Update tools/quran_search.py

tools/quran_search.py

@@ -1,6 +1,5 @@
 import requests
 import logging
-import pandas as pd
 import numpy as np
 from sentence_transformers import SentenceTransformer
 from sklearn.metrics.pairwise import cosine_similarity
@@ -27,13 +26,18 @@ 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
-                # Fetch full verses by iterating over surah IDs
+                self.surahs = response.json()  # Array of surah metadata (no 'id')
+                # 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()
-                    self.full_quran.append(surah_response.json())
+                    surah_data = surah_response.json()
+                    surah_data['id'] = surah_id  # Add id for consistency
+                    self.full_quran.append(surah_data)
                 break
             except Exception as e:
                 self.logger.error(f"Attempt {attempt + 1}/{max_retries} failed to fetch full Quran from API: {e}")
@@ -66,16 +70,25 @@ class QuranSearchEngine:
             return
         
         for surah in self.full_quran:
-            surah_id = surah.get('id', surah.get('surah_id', 1))  # Handle API vs fallback structure
-            verses = surah.get('verses', surah.get('arabic1', []))  # API uses 'arabic1', fallback uses 'verses'
-            for verse in verses:
-                verse_num = verse.get('id', verse.get('verse_id', 1))
-                text = verse.get('text', verse)  # Handle different key structures
-                self.all_verses.append({
-                    'surah_id': surah_id,
-                    'verse_num': verse_num,
-                    'text': text
-                })
+            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,
+                        '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)
@@ -93,7 +106,7 @@ class QuranSearchEngine:
     def get_surahs(self):
         if self.surahs:
             return [
-                (s['name'], s['id'])
+                (s.get('surahNameArabicLong', s.get('name', '')), s['id'])
                 for s in self.surahs
             ]
         return self._load_fallback_surahs()
@@ -101,10 +114,14 @@ class QuranSearchEngine:
     def get_surah_text(self, surah_id):
         if self.full_quran:
             try:
-                surah = next((s for s in self.full_quran if s.get('id', s.get('surah_id', 1)) == surah_id), None)
+                surah = next((s for s in self.full_quran if s['id'] == surah_id), None)
                 if surah:
-                    verses = surah.get('verses', surah.get('arabic1', []))
-                    return "\n\n".join(f"آية {v.get('id', i + 1)}: {v.get('text', v)}" for i, v in enumerate(verses))
+                    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}")
         
@@ -134,7 +151,7 @@ class QuranSearchEngine:
             results = []
             for idx in top_indices:
                 verse = self.all_verses[idx]
-                surah_name = next((s['name'] for s in self.surahs if s.get('id', s.get('surah_id', 1)) == verse['surah_id']), f"سورة {verse['surah_id']}")
+                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']}")
                 results.append(
                     f"سورة {surah_name} - آية {verse['verse_num']}:\n{verse['text']}"
                 )
@@ -148,33 +165,16 @@ class QuranSearchEngine:
         matches = []
         for verse in self.all_verses:
             if query_lower in verse['text'].lower():
-                surah_name = next((s['name'] for s in self.surahs if s.get('id', s.get('surah_id', 1)) == verse['surah_id']), f"سورة {verse['surah_id']}")
+                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']}")
                 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):
-        try:
-            for source in QURAN_DATA_SOURCES:
-                try:
-                    response = requests.get(source, timeout=10)
-                    response.raise_for_status()
-                    data = response.json()
-                    return [(s['name'], s['id']) for s in data]
-                except Exception as e:
-                    self.logger.error(f"Failed to load fallback surahs from {source}: {e}")
-                    continue
-            return [
-                ("الفاتحة", 1),
-                ("البقرة", 2),
-                ("آل عمران", 3)
-            ]
-        except Exception as e:
-            self.logger.error(f"Failed to load fallback surahs: {e}")
-            return [
-                ("الفاتحة", 1),
-                ("البقرة", 2),
-                ("آل عمران", 3)
-            ]
+        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"