Update tools/quran_search.py

tools/quran_search.py

@@ -15,10 +15,13 @@ class QuranSearchEngine:
         self.all_verses = []  # List of {'surah_id': int, 'verse_num': int, 'text': str}
         self.verse_embeddings = None
         self.model = None
+        print("Starting QuranSearchEngine initialization...")  # Debug
         self._load_full_quran()
         print(f"Surahs loaded: {len(self.surahs) if self.surahs else 0}")  # Debug
         self._load_all_verses_and_embeddings()
         print(f"Verses loaded: {len(self.all_verses)}")  # Debug
+        if not self.model:
+            self.logger.error("Model initialization failed, using fallback behavior")
         
     def _load_full_quran(self):
         max_retries = 3
@@ -77,111 +80,3 @@ class QuranSearchEngine:
                     self.all_verses.append({
                         'surah_id': surah_id,
                         'verse_num': verse_num,
-                        'text': text
-                    })
-            else:  # Fallback structure: verses as list of dict
-                verses = surah.get('verses', [])
-                for verse in verses:
-                    verse_num = verse.get('id', 1)
-                    text = verse.get('text', '')
-                    self.all_verses.append({
-                        'surah_id': surah_id,
-                        'verse_num': verse_num,
-                        'text': text
-                    })
-        
-        try:
-            self.model = SentenceTransformer(MODEL_NAME)
-            verse_texts = ["passage: " + v['text'] for v in self.all_verses]  # Add prefix for e5 model
-            self.verse_embeddings = []
-            for i in range(0, len(verse_texts), CHUNK_SIZE):
-                chunk = verse_texts[i:i + CHUNK_SIZE]
-                embeddings = self.model.encode(chunk, convert_to_tensor=False)
-                self.verse_embeddings.append(embeddings)
-            self.verse_embeddings = np.vstack(self.verse_embeddings)
-        except Exception as e:
-            self.logger.error(f"Failed to compute embeddings: {e}")
-            self.verse_embeddings = None
-    
-    def get_surahs(self):
-        if self.surahs:
-            return [
-                (s.get('surahNameArabicLong', s.get('name', '')), s['id'])
-                for s in self.surahs
-            ]
-        return self._load_fallback_surahs()
-    
-    def get_surah_text(self, surah_id):
-        if self.full_quran:
-            try:
-                surah = next((s for s in self.full_quran if s['id'] == surah_id), None)
-                if surah:
-                    if 'arabic1' in surah:  # API
-                        verses = surah['arabic1']
-                        return "\n\n".join(f"آية {i + 1}: {v}" for i, v in enumerate(verses))
-                    else:  # Fallback
-                        verses = surah['verses']
-                        return "\n\n".join(f"آية {v['id']}: {v['text']}" for v in verses)
-            except Exception as e:
-                self.logger.error(f"Error processing cached surah {surah_id}: {e}")
-        
-        max_retries = 3
-        for attempt in range(max_retries):
-            try:
-                response = requests.get(f"{self.api_url}{surah_id}.json", timeout=10)
-                response.raise_for_status()
-                data = response.json()
-                verses = data['arabic1']
-                return "\n\n".join(f"آية {i + 1}: {v}" for i, v in enumerate(verses))
-            except Exception as e:
-                self.logger.error(f"Attempt {attempt + 1}/{max_retries} failed to fetch surah {surah_id}: {e}")
-                if attempt == max_retries - 1:
-                    return self._load_fallback_verse()
-                time.sleep(2 ** attempt)
-    
-    def search_verses(self, query, top_k=5):
-        if self.verse_embeddings is None or not self.all_verses:
-            return self._keyword_fallback_search(query, top_k)
-        
-        try:
-            # Add context for single words or short queries
-            if len(query.split()) <= 1:
-                query = f"معنى كلمة {query}"  # Add context: "meaning of the word"
-            query_embedding = self.model.encode(["query: " + query], convert_to_tensor=False)
-            similarities = cosine_similarity(query_embedding, self.verse_embeddings)[0]
-            top_indices = np.argsort(similarities)[-top_k:][::-1]
-            
-            results = []
-            for idx in top_indices:
-                verse = self.all_verses[idx]
-                surah_name = next((s.get('surahNameArabicLong', s.get('name', '')) for s in self.surahs if s['id'] == verse['surah_id']), f"سورة {verse['surah_id']}")
-                if surah_name.startswith("سورة "):
-                    surah_name = surah_name[len("سورة "):]
-                results.append(
-                    f"سورة {surah_name} - آية {verse['verse_num']}:\n{verse['text']}"
-                )
-            return "\n\n".join(results) or "لا توجد نتائج ذات صلة."
-        except Exception as e:
-            self.logger.error(f"Search failed: {e}")
-            return "حدث خطأ أثناء البحث. جرب مرة أخرى."
-
-    def _keyword_fallback_search(self, query, top_k=5):
-        query_lower = query.lower()
-        matches = []
-        for verse in self.all_verses:
-            if query_lower in verse['text'].lower():
-                surah_name = next((s.get('surahNameArabicLong', s.get('name', '')) for s in self.surahs if s['id'] == verse['surah_id']), f"سورة {verse['surah_id']}")
-                if surah_name.startswith("سورة "):
-                    surah_name = surah_name[len("سورة "):]
-                matches.append(f"سورة {surah_name} - آية {verse['verse_num']}:\n{verse['text']}")
-        return "\n\n".join(matches[:top_k]) or "لا توجد نتائج مطابقة."
-
-    def _load_fallback_surahs(self):
-        return [
-            ("الفاتحة", 1),
-            ("البقرة", 2),
-            ("آل عمران", 3)
-        ]
-
-    def _load_fallback_verse(self):
-        return "بسم الله الرحمن الرحيم\nالله لا إله إلا هو الحي القيوم"