Update app.py

app.py

@@ -186,53 +186,63 @@ def generate_summary_and_assign_category(bookmark):
             else:
                 use_prior_knowledge = False
 
-            # Prepare the prompt
+            # Shortened prompts
             if use_prior_knowledge:
-                # Construct prompt to use prior knowledge
                 prompt = f"""
 You are a knowledgeable assistant with up-to-date information as of 2023.
 
-The user provided a URL: {bookmark.get('url')}
+URL: {bookmark.get('url')}
 
-Please provide:
-1. A concise summary in **no more than two sentences** about this website.
-2. Assign the most appropriate category from the list below for this website.
+Provide:
+1. A concise summary (max two sentences) about this website.
+2. Assign the most appropriate category from the list below.
 
 Categories:
 {', '.join([f'"{cat}"' for cat in CATEGORIES])}
 
-Provide your response in the following format:
-Summary: [Your summary here]
-Category: [One of the categories]
+Format:
+Summary: [Your summary]
+Category: [One category]
 """
             else:
-                # Construct the prompt with the extracted content
                 prompt = f"""
-You are a helpful assistant that creates concise webpage summaries and assigns categories.
-
-Analyze the following webpage content:
+You are an assistant that creates concise webpage summaries and assigns categories.
 
+Content:
 {content_text}
 
-Please provide:
-1. A concise summary in **no more than two sentences** focusing on the main purpose or topic of the page and key information or features.
-2. Assign the most appropriate category from the list below for this webpage. **Ensure the category directly reflects the content of the summary.**
+Provide:
+1. A concise summary (max two sentences) focusing on the main topic.
+2. Assign the most appropriate category from the list below.
 
 Categories:
 {', '.join([f'"{cat}"' for cat in CATEGORIES])}
 
-Provide your response in the following format:
-Summary: [Your summary here]
-Category: [One of the categories]
+Format:
+Summary: [Your summary]
+Category: [One category]
 """
 
+            # Estimate tokens
+            def estimate_tokens(text):
+                return len(text) / 4  # Approximate token estimation
+
+            prompt_tokens = estimate_tokens(prompt)
+            max_tokens = 150  # Reduced from 200
+            total_tokens = prompt_tokens + max_tokens
+
+            # Calculate required delay
+            tokens_per_second = 6000 / 60  # 100 tokens per second
+            required_delay = total_tokens / tokens_per_second
+            sleep_time = max(required_delay, 1)
+
             # Call the LLM via Groq Cloud API
             response = openai.ChatCompletion.create(
                 model='llama-3.1-70b-versatile',
                 messages=[
                     {"role": "user", "content": prompt}
                 ],
-                max_tokens=200,
+                max_tokens=int(max_tokens),
                 temperature=0.5,
             )
             content = response['choices'][0]['message']['content'].strip()
@@ -266,7 +276,7 @@ Category: [One of the categories]
                 bookmark['category'] = 'Reference and Knowledge Bases'
 
             logger.info("Successfully generated summary and assigned category")
-            time.sleep(1)  # Reduced sleep time
+            time.sleep(sleep_time)
             break  # Exit the retry loop upon success
 
         except openai.error.RateLimitError as e:
@@ -616,19 +626,37 @@ Bookmarks:
 Provide a concise and helpful response.
 """
 
+        # Estimate tokens
+        def estimate_tokens(text):
+            return len(text) / 4  # Approximate token estimation
+
+        prompt_tokens = estimate_tokens(prompt)
+        max_tokens = 300  # Adjust as needed
+        total_tokens = prompt_tokens + max_tokens
+
+        # Calculate required delay
+        tokens_per_second = 6000 / 60  # 100 tokens per second
+        required_delay = total_tokens / tokens_per_second
+        sleep_time = max(required_delay, 1)
+
         response = openai.ChatCompletion.create(
             model='llama-3.1-70b-versatile',
             messages=[
                 {"role": "user", "content": prompt}
             ],
-            max_tokens=500,
+            max_tokens=int(max_tokens),
             temperature=0.7,
         )
         answer = response['choices'][0]['message']['content'].strip()
         logger.info("Chatbot response generated")
-        time.sleep(1)  # Reduced sleep time
+        time.sleep(sleep_time)
         return answer
 
+    except openai.error.RateLimitError as e:
+        wait_time = int(e.headers.get("Retry-After", 5))
+        logger.warning(f"Rate limit reached. Waiting for {wait_time} seconds before retrying...")
+        time.sleep(wait_time)
+        return chatbot_response(user_query)  # Retry after waiting
     except Exception as e:
         error_message = f"⚠️ Error processing your query: {str(e)}"
         logger.error(error_message, exc_info=True)