Update app.py

app.py

@@ -31,6 +31,9 @@ processor = AutoProcessor.from_pretrained("lingshu-medical-mllm/Lingshu-7B")
 MAX_NUM_IMAGES = int(os.getenv("MAX_NUM_IMAGES", "5"))
 
 
+MAX_NUM_IMAGES = int(os.getenv("MAX_NUM_IMAGES", "5"))
+
+
 def count_files_in_new_message(paths: list[str]) -> tuple[int, int]:
     image_count = 0
     video_count = 0
@@ -111,7 +114,7 @@ def process_video(video_path: str) -> list[dict]:
         with tempfile.NamedTemporaryFile(delete=False, suffix=".png") as temp_file:
             pil_image.save(temp_file.name)
             content.append({"type": "text", "text": f"Frame {timestamp}:"})
-            content.append({"type": "image", "url": temp_file.name})
+            content.append({"type": "image", "image": temp_file.name})
     logger.debug(f"{content=}")
     return content
 
@@ -126,7 +129,7 @@ def process_interleaved_images(message: dict) -> list[dict]:
     for part in parts:
         logger.debug(f"{part=}")
         if part == "<image>":
-            content.append({"type": "image", "url": message["files"][image_index]})
+            content.append({"type": "image", "image": message["files"][image_index]})
             logger.debug(f"file: {message['files'][image_index]}")
             image_index += 1
         elif part.strip():
@@ -140,16 +143,13 @@ def process_interleaved_images(message: dict) -> list[dict]:
 def process_new_user_message(message: dict) -> list[dict]:
     if not message["files"]:
         return [{"type": "text", "text": message["text"]}]
-
     if message["files"][0].endswith(".mp4"):
         return [{"type": "text", "text": message["text"]}, *process_video(message["files"][0])]
-
     if "<image>" in message["text"]:
         return process_interleaved_images(message)
-
     return [
         {"type": "text", "text": message["text"]},
-        *[{"type": "image", "url": path} for path in message["files"]],
+        *[{"type": "image", "image": path} for path in message["files"]],
     ]
 
 
@@ -167,7 +167,7 @@ def process_history(history: list[dict]) -> list[dict]:
             if isinstance(content, str):
                 current_user_content.append({"type": "text", "text": content})
             else:
-                current_user_content.append({"type": "image", "url": content[0]})
+                current_user_content.append({"type": "image", "image": content[0]})
     return messages