Update run.py

run.py

@@ -33,6 +33,8 @@ def fast_caption(sys_prompt, usr_prompt, temp, top_p, max_tokens, model, key, en
         caption = api.get_caption(sys_prompt, usr_prompt, base64_list)
         return f"{caption}", f"Using model '{model}' with {len(frames)} frames extracted.", debug_image
     elif video_hf and video_hf_auth:
+        current_file_path = os.path.abspath(__file__)
+        current_directory = os.path.dirname(current_file_path)
         print('begin video_hf')
         # Process all videos in the dataset
         all_captions = []
@@ -51,15 +53,16 @@ def fast_caption(sys_prompt, usr_prompt, temp, top_p, max_tokens, model, key, en
 
             md5 = hashlib.md5(video).hexdigest()
             print(md5)
-            with open(md5, 'wb') as f:
-                fc = f.write(video)
+            with tempfile.NamedTemporaryFile(dir=current_directory) as temp_file:
+                temp_file.write(video)
+                video_path = temp_file.name
 
-            processor = VideoProcessor(frame_format=frame_format, frame_limit=frame_limit)
-            frames = processor._decode(md5)
-            base64_list = processor.to_base64_list(frames)
-            api = AzureAPI(key=key, endpoint=endpoint, model=model, temp=temp, top_p=top_p, max_tokens=max_tokens)
-            caption = api.get_caption(sys_prompt, usr_prompt, base64_list)
-            all_captions.append(caption)
+                processor = VideoProcessor(frame_format=frame_format, frame_limit=frame_limit)
+                frames = processor._decode(video_path)
+                base64_list = processor.to_base64_list(frames)
+                api = AzureAPI(key=key, endpoint=endpoint, model=model, temp=temp, top_p=top_p, max_tokens=max_tokens)
+                caption = api.get_caption(sys_prompt, usr_prompt, base64_list)
+                all_captions.append(caption)
 
         return "\n\n\n".join(all_captions), f"Processed {len(video_paths)} videos.", None