rhs) + { + set_value(rhs); + return *this; + } + + PUGI__FN xml_attribute& xml_attribute::operator=(float rhs) + { + set_value(rhs); + return *this; + } + + PUGI__FN xml_attribute& xml_attribute::operator=(bool rhs) + { + set_value(rhs); + return *this; + } + +#ifdef PUGIXML_HAS_LONG_LONG + PUGI__FN xml_attribute& xml_attribute::operator=(long long rhs) + { + set_value(rhs); + return *this; + } + + PUGI__FN xml_attribute& xml_attribute::operator=(unsigned long long rhs) + { + set_value(rhs); + return *this; + } +#endif + + PUGI__FN bool xml_attribute::set_name(const char_t* rhs) + { + if (!_attr) return false; + + return impl::strcpy_insitu(_attr->name, _attr->header, impl::xml_memory_page_name_allocated_mask, rhs, impl::strlength(rhs)); + } + + PUGI__FN bool xml_attribute::set_value(const char_t* rhs) + { + if (!_attr) return false; + + return impl::strcpy_insitu(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, impl::strlength(rhs)); + } + + PUGI__FN bool xml_attribute::set_value(int rhs) + { + if (!_attr) return false; + + return impl::set_value_integer(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, rhs < 0); + } + + PUGI__FN bool xml_attribute::set_value(unsigned int rhs) + { + if (!_attr) return false; + + return impl::set_value_integer(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, false); + } + + PUGI__FN bool xml_attribute::set_value(long rhs) + { + if (!_attr) return false; + + return impl::set_value_integer(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, rhs < 0); + } + + PUGI__FN bool xml_attribute::set_value(unsigned long rhs) + { + if (!_attr) return false; + + return impl::set_value_integer(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, false); + } + + PUGI__FN bool xml_attribute::set_value(double rhs) + { + if (!_attr) return false; + + return impl::set_value_convert(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs); + } + + PUGI__FN bool xml_attribute::set_value(float rhs) + { + if (!_attr) return false; + + return impl::set_value_convert(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs); + } + + PUGI__FN bool xml_attribute::set_value(bool rhs) + { + if (!_attr) return false; + + return impl::set_value_bool(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs); + } + +#ifdef PUGIXML_HAS_LONG_LONG + PUGI__FN bool xml_attribute::set_value(long long rhs) + { + if (!_attr) return false; + + return impl::set_value_integer(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, rhs < 0); + } + + PUGI__FN bool xml_attribute::set_value(unsigned long long rhs) + { + if (!_attr) return false; + + return impl::set_value_integer(_attr->value, _attr->header, impl::xml_memory_page_value_allocated_mask, rhs, false); + } +#endif + +#ifdef __BORLANDC__ + PUGI__FN bool operator&&(const xml_attribute& lhs, bool rhs) + { + return (bool)lhs && rhs; + } + + PUGI__FN bool operator||(const xml_attribute& lhs, bool rhs) + { + return (bool)lhs || rhs; + } +#endif + + PUGI__FN xml_node::xml_node(): _root(0) + { + } + + PUGI__FN xml_node::xml_node(xml_node_struct* p): _root(p) + { + } + + PUGI__FN static void unspecified_bool_xml_node(xml_node***) + { + } + + PUGI__FN xml_node::operator xml_node::unspecified_bool_type() const + { + return _root ? unspecified_bool_xml_node : 0; + } + + PUGI__FN bool xml_node::operator!() const + { + return !_root; + } + + PUGI__FN xml_node::iterator xml_node::begin() const + { + return iterator(_root ? _root->first_child + 0 : 0, _root); + } + + PUGI__FN xml_node::iterator xml_node::end() const + { + return iterator(0, _root); + } + + PUGI__FN xml_node::attribute_iterator xml_node::attributes_begin() const + { + return attribute_iterator(_root ? _root->first_attribute + 0 : 0, _root); + } + + PUGI__FN xml_node::attribute_iterator xml_node::attributes_end() const + { + return attribute_iterator(0, _root); + } + + PUGI__FN xml_object_range xml_node::children() const + { + return xml_object_range(begin(), end()); + } + + PUGI__FN xml_object_range xml_node::children(const char_t* name_) const + { + return xml_object_range(xml_named_node_iterator(child(name_)._root, _root, name_), xml_named_node_iterator(0, _root, name_)); + } + + PUGI__FN xml_object_range xml_node::attributes() const + { + return xml_object_range(attributes_begin(), attributes_end()); + } + + PUGI__FN bool xml_node::operator==(const xml_node& r) const + { + return (_root == r._root); + } + + PUGI__FN bool xml_node::operator!=(const xml_node& r) const + { + return (_root != r._root); + } + + PUGI__FN bool xml_node::operator<(const xml_node& r) const + { + return (_root < r._root); + } + + PUGI__FN bool xml_node::operator>(const xml_node& r) const + { + return (_root > r._root); + } + + PUGI__FN bool xml_node::operator<=(const xml_node& r) const + { + return (_root <= r._root); + } + + PUGI__FN bool xml_node::operator>=(const xml_node& r) const + { + return (_root >= r._root); + } + + PUGI__FN bool xml_node::empty() const + { + return !_root; + } + + PUGI__FN const char_t* xml_node::name() const + { + return (_root && _root->name) ? _root->name + 0 : PUGIXML_TEXT(""); + } + + PUGI__FN xml_node_type xml_node::type() const + { + return _root ? PUGI__NODETYPE(_root) : node_null; + } + + PUGI__FN const char_t* xml_node::value() const + { + return (_root && _root->value) ? _root->value + 0 : PUGIXML_TEXT(""); + } + + PUGI__FN xml_node xml_node::child(const char_t* name_) const + { + if (!_root) return xml_node(); + + for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling) + if (i->name && impl::strequal(name_, i->name)) return xml_node(i); + + return xml_node(); + } + + PUGI__FN xml_attribute xml_node::attribute(const char_t* name_) const + { + if (!_root) return xml_attribute(); + + for (xml_attribute_struct* i = _root->first_attribute; i; i = i->next_attribute) + if (i->name && impl::strequal(name_, i->name)) + return xml_attribute(i); + + return xml_attribute(); + } + + PUGI__FN xml_node xml_node::next_sibling(const char_t* name_) const + { + if (!_root) return xml_node(); + + for (xml_node_struct* i = _root->next_sibling; i; i = i->next_sibling) + if (i->name && impl::strequal(name_, i->name)) return xml_node(i); + + return xml_node(); + } + + PUGI__FN xml_node xml_node::next_sibling() const + { + return _root ? xml_node(_root->next_sibling) : xml_node(); + } + + PUGI__FN xml_node xml_node::previous_sibling(const char_t* name_) const + { + if (!_root) return xml_node(); + + for (xml_node_struct* i = _root->prev_sibling_c; i->next_sibling; i = i->prev_sibling_c) + if (i->name && impl::strequal(name_, i->name)) return xml_node(i); + + return xml_node(); + } + + PUGI__FN xml_attribute xml_node::attribute(const char_t* name_, xml_attribute& hint_) const + { + xml_attribute_struct* hint = hint_._attr; + + // if hint is not an attribute of node, behavior is not defined + assert(!hint || (_root && impl::is_attribute_of(hint, _root))); + + if (!_root) return xml_attribute(); + + // optimistically search from hint up until the end + for (xml_attribute_struct* i = hint; i; i = i->next_attribute) + if (i->name && impl::strequal(name_, i->name)) + { + // update hint to maximize efficiency of searching for consecutive attributes + hint_._attr = i->next_attribute; + + return xml_attribute(i); + } + + // wrap around and search from the first attribute until the hint + // 'j' null pointer check is technically redundant, but it prevents a crash in case the assertion above fails + for (xml_attribute_struct* j = _root->first_attribute; j && j != hint; j = j->next_attribute) + if (j->name && impl::strequal(name_, j->name)) + { + // update hint to maximize efficiency of searching for consecutive attributes + hint_._attr = j->next_attribute; + + return xml_attribute(j); + } + + return xml_attribute(); + } + + PUGI__FN xml_node xml_node::previous_sibling() const + { + if (!_root) return xml_node(); + + if (_root->prev_sibling_c->next_sibling) return xml_node(_root->prev_sibling_c); + else return xml_node(); + } + + PUGI__FN xml_node xml_node::parent() const + { + return _root ? xml_node(_root->parent) : xml_node(); + } + + PUGI__FN xml_node xml_node::root() const + { + return _root ? xml_node(&impl::get_document(_root)) : xml_node(); + } + + PUGI__FN xml_text xml_node::text() const + { + return xml_text(_root); + } + + PUGI__FN const char_t* xml_node::child_value() const + { + if (!_root) return PUGIXML_TEXT(""); + + // element nodes can have value if parse_embed_pcdata was used + if (PUGI__NODETYPE(_root) == node_element && _root->value) + return _root->value; + + for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling) + if (impl::is_text_node(i) && i->value) + return i->value; + + return PUGIXML_TEXT(""); + } + + PUGI__FN const char_t* xml_node::child_value(const char_t* name_) const + { + return child(name_).child_value(); + } + + PUGI__FN xml_attribute xml_node::first_attribute() const + { + return _root ? xml_attribute(_root->first_attribute) : xml_attribute(); + } + + PUGI__FN xml_attribute xml_node::last_attribute() const + { + return _root && _root->first_attribute ? xml_attribute(_root->first_attribute->prev_attribute_c) : xml_attribute(); + } + + PUGI__FN xml_node xml_node::first_child() const + { + return _root ? xml_node(_root->first_child) : xml_node(); + } + + PUGI__FN xml_node xml_node::last_child() const + { + return _root && _root->first_child ? xml_node(_root->first_child->prev_sibling_c) : xml_node(); + } + + PUGI__FN bool xml_node::set_name(const char_t* rhs) + { + xml_node_type type_ = _root ? PUGI__NODETYPE(_root) : node_null; + + if (type_ != node_element && type_ != node_pi && type_ != node_declaration) + return false; + + return impl::strcpy_insitu(_root->name, _root->header, impl::xml_memory_page_name_allocated_mask, rhs, impl::strlength(rhs)); + } + + PUGI__FN bool xml_node::set_value(const char_t* rhs) + { + xml_node_type type_ = _root ? PUGI__NODETYPE(_root) : node_null; + + if (type_ != node_pcdata && type_ != node_cdata && type_ != node_comment && type_ != node_pi && type_ != node_doctype) + return false; + + return impl::strcpy_insitu(_root->value, _root->header, impl::xml_memory_page_value_allocated_mask, rhs, impl::strlength(rhs)); + } + + PUGI__FN xml_attribute xml_node::append_attribute(const char_t* name_) + { + if (!impl::allow_insert_attribute(type())) return xml_attribute(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_attribute(); + + xml_attribute a(impl::allocate_attribute(alloc)); + if (!a) return xml_attribute(); + + impl::append_attribute(a._attr, _root); + + a.set_name(name_); + + return a; + } + + PUGI__FN xml_attribute xml_node::prepend_attribute(const char_t* name_) + { + if (!impl::allow_insert_attribute(type())) return xml_attribute(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_attribute(); + + xml_attribute a(impl::allocate_attribute(alloc)); + if (!a) return xml_attribute(); + + impl::prepend_attribute(a._attr, _root); + + a.set_name(name_); + + return a; + } + + PUGI__FN xml_attribute xml_node::insert_attribute_after(const char_t* name_, const xml_attribute& attr) + { + if (!impl::allow_insert_attribute(type())) return xml_attribute(); + if (!attr || !impl::is_attribute_of(attr._attr, _root)) return xml_attribute(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_attribute(); + + xml_attribute a(impl::allocate_attribute(alloc)); + if (!a) return xml_attribute(); + + impl::insert_attribute_after(a._attr, attr._attr, _root); + + a.set_name(name_); + + return a; + } + + PUGI__FN xml_attribute xml_node::insert_attribute_before(const char_t* name_, const xml_attribute& attr) + { + if (!impl::allow_insert_attribute(type())) return xml_attribute(); + if (!attr || !impl::is_attribute_of(attr._attr, _root)) return xml_attribute(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_attribute(); + + xml_attribute a(impl::allocate_attribute(alloc)); + if (!a) return xml_attribute(); + + impl::insert_attribute_before(a._attr, attr._attr, _root); + + a.set_name(name_); + + return a; + } + + PUGI__FN xml_attribute xml_node::append_copy(const xml_attribute& proto) + { + if (!proto) return xml_attribute(); + if (!impl::allow_insert_attribute(type())) return xml_attribute(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_attribute(); + + xml_attribute a(impl::allocate_attribute(alloc)); + if (!a) return xml_attribute(); + + impl::append_attribute(a._attr, _root); + impl::node_copy_attribute(a._attr, proto._attr); + + return a; + } + + PUGI__FN xml_attribute xml_node::prepend_copy(const xml_attribute& proto) + { + if (!proto) return xml_attribute(); + if (!impl::allow_insert_attribute(type())) return xml_attribute(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_attribute(); + + xml_attribute a(impl::allocate_attribute(alloc)); + if (!a) return xml_attribute(); + + impl::prepend_attribute(a._attr, _root); + impl::node_copy_attribute(a._attr, proto._attr); + + return a; + } + + PUGI__FN xml_attribute xml_node::insert_copy_after(const xml_attribute& proto, const xml_attribute& attr) + { + if (!proto) return xml_attribute(); + if (!impl::allow_insert_attribute(type())) return xml_attribute(); + if (!attr || !impl::is_attribute_of(attr._attr, _root)) return xml_attribute(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_attribute(); + + xml_attribute a(impl::allocate_attribute(alloc)); + if (!a) return xml_attribute(); + + impl::insert_attribute_after(a._attr, attr._attr, _root); + impl::node_copy_attribute(a._attr, proto._attr); + + return a; + } + + PUGI__FN xml_attribute xml_node::insert_copy_before(const xml_attribute& proto, const xml_attribute& attr) + { + if (!proto) return xml_attribute(); + if (!impl::allow_insert_attribute(type())) return xml_attribute(); + if (!attr || !impl::is_attribute_of(attr._attr, _root)) return xml_attribute(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_attribute(); + + xml_attribute a(impl::allocate_attribute(alloc)); + if (!a) return xml_attribute(); + + impl::insert_attribute_before(a._attr, attr._attr, _root); + impl::node_copy_attribute(a._attr, proto._attr); + + return a; + } + + PUGI__FN xml_node xml_node::append_child(xml_node_type type_) + { + if (!impl::allow_insert_child(type(), type_)) return xml_node(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_node(); + + xml_node n(impl::allocate_node(alloc, type_)); + if (!n) return xml_node(); + + impl::append_node(n._root, _root); + + if (type_ == node_declaration) n.set_name(PUGIXML_TEXT("xml")); + + return n; + } + + PUGI__FN xml_node xml_node::prepend_child(xml_node_type type_) + { + if (!impl::allow_insert_child(type(), type_)) return xml_node(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_node(); + + xml_node n(impl::allocate_node(alloc, type_)); + if (!n) return xml_node(); + + impl::prepend_node(n._root, _root); + + if (type_ == node_declaration) n.set_name(PUGIXML_TEXT("xml")); + + return n; + } + + PUGI__FN xml_node xml_node::insert_child_before(xml_node_type type_, const xml_node& node) + { + if (!impl::allow_insert_child(type(), type_)) return xml_node(); + if (!node._root || node._root->parent != _root) return xml_node(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_node(); + + xml_node n(impl::allocate_node(alloc, type_)); + if (!n) return xml_node(); + + impl::insert_node_before(n._root, node._root); + + if (type_ == node_declaration) n.set_name(PUGIXML_TEXT("xml")); + + return n; + } + + PUGI__FN xml_node xml_node::insert_child_after(xml_node_type type_, const xml_node& node) + { + if (!impl::allow_insert_child(type(), type_)) return xml_node(); + if (!node._root || node._root->parent != _root) return xml_node(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_node(); + + xml_node n(impl::allocate_node(alloc, type_)); + if (!n) return xml_node(); + + impl::insert_node_after(n._root, node._root); + + if (type_ == node_declaration) n.set_name(PUGIXML_TEXT("xml")); + + return n; + } + + PUGI__FN xml_node xml_node::append_child(const char_t* name_) + { + xml_node result = append_child(node_element); + + result.set_name(name_); + + return result; + } + + PUGI__FN xml_node xml_node::prepend_child(const char_t* name_) + { + xml_node result = prepend_child(node_element); + + result.set_name(name_); + + return result; + } + + PUGI__FN xml_node xml_node::insert_child_after(const char_t* name_, const xml_node& node) + { + xml_node result = insert_child_after(node_element, node); + + result.set_name(name_); + + return result; + } + + PUGI__FN xml_node xml_node::insert_child_before(const char_t* name_, const xml_node& node) + { + xml_node result = insert_child_before(node_element, node); + + result.set_name(name_); + + return result; + } + + PUGI__FN xml_node xml_node::append_copy(const xml_node& proto) + { + xml_node_type type_ = proto.type(); + if (!impl::allow_insert_child(type(), type_)) return xml_node(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_node(); + + xml_node n(impl::allocate_node(alloc, type_)); + if (!n) return xml_node(); + + impl::append_node(n._root, _root); + impl::node_copy_tree(n._root, proto._root); + + return n; + } + + PUGI__FN xml_node xml_node::prepend_copy(const xml_node& proto) + { + xml_node_type type_ = proto.type(); + if (!impl::allow_insert_child(type(), type_)) return xml_node(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_node(); + + xml_node n(impl::allocate_node(alloc, type_)); + if (!n) return xml_node(); + + impl::prepend_node(n._root, _root); + impl::node_copy_tree(n._root, proto._root); + + return n; + } + + PUGI__FN xml_node xml_node::insert_copy_after(const xml_node& proto, const xml_node& node) + { + xml_node_type type_ = proto.type(); + if (!impl::allow_insert_child(type(), type_)) return xml_node(); + if (!node._root || node._root->parent != _root) return xml_node(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_node(); + + xml_node n(impl::allocate_node(alloc, type_)); + if (!n) return xml_node(); + + impl::insert_node_after(n._root, node._root); + impl::node_copy_tree(n._root, proto._root); + + return n; + } + + PUGI__FN xml_node xml_node::insert_copy_before(const xml_node& proto, const xml_node& node) + { + xml_node_type type_ = proto.type(); + if (!impl::allow_insert_child(type(), type_)) return xml_node(); + if (!node._root || node._root->parent != _root) return xml_node(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_node(); + + xml_node n(impl::allocate_node(alloc, type_)); + if (!n) return xml_node(); + + impl::insert_node_before(n._root, node._root); + impl::node_copy_tree(n._root, proto._root); + + return n; + } + + PUGI__FN xml_node xml_node::append_move(const xml_node& moved) + { + if (!impl::allow_move(*this, moved)) return xml_node(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_node(); + + // disable document_buffer_order optimization since moving nodes around changes document order without changing buffer pointers + impl::get_document(_root).header |= impl::xml_memory_page_contents_shared_mask; + + impl::remove_node(moved._root); + impl::append_node(moved._root, _root); + + return moved; + } + + PUGI__FN xml_node xml_node::prepend_move(const xml_node& moved) + { + if (!impl::allow_move(*this, moved)) return xml_node(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_node(); + + // disable document_buffer_order optimization since moving nodes around changes document order without changing buffer pointers + impl::get_document(_root).header |= impl::xml_memory_page_contents_shared_mask; + + impl::remove_node(moved._root); + impl::prepend_node(moved._root, _root); + + return moved; + } + + PUGI__FN xml_node xml_node::insert_move_after(const xml_node& moved, const xml_node& node) + { + if (!impl::allow_move(*this, moved)) return xml_node(); + if (!node._root || node._root->parent != _root) return xml_node(); + if (moved._root == node._root) return xml_node(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_node(); + + // disable document_buffer_order optimization since moving nodes around changes document order without changing buffer pointers + impl::get_document(_root).header |= impl::xml_memory_page_contents_shared_mask; + + impl::remove_node(moved._root); + impl::insert_node_after(moved._root, node._root); + + return moved; + } + + PUGI__FN xml_node xml_node::insert_move_before(const xml_node& moved, const xml_node& node) + { + if (!impl::allow_move(*this, moved)) return xml_node(); + if (!node._root || node._root->parent != _root) return xml_node(); + if (moved._root == node._root) return xml_node(); + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return xml_node(); + + // disable document_buffer_order optimization since moving nodes around changes document order without changing buffer pointers + impl::get_document(_root).header |= impl::xml_memory_page_contents_shared_mask; + + impl::remove_node(moved._root); + impl::insert_node_before(moved._root, node._root); + + return moved; + } + + PUGI__FN bool xml_node::remove_attribute(const char_t* name_) + { + return remove_attribute(attribute(name_)); + } + + PUGI__FN bool xml_node::remove_attribute(const xml_attribute& a) + { + if (!_root || !a._attr) return false; + if (!impl::is_attribute_of(a._attr, _root)) return false; + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return false; + + impl::remove_attribute(a._attr, _root); + impl::destroy_attribute(a._attr, alloc); + + return true; + } + + PUGI__FN bool xml_node::remove_child(const char_t* name_) + { + return remove_child(child(name_)); + } + + PUGI__FN bool xml_node::remove_child(const xml_node& n) + { + if (!_root || !n._root || n._root->parent != _root) return false; + + impl::xml_allocator& alloc = impl::get_allocator(_root); + if (!alloc.reserve()) return false; + + impl::remove_node(n._root); + impl::destroy_node(n._root, alloc); + + return true; + } + + PUGI__FN xml_parse_result xml_node::append_buffer(const void* contents, size_t size, unsigned int options, xml_encoding encoding) + { + // append_buffer is only valid for elements/documents + if (!impl::allow_insert_child(type(), node_element)) return impl::make_parse_result(status_append_invalid_root); + + // get document node + impl::xml_document_struct* doc = &impl::get_document(_root); + + // disable document_buffer_order optimization since in a document with multiple buffers comparing buffer pointers does not make sense + doc->header |= impl::xml_memory_page_contents_shared_mask; + + // get extra buffer element (we'll store the document fragment buffer there so that we can deallocate it later) + impl::xml_memory_page* page = 0; + impl::xml_extra_buffer* extra = static_cast(doc->allocate_memory(sizeof(impl::xml_extra_buffer) + sizeof(void*), page)); + (void)page; + + if (!extra) return impl::make_parse_result(status_out_of_memory); + + #ifdef PUGIXML_COMPACT + // align the memory block to a pointer boundary; this is required for compact mode where memory allocations are only 4b aligned + // note that this requires up to sizeof(void*)-1 additional memory, which the allocation above takes into account + extra = reinterpret_cast((reinterpret_cast(extra) + (sizeof(void*) - 1)) & ~(sizeof(void*) - 1)); + #endif + + // add extra buffer to the list + extra->buffer = 0; + extra->next = doc->extra_buffers; + doc->extra_buffers = extra; + + // name of the root has to be NULL before parsing - otherwise closing node mismatches will not be detected at the top level + impl::name_null_sentry sentry(_root); + + return impl::load_buffer_impl(doc, _root, const_cast(contents), size, options, encoding, false, false, &extra->buffer); + } + + PUGI__FN xml_node xml_node::find_child_by_attribute(const char_t* name_, const char_t* attr_name, const char_t* attr_value) const + { + if (!_root) return xml_node(); + + for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling) + if (i->name && impl::strequal(name_, i->name)) + { + for (xml_attribute_struct* a = i->first_attribute; a; a = a->next_attribute) + if (a->name && impl::strequal(attr_name, a->name) && impl::strequal(attr_value, a->value ? a->value + 0 : PUGIXML_TEXT(""))) + return xml_node(i); + } + + return xml_node(); + } + + PUGI__FN xml_node xml_node::find_child_by_attribute(const char_t* attr_name, const char_t* attr_value) const + { + if (!_root) return xml_node(); + + for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling) + for (xml_attribute_struct* a = i->first_attribute; a; a = a->next_attribute) + if (a->name && impl::strequal(attr_name, a->name) && impl::strequal(attr_value, a->value ? a->value + 0 : PUGIXML_TEXT(""))) + return xml_node(i); + + return xml_node(); + } + +#ifndef PUGIXML_NO_STL + PUGI__FN string_t xml_node::path(char_t delimiter) const + { + if (!_root) return string_t(); + + size_t offset = 0; + + for (xml_node_struct* i = _root; i; i = i->parent) + { + offset += (i != _root); + offset += i->name ? impl::strlength(i->name) : 0; + } + + string_t result; + result.resize(offset); + + for (xml_node_struct* j = _root; j; j = j->parent) + { + if (j != _root) + result[--offset] = delimiter; + + if (j->name) + { + size_t length = impl::strlength(j->name); + + offset -= length; + memcpy(&result[offset], j->name, length * sizeof(char_t)); + } + } + + assert(offset == 0); + + return result; + } +#endif + + PUGI__FN xml_node xml_node::first_element_by_path(const char_t* path_, char_t delimiter) const + { + xml_node found = *this; // Current search context. + + if (!_root || !path_[0]) return found; + + if (path_[0] == delimiter) + { + // Absolute path; e.g. '/foo/bar' + found = found.root(); + ++path_; + } + + const char_t* path_segment = path_; + + while (*path_segment == delimiter) ++path_segment; + + const char_t* path_segment_end = path_segment; + + while (*path_segment_end && *path_segment_end != delimiter) ++path_segment_end; + + if (path_segment == path_segment_end) return found; + + const char_t* next_segment = path_segment_end; + + while (*next_segment == delimiter) ++next_segment; + + if (*path_segment == '.' && path_segment + 1 == path_segment_end) + return found.first_element_by_path(next_segment, delimiter); + else if (*path_segment == '.' && *(path_segment+1) == '.' && path_segment + 2 == path_segment_end) + return found.parent().first_element_by_path(next_segment, delimiter); + else + { + for (xml_node_struct* j = found._root->first_child; j; j = j->next_sibling) + { + if (j->name && impl::strequalrange(j->name, path_segment, static_cast(path_segment_end - path_segment))) + { + xml_node subsearch = xml_node(j).first_element_by_path(next_segment, delimiter); + + if (subsearch) return subsearch; + } + } + + return xml_node(); + } + } + + PUGI__FN bool xml_node::traverse(xml_tree_walker& walker) + { + walker._depth = -1; + + xml_node arg_begin(_root); + if (!walker.begin(arg_begin)) return false; + + xml_node_struct* cur = _root ? _root->first_child + 0 : 0; + + if (cur) + { + ++walker._depth; + + do + { + xml_node arg_for_each(cur); + if (!walker.for_each(arg_for_each)) + return false; + + if (cur->first_child) + { + ++walker._depth; + cur = cur->first_child; + } + else if (cur->next_sibling) + cur = cur->next_sibling; + else + { + while (!cur->next_sibling && cur != _root && cur->parent) + { + --walker._depth; + cur = cur->parent; + } + + if (cur != _root) + cur = cur->next_sibling; + } + } + while (cur && cur != _root); + } + + assert(walker._depth == -1); + + xml_node arg_end(_root); + return walker.end(arg_end); + } + + PUGI__FN size_t xml_node::hash_value() const + { + return static_cast(reinterpret_cast(_root) / sizeof(xml_node_struct)); + } + + PUGI__FN xml_node_struct* xml_node::internal_object() const + { + return _root; + } + + PUGI__FN void xml_node::print(xml_writer& writer, const char_t* indent, unsigned int flags, xml_encoding encoding, unsigned int depth) const + { + if (!_root) return; + + impl::xml_buffered_writer buffered_writer(writer, encoding); + + impl::node_output(buffered_writer, _root, indent, flags, depth); + + buffered_writer.flush(); + } + +#ifndef PUGIXML_NO_STL + PUGI__FN void xml_node::print(std::basic_ostream >& stream, const char_t* indent, unsigned int flags, xml_encoding encoding, unsigned int depth) const + { + xml_writer_stream writer(stream); + + print(writer, indent, flags, encoding, depth); + } + + PUGI__FN void xml_node::print(std::basic_ostream >& stream, const char_t* indent, unsigned int flags, unsigned int depth) const + { + xml_writer_stream writer(stream); + + print(writer, indent, flags, encoding_wchar, depth); + } +#endif + + PUGI__FN ptrdiff_t xml_node::offset_debug() const + { + if (!_root) return -1; + + impl::xml_document_struct& doc = impl::get_document(_root); + + // we can determine the offset reliably only if there is exactly once parse buffer + if (!doc.buffer || doc.extra_buffers) return -1; + + switch (type()) + { + case node_document: + return 0; + + case node_element: + case node_declaration: + case node_pi: + return _root->name && (_root->header & impl::xml_memory_page_name_allocated_or_shared_mask) == 0 ? _root->name - doc.buffer : -1; + + case node_pcdata: + case node_cdata: + case node_comment: + case node_doctype: + return _root->value && (_root->header & impl::xml_memory_page_value_allocated_or_shared_mask) == 0 ? _root->value - doc.buffer : -1; + + default: + assert(false && "Invalid node type"); // unreachable + return -1; + } + } + +#ifdef __BORLANDC__ + PUGI__FN bool operator&&(const xml_node& lhs, bool rhs) + { + return (bool)lhs && rhs; + } + + PUGI__FN bool operator||(const xml_node& lhs, bool rhs) + { + return (bool)lhs || rhs; + } +#endif + + PUGI__FN xml_text::xml_text(xml_node_struct* root): _root(root) + { + } + + PUGI__FN xml_node_struct* xml_text::_data() const + { + if (!