Update app.py

app.py

@@ -22,10 +22,9 @@ def extract_text_from_webpage(html_content):
 
 def search(query):
     term = query
-    print(f"Running web search for query: {term}")
     start = 0
     all_results = []
-    max_chars_per_page = 8000  
+    max_chars_per_page = 8000
     with requests.Session() as session:
         resp = session.get(
             url="https://www.google.com/search",
@@ -47,7 +46,7 @@ def search(query):
                 if len(visible_text) > max_chars_per_page:
                     visible_text = visible_text[:max_chars_per_page]
                 all_results.append({"link": link, "text": visible_text})
-            except requests.exceptions.RequestException as e:
+            except requests.exceptions.RequestException:
                 all_results.append({"link": link, "text": None})
     return all_results
 
@@ -55,23 +54,19 @@ def search(query):
 client_gemma = InferenceClient("google/gemma-1.1-7b-it")
 client_mixtral = InferenceClient("NousResearch/Nous-Hermes-2-Mixtral-8x7B-DPO")
 client_llama = InferenceClient("meta-llama/Meta-Llama-3-8B-Instruct")
-messages = []
+func_caller = []
 
 # Define the main chat function
 def respond(message, history):
-    global messages # Make messages global for persistent storage 
+    global func_caller
     vqa = ""
 
     # Handle image processing
     if message["files"]:
-        try:
-            for image in message["files"]:
-                vqa += "[CAPTION of IMAGE]  "
-                gr.Info("Analyzing image")
-                vqa += generate_caption_instructblip(image, message["text"])
-                print(vqa)
-        except:
-            vqa = ""
+        for image in message["files"]:
+            vqa += "[CAPTION of IMAGE]  "
+            gr.Info("Analyzing image")
+            vqa += generate_caption_instructblip(image, message["text"])
 
     # Define function metadata for user interface
     functions_metadata = [
@@ -82,11 +77,8 @@ def respond(message, history):
     ]
 
     message_text = message["text"]
+    func_caller.append({"role": "user", "content": f'[SYSTEM]You are a helpful assistant. You have access to the following functions: \n {str(functions_metadata)}\n\nTo use these functions respond with:\n<functioncall> {{ "name": "function_name", "arguments": {{ "arg_1": "value_1", "arg_1": "value_1", ... }} }}  </functioncall>  [USER] {message} {vqa}'})
 
-    # Append user messages and system instructions to the messages list
-    messages.append({"role": "user", "content": f'[SYSTEM]You are a helpful assistant. You have access to the following functions: \n {str(functions_metadata)}\n\nTo use these functions respond with:\n<functioncall> {{ "name": "function_name", "arguments": {{ "arg_1": "value_1", "arg_1": "value_1", ... }} }}  </functioncall>  [USER] {message} {vqa}'})
-
-    # Call the LLM for response generation
     response = client_gemma.chat_completion(messages, max_tokens=150)
     response = str(response)
     try:
@@ -98,9 +90,8 @@ def respond(message, history):
     response = response.replace('\\"', '"')
     print(f"\n{response}")
 
-    messages.append({"role": "assistant", "content": f"<functioncall>{response}</functioncall>"})
+    func_caller.append({"role": "assistant", "content": f"<functioncall>{response}</functioncall>"})
 
-    # Process and return the response based on the function call
     try:
         json_data = json.loads(str(response))
         if json_data["name"] == "web_search":