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）"""

from __future__ import annotations

import sys
import types
from unittest import mock

import pytest
from app.auth import AuthenticatedUser
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())


class _FakeYDLBase:
    """通用假 yt-dlp 基类（支持上下文管理器 with 语法）"""

    extract_info_result = None
    extract_info_raises = None
    prepare_filename_result = "/tmp/fake.mp4"

    def __init__(self, *a, **kw):
        pass

    def __enter__(self):
        return self

    def __exit__(self, *a):
        return False

    def extract_info(self, url, download=True):
        if self.__class__.extract_info_raises:
            raise self.__class__.extract_info_raises
        return self.__class__.extract_info_result

    def prepare_filename(self, info):
        return self.__class__.prepare_filename_result


def _install_fake_ytdlp(fake_ydl_class, *, download_error_cls=None):
    """把假 yt-dlp 注入 sys.modules，函数内 import yt_dlp 会拿到我们的假版本"""
    fake_mod = types.ModuleType("yt_dlp")
    fake_mod.YoutubeDL = fake_ydl_class
    if download_error_cls is None:
        download_error_cls = type("DownloadError", (Exception,), {})
    fake_mod.DownloadError = download_error_cls
    utils_mod = types.ModuleType("yt_dlp.utils")
    utils_mod.DownloadError = download_error_cls
    fake_mod.utils = utils_mod
    sys.modules["yt_dlp"] = fake_mod
    sys.modules["yt_dlp.utils"] = utils_mod
    return fake_mod


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

    return scripts_ai


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 test_download_http404_returns_400_not_500(fake_user):
    """无效短链 / 视频 404 → 应返回 400 业务错误，不能 500"""
    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/test123/")

    class DownloadError(Exception):
        pass

    class FailingYDL(_FakeYDLBase):
        extract_info_raises = DownloadError("ERROR: Unable to download webpage: HTTP Error 404: Not Found")

    _install_fake_ytdlp(FailingYDL, download_error_cls=DownloadError)

    with _fake_mk_unavailable():
        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_400_BAD_REQUEST
        ), f"应为400，实际 {exc.value.status_code}: {exc.value.detail}"
        assert "无法解析" in exc.value.detail or "抖音" in exc.value.detail


def test_download_network_error_returns_502_not_500(fake_user):
    """网络错误 / 上游异常 → 502，不能 500"""
    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/abc/")

    class DownloadError(Exception):
        pass

    class NetErrYDL(_FakeYDLBase):
        # 路径A（元信息解析）会吞异常返回 None；路径B（下载）抛网络错误
        @staticmethod
        def _raise():
            raise DownloadError("ERROR: Connection reset by peer")

        def extract_info(self, url, download=True):
            # 元信息探测返回 None（拿不到直链），下载时再抛
            if not download:
                return None
            self._raise()

    _install_fake_ytdlp(NetErrYDL, download_error_cls=DownloadError)

    with _fake_mk_unavailable():
        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


def test_info_none_returns_400(fake_user):
    """yt-dlp 返回 None info → 400"""
    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/abc/")

    class NoneInfoYDL(_FakeYDLBase):
        def extract_info(self, url, download=True):
            # 元信息探测返回 None；下载也返回 None
            return None

    _install_fake_ytdlp(NoneInfoYDL)

    with _fake_mk_unavailable():
        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_400_BAD_REQUEST


def test_asr_not_configured_returns_503(fake_user):
    scripts_ai = _import_target()
    from app.services.script_asr_service import ASRNotConfiguredError

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

    class OkYDL(_FakeYDLBase):
        def extract_info(self, url, download=True):
            # 元信息返回 None（不走 MediaKit）；下载返回正常 info
            if not download:
                return None
            return {"id": "x", "duration": 10, "title": "t"}

    _install_fake_ytdlp(OkYDL)
    with (
        _fake_mk_unavailable(),
        mock.patch.object(scripts_ai.os.path, "isfile", return_value=True),
        mock.patch.object(scripts_ai.os.path, "getsize", return_value=1024),
        mock.patch.object(scripts_ai, "transcribe_to_text", side_effect=ASRNotConfiguredError("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


def test_asr_failure_returns_502(fake_user):
    scripts_ai = _import_target()
    from app.services.script_asr_service import ASRTranscriptionError