_root || impl::is_text_node(_root)) return _root; + + // element nodes can have value if parse_embed_pcdata was used + if (PUGI__NODETYPE(_root) == node_element && _root->value) + return _root; + + for (xml_node_struct* node = _root->first_child; node; node = node->next_sibling) + if (impl::is_text_node(node)) + return node; + + return 0; + } + + PUGI__FN xml_node_struct* xml_text::_data_new() + { + xml_node_struct* d = _data(); + if (d) return d; + + return xml_node(_root).append_child(node_pcdata).internal_object(); + } + + PUGI__FN xml_text::xml_text(): _root(0) + { + } + + PUGI__FN static void unspecified_bool_xml_text(xml_text***) + { + } + + PUGI__FN xml_text::operator xml_text::unspecified_bool_type() const + { + return _data() ? unspecified_bool_xml_text : 0; + } + + PUGI__FN bool xml_text::operator!() const + { + return !_data(); + } + + PUGI__FN bool xml_text::empty() const + { + return _data() == 0; + } + + PUGI__FN const char_t* xml_text::get() const + { + xml_node_struct* d = _data(); + + return (d && d->value) ? d->value + 0 : PUGIXML_TEXT(""); + } + + PUGI__FN const char_t* xml_text::as_string(const char_t* def) const + { + xml_node_struct* d = _data(); + + return (d && d->value) ? d->value + 0 : def; + } + + PUGI__FN int xml_text::as_int(int def) const + { + xml_node_struct* d = _data(); + + return (d && d->value) ? impl::get_value_int(d->value) : def; + } + + PUGI__FN unsigned int xml_text::as_uint(unsigned int def) const + { + xml_node_struct* d = _data(); + + return (d && d->value) ? impl::get_value_uint(d->value) : def; + } + + PUGI__FN double xml_text::as_double(double def) const + { + xml_node_struct* d = _data(); + + return (d && d->value) ? impl::get_value_double(d->value) : def; + } + + PUGI__FN float xml_text::as_float(float def) const + { + xml_node_struct* d = _data(); + + return (d && d->value) ? impl::get_value_float(d->value) : def; + } + + PUGI__FN bool xml_text::as_bool(bool def) const + { + xml_node_struct* d = _data(); + + return (d && d->value) ? impl::get_value_bool(d->value) : def; + } + +#ifdef PUGIXML_HAS_LONG_LONG + PUGI__FN long long xml_text::as_llong(long long def) const + { + xml_node_struct* d = _data(); + + return (d && d->value) ? impl::get_value_llong(d->value) : def; + } + + PUGI__FN unsigned long long xml_text::as_ullong(unsigned long long def) const + { + xml_node_struct* d = _data(); + + return (d && d->value) ? impl::get_value_ullong(d->value) : def; + } +#endif + + PUGI__FN bool xml_text::set(const char_t* rhs) + { + xml_node_struct* dn = _data_new(); + + return dn ? impl::strcpy_insitu(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, impl::strlength(rhs)) : false; + } + + PUGI__FN bool xml_text::set(int rhs) + { + xml_node_struct* dn = _data_new(); + + return dn ? impl::set_value_integer(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, rhs < 0) : false; + } + + PUGI__FN bool xml_text::set(unsigned int rhs) + { + xml_node_struct* dn = _data_new(); + + return dn ? impl::set_value_integer(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, false) : false; + } + + PUGI__FN bool xml_text::set(long rhs) + { + xml_node_struct* dn = _data_new(); + + return dn ? impl::set_value_integer(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, rhs < 0) : false; + } + + PUGI__FN bool xml_text::set(unsigned long rhs) + { + xml_node_struct* dn = _data_new(); + + return dn ? impl::set_value_integer(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, false) : false; + } + + PUGI__FN bool xml_text::set(float rhs) + { + xml_node_struct* dn = _data_new(); + + return dn ? impl::set_value_convert(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs) : false; + } + + PUGI__FN bool xml_text::set(double rhs) + { + xml_node_struct* dn = _data_new(); + + return dn ? impl::set_value_convert(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs) : false; + } + + PUGI__FN bool xml_text::set(bool rhs) + { + xml_node_struct* dn = _data_new(); + + return dn ? impl::set_value_bool(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs) : false; + } + +#ifdef PUGIXML_HAS_LONG_LONG + PUGI__FN bool xml_text::set(long long rhs) + { + xml_node_struct* dn = _data_new(); + + return dn ? impl::set_value_integer(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, rhs < 0) : false; + } + + PUGI__FN bool xml_text::set(unsigned long long rhs) + { + xml_node_struct* dn = _data_new(); + + return dn ? impl::set_value_integer(dn->value, dn->header, impl::xml_memory_page_value_allocated_mask, rhs, false) : false; + } +#endif + + PUGI__FN xml_text& xml_text::operator=(const char_t* rhs) + { + set(rhs); + return *this; + } + + PUGI__FN xml_text& xml_text::operator=(int rhs) + { + set(rhs); + return *this; + } + + PUGI__FN xml_text& xml_text::operator=(unsigned int rhs) + { + set(rhs); + return *this; + } + + PUGI__FN xml_text& xml_text::operator=(long rhs) + { + set(rhs); + return *this; + } + + PUGI__FN xml_text& xml_text::operator=(unsigned long rhs) + { + set(rhs); + return *this; + } + + PUGI__FN xml_text& xml_text::operator=(double rhs) + { + set(rhs); + return *this; + } + + PUGI__FN xml_text& xml_text::operator=(float rhs) + { + set(rhs); + return *this; + } + + PUGI__FN xml_text& xml_text::operator=(bool rhs) + { + set(rhs); + return *this; + } + +#ifdef PUGIXML_HAS_LONG_LONG + PUGI__FN xml_text& xml_text::operator=(long long rhs) + { + set(rhs); + return *this; + } + + PUGI__FN xml_text& xml_text::operator=(unsigned long long rhs) + { + set(rhs); + return *this; + } +#endif + + PUGI__FN xml_node xml_text::data() const + { + return xml_node(_data()); + } + +#ifdef __BORLANDC__ + PUGI__FN bool operator&&(const xml_text& lhs, bool rhs) + { + return (bool)lhs && rhs; + } + + PUGI__FN bool operator||(const xml_text& lhs, bool rhs) + { + return (bool)lhs || rhs; + } +#endif + + PUGI__FN xml_node_iterator::xml_node_iterator() + { + } + + PUGI__FN xml_node_iterator::xml_node_iterator(const xml_node& node): _wrap(node), _parent(node.parent()) + { + } + + PUGI__FN xml_node_iterator::xml_node_iterator(xml_node_struct* ref, xml_node_struct* parent): _wrap(ref), _parent(parent) + { + } + + PUGI__FN bool xml_node_iterator::operator==(const xml_node_iterator& rhs) const + { + return _wrap._root == rhs._wrap._root && _parent._root == rhs._parent._root; + } + + PUGI__FN bool xml_node_iterator::operator!=(const xml_node_iterator& rhs) const + { + return _wrap._root != rhs._wrap._root || _parent._root != rhs._parent._root; + } + + PUGI__FN xml_node& xml_node_iterator::operator*() const + { + assert(_wrap._root); + return _wrap; + } + + PUGI__FN xml_node* xml_node_iterator::operator->() const + { + assert(_wrap._root); + return const_cast(&_wrap); // BCC5 workaround + } + + PUGI__FN const xml_node_iterator& xml_node_iterator::operator++() + { + assert(_wrap._root); + _wrap._root = _wrap._root->next_sibling; + return *this; + } + + PUGI__FN xml_node_iterator xml_node_iterator::operator++(int) + { + xml_node_iterator temp = *this; + ++*this; + return temp; + } + + PUGI__FN const xml_node_iterator& xml_node_iterator::operator--() + { + _wrap = _wrap._root ? _wrap.previous_sibling() : _parent.last_child(); + return *this; + } + + PUGI__FN xml_node_iterator xml_node_iterator::operator--(int) + { + xml_node_iterator temp = *this; + --*this; + return temp; + } + + PUGI__FN xml_attribute_iterator::xml_attribute_iterator() + { + } + + PUGI__FN xml_attribute_iterator::xml_attribute_iterator(const xml_attribute& attr, const xml_node& parent): _wrap(attr), _parent(parent) + { + } + + PUGI__FN xml_attribute_iterator::xml_attribute_iterator(xml_attribute_struct* ref, xml_node_struct* parent): _wrap(ref), _parent(parent) + { + } + + PUGI__FN bool xml_attribute_iterator::operator==(const xml_attribute_iterator& rhs) const + { + return _wrap._attr == rhs._wrap._attr && _parent._root == rhs._parent._root; + } + + PUGI__FN bool xml_attribute_iterator::operator!=(const xml_attribute_iterator& rhs) const + { + return _wrap._attr != rhs._wrap._attr || _parent._root != rhs._parent._root; + } + + PUGI__FN xml_attribute& xml_attribute_iterator::operator*() const + { + assert(_wrap._attr); + return _wrap; + } + + PUGI__FN xml_attribute* xml_attribute_iterator::operator->() const + { + assert(_wrap._attr); + return const_cast(&_wrap); // BCC5 workaround + } + + PUGI__FN const xml_attribute_iterator& xml_attribute_iterator::operator++() + { + assert(_wrap._attr); + _wrap._attr = _wrap._attr->next_attribute; + return *this; + } + + PUGI__FN xml_attribute_iterator xml_attribute_iterator::operator++(int) + { + xml_attribute_iterator temp = *this; + ++*this; + return temp; + } + + PUGI__FN const xml_attribute_iterator& xml_attribute_iterator::operator--() + { + _wrap = _wrap._attr ? _wrap.previous_attribute() : _parent.last_attribute(); + return *this; + } + + PUGI__FN xml_attribute_iterator xml_attribute_iterator::operator--(int) + { + xml_attribute_iterator temp = *this; + --*this; + return temp; + } + + PUGI__FN xml_named_node_iterator::xml_named_node_iterator(): _name(0) + { + } + + PUGI__FN xml_named_node_iterator::xml_named_node_iterator(const xml_node& node, const char_t* name): _wrap(node), _parent(node.parent()), _name(name) + { + } + + PUGI__FN xml_named_node_iterator::xml_named_node_iterator(xml_node_struct* ref, xml_node_struct* parent, const char_t* name): _wrap(ref), _parent(parent), _name(name) + { + } + + PUGI__FN bool xml_named_node_iterator::operator==(const xml_named_node_iterator& rhs) const + { + return _wrap._root == rhs._wrap._root && _parent._root == rhs._parent._root; + } + + PUGI__FN bool xml_named_node_iterator::operator!=(const xml_named_node_iterator& rhs) const + { + return _wrap._root != rhs._wrap._root || _parent._root != rhs._parent._root; + } + + PUGI__FN xml_node& xml_named_node_iterator::operator*() const + { + assert(_wrap._root); + return _wrap; + } + + PUGI__FN xml_node* xml_named_node_iterator::operator->() const + { + assert(_wrap._root); + return const_cast(&_wrap); // BCC5 workaround + } + + PUGI__FN const xml_named_node_iterator& xml_named_node_iterator::operator++() + { + assert(_wrap._root); + _wrap = _wrap.next_sibling(_name); + return *this; + } + + PUGI__FN xml_named_node_iterator xml_named_node_iterator::operator++(int) + { + xml_named_node_iterator temp = *this; + ++*this; + return temp; + } + + PUGI__FN const xml_named_node_iterator& xml_named_node_iterator::operator--() + { + if (_wrap._root) + _wrap = _wrap.previous_sibling(_name); + else + { + _wrap = _parent.last_child(); + + if (!impl::strequal(_wrap.name(), _name)) + _wrap = _wrap.previous_sibling(_name); + } + + return *this; + } + + PUGI__FN xml_named_node_iterator xml_named_node_iterator::operator--(int) + { + xml_named_node_iterator temp = *this; + --*this; + return temp; + } + + PUGI__FN xml_parse_result::xml_parse_result(): status(status_internal_error), offset(0), encoding(encoding_auto) + { + } + + PUGI__FN xml_parse_result::operator bool() const + { + return status == status_ok; + } + + PUGI__FN const char* xml_parse_result::description() const + { + switch (status) + { + case status_ok: return "No error"; + + case status_file_not_found: return "File was not found"; + case status_io_error: return "Error reading from file/stream"; + case status_out_of_memory: return "Could not allocate memory"; + case status_internal_error: return "Internal error occurred"; + + case status_unrecognized_tag: return "Could not determine tag type"; + + case status_bad_pi: return "Error parsing document declaration/processing instruction"; + case status_bad_comment: return "Error parsing comment"; + case status_bad_cdata: return "Error parsing CDATA section"; + case status_bad_doctype: return "Error parsing document type declaration"; + case status_bad_pcdata: return "Error parsing PCDATA section"; + case status_bad_start_element: return "Error parsing start element tag"; + case status_bad_attribute: return "Error parsing element attribute"; + case status_bad_end_element: return "Error parsing end element tag"; + case status_end_element_mismatch: return "Start-end tags mismatch"; + + case status_append_invalid_root: return "Unable to append nodes: root is not an element or document"; + + case status_no_document_element: return "No document element found"; + + default: return "Unknown error"; + } + } + + PUGI__FN xml_document::xml_document(): _buffer(0) + { + _create(); + } + + PUGI__FN xml_document::~xml_document() + { + _destroy(); + } + +#ifdef PUGIXML_HAS_MOVE + PUGI__FN xml_document::xml_document(xml_document&& rhs) PUGIXML_NOEXCEPT_IF_NOT_COMPACT: _buffer(0) + { + _create(); + _move(rhs); + } + + PUGI__FN xml_document& xml_document::operator=(xml_document&& rhs) PUGIXML_NOEXCEPT_IF_NOT_COMPACT + { + if (this == &rhs) return *this; + + _destroy(); + _create(); + _move(rhs); + + return *this; + } +#endif + + PUGI__FN void xml_document::reset() + { + _destroy(); + _create(); + } + + PUGI__FN void xml_document::reset(const xml_document& proto) + { + reset(); + + for (xml_node cur = proto.first_child(); cur; cur = cur.next_sibling()) + append_copy(cur); + } + + PUGI__FN void xml_document::_create() + { + assert(!_root); + + #ifdef PUGIXML_COMPACT + // space for page marker for the first page (uint32_t), rounded up to pointer size; assumes pointers are at least 32-bit + const size_t page_offset = sizeof(void*); + #else + const size_t page_offset = 0; + #endif + + // initialize sentinel page + PUGI__STATIC_ASSERT(sizeof(impl::xml_memory_page) + sizeof(impl::xml_document_struct) + page_offset <= sizeof(_memory)); + + // prepare page structure + impl::xml_memory_page* page = impl::xml_memory_page::construct(_memory); + assert(page); + + page->busy_size = impl::xml_memory_page_size; + + // setup first page marker + #ifdef PUGIXML_COMPACT + // round-trip through void* to avoid 'cast increases required alignment of target type' warning + page->compact_page_marker = reinterpret_cast(static_cast(reinterpret_cast(page) + sizeof(impl::xml_memory_page))); + *page->compact_page_marker = sizeof(impl::xml_memory_page); + #endif + + // allocate new root + _root = new (reinterpret_cast(page) + sizeof(impl::xml_memory_page) + page_offset) impl::xml_document_struct(page); + _root->prev_sibling_c = _root; + + // setup sentinel page + page->allocator = static_cast(_root); + + // setup hash table pointer in allocator + #ifdef PUGIXML_COMPACT + page->allocator->_hash = &static_cast(_root)->hash; + #endif + + // verify the document allocation + assert(reinterpret_cast(_root) + sizeof(impl::xml_document_struct) <= _memory + sizeof(_memory)); + } + + PUGI__FN void xml_document::_destroy() + { + assert(_root); + + // destroy static storage + if (_buffer) + { + impl::xml_memory::deallocate(_buffer); + _buffer = 0; + } + + // destroy extra buffers (note: no need to destroy linked list nodes, they're allocated using document allocator) + for (impl::xml_extra_buffer* extra = static_cast(_root)->extra_buffers; extra; extra = extra->next) + { + if (extra->buffer) impl::xml_memory::deallocate(extra->buffer); + } + + // destroy dynamic storage, leave sentinel page (it's in static memory) + impl::xml_memory_page* root_page = PUGI__GETPAGE(_root); + assert(root_page && !root_page->prev); + assert(reinterpret_cast(root_page) >= _memory && reinterpret_cast(root_page) < _memory + sizeof(_memory)); + + for (impl::xml_memory_page* page = root_page->next; page; ) + { + impl::xml_memory_page* next = page->next; + + impl::xml_allocator::deallocate_page(page); + + page = next; + } + + #ifdef PUGIXML_COMPACT + // destroy hash table + static_cast(_root)->hash.clear(); + #endif + + _root = 0; + } + +#ifdef PUGIXML_HAS_MOVE + PUGI__FN void xml_document::_move(xml_document& rhs) PUGIXML_NOEXCEPT_IF_NOT_COMPACT + { + impl::xml_document_struct* doc = static_cast(_root); + impl::xml_document_struct* other = static_cast(rhs._root); + + // save first child pointer for later; this needs hash access + xml_node_struct* other_first_child = other->first_child; + + #ifdef PUGIXML_COMPACT + // reserve space for the hash table up front; this is the only operation that can fail + // if it does, we have no choice but to throw (if we have exceptions) + if (other_first_child) + { + size_t other_children = 0; + for (xml_node_struct* node = other_first_child; node; node = node->next_sibling) + other_children++; + + // in compact mode, each pointer assignment could result in a hash table request + // during move, we have to relocate document first_child and parents of all children + // normally there's just one child and its parent has a pointerless encoding but + // we assume the worst here + if (!other->_hash->reserve(other_children + 1)) + { + #ifdef PUGIXML_NO_EXCEPTIONS + return; + #else + throw std::bad_alloc(); + #endif + } + } + #endif + + // move allocation state + doc->_root = other->_root; + doc->_busy_size = other->_busy_size; + + // move buffer state + doc->buffer = other->buffer; + doc->extra_buffers = other->extra_buffers; + _buffer = rhs._buffer; + + #ifdef PUGIXML_COMPACT + // move compact hash; note that the hash table can have pointers to other but they will be "inactive", similarly to nodes removed with remove_child + doc->hash = other->hash; + doc->_hash = &doc->hash; + + // make sure we don't access other hash up until the end when we reinitialize other document + other->_hash = 0; + #endif + + // move page structure + impl::xml_memory_page* doc_page = PUGI__GETPAGE(doc); + assert(doc_page && !doc_page->prev && !doc_page->next); + + impl::xml_memory_page* other_page = PUGI__GETPAGE(other); + assert(other_page && !other_page->prev); + + // relink pages since root page is embedded into xml_document + if (impl::xml_memory_page* page = other_page->next) + { + assert(page->prev == other_page); + + page->prev = doc_page; + + doc_page->next = page; + other_page->next = 0; + } + + // make sure pages point to the correct document state + for (impl::xml_memory_page* page = doc_page->next; page; page = page->next) + { + assert(page->allocator == other); + + page->allocator = doc; + + #ifdef PUGIXML_COMPACT + // this automatically migrates most children between documents and prevents ->parent assignment from allocating + if (page->compact_shared_parent == other) + page->compact_shared_parent = doc; + #endif + } + + // move tree structure + assert(!doc->first_child); + + doc->first_child = other_first_child; + + for (xml_node_struct* node = other_first_child; node; node = node->next_sibling) + { + #ifdef PUGIXML_COMPACT + // most children will have migrated when we reassigned compact_shared_parent + assert(node->parent == other || node->parent == doc); + + node->parent = doc; + #else + assert(node->parent == other); + node->parent = doc; + #endif + } + + // reset other document + new (other) impl::xml_document_struct(PUGI__GETPAGE(other)); + rhs._buffer = 0; + } +#endif + +#ifndef PUGIXML_NO_STL + PUGI__FN xml_parse_result xml_document::load(std::basic_istream >& stream, unsigned int options, xml_encoding encoding) + { + reset(); + + return impl::load_stream_impl(static_cast(_root), stream, options, encoding, &_buffer); + } + + PUGI__FN xml_parse_result xml_document::load(std::basic_istream >& stream, unsigned int options) + { + reset(); + + return impl::load_stream_impl(static_cast(_root), stream, options, encoding_wchar, &_buffer); + } +#endif + + PUGI__FN xml_parse_result xml_document::load_string(const char_t* contents, unsigned int options) + { + // Force native encoding (skip autodetection) + #ifdef PUGIXML_WCHAR_MODE + xml_encoding encoding = encoding_wchar; + #else + xml_encoding encoding = encoding_utf8; + #endif + + return load_buffer(contents, impl::strlength(contents) * sizeof(char_t), options, encoding); + } + + PUGI__FN xml_parse_result xml_document::load(const char_t* contents, unsigned int options) + { + return load_string(contents, options); + } + + PUGI__FN xml_parse_result xml_document::load_file(const char* path_, unsigned int options, xml_encoding encoding) + { + reset(); + + using impl::auto_deleter; // MSVC7 workaround + auto_deleter file(fopen(path_, "rb"), impl::close_file); + + return impl::load_file_impl(static_cast(_root), file.data, options, encoding, &_buffer); + } + + PUGI__FN xml_parse_result xml_document::load_file(const wchar_t* path_, unsigned int options, xml_encoding encoding) + { + reset(); + + using impl::auto_deleter; // MSVC7 workaround + auto_deleter file(impl::open_file_wide(path_, L"rb"), impl::close_file); + + return impl::load_file_impl(static_cast(_root), file.data, options, encoding, &_buffer); + } + + PUGI__FN xml_parse_result xml_document::load_buffer(const void* contents, size_t size, unsigned int options, xml_encoding encoding) + { + reset(); + + return impl::load_buffer_impl(static_cast(_root), _root, const_cast(contents), size, options, encoding, false, false, &_buffer); + } + + PUGI__FN xml_parse_result xml_document::load_buffer_inplace(void* contents, size_t size, unsigned int options, xml_encoding encoding) + { + reset(); + + return impl::load_buffer_impl(static_cast(_root), _root, contents, size, options, encoding, true, false, &_buffer); + } + + PUGI__FN xml_parse_result xml_document::load_buffer_inplace_own(void* contents, size_t size, unsigned int options, xml_encoding encoding) + { + reset(); + + return impl::load_buffer_impl(static_cast(_root), _root, contents, size, options, encoding, true, true, &_buffer); + } + + PUGI__FN void xml_document::save(xml_writer& writer, const char_t* indent, unsigned int flags, xml_encoding encoding) const + { + impl::xml_buffered_writer buffered_writer(writer, encoding); + + if ((flags & format_write_bom) && encoding != encoding_latin1) + { + // BOM always represents the codepoint U+FEFF, so just write it in native encoding + #ifdef PUGIXML_WCHAR_MODE + unsigned int bom = 0xfeff; + buffered_writer.write(static_cast(bom)); + #else + buffered_writer.write('\xef', '\xbb', '\xbf'); + #endif + } + + if (!(flags & format_no_declaration) && !impl::has_declaration(_root)) + { + buffered_writer.write_string(PUGIXML_TEXT("'); + if (!(flags & format_raw)) buffered_writer.write('\n'); + } + + impl::node_output(buffered_writer, _root, indent, flags, 0); + + buffered_writer.flush(); + } + +#ifndef PUGIXML_NO_STL + PUGI__FN void xml_document::save(std::basic_ostream >& stream, const char_t* indent, unsigned int flags, xml_encoding encoding) const + { + xml_writer_stream writer(stream); + + save(writer, indent, flags, encoding); + } + + PUGI__FN void xml_document::save(std::basic_ostream >& stream, const char_t* indent, unsigned int flags) const + { + xml_writer_stream writer(stream); + + save(writer, indent, flags, encoding_wchar); + } +#endif + + PUGI__FN bool xml_document::save_file(const char* path_, const char_t* indent, unsigned int flags, xml_encoding encoding) const + { + using impl::auto_deleter; // MSVC7 workaround + auto_deleter file(fopen(path_, (flags & format_save_file_text) ? "w" : "wb"), impl::close_file); + + return impl::save_file_impl(*this, file.data, indent, flags, encoding); + } + + PUGI__FN bool xml_document::save_file(const wchar_t* path_, const char_t* indent, unsigned int flags, xml_encoding encoding) const + { + using impl::auto_deleter; // MSVC7 workaround + auto_deleter file(impl::open_file_wide(path_, (flags & format_save_file_text) ? L"w" : L"wb"), impl::close_file); + + return impl::save_file_impl(*this, file.data, indent, flags, encoding); + } + + PUGI__FN xml_node xml_document::document_element() const + { + assert(_root); + + for (xml_node_struct* i = _root->first_child; i; i = i->next_sibling) + if (PUGI__NODETYPE(i) == node_element) + return xml_node(i); + + return xml_node(); + } + +#ifndef PUGIXML_NO_STL + PUGI__FN std::string PUGIXML_FUNCTION as_utf8(const wchar_t* str) + { + assert(str); + + return impl::as_utf8_impl(str, impl::strlength_wide(str)); + } + + PUGI__FN std::string PUGIXML_FUNCTION as_utf8(const std::basic_string& str) + { + return impl::as_utf8_impl(str.c_str(), str.size()); + } + + PUGI__FN std::basic_string PUGIXML_FUNCTION as_wide(const char* str) + { + assert(str); + + return impl::as_wide_impl(str, strlen(str)); + } + + PUGI__FN std::basic_string PUGIXML_FUNCTION as_wide(const std::string& str) + { + return impl::as_wide_impl(str.c_str(), str.size()); + } +#endif + + PUGI__FN void PUGIXML_FUNCTION set_memory_management_functions(allocation_function allocate, deallocation_function deallocate) + { + impl::xml_memory::allocate = allocate; + impl::xml_memory::deallocate = deallocate; + } + + PUGI__FN allocation_function PUGIXML_FUNCTION get_memory_allocation_function() + { + return impl::xml_memory::allocate; + } + + PUGI__FN deallocation_function PUGIXML_FUNCTION get_memory_deallocation_function() + { + return impl::xml_memory::deallocate; + } +} + +#if !defined(PUGIXML_NO_STL) && (defined(_MSC_VER) || defined(__ICC)) +namespace std +{ + // Workarounds for (non-standard) iterator category detection for older versions (MSVC7/IC8 and earlier) + PUGI__FN std::bidirectional_iterator_tag _Iter_cat(const pugi::xml_node_iterator&) + { + return std::bidirectional_iterator_tag(); + } + + PUGI__FN std::bidirectional_iterator_tag _Iter_cat(const pugi::xml_attribute_iterator&) + { + return std::bidirectional_iterator_tag(); + } + + PUGI__FN std::bidirectional_iterator_tag _Iter_cat(const pugi::xml_named_node_iterator&) + { + return std::bidirectional_iterator_tag(); + } +} +#endif + +#if !defined(PUGIXML_NO_STL) && defined(__SUNPRO_CC) +namespace std +{ + // Workarounds for (non-standard) iterator category detection + PUGI__FN std::bidirectional_iterator_tag __iterator_category(const pugi::xml_node_iterator&) + { + return std::bidirectional_iterator_tag(); + } + + PUGI__FN std::bidirectional_iterator_tag __iterator_category(const pugi::xml_attribute_iterator&) + { + return std::bidirectional_iterator_tag(); + } + + PUGI__FN std::bidirectional_iterator_tag __iterator_category(const pugi::xml_named_node_iterator&) + { + return std::bidirectional_iterator_tag(); + } +} +#endif + +#ifndef PUGIXML_NO_XPATH +// STL replacements +PUGI__NS_BEGIN + struct equal_to + { + template bool operator()(const T& lhs, const T& rhs) const + { + return lhs == rhs; + } + }; + + struct not_equal_to + { + template bool operator()(const T& lhs, const T& rhs) const + { + return lhs != rhs; + } + }; + + struct less + { + template bool operator()(const T& lhs, const T& rhs) const + { + return lhs < rhs; + } + }; + + struct less_equal + { + template bool operator()(const T& lhs, const T& rhs) const + { + return lhs <= rhs; + } + }; + + template inline void swap(T& lhs, T& rhs) + { + T temp = lhs; + lhs = rhs; + rhs = temp; + } + + template PUGI__FN I min_element(I begin, I end, const Pred& pred) + { + I result = begin; + + for (I it = begin + 1; it != end; ++it) + if (pred(*it, *result)) + result = it; + + return result; + } + + template PUGI__FN void reverse(I begin, I end) + { + while (end - begin > 1) + swap(*begin++, *--end); + } + + template PUGI__FN I unique(I begin, I end) + { + // fast skip head + while (end - begin > 1 && *begin != *(begin + 1)) + begin++; + + if (begin == end) + return begin; + + // last written element + I write = begin++; + + // merge unique elements + while (begin != end) + { + if (*begin != *write) + *++write = *begin++; + else + begin++; + } + + // past-the-end (write points to live element) + return write + 1; + } + + template PUGI__FN void insertion_sort(T* begin, T* end, const Pred& pred) + { + if (begin == end) + return; + + for (T* it = begin + 1; it != end; ++it) + { + T val = *it; + T* hole = it; + + // move hole backwards + while (hole > begin && pred(val, *(hole - 1))) + { + *hole = *(hole - 1); + hole--; + } + + // fill hole with element + *hole = val; + } + } + + template inline I median3(I first, I middle, I last, const Pred& pred) + { + if (pred(*middle, *first)) + swap(middle, first); + if (pred(*last, *middle)) + swap(last, middle); + if (pred(*middle, *first)) + swap(middle, first); + + return middle; + } + + template PUGI__FN void partition3(T* begin, T* end, T pivot, const Pred& pred, T** out_eqbeg, T** out_eqend) + { + // invariant: array is split into 4 groups: = < ? > (each variable denotes the boundary between the groups) + T* eq = begin; + T* lt = begin; + T* gt = end; + + while (lt < gt) + { + if (pred(*lt, pivot)) + lt++; + else if (*lt == pivot) + swap(*eq++, *lt++); + else + swap(*lt, *--gt); + } + + // we now have just 4 groups: = < >; move equal elements to the middle + T* eqbeg = gt; + + for (T* it = begin; it != eq; ++it) + swap(*it, *--eqbeg); + + *out_eqbeg = eqbeg; + *out_eqend = gt; + } + + template PUGI__FN void sort(I begin, I end, const Pred& pred) + { + // sort large chunks + while (end - begin > 16) + { + // find median element + I middle = begin + (end - begin) / 2; + I median = median3(begin, middle, end - 1, pred); + + // partition in three chunks (< = >) + I eqbeg, eqend; + partition3(begin, end, *median, pred, &eqbeg, &eqend); + + // loop on larger half + if (eqbeg - begin > end - eqend) + { + sort(eqend, end, pred); + end = eqbeg; + } + else + { + sort(begin, eqbeg, pred); + begin = eqend; + } + } + + // insertion sort small chunk + insertion_sort(begin, end, pred); + } + + PUGI__FN bool hash_insert(const void** table, size_t size, const void* key) + { + assert(key); + + unsigned int h = static_cast(reinterpret_cast(key)); + + // MurmurHash3 32-bit finalizer + h ^= h >> 16; + h *= 0x85ebca6bu; + h ^= h >> 13; + h *= 0xc2b2ae35u; + h ^= h >> 16; + + size_t hashmod = size - 1; + size_t bucket = h & hashmod; + + for (size_t probe = 0; probe <= hashmod; ++probe) + { + if (table[bucket] == 0) + { + table[bucket] = key; + return true; + } + + if (table[bucket] == key) + return false; + + // hash collision, quadratic probing + bucket = (bucket + probe + 1) & hashmod; + } + + assert(false && "Hash table is full"); // unreachable + return false; + } +PUGI__NS_END + +// Allocator used for AST and evaluation stacks +PUGI__NS_BEGIN + static const size_t xpath_memory_page_size = + #ifdef PUGIXML_MEMORY_XPATH_PAGE_SIZE + PUGIXML_MEMORY_XPATH_PAGE_SIZE + #else + 4096 + #endif + ; + + static const uintptr_t xpath_memory_block_alignment = sizeof(double) > sizeof(void*) ? sizeof(double) : sizeof(void*); + + struct xpath_memory_block + { + xpath_memory_block* next; + size_t capacity; + + union + { + char data[xpath_memory_page_size]; + double alignment; + }; + }; + + struct xpath_allocator + { + xpath_memory_block* _root; + size_t _root_size; + bool* _error; + + xpath_allocator(xpath_memory_block* root, bool* error = 0): _root(root), _root_size(0), _error(error) + { + } + + void* allocate(size_t size) + { + // round size up to block alignment boundary + size = (size + xpath_memory_block_alignment - 1) & ~(xpath_memory_block_alignment - 1); + + if (_root_size + size <= _root->capacity) + { + void* buf = &_root->data[0] + _root_size; + _root_size += size; + return buf; + } + else + { + // make sure we have at least 1/4th of the page free after allocation to satisfy subsequent allocation requests + size_t block_capacity_base = sizeof(_root->data); + size_t block_capacity_req = size + block_capacity_base / 4; + size_t block_capacity = (block_capacity_base > block_capacity_req) ? block_capacity_base : block_capacity_req; + + size_t block_size = block_capacity + offsetof(xpath_memory_block, data); + + xpath_memory_block* block = static_cast(xml_memory::allocate(block_size)); + if (!block) + { + if (_error) *_error = true; + return 0; + } + + block->next = _root; + block->capacity = block_capacity; + + _root = block; + _root_size = size; + + return block->data; + } + } + + void* reallocate(void* ptr, size_t old_size, size_t new_size) + { + // round size up to block alignment boundary + old_size = (old_size + xpath_memory_block_alignment - 1) & ~(xpath_memory_block_alignment - 1); + new_size = (new_size + xpath_memory_block_alignment - 1) & ~(xpath_memory_block_alignment - 1); + + // we can only reallocate the last object + assert(ptr == 0 || static_cast(ptr) + old_size == &_root->data[0] + _root_size); + + // try to reallocate the object inplace + if (ptr && _root_size - old_size + new_size <= _root->capacity) + { + _root_size = _root_size - old_size + new_size; + return ptr; + } + + // allocate a new block + void* result = allocate(new_size); + if (!result) return 0; + + // we have a new block + if (ptr) + { + // copy old data (we only support growing) + assert(new_size >= old_size); + memcpy(result, ptr, old_size); + + // free the previous page if it had no other objects + assert(_root->data == result); + assert(_root->next); + + if (_root->next->data == ptr) + { + // deallocate the whole page, unless it was the first one + xpath_memory_block* next = _root->next->next; + + if (next) + { + xml_memory::deallocate(_root->next); + _root->next = next; + } + } + } + + return result; + } + + void revert(const xpath_allocator& state) + { + // free all new pages + xpath_memory_block* cur = _root; + + while (cur != state._root) + { + xpath_memory_block* next = cur->next; + + xml_memory::deallocate(cur); + + cur = next; + } + + // restore state + _root = state._root; + _root_size = state._root_size; + } + + void release() + { + xpath_memory_block* cur = _root; + assert(cur); + + while (cur->next) + { + xpath_memory_block* next = cur->next; + + xml_memory::deallocate(cur); + + cur = next; + } + } + }; + + struct xpath_allocator_capture + { + xpath_allocator_capture(xpath_allocator* alloc): _target(alloc), _state(*alloc) + { + } + + ~xpath_allocator_capture() + { + _target->revert(_state); + } + + xpath_allocator* _target; + xpath_allocator _state; + }; + + struct xpath_stack + { + xpath_allocator* result; + xpath_allocator* temp; + }; + + struct xpath_stack_data + { + xpath_memory_block blocks[2]; + xpath_allocator result; + xpath_allocator temp; + xpath_stack stack; + bool oom; + + xpath_stack_data(): result(blocks + 0, &oom), temp(blocks + 1, &oom), oom(false) + { + blocks[0].next = blocks[1].next = 0; + blocks[0].capacity = blocks[1].capacity = sizeof(blocks[0].data); + + stack.result = &result; + stack.temp = &temp; + } + + ~xpath_stack_data() + { + result.release(); + temp.release(); + } + }; +PUGI__NS_END + +// String class +PUGI__NS_BEGIN + class xpath_string + { + const char_t* _buffer; + bool _uses_heap; + size_t _length_heap; + + static char_t* duplicate_string(const char_t* string, size_t length, xpath_allocator* alloc) + { + char_t* result = static_cast(alloc->allocate((length + 1) * sizeof(char_t))); + if (!result) return 0; + + memcpy(result, string, length * sizeof(char_t)); + result[length] = 0; + + return result; + } + + xpath_string(const char_t* buffer, bool uses_heap_, size_t length_heap): _buffer(buffer), _uses_heap(uses_heap_), _length_heap(length_heap) + { + } + + public: + static xpath_string from_const(const char_t* str) + { + return xpath_string(str, false, 0); + } + + static xpath_string from_heap_preallocated(const char_t* begin, const char_t* end) + { + assert(begin <= end && *end == 0); + + return xpath_string(begin, true, static_cast(end - begin)); + } + + static xpath_string from_heap(const char_t* begin, const char_t* end, xpath_allocator* alloc) + { + assert(begin <= end); + + if (begin == end) + return xpath_string(); + + size_t length = static_cast(end - begin); + const char_t* data = duplicate_string(begin, length, alloc); + + return data ? xpath_string(data, true, length) : xpath_string(); + } + + xpath_string(): _buffer(PUGIXML_TEXT("")), _uses_heap(false), _length_heap(0) + { + } + + void append(const xpath_string& o, xpath_allocator* alloc) + { + // skip empty sources + if (!*o._buffer) return; + + // fast append for constant empty target and constant source + if (!*_buffer && !_uses_heap && !o._uses_heap) + { + _buffer = o._buffer; + } + else + { + // need to make heap copy + size_t target_length = length(); + size_t source_length = o.length(); + size_t result_length = target_length + source_length; + + // allocate new buffer + char_t* result = static_cast(alloc->reallocate(_uses_heap ? const_cast(_buffer) : 0, (target_length + 1) * sizeof(char_t), (result_length + 1) * sizeof(char_t))); + if (!result) return; + + // append first string to the new buffer in case there was no reallocation + if (!_uses_heap) memcpy(result, _buffer, target_length * sizeof(char_t)); + + // append second string to the new buffer + memcpy(result + target_length, o._buffer, source_length * sizeof(char_t)); + result[result_length] = 0; + + // finalize + _buffer = result; + _uses_heap = true; + _length_heap = result_length; + } + } + + const char_t* c_str() const + { + return _buffer; + } + + size_t length() const + { + return _uses_heap ? _length_heap : strlength(_buffer); + } + + char_t* data(xpath_allocator* alloc) + { + // make private heap copy + if (!_uses_heap) + { + size_t length_ = strlength(_buffer); + const char_t* data_ = duplicate_string(_buffer, length_, alloc); + + if (!data_) return 0; + + _buffer = data_; + _uses_heap = true; + _length_heap = length_; + } + + return const_cast(_buffer); + } + + bool empty() const + { + return *_buffer == 0; + } + + bool operator==(const xpath_string& o) const + { + return strequal(_buffer, o._buffer); + } + + bool operator!=(const xpath_string& o) const + { + return !strequal(_buffer, o._buffer); + } + + bool uses_heap() const + { + return _uses_heap; + } + }; +PUGI__NS_END + +PUGI__NS_BEGIN + PUGI__FN bool starts_with(const char_t* string, const char_t* pattern) + { + while (*pattern && *string == *pattern) + { + string++; + pattern++; + } + + return *pattern == 0; + } + + PUGI__FN const char_t* find_char(const char_t* s, char_t c) + { + #ifdef PUGIXML_WCHAR_MODE + return wcschr(s, c); + #else + return strchr(s, c); + #endif + } + + PUGI__FN const char_t* find_substring(const char_t* s, const char_t* p) + { + #ifdef PUGIXML_WCHAR_MODE + // MSVC6 wcsstr bug workaround (if s is empty it always returns 0) + return (*p == 0) ? s : wcsstr(s, p); + #else + return strstr(s, p); + #endif + } + + // Converts symbol to lower case, if it is an ASCII one + PUGI__FN char_t tolower_ascii(char_t ch) + { + return static_cast(ch - 'A') < 26 ? static_cast(ch | ' ') : ch; + } + + PUGI__FN xpath_string string_value(const xpath_node& na, xpath_allocator* alloc) + { + if (na.attribute()) + return xpath_string::from_const(na.attribute().value()); + else + { + xml_node n = na.node(); + + switch (n.type()) + { + case node_pcdata: + case node_cdata: + case node_comment: + case node_pi: + return xpath_string::from_const(n.value()); + + case node_document: + case node_element: + { + xpath_string result; + + // element nodes can have value if parse_embed_pcdata was used + if (n.value()[0]) + result.append(xpath_string::from_const(n.value()), alloc); + + xml_node cur = n.first_child(); + + while (cur && cur != n) + { + if (cur.type() == node_pcdata || cur.type() == node_cdata) + result.append(xpath_string::from_const(cur.value()), alloc); + + if (cur.first_child()) + cur = cur.first_child(); + else if (cur.next_sibling()) + cur = cur.next_sibling(); + else + { + while (!cur.next_sibling() && cur != n) + cur = cur.parent(); + + if (cur != n) cur = cur.next_sibling(); + } + } + + return result; + } + + default: + return xpath_string(); + } + } + } + + PUGI__FN bool node_is_before_sibling(xml_node_struct* ln, xml_node_struct* rn) + { + assert(ln->parent == rn->parent); + + // there is no common ancestor (the shared parent is null), nodes are from different documents + if (!ln->parent) return ln < rn; + + // determine sibling order + xml_node_struct* ls = ln; + xml_node_struct* rs = rn; + + while (ls && rs) + { + if (ls == rn) return true; + if (rs == ln) return false; + + ls = ls->next_sibling; + rs = rs->next_sibling; + } + + // if rn sibling chain ended ln must be before rn + return !rs; + } + + PUGI__FN bool node_is_before(xml_node_struct* ln, xml_node_struct* rn) + { + // find common ancestor at the same depth, if any + xml_node_struct* lp = ln; + xml_node_struct* rp = rn; + + while (lp && rp && lp->parent != rp->parent) + { + lp = lp->parent; + rp = rp->parent; + } + + // parents are the same! + if (lp && rp) return node_is_before_sibling(lp, rp); + + // nodes are at different depths, need to normalize heights + bool left_higher = !lp; + + while (lp) + { + lp = lp->parent; + ln = ln->parent; + } + + while (rp) + { + rp = rp->parent; + rn = rn->parent; + } + + // one node is the ancestor of the other + if (ln == rn) return left_higher; + + // find common ancestor... again + while (ln->parent != rn->parent) + { + ln = ln->parent; + rn = rn->parent; + } + + return node_is_before_sibling(ln, rn); + } + + PUGI__FN bool node_is_ancestor(xml_node_struct* parent, xml_node_struct* node) + { + while (node && node != parent) node = node->parent; + + return parent && node == parent; + } + + PUGI__FN const void* document_buffer_order(const xpath_node& xnode) + { + xml_node_struct* node = xnode.node().internal_object(); + + if (node) + { + if ((get_document(node).header & xml_memory_page_contents_shared_mask) == 0) + { + if (node->name && (node->header & impl::xml_memory_page_name_allocated_or_shared_mask) == 0) return node->name; + if (node->value && (node->header & impl::xml_memory_page_value_allocated_or_shared_mask) == 0) return node->value; + } + + return 0; + } + + xml_attribute_struct* attr = xnode.attribute().internal_object(); + + if (attr) + { + if ((get_document(attr).header & xml_memory_page_contents_shared_mask) == 0) + { + if ((attr->header & impl::xml_memory_page_name_allocated_or_shared_mask) == 0) return attr->name; + if ((attr->header & impl::xml_memory_page_value_allocated_or_shared_mask) == 0) return attr->value; + } + + return 0; + } + + return 0; + } + + struct document_order_comparator + { + bool operator()(const xpath_node& lhs, const xpath_node& rhs) const + { + // optimized document order based check + const void* lo = document_buffer_order(lhs); + const void* ro = document_buffer_order(rhs); + + if (lo && ro) return lo < ro; + + // slow comparison + xml_node ln = lhs.node(), rn = rhs.node(); + + // compare attributes + if (lhs.attribute() && rhs.attribute()) + { + // shared parent + if (lhs.parent() == rhs.parent()) + { + // determine sibling order + for (xml_attribute a = lhs.attribute(); a; a = a.next_attribute()) + if (a == rhs.attribute()) + return true; + + return false; + } + + // compare attribute parents + ln = lhs.parent(); + rn = rhs.parent(); + } + else if (lhs.attribute()) + { + // attributes go after the parent element + if (lhs.parent() == rhs.node()) return false; + + ln = lhs.parent(); + } + else if (rhs.attribute()) + { + // attributes go after the parent element + if (rhs.parent() == lhs.node()) return true; + + rn = rhs.parent(); + } + + if (ln == rn) return false; + + if (!ln || !rn) return ln < rn; + + return node_is_before(ln.internal_object(), rn.internal_object()); + } + }; + + PUGI__FN double gen_nan() + { + #if defined(__STDC_IEC_559__) || ((FLT_RADIX - 0 == 2) && (FLT_MAX_EXP - 0 == 128) && (FLT_MANT_DIG - 0 == 24)) + PUGI__STATIC_ASSERT(sizeof(float) == sizeof(uint32_t)); + typedef uint32_t UI; // BCC5 workaround + union { float f; UI i; } u; + u.i = 0x7fc00000; + return double(u.f); + #else + // fallback + const volatile double zero = 0.0; + return zero / zero; + #endif + } + + PUGI__FN bool is_nan(double value) + { + #if defined(PUGI__MSVC_CRT_VERSION) || defined(__BORLANDC__) + return !!_isnan(value); + #elif defined(fpclassify) && defined(FP_NAN) + return fpclassify(value) == FP_NAN; + #else + // fallback + const volatile double v = value; + return v != v; + #endif + } + + PUGI__FN const char_t* convert_number_to_string_special(double value) + { + #if defined(PUGI__MSVC_CRT_VERSION) || defined(__BORLANDC__) + if (_finite(value)) return (value == 0) ? PUGIXML_TEXT("0") : 0; + if (_isnan(value)) return PUGIXML_TEXT("NaN"); + return value > 0 ? PUGIXML_TEXT("Infinity") : PUGIXML_TEXT("-Infinity"); + #elif defined(fpclassify) && defined(FP_NAN) && defined(FP_INFINITE) && defined(FP_ZERO) + switch (fpclassify(value)) + { + case FP_NAN: + return PUGIXML_TEXT("NaN"); + + case FP_INFINITE: + return value > 0 ? PUGIXML_TEXT("Infinity") : PUGIXML_TEXT("-Infinity"); + + case FP_ZERO: + return PUGIXML_TEXT("0"); + + default: + return 0; + } + #else + // fallback + const volatile double v = value; + + if (v == 0) return PUGIXML_TEXT("0"); + if (v != v) return PUGIXML_TEXT("NaN"); + if (v * 2 == v) return value > 0 ? PUGIXML_TEXT("Infinity") : PUGIXML_TEXT("-Infinity"); + return 0; + #endif + } + + PUGI__FN bool convert_number_to_boolean(double value) + { + return (value != 0 && !is_nan(value)); + } + + PUGI__FN void truncate_zeros(char* begin, char* end) + { + while (begin != end && end[-1] == '0') end--; + + *end = 0; + } + + // gets mantissa digits in the form of 0.xxxxx with 0. implied and the exponent +#if defined(PUGI__MSVC_CRT_VERSION) && PUGI__MSVC_CRT_VERSION >= 1400 && !defined(_WIN32_WCE) + PUGI__FN void convert_number_to_mantissa_exponent(double value, char (&buffer)[32], char** out_mantissa, int* out_exponent) + { + // get base values + int sign, exponent; + _ecvt_s(buffer, sizeof(buffer), value, DBL_DIG + 1, &exponent, &sign); + + // truncate redundant zeros + truncate_zeros(buffer, buffer + strlen(buffer)); + + // fill results + *out_mantissa = buffer; + *out_exponent = exponent; + } +#else + PUGI__FN void convert_number_to_mantissa_exponent(double value, char (&buffer)[32], char** out_mantissa, int* out_exponent) + { + // get a scientific notation value with IEEE DBL_DIG decimals + PUGI__SNPRINTF(buffer, "%.*e", DBL_DIG, value); + + // get the exponent (possibly negative) + char* exponent_string = strchr(buffer, 'e'); + assert(exponent_string); + + int exponent = atoi(exponent_string + 1); + + // extract mantissa string: skip sign + char* mantissa = buffer[0] == '-' ? buffer + 1 : buffer; + assert(mantissa[0] != '0' && mantissa[1] == '.'); + + // divide mantissa by 10 to eliminate integer part + mantissa[1] = mantissa[0]; + mantissa++; + exponent++; + + // remove extra mantissa digits and zero-terminate mantissa + truncate_zeros(mantissa, exponent_string); + + // fill results + *out_mantissa = mantissa; + *out_exponent = exponent; + } +#endif + + PUGI__FN xpath_string convert_number_to_string(double value, xpath_allocator* alloc) + { + // try special number conversion + const char_t* special = convert_number_to_string_special(value); + if (special) return xpath_string::from_const(special); + + // get mantissa + exponent form + char mantissa_buffer[32]; + + char* mantissa; + int exponent; + convert_number_to_mantissa_exponent(value, mantissa_buffer, &mantissa, &exponent); + + // allocate a buffer of suitable length for the number + size_t result_size = strlen(mantissa_buffer) + (exponent > 0 ? exponent : -exponent) + 4; + char_t* result = static_cast(alloc->allocate(sizeof(char_t) * result_size)); + if (!result) return xpath_string(); + + // make the number! + char_t* s = result; + + // sign + if (value < 0) *s++ = '-'; + + // integer part + if (exponent <= 0) + { + *s++ = '0'; + } + else + { + while (exponent > 0) + { + assert(*mantissa == 0 || static_cast(*mantissa - '0') <= 9); + *s++ = *mantissa ? *mantissa++ : '0'; + exponent--; + } + } + + // fractional part + if (*mantissa) + { + // decimal point + *s++ = '.'; + + // extra zeroes from negative exponent + while (exponent < 0) + { + *s++ = '0'; + exponent++; + } + + // extra mantissa digits + while (*mantissa) + { + assert(static_cast(*mantissa - '0') <= 9); + *s++ = *mantissa++; + } + } + + // zero-terminate + assert(s < result + result_size); + *s = 0; + + return xpath_string::from_heap_preallocated(result, s); + } + + PUGI__FN bool check_string_to_number_format(const char_t* string) + { + // parse leading whitespace + while (PUGI__IS_CHARTYPE(*string, ct_space)) ++string; + + // parse sign + if (*string == '-') ++string; + + if (!*string) return false; + + // if there is no integer part, there should be a decimal part with at least one digit + if (!PUGI__IS_CHARTYPEX(string[0], ctx_digit) && (string[0] != '.' || !PUGI__IS_CHARTYPEX(string[1], ctx_digit))) return false; + + // parse integer part + while (PUGI__IS_CHARTYPEX(*string, ctx_digit)) ++string; + + // parse decimal part + if (*string == '.') + { + ++string; + + while (PUGI__IS_CHARTYPEX(*string, ctx_digit)) ++string; + } + + // parse trailing whitespace + while (PUGI__IS_CHARTYPE(*string, ct_space)) ++string; + + return *string == 0; + } + + PUGI__FN double convert_string_to_number(const char_t* string) + { + // check string format + if (!check_string_to_number_format(string)) return gen_nan(); + + // parse string + #ifdef PUGIXML_WCHAR_MODE + return wcstod(string, 0); + #else + return strtod(string, 0); + #endif + } + + PUGI__FN bool convert_string_to_number_scratch(char_t (&buffer)[32], const char_t* begin, const char_t* end, double* out_result) + { + size_t length = static_cast(end - begin); + char_t* scratch = buffer; + + if (length >= sizeof(buffer) / sizeof(buffer[0])) + { + // need to make dummy on-heap copy + scratch = static_cast(xml_memory::allocate((length + 1) * sizeof(char_t))); + if (!scratch) return false; + } + + // copy string to zero-terminated buffer and perform conversion + memcpy(scratch, begin, length * sizeof(char_t)); + scratch[length] = 0; + + *out_result = convert_string_to_number(scratch); + + // free dummy buffer + if (scratch != buffer) xml_memory::deallocate(scratch); + + return true; + } + + PUGI__FN double round_nearest(double value) + { + return floor(value + 0.5); + } + + PUGI__FN double round_nearest_nzero(double value) + { + // same as round_nearest, but returns -0 for [-0.5, -0] + // ceil is used to differentiate between +0 and -0 (we return -0 for [-0.5, -0] and +0 for +0) + return (value >= -0.5 && value <= 0) ? ceil(value) : floor(value + 0.5); + } + + PUGI__FN const char_t* qualified_name(const xpath_node& node) + { + return node.attribute() ? node.attribute().name() : node.node().name(); + } + + PUGI__FN const char_t* local_name(const xpath_node& node) + { + const char_t* name = qualified_name(node); + const char_t* p = find_char(name, ':'); + + return p ? p + 1 : name; + } + + struct namespace_uri_predicate + { + const char_t* prefix; + size_t prefix_length; + + namespace_uri_predicate(const char_t* name) + { + const char_t* pos = find_char(name, ':'); + + prefix = pos ? name : 0; + prefix_length = pos ? static_cast(pos - name) : 0; + } + + bool operator()(xml_attribute a) const + { + const char_t* name = a.name(); + + if (!starts_with(name, PUGIXML_TEXT("xmlns"))) return false; + + return prefix ? name[5] == ':' && strequalrange(name + 6, prefix, prefix_length) : name[5] == 0; + } + }; + + PUGI__FN const char_t* namespace_uri(xml_node node) + { + namespace_uri_predicate pred = node.name(); + + xml_node p = node; + + while (p) + { + xml_attribute a = p.find_attribute(pred); + + if (a) return a.value(); + + p = p.parent(); + } + + return PUGIXML_TEXT(""); + } + + PUGI__FN const char_t* namespace_uri(xml_attribute attr, xml_node parent) + { + namespace_uri_predicate pred = attr.name(); + + // Default namespace does not apply to attributes + if (!pred.prefix) return PUGIXML_TEXT(""); + + xml_node p = parent; + + while (p) + { + xml_attribute a = p.find_attribute(pred); + + if (a) return a.value(); + + p = p.parent(); + } + + return PUGIXML_TEXT(""); + } + + PUGI__FN const char_t* namespace_uri(const xpath_node& node) + { + return node.attribute() ? namespace_uri(node.attribute(), node.parent()) : namespace_uri(node.node()); + } + + PUGI__FN char_t* normalize_space(char_t* buffer) + { + char_t* write = buffer; + + for (char_t* it = buffer; *it; ) + { + char_t ch = *it++; + + if (PUGI__IS_CHARTYPE(ch, ct_space)) + { + // replace whitespace sequence with single space + while (PUGI__IS_CHARTYPE(*it, ct_space)) it++; + + // avoid leading spaces + if (write != buffer) *write++ = ' '; + } + else *write++ = ch; + } + + // remove trailing space + if (write != buffer && PUGI__IS_CHARTYPE(write[-1], ct_space)) write--; + + // zero-terminate + *write = 0; + + return write; + } + + PUGI__FN char_t* translate(char_t* buffer, const char_t* from, const char_t* to, size_t to_length) + { + char_t* write = buffer; + + while (*buffer) + { + PUGI__DMC_VOLATILE char_t ch = *buffer++; + + const char_t* pos = find_char(from, ch); + + if (!pos) + *write++ = ch; // do not process + else if (static_cast(pos - from) < to_length) + *write++ = to[pos - from]; // replace + } + + // zero-terminate + *write = 0; + + return write; + } + + PUGI__FN unsigned char* translate_table_generate(xpath_allocator* alloc, const char_t* from, const char_t* to) + { + unsigned char table[128] = {0}; + + while (*from) + { + unsigned int fc = static_cast(*from); + unsigned int tc = static_cast(*to); + + if (fc >= 128 || tc >= 128) + return 0; + + // code=128 means "skip character" + if (!table[fc]) + table[fc] = static_cast(tc ? tc : 128); + + from++; + if (tc) to++; + } + + for (int i = 0; i < 128; ++i) + if (!table[i]) + table[i] = static_cast(i); + + void* result = alloc->allocate(sizeof(table)); + if (!result) return 0; + + memcpy(result, table, sizeof(table)); + + return static_cast(result); + } + + PUGI__FN char_t* translate_table(char_t* buffer, const unsigned char* table) + { + char_t* write = buffer; + + while (*buffer) + { + char_t ch = *buffer++; + unsigned int index = static_cast(ch); + + if (index < 128) + { + unsigned char code = table[index]; + + // code=128 means "skip character" (table size is 128 so 128 can be a special value) + // this code skips these characters without extra branches + *write = static_cast(code); + write += 1 - (code >> 7); + } + else + { + *write++ = ch; + } + } + + // zero-terminate + *write = 0; + + return write; + } + + inline bool is_xpath_attribute(const char_t* name) + { + return !(starts_with(name, PUGIXML_TEXT("xmlns")) && (name[5] == 0 || name[5] == ':')); + } + + struct xpath_variable_boolean: xpath_variable + { + xpath_variable_boolean(): xpath_variable(xpath_type_boolean), value(false) + { + } + + bool value; + char_t name[1]; + }; + + struct xpath_variable_number: xpath_variable + { + xpath_variable_number(): xpath_variable(xpath_type_number), value(0) + { + } + + double value; + char_t name[1]; + }; + + struct xpath_variable_string: xpath_variable + { + xpath_variable_string(): xpath_variable(xpath_type_string), value(0) + { + } + + ~xpath_variable_string() + { + if (value) xml_memory::deallocate(value); + } + + char_t* value; + char_t name[1]; + }; + + struct xpath_variable_node_set: xpath_variable + { + xpath_variable_node_set(): xpath_variable(xpath_type_node_set) + { + } + + xpath_node_set value; + char_t name[1]; + }; + + static const xpath_node_set dummy_node_set; + + PUGI__FN PUGI__UNSIGNED_OVERFLOW unsigned int hash_string(const char_t* str) + { + // Jenkins one-at-a-time hash (http://en.wikipedia.org/wiki/Jenkins_hash_function#one-at-a-time) + unsigned int result = 0; + + while (*str) + { + result += static_cast(*str++); + result += result << 10; + result ^= result >> 6; + } + + result += result << 3; + result ^= result >> 11; + result += result << 15; + + return result; + } + + template PUGI__FN T* new_xpath_variable(const char_t* name) + { + size_t length = strlength(name); + if (length == 0) return 0; // empty variable names are invalid + + // $$ we can't use offsetof(T, name) because T is non-POD, so we just allocate additional length characters + void* memory = xml_memory::allocate(sizeof(T) + length * sizeof(char_t)); + if (!memory) return 0; + + T* result = new (memory) T(); + + memcpy(result->name, name, (length + 1) * sizeof(char_t)); + + return result; + } + + PUGI__FN xpath_variable* new_xpath_variable(xpath_value_type type, const char_t* name) + { + switch (type) + { + case xpath_type_node_set: + return new_xpath_variable(name); + + case xpath_type_number: + return new_xpath_variable(name); + + case xpath_type_string: + return new_xpath_variable(name); + + case xpath_type_boolean: + return new_xpath_variable(name); + + default: + return 0; + } + } + + template PUGI__FN void delete_xpath_variable(T* var) + { + var->~T(); + xml_memory::deallocate(var); + } + + PUGI__FN void delete_xpath_variable(xpath_value_type type, xpath_variable* var) + { + switch (type) + { + case xpath_type_node_set: + delete_xpath_variable(static_cast(var)); + break; + + case xpath_type_number: + delete_xpath_variable(static_cast(var)); + break; + + case xpath_type_string: + delete_xpath_variable(static_cast(var)); + break; + + case xpath_type_boolean: + delete_xpath_variable(static_cast(var)); + break; + + default: + assert(false && "Invalid variable type"); // unreachable + } + } + + PUGI__FN bool copy_xpath_variable(xpath_variable* lhs, const xpath_variable* rhs) + { + switch (rhs->type()) + { + case xpath_type_node_set: + return lhs->set(static_cast(rhs)->value); + + case xpath_type_number: + return lhs->set(static_cast(rhs)->value); + + case xpath_type_string: + return lhs->set(static_cast(rhs)->value); + + case xpath_type_boolean: + return lhs->set(static_cast(rhs)->value); + + default: + assert(false && "Invalid variable type"); // unreachable + return false; + } + } + + PUGI__FN bool get_variable_scratch(char_t (&buffer)[32], xpath_variable_set* set, const char_t* begin, const char_t* end, xpath_variable** out_result) + { + size_t length = static_cast(end - begin); + char_t* scratch = buffer; + + if (length >= sizeof(buffer) / sizeof(buffer[0])) + { + // need to make dummy on-heap copy + scratch = static_cast(xml_memory::allocate((length + 1) * sizeof(char_t))); + if (!scratch) return false; + } + + // copy string to zero-terminated buffer and perform lookup + memcpy(scratch, begin, length * sizeof(char_t)); + scratch[length] = 0; + + *out_result = set->get(scratch); + + // free dummy buffer + if (scratch != buffer) xml_memory::deallocate(scratch); + + return true; + } +PUGI__NS_END + +// Internal node set class +PUGI__NS_BEGIN + PUGI__FN xpath_node_set::type_t xpath_get_order(const xpath_node* begin, const xpath_node* end) + { + if (end - begin < 2) + return xpath_node_set::type_sorted; + + document_order_comparator cmp; + + bool first = cmp(begin[0], begin[1]); + + for (const xpath_node* it = begin + 1; it + 1 < end; ++it) + if (cmp(it[0], it[1]) != first) + return xpath_node_set::type_unsorted; + + return first ? xpath_node_set::type_sorted : xpath_node_set::type_sorted_reverse; + } + + PUGI__FN xpath_node_set::type_t xpath_sort(xpath_node* begin, xpath_node* end, xpath_node_set::type_t type, bool rev) + { + xpath_node_set::type_t order = rev ? xpath_node_set::type_sorted_reverse : xpath_node_set::type_sorted; + + if (type == xpath_node_set::type_unsorted) + { + xpath_node_set::type_t sorted = xpath_get_order(begin, end); + + if (sorted == xpath_node_set::type_unsorted) + { + sort(begin, end, document_order_comparator()); + + type = xpath_node_set::type_sorted; + } + else + type = sorted; + } + + if (type != order) reverse(begin, end); + + return order; + } + + PUGI__FN xpath_node xpath_first(const xpath_node* begin, const xpath_node* end, xpath_node_set::type_t type) + { + if (begin == end) return xpath_node(); + + switch (type) + { + case xpath_node_set::type_sorted: + return *begin; + + case xpath_node_set::type_sorted_reverse: + return *(end - 1); + + case xpath_node_set::type_unsorted: + return *min_element(begin, end, document_order_comparator()); + + default: + assert(false && "Invalid node set type"); // unreachable + return xpath_node(); + } + } + + class xpath_node_set_raw + { + xpath_node_set::type_t _type; + + xpath_node* _begin; + xpath_node* _end; + xpath_node* _eos; + + public: + xpath_node_set_raw(): _type(xpath_node_set::type_unsorted), _begin(0), _end(0), _eos(0) + { + } + + xpath_node* begin() const + { + return _begin; + } + + xpath_node* end() const + { + return _end; + } + + bool empty() const + { + return _begin == _end; + } + + size_t size() const + { + return static_cast(_end - _begin); + } + + xpath_node first() const + { + return xpath_first(_begin, _end, _type); + } + + void push_back_grow(const xpath_node& node, xpath_allocator* alloc); + + void push_back(const xpath_node& node, xpath_allocator* alloc) + { + if (_end != _eos) + *_end++ = node; + else + push_back_grow(node, alloc); + } + + void append(const xpath_node* begin_, const xpath_node* end_, xpath_allocator* alloc) + { + if (begin_ == end_) return; + + size_t size_ = static_cast(_end - _begin); + size_t capacity = static_cast(_eos - _begin); + size_t count = static_cast(end_ - begin_); + + if (size_ + count > capacity) + { + // reallocate the old array or allocate a new one + xpath_node* data = static_cast(alloc->reallocate(_begin, capacity * sizeof(xpath_node), (size_ + count) * sizeof(xpath_node))); + if (!data) return; + + // finalize + _begin = data; + _end = data + size_; + _eos = data + size_ + count; + } + + memcpy(_end, begin_, count * sizeof(xpath_node)); + _end += count; + } + + void sort_do() + { + _type = xpath_sort(_begin, _end, _type, false); + } + + void truncate(xpath_node* pos) + { + assert(_begin <= pos && pos <= _end); + + _end = pos; + } + + void remove_duplicates(xpath_allocator* alloc) + { + if (_type == xpath_node_set::type_unsorted && _end - _begin > 2) + { + xpath_allocator_capture cr(alloc); + + size_t size_ = static_cast(_end - _begin); + + size_t hash_size = 1; + while (hash_size < size_ + size_ / 2) hash_size *= 2; + + const void** hash_data = static_cast(alloc->allocate(hash_size * sizeof(void**))); + if (!hash_data) return; + + memset(hash_data, 0, hash_size * sizeof(const void**)); + + xpath_node* write = _begin; + + for (xpath_node* it = _begin; it != _end; ++it) + { + const void* attr = it->attribute().internal_object(); + const void* node = it->node().internal_object(); + const void* key = attr ? attr : node; + + if (key && hash_insert(hash_data, hash_size, key)) + { + *write++ = *it; + } + } + + _end = write; + } + else + { + _end = unique(_begin, _end); + } + } + + xpath_node_set::type_t type() const + { + return _type; + } + + void set_type(xpath_node_set::type_t value) + { + _type = value; + } + }; + + PUGI__FN_NO_INLINE void xpath_node_set_raw::push_back_grow(const xpath_node& node, xpath_allocator* alloc) + { + size_t capacity = static_cast(_eos - _begin); + + // get new capacity (1.5x rule) + size_t new_capacity = capacity + capacity / 2 + 1; + + // reallocate the old array or allocate a new one + xpath_node* data = static_cast(alloc->reallocate(_begin, capacity * sizeof(xpath_node), new_capacity * sizeof(xpath_node))); + if (!data) return; + + // finalize + _begin = data; + _end = data + capacity; + _eos = data + new_capacity; + + // push + *_end++ = node; + } +PUGI__NS_END + +PUGI__NS_BEGIN + struct xpath_context + { + xpath_node n; + size_t position, size; + + xpath_context(const xpath_node& n_, size_t position_, size_t size_): n(n_), position(position_), size(size_) + { + } + }; + + enum lexeme_t + { + lex_none = 0, + lex_equal, + lex_not_equal, + lex_less, + lex_greater, + lex_less_or_equal, + lex_greater_or_equal, + lex_plus, + lex_minus, + lex_multiply, + lex_union, + lex_var_ref, + lex_open_brace, + lex_close_brace, + lex_quoted_string, + lex_number, + lex_slash, + lex_double_slash, + lex_open_square_brace, + lex_close_square_brace, + lex_string, + lex_comma, + lex_axis_attribute, + lex_dot, + lex_double_dot, + lex_double_colon, + lex_eof + }; + + struct xpath_lexer_string + { + const char_t* begin; + const char_t* end; + + xpath_lexer_string(): begin(0), end(0) + { + } + + bool operator==(const char_t* other) const + { + size_t length = static_cast(end - begin); + + return strequalrange(other, begin, length); + } + }; + + class xpath_lexer + { + const char_t* _cur; + const char_t* _cur_lexeme_pos; + xpath_lexer_string _cur_lexeme_contents; + + lexeme_t _cur_lexeme; + + public: + explicit xpath_lexer(const char_t* query): _cur(query) + { + next(); + } + + const char_t* state() const + { + return _cur; + } + + void next() + { + const char_t* cur = _cur; + + while (PUGI__IS_CHARTYPE(*cur, ct_space)) ++cur; + + // save lexeme position for error reporting + _cur_lexeme_pos = cur; + + switch (*cur) + { + case 0: + _cur_lexeme = lex_eof; + break; + + case '>': + if (*(cur+1) == '=') + { + cur += 2; + _cur_lexeme = lex_greater_or_equal; + } + else + { + cur += 1; + _cur_lexeme = lex_greater; + } + break; + + case '<': + if (*(cur+1) == '=') + { + cur += 2; + _cur_lexeme = lex_less_or_equal; + } + else + { + cur += 1; + _cur_lexeme = lex_less; + } + break; + + case '!': + if (*(cur+1) == '=') + { + cur += 2; + _cur_lexeme = lex_not_equal; + } + else + { + _cur_lexeme = lex_none; + } + break; + + case '=': + cur += 1; + _cur_lexeme = lex_equal; + + break; + + case '+': + cur += 1; + _cur_lexeme = lex_plus; + + break; + + case '-': + cur += 1; + _cur_lexeme = lex_minus; + + break; + + case '*': + cur += 1; + _cur_lexeme = lex_multiply; + + break; + + case '|': + cur += 1; + _cur_lexeme = lex_union; + + break; + + case '$': + cur += 1; + + if (PUGI__IS_CHARTYPEX(*cur, ctx_start_symbol)) + { + _cur_lexeme_contents.begin = cur; + + while (PUGI__IS_CHARTYPEX(*cur, ctx_symbol)) cur++; + + if (cur[0] == ':' && PUGI__IS_CHARTYPEX(cur[1], ctx_symbol)) // qname + { + cur++; // : + + while (PUGI__IS_CHARTYPEX(*cur, ctx_symbol)) cur++; + } + + _cur_lexeme_contents.end = cur; + + _cur_lexeme = lex_var_ref; + } + else + { + _cur_lexeme = lex_none; + } + + break; + + case '(': + cur += 1; + _cur_lexeme = lex_open_brace; + + break; + + case ')': + cur += 1; + _cur_lexeme = lex_close_brace; + + break; + + case '[': + cur += 1; + _cur_lexeme = lex_open_square_brace; + + break; + + case ']': + cur += 1; + _cur_lexeme = lex_close_square_brace; + + break; + + case ',': + cur += 1; + _cur_lexeme = lex_comma; + + break; + + case '/': + if (*(cur+1) == '/') + { + cur += 2; + _cur_lexeme = lex_double_slash; + } + else + { + cur += 1; + _cur_lexeme = lex_slash; + } + break; + + case '.': + if (*(cur+1) == '.') + { + cur += 2; + _cur_lexeme = lex_double_dot; + } + else if (PUGI__IS_CHARTYPEX(*(cur+1), ctx_digit)) + { + _cur_lexeme_contents.begin = cur; // . + + ++cur; + + while (PUGI__IS_CHARTYPEX(*cur, ctx_digit)) cur++; + + _cur_lexeme_contents.end = cur; + + _cur_lexeme = lex_number; + } + else + { + cur += 1; + _cur_lexeme = lex_dot; + } + break; + + case '@': + cur += 1; + _cur_lexeme = lex_axis_attribute; + + break; + + case '"': + case '\'': + { + char_t terminator = *cur; + + ++cur; + + _cur_lexeme_contents.begin = cur; + while (*cur && *cur != terminator) cur++; + _cur_lexeme_contents.end = cur; + + if (!*cur) + _cur_lexeme = lex_none; + else + { + cur += 1; + _cur_lexeme = lex_quoted_string; + } + + break; + } + + case ':': + if (*(cur+1) == ':') + { + cur += 2; + _cur_lexeme = lex_double_colon; + } + else + { + _cur_lexeme = lex_none; + } + break; + + default: + if (PUGI__IS_CHARTYPEX(*cur, ctx_digit)) + { + _cur_lexeme_contents.begin = cur; + + while (PUGI__IS_CHARTYPEX(*cur, ctx_digit)) cur++; + + if (*cur == '.') + { + cur++; + + while (PUGI__IS_CHARTYPEX(*cur, ctx_digit)) cur++; + } + + _cur_lexeme_contents.end = cur; + + _cur_lexeme = lex_number; + } + else if (PUGI__IS_CHARTYPEX(*cur, ctx_start_symbol)) + { + _cur_lexeme_contents.begin = cur; + + while (PUGI__IS_CHARTYPEX(*cur, ctx_symbol)) cur++; + + if (cur[0] == ':') + { + if (cur[1] == '*') // namespace test ncname:* + { + cur += 2; // :* + } + else if (PUGI__IS_CHARTYPEX(cur[1], ctx_symbol)) // namespace test qname + { + cur++; // : + + while (PUGI__IS_CHARTYPEX(*cur, ctx_symbol)) cur++; + } + } + + _cur_lexeme_contents.end = cur; + + _cur_lexeme = lex_string; + } + else + { + _cur_lexeme = lex_none; + } + } + + _cur = cur; + } + + lexeme_t current() const + { + return _cur_lexeme; + } + + const char_t* current_pos() const + { + return _cur_lexeme_pos; + } + + const xpath_lexer_string& contents() const + { + assert(_cur_lexeme == lex_var_ref || _cur_lexeme == lex_number || _cur_lexeme == lex_string || _cur_lexeme == lex_quoted_string); + + return _cur_lexeme_contents; + } + }; + + enum ast_type_t + { + ast_unknown, + ast_op_or, // left or right + ast_op_and, // left and right + ast_op_equal, // left = right + ast_op_not_equal, // left != right + ast_op_less, // left < right + ast_op_greater, // left > right + ast_op_less_or_equal, // left <= right + ast_op_greater_or_equal, // left >= right + ast_op_add, // left + right + ast_op_subtract, // left - right + ast_op_multiply, // left * right + ast_op_divide, // left / right + ast_op_mod, // left % right + ast_op_negate, // left - right + ast_op_union, // left | right + ast_predicate, // apply predicate to set; next points to next predicate + ast_filter, // select * from left where right + ast_string_constant, // string constant + ast_number_constant, // number constant + ast_variable, // variable + ast_func_last, // last() + ast_func_position, // position() + ast_func_count, // count(left) + ast_func_id, // id(left) + ast_func_local_name_0, // local-name() + ast_func_local_name_1, // local-name(left) + ast_func_namespace_uri_0, // namespace-uri() + ast_func_namespace_uri_1, // namespace-uri(left) + ast_func_name_0, // name() + ast_func_name_1, // name(left) + ast_func_string_0, // string() + ast_func_string_1, // string(left) + ast_func_concat, // concat(left, right, siblings) + ast_func_starts_with, // starts_with(left, right) + ast_func_contains, // contains(left, right) + ast_func_substring_before, // substring-before(left, right) + ast_func_substring_after, // substring-after(left, right) + ast_func_substring_2, // substring(left, right) + ast_func_substring_3, // substring(left, right, third) + ast_func_string_length_0, // string-length() + ast_func_string_length_1, // string-length(left) + ast_func_normalize_space_0, // normalize-space() + ast_func_normalize_space_1, // normalize-space(left) + ast_func_translate, // translate(left, right, third) + ast_func_boolean, // boolean(left) + ast_func_not, // not(left) + ast_func_true, // true() + ast_func_false, // false() + ast_func_lang, // lang(left) + ast_func_number_0, // number() + ast_func_number_1, // number(left) + ast_func_sum, // sum(left) + ast_func_floor, // floor(left) + ast_func_ceiling, // ceiling(left) + ast_func_round, // round(left) + ast_step, // process set left with step + ast_step_root, // select root node + + ast_opt_translate_table, // translate(left, right, third) where right/third are constants + ast_opt_compare_attribute // @name = 'string' + }; + + enum axis_t + { + axis_ancestor, + axis_ancestor_or_self, + axis_attribute, + axis_child, + axis_descendant, + axis_descendant_or_self, + axis_following, + axis_following_sibling, + axis_namespace, + axis_parent, + axis_preceding, + axis_preceding_sibling, + axis_self + }; + + enum nodetest_t + { + nodetest_none, + nodetest_name, + nodetest_type_node, + nodetest_type_comment, + nodetest_type_pi, + nodetest_type_text, + nodetest_pi, + nodetest_all, + nodetest_all_in_namespace + }; + + enum predicate_t + { + predicate_default, + predicate_posinv, + predicate_constant, + predicate_constant_one + }; + + enum nodeset_eval_t + { + nodeset_eval_all, + nodeset_eval_any, + nodeset_eval_first + }; + + template struct axis_to_type + { + static const axis_t axis; + }; + + template const axis_t axis_to_type::axis = N; + + class xpath_ast_node + { + private: + // node type + char _type; + char _rettype; + + // for ast_step + char _axis; + + // for ast_step/ast_predicate/ast_filter + char _test; + + // tree node structure + xpath_ast_node* _left; + xpath_ast_node* _right; + xpath_ast_node* _next; + + union + { + // value for ast_string_constant + const char_t* string; + // value for ast_number_constant + double number; + // variable for ast_variable + xpath_variable* variable; + // node test for ast_step (node name/namespace/node type/pi target) + const char_t* nodetest; + // table for ast_opt_translate_table + const unsigned char* table; + } _data; + + xpath_ast_node(const xpath_ast_node&); + xpath_ast_node& operator=(const xpath_ast_node&); + + template static bool compare_eq(xpath_ast_node* lhs, xpath_ast_node* rhs, const xpath_context& c, const xpath_stack& stack, const Comp& comp) + { + xpath_value_type lt = lhs->rettype(), rt = rhs->rettype(); + + if (lt != xpath_type_node_set && rt != xpath_type_node_set) + { + if (lt == xpath_type_boolean || rt == xpath_type_boolean) + return comp(lhs->eval_boolean(c, stack), rhs->eval_boolean(c, stack)); + else if (lt == xpath_type_number || rt == xpath_type_number) + return comp(lhs->eval_number(c, stack), rhs->eval_number(c, stack)); + else if (lt == xpath_type_string || rt == xpath_type_string) + { + xpath_allocator_capture cr(stack.result); + + xpath_string ls = lhs->eval_string(c, stack); + xpath_string rs = rhs->eval_string(c, stack); + + return comp(ls, rs); + } + } + else if (lt == xpath_type_node_set && rt == xpath_type_node_set) + { + xpath_allocator_capture cr(stack.result); + + xpath_node_set_raw ls = lhs->eval_node_set(c, stack, nodeset_eval_all); + xpath_node_set_raw rs = rhs->eval_node_set(c, stack, nodeset_eval_all); + + for (const xpath_node* li = ls.begin(); li != ls.end(); ++li) + for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri) + { + xpath_allocator_capture cri(stack.result); + + if (comp(string_value(*li, stack.result), string_value(*ri, stack.result))) + return true; + } + + return false; + } + else + { + if (lt == xpath_type_node_set) + { + swap(lhs, rhs); + swap(lt, rt); + } + + if (lt == xpath_type_boolean) + return comp(lhs->eval_boolean(c, stack), rhs->eval_boolean(c, stack)); + else if (lt == xpath_type_number) + { + xpath_allocator_capture cr(stack.result); + + double l = lhs->eval_number(c, stack); + xpath_node_set_raw rs = rhs->eval_node_set(c, stack, nodeset_eval_all); + + for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri) + { + xpath_allocator_capture cri(stack.result); + + if (comp(l, convert_string_to_number(string_value(*ri, stack.result).c_str()))) + return true; + } + + return false; + } + else if (lt == xpath_type_string) + { + xpath_allocator_capture cr(stack.result); + + xpath_string l = lhs->eval_string(c, stack); + xpath_node_set_raw rs = rhs->eval_node_set(c, stack, nodeset_eval_all); + + for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri) + { + xpath_allocator_capture cri(stack.result); + + if (comp(l, string_value(*ri, stack.result))) + return true; + } + + return false; + } + } + + assert(false && "Wrong types"); // unreachable + return false; + } + + static bool eval_once(xpath_node_set::type_t type, nodeset_eval_t eval) + { + return type == xpath_node_set::type_sorted ? eval != nodeset_eval_all : eval == nodeset_eval_any; + } + + template static bool compare_rel(xpath_ast_node* lhs, xpath_ast_node* rhs, const xpath_context& c, const xpath_stack& stack, const Comp& comp) + { + xpath_value_type lt = lhs->rettype(), rt = rhs->rettype(); + + if (lt != xpath_type_node_set && rt != xpath_type_node_set) + return comp(lhs->eval_number(c, stack), rhs->eval_number(c, stack)); + else if (lt == xpath_type_node_set && rt == xpath_type_node_set) + { + xpath_allocator_capture cr(stack.result); + + xpath_node_set_raw ls = lhs->eval_node_set(c, stack, nodeset_eval_all); + xpath_node_set_raw rs = rhs->eval_node_set(c, stack, nodeset_eval_all); + + for (const xpath_node* li = ls.begin(); li != ls.end(); ++li) + { + xpath_allocator_capture cri(stack.result); + + double l = convert_string_to_number(string_value(*li, stack.result).c_str()); + + for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri) + { + xpath_allocator_capture crii(stack.result); + + if (comp(l, convert_string_to_number(string_value(*ri, stack.result).c_str()))) + return true; + } + } + + return false; + } + else if (lt != xpath_type_node_set && rt == xpath_type_node_set) + { + xpath_allocator_capture cr(stack.result); + + double l = lhs->eval_number(c, stack); + xpath_node_set_raw rs = rhs->eval_node_set(c, stack, nodeset_eval_all); + + for (const xpath_node* ri = rs.begin(); ri != rs.end(); ++ri) + { + xpath_allocator_capture cri(stack.result); + + if (comp(l, convert_string_to_number(string_value(*ri, stack.result).c_str()))) + return true; + } + + return false; + } + else if (lt == xpath_type_node_set && rt != xpath_type_node_set) + { + xpath_allocator_capture cr(stack.result); + + xpath_node_set_raw ls = lhs->eval_node_set(c, stack, nodeset_eval_all); + double r = rhs->eval_number(c, stack); + + for (const xpath_node* li = ls.begin(); li != ls.end(); ++li) + { + xpath_allocator_capture cri(stack.result); + + if (comp(convert_string_to_number(string_value(*li, stack.result).c_str()), r)) + return true; + } + + return false; + } + else + { + assert(false && "Wrong types"); // unreachable + return false; + } + } + + static void apply_predicate_boolean(xpath_node_set_raw& ns, size_t first, xpath_ast_node* expr, const xpath_stack& stack, bool once) + { + assert(ns.size() >= first); + assert(expr->rettype() != xpath_type_number); + + size_t i = 1; + size_t size = ns.size() - first; + + xpath_node* last = ns.begin() + first; + + // remove_if... or well, sort of + for (xpath_node* it = last; it != ns.end(); ++it, ++i) + { + xpath_context c(*it, i, size); + + if (expr->eval_boolean(c, stack)) + { + *last++ = *it; + + if (once) break; + } + } + + ns.truncate(last); + } + + static void apply_predicate_number(xpath_node_set_raw& ns, size_t first, xpath_ast_node* expr, const xpath_stack& stack, bool once) + { + assert(ns.size() >= first); + assert(expr->rettype() == xpath_type_number); + + size_t i = 1; + size_t size = ns.size() - first; + + xpath_node* last = ns.begin() + first; + + // remove_if... or well, sort of + for (xpath_node* it = last; it != ns.end(); ++it, ++i) + { + xpath_context c(*it, i, size); + + if (expr->eval_number(c, stack) == i) + { + *last++ = *it; + + if (once) break; + } + } + + ns.truncate(last); + } + + static void apply_predicate_number_const(xpath_node_set_raw& ns, size_t first, xpath_ast_node* expr, const xpath_stack& stack) + { + assert(ns.size() >= first); + assert(expr->rettype() == xpath_type_number); + + size_t size = ns.size() - first; + + xpath_node* last = ns.begin() + first; + + xpath_context c(xpath_node(), 1, size); + + double er = expr->eval_number(c, stack); + + if (er >= 1.0 && er <= size) + { + size_t eri = static_cast(er); + + if (er == eri) + { + xpath_node r = last[eri - 1]; + + *last++ = r; + } + } + + ns.truncate(last); + } + + void apply_predicate(xpath_node_set_raw& ns, size_t first, const xpath_stack& stack, bool once) + { + if (ns.size() == first) return; + + assert(_type == ast_filter || _type == ast_predicate); + + if (_test == predicate_constant || _test == predicate_constant_one) + apply_predicate_number_const(ns, first, _right, stack); + else if (_right->rettype() == xpath_type_number) + apply_predicate_number(ns, first, _right, stack, once); + else + apply_predicate_boolean(ns, first, _right, stack, once); + } + + void apply_predicates(xpath_node_set_raw& ns, size_t first, const xpath_stack& stack, nodeset_eval_t eval) + { + if (ns.size() == first) return; + + bool last_once = eval_once(ns.type(), eval); + + for (xpath_ast_node* pred = _right; pred; pred = pred->_next) + pred->apply_predicate(ns, first, stack, !pred->_next && last_once); + } + + bool step_push(xpath_node_set_raw& ns, xml_attribute_struct* a, xml_node_struct* parent, xpath_allocator* alloc) + { + assert(a); + + const char_t* name = a->name ? a->name + 0 : PUGIXML_TEXT(""); + + switch (_test) + { + case nodetest_name: + if (strequal(name, _data.nodetest) && is_xpath_attribute(name)) + { + ns.push_back(xpath_node(xml_attribute(a), xml_node(parent)), alloc); + return true; + } + break; + + case nodetest_type_node: + case nodetest_all: + if (is_xpath_attribute(name)) + { + ns.push_back(xpath_node(xml_attribute(a), xml_node(parent)), alloc); + return true; + } + break; + + case nodetest_all_in_namespace: + if (starts_with(name, _data.nodetest) && is_xpath_attribute(name)) + { + ns.push_back(xpath_node(xml_attribute(a), xml_node(parent)), alloc); + return true; + } + break; + + default: + ; + } + + return false; + } + + bool step_push(xpath_node_set_raw& ns, xml_node_struct* n, xpath_allocator* alloc) + { + assert(n); + + xml_node_type type = PUGI__NODETYPE(n); + + switch (_test) + { + case nodetest_name: + if (type == node_element && n->name && strequal(n->name, _data.nodetest)) + { + ns.push_back(xml_node(n), alloc); + return true; + } + break; + + case nodetest_type_node: + ns.push_back(xml_node(n), alloc); + return true; + + case nodetest_type_comment: + if (type == node_comment) + { + ns.push_back(xml_node(n), alloc); + return true; + } + break; + + case nodetest_type_text: + if (type == node_pcdata || type == node_cdata) + { + ns.push_back(xml_node(n), alloc); + return true; + } + break; + + case nodetest_type_pi: + if (type == node_pi) + { + ns.push_back(xml_node(n), alloc); + return true; + } + break; + + case nodetest_pi: + if (type == node_pi && n->name && strequal(n->name, _data.nodetest)) + { + ns.push_back(xml_node(n), alloc); + return true; + } + break; + + case nodetest_all: + if (type == node_element) + { + ns.push_back(xml_node(n), alloc); + return true; + } + break; + + case nodetest_all_in_namespace: + if (type == node_element && n->name && starts_with(n->name, _data.nodetest)) + { + ns.push_back(xml_node(n), alloc); + return true; + } + break; + + default: + assert(false && "Unknown axis"); // unreachable + } + + return false; + } + + template void step_fill(xpath_node_set_raw& ns, xml_node_struct* n, xpath_allocator* alloc, bool once, T) + { + const axis_t axis = T::axis; + + switch (axis) + { + case axis_attribute: + { + for (xml_attribute_struct* a = n->first_attribute; a; a = a->next_attribute) + if (step_push(ns, a, n, alloc) & once) + return; + + break; + } + + case axis_child: + { + for (xml_node_struct* c = n->first_child; c; c = c->next_sibling) + if (step_push(ns, c, alloc) & once) + return; + + break; + } + + case axis_descendant: + case axis_descendant_or_self: + { + if (axis == axis_descendant_or_self) + if (step_push(ns, n, alloc) & once) + return; + + xml_node_struct* cur = n->first_child; + + while (cur) + { + if (step_push(ns, cur, alloc) & once) + return; + + if (cur->first_child) + cur = cur->first_child; + else + { + while (!cur->next_sibling) + { + cur = cur->parent; + + if (cur == n) return; + } + + cur = cur->next_sibling; + } + } + + break; + } + + case axis_following_sibling: + { + for (xml_node_struct* c = n->next_sibling; c; c = c->next_sibling) + if (step_push(ns, c, alloc) & once) + return; + + break; + } + + case axis_preceding_sibling: + { + for (xml_node_struct* c = n->prev_sibling_c; c->next_sibling; c = c->prev_sibling_c) + if (step_push(ns, c, alloc) & once) + return; + + break; + } + + case axis_following: + { + xml_node_struct* cur = n; + + // exit from this node so that we don't include descendants + while (!cur->next_sibling) + { + cur = cur->parent; + + if (!cur) return; + } + + cur = cur->next_sibling; + + while (cur) + { + if (step_push(ns, cur, alloc) & once) + return; + + if (cur->first_child) + cur = cur->first_child; + else + { + while (!cur->next_sibling) + { + cur = cur->parent; + + if (!cur) return; + } + + cur = cur->next_sibling; + } + } + + break; + } + + case axis_preceding: + { + xml_node_struct* cur = n; + + // exit from this node so that we don't include descendants + while (!cur->prev_sibling_c->next_sibling) + { + cur = cur->parent; + + if (!cur) return; + } + + cur = cur->prev_sibling_c; + + while (cur) + { + if (cur->first_child) + cur = cur->first_child->prev_sibling_c; + else + { + // leaf node, can't be ancestor + if (step_push(ns, cur, alloc) & once) + return; + + while (!cur->prev_sibling_c->next_sibling) + { + cur = cur->parent; + + if (!cur) return; + + if (!node_is_ancestor(cur, n)) + if (step_push(ns, cur, alloc) & once) + return; + } + + cur = cur->prev_sibling_c; + } + } + + break; + } + + case axis_ancestor: + case axis_ancestor_or_self: + { + if (axis == axis_ancestor_or_self) + if (step_push(ns, n, alloc) & once) + return; + + xml_node_struct* cur = n->parent; + + while (cur) + { + if (step_push(ns, cur, alloc) & once) + return; + + cur = cur->parent; + } + + break; + } + + case axis_self: + { + step_push(ns, n, alloc); + + break; + } + + case axis_parent: + { + if (n->parent) + step_push(ns, n->parent, alloc); + + break; + } + + default: + assert(false && "Unimplemented axis"); // unreachable + } + } + + template void step_fill(xpath_node_set_raw& ns, xml_attribute_struct* a, xml_node_struct* p, xpath_allocator* alloc, bool once, T v) + { + const axis_t axis = T::axis; + + switch (axis) + { + case axis_ancestor: + case axis_ancestor_or_self: + { + if (axis == axis_ancestor_or_self && _test == nodetest_type_node) // reject attributes based on principal node type test + if (step_push(ns, a, p, alloc) & once) + return; + + xml_node_struct* cur = p; + + while (cur) + { + if (step_push(ns, cur, alloc) & once) + return; + + cur = cur->parent; + } + + break; + } + + case axis_descendant_or_self: + case axis_self: + { + if (_test == nodetest_type_node) // reject attributes based on principal node type test + step_push(ns, a, p, alloc); + + break; + } + + case axis_following: + { + xml_node_struct* cur = p; + + while (cur) + { + if (cur->first_child) + cur = cur->first_child; + else + { + while (!cur->next_sibling) + { + cur = cur->parent; + + if (!cur) return; + } + + cur = cur->next_sibling; + } + + if (step_push(ns, cur, alloc) & once) + return; + } + + break; + } + + case axis_parent: + { + step_push(ns, p, alloc); + + break; + } + + case axis_preceding: + { + // preceding:: axis does not include attribute nodes and attribute ancestors (they are the same as parent's ancestors), so we can reuse node preceding + step_fill(ns, p, alloc, once, v); + break; + } + + default: + assert(false && "Unimplemented axis"); // unreachable + } + } + + template void step_fill(xpath_node_set_raw& ns, const xpath_node& xn, xpath_allocator* alloc, bool once, T v) + { + const axis_t axis = T::axis; + const bool axis_has_attributes = (axis == axis_ancestor || axis == axis_ancestor_or_self || axis == axis_descendant_or_self || axis == axis_following || axis == axis_parent || axis == axis_preceding || axis == axis_self); + + if (xn.node()) + step_fill(ns, xn.node().internal_object(), alloc, once, v); + else if (axis_has_attributes && xn.attribute() && xn.parent()) + step_fill(ns, xn.attribute().internal_object(), xn.parent().internal_object(), alloc, once, v); + } + + template xpath_node_set_raw step_do(const xpath_context& c, const xpath_stack& stack, nodeset_eval_t eval, T v) + { + const axis_t axis = T::axis; + const bool axis_reverse = (axis == axis_ancestor || axis == axis_ancestor_or_self || axis == axis_preceding || axis == axis_preceding_sibling); + const xpath_node_set::type_t axis_type = axis_reverse ? xpath_node_set::type_sorted_reverse : xpath_node_set::type_sorted; + + bool once = + (axis == axis_attribute && _test == nodetest_name) || + (!_right && eval_once(axis_type, eval)) || + // coverity[mixed_enums] + (_right && !