chore: Refactor process_input function to support local and online browsing

main.py

@@ -47,11 +47,11 @@ class MultiOnInputBrowse(BaseModel):
     Attributes:
         cmd (str): The command to execute. Example: "post 'hello world - I love multion' on twitter".
         url (str): The URL where the action should be performed. Example: "https://twitter.com".
-        local (bool): Flag indicating whether the action should be performed locally. Default is True.
+        local (bool): Flag indicating whether the action should be performed locally. Default is False.
     """
     cmd: str
     url: str
-    local: bool = True
+    local: bool = False
 
 async def process_image_file(file: UploadFile) -> str:
     """
@@ -69,12 +69,16 @@ async def process_image_file(file: UploadFile) -> str:
     if file.content_type not in ["image/jpeg", "image/png"]:
         raise HTTPException(status_code=400, detail="Invalid file type. Only JPEG and PNG are supported.")
     
+    print("Reading image file")
     image_data = await file.read()
     image = Image.open(io.BytesIO(image_data))
+    print("Image loaded")
 
     try:
+        print("Encoding image")
         enc_image = model.encode_image(image)
         description = model.answer_question(enc_image, "Describe this image.", tokenizer)
+        print("Image description generated")
         return description
     except Exception as e:
         raise HTTPException(status_code=500, detail=str(e))
@@ -84,10 +88,11 @@ def read_root():
     return {"Hello": "World"}
 
 @app.post("/process-input/")
-async def process_input(text: str = Form(...), file: UploadFile = File(None)):
+async def process_input(text: str = Form(...), file: UploadFile = File(None), online: bool = Form(False)):
     if file is not None:
         try:
             print("Processing image file")
+            print(f"File type: type(file) = {type(file)}, Filename: {file.filename}, Content type: {file.content_type}")
             image_description = await process_image_file(file)
             print(f"Image description: {image_description}")
         except HTTPException as e:
@@ -104,21 +109,23 @@ async def process_input(text: str = Form(...), file: UploadFile = File(None)):
     
     print(f"Processed text: {processed_text}")
     command = await generate_command(processed_text)
-    print(f"Command generated: {command.message}")
-
-    try:
-        print("Calling MultiOn API")
-        response = multion.browse(
-            cmd=command.cmd,
-            url=command.url,
-            local=command.local
-        )
-        print(f"Response received: {response.message}")
-        return JSONResponse(content={"response": response.message, "command": command.model_dump()})
-
-    except Exception as e:
-        raise HTTPException(status_code=500, detail=f"Mution API error: {str(e)}")
+    print(f"Command generated: {command}")
 
+    if not online and not command.local:
+        try:
+            print("Calling MultiOn API with online=True")
+            response = multion.browse(
+                cmd=command.cmd,
+                url=command.url,
+                local=command.local
+            )
+            print(f"Response received: {response.message}")
+            return JSONResponse(content={"response": response.message, "command": command.model_dump()})
+
+        except Exception as e:
+            raise HTTPException(status_code=500, detail=f"Mution API error: {str(e)}")
+    else:
+        return JSONResponse(content={"response": "This command is for local browsing", "command": command.model_dump()})
 
 async def generate_command(content: str) -> MultiOnInputBrowse:
     try: