optimize system and cleanup

src/Intent/CatalogIntentLite.php

@@ -11,9 +11,7 @@ use App\Tag\TagTypes;
 /**
  * CatalogIntentLite
  *
- * Reiner Entity-Detector.
- *
- * Verantwortlich nur für:
+ * Verantwortlich ausschließlich für:
  * - Vector-Tag-Erkennung
  * - Score-Gate
  * - Ambiguity-Check
@@ -26,27 +24,14 @@ use App\Tag\TagTypes;
  */
 final class CatalogIntentLite
 {
-    /**
-     * Minimaler Similarity-Score.
-     * Verhindert Rauschen.
-     */
     private const MIN_SCORE = 0.72;
-
-    /**
-     * Differenz zwischen Top1 und Top2,
-     * damit kein unsicherer Treffer akzeptiert wird.
-     */
     private const AMBIGUITY_DELTA = 0.03;
 
     public function __construct(
         private readonly TagVectorSearchClient $tagVectorClient,
-        private readonly QueryCleaner   $queryCleaner,
+        private readonly QueryCleaner $queryCleaner,
     ) {}
 
-    /**
-     * Gibt das canonical Label der erkannten catalog_entity zurück
-     * oder null, wenn kein sauberer Treffer.
-     */
     public function detect(string $prompt): ?string
     {
         $prompt = trim($prompt);
@@ -54,10 +39,82 @@ final class CatalogIntentLite
             return null;
         }
 
-        $promptTag = $this->queryCleaner->clean($prompt);
+        $clean = $this->queryCleaner->clean($prompt);
+        if ($clean === '') {
+            $clean = $prompt;
+        }
 
-        // 1) Tag-Vector-Suche
-        $hits = $this->tagVectorClient->search($promptTag, 3);
+        // ----------------------------------------------------
+        // 1️⃣ Primär: Vollquery testen
+        // ----------------------------------------------------
+
+        $label = $this->detectFromQuery($clean);
+        if ($label !== null) {
+            return $label;
+        }
+
+        // ----------------------------------------------------
+        // 2️⃣ Fallback: Tokenweise testen
+        //    (wichtig für "geräteliste testomat")
+        // ----------------------------------------------------
+
+        $tokens = $this->tokenize($clean);
+
+        $bestLabel = null;
+        $bestScore = 0.0;
+
+        foreach ($tokens as $token) {
+
+            // sehr kurze Tokens ignorieren (Noise)
+            if (mb_strlen($token) < 3) {
+                continue;
+            }
+
+            $hits = $this->tagVectorClient->search($token, 3);
+
+            if ($hits === []) {
+                continue;
+            }
+
+            $top = $hits[0] ?? null;
+            if (!is_array($top)) {
+                continue;
+            }
+
+            $score = (float)($top['score'] ?? 0.0);
+
+            if ($score < self::MIN_SCORE) {
+                continue;
+            }
+
+            // Ambiguity-Check
+            if (isset($hits[1])) {
+                $secondScore = (float)($hits[1]['score'] ?? 0.0);
+                if (abs($score - $secondScore) < self::AMBIGUITY_DELTA) {
+                    continue;
+                }
+            }
+
+            if (($top['tag_type'] ?? null) !== TagTypes::CATALOG_ENTITY) {
+                continue;
+            }
+
+            if ($score > $bestScore) {
+                $bestScore = $score;
+                $bestLabel = trim((string)($top['label'] ?? ''));
+            }
+        }
+
+        if ($bestLabel === null || $bestLabel === '') {
+            return null;
+        }
+
+        return mb_strtolower($bestLabel);
+    }
+
+    private function detectFromQuery(string $query): ?string
+    {
+        $hits = $this->tagVectorClient->search($query, 3);
 
         if ($hits === []) {
             return null;
@@ -66,26 +123,21 @@ final class CatalogIntentLite
         $best = $hits[0];
         $bestScore = (float)($best['score'] ?? 0.0);
 
-        // 2) Score-Tags
         if ($bestScore < self::MIN_SCORE) {
             return null;
         }
 
-        // 3) Ambiguity-Check
         if (isset($hits[1])) {
             $secondScore = (float)($hits[1]['score'] ?? 0.0);
-
             if (abs($bestScore - $secondScore) < self::AMBIGUITY_DELTA) {
                 return null;
             }
         }
 
-        // 4) Nur catalog_entity zulassen
         if (($best['tag_type'] ?? null) !== TagTypes::CATALOG_ENTITY) {
             return null;
         }
 
-        // 5) Canonical Label
         $label = trim((string)($best['label'] ?? ''));
 
         if ($label === '') {
@@ -94,4 +146,31 @@ final class CatalogIntentLite
 
         return mb_strtolower($label);
     }
+
+    private function tokenize(string $text): array
+    {
+        $parts = preg_split('/\s+/u', trim($text));
+        if (!$parts) {
+            return [];
+        }
+
+        $seen = [];
+        $out = [];
+
+        foreach ($parts as $p) {
+            $p = trim($p);
+            if ($p === '') {
+                continue;
+            }
+
+            if (isset($seen[$p])) {
+                continue;
+            }
+
+            $seen[$p] = true;
+            $out[] = $p;
+        }
+
+        return $out;
+    }
 }