Update app.py

app.py

@@ -38,8 +38,9 @@ except ImportError: PIL_TESSERACT_AVAILABLE = False; print("WARNING: Pillow or P
 try: import whisper; WHISPER_AVAILABLE = True
 except ImportError: WHISPER_AVAILABLE = False; print("WARNING: OpenAI Whisper not found, Audio Transcription tool will be disabled.")
 
-# Google GenAI (Used by LangChain integration AND direct client)
+# Google GenAI SDK types (used by langchain-google-genai and direct client)
 from google.genai.types import HarmCategory, HarmBlockThreshold
+from google.ai import generativelanguage as glm # For FileState enum
 
 # LangChain
 from langchain_core.messages import HumanMessage, AIMessage, SystemMessage, ToolMessage
@@ -337,10 +338,13 @@ def direct_multimodal_gemini_tool(action_input_json_str: str) -> str:
         try:
             pil_image = Image.open(local_image_path)
         except Exception as e_img_open: return f"Error opening image {local_image_path}: {str(e_img_open)}"
-        response = google_genai_client.models.generate_content(
-            model=GEMINI_FLASH_MULTIMODAL_MODEL_NAME, contents=[pil_image, text_prompt]
+        
+        model_id_for_client = f"models/{GEMINI_FLASH_MULTIMODAL_MODEL_NAME}" if not GEMINI_FLASH_MULTIMODAL_MODEL_NAME.startswith("models/") else GEMINI_FLASH_MULTIMODAL_MODEL_NAME
+        response = google_genai_client.generate_content( # Corrected to use the alias
+            model=model_id_for_client,
+            contents=[pil_image, text_prompt]
         )
-        logger.info(f"Direct Multimodal Tool: Response from {GEMINI_FLASH_MULTIMODAL_MODEL_NAME} received.")
+        logger.info(f"Direct Multimodal Tool: Response received from {model_id_for_client} received.")
         return response.text[:40000]
     except json.JSONDecodeError as e_json_mm: return f"Error parsing JSON for Direct MM Tool: {str(e_json_mm)}. Input: {action_input_json_str}"
     except Exception as e_tool_mm:
@@ -386,9 +390,8 @@ def initialize_agent_and_tools(force_reinit=False):
     logger.info("Initializing agent and tools...")
     if not GOOGLE_API_KEY: raise ValueError("GOOGLE_API_KEY not set for LangChain LLM.")
     
-    # CORRECTED safety_settings format for ChatGoogleGenerativeAI
-    # Using INTEGER VALUES for HarmCategory keys and HarmBlockThreshold enum members for values.
-    llm_safety_settings_corrected_again = {
+    # Using INTEGER VALUES for HarmCategory keys AND HarmBlockThreshold enum .value for values.
+    llm_safety_settings_final_corrected = {
         HarmCategory.HARM_CATEGORY_HARASSMENT.value: HarmBlockThreshold.BLOCK_NONE.value,
         HarmCategory.HARM_CATEGORY_HATE_SPEECH.value: HarmBlockThreshold.BLOCK_NONE.value,
         HarmCategory.HARM_CATEGORY_SEXUALLY_EXPLICIT.value: HarmBlockThreshold.BLOCK_NONE.value,
@@ -400,14 +403,14 @@ def initialize_agent_and_tools(force_reinit=False):
             model=GEMINI_MODEL_NAME,
             google_api_key=GOOGLE_API_KEY,
             temperature=0.0,
-            safety_settings=llm_safety_settings_corrected_again, # USE THE DICTIONARY WITH INT VALUES FOR BOTH
-            timeout=120,                                        # Corrected: 'timeout'
+            safety_settings=llm_safety_settings_final_corrected, # USE THE DICTIONARY WITH INT VALUES FOR BOTH
+            timeout=120,
             convert_system_message_to_human=True
         )
         logger.info(f"LangChain LLM (Planner) initialized: {GEMINI_MODEL_NAME}")
     except Exception as e:
         logger.error(f"LangChain LLM init failed: {e}", exc_info=True)
-        logger.error(f"Type of safety_settings attempted: {type(llm_safety_settings_corrected_again)}, content: {llm_safety_settings_corrected_again}")
+        logger.error(f"Type of safety_settings attempted: {type(llm_safety_settings_final_corrected)}, content: {llm_safety_settings_final_corrected}")
         raise
 
     TOOLS = []
@@ -631,7 +634,7 @@ with gr.Blocks(css=".gradio-container {max-width:1280px !important;margin:auto !
     demo.load(update_ui_on_load_fn_within_context, [], [agent_status_display, missing_secrets_display])
 
 if __name__ == "__main__":
-    logger.info(f"Application starting up (v7 - SafetySettings INT Values)...")
+    logger.info(f"Application starting up (v7 - SafetySettings INT Values FINAL fix)...")
     if not PYPDF2_AVAILABLE: logger.warning("PyPDF2 (PDF tool) NOT AVAILABLE.")
     if not PIL_TESSERACT_AVAILABLE: logger.warning("Pillow/Pytesseract (OCR tool) NOT AVAILABLE.")
     if not WHISPER_AVAILABLE: logger.warning("Whisper (Audio tool) NOT AVAILABLE.")
@@ -659,5 +662,4 @@ if __name__ == "__main__":
 
     logger.info(f"Space ID: {os.getenv('SPACE_ID', 'Not Set')}")
     logger.info("Gradio Interface launching...")
-    demo.queue().launch(debug=os.getenv("GRADIO_DEBUG","false").lower()=="true", share=False, max_threads=20)
-    
+    demo.queue().launch(debug=os.getenv("GRADIO_DEBUG","false").lower()=="true", share=False, max_threads=20)