mirror of
https://github.com/jxxghp/MoviePilot.git
synced 2026-08-19 05:03:57 +08:00
423 lines
16 KiB
Python
423 lines
16 KiB
Python
"""
|
||
订阅的写入路径：app/application/subscribe.py 的 add_subscribe / async_add_subscribe。
|
||
|
||
这两个函数是订阅表的唯一写入口，把 MediaInfo / MusicInfo 翻译成一行订阅：
|
||
标题、年份、类型、海报背景、评分简介、剧集组、音乐实体与曲目数，再叠上
|
||
持久化类型强转（布尔开关转整型、年份转字符串）。字段映射错了不会报错，
|
||
只会让订阅静静地记错——而搜索、洗版、完成判定、去重全都读这张表。
|
||
|
||
因此这里断言的是「落库后每个字段的实际值」，不是「调用了什么」：
|
||
同目录的 test_subscribe_oper.py 用替身钉的是查重语义（谁被查、查几次），
|
||
证明不了写进去的到底是什么。两者互补，缺一不可。
|
||
|
||
唯一的例外是**持久化类型强转**（年份转字符串、布尔开关转整型）。这两步是
|
||
为 PostgreSQL 的严格类型检查而存在的，而测试库是 SQLite——SQLite 的类型
|
||
亲和会在写入时把 ``2003`` 悄悄转成 ``'2003'``、把 ``True`` 转成 ``1``，
|
||
落库后的值对「写入路径自己有没有转」完全无感（已用变异验证确认：删掉
|
||
``_normalize_year`` 后按落库值断言的用例全部照过）。所以这两类契约必须在
|
||
建模那一刻、即 ``Subscribe(**kwargs)`` 的入参上断言，见 ``payloads`` 夹具。
|
||
强转跟着订阅表的列走，因此仍留在 ``app/db/oper/subscribe.py``，夹具也就
|
||
仍然钉在那个模块的 ``Subscribe`` 上。
|
||
|
||
同步与异步两条链路是两份逐字复制的实现，任何一条改了另一条没跟上都属于
|
||
真实缺陷，故每个字段契约都在两条链路上各断言一次。
|
||
"""
|
||
import asyncio
|
||
|
||
import pytest
|
||
|
||
from app.application.subscribe import add_subscribe, async_add_subscribe
|
||
from app.db.models.subscribe import Subscribe
|
||
from app.db.oper.subscribe import SubscribeOper
|
||
from app.domain.context import MediaInfo, MusicInfo
|
||
from app.schemas.types import MediaSource, MediaType
|
||
|
||
|
||
@pytest.fixture(autouse=True)
|
||
def _track(db):
|
||
"""把订阅表纳入用例级回收。"""
|
||
db.watermark(Subscribe)
|
||
|
||
|
||
def _media_id(tag: str) -> str:
|
||
"""给每个用例分配独立媒体 ID，避免共用测试库时互相去重。"""
|
||
return f"wp-{tag}"
|
||
|
||
|
||
def _mediainfo(media_id: str, title: str = "识别标题",
|
||
mtype: MediaType = MediaType.TV, year: str = "2026",
|
||
episode_group: str = None) -> MediaInfo:
|
||
"""构造带完整展示字段的识别结果。"""
|
||
media = MediaInfo()
|
||
media.type = mtype
|
||
media.title = title
|
||
media.year = year
|
||
media.media_source = MediaSource.TMDB
|
||
media.media_id = media_id
|
||
media.episode_group = episode_group
|
||
media.vote_average = 8.5
|
||
media.overview = "测试简介"
|
||
media.poster_path = "https://image.tmdb.org/t/p/original/poster.jpg"
|
||
media.backdrop_path = "https://image.tmdb.org/t/p/original/backdrop.jpg"
|
||
return media
|
||
|
||
|
||
def _musicinfo(media_id: str, music_type: str, **kwargs) -> MusicInfo:
|
||
"""构造音乐订阅所需的标准音乐信息。"""
|
||
return MusicInfo(media_source=MediaSource.MusicBrainz, media_id=media_id,
|
||
music_type=music_type, **kwargs)
|
||
|
||
|
||
def _add(oper: SubscribeOper, is_async: bool, **kwargs):
|
||
"""按链路分派到同步或异步新增，让同一份字段契约跑两遍。"""
|
||
if is_async:
|
||
