Update app.py

app.py

@@ -62,15 +62,18 @@ MOCK_ANSWERS = {
 # --- Task Handlers ---
 def handle_grocery_list(question: str) -> str:
     logger.info("Handling grocery list task")
-    match = re.search(r'(?:list\s*[:\s]*|items\s*[:\s]*)([^.]+)', question, re.IGNORECASE)
+    # Extract items
+    match = re.search(r'list\s*[:\s]*([^.]+)', question, re.IGNORECASE)
     if not match:
         logger.error("Failed to extract grocery list")
         return "Error: Invalid input"
     items = match.group(1).strip()
+    # Define botanical fruits
     botanical_fruits = [
         "bell pepper", "zucchini", "corn", "plums", "cucumber", "tomato",
         "eggplant", "pumpkin", "squash", "avocado", "olive"
     ]
+    # Filter vegetables
     vegetables = [
         item.strip().lower() for item in items.split(',')
         if item.strip().lower() not in botanical_fruits and any(
@@ -82,27 +85,33 @@ def handle_grocery_list(question: str) -> str:
     ]
     vegetables = sorted(vegetables)
     result = ','.join(vegetables) if vegetables else "None"
-    logger.info(f"Grocery list result: {result}")
-    # Force Groq API call for formatting
+    logger.info(f"Raw grocery list result: {result}")
+    # Use Groq for formatting
     rate_limiter.acquire()
     @backoff.on_exception(backoff.expo, Exception, max_tries=3)
     def groq_format():
         client = Groq(api_key=GROQ_API_KEY)
+        prompt = f"""
+        Format the following list as a comma-separated string with no spaces, no extra characters, and exact item names in alphabetical order:
+        Input: {result}
+        Example: broccoli,carrot,lettuce
+        Output: {result}
+        """
         response = client.chat.completions.create(
             model="llama-3.3-70b-versatile",
-            messages=[{
-                "role": "user",
-                "content": f"Return the input as a comma-separated list with no spaces, no extra characters, and exact item names: {result}"
-            }],
+            messages=[{"role": "user", "content": prompt}],
             temperature=0
         )
         formatted = response.choices[0].message.content.strip()
-        debug_log.append({"task_id": "8e867cd7-cff9-4e6c-867a-ff5ddc2550be", "groq_response": formatted})
+        debug_log.append({"task_id": "8e867cd7-cff9-4e6c-867a-ff5ddc2550be", "groq_prompt": prompt, "groq_response": formatted})
         # Validate response
-        if len(formatted) > 100 or any(c in formatted for c in '|;'):
+        if len(formatted) > 100 or any(c in formatted for c in ':|;'):
             logger.error(f"Invalid Groq response: {formatted}")
             return result
-        logger.info(f"Groq API call successful for grocery list: {formatted}")
+        if not re.match(r'^[a-z,]+$', formatted):
+            logger.error(f"Groq response format incorrect: {formatted}")
+            return result
+        logger.info(f"Groq formatted result: {formatted}")
         return formatted
     try:
         return groq_format()
@@ -118,25 +127,37 @@ def handle_sales(file_path: str) -> str:
     try:
         df = pd.read_excel(file_path)
         logger.info(f"Excel columns: {list(df.columns)}")
-        df.columns = df.columns.str.lower()
-        if 'category' not in df.columns or 'sales' not in df.columns:
-            logger.error("Missing required columns")
+        # Case-insensitive column search
+        df.columns = [col.lower().strip() for col in df.columns]
+        category_col = next((col for col in df.columns if 'category' in col), None)
+        sales_col = next((col for col in df.columns if 'sales' in col), None)
+        if not category_col or not sales_col:
+            logger.error(f"Columns missing: category={category_col}, sales={sales_col}")
             return "Error: Missing 'category' or 'sales' columns"
-        food_sales = df[df['category'].str.lower() == 'food']['sales'].sum()
+        food_sales = df[df[category_col].str.lower() == 'food'][sales_col].sum()
         result = f"{food_sales:.2f}"
         logger.info(f"Sales result: {result}")
         rate_limiter.acquire()
         @backoff.on_exception(backoff.expo, Exception, max_tries=3)
         def groq_format():
             client = Groq(api_key=GROQ_API_KEY)
+            prompt = f"""
+            Format the number as a string with two decimal places, no extra characters:
+            Input: {result}
+            Example: 1234.56
+            Output: {result}
+            """
             response = client.chat.completions.create(
                 model="llama-3.3-70b-versatile",
-                messages=[{"role": "user", "content": f"Format number with two decimal places, no extra characters: {result}"}],
+                messages=[{"role": "user", "content": prompt}],
                 temperature=0
             )
             formatted = response.choices[0].message.content.strip()
-            debug_log.append({"task_id": "7bd855d8-463d-4ed5-93ca-5fe35145f733", "groq_response": formatted})
-            logger.info(f"Groq API call successful for sales: {formatted}")
+            debug_log.append({"task_id": "7bd855d8-463d-4ed5-93ca-5fe35145f733", "groq_prompt": prompt, "groq_response": formatted})
+            if not re.match(r'^\d+\.\d{2}$', formatted):
+                logger.error(f"Invalid Groq sales format: {formatted}")
+                return result
+            logger.info(f"Groq formatted sales: {formatted}")
             return formatted
         try:
             return groq_format()
@@ -156,14 +177,23 @@ def handle_web(task_id: str, question: str) -> str:
         @backoff.on_exception(backoff.expo, Exception, max_tries=3)
         def groq_validate():
             client = Groq(api_key=GROQ_API_KEY)
+            prompt = f"""
+            Return the input exactly as provided, no extra characters, no spaces:
+            Input: {result}
+            Example: Input: John,Doe Output: John,Doe
+            Output: {result}
+            """
             response = client.chat.completions.create(
                 model="llama-3.3-70b-versatile",
-                messages=[{"role": "user", "content": f"Return the input exactly as provided, no extra characters: {result}"}],
+                messages=[{"role": "user", "content": prompt}],
                 temperature=0
             )
             formatted = response.choices[0].message.content.strip()
-            debug_log.append({"task_id": task_id, "groq_response": formatted})
-            logger.info(f"Groq API call successful for web task {task_id}: {formatted}")
+            debug_log.append({"task_id": task_id, "groq_prompt": prompt, "groq_response": formatted})
+            if formatted != result:
+                logger.error(f"Groq altered web task output: {formatted}")
+                return result
+            logger.info(f"Groq validated web task {task_id}: {formatted}")
             return formatted
         try:
             return groq_validate()
@@ -183,6 +213,7 @@ TASK_HANDLERS = {
 }
 