_right->_next && _right->_test == predicate_constant_one); + + xpath_node_set_raw ns; + ns.set_type(axis_type); + + if (_left) + { + xpath_node_set_raw s = _left->eval_node_set(c, stack, nodeset_eval_all); + + // self axis preserves the original order + if (axis == axis_self) ns.set_type(s.type()); + + for (const xpath_node* it = s.begin(); it != s.end(); ++it) + { + size_t size = ns.size(); + + // in general, all axes generate elements in a particular order, but there is no order guarantee if axis is applied to two nodes + if (axis != axis_self && size != 0) ns.set_type(xpath_node_set::type_unsorted); + + step_fill(ns, *it, stack.result, once, v); + if (_right) apply_predicates(ns, size, stack, eval); + } + } + else + { + step_fill(ns, c.n, stack.result, once, v); + if (_right) apply_predicates(ns, 0, stack, eval); + } + + // child, attribute and self axes always generate unique set of nodes + // for other axis, if the set stayed sorted, it stayed unique because the traversal algorithms do not visit the same node twice + if (axis != axis_child && axis != axis_attribute && axis != axis_self && ns.type() == xpath_node_set::type_unsorted) + ns.remove_duplicates(stack.temp); + + return ns; + } + + public: + xpath_ast_node(ast_type_t type, xpath_value_type rettype_, const char_t* value): + _type(static_cast(type)), _rettype(static_cast(rettype_)), _axis(0), _test(0), _left(0), _right(0), _next(0) + { + assert(type == ast_string_constant); + _data.string = value; + } + + xpath_ast_node(ast_type_t type, xpath_value_type rettype_, double value): + _type(static_cast(type)), _rettype(static_cast(rettype_)), _axis(0), _test(0), _left(0), _right(0), _next(0) + { + assert(type == ast_number_constant); + _data.number = value; + } + + xpath_ast_node(ast_type_t type, xpath_value_type rettype_, xpath_variable* value): + _type(static_cast(type)), _rettype(static_cast(rettype_)), _axis(0), _test(0), _left(0), _right(0), _next(0) + { + assert(type == ast_variable); + _data.variable = value; + } + + xpath_ast_node(ast_type_t type, xpath_value_type rettype_, xpath_ast_node* left = 0, xpath_ast_node* right = 0): + _type(static_cast(type)), _rettype(static_cast(rettype_)), _axis(0), _test(0), _left(left), _right(right), _next(0) + { + } + + xpath_ast_node(ast_type_t type, xpath_ast_node* left, axis_t axis, nodetest_t test, const char_t* contents): + _type(static_cast(type)), _rettype(xpath_type_node_set), _axis(static_cast(axis)), _test(static_cast(test)), _left(left), _right(0), _next(0) + { + assert(type == ast_step); + _data.nodetest = contents; + } + + xpath_ast_node(ast_type_t type, xpath_ast_node* left, xpath_ast_node* right, predicate_t test): + _type(static_cast(type)), _rettype(xpath_type_node_set), _axis(0), _test(static_cast(test)), _left(left), _right(right), _next(0) + { + assert(type == ast_filter || type == ast_predicate); + } + + void set_next(xpath_ast_node* value) + { + _next = value; + } + + void set_right(xpath_ast_node* value) + { + _right = value; + } + + bool eval_boolean(const xpath_context& c, const xpath_stack& stack) + { + switch (_type) + { + case ast_op_or: + return _left->eval_boolean(c, stack) || _right->eval_boolean(c, stack); + + case ast_op_and: + return _left->eval_boolean(c, stack) && _right->eval_boolean(c, stack); + + case ast_op_equal: + return compare_eq(_left, _right, c, stack, equal_to()); + + case ast_op_not_equal: + return compare_eq(_left, _right, c, stack, not_equal_to()); + + case ast_op_less: + return compare_rel(_left, _right, c, stack, less()); + + case ast_op_greater: + return compare_rel(_right, _left, c, stack, less()); + + case ast_op_less_or_equal: + return compare_rel(_left, _right, c, stack, less_equal()); + + case ast_op_greater_or_equal: + return compare_rel(_right, _left, c, stack, less_equal()); + + case ast_func_starts_with: + { + xpath_allocator_capture cr(stack.result); + + xpath_string lr = _left->eval_string(c, stack); + xpath_string rr = _right->eval_string(c, stack); + + return starts_with(lr.c_str(), rr.c_str()); + } + + case ast_func_contains: + { + xpath_allocator_capture cr(stack.result); + + xpath_string lr = _left->eval_string(c, stack); + xpath_string rr = _right->eval_string(c, stack); + + return find_substring(lr.c_str(), rr.c_str()) != 0; + } + + case ast_func_boolean: + return _left->eval_boolean(c, stack); + + case ast_func_not: + return !_left->eval_boolean(c, stack); + + case ast_func_true: + return true; + + case ast_func_false: + return false; + + case ast_func_lang: + { + if (c.n.attribute()) return false; + + xpath_allocator_capture cr(stack.result); + + xpath_string lang = _left->eval_string(c, stack); + + for (xml_node n = c.n.node(); n; n = n.parent()) + { + xml_attribute a = n.attribute(PUGIXML_TEXT("xml:lang")); + + if (a) + { + const char_t* value = a.value(); + + // strnicmp / strncasecmp is not portable + for (const char_t* lit = lang.c_str(); *lit; ++lit) + { + if (tolower_ascii(*lit) != tolower_ascii(*value)) return false; + ++value; + } + + return *value == 0 || *value == '-'; + } + } + + return false; + } + + case ast_opt_compare_attribute: + { + const char_t* value = (_right->_type == ast_string_constant) ? _right->_data.string : _right->_data.variable->get_string(); + + xml_attribute attr = c.n.node().attribute(_left->_data.nodetest); + + return attr && strequal(attr.value(), value) && is_xpath_attribute(attr.name()); + } + + case ast_variable: + { + assert(_rettype == _data.variable->type()); + + if (_rettype == xpath_type_boolean) + return _data.variable->get_boolean(); + } + + // fallthrough + default: + { + switch (_rettype) + { + case xpath_type_number: + return convert_number_to_boolean(eval_number(c, stack)); + + case xpath_type_string: + { + xpath_allocator_capture cr(stack.result); + + return !eval_string(c, stack).empty(); + } + + case xpath_type_node_set: + { + xpath_allocator_capture cr(stack.result); + + return !eval_node_set(c, stack, nodeset_eval_any).empty(); + } + + default: + assert(false && "Wrong expression for return type boolean"); // unreachable + return false; + } + } + } + } + + double eval_number(const xpath_context& c, const xpath_stack& stack) + { + switch (_type) + { + case ast_op_add: + return _left->eval_number(c, stack) + _right->eval_number(c, stack); + + case ast_op_subtract: + return _left->eval_number(c, stack) - _right->eval_number(c, stack); + + case ast_op_multiply: + return _left->eval_number(c, stack) * _right->eval_number(c, stack); + + case ast_op_divide: + return _left->eval_number(c, stack) / _right->eval_number(c, stack); + + case ast_op_mod: + return fmod(_left->eval_number(c, stack), _right->eval_number(c, stack)); + + case ast_op_negate: + return -_left->eval_number(c, stack); + + case ast_number_constant: + return _data.number; + + case ast_func_last: + return static_cast(c.size); + + case ast_func_position: + return static_cast(c.position); + + case ast_func_count: + { + xpath_allocator_capture cr(stack.result); + + return static_cast(_left->eval_node_set(c, stack, nodeset_eval_all).size()); + } + + case ast_func_string_length_0: + { + xpath_allocator_capture cr(stack.result); + + return static_cast(string_value(c.n, stack.result).length()); + } + + case ast_func_string_length_1: + { + xpath_allocator_capture cr(stack.result); + + return static_cast(_left->eval_string(c, stack).length()); + } + + case ast_func_number_0: + { + xpath_allocator_capture cr(stack.result); + + return convert_string_to_number(string_value(c.n, stack.result).c_str()); + } + + case ast_func_number_1: + return _left->eval_number(c, stack); + + case ast_func_sum: + { + xpath_allocator_capture cr(stack.result); + + double r = 0; + + xpath_node_set_raw ns = _left->eval_node_set(c, stack, nodeset_eval_all); + + for (const xpath_node* it = ns.begin(); it != ns.end(); ++it) + { + xpath_allocator_capture cri(stack.result); + + r += convert_string_to_number(string_value(*it, stack.result).c_str()); + } + + return r; + } + + case ast_func_floor: + { + double r = _left->eval_number(c, stack); + + return r == r ? floor(r) : r; + } + + case ast_func_ceiling: + { + double r = _left->eval_number(c, stack); + + return r == r ? ceil(r) : r; + } + + case ast_func_round: + return round_nearest_nzero(_left->eval_number(c, stack)); + + case ast_variable: + { + assert(_rettype == _data.variable->type()); + + if (_rettype == xpath_type_number) + return _data.variable->get_number(); + } + + // fallthrough + default: + { + switch (_rettype) + { + case xpath_type_boolean: + return eval_boolean(c, stack) ? 1 : 0; + + case xpath_type_string: + { + xpath_allocator_capture cr(stack.result); + + return convert_string_to_number(eval_string(c, stack).c_str()); + } + + case xpath_type_node_set: + { + xpath_allocator_capture cr(stack.result); + + return convert_string_to_number(eval_string(c, stack).c_str()); + } + + default: + assert(false && "Wrong expression for return type number"); // unreachable + return 0; + } + + } + } + } + + xpath_string eval_string_concat(const xpath_context& c, const xpath_stack& stack) + { + assert(_type == ast_func_concat); + + xpath_allocator_capture ct(stack.temp); + + // count the string number + size_t count = 1; + for (xpath_ast_node* nc = _right; nc; nc = nc->_next) count++; + + // allocate a buffer for temporary string objects + xpath_string* buffer = static_cast(stack.temp->allocate(count * sizeof(xpath_string))); + if (!buffer) return xpath_string(); + + // evaluate all strings to temporary stack + xpath_stack swapped_stack = {stack.temp, stack.result}; + + buffer[0] = _left->eval_string(c, swapped_stack); + + size_t pos = 1; + for (xpath_ast_node* n = _right; n; n = n->_next, ++pos) buffer[pos] = n->eval_string(c, swapped_stack); + assert(pos == count); + + // get total length + size_t length = 0; + for (size_t i = 0; i < count; ++i) length += buffer[i].length(); + + // create final string + char_t* result = static_cast(stack.result->allocate((length + 1) * sizeof(char_t))); + if (!result) return xpath_string(); + + char_t* ri = result; + + for (size_t j = 0; j < count; ++j) + for (const char_t* bi = buffer[j].c_str(); *bi; ++bi) + *ri++ = *bi; + + *ri = 0; + + return xpath_string::from_heap_preallocated(result, ri); + } + + xpath_string eval_string(const xpath_context& c, const xpath_stack& stack) + { + switch (_type) + { + case ast_string_constant: + return xpath_string::from_const(_data.string); + + case ast_func_local_name_0: + { + xpath_node na = c.n; + + return xpath_string::from_const(local_name(na)); + } + + case ast_func_local_name_1: + { + xpath_allocator_capture cr(stack.result); + + xpath_node_set_raw ns = _left->eval_node_set(c, stack, nodeset_eval_first); + xpath_node na = ns.first(); + + return xpath_string::from_const(local_name(na)); + } + + case ast_func_name_0: + { + xpath_node na = c.n; + + return xpath_string::from_const(qualified_name(na)); + } + + case ast_func_name_1: + { + xpath_allocator_capture cr(stack.result); + + xpath_node_set_raw ns = _left->eval_node_set(c, stack, nodeset_eval_first); + xpath_node na = ns.first(); + + return xpath_string::from_const(qualified_name(na)); + } + + case ast_func_namespace_uri_0: + { + xpath_node na = c.n; + + return xpath_string::from_const(namespace_uri(na)); + } + + case ast_func_namespace_uri_1: + { + xpath_allocator_capture cr(stack.result); + + xpath_node_set_raw ns = _left->eval_node_set(c, stack, nodeset_eval_first); + xpath_node na = ns.first(); + + return xpath_string::from_const(namespace_uri(na)); + } + + case ast_func_string_0: + return string_value(c.n, stack.result); + + case ast_func_string_1: + return _left->eval_string(c, stack); + + case ast_func_concat: + return eval_string_concat(c, stack); + + case ast_func_substring_before: + { + xpath_allocator_capture cr(stack.temp); + + xpath_stack swapped_stack = {stack.temp, stack.result}; + + xpath_string s = _left->eval_string(c, swapped_stack); + xpath_string p = _right->eval_string(c, swapped_stack); + + const char_t* pos = find_substring(s.c_str(), p.c_str()); + + return pos ? xpath_string::from_heap(s.c_str(), pos, stack.result) : xpath_string(); + } + + case ast_func_substring_after: + { + xpath_allocator_capture cr(stack.temp); + + xpath_stack swapped_stack = {stack.temp, stack.result}; + + xpath_string s = _left->eval_string(c, swapped_stack); + xpath_string p = _right->eval_string(c, swapped_stack); + + const char_t* pos = find_substring(s.c_str(), p.c_str()); + if (!pos) return xpath_string(); + + const char_t* rbegin = pos + p.length(); + const char_t* rend = s.c_str() + s.length(); + + return s.uses_heap() ? xpath_string::from_heap(rbegin, rend, stack.result) : xpath_string::from_const(rbegin); + } + + case ast_func_substring_2: + { + xpath_allocator_capture cr(stack.temp); + + xpath_stack swapped_stack = {stack.temp, stack.result}; + + xpath_string s = _left->eval_string(c, swapped_stack); + size_t s_length = s.length(); + + double first = round_nearest(_right->eval_number(c, stack)); + + if (is_nan(first)) return xpath_string(); // NaN + else if (first >= s_length + 1) return xpath_string(); + + size_t pos = first < 1 ? 1 : static_cast(first); + assert(1 <= pos && pos <= s_length + 1); + + const char_t* rbegin = s.c_str() + (pos - 1); + const char_t* rend = s.c_str() + s.length(); + + return s.uses_heap() ? xpath_string::from_heap(rbegin, rend, stack.result) : xpath_string::from_const(rbegin); + } + + case ast_func_substring_3: + { + xpath_allocator_capture cr(stack.temp); + + xpath_stack swapped_stack = {stack.temp, stack.result}; + + xpath_string s = _left->eval_string(c, swapped_stack); + size_t s_length = s.length(); + + double first = round_nearest(_right->eval_number(c, stack)); + double last = first + round_nearest(_right->_next->eval_number(c, stack)); + + if (is_nan(first) || is_nan(last)) return xpath_string(); + else if (first >= s_length + 1) return xpath_string(); + else if (first >= last) return xpath_string(); + else if (last < 1) return xpath_string(); + + size_t pos = first < 1 ? 1 : static_cast(first); + size_t end = last >= s_length + 1 ? s_length + 1 : static_cast(last); + + assert(1 <= pos && pos <= end && end <= s_length + 1); + const char_t* rbegin = s.c_str() + (pos - 1); + const char_t* rend = s.c_str() + (end - 1); + + return (end == s_length + 1 && !s.uses_heap()) ? xpath_string::from_const(rbegin) : xpath_string::from_heap(rbegin, rend, stack.result); + } + + case ast_func_normalize_space_0: + { + xpath_string s = string_value(c.n, stack.result); + + char_t* begin = s.data(stack.result); + if (!begin) return xpath_string(); + + char_t* end = normalize_space(begin); + + return xpath_string::from_heap_preallocated(begin, end); + } + + case ast_func_normalize_space_1: + { + xpath_string s = _left->eval_string(c, stack); + + char_t* begin = s.data(stack.result); + if (!begin) return xpath_string(); + + char_t* end = normalize_space(begin); + + return xpath_string::from_heap_preallocated(begin, end); + } + + case ast_func_translate: + { + xpath_allocator_capture cr(stack.temp); + + xpath_stack swapped_stack = {stack.temp, stack.result}; + + xpath_string s = _left->eval_string(c, stack); + xpath_string from = _right->eval_string(c, swapped_stack); + xpath_string to = _right->_next->eval_string(c, swapped_stack); + + char_t* begin = s.data(stack.result); + if (!begin) return xpath_string(); + + char_t* end = translate(begin, from.c_str(), to.c_str(), to.length()); + + return xpath_string::from_heap_preallocated(begin, end); + } + + case ast_opt_translate_table: + { + xpath_string s = _left->eval_string(c, stack); + + char_t* begin = s.data(stack.result); + if (!begin) return xpath_string(); + + char_t* end = translate_table(begin, _data.table); + + return xpath_string::from_heap_preallocated(begin, end); + } + + case ast_variable: + { + assert(_rettype == _data.variable->type()); + + if (_rettype == xpath_type_string) + return xpath_string::from_const(_data.variable->get_string()); + } + + // fallthrough + default: + { + switch (_rettype) + { + case xpath_type_boolean: + return xpath_string::from_const(eval_boolean(c, stack) ? PUGIXML_TEXT("true") : PUGIXML_TEXT("false")); + + case xpath_type_number: + return convert_number_to_string(eval_number(c, stack), stack.result); + + case xpath_type_node_set: + { + xpath_allocator_capture cr(stack.temp); + + xpath_stack swapped_stack = {stack.temp, stack.result}; + + xpath_node_set_raw ns = eval_node_set(c, swapped_stack, nodeset_eval_first); + return ns.empty() ? xpath_string() : string_value(ns.first(), stack.result); + } + + default: + assert(false && "Wrong expression for return type string"); // unreachable + return xpath_string(); + } + } + } + } + + xpath_node_set_raw eval_node_set(const xpath_context& c, const xpath_stack& stack, nodeset_eval_t eval) + { + switch (_type) + { + case ast_op_union: + { + xpath_allocator_capture cr(stack.temp); + + xpath_stack swapped_stack = {stack.temp, stack.result}; + + xpath_node_set_raw ls = _left->eval_node_set(c, stack, eval); + xpath_node_set_raw rs = _right->eval_node_set(c, swapped_stack, eval); + + // we can optimize merging two sorted sets, but this is a very rare operation, so don't bother + ls.set_type(xpath_node_set::type_unsorted); + + ls.append(rs.begin(), rs.end(), stack.result); + ls.remove_duplicates(stack.temp); + + return ls; + } + + case ast_filter: + { + xpath_node_set_raw set = _left->eval_node_set(c, stack, _test == predicate_constant_one ? nodeset_eval_first : nodeset_eval_all); + + // either expression is a number or it contains position() call; sort by document order + if (_test != predicate_posinv) set.sort_do(); + + bool once = eval_once(set.type(), eval); + + apply_predicate(set, 0, stack, once); + + return set; + } + + case ast_func_id: + return xpath_node_set_raw(); + + case ast_step: + { + switch (_axis) + { + case axis_ancestor: + return step_do(c, stack, eval, axis_to_type()); + + case axis_ancestor_or_self: + return step_do(c, stack, eval, axis_to_type()); + + case axis_attribute: + return step_do(c, stack, eval, axis_to_type()); + + case axis_child: + return step_do(c, stack, eval, axis_to_type()); + + case axis_descendant: + return step_do(c, stack, eval, axis_to_type()); + + case axis_descendant_or_self: + return step_do(c, stack, eval, axis_to_type()); + + case axis_following: + return step_do(c, stack, eval, axis_to_type()); + + case axis_following_sibling: + return step_do(c, stack, eval, axis_to_type()); + + case axis_namespace: + // namespaced axis is not supported + return xpath_node_set_raw(); + + case axis_parent: + return step_do(c, stack, eval, axis_to_type()); + + case axis_preceding: + return step_do(c, stack, eval, axis_to_type()); + + case axis_preceding_sibling: + return step_do(c, stack, eval, axis_to_type()); + + case axis_self: + return step_do(c, stack, eval, axis_to_type()); + + default: + assert(false && "Unknown axis"); // unreachable + return xpath_node_set_raw(); + } + } + + case ast_step_root: + { + assert(!_right); // root step can't have any predicates + + xpath_node_set_raw ns; + + ns.set_type(xpath_node_set::type_sorted); + + if (c.n.node()) ns.push_back(c.n.node().root(), stack.result); + else if (c.n.attribute()) ns.push_back(c.n.parent().root(), stack.result); + + return ns; + } + + case ast_variable: + { + assert(_rettype == _data.variable->type()); + + if (_rettype == xpath_type_node_set) + { + const xpath_node_set& s = _data.variable->get_node_set(); + + xpath_node_set_raw ns; + + ns.set_type(s.type()); + ns.append(s.begin(), s.end(), stack.result); + + return ns; + } + } + + // fallthrough + default: + assert(false && "Wrong expression for return type node set"); // unreachable + return xpath_node_set_raw(); + } + } + + void optimize(xpath_allocator* alloc) + { + if (_left) + _left->optimize(alloc); + + if (_right) + _right->optimize(alloc); + + if (_next) + _next->optimize(alloc); + + // coverity[var_deref_model] + optimize_self(alloc); + } + + void optimize_self(xpath_allocator* alloc) + { + // Rewrite [position()=expr] with [expr] + // Note that this step has to go before classification to recognize [position()=1] + if ((_type == ast_filter || _type == ast_predicate) && + _right && // workaround for clang static analyzer (_right is never null for ast_filter/ast_predicate) + _right->_type == ast_op_equal && _right->_left->_type == ast_func_position && _right->_right->_rettype == xpath_type_number) + { + _right = _right->_right; + } + + // Classify filter/predicate ops to perform various optimizations during evaluation + if ((_type == ast_filter || _type == ast_predicate) && _right) // workaround for clang static analyzer (_right is never null for ast_filter/ast_predicate) + { + assert(_test == predicate_default); + + if (_right->_type == ast_number_constant && _right->_data.number == 1.0) + _test = predicate_constant_one; + else if (_right->_rettype == xpath_type_number && (_right->_type == ast_number_constant || _right->_type == ast_variable || _right->_type == ast_func_last)) + _test = predicate_constant; + else if (_right->_rettype != xpath_type_number && _right->is_posinv_expr()) + _test = predicate_posinv; + } + + // Rewrite descendant-or-self::node()/child::foo with descendant::foo + // The former is a full form of //foo, the latter is much faster since it executes the node test immediately + // Do a similar kind of rewrite for self/descendant/descendant-or-self axes + // Note that we only rewrite positionally invariant steps (//foo[1] != /descendant::foo[1]) + if (_type == ast_step && (_axis == axis_child || _axis == axis_self || _axis == axis_descendant || _axis == axis_descendant_or_self) && + _left && _left->_type == ast_step && _left->_axis == axis_descendant_or_self && _left->_test == nodetest_type_node && !_left->_right && + is_posinv_step()) + { + if (_axis == axis_child || _axis == axis_descendant) + _axis = axis_descendant; + else + _axis = axis_descendant_or_self; + + _left = _left->_left; + } + + // Use optimized lookup table implementation for translate() with constant arguments + if (_type == ast_func_translate && + _right && // workaround for clang static analyzer (_right is never null for ast_func_translate) + _right->_type == ast_string_constant && _right->_next->_type == ast_string_constant) + { + unsigned char* table = translate_table_generate(alloc, _right->_data.string, _right->_next->_data.string); + + if (table) + { + _type = ast_opt_translate_table; + _data.table = table; + } + } + + // Use optimized path for @attr = 'value' or @attr = $value + if (_type == ast_op_equal && + _left && _right && // workaround for clang static analyzer and Coverity (_left and _right are never null for ast_op_equal) + // coverity[mixed_enums] + _left->_type == ast_step && _left->_axis == axis_attribute && _left->_test == nodetest_name && !_left->_left && !_left->_right && + (_right->_type == ast_string_constant || (_right->_type == ast_variable && _right->_rettype == xpath_type_string))) + { + _type = ast_opt_compare_attribute; + } + } + + bool is_posinv_expr() const + { + switch (_type) + { + case ast_func_position: + case ast_func_last: + return false; + + case ast_string_constant: + case ast_number_constant: + case ast_variable: + return true; + + case ast_step: + case ast_step_root: + return true; + + case ast_predicate: + case ast_filter: + return true; + + default: + if (_left && !_left->is_posinv_expr()) return false; + + for (xpath_ast_node* n = _right; n; n = n->_next) + if (!n->is_posinv_expr()) return false; + + return true; + } + } + + bool is_posinv_step() const + { + assert(_type == ast_step); + + for (xpath_ast_node* n = _right; n; n = n->_next) + { + assert(n->_type == ast_predicate); + + if (n->_test != predicate_posinv) + return false; + } + + return true; + } + + xpath_value_type rettype() const + { + return static_cast(_rettype); + } + }; + + struct xpath_parser + { + xpath_allocator* _alloc; + xpath_lexer _lexer; + + const char_t* _query; + xpath_variable_set* _variables; + + xpath_parse_result* _result; + + char_t _scratch[32]; + + xpath_ast_node* error(const char* message) + { + _result->error = message; + _result->offset = _lexer.current_pos() - _query; + + return 0; + } + + xpath_ast_node* error_oom() + { + assert(_alloc->_error); + *_alloc->_error = true; + + return 0; + } + + void* alloc_node() + { + return _alloc->allocate(sizeof(xpath_ast_node)); + } + + xpath_ast_node* alloc_node(ast_type_t type, xpath_value_type rettype, const char_t* value) + { + void* memory = alloc_node(); + return memory ? new (memory) xpath_ast_node(type, rettype, value) : 0; + } + + xpath_ast_node* alloc_node(ast_type_t type, xpath_value_type rettype, double value) + { + void* memory = alloc_node(); + return memory ? new (memory) xpath_ast_node(type, rettype, value) : 0; + } + + xpath_ast_node* alloc_node(ast_type_t type, xpath_value_type rettype, xpath_variable* value) + { + void* memory = alloc_node(); + return memory ? new (memory) xpath_ast_node(type, rettype, value) : 0; + } + + xpath_ast_node* alloc_node(ast_type_t type, xpath_value_type rettype, xpath_ast_node* left = 0, xpath_ast_node* right = 0) + { + void* memory = alloc_node(); + return memory ? new (memory) xpath_ast_node(type, rettype, left, right) : 0; + } + + xpath_ast_node* alloc_node(ast_type_t type, xpath_ast_node* left, axis_t axis, nodetest_t test, const char_t* contents) + { + void* memory = alloc_node(); + return memory ? new (memory) xpath_ast_node(type, left, axis, test, contents) : 0; + } + + xpath_ast_node* alloc_node(ast_type_t type, xpath_ast_node* left, xpath_ast_node* right, predicate_t test) + { + void* memory = alloc_node(); + return memory ? new (memory) xpath_ast_node(type, left, right, test) : 0; + } + + const char_t* alloc_string(const xpath_lexer_string& value) + { + if (!value.begin) + return PUGIXML_TEXT(""); + + size_t length = static_cast(value.end - value.begin); + + char_t* c = static_cast(_alloc->allocate((length + 1) * sizeof(char_t))); + if (!c) return 0; + + memcpy(c, value.begin, length * sizeof(char_t)); + c[length] = 0; + + return c; + } + + xpath_ast_node* parse_function(const xpath_lexer_string& name, size_t argc, xpath_ast_node* args[2]) + { + switch (name.begin[0]) + { + case 'b': + if (name == PUGIXML_TEXT("boolean") && argc == 1) + return alloc_node(ast_func_boolean, xpath_type_boolean, args[0]); + + break; + + case 'c': + if (name == PUGIXML_TEXT("count") && argc == 1) + { + if (args[0]->rettype() != xpath_type_node_set) return error("Function has to be applied to node set"); + return alloc_node(ast_func_count, xpath_type_number, args[0]); + } + else if (name == PUGIXML_TEXT("contains") && argc == 2) + return alloc_node(ast_func_contains, xpath_type_boolean, args[0], args[1]); + else if (name == PUGIXML_TEXT("concat") && argc >= 2) + return alloc_node(ast_func_concat, xpath_type_string, args[0], args[1]); + else if (name == PUGIXML_TEXT("ceiling") && argc == 1) + return alloc_node(ast_func_ceiling, xpath_type_number, args[0]); + + break; + + case 'f': + if (name == PUGIXML_TEXT("false") && argc == 0) + return alloc_node(ast_func_false, xpath_type_boolean); + else if (name == PUGIXML_TEXT("floor") && argc == 1) + return alloc_node(ast_func_floor, xpath_type_number, args[0]); + + break; + + case 'i': + if (name == PUGIXML_TEXT("id") && argc == 1) + return alloc_node(ast_func_id, xpath_type_node_set, args[0]); + + break; + + case 'l': + if (name == PUGIXML_TEXT("last") && argc == 0) + return alloc_node(ast_func_last, xpath_type_number); + else if (name == PUGIXML_TEXT("lang") && argc == 1) + return alloc_node(ast_func_lang, xpath_type_boolean, args[0]); + else if (name == PUGIXML_TEXT("local-name") && argc <= 1) + { + if (argc == 1 && args[0]->rettype() != xpath_type_node_set) return error("Function has to be applied to node set"); + return alloc_node(argc == 0 ? ast_func_local_name_0 : ast_func_local_name_1, xpath_type_string, args[0]); + } + + break; + + case 'n': + if (name == PUGIXML_TEXT("name") && argc <= 1) + { + if (argc == 1 && args[0]->rettype() != xpath_type_node_set) return error("Function has to be applied to node set"); + return alloc_node(argc == 0 ? ast_func_name_0 : ast_func_name_1, xpath_type_string, args[0]); + } + else if (name == PUGIXML_TEXT("namespace-uri") && argc <= 1) + { + if (argc == 1 && args[0]->rettype() != xpath_type_node_set) return error("Function has to be applied to node set"); + return alloc_node(argc == 0 ? ast_func_namespace_uri_0 : ast_func_namespace_uri_1, xpath_type_string, args[0]); + } + else if (name == PUGIXML_TEXT("normalize-space") && argc <= 1) + return alloc_node(argc == 0 ? ast_func_normalize_space_0 : ast_func_normalize_space_1, xpath_type_string, args[0], args[1]); + else if (name == PUGIXML_TEXT("not") && argc == 1) + return alloc_node(ast_func_not, xpath_type_boolean, args[0]); + else if (name == PUGIXML_TEXT("number") && argc <= 1) + return alloc_node(argc == 0 ? ast_func_number_0 : ast_func_number_1, xpath_type_number, args[0]); + + break; + + case 'p': + if (name == PUGIXML_TEXT("position") && argc == 0) + return alloc_node(ast_func_position, xpath_type_number); + + break; + + case 'r': + if (name == PUGIXML_TEXT("round") && argc == 1) + return alloc_node(ast_func_round, xpath_type_number, args[0]); + + break; + + case 's': + if (name == PUGIXML_TEXT("string") && argc <= 1) + return alloc_node(argc == 0 ? ast_func_string_0 : ast_func_string_1, xpath_type_string, args[0]); + else if (name == PUGIXML_TEXT("string-length") && argc <= 1) + return alloc_node(argc == 0 ? ast_func_string_length_0 : ast_func_string_length_1, xpath_type_number, args[0]); + else if (name == PUGIXML_TEXT("starts-with") && argc == 2) + return alloc_node(ast_func_starts_with, xpath_type_boolean, args[0], args[1]); + else if (name == PUGIXML_TEXT("substring-before") && argc == 2) + return alloc_node(ast_func_substring_before, xpath_type_string, args[0], args[1]); + else if (name == PUGIXML_TEXT("substring-after") && argc == 2) + return alloc_node(ast_func_substring_after, xpath_type_string, args[0], args[1]); + else if (name == PUGIXML_TEXT("substring") && (argc == 2 || argc == 3)) + return alloc_node(argc == 2 ? ast_func_substring_2 : ast_func_substring_3, xpath_type_string, args[0], args[1]); + else if (name == PUGIXML_TEXT("sum") && argc == 1) + { + if (args[0]->rettype() != xpath_type_node_set) return error("Function has to be applied to node set"); + return alloc_node(ast_func_sum, xpath_type_number, args[0]); + } + + break; + + case 't': + if (name == PUGIXML_TEXT("translate") && argc == 3) + return alloc_node(ast_func_translate, xpath_type_string, args[0], args[1]); + else if (name == PUGIXML_TEXT("true") && argc == 0) + return alloc_node(ast_func_true, xpath_type_boolean); + + break; + + default: + break; + } + + return error("Unrecognized function or wrong parameter count"); + } + + axis_t parse_axis_name(const xpath_lexer_string& name, bool& specified) + { + specified = true; + + switch (name.begin[0]) + { + case 'a': + if (name == PUGIXML_TEXT("ancestor")) + return axis_ancestor; + else if (name == PUGIXML_TEXT("ancestor-or-self")) + return axis_ancestor_or_self; + else if (name == PUGIXML_TEXT("attribute")) + return axis_attribute; + + break; + + case 'c': + if (name == PUGIXML_TEXT("child")) + return axis_child; + + break; + + case 'd': + if (name == PUGIXML_TEXT("descendant")) + return axis_descendant; + else if (name == PUGIXML_TEXT("descendant-or-self")) + return axis_descendant_or_self; + + break; + + case 'f': + if (name == PUGIXML_TEXT("following")) + return axis_following; + else if (name == PUGIXML_TEXT("following-sibling")) + return axis_following_sibling; + + break; + + case 'n': + if (name == PUGIXML_TEXT("namespace")) + return axis_namespace; + + break; + + case 'p': + if (name == PUGIXML_TEXT("parent")) + return axis_parent; + else if (name == PUGIXML_TEXT("preceding")) + return axis_preceding; + else if (name == PUGIXML_TEXT("preceding-sibling")) + return axis_preceding_sibling; + + break; + + case 's': + if (name == PUGIXML_TEXT("self")) + return axis_self; + + break; + + default: + break; + } + + specified = false; + return axis_child; + } + + nodetest_t parse_node_test_type(const xpath_lexer_string& name) + { + switch (name.begin[0]) + { + case 'c': + if (name == PUGIXML_TEXT("comment")) + return nodetest_type_comment; + + break; + + case 'n': + if (name == PUGIXML_TEXT("node")) + return nodetest_type_node; + + break; + + case 'p': + if (name == PUGIXML_TEXT("processing-instruction")) + return nodetest_type_pi; + + break; + + case 't': + if (name == PUGIXML_TEXT("text")) + return nodetest_type_text; + + break; + + default: + break; + } + + return nodetest_none; + } + + // PrimaryExpr ::= VariableReference | '(' Expr ')' | Literal | Number | FunctionCall + xpath_ast_node* parse_primary_expression() + { + switch (_lexer.current()) + { + case lex_var_ref: + { + xpath_lexer_string name = _lexer.contents(); + + if (!