biliup-next/tests/test_song_detect_providers.py

from __future__ import annotations

import json
import os
import tempfile
import unittest
from pathlib import Path
from unittest.mock import patch

from biliup_next.core.models import Artifact, Task, utc_now_iso
from biliup_next.infra.adapters.codex_cli import CodexCliAdapter
from biliup_next.modules.song_detect.providers.codex import CodexSongDetector
from biliup_next.modules.song_detect.providers.qwen_cli import QwenCliSongDetector


class FakeQwenCliAdapter:
    def __init__(self, returncode: int = 0) -> None:
        self.returncode = returncode
        self.last_qwen_cmd: str | None = None

    def run_song_detect(self, *, qwen_cmd: str, work_dir: Path, prompt: str):  # noqa: ANN001
        self.last_qwen_cmd = qwen_cmd
        songs_json_path = work_dir / "songs.json"
        songs_json_path.write_text(
            json.dumps(
                {
                    "songs": [
                        {
                            "start": "00:01:23,000",
                            "end": "00:03:45,000",
                            "title": "测试歌曲",
                            "artist": "测试歌手",
                            "confidence": 0.93,
                            "evidence": "歌词命中",
                        }
                    ]
                },
                ensure_ascii=False,
            ),
            encoding="utf-8",
        )
        return type("Result", (), {"returncode": self.returncode, "stdout": "ok", "stderr": ""})()


class FakeCodexCliAdapter:
    def __init__(self, returncode: int = 0) -> None:
        self.returncode = returncode

    def run_song_detect(self, *, codex_cmd: str, work_dir: Path, prompt: str):  # noqa: ANN001
        songs_json_path = work_dir / "songs.json"
        songs_json_path.write_text(
            json.dumps(
                {
                    "songs": [
                        {
                            "start": "00:01:23,000",
                            "end": "00:03:45,000",
                            "title": "测试歌曲",
                            "artist": "测试歌手",
                            "confidence": 0.93,
                            "evidence": "歌词命中",
                        }
                    ]
                },
                ensure_ascii=False,
            ),
            encoding="utf-8",
        )
        return type("Result", (), {"returncode": self.returncode, "stdout": "codex stdout", "stderr": "codex stderr"})()


class SongDetectProviderTests(unittest.TestCase):
    def test_qwen_cli_provider_generates_json_and_txt_artifacts(self) -> None:
        with tempfile.TemporaryDirectory() as tmpdir:
            work_dir = Path(tmpdir)
            subtitle_path = work_dir / "subtitle.srt"
            subtitle_path.write_text("1\n00:00:00,000 --> 00:00:03,000\n测试字幕\n", encoding="utf-8")
            provider = QwenCliSongDetector(adapter=FakeQwenCliAdapter())

            task = Task(
                id="task-1",
                source_type="local_file",
                source_path=str(work_dir / "video.mp4"),
                title="task-1",
                status="transcribed",
                created_at=utc_now_iso(),
                updated_at=utc_now_iso(),
            )
            subtitle = Artifact(
                id=None,
                task_id=task.id,
                artifact_type="subtitle_srt",
                path=str(subtitle_path),
                metadata_json=None,
                created_at=utc_now_iso(),
            )

            songs_json, songs_txt = provider.detect(task, subtitle, {"qwen_cmd": "qwen"})

            self.assertEqual(json.loads(songs_json.metadata_json)["provider"], "qwen_cli")
            self.assertEqual(json.loads(songs_txt.metadata_json)["provider"], "qwen_cli")
            self.assertTrue(Path(songs_json.path).exists())
            self.assertTrue(Path(songs_txt.path).exists())
            self.assertIn("测试歌曲", Path(songs_txt.path).read_text(encoding="utf-8"))

    def test_codex_provider_writes_execution_output_to_session_log(self) -> None:
        with tempfile.TemporaryDirectory() as tmpdir:
            work_dir = Path(tmpdir)
            subtitle_path = work_dir / "subtitle.srt"
            subtitle_path.write_text("1\n00:00:00,000 --> 00:00:03,000\n测试字幕\n", encoding="utf-8")
            provider = CodexSongDetector(adapter=FakeCodexCliAdapter())

            task = Task(
                id="task-1",
                source_type="local_file",
                source_path=str(work_dir / "video.mp4"),
                title="task-1",
                status="transcribed",
                created_at=utc_now_iso(),
                updated_at=utc_now_iso(),
            )
            subtitle = Artifact(
                id=None,
                task_id=task.id,
                artifact_type="subtitle_srt",
                path=str(subtitle_path),
                metadata_json=None,
                created_at=utc_now_iso(),
            )

            songs_json, songs_txt = provider.detect(task, subtitle, {"codex_cmd": "codex"})

            json_metadata = json.loads(songs_json.metadata_json)
            txt_metadata = json.loads(songs_txt.metadata_json)
            self.assertEqual(json_metadata["provider"], "codex")
            self.assertEqual(txt_metadata["provider"], "codex")
            self.assertNotIn("execution", json_metadata)
            codex_log = work_dir / "codex.log"
            self.assertTrue(codex_log.exists())
            log_text = codex_log.read_text(encoding="utf-8")
            self.assertIn("returncode: 0", log_text)
            self.assertIn("codex stdout", log_text)
            self.assertIn("codex stderr", log_text)

    def test_codex_cli_adapter_disables_inner_sandbox_and_normalizes_proxy_env(self) -> None:
        with tempfile.TemporaryDirectory() as tmpdir:
            calls = []

            def fake_run(cmd, **kwargs):  # noqa: ANN001
                calls.append((cmd, kwargs))
                return type("Result", (), {"returncode": 0, "stdout": "", "stderr": ""})()

            with patch.dict(os.environ, {"HTTPS_PROXY": "192.168.1.100:7897"}, clear=True):
                with patch("subprocess.run", side_effect=fake_run):
                    CodexCliAdapter().run_song_detect(
                        codex_cmd="codex",
                        work_dir=Path(tmpdir),
                        prompt="detect songs",
                    )

            cmd, kwargs = calls[0]
            self.assertIn("--dangerously-bypass-approvals-and-sandbox", cmd)
            self.assertNotIn("--full-auto", cmd)
            self.assertNotIn("workspace-write", cmd)
            self.assertEqual(kwargs["env"]["HTTPS_PROXY"], "http://192.168.1.100:7897")


if __name__ == "__main__":
    unittest.main()