fix retrieve final technical questions

src/Knowledge/Retrieval/NdjsonChunkLookup.php

@@ -10,8 +10,7 @@ final readonly class NdjsonChunkLookup
 {
     public function __construct(
         private ChunkManager $chunkManager
-    )
-    {
+    ) {
     }
 
     /**
@@ -64,12 +63,12 @@ final readonly class NdjsonChunkLookup
     }
 
     /**
-     * Resolves the best exact document title match from the user prompt.
+     * Resolves the best document-title match from the user prompt.
      *
      * Matching rules:
-     * - the normalized prompt must contain the full normalized document title
-     * - titles containing digits are preferred, e.g. "Testomat 808"
-     * - longer exact titles win over shorter generic titles
+     * - exact contiguous title matches still win
+     * - if no exact match exists, all significant title tokens may match in any order
+     * - token fallback is intentionally conservative and handles light German suffixes
      *
      * @return array{
      *     document_id:string,
@@ -135,6 +134,27 @@ final readonly class NdjsonChunkLookup
             }
         }
 
+        if ($best === null) {
+            foreach ($documents as $document) {
+                $normalizedTitle = $document['normalized_title'];
+
+                if (!$this->isConfidentTitleTokenMatch($normalizedPrompt, $normalizedTitle)) {
+                    continue;
+                }
+
+                $score = 500 + mb_strlen($normalizedTitle, 'UTF-8');
+
+                if (preg_match('/\d/u', $normalizedTitle) === 1) {
+                    $score += 1000;
+                }
+
+                if ($best === null || $score > $bestScore) {
+                    $best = $document;
+                    $bestScore = $score;
+                }
+            }
+        }
+
         if ($best === null) {
             return null;
         }
@@ -192,6 +212,115 @@ final readonly class NdjsonChunkLookup
         return count($significantTokens) >= 2 || preg_match('/\d/u', $normalizedTitle) === 1;
     }
 
+    /**
+     * Fallback for factual prompts that mention a document title by its terms,
+     * but not as one contiguous phrase.
+     */
+    private function isConfidentTitleTokenMatch(string $normalizedPrompt, string $normalizedTitle): bool
+    {
+        if ($normalizedPrompt === '' || $normalizedTitle === '') {
+            return false;
+        }
+
+        $titleTokens = $this->significantTitleTokens($normalizedTitle);
+
+        if (count($titleTokens) < 3 && preg_match('/\d/u', $normalizedTitle) !== 1) {
+            return false;
+        }
+
+        $promptTokenVariants = $this->tokenVariantLookup($normalizedPrompt);
+
+        foreach ($titleTokens as $titleToken) {
+            $matched = false;
+
+            foreach ($this->tokenVariants($titleToken) as $variant) {
+                if (isset($promptTokenVariants[$variant])) {
+                    $matched = true;
+                    break;
+                }
+            }
+
+            if (!$matched) {
+                return false;
+            }
+        }
+
+        return true;
+    }
+
+    /**
+     * @return string[]
+     */
+    private function significantTitleTokens(string $normalizedTitle): array
+    {
+        $tokens = preg_split('/\s+/u', $normalizedTitle, -1, PREG_SPLIT_NO_EMPTY) ?: [];
+        $out = [];
+
+        foreach ($tokens as $token) {
+            $token = trim($token);
+
+            if ($token === '') {
+                continue;
+            }
+
+            if (mb_strlen($token, 'UTF-8') < 3 && preg_match('/\d/u', $token) !== 1) {
+                continue;
+            }
+
+            $out[] = $token;
+        }
+
+        return array_values(array_unique($out));
+    }
+
+    /**
+     * @return array<string,bool>
+     */
+    private function tokenVariantLookup(string $normalizedText): array
+    {
+        $tokens = preg_split('/\s+/u', $normalizedText, -1, PREG_SPLIT_NO_EMPTY) ?: [];
+        $lookup = [];
+
+        foreach ($tokens as $token) {
+            foreach ($this->tokenVariants($token) as $variant) {
+                $lookup[$variant] = true;
+            }
+        }
+
+        return $lookup;
+    }
+
+    /**
+     * @return string[]
+     */
+    private function tokenVariants(string $token): array
+    {
+        $token = trim($token);
+
+        if ($token === '') {
+            return [];
+        }
+
+        $variants = [$token];
+        $length = mb_strlen($token, 'UTF-8');
+
+        if ($length >= 5) {
+            foreach (['innen', 'enen', 'ern', 'en', 'er', 'es', 'e', 's', 'n'] as $suffix) {
+                if (!str_ends_with($token, $suffix)) {
+                    continue;
+                }
+
+                $stem = mb_substr($token, 0, $length - mb_strlen($suffix, 'UTF-8'), 'UTF-8');
+
+                if (mb_strlen($stem, 'UTF-8') >= 3) {
+                    $variants[] = $stem;
+                }
+            }
+        }
+
+        return array_values(array_unique($variants));
+    }
+
     private function normalizeText(string $value): string
     {
         $value = mb_strtolower(trim($value), 'UTF-8');
@@ -201,4 +330,4 @@ final readonly class NdjsonChunkLookup
 
         return trim($value);
     }
-}
+}