_variables) + return error("Unknown variable: variable set is not provided"); + + xpath_variable* var = 0; + if (!get_variable_scratch(_scratch, _variables, name.begin, name.end, &var)) + return error_oom(); + + if (!var) + return error("Unknown variable: variable set does not contain the given name"); + + _lexer.next(); + + return alloc_node(ast_variable, var->type(), var); + } + + case lex_open_brace: + { + _lexer.next(); + + xpath_ast_node* n = parse_expression(); + if (!n) return 0; + + if (_lexer.current() != lex_close_brace) + return error("Expected ')' to match an opening '('"); + + _lexer.next(); + + return n; + } + + case lex_quoted_string: + { + const char_t* value = alloc_string(_lexer.contents()); + if (!value) return 0; + + _lexer.next(); + + return alloc_node(ast_string_constant, xpath_type_string, value); + } + + case lex_number: + { + double value = 0; + + if (!convert_string_to_number_scratch(_scratch, _lexer.contents().begin, _lexer.contents().end, &value)) + return error_oom(); + + _lexer.next(); + + return alloc_node(ast_number_constant, xpath_type_number, value); + } + + case lex_string: + { + xpath_ast_node* args[2] = {0}; + size_t argc = 0; + + xpath_lexer_string function = _lexer.contents(); + _lexer.next(); + + xpath_ast_node* last_arg = 0; + + if (_lexer.current() != lex_open_brace) + return error("Unrecognized function call"); + _lexer.next(); + + while (_lexer.current() != lex_close_brace) + { + if (argc > 0) + { + if (_lexer.current() != lex_comma) + return error("No comma between function arguments"); + _lexer.next(); + } + + xpath_ast_node* n = parse_expression(); + if (!n) return 0; + + if (argc < 2) args[argc] = n; + else last_arg->set_next(n); + + argc++; + last_arg = n; + } + + _lexer.next(); + + return parse_function(function, argc, args); + } + + default: + return error("Unrecognizable primary expression"); + } + } + + // FilterExpr ::= PrimaryExpr | FilterExpr Predicate + // Predicate ::= '[' PredicateExpr ']' + // PredicateExpr ::= Expr + xpath_ast_node* parse_filter_expression() + { + xpath_ast_node* n = parse_primary_expression(); + if (!n) return 0; + + while (_lexer.current() == lex_open_square_brace) + { + _lexer.next(); + + if (n->rettype() != xpath_type_node_set) + return error("Predicate has to be applied to node set"); + + xpath_ast_node* expr = parse_expression(); + if (!expr) return 0; + + n = alloc_node(ast_filter, n, expr, predicate_default); + if (!n) return 0; + + if (_lexer.current() != lex_close_square_brace) + return error("Expected ']' to match an opening '['"); + + _lexer.next(); + } + + return n; + } + + // Step ::= AxisSpecifier NodeTest Predicate* | AbbreviatedStep + // AxisSpecifier ::= AxisName '::' | '@'? + // NodeTest ::= NameTest | NodeType '(' ')' | 'processing-instruction' '(' Literal ')' + // NameTest ::= '*' | NCName ':' '*' | QName + // AbbreviatedStep ::= '.' | '..' + xpath_ast_node* parse_step(xpath_ast_node* set) + { + if (set && set->rettype() != xpath_type_node_set) + return error("Step has to be applied to node set"); + + bool axis_specified = false; + axis_t axis = axis_child; // implied child axis + + if (_lexer.current() == lex_axis_attribute) + { + axis = axis_attribute; + axis_specified = true; + + _lexer.next(); + } + else if (_lexer.current() == lex_dot) + { + _lexer.next(); + + if (_lexer.current() == lex_open_square_brace) + return error("Predicates are not allowed after an abbreviated step"); + + return alloc_node(ast_step, set, axis_self, nodetest_type_node, 0); + } + else if (_lexer.current() == lex_double_dot) + { + _lexer.next(); + + if (_lexer.current() == lex_open_square_brace) + return error("Predicates are not allowed after an abbreviated step"); + + return alloc_node(ast_step, set, axis_parent, nodetest_type_node, 0); + } + + nodetest_t nt_type = nodetest_none; + xpath_lexer_string nt_name; + + if (_lexer.current() == lex_string) + { + // node name test + nt_name = _lexer.contents(); + _lexer.next(); + + // was it an axis name? + if (_lexer.current() == lex_double_colon) + { + // parse axis name + if (axis_specified) + return error("Two axis specifiers in one step"); + + axis = parse_axis_name(nt_name, axis_specified); + + if (!axis_specified) + return error("Unknown axis"); + + // read actual node test + _lexer.next(); + + if (_lexer.current() == lex_multiply) + { + nt_type = nodetest_all; + nt_name = xpath_lexer_string(); + _lexer.next(); + } + else if (_lexer.current() == lex_string) + { + nt_name = _lexer.contents(); + _lexer.next(); + } + else + { + return error("Unrecognized node test"); + } + } + + if (nt_type == nodetest_none) + { + // node type test or processing-instruction + if (_lexer.current() == lex_open_brace) + { + _lexer.next(); + + if (_lexer.current() == lex_close_brace) + { + _lexer.next(); + + nt_type = parse_node_test_type(nt_name); + + if (nt_type == nodetest_none) + return error("Unrecognized node type"); + + nt_name = xpath_lexer_string(); + } + else if (nt_name == PUGIXML_TEXT("processing-instruction")) + { + if (_lexer.current() != lex_quoted_string) + return error("Only literals are allowed as arguments to processing-instruction()"); + + nt_type = nodetest_pi; + nt_name = _lexer.contents(); + _lexer.next(); + + if (_lexer.current() != lex_close_brace) + return error("Unmatched brace near processing-instruction()"); + _lexer.next(); + } + else + { + return error("Unmatched brace near node type test"); + } + } + // QName or NCName:* + else + { + if (nt_name.end - nt_name.begin > 2 && nt_name.end[-2] == ':' && nt_name.end[-1] == '*') // NCName:* + { + nt_name.end--; // erase * + + nt_type = nodetest_all_in_namespace; + } + else + { + nt_type = nodetest_name; + } + } + } + } + else if (_lexer.current() == lex_multiply) + { + nt_type = nodetest_all; + _lexer.next(); + } + else + { + return error("Unrecognized node test"); + } + + const char_t* nt_name_copy = alloc_string(nt_name); + if (!nt_name_copy) return 0; + + xpath_ast_node* n = alloc_node(ast_step, set, axis, nt_type, nt_name_copy); + if (!n) return 0; + + xpath_ast_node* last = 0; + + while (_lexer.current() == lex_open_square_brace) + { + _lexer.next(); + + xpath_ast_node* expr = parse_expression(); + if (!expr) return 0; + + xpath_ast_node* pred = alloc_node(ast_predicate, 0, expr, predicate_default); + if (!pred) return 0; + + if (_lexer.current() != lex_close_square_brace) + return error("Expected ']' to match an opening '['"); + _lexer.next(); + + if (last) last->set_next(pred); + else n->set_right(pred); + + last = pred; + } + + return n; + } + + // RelativeLocationPath ::= Step | RelativeLocationPath '/' Step | RelativeLocationPath '//' Step + xpath_ast_node* parse_relative_location_path(xpath_ast_node* set) + { + xpath_ast_node* n = parse_step(set); + if (!n) return 0; + + while (_lexer.current() == lex_slash || _lexer.current() == lex_double_slash) + { + lexeme_t l = _lexer.current(); + _lexer.next(); + + if (l == lex_double_slash) + { + n = alloc_node(ast_step, n, axis_descendant_or_self, nodetest_type_node, 0); + if (!n) return 0; + } + + n = parse_step(n); + if (!n) return 0; + } + + return n; + } + + // LocationPath ::= RelativeLocationPath | AbsoluteLocationPath + // AbsoluteLocationPath ::= '/' RelativeLocationPath? | '//' RelativeLocationPath + xpath_ast_node* parse_location_path() + { + if (_lexer.current() == lex_slash) + { + _lexer.next(); + + xpath_ast_node* n = alloc_node(ast_step_root, xpath_type_node_set); + if (!n) return 0; + + // relative location path can start from axis_attribute, dot, double_dot, multiply and string lexemes; any other lexeme means standalone root path + lexeme_t l = _lexer.current(); + + if (l == lex_string || l == lex_axis_attribute || l == lex_dot || l == lex_double_dot || l == lex_multiply) + return parse_relative_location_path(n); + else + return n; + } + else if (_lexer.current() == lex_double_slash) + { + _lexer.next(); + + xpath_ast_node* n = alloc_node(ast_step_root, xpath_type_node_set); + if (!n) return 0; + + n = alloc_node(ast_step, n, axis_descendant_or_self, nodetest_type_node, 0); + if (!n) return 0; + + return parse_relative_location_path(n); + } + + // else clause moved outside of if because of bogus warning 'control may reach end of non-void function being inlined' in gcc 4.0.1 + return parse_relative_location_path(0); + } + + // PathExpr ::= LocationPath + // | FilterExpr + // | FilterExpr '/' RelativeLocationPath + // | FilterExpr '//' RelativeLocationPath + // UnionExpr ::= PathExpr | UnionExpr '|' PathExpr + // UnaryExpr ::= UnionExpr | '-' UnaryExpr + xpath_ast_node* parse_path_or_unary_expression() + { + // Clarification. + // PathExpr begins with either LocationPath or FilterExpr. + // FilterExpr begins with PrimaryExpr + // PrimaryExpr begins with '$' in case of it being a variable reference, + // '(' in case of it being an expression, string literal, number constant or + // function call. + if (_lexer.current() == lex_var_ref || _lexer.current() == lex_open_brace || + _lexer.current() == lex_quoted_string || _lexer.current() == lex_number || + _lexer.current() == lex_string) + { + if (_lexer.current() == lex_string) + { + // This is either a function call, or not - if not, we shall proceed with location path + const char_t* state = _lexer.state(); + + while (PUGI__IS_CHARTYPE(*state, ct_space)) ++state; + + if (*state != '(') + return parse_location_path(); + + // This looks like a function call; however this still can be a node-test. Check it. + if (parse_node_test_type(_lexer.contents()) != nodetest_none) + return parse_location_path(); + } + + xpath_ast_node* n = parse_filter_expression(); + if (!n) return 0; + + if (_lexer.current() == lex_slash || _lexer.current() == lex_double_slash) + { + lexeme_t l = _lexer.current(); + _lexer.next(); + + if (l == lex_double_slash) + { + if (n->rettype() != xpath_type_node_set) + return error("Step has to be applied to node set"); + + n = alloc_node(ast_step, n, axis_descendant_or_self, nodetest_type_node, 0); + if (!n) return 0; + } + + // select from location path + return parse_relative_location_path(n); + } + + return n; + } + else if (_lexer.current() == lex_minus) + { + _lexer.next(); + + // precedence 7+ - only parses union expressions + xpath_ast_node* n = parse_expression(7); + if (!n) return 0; + + return alloc_node(ast_op_negate, xpath_type_number, n); + } + else + { + return parse_location_path(); + } + } + + struct binary_op_t + { + ast_type_t asttype; + xpath_value_type rettype; + int precedence; + + binary_op_t(): asttype(ast_unknown), rettype(xpath_type_none), precedence(0) + { + } + + binary_op_t(ast_type_t asttype_, xpath_value_type rettype_, int precedence_): asttype(asttype_), rettype(rettype_), precedence(precedence_) + { + } + + static binary_op_t parse(xpath_lexer& lexer) + { + switch (lexer.current()) + { + case lex_string: + if (lexer.contents() == PUGIXML_TEXT("or")) + return binary_op_t(ast_op_or, xpath_type_boolean, 1); + else if (lexer.contents() == PUGIXML_TEXT("and")) + return binary_op_t(ast_op_and, xpath_type_boolean, 2); + else if (lexer.contents() == PUGIXML_TEXT("div")) + return binary_op_t(ast_op_divide, xpath_type_number, 6); + else if (lexer.contents() == PUGIXML_TEXT("mod")) + return binary_op_t(ast_op_mod, xpath_type_number, 6); + else + return binary_op_t(); + + case lex_equal: + return binary_op_t(ast_op_equal, xpath_type_boolean, 3); + + case lex_not_equal: + return binary_op_t(ast_op_not_equal, xpath_type_boolean, 3); + + case lex_less: + return binary_op_t(ast_op_less, xpath_type_boolean, 4); + + case lex_greater: + return binary_op_t(ast_op_greater, xpath_type_boolean, 4); + + case lex_less_or_equal: + return binary_op_t(ast_op_less_or_equal, xpath_type_boolean, 4); + + case lex_greater_or_equal: + return binary_op_t(ast_op_greater_or_equal, xpath_type_boolean, 4); + + case lex_plus: + return binary_op_t(ast_op_add, xpath_type_number, 5); + + case lex_minus: + return binary_op_t(ast_op_subtract, xpath_type_number, 5); + + case lex_multiply: + return binary_op_t(ast_op_multiply, xpath_type_number, 6); + + case lex_union: + return binary_op_t(ast_op_union, xpath_type_node_set, 7); + + default: + return binary_op_t(); + } + } + }; + + xpath_ast_node* parse_expression_rec(xpath_ast_node* lhs, int limit) + { + binary_op_t op = binary_op_t::parse(_lexer); + + while (op.asttype != ast_unknown && op.precedence >= limit) + { + _lexer.next(); + + xpath_ast_node* rhs = parse_path_or_unary_expression(); + if (!rhs) return 0; + + binary_op_t nextop = binary_op_t::parse(_lexer); + + while (nextop.asttype != ast_unknown && nextop.precedence > op.precedence) + { + rhs = parse_expression_rec(rhs, nextop.precedence); + if (!rhs) return 0; + + nextop = binary_op_t::parse(_lexer); + } + + if (op.asttype == ast_op_union && (lhs->rettype() != xpath_type_node_set || rhs->rettype() != xpath_type_node_set)) + return error("Union operator has to be applied to node sets"); + + lhs = alloc_node(op.asttype, op.rettype, lhs, rhs); + if (!lhs) return 0; + + op = binary_op_t::parse(_lexer); + } + + return lhs; + } + + // Expr ::= OrExpr + // OrExpr ::= AndExpr | OrExpr 'or' AndExpr + // AndExpr ::= EqualityExpr | AndExpr 'and' EqualityExpr + // EqualityExpr ::= RelationalExpr + // | EqualityExpr '=' RelationalExpr + // | EqualityExpr '!=' RelationalExpr + // RelationalExpr ::= AdditiveExpr + // | RelationalExpr '<' AdditiveExpr + // | RelationalExpr '>' AdditiveExpr + // | RelationalExpr '<=' AdditiveExpr + // | RelationalExpr '>=' AdditiveExpr + // AdditiveExpr ::= MultiplicativeExpr + // | AdditiveExpr '+' MultiplicativeExpr + // | AdditiveExpr '-' MultiplicativeExpr + // MultiplicativeExpr ::= UnaryExpr + // | MultiplicativeExpr '*' UnaryExpr + // | MultiplicativeExpr 'div' UnaryExpr + // | MultiplicativeExpr 'mod' UnaryExpr + xpath_ast_node* parse_expression(int limit = 0) + { + xpath_ast_node* n = parse_path_or_unary_expression(); + if (!n) return 0; + + return parse_expression_rec(n, limit); + } + + xpath_parser(const char_t* query, xpath_variable_set* variables, xpath_allocator* alloc, xpath_parse_result* result): _alloc(alloc), _lexer(query), _query(query), _variables(variables), _result(result) + { + } + + xpath_ast_node* parse() + { + xpath_ast_node* n = parse_expression(); + if (!n) return 0; + + // check if there are unparsed tokens left + if (_lexer.current() != lex_eof) + return error("Incorrect query"); + + return n; + } + + static xpath_ast_node* parse(const char_t* query, xpath_variable_set* variables, xpath_allocator* alloc, xpath_parse_result* result) + { + xpath_parser parser(query, variables, alloc, result); + + return parser.parse(); + } + }; + + struct xpath_query_impl + { + static xpath_query_impl* create() + { + void* memory = xml_memory::allocate(sizeof(xpath_query_impl)); + if (!memory) return 0; + + return new (memory) xpath_query_impl(); + } + + static void destroy(xpath_query_impl* impl) + { + // free all allocated pages + impl->alloc.release(); + + // free allocator memory (with the first page) + xml_memory::deallocate(impl); + } + + xpath_query_impl(): root(0), alloc(&block, &oom), oom(false) + { + block.next = 0; + block.capacity = sizeof(block.data); + } + + xpath_ast_node* root; + xpath_allocator alloc; + xpath_memory_block block; + bool oom; + }; + + PUGI__FN impl::xpath_ast_node* evaluate_node_set_prepare(xpath_query_impl* impl) + { + if (!impl) return 0; + + if (impl->root->rettype() != xpath_type_node_set) + { + #ifdef PUGIXML_NO_EXCEPTIONS + return 0; + #else + xpath_parse_result res; + res.error = "Expression does not evaluate to node set"; + + throw xpath_exception(res); + #endif + } + + return impl->root; + } +PUGI__NS_END + +namespace pugi +{ +#ifndef PUGIXML_NO_EXCEPTIONS + PUGI__FN xpath_exception::xpath_exception(const xpath_parse_result& result_): _result(result_) + { + assert(_result.error); + } + + PUGI__FN const char* xpath_exception::what() const throw() + { + return _result.error; + } + + PUGI__FN const xpath_parse_result& xpath_exception::result() const + { + return _result; + } +#endif + + PUGI__FN xpath_node::xpath_node() + { + } + + PUGI__FN xpath_node::xpath_node(const xml_node& node_): _node(node_) + { + } + + PUGI__FN xpath_node::xpath_node(const xml_attribute& attribute_, const xml_node& parent_): _node(attribute_ ? parent_ : xml_node()), _attribute(attribute_) + { + } + + PUGI__FN xml_node xpath_node::node() const + { + return _attribute ? xml_node() : _node; + } + + PUGI__FN xml_attribute xpath_node::attribute() const + { + return _attribute; + } + + PUGI__FN xml_node xpath_node::parent() const + { + return _attribute ? _node : _node.parent(); + } + + PUGI__FN static void unspecified_bool_xpath_node(xpath_node***) + { + } + + PUGI__FN xpath_node::operator xpath_node::unspecified_bool_type() const + { + return (_node || _attribute) ? unspecified_bool_xpath_node : 0; + } + + PUGI__FN bool xpath_node::operator!() const + { + return !(_node || _attribute); + } + + PUGI__FN bool xpath_node::operator==(const xpath_node& n) const + { + return _node == n._node && _attribute == n._attribute; + } + + PUGI__FN bool xpath_node::operator!=(const xpath_node& n) const + { + return _node != n._node || _attribute != n._attribute; + } + +#ifdef __BORLANDC__ + PUGI__FN bool operator&&(const xpath_node& lhs, bool rhs) + { + return (bool)lhs && rhs; + } + + PUGI__FN bool operator||(const xpath_node& lhs, bool rhs) + { + return (bool)lhs || rhs; + } +#endif + + PUGI__FN void xpath_node_set::_assign(const_iterator begin_, const_iterator end_, type_t type_) + { + assert(begin_ <= end_); + + size_t size_ = static_cast(end_ - begin_); + + // use internal buffer for 0 or 1 elements, heap buffer otherwise + xpath_node* storage = (size_ <= 1) ? _storage : static_cast(impl::xml_memory::allocate(size_ * sizeof(xpath_node))); + + if (!storage) + { + #ifdef PUGIXML_NO_EXCEPTIONS + return; + #else + throw std::bad_alloc(); + #endif + } + + // deallocate old buffer + if (_begin != _storage) + impl::xml_memory::deallocate(_begin); + + // size check is necessary because for begin_ = end_ = nullptr, memcpy is UB + if (size_) + memcpy(storage, begin_, size_ * sizeof(xpath_node)); + + _begin = storage; + _end = storage + size_; + _type = type_; + } + +#ifdef PUGIXML_HAS_MOVE + PUGI__FN void xpath_node_set::_move(xpath_node_set& rhs) PUGIXML_NOEXCEPT + { + _type = rhs._type; + _storage[0] = rhs._storage[0]; + _begin = (rhs._begin == rhs._storage) ? _storage : rhs._begin; + _end = _begin + (rhs._end - rhs._begin); + + rhs._type = type_unsorted; + rhs._begin = rhs._storage; + rhs._end = rhs._storage; + } +#endif + + PUGI__FN xpath_node_set::xpath_node_set(): _type(type_unsorted), _begin(_storage), _end(_storage) + { + } + + PUGI__FN xpath_node_set::xpath_node_set(const_iterator begin_, const_iterator end_, type_t type_): _type(type_unsorted), _begin(_storage), _end(_storage) + { + _assign(begin_, end_, type_); + } + + PUGI__FN xpath_node_set::~xpath_node_set() + { + if (_begin != _storage) + impl::xml_memory::deallocate(_begin); + } + + PUGI__FN xpath_node_set::xpath_node_set(const xpath_node_set& ns): _type(type_unsorted), _begin(_storage), _end(_storage) + { + _assign(ns._begin, ns._end, ns._type); + } + + PUGI__FN xpath_node_set& xpath_node_set::operator=(const xpath_node_set& ns) + { + if (this == &ns) return *this; + + _assign(ns._begin, ns._end, ns._type); + + return *this; + } + +#ifdef PUGIXML_HAS_MOVE + PUGI__FN xpath_node_set::xpath_node_set(xpath_node_set&& rhs) PUGIXML_NOEXCEPT: _type(type_unsorted), _begin(_storage), _end(_storage) + { + _move(rhs); + } + + PUGI__FN xpath_node_set& xpath_node_set::operator=(xpath_node_set&& rhs) PUGIXML_NOEXCEPT + { + if (this == &rhs) return *this; + + if (_begin != _storage) + impl::xml_memory::deallocate(_begin); + + _move(rhs); + + return *this; + } +#endif + + PUGI__FN xpath_node_set::type_t xpath_node_set::type() const + { + return _type; + } + + PUGI__FN size_t xpath_node_set::size() const + { + return _end - _begin; + } + + PUGI__FN bool xpath_node_set::empty() const + { + return _begin == _end; + } + + PUGI__FN const xpath_node& xpath_node_set::operator[](size_t index) const + { + assert(index < size()); + return _begin[index]; + } + + PUGI__FN xpath_node_set::const_iterator xpath_node_set::begin() const + { + return _begin; + } + + PUGI__FN xpath_node_set::const_iterator xpath_node_set::end() const + { + return _end; + } + + PUGI__FN void xpath_node_set::sort(bool reverse) + { + _type = impl::xpath_sort(_begin, _end, _type, reverse); + } + + PUGI__FN xpath_node xpath_node_set::first() const + { + return impl::xpath_first(_begin, _end, _type); + } + + PUGI__FN xpath_parse_result::xpath_parse_result(): error("Internal error"), offset(0) + { + } + + PUGI__FN xpath_parse_result::operator bool() const + { + return error == 0; + } + + PUGI__FN const char* xpath_parse_result::description() const + { + return error ? error : "No error"; + } + + PUGI__FN xpath_variable::xpath_variable(xpath_value_type type_): _type(type_), _next(0) + { + } + + PUGI__FN const char_t* xpath_variable::name() const + { + switch (_type) + { + case xpath_type_node_set: + return static_cast(this)->name; + + case xpath_type_number: + return static_cast(this)->name; + + case xpath_type_string: + return static_cast(this)->name; + + case xpath_type_boolean: + return static_cast(this)->name; + + default: + assert(false && "Invalid variable type"); // unreachable + return 0; + } + } + + PUGI__FN xpath_value_type xpath_variable::type() const + { + return _type; + } + + PUGI__FN bool xpath_variable::get_boolean() const + { + return (_type == xpath_type_boolean) ? static_cast(this)->value : false; + } + + PUGI__FN double xpath_variable::get_number() const + { + return (_type == xpath_type_number) ? static_cast(this)->value : impl::gen_nan(); + } + + PUGI__FN const char_t* xpath_variable::get_string() const + { + const char_t* value = (_type == xpath_type_string) ? static_cast(this)->value : 0; + return value ? value : PUGIXML_TEXT(""); + } + + PUGI__FN const xpath_node_set& xpath_variable::get_node_set() const + { + return (_type == xpath_type_node_set) ? static_cast(this)->value : impl::dummy_node_set; + } + + PUGI__FN bool xpath_variable::set(bool value) + { + if (_type != xpath_type_boolean) return false; + + static_cast(this)->value = value; + return true; + } + + PUGI__FN bool xpath_variable::set(double value) + { + if (_type != xpath_type_number) return false; + + static_cast(this)->value = value; + return true; + } + + PUGI__FN bool xpath_variable::set(const char_t* value) + { + if (_type != xpath_type_string) return false; + + impl::xpath_variable_string* var = static_cast(this); + + // duplicate string + size_t size = (impl::strlength(value) + 1) * sizeof(char_t); + + char_t* copy = static_cast(impl::xml_memory::allocate(size)); + if (!copy) return false; + + memcpy(copy, value, size); + + // replace old string + if (var->value) impl::xml_memory::deallocate(var->value); + var->value = copy; + + return true; + } + + PUGI__FN bool xpath_variable::set(const xpath_node_set& value) + { + if (_type != xpath_type_node_set) return false; + + static_cast(this)->value = value; + return true; + } + + PUGI__FN xpath_variable_set::xpath_variable_set() + { + for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i) + _data[i] = 0; + } + + PUGI__FN xpath_variable_set::~xpath_variable_set() + { + for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i) + _destroy(_data[i]); + } + + PUGI__FN xpath_variable_set::xpath_variable_set(const xpath_variable_set& rhs) + { + for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i) + _data[i] = 0; + + _assign(rhs); + } + + PUGI__FN xpath_variable_set& xpath_variable_set::operator=(const xpath_variable_set& rhs) + { + if (this == &rhs) return *this; + + _assign(rhs); + + return *this; + } + +#ifdef PUGIXML_HAS_MOVE + PUGI__FN xpath_variable_set::xpath_variable_set(xpath_variable_set&& rhs) PUGIXML_NOEXCEPT + { + for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i) + { + _data[i] = rhs._data[i]; + rhs._data[i] = 0; + } + } + + PUGI__FN xpath_variable_set& xpath_variable_set::operator=(xpath_variable_set&& rhs) PUGIXML_NOEXCEPT + { + for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i) + { + _destroy(_data[i]); + + _data[i] = rhs._data[i]; + rhs._data[i] = 0; + } + + return *this; + } +#endif + + PUGI__FN void xpath_variable_set::_assign(const xpath_variable_set& rhs) + { + xpath_variable_set temp; + + for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i) + if (rhs._data[i] && !_clone(rhs._data[i], &temp._data[i])) + return; + + _swap(temp); + } + + PUGI__FN void xpath_variable_set::_swap(xpath_variable_set& rhs) + { + for (size_t i = 0; i < sizeof(_data) / sizeof(_data[0]); ++i) + { + xpath_variable* chain = _data[i]; + + _data[i] = rhs._data[i]; + rhs._data[i] = chain; + } + } + + PUGI__FN xpath_variable* xpath_variable_set::_find(const char_t* name) const + { + const size_t hash_size = sizeof(_data) / sizeof(_data[0]); + size_t hash = impl::hash_string(name) % hash_size; + + // look for existing variable + for (xpath_variable* var = _data[hash]; var; var = var->_next) + if (impl::strequal(var->name(), name)) + return var; + + return 0; + } + + PUGI__FN bool xpath_variable_set::_clone(xpath_variable* var, xpath_variable** out_result) + { + xpath_variable* last = 0; + + while (var) + { + // allocate storage for new variable + xpath_variable* nvar = impl::new_xpath_variable(var->_type, var->name()); + if (!nvar) return false; + + // link the variable to the result immediately to handle failures gracefully + if (last) + last->_next = nvar; + else + *out_result = nvar; + + last = nvar; + + // copy the value; this can fail due to out-of-memory conditions + if (!impl::copy_xpath_variable(nvar, var)) return false; + + var = var->_next; + } + + return true; + } + + PUGI__FN void xpath_variable_set::_destroy(xpath_variable* var) + { + while (var) + { + xpath_variable* next = var->_next; + + impl::delete_xpath_variable(var->_type, var); + + var = next; + } + } + + PUGI__FN xpath_variable* xpath_variable_set::add(const char_t* name, xpath_value_type type) + { + const size_t hash_size = sizeof(_data) / sizeof(_data[0]); + size_t hash = impl::hash_string(name) % hash_size; + + // look for existing variable + for (xpath_variable* var = _data[hash]; var; var = var->_next) + if (impl::strequal(var->name(), name)) + return var->type() == type ? var : 0; + + // add new variable + xpath_variable* result = impl::new_xpath_variable(type, name); + + if (result) + { + result->_next = _data[hash]; + + _data[hash] = result; + } + + return result; + } + + PUGI__FN bool xpath_variable_set::set(const char_t* name, bool value) + { + xpath_variable* var = add(name, xpath_type_boolean); + return var ? var->set(value) : false; + } + + PUGI__FN bool xpath_variable_set::set(const char_t* name, double value) + { + xpath_variable* var = add(name, xpath_type_number); + return var ? var->set(value) : false; + } + + PUGI__FN bool xpath_variable_set::set(const char_t* name, const char_t* value) + { + xpath_variable* var = add(name, xpath_type_string); + return var ? var->set(value) : false; + } + + PUGI__FN bool xpath_variable_set::set(const char_t* name, const xpath_node_set& value) + { + xpath_variable* var = add(name, xpath_type_node_set); + return var ? var->set(value) : false; + } + + PUGI__FN xpath_variable* xpath_variable_set::get(const char_t* name) + { + return _find(name); + } + + PUGI__FN const xpath_variable* xpath_variable_set::get(const char_t* name) const + { + return _find(name); + } + + PUGI__FN xpath_query::xpath_query(const char_t* query, xpath_variable_set* variables): _impl(0) + { + impl::xpath_query_impl* qimpl = impl::xpath_query_impl::create(); + + if (!qimpl) + { + #ifdef PUGIXML_NO_EXCEPTIONS + _result.error = "Out of memory"; + #else + throw std::bad_alloc(); + #endif + } + else + { + using impl::auto_deleter; // MSVC7 workaround + auto_deleter impl(qimpl, impl::xpath_query_impl::destroy); + + qimpl->root = impl::xpath_parser::parse(query, variables, &qimpl->alloc, &_result); + + if (qimpl->root) + { + qimpl->root->optimize(&qimpl->alloc); + + _impl = impl.release(); + _result.error = 0; + } + else + { + #ifdef PUGIXML_NO_EXCEPTIONS + if (qimpl->oom) _result.error = "Out of memory"; + #else + if (qimpl->oom) throw std::bad_alloc(); + throw xpath_exception(_result); + #endif + } + } + } + + PUGI__FN xpath_query::xpath_query(): _impl(0) + { + } + + PUGI__FN xpath_query::~xpath_query() + { + if (_impl) + impl::xpath_query_impl::destroy(static_cast(_impl)); + } + +#ifdef PUGIXML_HAS_MOVE + PUGI__FN xpath_query::xpath_query(xpath_query&& rhs) PUGIXML_NOEXCEPT + { + _impl = rhs._impl; + _result = rhs._result; + rhs._impl = 0; + rhs._result = xpath_parse_result(); + } + + PUGI__FN xpath_query& xpath_query::operator=(xpath_query&& rhs) PUGIXML_NOEXCEPT + { + if (this == &rhs) return *this; + + if (_impl) + impl::xpath_query_impl::destroy(static_cast(_impl)); + + _impl = rhs._impl; + _result = rhs._result; + rhs._impl = 0; + rhs._result = xpath_parse_result(); + + return *this; + } +#endif + + PUGI__FN xpath_value_type xpath_query::return_type() const + { + if (!