Neue Bridge für Dubesors LLM Benchmark Blog

2025-07-27 14:48:18 +02:00
parent cd04d1fd0c
commit 6d36fb181b
1 changed files with 75 additions and 0 deletions
--- a/DubesorBridge.php
+++ b/DubesorBridge.php
@@ -0,0 +1,75 @@
+<?php
+class DubesorBridge extends BridgeAbstract
+{
+
+    const MAINTAINER = 'Brawl, Claude';
+    const NAME = 'Dubesor LLM Benchmark First Impressions';
+    const URI = 'https://dubesor.de/first-impressions';
+    const CACHE_TIMEOUT = 43200; // 12h
+    const DESCRIPTION = 'First impressions blog from Dubesor';
+
+    public function getIcon()
+    {
+        return 'https://dubesor.de/favicon.ico';
+    }
+
+    public function collectData()
+    {
+        $pageUrl = $this->getURI();
+        $html = getSimpleHTMLDOM($pageUrl)
+            or returnServerError('Could not request webpage: ' . $pageUrl);
+
+        foreach ($html->find('.entry') as $element) {
+            if (count($this->items) >= 20) {
+                break;
+            }
+
+            $modelTagsContainer = $element->find('.model-tags-container', 0);
+            $dateElement = $element->find('.entry-date', 0);
+            $contentElement = $element->find('.entry-content', 0);
+
+            if (!$modelTagsContainer || !$dateElement || !$contentElement) {
+                continue;
+            }
+
+            $modelTags = $modelTagsContainer->find('.model-tag');
+            $modelNames = array();
+            foreach ($modelTags as $tag) {
+                $modelNames[] = trim(strip_tags($tag->innertext));
+            }
+
+            if (empty($modelNames)) {
+                continue;
+            }
+
+            $dateStr = trim(strip_tags($dateElement->innertext));
+            $content = $contentElement->innertext;
+
+            $timestamp = strtotime($dateStr);
+            $modelList = implode(', ', $modelNames);
+            $title = 'First Impression: ' . $modelList;
+
+            $anchorLink = $modelTagsContainer->find('.anchor-link', 0);
+            $anchorId = '';
+            if ($anchorLink) {
+                $href = $anchorLink->href;
+                $anchorId = str_replace('#', '', $href);
+            } else {
+                $anchorId = strtolower(str_replace(' ', '-', $modelNames[0]));
+                $anchorId = preg_replace('/[^a-z0-9\-]/', '', $anchorId);
+            }
+
+            $uri = $this->getURI() . '#' . $anchorId;
+
+            $item = array();
+            $item['uid'] = $uri;
+            $item['uri'] = $uri;
+            $item['title'] = $title;
+            $item['content'] = $content;
+            $item['timestamp'] = $timestamp;
+            $item['author'] = 'Dubesor';
+
+            $this->items[] = $item;
+        }
+    }
+}