Update app.py

app.py

@@ -177,53 +177,63 @@ async def rephrase_query(chat_history, query, temperature=0.2) -> str:
     """Rephrase the query based on chat history and context."""
     logger.info(f'Rephrasing query: {query}')
     try:
-        # Format chat history for context
+        # Format recent conversation history (last 3 turns for context)
         formatted_history = []
-        for user_msg, assistant_msg in chat_history:
-            formatted_history.append({"role": "user", "content": user_msg})
-            if assistant_msg:  # Only add if there's an assistant message
-                formatted_history.append({"role": "assistant", "content": assistant_msg})
+        for i, (user_msg, assistant_msg) in enumerate(chat_history[-3:], 1):
+            formatted_history.append(f"Turn {i}:")
+            formatted_history.append(f"User: {user_msg}")
+            if assistant_msg:
+                formatted_history.append(f"Assistant: {assistant_msg}")
         
+        chat_context = "\n".join(formatted_history)
         current_year = datetime.now().year 
-        system_prompt = """You are a highly intelligent and context-aware query rephrasing assistant. Your task is to rephrase search queries while following these strict rules:
-
-1. Entity Handling:
-   - Identify main entities (organizations, brands, products, locations)
-   - Enclose ONLY the entity names in double quotes
-   - Example: "Apple" stock price, not "Apple stock price"
-
-2. Date Handling Rules (VERY IMPORTANT):
-   - For queries about current/latest/recent information:
-     * If query contains words like "latest", "current", "recent", "now", "today":
-       - Keep these words in the query
-       - ALWAYS append "after: YYYY" (current year) at the end
-     * Example: "latest news on "Apple"" becomes "latest news on "Apple" after: 2024"
-   
-   - For queries with specific time periods:
-     * Keep the original time reference
-     * Add appropriate "after: YYYY" based on the mentioned year
-     * Example: "How did "Bank of America" perform in Q2 2023" becomes
-       "How did "Bank of America" perform in Q2 2023 after: 2023"
-
-   - For queries without any time reference:
-     * ALWAYS append "after: YYYY" (current year) at the end
-     * Example: ""Toyota" market share" becomes ""Toyota" market share after: 2024"
-
-3. Output Format:
-   - First letter should be capitalized
-   - No period at the end
-   - Include all specified date operators
-   - Maintain the entire original query's meaning and context
-
-Remember: EVERY query must end with a date operator unless it explicitly references a past date/year."""
-
-        # Prepare messages for the API call
+
+        system_prompt = """You are a highly intelligent query rephrasing assistant. Your task is to analyze the conversation history and current query to generate a complete, contextual search query.
+
+Key Rules:
+1. For follow-up questions or queries referencing previous conversation:
+   - Extract the main topic/subject from previous messages
+   - Combine previous context with the current query
+   - Example: 
+     Previous: "What is the structure of German banking industry?"
+     Current: "can you do more latest web search on my previous query"
+     Should become: "Latest structure and developments in German banking industry after: 2024"
+
+2. Entity Handling:
+   - Identify and preserve main entities from context
+   - Enclose ONLY entity names in double quotes
+   - Example: "Deutsche Bank" profits, not "Deutsche Bank profits"
+
+3. Date and Time Context:
+   - For queries about current/latest information:
+     * Keep time-related words (latest, current, recent, now)
+     * ALWAYS append "after: YYYY" (current year)
+   - For specific time periods:
+     * Preserve the original time reference
+     * Add appropriate "after: YYYY" based on context
+   - For queries without time reference:
+     * Add "after: YYYY" if about current state/status
+
+4. Query Formatting:
+   - Capitalize first letter
+   - No period at end
+   - Include all relevant context
+   - Maintain clear and searchable structure
+
+Remember: Your goal is to create a complete, self-contained query that includes all necessary context from the conversation history."""
+
         messages = [
             {"role": "system", "content": system_prompt},
-            {"role": "user", "content": f"Current year is {current_year}. Rephrase this query: {query}"}
+            {"role": "user", "content": f"""Current year: {current_year}
+
+Recent conversation history:
+{chat_context}
+
+Current query: {query}
+
+Please rephrase this query into a complete, contextual search query following the rules above. The rephrased query should be clear and complete even without the conversation context."""}
         ]
 
-        # Call Groq API
         response = groq_client.chat.completions.create(
             messages=messages,
             model="llama-3.1-70b-versatile",
@@ -238,7 +248,14 @@ Remember: EVERY query must end with a date operator unless it explicitly referen
 
     except Exception as e:
         logger.error(f'Error rephrasing query: {e}')
-        return query  # Return original query if rephrasing fails
+        # If rephrasing fails, construct a basic contextual query
+        try:
+            last_query = chat_history[-1][0] if chat_history else ""
+            if any(word in query.lower() for word in ['latest', 'recent', 'current', 'now', 'update']):
+                return f"{last_query} latest updates after: {datetime.now().year}"
+            return query
+        except:
+            return query  # Return original query as last resort
 
 class ParallelScraper:
     def __init__(self, max_workers: int = 5):