_impl) return xpath_type_none; + + return static_cast(_impl)->root->rettype(); + } + + PUGI__FN bool xpath_query::evaluate_boolean(const xpath_node& n) const + { + if (!_impl) return false; + + impl::xpath_context c(n, 1, 1); + impl::xpath_stack_data sd; + + bool r = static_cast(_impl)->root->eval_boolean(c, sd.stack); + + if (sd.oom) + { + #ifdef PUGIXML_NO_EXCEPTIONS + return false; + #else + throw std::bad_alloc(); + #endif + } + + return r; + } + + PUGI__FN double xpath_query::evaluate_number(const xpath_node& n) const + { + if (!_impl) return impl::gen_nan(); + + impl::xpath_context c(n, 1, 1); + impl::xpath_stack_data sd; + + double r = static_cast(_impl)->root->eval_number(c, sd.stack); + + if (sd.oom) + { + #ifdef PUGIXML_NO_EXCEPTIONS + return impl::gen_nan(); + #else + throw std::bad_alloc(); + #endif + } + + return r; + } + +#ifndef PUGIXML_NO_STL + PUGI__FN string_t xpath_query::evaluate_string(const xpath_node& n) const + { + if (!_impl) return string_t(); + + impl::xpath_context c(n, 1, 1); + impl::xpath_stack_data sd; + + impl::xpath_string r = static_cast(_impl)->root->eval_string(c, sd.stack); + + if (sd.oom) + { + #ifdef PUGIXML_NO_EXCEPTIONS + return string_t(); + #else + throw std::bad_alloc(); + #endif + } + + return string_t(r.c_str(), r.length()); + } +#endif + + PUGI__FN size_t xpath_query::evaluate_string(char_t* buffer, size_t capacity, const xpath_node& n) const + { + impl::xpath_context c(n, 1, 1); + impl::xpath_stack_data sd; + + impl::xpath_string r = _impl ? static_cast(_impl)->root->eval_string(c, sd.stack) : impl::xpath_string(); + + if (sd.oom) + { + #ifdef PUGIXML_NO_EXCEPTIONS + r = impl::xpath_string(); + #else + throw std::bad_alloc(); + #endif + } + + size_t full_size = r.length() + 1; + + if (capacity > 0) + { + size_t size = (full_size < capacity) ? full_size : capacity; + assert(size > 0); + + memcpy(buffer, r.c_str(), (size - 1) * sizeof(char_t)); + buffer[size - 1] = 0; + } + + return full_size; + } + + PUGI__FN xpath_node_set xpath_query::evaluate_node_set(const xpath_node& n) const + { + impl::xpath_ast_node* root = impl::evaluate_node_set_prepare(static_cast(_impl)); + if (!root) return xpath_node_set(); + + impl::xpath_context c(n, 1, 1); + impl::xpath_stack_data sd; + + impl::xpath_node_set_raw r = root->eval_node_set(c, sd.stack, impl::nodeset_eval_all); + + if (sd.oom) + { + #ifdef PUGIXML_NO_EXCEPTIONS + return xpath_node_set(); + #else + throw std::bad_alloc(); + #endif + } + + return xpath_node_set(r.begin(), r.end(), r.type()); + } + + PUGI__FN xpath_node xpath_query::evaluate_node(const xpath_node& n) const + { + impl::xpath_ast_node* root = impl::evaluate_node_set_prepare(static_cast(_impl)); + if (!root) return xpath_node(); + + impl::xpath_context c(n, 1, 1); + impl::xpath_stack_data sd; + + impl::xpath_node_set_raw r = root->eval_node_set(c, sd.stack, impl::nodeset_eval_first); + + if (sd.oom) + { + #ifdef PUGIXML_NO_EXCEPTIONS + return xpath_node(); + #else + throw std::bad_alloc(); + #endif + } + + return r.first(); + } + + PUGI__FN const xpath_parse_result& xpath_query::result() const + { + return _result; + } + + PUGI__FN static void unspecified_bool_xpath_query(xpath_query***) + { + } + + PUGI__FN xpath_query::operator xpath_query::unspecified_bool_type() const + { + return _impl ? unspecified_bool_xpath_query : 0; + } + + PUGI__FN bool xpath_query::operator!() const + { + return !_impl; + } + + PUGI__FN xpath_node xml_node::select_node(const char_t* query, xpath_variable_set* variables) const + { + xpath_query q(query, variables); + return q.evaluate_node(*this); + } + + PUGI__FN xpath_node xml_node::select_node(const xpath_query& query) const + { + return query.evaluate_node(*this); + } + + PUGI__FN xpath_node_set xml_node::select_nodes(const char_t* query, xpath_variable_set* variables) const + { + xpath_query q(query, variables); + return q.evaluate_node_set(*this); + } + + PUGI__FN xpath_node_set xml_node::select_nodes(const xpath_query& query) const + { + return query.evaluate_node_set(*this); + } + + PUGI__FN xpath_node xml_node::select_single_node(const char_t* query, xpath_variable_set* variables) const + { + xpath_query q(query, variables); + return q.evaluate_node(*this); + } + + PUGI__FN xpath_node xml_node::select_single_node(const xpath_query& query) const + { + return query.evaluate_node(*this); + } +} + +#endif + +#ifdef __BORLANDC__ +# pragma option pop +#endif + +// Intel C++ does not properly keep warning state for function templates, +// so popping warning state at the end of translation unit leads to warnings in the middle. +#if defined(_MSC_VER) && !defined(__INTEL_COMPILER) +# pragma warning(pop) +#endif + +#if defined(_MSC_VER) && defined(__c2__) +# pragma clang diagnostic pop +#endif + +// Undefine all local macros (makes sure we're not leaking macros in header-only mode) +#undef PUGI__NO_INLINE +#undef PUGI__UNLIKELY +#undef PUGI__STATIC_ASSERT +#undef PUGI__DMC_VOLATILE +#undef PUGI__UNSIGNED_OVERFLOW +#undef PUGI__MSVC_CRT_VERSION +#undef PUGI__SNPRINTF +#undef PUGI__NS_BEGIN +#undef PUGI__NS_END +#undef PUGI__FN +#undef PUGI__FN_NO_INLINE +#undef PUGI__GETHEADER_IMPL +#undef PUGI__GETPAGE_IMPL +#undef PUGI__GETPAGE +#undef PUGI__NODETYPE +#undef PUGI__IS_CHARTYPE_IMPL +#undef PUGI__IS_CHARTYPE +#undef PUGI__IS_CHARTYPEX +#undef PUGI__ENDSWITH +#undef PUGI__SKIPWS +#undef PUGI__OPTSET +#undef PUGI__PUSHNODE +#undef PUGI__POPNODE +#undef PUGI__SCANFOR +#undef PUGI__SCANWHILE +#undef PUGI__SCANWHILE_UNROLL +#undef PUGI__ENDSEG +#undef PUGI__THROW_ERROR +#undef PUGI__CHECK_ERROR + +#endif + +/** + * Copyright (c) 2006-2019 Arseny Kapoulkine + * + * Permission is hereby granted, free of charge, to any person + * obtaining a copy of this software and associated documentation + * files (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use, + * copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following + * conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES + * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ diff --git a/source/utils/pugixml.hpp b/source/utils/pugixml.hpp new file mode 100644 index 0000000..64e735e --- /dev/null +++ b/source/utils/pugixml.hpp @@ -0,0 +1,1477 @@ +/** + * pugixml parser - version 1.10 + * -------------------------------------------------------- + * Copyright (C) 2006-2019, by Arseny Kapoulkine (arseny.kapoulkine@gmail.com) + * Report bugs and download new versions at https://pugixml.org/ + * + * This library is distributed under the MIT License. See notice at the end + * of this file. + * + * This work is based on the pugxml parser, which is: + * Copyright (C) 2003, by Kristen Wegner (kristen@tima.net) + */ + +#ifndef PUGIXML_VERSION +// Define version macro; evaluates to major * 1000 + minor * 10 + patch so that it's safe to use in less-than comparisons +// Note: pugixml used major * 100 + minor * 10 + patch format up until 1.9 (which had version identifier 190); starting from pugixml 1.10, the minor version number is two digits +# define PUGIXML_VERSION 1100 +#endif + +// Include user configuration file (this can define various configuration macros) +#include "pugiconfig.hpp" + +#ifndef HEADER_PUGIXML_HPP +#define HEADER_PUGIXML_HPP + +// Include stddef.h for size_t and ptrdiff_t +#include + +// Include exception header for XPath +#if !defined(PUGIXML_NO_XPATH) && !defined(PUGIXML_NO_EXCEPTIONS) +# include +#endif + +// Include STL headers +#ifndef PUGIXML_NO_STL +# include +# include +# include +#endif + +// Macro for deprecated features +#ifndef PUGIXML_DEPRECATED +# if defined(__GNUC__) +# define PUGIXML_DEPRECATED __attribute__((deprecated)) +# elif defined(_MSC_VER) && _MSC_VER >= 1300 +# define PUGIXML_DEPRECATED __declspec(deprecated) +# else +# define PUGIXML_DEPRECATED +# endif +#endif + +// If no API is defined, assume default +#ifndef PUGIXML_API +# define PUGIXML_API +#endif + +// If no API for classes is defined, assume default +#ifndef PUGIXML_CLASS +# define PUGIXML_CLASS PUGIXML_API +#endif + +// If no API for functions is defined, assume default +#ifndef PUGIXML_FUNCTION +# define PUGIXML_FUNCTION PUGIXML_API +#endif + +// If the platform is known to have long long support, enable long long functions +#ifndef PUGIXML_HAS_LONG_LONG +# if __cplusplus >= 201103 +# define PUGIXML_HAS_LONG_LONG +# elif defined(_MSC_VER) && _MSC_VER >= 1400 +# define PUGIXML_HAS_LONG_LONG +# endif +#endif + +// If the platform is known to have move semantics support, compile move ctor/operator implementation +#ifndef PUGIXML_HAS_MOVE +# if __cplusplus >= 201103 +# define PUGIXML_HAS_MOVE +# elif defined(_MSC_VER) && _MSC_VER >= 1600 +# define PUGIXML_HAS_MOVE +# endif +#endif + +// If C++ is 2011 or higher, add 'noexcept' specifiers +#ifndef PUGIXML_NOEXCEPT +# if __cplusplus >= 201103 +# define PUGIXML_NOEXCEPT noexcept +# elif defined(_MSC_VER) && _MSC_VER >= 1900 +# define PUGIXML_NOEXCEPT noexcept +# else +# define PUGIXML_NOEXCEPT +# endif +#endif + +// Some functions can not be noexcept in compact mode +#ifdef PUGIXML_COMPACT +# define PUGIXML_NOEXCEPT_IF_NOT_COMPACT +#else +# define PUGIXML_NOEXCEPT_IF_NOT_COMPACT PUGIXML_NOEXCEPT +#endif + +// If C++ is 2011 or higher, add 'override' qualifiers +#ifndef PUGIXML_OVERRIDE +# if __cplusplus >= 201103 +# define PUGIXML_OVERRIDE override +# elif defined(_MSC_VER) && _MSC_VER >= 1700 +# define PUGIXML_OVERRIDE override +# else +# define PUGIXML_OVERRIDE +# endif +#endif + +// Character interface macros +#ifdef PUGIXML_WCHAR_MODE +# define PUGIXML_TEXT(t) L ## t +# define PUGIXML_CHAR wchar_t +#else +# define PUGIXML_TEXT(t) t +# define PUGIXML_CHAR char +#endif + +namespace pugi +{ + // Character type used for all internal storage and operations; depends on PUGIXML_WCHAR_MODE + typedef PUGIXML_CHAR char_t; + +#ifndef PUGIXML_NO_STL + // String type used for operations that work with STL string; depends on PUGIXML_WCHAR_MODE + typedef std::basic_string, std::allocator > string_t; +#endif +} + +// The PugiXML namespace +namespace pugi +{ + // Tree node types + enum xml_node_type + { + node_null, // Empty (null) node handle + node_document, // A document tree's absolute root + node_element, // Element tag, i.e. '' + node_pcdata, // Plain character data, i.e. 'text' + node_cdata, // Character data, i.e. '' + node_comment, // Comment tag, i.e. '' + node_pi, // Processing instruction, i.e. '' + node_declaration, // Document declaration, i.e. '' + node_doctype // Document type declaration, i.e. '' + }; + + // Parsing options + + // Minimal parsing mode (equivalent to turning all other flags off). + // Only elements and PCDATA sections are added to the DOM tree, no text conversions are performed. + const unsigned int parse_minimal = 0x0000; + + // This flag determines if processing instructions (node_pi) are added to the DOM tree. This flag is off by default. + const unsigned int parse_pi = 0x0001; + + // This flag determines if comments (node_comment) are added to the DOM tree. This flag is off by default. + const unsigned int parse_comments = 0x0002; + + // This flag determines if CDATA sections (node_cdata) are added to the DOM tree. This flag is on by default. + const unsigned int parse_cdata = 0x0004; + + // This flag determines if plain character data (node_pcdata) that consist only of whitespace are added to the DOM tree. + // This flag is off by default; turning it on usually results in slower parsing and more memory consumption. + const unsigned int parse_ws_pcdata = 0x0008; + + // This flag determines if character and entity references are expanded during parsing. This flag is on by default. + const unsigned int parse_escapes = 0x0010; + + // This flag determines if EOL characters are normalized (converted to #xA) during parsing. This flag is on by default. + const unsigned int parse_eol = 0x0020; + + // This flag determines if attribute values are normalized using CDATA normalization rules during parsing. This flag is on by default. + const unsigned int parse_wconv_attribute = 0x0040; + + // This flag determines if attribute values are normalized using NMTOKENS normalization rules during parsing. This flag is off by default. + const unsigned int parse_wnorm_attribute = 0x0080; + + // This flag determines if document declaration (node_declaration) is added to the DOM tree. This flag is off by default. + const unsigned int parse_declaration = 0x0100; + + // This flag determines if document type declaration (node_doctype) is added to the DOM tree. This flag is off by default. + const unsigned int parse_doctype = 0x0200; + + // This flag determines if plain character data (node_pcdata) that is the only child of the parent node and that consists only + // of whitespace is added to the DOM tree. + // This flag is off by default; turning it on may result in slower parsing and more memory consumption. + const unsigned int parse_ws_pcdata_single = 0x0400; + + // This flag determines if leading and trailing whitespace is to be removed from plain character data. This flag is off by default. + const unsigned int parse_trim_pcdata = 0x0800; + + // This flag determines if plain character data that does not have a parent node is added to the DOM tree, and if an empty document + // is a valid document. This flag is off by default. + const unsigned int parse_fragment = 0x1000; + + // This flag determines if plain character data is be stored in the parent element's value. This significantly changes the structure of + // the document; this flag is only recommended for parsing documents with many PCDATA nodes in memory-constrained environments. + // This flag is off by default. + const unsigned int parse_embed_pcdata = 0x2000; + + // The default parsing mode. + // Elements, PCDATA and CDATA sections are added to the DOM tree, character/reference entities are expanded, + // End-of-Line characters are normalized, attribute values are normalized using CDATA normalization rules. + const unsigned int parse_default = parse_cdata | parse_escapes | parse_wconv_attribute | parse_eol; + + // The full parsing mode. + // Nodes of all types are added to the DOM tree, character/reference entities are expanded, + // End-of-Line characters are normalized, attribute values are normalized using CDATA normalization rules. + const unsigned int parse_full = parse_default | parse_pi | parse_comments | parse_declaration | parse_doctype; + + // These flags determine the encoding of input data for XML document + enum xml_encoding + { + encoding_auto, // Auto-detect input encoding using BOM or < / class xml_object_range + { + public: + typedef It const_iterator; + typedef It iterator; + + xml_object_range(It b, It e): _begin(b), _end(e) + { + } + + It begin() const { return _begin; } + It end() const { return _end; } + + private: + It _begin, _end; + }; + + // Writer interface for node printing (see xml_node::print) + class PUGIXML_CLASS xml_writer + { + public: + virtual ~xml_writer() {} + + // Write memory chunk into stream/file/whatever + virtual void write(const void* data, size_t size) = 0; + }; + + // xml_writer implementation for FILE* + class PUGIXML_CLASS xml_writer_file: public xml_writer + { + public: + // Construct writer from a FILE* object; void* is used to avoid header dependencies on stdio + xml_writer_file(void* file); + + virtual void write(const void* data, size_t size) PUGIXML_OVERRIDE; + + private: + void* file; + }; + + #ifndef PUGIXML_NO_STL + // xml_writer implementation for streams + class PUGIXML_CLASS xml_writer_stream: public xml_writer + { + public: + // Construct writer from an output stream object + xml_writer_stream(std::basic_ostream >& stream); + xml_writer_stream(std::basic_ostream >& stream); + + virtual void write(const void* data, size_t size) PUGIXML_OVERRIDE; + + private: + std::basic_ostream >* narrow_stream; + std::basic_ostream >* wide_stream; + }; + #endif + + // A light-weight handle for manipulating attributes in DOM tree + class PUGIXML_CLASS xml_attribute + { + friend class xml_attribute_iterator; + friend class xml_node; + + private: + xml_attribute_struct* _attr; + + typedef void (*unspecified_bool_type)(xml_attribute***); + + public: + // Default constructor. Constructs an empty attribute. + xml_attribute(); + + // Constructs attribute from internal pointer + explicit xml_attribute(xml_attribute_struct* attr); + + // Safe bool conversion operator + operator unspecified_bool_type() const; + + // Borland C++ workaround + bool operator!() const; + + // Comparison operators (compares wrapped attribute pointers) + bool operator==(const xml_attribute& r) const; + bool operator!=(const xml_attribute& r) const; + bool operator<(const xml_attribute& r) const; + bool operator>(const xml_attribute& r) const; + bool operator<=(const xml_attribute& r) const; + bool operator>=(const xml_attribute& r) const; + + // Check if attribute is empty + bool empty() const; + + // Get attribute name/value, or "" if attribute is empty + const char_t* name() const; + const char_t* value() const; + + // Get attribute value, or the default value if attribute is empty + const char_t* as_string(const char_t* def = PUGIXML_TEXT("")) const; + + // Get attribute value as a number, or the default value if conversion did not succeed or attribute is empty + int as_int(int def = 0) const; + unsigned int as_uint(unsigned int def = 0) const; + double as_double(double def = 0) const; + float as_float(float def = 0) const; + + #ifdef PUGIXML_HAS_LONG_LONG + long long as_llong(long long def = 0) const; + unsigned long long as_ullong(unsigned long long def = 0) const; + #endif + + // Get attribute value as bool (returns true if first character is in '1tTyY' set), or the default value if attribute is empty + bool as_bool(bool def = false) const; + + // Set attribute name/value (returns false if attribute is empty or there is not enough memory) + bool set_name(const char_t* rhs); + bool set_value(const char_t* rhs); + + // Set attribute value with type conversion (numbers are converted to strings, boolean is converted to "true"/"false") + bool set_value(int rhs); + bool set_value(unsigned int rhs); + bool set_value(long rhs); + bool set_value(unsigned long rhs); + bool set_value(double rhs); + bool set_value(float rhs); + bool set_value(bool rhs); + + #ifdef PUGIXML_HAS_LONG_LONG + bool set_value(long long rhs); + bool set_value(unsigned long long rhs); + #endif + + // Set attribute value (equivalent to set_value without error checking) + xml_attribute& operator=(const char_t* rhs); + xml_attribute& operator=(int rhs); + xml_attribute& operator=(unsigned int rhs); + xml_attribute& operator=(long rhs); + xml_attribute& operator=(unsigned long rhs); + xml_attribute& operator=(double rhs); + xml_attribute& operator=(float rhs); + xml_attribute& operator=(bool rhs); + + #ifdef PUGIXML_HAS_LONG_LONG + xml_attribute& operator=(long long rhs); + xml_attribute& operator=(unsigned long long rhs); + #endif + + // Get next/previous attribute in the attribute list of the parent node + xml_attribute next_attribute() const; + xml_attribute previous_attribute() const; + + // Get hash value (unique for handles to the same object) + size_t hash_value() const; + + // Get internal pointer + xml_attribute_struct* internal_object() const; + }; + +#ifdef __BORLANDC__ + // Borland C++ workaround + bool PUGIXML_FUNCTION operator&&(const xml_attribute& lhs, bool rhs); + bool PUGIXML_FUNCTION operator||(const xml_attribute& lhs, bool rhs); +#endif + + // A light-weight handle for manipulating nodes in DOM tree + class PUGIXML_CLASS xml_node + { + friend class xml_attribute_iterator; + friend class xml_node_iterator; + friend class xml_named_node_iterator; + + protected: + xml_node_struct* _root; + + typedef void (*unspecified_bool_type)(xml_node***); + + public: + // Default constructor. Constructs an empty node. + xml_node(); + + // Constructs node from internal pointer + explicit xml_node(xml_node_struct* p); + + // Safe bool conversion operator + operator unspecified_bool_type() const; + + // Borland C++ workaround + bool operator!() const; + + // Comparison operators (compares wrapped node pointers) + bool operator==(const xml_node& r) const; + bool operator!=(const xml_node& r) const; + bool operator<(const xml_node& r) const; + bool operator>(const xml_node& r) const; + bool operator<=(const xml_node& r) const; + bool operator>=(const xml_node& r) const; + + // Check if node is empty. + bool empty() const; + + // Get node type + xml_node_type type() const; + + // Get node name, or "" if node is empty or it has no name + const char_t* name() const; + + // Get node value, or "" if node is empty or it has no value + // Note: For text node.value() does not return "text"! Use child_value() or text() methods to access text inside nodes. + const char_t* value() const; + + // Get attribute list + xml_attribute first_attribute() const; + xml_attribute last_attribute() const; + + // Get children list + xml_node first_child() const; + xml_node last_child() const; + + // Get next/previous sibling in the children list of the parent node + xml_node next_sibling() const; + xml_node previous_sibling() const; + + // Get parent node + xml_node parent() const; + + // Get root of DOM tree this node belongs to + xml_node root() const; + + // Get text object for the current node + xml_text text() const; + + // Get child, attribute or next/previous sibling with the specified name + xml_node child(const char_t* name) const; + xml_attribute attribute(const char_t* name) const; + xml_node next_sibling(const char_t* name) const; + xml_node previous_sibling(const char_t* name) const; + + // Get attribute, starting the search from a hint (and updating hint so that searching for a sequence of attributes is fast) + xml_attribute attribute(const char_t* name, xml_attribute& hint) const; + + // Get child value of current node; that is, value of the first child node of type PCDATA/CDATA + const char_t* child_value() const; + + // Get child value of child with specified name. Equivalent to child(name).child_value(). + const char_t* child_value(const char_t* name) const; + + // Set node name/value (returns false if node is empty, there is not enough memory, or node can not have name/value) + bool set_name(const char_t* rhs); + bool set_value(const char_t* rhs); + + // Add attribute with specified name. Returns added attribute, or empty attribute on errors. + xml_attribute append_attribute(const char_t* name); + xml_attribute prepend_attribute(const char_t* name); + xml_attribute insert_attribute_after(const char_t* name, const xml_attribute& attr); + xml_attribute insert_attribute_before(const char_t* name, const xml_attribute& attr); + + // Add a copy of the specified attribute. Returns added attribute, or empty attribute on errors. + xml_attribute append_copy(const xml_attribute& proto); + xml_attribute prepend_copy(const xml_attribute& proto); + xml_attribute insert_copy_after(const xml_attribute& proto, const xml_attribute& attr); + xml_attribute insert_copy_before(const xml_attribute& proto, const xml_attribute& attr); + + // Add child node with specified type. Returns added node, or empty node on errors. + xml_node append_child(xml_node_type type = node_element); + xml_node prepend_child(xml_node_type type = node_element); + xml_node insert_child_after(xml_node_type type, const xml_node& node); + xml_node insert_child_before(xml_node_type type, const xml_node& node); + + // Add child element with specified name. Returns added node, or empty node on errors. + xml_node append_child(const char_t* name); + xml_node prepend_child(const char_t* name); + xml_node insert_child_after(const char_t* name, const xml_node& node); + xml_node insert_child_before(const char_t* name, const xml_node& node); + + // Add a copy of the specified node as a child. Returns added node, or empty node on errors. + xml_node append_copy(const xml_node& proto); + xml_node prepend_copy(const xml_node& proto); + xml_node insert_copy_after(const xml_node& proto, const xml_node& node); + xml_node insert_copy_before(const xml_node& proto, const xml_node& node); + + // Move the specified node to become a child of this node. Returns moved node, or empty node on errors. + xml_node append_move(const xml_node& moved); + xml_node prepend_move(const xml_node& moved); + xml_node insert_move_after(const xml_node& moved, const xml_node& node); + xml_node insert_move_before(const xml_node& moved, const xml_node& node); + + // Remove specified attribute + bool remove_attribute(const xml_attribute& a); + bool remove_attribute(const char_t* name); + + // Remove specified child + bool remove_child(const xml_node& n); + bool remove_child(const char_t* name); + + // Parses buffer as an XML document fragment and appends all nodes as children of the current node. + // Copies/converts the buffer, so it may be deleted or changed after the function returns. + // Note: append_buffer allocates memory that has the lifetime of the owning document; removing the appended nodes does not immediately reclaim that memory. + xml_parse_result append_buffer(const void* contents, size_t size, unsigned int options = parse_default, xml_encoding encoding = encoding_auto); + + // Find attribute using predicate. Returns first attribute for which predicate returned true. + template xml_attribute find_attribute(Predicate pred) const + { + if (!_root) return xml_attribute(); + + for (xml_attribute attrib = first_attribute(); attrib; attrib = attrib.next_attribute()) + if (pred(attrib)) + return attrib; + + return xml_attribute(); + } + + // Find child node using predicate. Returns first child for which predicate returned true. + template xml_node find_child(Predicate pred) const + { + if (!_root) return xml_node(); + + for (xml_node node = first_child(); node; node = node.next_sibling()) + if (pred(node)) + return node; + + return xml_node(); + } + + // Find node from subtree using predicate. Returns first node from subtree (depth-first), for which predicate returned true. + template xml_node find_node(Predicate pred) const + { + if (!_root) return xml_node(); + + xml_node cur = first_child(); + + while (cur._root && cur._root != _root) + { + if (pred(cur)) return cur; + + if (cur.first_child()) cur = cur.first_child(); + else if (cur.next_sibling()) cur = cur.next_sibling(); + else + { + while (!cur.next_sibling() && cur._root != _root) cur = cur.parent(); + + if (cur._root != _root) cur = cur.next_sibling(); + } + } + + return xml_node(); + } + + // Find child node by attribute name/value + xml_node find_child_by_attribute(const char_t* name, const char_t* attr_name, const char_t* attr_value) const; + xml_node find_child_by_attribute(const char_t* attr_name, const char_t* attr_value) const; + + #ifndef PUGIXML_NO_STL + // Get the absolute node path from root as a text string. + string_t path(char_t delimiter = '/') const; + #endif + + // Search for a node by path consisting of node names and . or .. elements. + xml_node first_element_by_path(const char_t* path, char_t delimiter = '/') const; + + // Recursively traverse subtree with xml_tree_walker + bool traverse(xml_tree_walker& walker); + + #ifndef PUGIXML_NO_XPATH + // Select single node by evaluating XPath query. Returns first node from the resulting node set. + xpath_node select_node(const char_t* query, xpath_variable_set* variables = 0) const; + xpath_node select_node(const xpath_query& query) const; + + // Select node set by evaluating XPath query + xpath_node_set select_nodes(const char_t* query, xpath_variable_set* variables = 0) const; + xpath_node_set select_nodes(const xpath_query& query) const; + + // (deprecated: use select_node instead) Select single node by evaluating XPath query. + PUGIXML_DEPRECATED xpath_node select_single_node(const char_t* query, xpath_variable_set* variables = 0) const; + PUGIXML_DEPRECATED xpath_node select_single_node(const xpath_query& query) const; + + #endif + + // Print subtree using a writer object + void print(xml_writer& writer, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto, unsigned int depth = 0) const; + + #ifndef PUGIXML_NO_STL + // Print subtree to stream + void print(std::basic_ostream >& os, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto, unsigned int depth = 0) const; + void print(std::basic_ostream >& os, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, unsigned int depth = 0) const; + #endif + + // Child nodes iterators + typedef xml_node_iterator iterator; + + iterator begin() const; + iterator end() const; + + // Attribute iterators + typedef xml_attribute_iterator attribute_iterator; + + attribute_iterator attributes_begin() const; + attribute_iterator attributes_end() const; + + // Range-based for support + xml_object_range children() const; + xml_object_range children(const char_t* name) const; + xml_object_range attributes() const; + + // Get node offset in parsed file/string (in char_t units) for debugging purposes + ptrdiff_t offset_debug() const; + + // Get hash value (unique for handles to the same object) + size_t hash_value() const; + + // Get internal pointer + xml_node_struct* internal_object() const; + }; + +#ifdef __BORLANDC__ + // Borland C++ workaround + bool PUGIXML_FUNCTION operator&&(const xml_node& lhs, bool rhs); + bool PUGIXML_FUNCTION operator||(const xml_node& lhs, bool rhs); +#endif + + // A helper for working with text inside PCDATA nodes + class PUGIXML_CLASS xml_text + { + friend class xml_node; + + xml_node_struct* _root; + + typedef void (*unspecified_bool_type)(xml_text***); + + explicit xml_text(xml_node_struct* root); + + xml_node_struct* _data_new(); + xml_node_struct* _data() const; + + public: + // Default constructor. Constructs an empty object. + xml_text(); + + // Safe bool conversion operator + operator unspecified_bool_type() const; + + // Borland C++ workaround + bool operator!