return asyncio.run(async_add_subscribe(subscribe_oper=oper, **kwargs))
|
||
return add_subscribe(subscribe_oper=oper, **kwargs)
|
||
|
||
|
||
def _row(db, subscribe_id: int) -> Subscribe:
|
||
"""按主键读回落库的订阅行。"""
|
||
db.session.expire_all()
|
||
return Subscribe.get(db.session, subscribe_id)
|
||
|
||
|
||
class _SubscribeSpy:
|
||
"""记录建模入参并转交真实模型，保持写入路径仍然真的落库。"""
|
||
|
||
def __init__(self, recorded: list):
|
||
self._recorded = recorded
|
||
|
||
def __call__(self, **kwargs):
|
||
"""截获 ``Subscribe(**kwargs)`` 的入参后构造真实模型实例。"""
|
||
self._recorded.append(dict(kwargs))
|
||
return Subscribe(**kwargs)
|
||
|
||
def __getattr__(self, name):
|
||
"""查重用的类方法（exists / async_exists 等）原样透传给真实模型。"""
|
||
return getattr(Subscribe, name)
|
||
|
||
|
||
@pytest.fixture
|
||
def payloads(monkeypatch):
|
||
"""
|
||
捕获写入路径建模时的原始 kwargs，即落库**前**的值与类型。
|
||
|
||
只用于持久化类型强转这一类契约：SQLite 的类型亲和会在写入时替写入路径
|
||
把类型「修好」，落库后的值证明不了转换真的发生过；而这两步转换恰恰是为
|
||
PostgreSQL 而写的，漏了只会在生产库上炸。
|
||
"""
|
||
recorded: list = []
|
||
monkeypatch.setattr("app.db.oper.subscribe.Subscribe", _SubscribeSpy(recorded))
|
||
return recorded
|
||
|
||
|
||
# 每个字段契约都在同步与异步两条链路上跑一遍：两份实现是逐字复制的，
|
||
# 只测一条等于放任另一条漂移
|
||
_BOTH_PATHS = pytest.mark.parametrize(
|
||
"is_async", [pytest.param(False, id="sync"), pytest.param(True, id="async")]
|
||
)
|
||
|
||
|
||
# --------------------------------------------------------------------------- #
|
||
# 展示字段的翻译
|
||
# --------------------------------------------------------------------------- #
|
||
|
||
@_BOTH_PATHS
|
||
def test_add_maps_every_display_field_onto_the_row(db, is_async):
|
||
"""
|
||
识别结果的展示字段必须完整落库。
|
||
|
||
标题、年份、类型、评分、简介都是订阅列表和通知的唯一数据来源，
|
||
错一项用户就看到一条张冠李戴的订阅。
|
||
"""
|
||
oper = SubscribeOper()
|
||
media_id = _media_id(f"display-{is_async}")
|
||
|
||
sid, message = _add(oper, is_async, mediainfo=_mediainfo(media_id), season=1)
|
||
|
||
assert message == "新增订阅成功"
|
||
row = _row(db, sid)
|
||
assert row.name == "识别标题"
|
||
assert row.year == "2026"
|
||
assert row.type == MediaType.TV.value
|
||
assert row.media_source == MediaSource.TMDB.value
|
||
assert row.media_id == media_id
|
||
assert row.season == 1
|
||
assert row.vote == 8.5
|
||
assert row.description == "测试简介"
|
||
|
||
|
||
@_BOTH_PATHS
|
||
def test_add_persists_poster_and_backdrop_from_media(db, is_async):
|
||
"""
|
||
海报与背景取自识别结果的图片接口，而不是原始路径字段。
|
||
|
||
接口会把 original 尺寸换成 w500，直接存 poster_path 会让列表页
|
||
每张卡片都去拉原图。
|
||
"""
|
||
oper = SubscribeOper()
|
||
media_id = _media_id(f"image-{is_async}")
|
||
|
||
sid, _ = _add(oper, is_async, mediainfo=_mediainfo(media_id), season=1)
|
||
|
||
row = _row(db, sid)
|
||
