feat(autotag) improve selection of relevant tags

auto_tag_plugin/auto_tag_plugin.php

@@ -70,7 +70,7 @@ function auto_tag_plugin_init(ConfigManager $conf)
         'gratuit,free' => 'free',
         'gitops,gitlab,github actions,devops,SRE,ci/cd,platform-engineering,ci pipeline,application deployment,dagger,renovatebot,dependabot,continuous integration,site reliability eng' => 'devops,automation',
         'git,gitlab,jujutsu,pijul,mercurial,svn,version control,contrôle de version' => 'version-control',
-        'gpt,chatgpt,llm,llms,artificial intelligence,intelligence artificielle,IA' => 'ai',
+        'gpt,chatgpt,llm,llms,artificial intelligence,intelligence artificielle,IA,l\'ia' => 'ai',
         'hacking,piratage' => 'hacking',
         'health,healthy,nutrition,food,alimentation,nourriture,santé' => 'health-and-wellness,food',
         'humans,humains' => 'humans',
@@ -112,7 +112,7 @@ function auto_tag_plugin_init(ConfigManager $conf)
         'scripting,jq,curl,wget,script,bash' => 'scripting',
         'search engine,moteur de recherche' => 'search-engines',
         'security,permission,sécurité,secure,privacy,private,degoogl,gdpr,data protection,online tracking,user profiling,anonymo,anonymi,surveillance,malware,spyware,decentrali,secrets,privacy matters,vpn,passkey,protéger' => 'privacy-and-security',
-        'simplicity,minimal,less,declutter,stopped using,simple,simplification,simplifier,reduction in,fewer,no longer needed,minimalisme,simplicité,réduction' => 'minimalism',
+        'simplicity,minimal,declutter,stopped using,simple,simplification,simplifier,reduction in,no longer needed,minimalisme,simplicité,réduction' => 'minimalism',
         'small web,indie web,indieweb,petit web' => 'small-web',
         'smartphone,android,mobile,phone,téléphone,portables' => 'mobile',
         'snippet,extrait de code' => 'code',
@@ -147,11 +147,11 @@ function fetch_page_content($url)
         return '';
     }
 
+    // Remove unnecessary tags and strip content to plain text
     $htmlContent = preg_replace('/<(script|style|head|noscript)[^>]*>.*?<\/\1>/is', '', $htmlContent);
     return strip_tags($htmlContent);
 }
 
-
 function calculate_tags(array $keywordsToTags, array $searchContents): array
 {
     $contextWeights = [
@@ -175,10 +175,7 @@ function calculate_tags(array $keywordsToTags, array $searchContents): array
                     $tagList = explode(',', $tags);
                     foreach ($tagList as $tag) {
                         $tag = trim($tag);
-                        if (!isset($tagScores[$tag])) {
+                        $tagScores[$tag] = ($tagScores[$tag] ?? 0) + $contextWeights[$context];
-                            $tagScores[$tag] = 0;
-                        }
-                        $tagScores[$tag] += $contextWeights[$context];
                     }
                 }
             }
@@ -188,13 +185,21 @@ function calculate_tags(array $keywordsToTags, array $searchContents): array
     return $tagScores;
 }
 
-function filter_and_limit_tags(array $tagScores, int $minScore = 2, int $maxTags = 3): array
+function filter_and_limit_tags(array $tagScores, int $minScore = 2, int $maxTags = 4): array
 {
     // Filter tags with a score greater than or equal to the minimum required
     $filteredTags = array_filter($tagScores, function ($score) use ($minScore) {
         return $score >= $minScore;
     });
 
+    // Determine the maximum score
+    $maxScore = max($filteredTags);
+
+    // Remove tags with a score less than half of the maximum score
+    $filteredTags = array_filter($filteredTags, function ($score) use ($maxScore) {
+        return $score > ($maxScore / 2);
+    });
+
     // Sort tags by descending score
     arsort($filteredTags);
 
@@ -202,6 +207,17 @@ function filter_and_limit_tags(array $tagScores, int $minScore = 2, int $maxTags
     return array_slice(array_keys($filteredTags), 0, $maxTags);
 }
 
+function append_tag_stats(array $data, array $tagScores): array
+{
+    $stats = [];
+    foreach ($tagScores as $tag => $score) {
+        $stats[] = "$tag: score=$score";
+    }
+
+    $data['link']['description'] .= "\n\nTag Stats:\n" . implode("\n", $stats);
+    return $data;
+}
+
 function apply_auto_tags(array $data, ConfigManager $conf): array
 {
     if (empty($data['link']['url'])) {
@@ -222,6 +238,9 @@ function apply_auto_tags(array $data, ConfigManager $conf): array
     // Calculate scores for each tag
     $tagScores = calculate_tags($keywordsToTags, $searchContents);
 
+    // Append tag stats to the description
+    $data = append_tag_stats($data, $tagScores);
+
     // Filter and limit tags
     $tagsToAdd = filter_and_limit_tags($tagScores);
 
@@ -237,10 +256,7 @@ function apply_auto_tags(array $data, ConfigManager $conf): array
 
 function hook_auto_tag_plugin_render_editlink(array $data, ConfigManager $conf): array
 {
-    if (!$data['link_is_new']) {
+    // Always apply auto-tagging, even for existing links
-      // return $data; // Only auto-tag new links
-    }
-
     return apply_auto_tags($data, $conf);
 }
 
@@ -248,3 +264,4 @@ function hook_auto_tag_plugin_save_link(array $data, ConfigManager $conf): array
 {
     return apply_auto_tags($data, $conf);
 }
+