Skip to content

利用 AI 大模型和自动化工作流,根据主题或关键词一键生成高清短视频。Generate HD short videos from a topic or keyword with an automated AI workflow.

fix(task): avoid visual term generation for narration-only jobs

Rudy Celekli245fd113500e
diff --git a/app/services/task.py b/app/services/task.py
index 6334b721..12b18ba2 100644
--- a/app/services/task.py
+++ b/app/services/task.py
@@ -1614,7 +1614,7 @@ def _run_pipeline(
# 2. Generate terms
video_terms = ""
- if params.video_source != "local":
+ if stop_at in {"terms", "materials", "video"} and params.video_source != "local":
video_terms = generate_terms(task_id, params, video_script)
if not video_terms:
return _mark_task_failed(
diff --git a/test/services/test_audio_stage_terms.py b/test/services/test_audio_stage_terms.py
new file mode 100644
index 00000000..a7b0dc23
--- /dev/null
+++ b/test/services/test_audio_stage_terms.py
@@ -0,0 +1,30 @@
+from unittest.mock import patch
+import pytest
+from app.models.schema import VideoParams
+from app.services import task as tm
+from app.services.state import MemoryState
+
[email protected]("stage", ["audio", "subtitle"])
+def test_narration_stages_do_not_depend_on_visual_search_terms(stage):
+ params = VideoParams(video_subject="Coffee", video_script="A supplied script", video_source="pexels")
+ state = MemoryState()
+ with (patch.object(tm.sm, "state", state),
+ patch.object(tm.utils, "check_ffmpeg_ready", return_value=True),
+ patch.object(tm, "generate_terms", side_effect=RuntimeError("search terms unavailable")) as terms,
+ patch.object(tm, "save_script_data") as save,
+ patch.object(tm, "generate_audio", return_value=("audio.mp3", 3.0, object())) as audio,
+ patch.object(tm, "generate_subtitle", return_value="captions.srt")):
+ result = tm.start("task-1", params, stop_at=stage)
+ terms.assert_not_called()
+ audio.assert_called_once()
+ assert state.get_task("task-1")["state"] == tm.const.TASK_STATE_COMPLETE
+ assert result == ({"audio_file": "audio.mp3", "audio_duration": 3.0} if stage == "audio" else {"subtitle_path": "captions.srt"})
+ assert save.call_args.args[2] == ""
+
+def test_terms_stage_still_generates_visual_search_terms():
+ params = VideoParams(video_subject="Coffee", video_script="A supplied script")
+ with (patch.object(tm.sm, "state", MemoryState()), patch.object(tm, "save_script_data"),
+ patch.object(tm, "generate_terms", return_value=["coffee beans"]) as terms):
+ result = tm.start("task-1", params, stop_at="terms")
+ terms.assert_called_once()
+ assert result["terms"] == ["coffee beans"]