assert row.poster == "https://image.tmdb.org/t/p/w500/poster.jpg"
|
||
assert row.backdrop == "https://image.tmdb.org/t/p/w500/backdrop.jpg"
|
||
|
||
|
||
@_BOTH_PATHS
|
||
def test_add_persists_episode_group(db, is_async):
|
||
"""
|
||
剧集组必须来自识别结果并落库。
|
||
|
||
订阅按剧集组去重、搜索也按剧集组匹配集数，丢了它主季与自定义组会互相顶替。
|
||
"""
|
||
oper = SubscribeOper()
|
||
media_id = _media_id(f"eg-{is_async}")
|
||
|
||
sid, _ = _add(oper, is_async,
|
||
mediainfo=_mediainfo(media_id, episode_group="eg-1"), season=1)
|
||
|
||
assert _row(db, sid).episode_group == "eg-1"
|
||
|
||
|
||
@_BOTH_PATHS
|
||
def test_add_stamps_creation_date(db, is_async):
|
||
"""
|
||
新增时间由写入路径盖戳，调用方传入的值不作数。
|
||
|
||
订阅列表默认按 date 排序、过期清理也读它，留空会让这条订阅永远排在最后。
|
||
"""
|
||
oper = SubscribeOper()
|
||
media_id = _media_id(f"date-{is_async}")
|
||
|
||
sid, _ = _add(oper, is_async, mediainfo=_mediainfo(media_id), season=1,
|
||
date="1970-01-01 00:00:00")
|
||
|
||
date = _row(db, sid).date
|
||
assert date is not None
|
||
assert date != "1970-01-01 00:00:00"
|
||
# 形如 2026-08-14 12:34:56
|
||
assert len(date) == 19 and date[4] == "-" and date[13] == ":"
|
||
|
||
|
||
# --------------------------------------------------------------------------- #
|
||
# 持久化类型强转
|
||
# --------------------------------------------------------------------------- #
|
||
|
||
@_BOTH_PATHS
|
||
def test_add_converts_boolean_flags_to_integers(db, payloads, is_async):
|
||
"""
|
||
历史兼容的布尔开关建模前必须转成整型。
|
||
|
||
PostgreSQL 的整型列拒收布尔值，不转会让新增订阅在 PG 上直接抛类型错误。
|
||
断言落在建模入参上而非落库值：SQLite 会替我们把 True 存成 1，
|
||
按落库值断言的话删掉转换也照样通过。
|
||
"""
|
||
oper = SubscribeOper()
|
||
media_id = _media_id(f"flags-{is_async}")
|
||
|
||
_add(oper, is_async, mediainfo=_mediainfo(media_id), season=1,
|
||
best_version=True, best_version_full=False, manual_total_episode=True)
|
||
|
||
payload = payloads[-1]
|
||
for field, expected in (("best_version", 1), ("best_version_full", 0),
|
||
("manual_total_episode", 1)):
|
||
assert payload[field] == expected
|
||
assert type(payload[field]) is int, f"{field} 仍是 {type(payload[field])}"
|
||
|
||
|
||
@_BOTH_PATHS
|
||
@pytest.mark.parametrize(
|
||
"supplied, expected",
|
||
[
|
||
pytest.param(True, 1, id="真"),
|
||
pytest.param(False, 0, id="假"),
|
||
pytest.param(None, 0, id="缺省"),
|
||
],
|
||
)
|
||
def test_add_normalizes_search_imdbid_to_zero_or_one(db, payloads, is_async,
|
||
supplied, expected):
|
||
"""
|
||
search_imdbid 无论传什么都归一到整型 0/1。
|
||
|
||
这一列参与搜索分支判定，存进 None 或 True 会让「是否用 imdbid 搜」
|
||
在不同订阅上表现不一致，在 PG 上还会直接拒写。
|
||
"""
|
||
oper = SubscribeOper()
|
||
media_id = _media_id(f"imdb-{is_async}-{supplied}")
|
||
|
||
