fix FastAPI模型FFMPEGSubtitleOverlayRequest解析后没有返回Self以及可选参数没有给default值

src/BowongModalFunctions/config.py

@@ -20,6 +20,7 @@ class WorkerConfig(BaseSettings):
     modal_product_kv_name: str = Field(default='live-product-cache', description="Modal抖音直播间商品缓存KV库")
     modal_environment: str = Field(default="dev", description="Modal worker运行环境")
     modal_app_name: str = Field(default='bowong-ai-video', description="Modal App集群名称")
+    modal_is_local: bool = Field(default=False, description="本地开发环境")
     comfyui_s3_input: Optional[str] = Field(default="comfyui-input", description="ComfyUI input S3文件夹名")
     comfyui_s3_output: Optional[str] = Field(default="comfyui-output", description="ComfyUI output S3文件夹名")
 

src/BowongModalFunctions/models/media_model.py

@@ -80,7 +80,7 @@ class MediaSource(BaseModel):
             cache_path = os.path.join(s3_mount_path, media_source.cache_filepath)
 
             # 校验媒体文件是否存在缓存中
-            if not os.path.exists(cache_path):
+            if not config.modal_is_local and not os.path.exists(cache_path):
                 raise ValueError(f"媒体文件 {media_source.cache_filepath} 不存在于缓存中")
 
             return media_source

src/BowongModalFunctions/models/web_model.py

@@ -286,9 +286,9 @@ class FFMPEGOverlayGifTaskStatusResponse(BaseFFMPEGTaskStatusResponse):
 
 class FFMPEGSubtitleOverlayRequest(BaseFFMPEGTaskRequest):
     media: MediaSource = Field(description="需要处理的媒体源")
-    subtitle: Optional[MediaSource] = Field(description="需要叠加的字幕文件")
+    subtitle: Optional[MediaSource] = Field(default=None, description="需要叠加的字幕文件")
-    embedded_subtitle: Optional[MediaSource] = Field(description="需要内嵌的字幕文件")
+    embedded_subtitle: Optional[MediaSource] = Field(default=None, description="需要内嵌的字幕文件")
-    fonts: Optional[List[MediaSource]] = Field(description="字幕文件内使用到的字体文件")
+    fonts: Optional[List[MediaSource]] = Field(default=None, description="字幕文件内使用到的字体文件")
 
     @field_validator('media', mode='before')
     @classmethod
@@ -351,6 +351,7 @@ class FFMPEGSubtitleOverlayRequest(BaseFFMPEGTaskRequest):
             raise pydantic.ValidationError("至少需要提供一个有效字幕")
         if self.subtitle is not None and self.fonts is None:
             raise pydantic.ValidationError("使用叠加字幕时需要指定使用的字体文件")
+        return self
 
 
 class FFMPEGSubtitleTaskStatusResponse(BaseFFMPEGTaskStatusResponse):
@@ -504,7 +505,7 @@ class GeminiRequest(BaseFFMPEGTaskRequest):
     end_time: str = Field(default="00:20:00.000", description="结束时间(hls)")
     options: FFMPEGSliceOptions = Field(default=FFMPEGSliceOptions(), description="输出质量选项")
     scale: float = Field(default=0.85, description="视频尺寸缩放倍率")
-    last_product_text:str = Field(default="", description="上一段视频结尾介绍的商品以及标签")
+    last_product_text: str = Field(default="", description="上一段视频结尾介绍的商品以及标签")
 
     @field_validator('media_hls_url', mode='before')
     @classmethod
@@ -673,6 +674,7 @@ class GeminiSecondStagePromptVariables(BaseModel):
     @computed_field(description="xml格式排列的识别出的商品JSON列表")
     @property
     def product_json_list_xml(self) -> str:
-        xml_items = [f"              <product>{json.dumps(product,ensure_ascii=False)}</product>" for product in self.product_json_list]
+        xml_items = [f"              <product>{json.dumps(product, ensure_ascii=False)}</product>" for product in
+                     self.product_json_list]
         xml_string = "\n".join(xml_items)
         return f"<products>\n{xml_string}\n            </products>"

tests/ffmpeg_test_case.py

@@ -111,21 +111,16 @@ class FFMPEGTestCase(unittest.IsolatedAsyncioTestCase):
 
     async def test_ffmpeg_slice_stream(self):
         from pydantic import TypeAdapter
+        # #EXT-X-PROGRAM-DATE-TIME:2025-06-23T06:20:30.720+0000
+        # #EXT-X-PROGRAM-DATE-TIME:2025-06-23T10:09:38.476+0000
 
-        stream_url = "https://cdn.roasmax.cn/test/records/hls/fc-01JXY1AS1HDGS300EQ54ATAKHF/playlist.m3u8"
+        # stream_url = "https://cdn.roasmax.cn/test/records/hls/fc-01JXY1AS1HDGS300EQ54ATAKHF/playlist.m3u8"
+        stream_url = "https://cdn.roasmax.cn/test/records/hls/fc-01JYDQ3RVMPAEKMYHA0HQ14WAJ/playlist.m3u8"
         adapter = TypeAdapter(List[FFMpegSliceSegment])
         segments = adapter.validate_json("""[
         {
-          "start": 920.425,
+          "start": 12600,
-          "end": 921.688
+          "end": 13500
-        },
-        {
-          "start": 922.425,
-          "end": 923.688
-        },
-        {
-          "start": 940.425,
-          "end": 941.688
         }        
       ]""")
         for segment in segments:

tests/pydantic_model_validation_test.py

@@ -3,6 +3,7 @@ import unittest
 
 import httpx
 from loguru import logger
+from BowongModalFunctions.models.web_model import FFMPEGSubtitleOverlayRequest
 
 
 class PydanticModelTestCase(unittest.TestCase):
@@ -25,6 +26,14 @@ class PydanticModelTestCase(unittest.TestCase):
             logger.info(f"Login response: {session_json}")
         self.assertEqual(True, True)
 
+    def test_subtitle_apply(self):
+        subtitle_apply_input = """
+        {\"media\": \"s3://ap-northeast-2/modal-media-cache/test/bgm_nosie_reduce/outputs/fc-01JYG6NVQ1AEJMARECGN8FXDYH/output.mp4\",\n  \"subtitle\": \"s3://ap-northeast-2/modal-media-cache/upload/2d5e2674-2c03-4a83-a589-3dce96003470/1183fed3-770d-4ded-a314-ef47c37d84d7.ass\",\n  \"fonts\": [\n    \"s3://ap-northeast-2/modal-media-cache/upload/test/fonts/荆南俊俊体.ttf\"\n  ]}
+        """
+
+        request = FFMPEGSubtitleOverlayRequest.model_validate_json(subtitle_apply_input)
+        logger.info(f"request = {request.model_dump_json(indent=2, exclude_none=True)}")
+
 
 if __name__ == '__main__':
     unittest.main()