() const; + + // Check if text object is empty + bool empty() const; + + // Get text, or "" if object is empty + const char_t* get() const; + + // Get text, or the default value if object is empty + const char_t* as_string(const char_t* def = PUGIXML_TEXT("")) const; + + // Get text as a number, or the default value if conversion did not succeed or object is empty + int as_int(int def = 0) const; + unsigned int as_uint(unsigned int def = 0) const; + double as_double(double def = 0) const; + float as_float(float def = 0) const; + + #ifdef PUGIXML_HAS_LONG_LONG + long long as_llong(long long def = 0) const; + unsigned long long as_ullong(unsigned long long def = 0) const; + #endif + + // Get text as bool (returns true if first character is in '1tTyY' set), or the default value if object is empty + bool as_bool(bool def = false) const; + + // Set text (returns false if object is empty or there is not enough memory) + bool set(const char_t* rhs); + + // Set text with type conversion (numbers are converted to strings, boolean is converted to "true"/"false") + bool set(int rhs); + bool set(unsigned int rhs); + bool set(long rhs); + bool set(unsigned long rhs); + bool set(double rhs); + bool set(float rhs); + bool set(bool rhs); + + #ifdef PUGIXML_HAS_LONG_LONG + bool set(long long rhs); + bool set(unsigned long long rhs); + #endif + + // Set text (equivalent to set without error checking) + xml_text& operator=(const char_t* rhs); + xml_text& operator=(int rhs); + xml_text& operator=(unsigned int rhs); + xml_text& operator=(long rhs); + xml_text& operator=(unsigned long rhs); + xml_text& operator=(double rhs); + xml_text& operator=(float rhs); + xml_text& operator=(bool rhs); + + #ifdef PUGIXML_HAS_LONG_LONG + xml_text& operator=(long long rhs); + xml_text& operator=(unsigned long long rhs); + #endif + + // Get the data node (node_pcdata or node_cdata) for this object + xml_node data() const; + }; + +#ifdef __BORLANDC__ + // Borland C++ workaround + bool PUGIXML_FUNCTION operator&&(const xml_text& lhs, bool rhs); + bool PUGIXML_FUNCTION operator||(const xml_text& lhs, bool rhs); +#endif + + // Child node iterator (a bidirectional iterator over a collection of xml_node) + class PUGIXML_CLASS xml_node_iterator + { + friend class xml_node; + + private: + mutable xml_node _wrap; + xml_node _parent; + + xml_node_iterator(xml_node_struct* ref, xml_node_struct* parent); + + public: + // Iterator traits + typedef ptrdiff_t difference_type; + typedef xml_node value_type; + typedef xml_node* pointer; + typedef xml_node& reference; + + #ifndef PUGIXML_NO_STL + typedef std::bidirectional_iterator_tag iterator_category; + #endif + + // Default constructor + xml_node_iterator(); + + // Construct an iterator which points to the specified node + xml_node_iterator(const xml_node& node); + + // Iterator operators + bool operator==(const xml_node_iterator& rhs) const; + bool operator!=(const xml_node_iterator& rhs) const; + + xml_node& operator*() const; + xml_node* operator->() const; + + const xml_node_iterator& operator++(); + xml_node_iterator operator++(int); + + const xml_node_iterator& operator--(); + xml_node_iterator operator--(int); + }; + + // Attribute iterator (a bidirectional iterator over a collection of xml_attribute) + class PUGIXML_CLASS xml_attribute_iterator + { + friend class xml_node; + + private: + mutable xml_attribute _wrap; + xml_node _parent; + + xml_attribute_iterator(xml_attribute_struct* ref, xml_node_struct* parent); + + public: + // Iterator traits + typedef ptrdiff_t difference_type; + typedef xml_attribute value_type; + typedef xml_attribute* pointer; + typedef xml_attribute& reference; + + #ifndef PUGIXML_NO_STL + typedef std::bidirectional_iterator_tag iterator_category; + #endif + + // Default constructor + xml_attribute_iterator(); + + // Construct an iterator which points to the specified attribute + xml_attribute_iterator(const xml_attribute& attr, const xml_node& parent); + + // Iterator operators + bool operator==(const xml_attribute_iterator& rhs) const; + bool operator!=(const xml_attribute_iterator& rhs) const; + + xml_attribute& operator*() const; + xml_attribute* operator->() const; + + const xml_attribute_iterator& operator++(); + xml_attribute_iterator operator++(int); + + const xml_attribute_iterator& operator--(); + xml_attribute_iterator operator--(int); + }; + + // Named node range helper + class PUGIXML_CLASS xml_named_node_iterator + { + friend class xml_node; + + public: + // Iterator traits + typedef ptrdiff_t difference_type; + typedef xml_node value_type; + typedef xml_node* pointer; + typedef xml_node& reference; + + #ifndef PUGIXML_NO_STL + typedef std::bidirectional_iterator_tag iterator_category; + #endif + + // Default constructor + xml_named_node_iterator(); + + // Construct an iterator which points to the specified node + xml_named_node_iterator(const xml_node& node, const char_t* name); + + // Iterator operators + bool operator==(const xml_named_node_iterator& rhs) const; + bool operator!=(const xml_named_node_iterator& rhs) const; + + xml_node& operator*() const; + xml_node* operator->() const; + + const xml_named_node_iterator& operator++(); + xml_named_node_iterator operator++(int); + + const xml_named_node_iterator& operator--(); + xml_named_node_iterator operator--(int); + + private: + mutable xml_node _wrap; + xml_node _parent; + const char_t* _name; + + xml_named_node_iterator(xml_node_struct* ref, xml_node_struct* parent, const char_t* name); + }; + + // Abstract tree walker class (see xml_node::traverse) + class PUGIXML_CLASS xml_tree_walker + { + friend class xml_node; + + private: + int _depth; + + protected: + // Get current traversal depth + int depth() const; + + public: + xml_tree_walker(); + virtual ~xml_tree_walker(); + + // Callback that is called when traversal begins + virtual bool begin(xml_node& node); + + // Callback that is called for each node traversed + virtual bool for_each(xml_node& node) = 0; + + // Callback that is called when traversal ends + virtual bool end(xml_node& node); + }; + + // Parsing status, returned as part of xml_parse_result object + enum xml_parse_status + { + status_ok = 0, // No error + + status_file_not_found, // File was not found during load_file() + status_io_error, // Error reading from file/stream + status_out_of_memory, // Could not allocate memory + status_internal_error, // Internal error occurred + + status_unrecognized_tag, // Parser could not determine tag type + + status_bad_pi, // Parsing error occurred while parsing document declaration/processing instruction + status_bad_comment, // Parsing error occurred while parsing comment + status_bad_cdata, // Parsing error occurred while parsing CDATA section + status_bad_doctype, // Parsing error occurred while parsing document type declaration + status_bad_pcdata, // Parsing error occurred while parsing PCDATA section + status_bad_start_element, // Parsing error occurred while parsing start element tag + status_bad_attribute, // Parsing error occurred while parsing element attribute + status_bad_end_element, // Parsing error occurred while parsing end element tag + status_end_element_mismatch,// There was a mismatch of start-end tags (closing tag had incorrect name, some tag was not closed or there was an excessive closing tag) + + status_append_invalid_root, // Unable to append nodes since root type is not node_element or node_document (exclusive to xml_node::append_buffer) + + status_no_document_element // Parsing resulted in a document without element nodes + }; + + // Parsing result + struct PUGIXML_CLASS xml_parse_result + { + // Parsing status (see xml_parse_status) + xml_parse_status status; + + // Last parsed offset (in char_t units from start of input data) + ptrdiff_t offset; + + // Source document encoding + xml_encoding encoding; + + // Default constructor, initializes object to failed state + xml_parse_result(); + + // Cast to bool operator + operator bool() const; + + // Get error description + const char* description() const; + }; + + // Document class (DOM tree root) + class PUGIXML_CLASS xml_document: public xml_node + { + private: + char_t* _buffer; + + char _memory[192]; + + // Non-copyable semantics + xml_document(const xml_document&); + xml_document& operator=(const xml_document&); + + void _create(); + void _destroy(); + void _move(xml_document& rhs) PUGIXML_NOEXCEPT_IF_NOT_COMPACT; + + public: + // Default constructor, makes empty document + xml_document(); + + // Destructor, invalidates all node/attribute handles to this document + ~xml_document(); + + #ifdef PUGIXML_HAS_MOVE + // Move semantics support + xml_document(xml_document&& rhs) PUGIXML_NOEXCEPT_IF_NOT_COMPACT; + xml_document& operator=(xml_document&& rhs) PUGIXML_NOEXCEPT_IF_NOT_COMPACT; + #endif + + // Removes all nodes, leaving the empty document + void reset(); + + // Removes all nodes, then copies the entire contents of the specified document + void reset(const xml_document& proto); + + #ifndef PUGIXML_NO_STL + // Load document from stream. + xml_parse_result load(std::basic_istream >& stream, unsigned int options = parse_default, xml_encoding encoding = encoding_auto); + xml_parse_result load(std::basic_istream >& stream, unsigned int options = parse_default); + #endif + + // (deprecated: use load_string instead) Load document from zero-terminated string. No encoding conversions are applied. + PUGIXML_DEPRECATED xml_parse_result load(const char_t* contents, unsigned int options = parse_default); + + // Load document from zero-terminated string. No encoding conversions are applied. + xml_parse_result load_string(const char_t* contents, unsigned int options = parse_default); + + // Load document from file + xml_parse_result load_file(const char* path, unsigned int options = parse_default, xml_encoding encoding = encoding_auto); + xml_parse_result load_file(const wchar_t* path, unsigned int options = parse_default, xml_encoding encoding = encoding_auto); + + // Load document from buffer. Copies/converts the buffer, so it may be deleted or changed after the function returns. + xml_parse_result load_buffer(const void* contents, size_t size, unsigned int options = parse_default, xml_encoding encoding = encoding_auto); + + // Load document from buffer, using the buffer for in-place parsing (the buffer is modified and used for storage of document data). + // You should ensure that buffer data will persist throughout the document's lifetime, and free the buffer memory manually once document is destroyed. + xml_parse_result load_buffer_inplace(void* contents, size_t size, unsigned int options = parse_default, xml_encoding encoding = encoding_auto); + + // Load document from buffer, using the buffer for in-place parsing (the buffer is modified and used for storage of document data). + // You should allocate the buffer with pugixml allocation function; document will free the buffer when it is no longer needed (you can't use it anymore). + xml_parse_result load_buffer_inplace_own(void* contents, size_t size, unsigned int options = parse_default, xml_encoding encoding = encoding_auto); + + // Save XML document to writer (semantics is slightly different from xml_node::print, see documentation for details). + void save(xml_writer& writer, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto) const; + + #ifndef PUGIXML_NO_STL + // Save XML document to stream (semantics is slightly different from xml_node::print, see documentation for details). + void save(std::basic_ostream >& stream, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto) const; + void save(std::basic_ostream >& stream, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default) const; + #endif + + // Save XML to file + bool save_file(const char* path, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto) const; + bool save_file(const wchar_t* path, const char_t* indent = PUGIXML_TEXT("\t"), unsigned int flags = format_default, xml_encoding encoding = encoding_auto) const; + + // Get document element + xml_node document_element() const; + }; + +#ifndef PUGIXML_NO_XPATH + // XPath query return type + enum xpath_value_type + { + xpath_type_none, // Unknown type (query failed to compile) + xpath_type_node_set, // Node set (xpath_node_set) + xpath_type_number, // Number + xpath_type_string, // String + xpath_type_boolean // Boolean + }; + + // XPath parsing result + struct PUGIXML_CLASS xpath_parse_result + { + // Error message (0 if no error) + const char* error; + + // Last parsed offset (in char_t units from string start) + ptrdiff_t offset; + + // Default constructor, initializes object to failed state + xpath_parse_result(); + + // Cast to bool operator + operator bool() const; + + // Get error description + const char* description() const; + }; + + // A single XPath variable + class PUGIXML_CLASS xpath_variable + { + friend class xpath_variable_set; + + protected: + xpath_value_type _type; + xpath_variable* _next; + + xpath_variable(xpath_value_type type); + + // Non-copyable semantics + xpath_variable(const xpath_variable&); + xpath_variable& operator=(const xpath_variable&); + + public: + // Get variable name + const char_t* name() const; + + // Get variable type + xpath_value_type type() const; + + // Get variable value; no type conversion is performed, default value (false, NaN, empty string, empty node set) is returned on type mismatch error + bool get_boolean() const; + double get_number() const; + const char_t* get_string() const; + const xpath_node_set& get_node_set() const; + + // Set variable value; no type conversion is performed, false is returned on type mismatch error + bool set(bool value); + bool set(double value); + bool set(const char_t* value); + bool set(const xpath_node_set& value); + }; + + // A set of XPath variables + class PUGIXML_CLASS xpath_variable_set + { + private: + xpath_variable* _data[64]; + + void _assign(const xpath_variable_set& rhs); + void _swap(xpath_variable_set& rhs); + + xpath_variable* _find(const char_t* name) const; + + static bool _clone(xpath_variable* var, xpath_variable** out_result); + static void _destroy(xpath_variable* var); + + public: + // Default constructor/destructor + xpath_variable_set(); + ~xpath_variable_set(); + + // Copy constructor/assignment operator + xpath_variable_set(const xpath_variable_set& rhs); + xpath_variable_set& operator=(const xpath_variable_set& rhs); + + #ifdef PUGIXML_HAS_MOVE + // Move semantics support + xpath_variable_set(xpath_variable_set&& rhs) PUGIXML_NOEXCEPT; + xpath_variable_set& operator=(xpath_variable_set&& rhs) PUGIXML_NOEXCEPT; + #endif + + // Add a new variable or get the existing one, if the types match + xpath_variable* add(const char_t* name, xpath_value_type type); + + // Set value of an existing variable; no type conversion is performed, false is returned if there is no such variable or if types mismatch + bool set(const char_t* name, bool value); + bool set(const char_t* name, double value); + bool set(const char_t* name, const char_t* value); + bool set(const char_t* name, const xpath_node_set& value); + + // Get existing variable by name + xpath_variable* get(const char_t* name); + const xpath_variable* get(const char_t* name) const; + }; + + // A compiled XPath query object + class PUGIXML_CLASS xpath_query + { + private: + void* _impl; + xpath_parse_result _result; + + typedef void (*unspecified_bool_type)(xpath_query***); + + // Non-copyable semantics + xpath_query(const xpath_query&); + xpath_query& operator=(const xpath_query&); + + public: + // Construct a compiled object from XPath expression. + // If PUGIXML_NO_EXCEPTIONS is not defined, throws xpath_exception on compilation errors. + explicit xpath_query(const char_t* query, xpath_variable_set* variables = 0); + + // Constructor + xpath_query(); + + // Destructor + ~xpath_query(); + + #ifdef PUGIXML_HAS_MOVE + // Move semantics support + xpath_query(xpath_query&& rhs) PUGIXML_NOEXCEPT; + xpath_query& operator=(xpath_query&& rhs) PUGIXML_NOEXCEPT; + #endif + + // Get query expression return type + xpath_value_type return_type() const; + + // Evaluate expression as boolean value in the specified context; performs type conversion if necessary. + // If PUGIXML_NO_EXCEPTIONS is not defined, throws std::bad_alloc on out of memory errors. + bool evaluate_boolean(const xpath_node& n) const; + + // Evaluate expression as double value in the specified context; performs type conversion if necessary. + // If PUGIXML_NO_EXCEPTIONS is not defined, throws std::bad_alloc on out of memory errors. + double evaluate_number(const xpath_node& n) const; + + #ifndef PUGIXML_NO_STL + // Evaluate expression as string value in the specified context; performs type conversion if necessary. + // If PUGIXML_NO_EXCEPTIONS is not defined, throws std::bad_alloc on out of memory errors. + string_t evaluate_string(const xpath_node& n) const; + #endif + + // Evaluate expression as string value in the specified context; performs type conversion if necessary. + // At most capacity characters are written to the destination buffer, full result size is returned (includes terminating zero). + // If PUGIXML_NO_EXCEPTIONS is not defined, throws std::bad_alloc on out of memory errors. + // If PUGIXML_NO_EXCEPTIONS is defined, returns empty set instead. + size_t evaluate_string(char_t* buffer, size_t capacity, const xpath_node& n) const; + + // Evaluate expression as node set in the specified context. + // If PUGIXML_NO_EXCEPTIONS is not defined, throws xpath_exception on type mismatch and std::bad_alloc on out of memory errors. + // If PUGIXML_NO_EXCEPTIONS is defined, returns empty node set instead. + xpath_node_set evaluate_node_set(const xpath_node& n) const; + + // Evaluate expression as node set in the specified context. + // Return first node in document order, or empty node if node set is empty. + // If PUGIXML_NO_EXCEPTIONS is not defined, throws xpath_exception on type mismatch and std::bad_alloc on out of memory errors. + // If PUGIXML_NO_EXCEPTIONS is defined, returns empty node instead. + xpath_node evaluate_node(const xpath_node& n) const; + + // Get parsing result (used to get compilation errors in PUGIXML_NO_EXCEPTIONS mode) + const xpath_parse_result& result() const; + + // Safe bool conversion operator + operator unspecified_bool_type() const; + + // Borland C++ workaround + bool operator!() const; + }; + + #ifndef PUGIXML_NO_EXCEPTIONS + #if defined(_MSC_VER) + // C4275 can be ignored in Visual C++ if you are deriving + // from a type in the Standard C++ Library + #pragma warning(push) + #pragma warning(disable: 4275) + #endif + // XPath exception class + class PUGIXML_CLASS xpath_exception: public std::exception + { + private: + xpath_parse_result _result; + + public: + // Construct exception from parse result + explicit xpath_exception(const xpath_parse_result& result); + + // Get error message + virtual const char* what() const throw() PUGIXML_OVERRIDE; + + // Get parse result + const xpath_parse_result& result() const; + }; + #if defined(_MSC_VER) + #pragma warning(pop) + #endif + #endif + + // XPath node class (either xml_node or xml_attribute) + class PUGIXML_CLASS xpath_node + { + private: + xml_node _node; + xml_attribute _attribute; + + typedef void (*unspecified_bool_type)(xpath_node***); + + public: + // Default constructor; constructs empty XPath node + xpath_node(); + + // Construct XPath node from XML node/attribute + xpath_node(const xml_node& node); + xpath_node(const xml_attribute& attribute, const xml_node& parent); + + // Get node/attribute, if any + xml_node node() const; + xml_attribute attribute() const; + + // Get parent of contained node/attribute + xml_node parent() const; + + // Safe bool conversion operator + operator unspecified_bool_type() const; + + // Borland C++ workaround + bool operator!() const; + + // Comparison operators + bool operator==(const xpath_node& n) const; + bool operator!=(const xpath_node& n) const; + }; + +#ifdef __BORLANDC__ + // Borland C++ workaround + bool PUGIXML_FUNCTION operator&&(const xpath_node& lhs, bool rhs); + bool PUGIXML_FUNCTION operator||(const xpath_node& lhs, bool rhs); +#endif + + // A fixed-size collection of XPath nodes + class PUGIXML_CLASS xpath_node_set + { + public: + // Collection type + enum type_t + { + type_unsorted, // Not ordered + type_sorted, // Sorted by document order (ascending) + type_sorted_reverse // Sorted by document order (descending) + }; + + // Constant iterator type + typedef const xpath_node* const_iterator; + + // We define non-constant iterator to be the same as constant iterator so that various generic algorithms (i.e. boost foreach) work + typedef const xpath_node* iterator; + + // Default constructor. Constructs empty set. + xpath_node_set(); + + // Constructs a set from iterator range; data is not checked for duplicates and is not sorted according to provided type, so be careful + xpath_node_set(const_iterator begin, const_iterator end, type_t type = type_unsorted); + + // Destructor + ~xpath_node_set(); + + // Copy constructor/assignment operator + xpath_node_set(const xpath_node_set& ns); + xpath_node_set& operator=(const xpath_node_set& ns); + + #ifdef PUGIXML_HAS_MOVE + // Move semantics support + xpath_node_set(xpath_node_set&& rhs) PUGIXML_NOEXCEPT; + xpath_node_set& operator=(xpath_node_set&& rhs) PUGIXML_NOEXCEPT; + #endif + + // Get collection type + type_t type() const; + + // Get collection size + size_t size() const; + + // Indexing operator + const xpath_node& operator[](size_t index) const; + + // Collection iterators + const_iterator begin() const; + const_iterator end() const; + + // Sort the collection in ascending/descending order by document order + void sort(bool reverse = false); + + // Get first node in the collection by document order + xpath_node first() const; + + // Check if collection is empty + bool empty() const; + + private: + type_t _type; + + xpath_node _storage[1]; + + xpath_node* _begin; + xpath_node* _end; + + void _assign(const_iterator begin, const_iterator end, type_t type); + void _move(xpath_node_set& rhs) PUGIXML_NOEXCEPT; + }; +#endif + +#ifndef PUGIXML_NO_STL + // Convert wide string to UTF8 + std::basic_string, std::allocator > PUGIXML_FUNCTION as_utf8(const wchar_t* str); + std::basic_string, std::allocator > PUGIXML_FUNCTION as_utf8(const std::basic_string, std::allocator >& str); + + // Convert UTF8 to wide string + std::basic_string, std::allocator > PUGIXML_FUNCTION as_wide(const char* str); + std::basic_string, std::allocator > PUGIXML_FUNCTION as_wide(const std::basic_string, std::allocator >& str); +#endif + + // Memory allocation function interface; returns pointer to allocated memory or NULL on failure + typedef void* (*allocation_function)(size_t size); + + // Memory deallocation function interface + typedef void (*deallocation_function)(void* ptr); + + // Override default memory management functions. All subsequent allocations/deallocations will be performed via supplied functions. + void PUGIXML_FUNCTION set_memory_management_functions(allocation_function allocate, deallocation_function deallocate); + + // Get current memory management functions + allocation_function PUGIXML_FUNCTION get_memory_allocation_function(); + deallocation_function PUGIXML_FUNCTION get_memory_deallocation_function(); +} + +#if !defined(PUGIXML_NO_STL) && (defined(_MSC_VER) || defined(__ICC)) +namespace std +{ + // Workarounds for (non-standard) iterator category detection for older versions (MSVC7/IC8 and earlier) + std::bidirectional_iterator_tag PUGIXML_FUNCTION _Iter_cat(const pugi::xml_node_iterator&); + std::bidirectional_iterator_tag PUGIXML_FUNCTION _Iter_cat(const pugi::xml_attribute_iterator&); + std::bidirectional_iterator_tag PUGIXML_FUNCTION _Iter_cat(const pugi::xml_named_node_iterator&); +} +#endif + +#if !defined(PUGIXML_NO_STL) && defined(__SUNPRO_CC) +namespace std +{ + // Workarounds for (non-standard) iterator category detection + std::bidirectional_iterator_tag PUGIXML_FUNCTION __iterator_category(const pugi::xml_node_iterator&); + std::bidirectional_iterator_tag PUGIXML_FUNCTION __iterator_category(const pugi::xml_attribute_iterator&); + std::bidirectional_iterator_tag PUGIXML_FUNCTION __iterator_category(const pugi::xml_named_node_iterator&); +} +#endif + +#endif + +// Make sure implementation is included in header-only mode +// Use macro expansion in #include to work around QMake (QTBUG-11923) +#if defined(PUGIXML_HEADER_ONLY) && !defined(PUGIXML_SOURCE) +# define PUGIXML_SOURCE "pugixml.cpp" +# include PUGIXML_SOURCE +#endif + +/** + * Copyright (c) 2006-2019 Arseny Kapoulkine + * + * Permission is hereby granted, free of charge, to any person + * obtaining a copy of this software and associated documentation + * files (the "Software"), to deal in the Software without + * restriction, including without limitation the rights to use, + * copy, modify, merge, publish, distribute, sublicense, and/or sell + * copies of the Software, and to permit persons to whom the + * Software is furnished to do so, subject to the following + * conditions: + * + * The above copyright notice and this permission notice shall be + * included in all copies or substantial portions of the Software. + * + * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, + * EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES + * OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND + * NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT + * HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, + * WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING + * FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR + * OTHER DEALINGS IN THE SOFTWARE. + */ diff --git a/source/utils/utils.cpp b/source/utils/utils.cpp new file mode 100644 index 0000000..1fb2150 --- /dev/null +++ b/source/utils/utils.cpp @@ -0,0 +1,64 @@ +#include +#include +#include +#include +#include "logger.h" +#include "utils.h" +#include "TinySHA1.hpp" + +// https://gist.github.com/ccbrown/9722406 +void Utils::dumpHex(const void *data, size_t size) { + char ascii[17]; + size_t i, j; + ascii[16] = '\0'; + DEBUG_FUNCTION_LINE("0x%08X (0x0000): ", data); + for (i = 0; i < size; ++i) { + WHBLogWritef("%02X ", ((unsigned char *) data)[i]); + if (((unsigned char *) data)[i] >= ' ' && ((unsigned char *) data)[i] <= '~') { + ascii[i % 16] = ((unsigned char *) data)[i]; + } else { + ascii[i % 16] = '.'; + } + if ((i + 1) % 8 == 0 || i + 1 == size) { + WHBLogWritef(" "); + if ((i + 1) % 16 == 0) { + WHBLogPrintf("| %s ", ascii); + if (i + 1 < size) { + DEBUG_FUNCTION_LINE("0x%08X (0x%04X); ", data + i + 1, i + 1); + } + } else if (i + 1 == size) { + ascii[(i + 1) % 16] = '\0'; + if ((i + 1) % 16 <= 8) { + WHBLogWritef(" "); + } + for (j = (i + 1) % 16; j < 16; ++j) { + WHBLogWritef(" "); + } + WHBLogPrintf("| %s ", ascii); + } + } + } +} + + +std::string Utils::calculateSHA1(const char *buffer, size_t size) { + sha1::SHA1 s; + s.processBytes(buffer, size); + uint32_t digest[5]; + s.getDigest(digest); + char tmp[48]; + snprintf(tmp, 45, "%08X%08X%08X%08X%08X", digest[0], digest[1], digest[2], digest[3], digest[4]); + return tmp; +} + +std::string Utils::hashFile(const std::string &path) { + uint8_t *data = NULL; + uint32_t size = 0; + FSUtils::LoadFileToMem(path.c_str(), &data, &size); + if (data == NULL) { + return calculateSHA1(NULL, 0); + } + std::string result = calculateSHA1(reinterpret_cast(data), size); + free(data); + return result; +} diff --git a/source/utils/utils.h b/source/utils/utils.h new file mode 100644 index 0000000..d4223be --- /dev/null +++ b/source/utils/utils.h @@ -0,0 +1,45 @@ +#pragma once + +#include +#include + +#ifdef __cplusplus +extern "C" { +#endif + +#define LIMIT(x, min, max) \ + ({ \ + typeof( x ) _x = x; \ + typeof( min ) _min = min; \ + typeof( max ) _max = max; \ + ( ( ( _x ) < ( _min ) ) ? ( _min ) : ( ( _x ) > ( _max ) ) ? ( _max) : ( _x ) ); \ +}) + +#define DegToRad(a) ( (a) * 0.01745329252f ) +#define RadToDeg(a) ( (a) * 57.29577951f ) + +#define ALIGN4(x) (((x) + 3) & ~3) +#define ALIGN32(x) (((x) + 31) & ~31) + +// those work only in powers of 2 +#define ROUNDDOWN(val, align) ((val) & ~(align-1)) +#define ROUNDUP(val, align) ROUNDDOWN(((val) + (align-1)), align) + + +#define le16(i) ((((uint16_t) ((i) & 0xFF)) << 8) | ((uint16_t) (((i) & 0xFF00) >> 8))) +#define le32(i) ((((uint32_t)le16((i) & 0xFFFF)) << 16) | ((uint32_t)le16(((i) & 0xFFFF0000) >> 16))) +#define le64(i) ((((uint64_t)le32((i) & 0xFFFFFFFFLL)) << 32) | ((uint64_t)le32(((i) & 0xFFFFFFFF00000000LL) >> 32))) + +#ifdef __cplusplus +} +#endif + + +class Utils { +public: + static void dumpHex(const void *data, size_t size); + + static std::string calculateSHA1(const char *buffer, size_t size); + + static std::string hashFile(const std::string &path); +}; \ No newline at end of file