Update app.py

app.py

@@ -8,7 +8,7 @@ import spaces
 import cv2
 import numpy as np
 from PIL import Image
-
+import html # <--- 1. 导入 html 模块
 def progress_bar_html(label: str) -> str:
     """
     Returns an HTML snippet for a thin progress bar with a label.
@@ -68,52 +68,6 @@ def model_inference(input_dict, history):
     text = input_dict["text"]
     files = input_dict["files"]
 
-    if text.strip().lower().startswith("@video-infer"):
-        # Remove the tag from the query.
-        text = text[len("@video-infer"):].strip()
-        if not files:
-            gr.Error("Please upload a video file along with your @video-infer query.")
-            return
-        # Assume the first file is a video.
-        video_path = files[0]
-        frames = downsample_video(video_path)
-        if not frames:
-            gr.Error("Could not process video.")
-            return
-        # Build messages: start with the text prompt.
-        messages = [
-            {
-                "role": "user",
-                "content": [{"type": "text", "text": text}]
-            }
-        ]
-        # Append each frame with a timestamp label.
-        for image, timestamp in frames:
-            messages[0]["content"].append({"type": "text", "text": f"Frame {timestamp}:"})
-            messages[0]["content"].append({"type": "image", "image": image})
-        # Collect only the images from the frames.
-        video_images = [image for image, _ in frames]
-        # Prepare the prompt.
-        prompt = processor.apply_chat_template(messages, tokenize=False, add_generation_prompt=True)
-        inputs = processor(
-            text=[prompt],
-            images=video_images,
-            return_tensors="pt",
-            padding=True,
-        ).to("cuda")
-        # Set up streaming generation.
-        streamer = TextIteratorStreamer(processor, skip_prompt=True, skip_special_tokens=False)
-        generation_kwargs = dict(inputs, streamer=streamer, max_new_tokens=1024)
-        thread = Thread(target=model.generate, kwargs=generation_kwargs)
-        thread.start()
-        buffer = ""
-        yield progress_bar_html("Processing video with MiMo-VL-7B-RL Model")
-        for new_text in streamer:
-            buffer += new_text
-            time.sleep(0.01)
-            yield buffer
-        return
-
     if len(files) > 1:
         images = [load_image(image) for image in files]
     elif len(files) == 1:
@@ -151,7 +105,9 @@ def model_inference(input_dict, history):
     buffer = ""
     yield progress_bar_html("Processing with MiMo-VL-7B-RL Model")
     for new_text in streamer:
-        buffer += new_text
+        escaped_new_text = html.escape(new_text)
+        buffer += escaped_new_text
+        
         time.sleep(0.01)
         yield buffer