xiaoxia-saas/tests/unit/test_extract_from_douyin_errors.py

"""验证 extract-from-douyin 在各种失败场景返回正确的 HTTP 状态码（绝不能 500）

新版架构：douyin_resolver 多源轮询 → MediaKit ASR → 本地下载+ASR → desc 兜底。
所有外部依赖（resolver、MediaKit、transcribe_to_text）均通过 mock 隔离。
"""

from __future__ import annotations

from unittest import mock

import pytest
from app.auth import AuthenticatedUser
from app.services.douyin_resolver import ResolveResult
from fastapi import HTTPException, status


class _FakeUser:
    id = "u-test"
    is_member = False
    member_type = None


@pytest.fixture
def fake_user():
    return AuthenticatedUser(user=_FakeUser())


def _import_target():
    from app.api.routes import scripts_ai

    return scripts_ai


def _fake_mk_available(text="识别成功的文案", duration=5.0):
    """Mock MediaKitClient 可用并返回指定 ASR 结果。"""
    fake_mk = mock.MagicMock()
    fake_mk.is_available = True
    fake_mk.asr_submit.return_value = "tk1"
    fake_mk.asr_poll.return_value = (text, duration)
    return mock.patch("app.api.routes.scripts_ai.get_mediakit_client", return_value=fake_mk)


def _fake_mk_unavailable():
    """Mock MediaKitClient 不可用，强制走下载+本地 ASR 路径。"""
    fake_mk = mock.MagicMock()
    fake_mk.is_available = False
    return mock.patch("app.api.routes.scripts_ai.get_mediakit_client", return_value=fake_mk)


def _fake_resolver_success(video_url="https://example.com/direct.mp4", desc="", source="app_feed"):
    """Mock resolver 返回成功。"""
    result = ResolveResult(video_url=video_url, desc=desc, source=source)
    return mock.patch("app.api.routes.scripts_ai.resolve_douyin_video", return_value=result)


def _fake_resolver_image(desc="图文文案内容", source="app_feed_image"):
    """Mock resolver 返回图文视频（video_url 为空）。"""
    result = ResolveResult(video_url="", desc=desc, source=source)
    return mock.patch("app.api.routes.scripts_ai.resolve_douyin_video", return_value=result)


def _fake_resolver_failure():
    """Mock resolver 所有源均失败，返回 None。"""
    return mock.patch("app.api.routes.scripts_ai.resolve_douyin_video", return_value=None)


# ── 解析阶段失败 ──────────────────────────────────────────────────


def test_resolver_all_fail_returns_503_parse(fake_user):
    """所有解析源均失败 → 503 解析失败。"""
    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/test123/")

    with _fake_resolver_failure():
        with pytest.raises(HTTPException) as exc:
            scripts_ai.extract_from_douyin(request=body, current_user=fake_user, db=mock.MagicMock())
        assert exc.value.status_code == status.HTTP_503_SERVICE_UNAVAILABLE
        assert "解析失败" in exc.value.detail or "链接" in exc.value.detail


# ── 图文视频（无需ASR） ─────────────────────────────────────────


def test_image_post_returns_desc_directly(fake_user):
    """图文视频（resolver返回空video_url有desc）→ 直接返回 desc，不走 ASR。"""
    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/abcdeFG/")

    with _fake_resolver_image(desc="这是图文文案"):
        resp = scripts_ai.extract_from_douyin(request=body, current_user=fake_user, db=mock.MagicMock())
        assert resp.text == "这是图文文案"
        assert resp.duration_seconds == 0.0


# ── MediaKit ASR 成功路径 ────────────────────────────────────────


def test_mediakit_asr_success(fake_user):
    """正常流程：resolver 成功 + MediaKit ASR 成功 → 返回文本。"""
    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/xxxxx/ 快来看看！")

    with _fake_resolver_success(desc="Feed标题"):
        with _fake_mk_available(text="这是MediaKit识别的文案", duration=12.5):
            result = scripts_ai.extract_from_douyin(
                request=body, current_user=fake_user, db=mock.MagicMock()
            )
    assert result.text == "这是MediaKit识别的文案"
    assert result.duration_seconds == 12.5


# ── 分享文本含前后文字 ──────────────────────────────────────────


def test_share_text_input_extracts_url_correctly(fake_user):
    """分享文本（含前后说明文字）应能正确提取 URL。"""
    scripts_ai = _import_target()
    share_text = "这个视频太搞笑了 https://v.douyin.com/abcdeFG/ 快来看看！#搞笑 #日常"
    body = scripts_ai.ExtractFromDouyinRequest(url=share_text)

    with _fake_resolver_success():
        with _fake_mk_available(text="识别成功的文案", duration=5.0) as mk_mock:
            resp = scripts_ai.extract_from_douyin(
                request=body, current_user=fake_user, db=mock.MagicMock()
            )
    assert resp.source_url == "https://v.douyin.com/abcdeFG/"
    assert resp.text == "识别成功的文案"
    assert resp.duration_seconds == 5.0


