Update app.py

app.py

@@ -41,39 +41,6 @@ def sample_frames(video_file) :
         frames=[]
         return frames
 
-def llava(user_prompt, history):
-    image = user_prompt["files"][-1]    
-    txt = user_prompt["text"]
-    img = user_prompt["files"]
-    
-    video_extensions = ("avi", "mp4", "mov", "mkv", "flv", "wmv", "mjpeg", "wav", "gif", "webm", "m4v", "3gp")
-    image_extensions = Image.registered_extensions()
-    image_extensions = tuple([ex for ex, f in image_extensions.items()])
-        
-    if image.endswith(video_extensions):
-        image = sample_frames(image)
-        image_tokens = "<image>" * int(len(image))
-        prompt = f"<|im_start|>user {image_tokens}\n{user_prompt}<|im_end|><|im_start|>assistant"
-          
-    elif image.endswith(image_extensions):
-        image = Image.open(image).convert("RGB")
-        prompt = f"<|im_start|>user <image>\n{user_prompt}<|im_end|><|im_start|>assistant"
-
-    print(len(image))
-
-    inputs = processor(prompt, image, return_tensors="pt")
-    streamer = TextIteratorStreamer(processor, skip_prompt=True, **{"skip_special_tokens": True})
-    generation_kwargs = dict(inputs, streamer=streamer, max_new_tokens=1024)
-    generated_text = ""
-
-    thread = Thread(target=model.generate, kwargs=generation_kwargs)
-    thread.start()
-
-    buffer = ""
-    for new_text in streamer:
-        buffer += new_text
-        yield buffer
-
 def extract_text_from_webpage(html_content):
     soup = BeautifulSoup(html_content, 'html.parser')
     for tag in soup(["script", "style", "header", "footer"]):
@@ -122,7 +89,37 @@ def respond(message, history):
 
     # Handle image processing
     if message["files"]:
-        llava(message, history)
+        image = user_prompt["files"][-1]    
+        txt = user_prompt["text"]
+        img = user_prompt["files"]
+        
+        video_extensions = ("avi", "mp4", "mov", "mkv", "flv", "wmv", "mjpeg", "wav", "gif", "webm", "m4v", "3gp")
+        image_extensions = Image.registered_extensions()
+        image_extensions = tuple([ex for ex, f in image_extensions.items()])
+            
+        if image.endswith(video_extensions):
+            image = sample_frames(image)
+            image_tokens = "<image>" * int(len(image))
+            prompt = f"<|im_start|>user {image_tokens}\n{user_prompt}<|im_end|><|im_start|>assistant"
+              
+        elif image.endswith(image_extensions):
+            image = Image.open(image).convert("RGB")
+            prompt = f"<|im_start|>user <image>\n{user_prompt}<|im_end|><|im_start|>assistant"
+    
+        print(len(image))
+    
+        inputs = processor(prompt, image, return_tensors="pt")
+        streamer = TextIteratorStreamer(processor, skip_prompt=True, **{"skip_special_tokens": True})
+        generation_kwargs = dict(inputs, streamer=streamer, max_new_tokens=1024)
+        generated_text = ""
+    
+        thread = Thread(target=model.generate, kwargs=generation_kwargs)
+        thread.start()
+    
+        buffer = ""
+        for new_text in streamer:
+            buffer += new_text
+            yield buffer
         
 
     # Define function metadata for user interface