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

    class OkYDL(_FakeYDLBase):
        def extract_info(self, url, download=True):
            if not download:
                return None
            return {"id": "x", "duration": 10, "title": "t"}

    _install_fake_ytdlp(OkYDL)
    with (
        _fake_mk_unavailable(),
        mock.patch.object(scripts_ai.os.path, "isfile", return_value=True),
        mock.patch.object(scripts_ai.os.path, "getsize", return_value=1024),
        mock.patch.object(scripts_ai, "transcribe_to_text", side_effect=ASRTranscriptionError("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


def test_asr_unexpected_error_returns_502_not_500(fake_user):
    """ASR 抛未预期异常也应被兜住，不能 500"""
    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/abc/")

    class OkYDL(_FakeYDLBase):
        def extract_info(self, url, download=True):
            if not download:
                return None
            return {"id": "x", "duration": 10, "title": "t"}

    _install_fake_ytdlp(OkYDL)
    with (
        _fake_mk_unavailable(),
        mock.patch.object(scripts_ai.os.path, "isfile", return_value=True),
        mock.patch.object(scripts_ai.os.path, "getsize", return_value=1024),
        mock.patch.object(scripts_ai, "transcribe_to_text", side_effect=RuntimeError("ffmpeg crashed")),
    ):
        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, f"应为502，实际 {exc.value.status_code}"


def test_missing_downloaded_file_returns_502_not_500(fake_user):
    """yt-dlp 返回 info 但文件未落地（isfile False）→ 502"""
    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/abc/")

    class OkYDL(_FakeYDLBase):
        def extract_info(self, url, download=True):
            if not download:
                return None
            return {"id": "x", "duration": 10, "title": "t"}

    _install_fake_ytdlp(OkYDL)
    with (
        _fake_mk_unavailable(),
        mock.patch.object(scripts_ai.os.path, "isfile", return_value=False),
    ):
        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 != 500
        assert "下载异常" in exc.value.detail or "文件" in exc.value.detail


def test_any_unexpected_error_does_not_return_500_raw(fake_user):
    """兜底：prepare_filename 抛未预期异常也应被捕获，返回500 code但含业务detail"""
    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/abc/")

    class BuggyYDL(_FakeYDLBase):
        def extract_info(self, url, download=True):
            if not download:
                return None
            return {"id": "x", "duration": "not_a_number", "title": "t"}

        def prepare_filename(self, info):
            raise RuntimeError("some internal bug")

    _install_fake_ytdlp(BuggyYDL)
    with _fake_mk_unavailable():
        with pytest.raises(HTTPException) as exc:
            scripts_ai.extract_from_douyin(request=body, current_user=fake_user, db=mock.MagicMock())
        assert "抖音" in exc.value.detail or "失败" in exc.value.detail or exc.value.status_code != 500


# ── cookies 相关测试 ──────────────────────────────────────────────


def test_cookies_error_returns_503_friendly_message(fake_user):
    """cookies 缺失/过期 → 返回 503 + 友好文案，不暴露原始错误"""
    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/test123/")

    class DownloadError(Exception):
        pass

    class CookiesYDL(_FakeYDLBase):
        def extract_info(self, url, download=True):
            raise DownloadError(
                "ERROR: [Douyin] 7623712911260650802: Fresh cookies (not necessarily logged in) are needed"
            )

    _install_fake_ytdlp(CookiesYDL, download_error_cls=DownloadError)

    with _fake_mk_unavailable():
        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, f"应为503，实际 {exc.value.status_code}"
        assert (
            "暂时不可用" in exc.value.detail or "稍后重试" in exc.value.detail
        ), f"应有友好提示，实际: {exc.value.detail}"
        assert "Fresh cookies" not in exc.value.detail


def test_cookies_error_in_generic_except_also_returns_503(fake_user):
    """cookies 错误绕过 DownloadError 时，兜底异常分支也应识别并返回 503"""
    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/abc/")

    class CookieBugYDL(_FakeYDLBase):
        def extract_info(self, url, download=True):
            raise RuntimeError("Fresh cookies are needed to access this video")

