| 1 | import asyncio |
| 2 | |
| 3 | from config import ConfigLoader |
| 4 | from core.transcript_manager import TranscriptManager |
| 5 | from storage import FileManager |
| 6 | |
| 7 | |
| 8 | class _FakeDatabase: |
| 9 | def __init__(self): |
| 10 | self.rows = [] |
| 11 | |
| 12 | async def upsert_transcript_job(self, payload): |
| 13 | self.rows.append(payload) |
| 14 | |
| 15 | |
| 16 | def test_transcript_default_disabled(): |
| 17 | loader = ConfigLoader() |
| 18 | transcript_cfg = loader.get("transcript", {}) |
| 19 | |
| 20 | assert transcript_cfg.get("enabled") is False |
| 21 | assert transcript_cfg.get("api_url") == "https://api.openai.com/v1/audio/transcriptions" |
| 22 | |
| 23 | |
| 24 | def test_transcript_skip_when_missing_api_key(tmp_path): |
| 25 | config = ConfigLoader() |
| 26 | config.update( |
| 27 | transcript={ |
| 28 | "enabled": True, |
| 29 | "api_key_env": "OPENAI_API_KEY", |
| 30 | "api_key": "", |
| 31 | "output_dir": "", |
| 32 | "response_formats": ["txt", "json"], |
| 33 | } |
| 34 | ) |
| 35 | |
| 36 | file_manager = FileManager(str(tmp_path / "Downloaded")) |
| 37 | database = _FakeDatabase() |
| 38 | manager = TranscriptManager(config, file_manager, database=database) |
| 39 | |
| 40 | video_path = tmp_path / "Downloaded" / "author" / "post" / "demo.mp4" |
| 41 | video_path.parent.mkdir(parents=True, exist_ok=True) |
| 42 | video_path.write_bytes(b"video") |
| 43 | |
| 44 | result = asyncio.run(manager.process_video(video_path, aweme_id="123")) |
| 45 | |
| 46 | assert result["status"] == "skipped" |
| 47 | assert result["reason"] == "missing_api_key" |
| 48 | assert database.rows[-1]["status"] == "skipped" |
| 49 | assert database.rows[-1]["skip_reason"] == "missing_api_key" |
| 50 | |
| 51 | |
| 52 | def test_transcript_output_dir_defaults_to_video_dir(tmp_path): |
| 53 | config = ConfigLoader() |
| 54 | config.update(transcript={"enabled": True, "output_dir": ""}) |
| 55 | file_manager = FileManager(str(tmp_path / "Downloaded")) |
| 56 | manager = TranscriptManager(config, file_manager, database=None) |
| 57 | |
| 58 | video_path = tmp_path / "Downloaded" / "a" / "post" / "x.mp4" |
| 59 | video_path.parent.mkdir(parents=True, exist_ok=True) |
| 60 | video_path.write_bytes(b"video") |
| 61 | |
| 62 | resolved = manager.resolve_output_dir(video_path) |
| 63 | assert resolved == video_path.parent |
| 64 | |
| 65 | |
| 66 | def test_transcript_output_dir_mirrors_video_tree(tmp_path): |
| 67 | config = ConfigLoader() |
| 68 | output_root = tmp_path / "Transcripts" |
| 69 | config.update(transcript={"enabled": True, "output_dir": str(output_root)}) |
| 70 | file_manager = FileManager(str(tmp_path / "Downloaded")) |
| 71 | manager = TranscriptManager(config, file_manager, database=None) |
| 72 | |
| 73 | video_path = tmp_path / "Downloaded" / "a" / "post" / "2026-02-18_demo" / "x.mp4" |
| 74 | video_path.parent.mkdir(parents=True, exist_ok=True) |
| 75 | video_path.write_bytes(b"video") |
| 76 | |
| 77 | resolved = manager.resolve_output_dir(video_path) |
| 78 | expected = output_root / "a" / "post" / "2026-02-18_demo" |
| 79 | assert resolved == expected |
| 80 | |
| 81 | |
| 82 | def test_transcript_file_names(tmp_path): |
| 83 | config = ConfigLoader() |
| 84 | config.update(transcript={"enabled": True, "output_dir": ""}) |
| 85 | file_manager = FileManager(str(tmp_path / "Downloaded")) |
| 86 | manager = TranscriptManager(config, file_manager, database=None) |
| 87 | |
| 88 | video_path = tmp_path / "Downloaded" / "a" / "post" / "demo.mp4" |
| 89 | video_path.parent.mkdir(parents=True, exist_ok=True) |
| 90 | video_path.write_bytes(b"video") |
| 91 | |
| 92 | text_path, json_path = manager.build_output_paths(video_path) |
| 93 | assert text_path.name == "demo.transcript.txt" |
| 94 | assert json_path.name == "demo.transcript.json" |
| 95 |