Update tools/quran_search.py

tools/quran_search.py

@@ -10,7 +10,6 @@ class QuranSearchEngine:
     def __init__(self):
         self.api_url = "https://quranapi.pages.dev/api/"
         self.logger = logging.getLogger(__name__)
-        self.full_quran = None
         self.surahs = None
         self.all_verses = []  # List of {'surah_id': int, 'verse_num': int, 'text': str}
         self.verse_embeddings = None
@@ -20,8 +19,6 @@ class QuranSearchEngine:
         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
@@ -29,54 +26,113 @@ class QuranSearchEngine:
             try:
                 response = requests.get(f"{self.api_url}surah.json", timeout=10)
                 response.raise_for_status()
-                self.surahs = response.json()  # Array of surah metadata (no 'id')
+                self.surahs = response.json()
                 # Add 'id' to surahs for consistency
                 for i, s in enumerate(self.surahs):
                     s['id'] = i + 1
                 # Fetch full verses
-                self.full_quran = []
                 for surah_id in range(1, 115):
                     surah_response = requests.get(f"{self.api_url}{surah_id}.json", timeout=10)
                     surah_response.raise_for_status()
-                    surah_data = surah_response.json()
-                    surah_data['id'] = surah_id  # Add id for consistency
-                    self.full_quran.append(surah_data)
+                    data = surah_response.json()
+                    verses = data['arabic1']  # Arabic with tashkeel
+                    for verse_num, text in enumerate(verses, start=1):
+                        self.all_verses.append({
+                            'surah_id': surah_id,
+                            'verse_num': verse_num,
+                            'text': text
+                        })
                 break
             except Exception as e:
-                self.logger.error(f"Attempt {attempt + 1}/{max_retries} failed to fetch full Quran from API: {e}")
+                self.logger.error(f"Attempt {attempt + 1}/{max_retries} failed to fetch Quran data: {e}")
                 if attempt == max_retries - 1:
-                    self._load_fallback_quran()
+                    self._load_fallback_data()
                 time.sleep(2 ** attempt)
     
-    def _load_fallback_quran(self):
-        max_retries = 3
-        for source in QURAN_DATA_SOURCES:
-            for attempt in range(max_retries):
-                try:
-                    response = requests.get(source, timeout=10)
-                    response.raise_for_status()
-                    self.full_quran = response.json()  # Array of surah dicts
-                    self.surahs = self.full_quran
-                    break
-                except Exception as e:
-                    self.logger.error(f"Attempt {attempt + 1}/{max_retries} failed to fetch fallback from {source}: {e}")
-                    if attempt == max_retries - 1 and source == QURAN_DATA_SOURCES[-1]:
-                        self.surahs = self._load_fallback_surahs()
-                    time.sleep(2 ** attempt)
+    def _load_fallback_data(self):
+        self.surahs = self._load_fallback_surahs()
+        self.all_verses = [
+            {'surah_id': 1, 'verse_num': 1, 'text': "بِسْمِ ٱللَّهِ ٱلرَّحْمَـٰنِ ٱلرَّحِيمِ"},
+            {'surah_id': 1, 'verse_num': 2, 'text': "ٱلْحَمْدُ لِلَّهِ رَبِّ ٱلْعَٰلَمِينَ"}
+            # Add more hardcoded verses if needed, but limited
+        ]
     
     def _load_all_verses_and_embeddings(self):
-        if not self.full_quran:
-            self.logger.error("No full Quran loaded, skipping verse loading")
-            self.all_verses = [
-                {'surah_id': 1, 'verse_num': 1, 'text': "بِسْمِ ٱللَّهِ ٱلرَّحْمَـٰنِ ٱلرَّحِيمِ"},
-            ]
+        if not self.all_verses:
             return
         
-        for surah in self.full_quran:
-            surah_id = surah.get('id', 1)
-            if 'arabic1' in surah:  # API structure: verses as list of str
-                verses = surah['arabic1']
-                for verse_num, text in enumerate(verses, start=1):
-                    self.all_verses.append({
-                        'surah_id': surah_id,
-                        'verse_num': verse_num,
+        try:
+            self.model = SentenceTransformer(MODEL_NAME)
+            verse_texts = [v['text'] for v in self.all_verses]
+            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['surahNameArabicLong'], s['id'])
+                for s in self.surahs
+            ]
+        return self._load_fallback_surahs()
+    
+    def get_surah_text(self, surah_id):
+        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:
+            query_embedding = self.model.encode([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 = self.surahs[verse['surah_id'] - 1]['surahNameArabicLong']
+                results.append(
+                    f"سورة {surah_name} - آية {verse['verse_num']}:\n{verse['text']}"
+                )
+            return "\n\n".join(results)
+        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 = self.surahs[verse['surah_id'] - 1]['surahNameArabicLong'] if self.surahs else f"سورة {verse['surah_id']}"
+                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"