Update app.py

app.py

@@ -1,51 +1,185 @@
-import gradio as gr
-from tools.arabic_generator import ArabicTextGenerator
-from tools.quran_search import QuranSearchEngine
+import requests
+import logging
+import numpy as np
+from sentence_transformers import SentenceTransformer
+from sklearn.metrics.pairwise import cosine_similarity
+from config import QURAN_DATA_SOURCES, MODEL_NAME, CHUNK_SIZE
+import time
 
-text_gen = ArabicTextGenerator()
-quran = QuranSearchEngine()
-
-with gr.Blocks(title="الأدوات العربية") as app:
-    # Tab 1: Arabic Generator (unchanged)
-    with gr.Tab("🖊️ مولد النصوص"):
-        text_input = gr.Textbox(label="النص الأولي")
-        length_slider = gr.Slider(50, 300, value=100, label="طول النص") 
-        gen_btn = gr.Button("توليد")
-        text_output = gr.Textbox(label="النص المولد", lines=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
+        self.model = None
+        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
         
-        gen_btn.click(
-            text_gen.generate,
-            inputs=[text_input, length_slider],
-            outputs=text_output
-        )
-
-    # Tab 2: Quran Surah Viewer (unchanged)
-    with gr.Tab("📖 القرآن الكريم"):
-        surah_dropdown = gr.Dropdown(
-            label="اختر سورة",
-            choices=quran.get_surahs(),
-            value=quran.get_surahs()[0][1] if quran.get_surahs() else 1
-        )
-        show_btn = gr.Button("عرض السورة")
-        quran_output = gr.Textbox(label="النص القرآني", lines=15)
+    def _load_full_quran(self):
+        max_retries = 3
+        for attempt in range(max_retries):
+            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')
+                # 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)
+                break
+            except Exception as e:
+                self.logger.error(f"Attempt {attempt + 1}/{max_retries} failed to fetch full Quran from API: {e}")
+                if attempt == max_retries - 1:
+                    self._load_fallback_quran()
+                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_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"},
+            ]
+            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,
+                        '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
+                    })
         
-        show_btn.click(
-            quran.get_surah_text,
-            inputs=surah_dropdown,
-            outputs=quran_output
-        )
+        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, prompt_name='retrieval.passage')
+                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
     
-    # New Tab 3: Quranic Verse Search (Semantic)
-    with gr.Tab("🔍 بحث آيات قرآنية"):
-        search_input = gr.Textbox(label="أدخل استعلام البحث (مثال: الرحمة والمغفرة)")
-        top_k_slider = gr.Slider(1, 10, value=5, label="عدد النتائج") 
-        search_btn = gr.Button("بحث")
-        search_output = gr.Textbox(label="الآيات ذات الصلة", lines=15)
+    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)
         
-        search_btn.click(
-            quran.search_verses,
-            inputs=[search_input, top_k_slider],
-            outputs=search_output
-        )
+        try:
+            query_embedding = self.model.encode([query], prompt_name='retrieval.query')
+            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']}")
+                # Avoid double "سورة"
+                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)
+        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)
+        ]
 
-app.launch()
+    def _load_fallback_verse(self):
+        return "بسم الله الرحمن الرحيم\nالله لا إله إلا هو الحي القيوم"