fix: separate video payload from images in task API

server/tasks.py

@@ -1,5 +1,6 @@
 """简单任务 API：将聊天任务与 WebSocket 解耦，支持后台运行与轮询。"""
 from __future__ import annotations
+import mimetypes
 import time
 import threading
 import uuid
@@ -101,6 +102,7 @@ class TaskManager:
         message: str,
         images: List[Any],
         conversation_id: Optional[str],
+        videos: Optional[List[Any]] = None,
         model_key: Optional[str] = None,
         thinking_mode: Optional[bool] = None,
         run_mode: Optional[str] = None,
@@ -137,7 +139,7 @@ class TaskManager:
                 record.session_data = {}
         with self._lock:
             self._tasks[task_id] = record
-        thread = threading.Thread(target=self._run_chat_task, args=(record, images), daemon=True)
+        thread = threading.Thread(target=self._run_chat_task, args=(record, images, videos or []), daemon=True)
         record.thread = thread
         record.status = "running"
         record.updated_at = time.time()
@@ -199,7 +201,7 @@ class TaskManager:
             })
             rec.updated_at = time.time()
 
-    def _run_chat_task(self, rec: TaskRecord, images: List[Any]):
+    def _run_chat_task(self, rec: TaskRecord, images: List[Any], videos: List[Any]):
         username = rec.username
         workspace_id = rec.workspace_id
         terminal = None
@@ -269,6 +271,7 @@ class TaskManager:
                 client_sid=rec.task_id,
                 workspace=workspace,
                 username=username,
+                videos=videos,
             )
 
             # 结束状态
@@ -347,8 +350,7 @@ def create_task_api():
     workspace_id = session.get("workspace_id") or "default"
     payload = request.get_json() or {}
     message = (payload.get("message") or "").strip()
-    images = payload.get("images") or []
+    images, videos = _normalize_media_payload(payload.get("images") or [], payload.get("videos") or [])
-    videos = payload.get("videos") or []
     conversation_id = payload.get("conversation_id")
     if not message and not images and not videos:
         return jsonify({"success": False, "error": "消息不能为空"}), 400
@@ -357,16 +359,14 @@ def create_task_api():
     run_mode = payload.get("run_mode")
     max_iterations = payload.get("max_iterations")
 
-    # 合并 images 和 videos 到 images 参数（后端统一处理）
-    all_media = images + videos
-
     try:
         rec = task_manager.create_chat_task(
             username,
             workspace_id,
             message,
-            all_media,
+            images,
             conversation_id,
+            videos=videos,
             model_key=model_key,
             thinking_mode=thinking_mode,
             run_mode=run_mode,
@@ -385,6 +385,48 @@ def create_task_api():
     }), 202
 
 
+def _media_path(item: Any) -> str:
+    if isinstance(item, dict):
+        return str(item.get("path") or "")
+    return str(item or "")
+
+
+def _is_video_item(item: Any) -> bool:
+    path = _media_path(item)
+    if not path:
+        return False
+    mime, _ = mimetypes.guess_type(path)
+    return bool(mime and mime.startswith("video/"))
+
+
+def _is_image_item(item: Any) -> bool:
+    path = _media_path(item)
+    if not path:
+        return False
+    mime, _ = mimetypes.guess_type(path)
+    return bool(mime and mime.startswith("image/"))
+
+
+def _normalize_media_payload(images: List[Any], videos: List[Any]) -> tuple[List[Any], List[Any]]:
+    """纠偏媒体字段：把误传到 images 的视频项自动归入 videos。"""
+    fixed_images: List[Any] = []
+    fixed_videos: List[Any] = []
+
+    for item in images or []:
+        if _is_video_item(item):
+            fixed_videos.append(item)
+        else:
+            fixed_images.append(item)
+
+    for item in videos or []:
+        if _is_image_item(item):
+            fixed_images.append(item)
+        else:
+            fixed_videos.append(item)
+
+    return fixed_images, fixed_videos
+
+
 @tasks_bp.route("/api/tasks/<task_id>", methods=["GET"])
 @api_login_required
 def get_task_api(task_id: str):

utils/api_client.py

@@ -205,6 +205,7 @@ class DeepSeekClient:
             return text
         qwen_video_fps = 2
         parts: List[Dict[str, Any]] = []
+        extra_videos: List[Any] = []
         if text:
             parts.append({"type": "text", "text": text})
         base_path = Path(self.project_path or ".")
@@ -214,6 +215,11 @@ class DeepSeekClient:
                 if not abs_path.exists() or not abs_path.is_file():
                     continue
                 mime, _ = mimetypes.guess_type(abs_path.name)
+                if mime and mime.startswith("video/"):
+                    extra_videos.append(path)
+                    continue
+                if mime and not mime.startswith("image/"):
+                    continue
                 if not mime:
                     mime = "image/png"
                 data = abs_path.read_bytes()
@@ -221,7 +227,7 @@ class DeepSeekClient:
                 parts.append({"type": "image_url", "image_url": {"url": f"data:{mime};base64,{b64}"}})
             except Exception:
                 continue
-        for item in videos:
+        for item in [*videos, *extra_videos]:
             try:
                 if isinstance(item, dict):
                     path = item.get("path") or ""

utils/context_manager.py

@@ -1608,6 +1608,7 @@ class ContextManager:
         if not images and not videos:
             return text
         parts: List[Dict[str, Any]] = []
+        extra_videos: List[Any] = []
         supports_video_fps = getattr(getattr(self, "main_terminal", None), "model_key", None) == "qwen3-vl-plus"
         qwen_video_fps = 2
         if text:
@@ -1617,9 +1618,14 @@ class ContextManager:
                 abs_path = Path(self.project_path) / path
                 if not abs_path.exists() or not abs_path.is_file():
                     continue
+                mime, _ = mimetypes.guess_type(abs_path.name)
+                if mime and mime.startswith("video/"):
+                    extra_videos.append(path)
+                    continue
+                if mime and not mime.startswith("image/"):
+                    continue
                 data_url = self._compress_image_if_needed(abs_path)
                 if not data_url:
-                    mime, _ = mimetypes.guess_type(abs_path.name)
                     if not mime:
                         mime = "image/png"
                     data = abs_path.read_bytes()
@@ -1628,7 +1634,7 @@ class ContextManager:
                 parts.append({"type": "image_url", "image_url": {"url": data_url}})
             except Exception:
                 continue
-        for item in videos:
+        for item in [*videos, *extra_videos]:
             try:
                 if isinstance(item, dict):
                     path = item.get("path") or ""