_add(oper, is_async, mediainfo=_mediainfo(media_id), season=1,
|
||
search_imdbid=supplied)
|
||
|
||
payload = payloads[-1]
|
||
assert payload["search_imdbid"] == expected
|
||
assert type(payload["search_imdbid"]) is int
|
||
|
||
|
||
@_BOTH_PATHS
|
||
def test_add_converts_numeric_year_to_string(db, payloads, is_async):
|
||
"""
|
||
音乐链路的年份是数字，而 year 列是字符串，建模前必须转换。
|
||
|
||
不转在 PostgreSQL 上直接抛类型错误。同样只能在建模入参上验证：
|
||
SQLite 的 TEXT 亲和会把整数 2003 自动存成 '2003'，读回来看不出差别。
|
||
"""
|
||
oper = SubscribeOper()
|
||
media_id = _media_id(f"year-{is_async}")
|
||
|
||
sid, _ = _add(oper, is_async,
|
||
mediainfo=_musicinfo(media_id, "album", title="叶惠美", year=2003))
|
||
|
||
assert payloads[-1]["year"] == "2003"
|
||
assert type(payloads[-1]["year"]) is str
|
||
# 落库值也要对得上，转换不能只发生在建模而在写入时被改回去
|
||
assert _row(db, sid).year == "2003"
|
||
|
||
|
||
@_BOTH_PATHS
|
||
def test_add_keeps_missing_year_as_null(db, payloads, is_async):
|
||
"""
|
||
年份缺失时留空，不能变成字符串 "None"。
|
||
|
||
转字符串离无脑 str() 只有一步之遥，写成 "None" 后年份筛选会命中一个
|
||
不存在的年份，而这条订阅从此在按年份筛选的界面里凭空消失。
|
||
"""
|
||
oper = SubscribeOper()
|
||
media_id = _media_id(f"noyear-{is_async}")
|
||
|
||
sid, _ = _add(oper, is_async,
|
||
mediainfo=_musicinfo(media_id, "album", title="无年份专辑"))
|
||
|
||
assert payloads[-1]["year"] is None
|
||
assert _row(db, sid).year is None
|
||
|
||
|
||
# --------------------------------------------------------------------------- #
|
||
# 音乐字段
|
||
# --------------------------------------------------------------------------- #
|
||
|
||
@_BOTH_PATHS
|
||
def test_add_persists_album_entity_and_track_count(db, is_async):
|
||
"""
|
||
专辑订阅要落实体类型和总曲目数。
|
||
|
||
整专完成判定拿 total_tracks 当分母，缺了这条订阅永远判不到完成。
|
||
"""
|
||
oper = SubscribeOper()
|
||
media_id = _media_id(f"album-{is_async}")
|
||
|
||
sid, _ = _add(oper, is_async,
|
||
mediainfo=_musicinfo(media_id, "album", title="叶惠美", total_tracks=11))
|
||
|
||
row = _row(db, sid)
|
||
assert row.type == MediaType.MUSIC.value
|
||
assert row.music_type == "album"
|
||
assert row.total_tracks == 11
|
||
|
||
|
||
@_BOTH_PATHS
|
||
def test_add_drops_track_count_for_single_recording(db, is_async):
|
||
"""
|
||
单曲订阅只留实体类型，专辑曲目数必须丢弃。
|
||
|
||
单曲带着专辑的 total_tracks 会让完成判定把一首歌当整专等，永远不完成。
|
||
"""
|
||
oper = SubscribeOper()
|
||
media_id = _media_id(f"recording-{is_async}")
|
||
|
||
sid, _ = _add(oper, is_async,
|
||
mediainfo=_musicinfo(media_id, "recording", title="晴天", total_tracks=11))
|
||
|
||
row = _row(db, sid)
|
||
assert row.music_type == "recording"
|
||
assert row.total_tracks is None
|
||
|
||
|
||
@_BOTH_PATHS
|
||
def test_add_clears_music_fields_for_non_music_media(db, is_async):