 # --- File Download ---
+@backoff.on_exception(backoff.expo, RequestException, max_tries=3)
 def download_file_if_any(base_api_url: str, task_id: str) -> Optional[str]:
     if task_id != "7bd855d8-463d-4ed5-93ca-5fe35145f733":
         logger.info(f"No file needed for task_id: {task_id}")
@@ -196,7 +227,7 @@ def download_file_if_any(base_api_url: str, task_id: str) -> Optional[str]:
             return None
         resp.raise_for_status()
         cdisp = resp.headers.get("content-disposition", "")
-        filename = task_id
+        filename = task_id + ".xlsx"
         if "filename=" in cdisp:
             m = re.search(r'filename="([^"]+)"', cdisp)
             if m:
@@ -214,12 +245,12 @@ def download_file_if_any(base_api_url: str, task_id: str) -> Optional[str]:
 
 # --- Answer Formatter ---
 def format_answer(answer: str, task_id: str) -> str:
-    space_preserved_tasks = ["3cef3a44-215e-4aed-8e3b-b1e3f08063b7", "99c9cc74-fdc8-46c6-8f8d-3ce2d3bfeea3"]
     answer = answer.strip()
-    if task_id not in space_preserved_tasks:
-        answer = re.sub(r'\s+', '', answer)
-        answer = re.sub(r',\s*', ',', answer)
-        answer = re.sub(r'[^a-zA-Z0-9,.]', '', answer)  # Remove |, ;, etc.
+    # Strict formatting: remove spaces, non-alphanumeric except commas/periods
+    answer = re.sub(r'\s+', '', answer)
+    answer = re.sub(r'[^a-zA-Z0-9,.]', '', answer)
+    # Ensure no trailing commas or periods
+    answer = answer.rstrip(',.')
     logger.info(f"Formatted answer for task_id {task_id}: {answer}")
     return answer
 
@@ -263,20 +294,25 @@ class GroqAgent:
             rate_limiter.acquire()
             @backoff.on_exception(backoff.expo, Exception, max_tries=3)
             def groq_fallback():
+                prompt = f"""
+                Answer in a single value or comma-separated list, no spaces, no extra characters:
+                Question: {question[:500]}
+                Example: apple,banana or 123.45
+                """
                 response = self.client.chat.completions.create(
                     model="llama-3.3-70b-versatile",
-                    messages=[{"role": "user", "content": f"Answer in a single value or comma-separated list, no spaces, no extra characters: {question[:500]}"}],
+                    messages=[{"role": "user", "content": prompt}],
                     temperature=0
                 )
                 result = response.choices[0].message.content.strip()
-                debug_log.append({"task_id": task_id, "groq_response": result})
-                logger.info(f"Groq API call successful for unknown task {task_id}: {result}")
+                debug_log.append({"task_id": task_id, "groq_prompt": prompt, "groq_response": result})
+                logger.info(f"Groq fallback result for {task_id}: {result}")
                 return result
             try:
                 result = groq_fallback()
                 return format_answer(result, task_id)
             except Exception as e:
-                logger.error(f"Groq error for unknown task {task_id}: {e}")
+                logger.error(f"Groq fallback error for task_id {task_id}: {e}")
                 return format_answer(MOCK_ANSWERS.get(task_id, "Error"), task_id)
         file_path = download_file_if_any(DEFAULT_API_URL, task_id)
         try: