Update app.py

app.py

@@ -3,57 +3,27 @@ import os
 from google import genai
 from google.genai import types
 import gradio as gr
-# removed unused IPython imports
-# from IPython.display import display
-# from IPython.display import Markdown
-
-# Ensure the API key environment variable is set
-api_key = os.environ.get("GEMINI_API_KEY")
-if not api_key:
-    raise ValueError("GEMINI_API_KEY environment variable not set.")
-
-# Configure the generative AI client
-#genai.configure(api_key=api_key)
-
-# Note: Using genai.GenerativeModel is the more current practice
-# compared to genai.Client().models... but client works too.
-# Let's stick to the user's client approach for consistency.
-client = genai.Client(api_key=api_key)
-
-# Define the generation function compatible with Gradio ChatInterface streaming
-def generate(message: str, history: list[list[str]]):
-    """
-    Generates a response from the Gemini model, streaming the output.
-
-    Args:
-        message: The latest user message.
-        history: A list of previous user/assistant message pairs.
-                 Example: [["Hello", "Hi there!"], ["How are you?", "I'm doing well."]]
-
-    Yields:
-        str: The cumulative response text as it's generated chunk by chunk.
-    """
-
-    model_name = "gemini-2.5-pro-exp-03-25"
-    # model_name = "gemini-1.5-flash-latest" # Alternative models
-    # model_name = "gemini-1.5-pro-latest"
-    # model_name = "gemini-pro" # Standard pro model
-
-    # --- Construct conversation history for the API ---
-    # The API expects a list of Content objects (user, model, user, model...)
-    api_history = []
-    for user_msg, model_msg in history:
-        api_history.append(types.Content(role="user", parts=[types.Part.from_text(user_msg)]))
-        # Ensure model messages are also added correctly
-        if model_msg: # Avoid adding empty model messages if something went wrong previously
-             api_history.append(types.Content(role="model", parts=[types.Part.from_text(model_msg)]))
-
-    # Add the current user message to the end
-    contents = api_history + [
-        types.Content(role="user", parts=[types.Part.from_text(message)])
+from IPython.display import display
+from IPython.display import Markdown
+
+client = genai.Client(
+    api_key=os.environ.get("GEMINI_API_KEY"),
+)
+
+def generate(prompt, *args):
+
+    model = "gemini-2.5-pro-exp-03-25"
+    #model = "gemini-2.0-flash"
+    #model = "gemini-2.0-pro-exp-02-05"
+    #model = "gemini-2.0-flash-thinking-exp-01-21"
+    contents = [
+        types.Content(
+            role="user",
+            parts=[
+                types.Part.from_text(text=f"{prompt}"),
+            ],
+        ),
     ]
-
-    # --- Configure generation parameters ---
     tools = [
         types.Tool(google_search=types.GoogleSearch())
     ]
@@ -61,70 +31,28 @@ def generate(message: str, history: list[list[str]]):
         temperature=0.35,
         top_p=0.95,
         top_k=40,
-        # max_output_tokens=65536, # Use model's default unless specific limit needed
-        # max_output_tokens=8192,
+        max_output_tokens=65536,
+        #max_output_tokens=8192,
         tools=tools,
         response_mime_type="text/plain",
     )
 
-    # --- Call the API and stream the response ---
-    stream = None # Initialize stream variable
-    try:
-        stream = client.generate_content( # Use generate_content with stream=True
-            model=f"models/{model_name}", # Model name needs 'models/' prefix for generate_content
-            contents=contents,
-            generation_config=generate_content_config,
-            stream=True,
-            # safety_settings=... # Optional: configure safety settings if needed
-        )
-
-        full_response = ""
-        for chunk in stream:
-            # Check if the chunk contains text parts
-            # Handle potential errors like blocked content or empty chunks
-            try:
-                if chunk.text:
-                    full_response += chunk.text
-                    yield full_response # Yield the cumulative response so far
-            except ValueError:
-                 # Handle cases where the chunk might be blocked or have no text
-                 # (e.g., safety settings blocked response)
-                 # You could yield an error message or just skip
-                 print(f"Warning: Received chunk without valid text: {chunk.prompt_feedback}")
-                 # yield full_response + "\n\n[Content generation issue]" # Option to show user
-                 pass # Silently ignore chunks without text
+    response = ""
+    for chunk in client.models.generate_content_stream(
+        model=model,
+        contents=contents,
+        config=generate_content_config,
+    ):
+        response += chunk.text
+        
+    display(Markdown(response))
+    return response
 
-    except Exception as e:
-        print(f"An error occurred during generation: {e}")
-        # Yield a user-friendly error message
-        yield f"Sorry, an error occurred: {str(e)}"
-    finally:
-        # Although Python generators handle cleanup, explicitly clearing
-        # resources if needed could go here. For this stream, it's usually automatic.
-        pass
 
-# --- Launch the Gradio Interface ---
 if __name__ == "__main__":
     iface = gr.ChatInterface(
-        fn=generate, # The generator function handles streaming
-        title=gr.Markdown("<h1><center>Chatbot with Google Search</center></h1>"), # Centered title
-        description="Chatbot powered by Gemini and Google Search. Type your message and press Enter.",
-        chatbot=gr.Chatbot(
-            height=600, # Adjust chatbot window height
-            show_copy_button=True,
-            bubble_full_width=False, # Make bubbles wrap text
-        ),
-        textbox=gr.Textbox(placeholder="Ask me anything...", container=False, scale=7),
-        retry_btn="Retry",
-        undo_btn="Undo",
-        clear_btn="Clear",
-        # Examples can be useful for users
-        examples=[
-            "What's the latest news about generative AI?",
-            "Explain the concept of quantum entanglement simply.",
-            "Write a short poem about a rainy day.",
-            "Summarize the main points of the Paris Agreement on climate change.",
-        ],
-        cache_examples=False # Disable caching if examples should always trigger a fresh generation
+        fn=generate,
+        title=gr.Markdown("# Chatbot with Google Search"),
+        description="Chatbot powered by Gemini-2.5 Pro and Google Search.",
     )
-    iface.launch(share=False) # Set share=True to get a public link (use with caution)
+    iface.launch()