Update response_generator.py

response_generator.py

@@ -30,31 +30,24 @@ class ResponseGenerator:
         """
         Smart extraction with sentence selection
         """
-        # Get top chunks
         top_chunks = relevant_docs.head(3)
         
-        # Extract query keywords
         query_keywords = self._extract_keywords(query)
         
-        # Find best sentences
         best_sentences = []
         
         for idx, row in top_chunks.iterrows():
             content = row['content']
             similarity = row.get('similarity_score', 0)
             
-            # Split into sentences
             sentences = self._split_sentences(content)
             
-            # Score each sentence
             for sentence in sentences:
-                if len(sentence.strip()) < 20:  # Skip very short sentences
+                if len(sentence.strip()) < 20:
                     continue
                 
-                # Calculate sentence score
                 score = similarity
                 
-                # Boost if contains query keywords
                 sentence_lower = sentence.lower()
                 for keyword in query_keywords:
                     if keyword in sentence_lower:
@@ -66,17 +59,13 @@ class ResponseGenerator:
                     'chunk_id': idx
                 })
         
-        # Sort by score
         best_sentences.sort(key=lambda x: x['score'], reverse=True)
         
-        # Take top 3 sentences
         top_sentences = best_sentences[:3]
         
-        # Format answer
         if not top_sentences:
             return self._format_simple_answer(top_chunks)
         
-        # Build answer
         answer_parts = ["**الإجابة:**\n"]
         
         for i, sent in enumerate(top_sentences, 1):
@@ -85,7 +74,6 @@ class ResponseGenerator:
             else:
                 answer_parts.append(f"\n{sent['text']}")
         
-        # Add metadata
         answer_parts.append("\n\n---")
         
         scores = top_chunks['similarity_score'].values if 'similarity_score' in top_chunks.columns else []
@@ -100,10 +88,8 @@ class ResponseGenerator:
         """
         Extract meaningful keywords from query
         """
-        # Remove stop words
         stop_words = {'ما', 'هي', 'هو', 'في', 'من', 'إلى', 'على', 'عن', 'ال', 'و', 'أو'}
         
-        # Split and clean
         words = query.lower().split()
         keywords = [w for w in words if len(w) > 2 and w not in stop_words]
         
@@ -113,7 +99,6 @@ class ResponseGenerator:
         """
         Split text into sentences
         """
-        # Split by Arabic and English punctuation
         sentences = re.split(r'[.؟!]\s+', text)
         return [s.strip() for s in sentences if s.strip()]
     
@@ -124,10 +109,8 @@ class ResponseGenerator:
         best_chunk = top_chunks.iloc[0]
         content = best_chunk['content']
         
-        # Truncate if too long
         if len(content) > 400:
             content = content[:400]
-            # Try to end at sentence
             last_period = max(content.rfind('.'), content.rfind('؟'), content.rfind('!'))
             if last_period > 100:
                 content = content[:last_period + 1]
@@ -136,6 +119,6 @@ class ResponseGenerator:
         
         score = best_chunk.get('similarity_score', 0)
         
-        return f"""**الإجابة:**
-
-{content}
+        answer = f"**الإجابة:**\n\n{content}\n\n---\n**دقة المطابقة:** {score:.0%}"
+        
+        return answer