Update app.py

app.py

@@ -1,6 +1,7 @@
 import os
 import logging
 import json
+import time
 import gradio as gr
 from huggingface_hub import InferenceClient
 from langchain.embeddings import HuggingFaceEmbeddings
@@ -9,6 +10,7 @@ from langchain.schema import Document
 from duckduckgo_search import DDGS
 from dotenv import load_dotenv
 from functools import lru_cache
+from tenacity import retry, stop_after_attempt, wait_fixed
 
 # Load environment variables
 load_dotenv()
@@ -70,7 +72,11 @@ def create_web_search_vectors(search_results):
     logger.info(f"Created vectors for {len(documents)} search results.")
     return FAISS.from_documents(documents, embed)
 
-def get_response_with_search(query, system_prompt, model, use_embeddings, num_calls=3, temperature=0.2):
+@retry(stop=stop_after_attempt(3), wait=wait_fixed(2))
+def make_api_call(client, api_params):
+    return client.chat_completion(**api_params)
+
+def get_response_with_search(query, system_prompt, model, use_embeddings, history, num_calls=3, temperature=0.2):
     searcher = WebSearcher()
     search_results = searcher.search(query)
     
@@ -91,7 +97,11 @@ def get_response_with_search(query, system_prompt, model, use_embeddings, num_ca
 
     logger.info(f"Context created for query: {query}")
 
-    user_message = f"""Using the following context from web search results:
+    chat_history = "\n".join([f"Human: {h[0]}\nAI: {h[1]}" for h in history])
+    user_message = f"""Chat history:
+{chat_history}
+
+Using the following context from web search results:
 {context}
 
 Write a detailed and complete research document that fulfills the following user request: '{query}'."""
@@ -110,7 +120,7 @@ Write a detailed and complete research document that fulfills the following user
                 "top_p": 0.8,
             }
             logger.info(f"Sending request to API with params: {json.dumps(api_params, indent=2, default=str)}")
-            response = client.chat_completion(**api_params)
+            response = make_api_call(client, api_params)
             logger.info(f"Raw response from model: {response}")
 
             if isinstance(response, dict):
@@ -128,6 +138,8 @@ Write a detailed and complete research document that fulfills the following user
             else:
                 logger.error(f"Unexpected response format from the model: {type(response)}")
                 return "Unexpected response format from the model. Please try again.", ""
+            
+            time.sleep(1)  # Add a 1-second delay between API calls
     except Exception as e:
         logger.error(f"Error in get_response_with_search: {str(e)}")
         logger.info(f"Attempting fallback to {FALLBACK_MODEL}")
@@ -149,9 +161,10 @@ def respond(message, system_prompt, history, model, temperature, num_calls, use_
     logger.info(f"Number of API Calls: {num_calls}")
     logger.info(f"Use Embeddings: {use_embeddings}")
     logger.info(f"System Prompt: {system_prompt}")
+    logger.info(f"History: {history}")
 
     try:
-        main_content, sources = get_response_with_search(message, system_prompt, model, use_embeddings, num_calls=num_calls, temperature=temperature)
+        main_content, sources = get_response_with_search(message, system_prompt, model, use_embeddings, history, num_calls=num_calls, temperature=temperature)
         return main_content
     except Exception as e:
         logger.error(f"Error in respond function: {str(e)}")