# ── 非抖音链接 ─────────────────────────────────────────────────


def test_non_douyin_share_text_returns_400(fake_user):
    """粘贴非抖音分享链接 → 400。"""
    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="看看这个 https://www.bilibili.com/video/BV1xx 哈哈哈")

    with pytest.raises(HTTPException) as exc:
        scripts_ai.extract_from_douyin(request=body, current_user=fake_user, db=mock.MagicMock())
    assert exc.value.status_code == 400


# ── ASR 空结果 → desc 兜底 ────────────────────────────────────


def test_asr_empty_falls_back_to_desc(fake_user):
    """MediaKit 和本地 ASR 都返回空文本（无旁白视频）→ 使用 desc 兜底。"""
    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/xxxxx/")

    # MediaKit 返回空文本
    fake_mk = mock.MagicMock()
    fake_mk.is_available = True
    fake_mk.asr_submit.return_value = "tk1"
    fake_mk.asr_poll.return_value = ("", 4.5)

    # 本地下载+ASR 也返回空（通过mock _direct_url_download_and_local_asr）
    with _fake_resolver_success(desc="Feed描述文案"):
        with mock.patch("app.api.routes.scripts_ai.get_mediakit_client", return_value=fake_mk):
            with mock.patch(
                "app.api.routes.scripts_ai._direct_url_download_and_local_asr",
                return_value=("", 0.0),
            ):
                resp = scripts_ai.extract_from_douyin(
                    request=body, current_user=fake_user, db=mock.MagicMock()
                )
    assert resp.text == "Feed描述文案"


# ── ASR 未配置 → 503 ─────────────────────────────────────────


def test_asr_not_configured_returns_503(fake_user):
    """本地 ASR 未配置 → 503。"""
    from app.services.script_asr_service import ASRNotConfiguredError

    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/xxxxx/")

    with _fake_resolver_success():
        with _fake_mk_unavailable():
            with mock.patch(
                "app.api.routes.scripts_ai._direct_url_download_and_local_asr",
                side_effect=HTTPException(status_code=503, detail="ASR未配置"),
            ):
                with pytest.raises(HTTPException) as exc:
                    scripts_ai.extract_from_douyin(
                        request=body, current_user=fake_user, db=mock.MagicMock()
                    )
                assert exc.value.status_code == status.HTTP_503_SERVICE_UNAVAILABLE


# ── ASR 转写失败 → 502 ────────────────────────────────────────


def test_asr_transcription_failure_returns_502(fake_user):
    """ASR 转写异常 → 502（被 _direct_url_download_and_local_asr 包装）。"""
    from app.services.script_asr_service import ASRTranscriptionError

    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/xxxxx/")

    # MediaKit 失败
    fake_mk = mock.MagicMock()
    fake_mk.is_available = True
    from app.services.mediakit_client import MediaKitError

    fake_mk.asr_submit.side_effect = MediaKitError("ASR failed", code="TaskFailed")

    with _fake_resolver_success(desc=""):
        with mock.patch("app.api.routes.scripts_ai.get_mediakit_client", return_value=fake_mk):
            with mock.patch(
                "app.api.routes.scripts_ai._direct_url_download_and_local_asr",
                side_effect=HTTPException(status_code=502, detail="语音识别失败"),
            ):
                with pytest.raises(HTTPException) as exc:
                    scripts_ai.extract_from_douyin(
                        request=body, current_user=fake_user, db=mock.MagicMock()
                    )
                assert exc.value.status_code == status.HTTP_502_BAD_GATEWAY


# ── 下载超时 → 504 ────────────────────────────────────────────


def test_download_timeout_returns_504(fake_user):
    """视频下载超时 → 504。"""
    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/xxxxx/")

    with _fake_resolver_success():
        with _fake_mk_unavailable():
            with mock.patch(
                "app.api.routes.scripts_ai._direct_url_download_and_local_asr",
                side_effect=HTTPException(status_code=504, detail="视频下载超时"),
            ):
                with pytest.raises(HTTPException) as exc:
                    scripts_ai.extract_from_douyin(
                        request=body, current_user=fake_user, db=mock.MagicMock()
                    )
                assert exc.value.status_code == status.HTTP_504_GATEWAY_TIMEOUT