    _install_fake_ytdlp(CookieBugYDL)

    with _fake_mk_unavailable():
        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


def test_ydl_opts_includes_cookiefile_when_file_exists(fake_user):
    """cookies 文件存在时，ydl_opts 应包含 cookiefile 指向该路径（在下载分支）"""
    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/abc/")

    captured_opts_download = {}

    class CaptureOptsYDL(_FakeYDLBase):
        def __init__(self, opts):
            # 下载分支会触发 download=True；元信息探测 download=False
            # 元信息也会传 cookiefile，但我们只在下载分支记录（更接近真实）
            super().__init__()
            self._opts = opts
            # 总是记录最后一次的 opts，方便断言
            captured_opts_download.clear()
            captured_opts_download.update(opts)

        def extract_info(self, url, download=True):
            if not download:
                return None  # 元信息失败，走下载分支
            return {"id": "x", "duration": 5, "title": "t"}

    _install_fake_ytdlp(CaptureOptsYDL)

    with (
        _fake_mk_unavailable(),
        mock.patch.object(scripts_ai, "_resolve_cookies_file", return_value="/tmp/fake_cookies.txt"),
        mock.patch.object(scripts_ai.os.path, "isfile", return_value=True),
        mock.patch.object(scripts_ai.os.path, "getsize", return_value=1024),
        mock.patch.object(scripts_ai, "transcribe_to_text", return_value="ok"),
    ):
        scripts_ai.extract_from_douyin(request=body, current_user=fake_user, db=mock.MagicMock())

    assert (
        captured_opts_download.get("cookiefile") == "/tmp/fake_cookies.txt"
    ), f"cookiefile 应被设置，opts={captured_opts_download}"


def test_ydl_opts_no_cookiefile_when_file_missing(fake_user):
    """cookies 文件不存在时，ydl_opts 不应包含 cookiefile 键"""
    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/abc/")

    captured_opts_download = {}

    class CaptureOptsYDL(_FakeYDLBase):
        def __init__(self, opts):
            super().__init__()
            captured_opts_download.clear()
            captured_opts_download.update(opts)

        def extract_info(self, url, download=True):
            if not download:
                return None
            return {"id": "x", "duration": 5, "title": "t"}

    _install_fake_ytdlp(CaptureOptsYDL)

    with (
        _fake_mk_unavailable(),
        mock.patch.object(scripts_ai, "_resolve_cookies_file", return_value=None),
        mock.patch.object(scripts_ai.os.path, "isfile", return_value=True),
        mock.patch.object(scripts_ai.os.path, "getsize", return_value=1024),
        mock.patch.object(scripts_ai, "transcribe_to_text", return_value="ok"),
    ):
        scripts_ai.extract_from_douyin(request=body, current_user=fake_user, db=mock.MagicMock())

    assert (
        "cookiefile" not in captured_opts_download
    ), f"cookies 文件缺失时不应设置 cookiefile，opts={captured_opts_download}"


def test_generic_download_error_hides_raw_message(fake_user):
    """非 cookies 非 404 的通用下载错误 → 502，且不暴露 yt-dlp 原始错误文本"""
    scripts_ai = _import_target()
    body = scripts_ai.ExtractFromDouyinRequest(url="https://v.douyin.com/abc/")

    class DownloadError(Exception):
        pass

    class GenErrYDL(_FakeYDLBase):
        def extract_info(self, url, download=True):
            raise DownloadError("ERROR: some internal yt-dlp weird failure with trace")

    _install_fake_ytdlp(GenErrYDL, download_error_cls=DownloadError)

    with _fake_mk_unavailable():
        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
        assert "下载失败" in exc.value.detail
        assert "weird failure" not in exc.value.detail, "不应暴露 yt-dlp 内部错误文本"


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)

    class OkYDL(_FakeYDLBase):
        def extract_info(self, url, download=True):
            if not download:
                return {"url": "https://example.com/direct.mp4", "duration": 5}
            return {"id": "x", "duration": 5, "title": "t"}

    _install_fake_ytdlp(OkYDL)
    fake_mk = mock.MagicMock()
    fake_mk.is_available = True
    fake_mk.asr_submit.return_value = "tk1"
    fake_mk.asr_poll.return_value = ("识别成功的文案", 5.0)
    with (
        mock.patch("app.api.routes.scripts_ai.get_mediakit_client", return_value=fake_mk),
        mock.patch.object(scripts_ai, "_ytdlp_extract_video_url", return_value=("https://example.com/direct.mp4", 5.0)),
    ):
        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


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 _fake_mk_unavailable():
        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