|
||
"""
|
||
非音乐媒体的音乐字段一律置空，调用方传进来的也要被覆盖。
|
||
|
||
影视订阅带上 music_type 会被音乐去重逻辑当成音乐实体，造成串号。
|
||
"""
|
||
oper = SubscribeOper()
|
||
media_id = _media_id(f"nonmusic-{is_async}")
|
||
|
||
sid, _ = _add(oper, is_async, mediainfo=_mediainfo(media_id), season=1,
|
||
music_type="album", total_tracks=99)
|
||
|
||
row = _row(db, sid)
|
||
assert row.music_type is None
|
||
assert row.total_tracks is None
|
||
|
||
|
||
@_BOTH_PATHS
|
||
def test_add_persists_music_cover_as_poster_and_backdrop(db, is_async):
|
||
"""
|
||
音乐订阅的海报与背景都用发行封面。
|
||
|
||
音乐没有独立背景图，留空会让订阅卡片在列表里显示成一块空白。
|
||
"""
|
||
oper = SubscribeOper()
|
||
media_id = _media_id(f"cover-{is_async}")
|
||
cover = "https://coverartarchive.org/release-group/example/front-500"
|
||
|
||
sid, _ = _add(oper, is_async,
|
||
mediainfo=_musicinfo(media_id, "album", title="封面专辑", cover_url=cover))
|
||
|
||
row = _row(db, sid)
|
||
assert row.poster == cover
|
||
assert row.backdrop == cover
|
||
|
||
|
||
# --------------------------------------------------------------------------- #
|
||
# 调用方字段的保留
|
||
# --------------------------------------------------------------------------- #
|
||
|
||
@_BOTH_PATHS
|
||
def test_add_keeps_caller_supplied_subscription_settings(db, is_async):
|
||
"""
|
||
调用方传入的订阅设置要原样保留，不被媒体翻译覆盖。
|
||
|
||
写入路径只负责翻译媒体身份与展示字段；把用户填的保存路径、过滤词、
|
||
总集数一并覆盖掉，等于用户每次新增订阅的设置都白填。
|
||
"""
|
||
oper = SubscribeOper()
|
||
media_id = _media_id(f"settings-{is_async}")
|
||
|
||
sid, _ = _add(oper, is_async, mediainfo=_mediainfo(media_id), season=1,
|
||
username="alice", save_path="/media/tv", keyword="关键字",
|
||
include="内嵌", exclude="预告", total_episode=12,
|
||
start_episode=3, downloader="qbittorrent", state="R")
|
||
|
||
row = _row(db, sid)
|
||
assert row.username == "alice"
|
||
assert row.save_path == "/media/tv"
|
||
assert row.keyword == "关键字"
|
||
assert (row.include, row.exclude) == ("内嵌", "预告")
|
||
assert (row.total_episode, row.start_episode) == (12, 3)
|
||
assert row.downloader == "qbittorrent"
|
||
assert row.state == "R"
|
||
|
||
|
||
@_BOTH_PATHS
|
||
def test_add_overrides_caller_supplied_media_fields(db, is_async):
|
||
"""
|
||
媒体身份与展示字段以识别结果为准，调用方传的同名值要被覆盖。
|
||
|
||
否则上游一个陈旧的 name/year 就能让订阅记成另一部剧，而去重按身份走、
|
||
发现不了这种错位。
|
||
"""
|
||
oper = SubscribeOper()
|
||
media_id = _media_id(f"override-{is_async}")
|
||
|
||
sid, _ = _add(oper, is_async,
|
||
mediainfo=_mediainfo(media_id, title="正确标题", year="2026"),
|
||
season=1, name="错误标题", year="1999", type="电影")
|
||
|
||
row = _row(db, sid)
|
||
assert row.name == "正确标题"
|
||
assert row.year == "2026"
|
||
assert row.type == MediaType.TV.value
|