Update app.py

app.py

@@ -104,6 +104,7 @@ from fastapi import FastAPI, HTTPException
 from fastapi.responses import JSONResponse
 from pydantic import BaseModel
 from huggingface_hub import InferenceClient
+from requests.exceptions import HTTPError
 
 # Set up logging
 logging.basicConfig(level=logging.INFO)
@@ -138,7 +139,7 @@ def llm_chat_response(text: str, image_url: Optional[str] = None) -> str:
             api_key=HF_TOKEN
         )
         
-        # Build the messages payload dynamically.
+        # Build the messages payload.
         # For text-only queries, append a default instruction.
         message_content = [{
             "type": "text",
@@ -147,14 +148,14 @@ def llm_chat_response(text: str, image_url: Optional[str] = None) -> str:
         
         if image_url:
             logger.info("Downloading and converting image to base64 data URI...")
-            # Download the image from the URL
+            # Download the image
             image_response = requests.get(image_url)
             if image_response.status_code != 200:
                 logger.error("Failed to download image from URL")
                 raise HTTPException(status_code=500, detail="Failed to download image from provided URL")
             image_bytes = image_response.content
-            
-            # Determine the MIME type
+
+            # Get the MIME type from the response headers
             mime_type = image_response.headers.get("Content-Type")
             if not mime_type or not mime_type.startswith("image/"):
                 logger.error("Invalid image MIME type")
@@ -163,6 +164,7 @@ def llm_chat_response(text: str, image_url: Optional[str] = None) -> str:
             # Encode image in base64 and format as a data URI
             base64_image = base64.b64encode(image_bytes).decode("utf-8")
             data_uri = f"data:{mime_type};base64,{base64_image}"
+            logger.info(f"Data URI created: {data_uri[:50]}...")  # log first 50 chars for verification
             
             message_content.append({
                 "type": "image_url",
@@ -175,19 +177,31 @@ def llm_chat_response(text: str, image_url: Optional[str] = None) -> str:
         }]
         
         logger.info("Sending request to model...")
-        completion = client.chat.completions.create(
-            model="meta-llama/Llama-3.2-11B-Vision-Instruct",
-            messages=messages,
-            max_tokens=500
-        )
+        try:
+            completion = client.chat.completions.create(
+                model="meta-llama/Llama-3.2-11B-Vision-Instruct",
+                messages=messages,
+                max_tokens=500
+            )
+        except HTTPError as http_err:
+            # Log HTTP errors from the request
+            logger.error(f"HTTP error occurred: {http_err.response.text}")
+            raise HTTPException(status_code=500, detail=http_err.response.text)
         
         logger.info(f"Raw model response: {completion}")
         
+        # If the model returned an error field, capture and return that error.
+        if getattr(completion, "error", None):
+            error_details = completion.error
+            error_message = error_details.get("message", "Unknown error")
+            logger.error(f"Model returned error: {error_message}")
+            raise HTTPException(status_code=500, detail=f"Model returned error: {error_message}")
+        
         if not completion.choices or len(completion.choices) == 0:
             logger.error("No choices returned from model.")
             raise HTTPException(status_code=500, detail="Model returned no choices.")
         
-        # Extract the message content from the first choice.
+        # Extract the response message from the first choice.
         choice = completion.choices[0]
         response_message = None
         if hasattr(choice, "message"):
@@ -247,5 +261,3 @@ async def method_not_allowed_handler(request, exc):
         status_code=405,
         content={"error": "Method not allowed. Please check the API documentation."}
     )
-
-