From c82457305db6f6ff95a18e3e4cad898171a2da4b Mon Sep 17 00:00:00 2001
From: SansGuidon <sansguidon@noreply.gitea.zoemp.be>
Date: Fri, 2 May 2025 14:55:09 +0000
Subject: [PATCH] fix(shaarli/autotag) fallback to dumb keyword (no AI) tagging
 when AI API key is not set

---
 .../auto_tag_plugin_with_ai.php               | 91 ++++++++++++++++++-
 1 file changed, 88 insertions(+), 3 deletions(-)

diff --git a/auto_tag_plugin_with_ai/auto_tag_plugin_with_ai.php b/auto_tag_plugin_with_ai/auto_tag_plugin_with_ai.php
index 6a10876..5509444 100644
--- a/auto_tag_plugin_with_ai/auto_tag_plugin_with_ai.php
+++ b/auto_tag_plugin_with_ai/auto_tag_plugin_with_ai.php
@@ -91,7 +91,7 @@ function auto_tag_plugin_init(ConfigManager $conf)
         'music,spotify,radios,webradios,soundtrack,bande originale,musique' => 'music',
         'newsletter,news' => 'newsletter,news',
         'nostalgia,nostalgie,things used to be better,internet archive' => 'nostalgia',
-        'obsidian,note taking,note-taking,takings notes,note-geek,capturing knowledge,knowledge management,prise de notes,gestion de connaissances,knowledge transfer,transferring knowledge,your notes,my notes' => 'knowledge-management,note-taking',
+        'obsidian,note taking,note-taking,taking notes,note-geek,capturing knowledge,knowledge management,prise de notes,gestion de connaissances,knowledge transfer,transferring knowledge,your notes,my notes' => 'knowledge-management,note-taking',
         'ocr' => 'ocr',
         'open-source,open source,code source libre' => 'open-source,free',
         'optimize,optimization,speed up,an efficient,optimiser,optimisation' => 'optimization',
@@ -245,12 +245,36 @@ function apply_auto_tags(array $data, ConfigManager $conf): array
         return $data;
     }
 
+    // Check if Mistral API key is set to default value
+    if ($conf->get('plugins.MISTRAL_API_KEY') === 'YOUR_MISTRAL_API_KEY') {
+        // Fallback to non-AI mode
+        $keywordsToTags = $conf->get('plugins.AUTO_TAG_KEYWORDS', []);
+        $searchContents = [
+            'title' => $title,
+            'url' => $url,
+            'description' => $desc,
+            'existing' => implode(' ', explode(' ', $data['link']['tags'] ?? '')),
+            'content' => $body
+        ];
+
+        // Calculate scores for each tag
+        $tagScores = calculate_tags($keywordsToTags, $searchContents);
+
+        // Filter and limit tags
+        $tagsToAdd = filter_and_limit_tags($tagScores);
+
+        // Merge, remove duplicates, and clean up
+        $data['link']['tags'] = trim(implode(' ', array_unique(array_merge(explode(' ', $data['link']['tags'] ?? ''), $tagsToAdd, ['auto-tagged']))));
+
+        return $data;
+    }
+
     $text  = "$title\n$url\n\n$body";
 
-    $prompt  = "TLDr de l'article, style gilfoyle/critique sans blabla et sans markdown/formating, max une ou deux phrases... ";
+    $prompt  = "TLDr de l'article, style Gilfoyle sans blabla et sans markdown/formating, max une ou deux phrases... ";
     $prompt .= "EN FRANÇAIS et si possible avec une mini conclusion crue/honnête, pas de language prétentieux ici, soyons bruts... ";
     $prompt .= "et drôles quand c le cas, sinon francs, éducatifs/informatifs, optimistes mais pas dupes. ";
-    $prompt .= "Si possible en français, pas franglais, et si possible pas de \"cela\" mais des \"ça\", pas de zut mais des \"merde\" etc, sans être vulgaire ni raciste ).\n\n";
+    $prompt .= "Si possible en français, pas franglais, et si possible pas de 'cela' mais des 'ça', pas de zut mais des 'merde' etc.\n\n";
     $prompt .= $text . "\n\n";
     $prompt .= "Ensuite, parmi ces tags (" . implode(', ', $conf->get('plugins.AUTO_TAG_WHITELIST')) . "), ";
     $prompt .= "propose jusqu'à max " . $conf->get('plugins.AUTO_TAG_MAX_TAGS') . " tags pertinents (si possible moins) ";
@@ -283,6 +307,67 @@ function apply_auto_tags(array $data, ConfigManager $conf): array
     return $data;
 }
 
+function calculate_tags(array $keywordsToTags, array $searchContents): array
+{
+    $contextWeights = [
+        'title' => 3,
+        'url' => 3,
+        'description' => 3,
+        'existing' => 3,
+        'content' => 1
+    ];
+
+    $tagScores = [];
+
+    foreach ($keywordsToTags as $keywords => $tags) {
+        $keywordList = explode(',', $keywords);
+        foreach ($keywordList as $keyword) {
+            $keyword = preg_quote(trim($keyword), '/');
+            $regex = '/\b' . $keyword . '\b/i';
+
+            foreach ($searchContents as $context => $content) {
+                if (preg_match($regex, $content)) {
+                    $tagList = explode(',', $tags);
+                    foreach ($tagList as $tag) {
+                        $tag = trim($tag);
+                        $tagScores[$tag] = ($tagScores[$tag] ?? 0) + $contextWeights[$context];
+                    }
+                }
+            }
+        }
+    }
+
+    return $tagScores;
+}
+
+function filter_and_limit_tags(array $tagScores, int $minScore = 2, int $maxTags = 10): 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;
+    });
+
+    // Check if $filteredTags is not empty to avoid max() error
+    if (!empty($filteredTags)) {
+        // 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);
+        });
+    } else {
+        // If no tags pass the filter, return an empty array
+        $filteredTags = [];
+    }
+
+    // Sort tags by descending score
+    arsort($filteredTags);
+
+    // Limit the number of tags to the maximum allowed
+    return array_slice(array_keys($filteredTags), 0, $maxTags);
+}
+
 function hook_auto_tag_plugin_render_editlink(array $data, ConfigManager $conf): array
 {
     return apply_auto_tags($data, $conf);