GithubBackup/MoviePilot
1
0
Fork 0
mirror of https://github.com/jxxghp/MoviePilot.git synced 2026-09-08 17:08:35 +08:00
Code Issues Packages Projects Releases Wiki Activity

Merge pull request #6583 from dogodefi/codex/music-album-scrape-context

Browse Source
This commit is contained in:
jxxghp
2026-09-05 15:29:15 +08:00
committed by GitHub
parent c67622f1ad 20257a88cb
commit 5c2c62f7b6
3 changed files with 255 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files
app/chain/transfer/filter.py
+19 -1
View File
@@ -26,6 +26,7 @@ from app.runtime.log import logger
from app.schemas.transfer import TransferInfo
from app.schemas.types import (
MUSIC_ENTITY_ALBUM,
MUSIC_ENTITY_RECORDING,
MediaType,
)
from app.schemas.workflow import FileItem
@@ -251,8 +252,13 @@ class FileFilterMixin(_TransferOwnerBase):
cls,
download_history: Optional[DownloadHistorySnapshot],
file_path: Path,
discard_recording_identity: bool = False,
) -> tuple[Optional[MetaMusic], Optional[MusicInfo]]:
"""从下载历史恢复音乐上下文，并用当前音频标签覆盖曲目级字段。"""
"""从下载历史恢复音乐上下文，并用当前音频标签覆盖曲目级字段。
多音轨批次误带单曲身份时只保留文件自身标签，避免把同一 recording
身份传播到整张专辑；调用方随后可使用目录级证据重新匹配专辑。
"""
note = getattr(download_history, "note", None)
music_note = note.get("music") if isinstance(note, dict) else None
if not isinstance(music_note, dict) or music_note.get("version") != 1:
@@ -264,6 +270,18 @@ class FileFilterMixin(_TransferOwnerBase):
return None, None
file_tags = MediaChain.read_path_meta(file_path)
should_discard_identity = (
discard_recording_identity
and saved_info.music_type == MUSIC_ENTITY_RECORDING
)
if should_discard_identity:
# 共享 recording 上下文可能包含错误的专辑、年份等字段；整张丢弃，
# 只保留当前文件实际标签，目录级匹配失败时也不会回落到错误身份。
file_meta = deepcopy(file_tags)
file_meta.org_string = file_path.name
file_meta.title = file_meta.title or file_path.stem
return file_meta, None
file_meta = deepcopy(saved_meta)
file_meta.org_string = file_path.name
# 曲目标题始终优先使用当前文件自身的标签（缺失时回退为文件名），
app/chain/transfer/workflow.py
+41 -2
View File
@@ -36,6 +36,7 @@ from app.schemas.media import resolve_media_identity
from app.schemas.system import TransferDirectoryConf
from app.schemas.transfer import EpisodeFormat, TransferInfo
from app.schemas.types import (
MUSIC_ENTITY_RECORDING,
MediaSource,
MediaType,
ProgressKey,
@@ -46,6 +47,25 @@ from app.schemas.workflow import FileItem
from .request import _TransferCandidatePlanner
def _should_discard_batch_recording_identity(
*,
multi_track_music_batch: bool,
manual: bool,
media_source: Optional[MediaSource],
media_id: Optional[str],
mediainfo: Optional[MediaInfo | MusicInfo],
history_music_type: Optional[str],
) -> bool:
"""判断自动整专是否误带了共享单曲身份。"""
if not multi_track_music_batch or (manual and media_source and media_id):
return False
batch_music_type = getattr(mediainfo, "music_type", None)
return (
batch_music_type == MUSIC_ENTITY_RECORDING
or (not batch_music_type and history_music_type == MUSIC_ENTITY_RECORDING)
)
class TransferWorkflowOwner(_TransferOwnerBase):
"""协调请求级候选构建并委托规划、执行与结算 owner。"""
@@ -654,6 +674,10 @@ class TransferWorkflowOwner(_TransferOwnerBase):
skipped_history_count = 0
skipped_torrents = set()
cleanup_intent_assigned = False
multi_track_music_batch = (
batch_mtype == MediaType.MUSIC
and sum(self._is_audio_file(item) for item, _ in file_items) > 1
)
try:
for file_item, bluray_dir in file_items:
if runtime_stop_state.is_system_stopped:
@@ -749,9 +773,18 @@ class TransferWorkflowOwner(_TransferOwnerBase):
download_hash=download_hash,
)
discard_recording_identity = _should_discard_batch_recording_identity(
multi_track_music_batch=multi_track_music_batch,
manual=manual,
media_source=media_source,
media_id=media_id,
mediainfo=mediainfo,
history_music_type=self._download_history_music_type(download_history),
)
history_music_meta, history_music_info = self._restore_music_download_context(
download_history=download_history,
file_path=file_path,
discard_recording_identity=discard_recording_identity,
)
if not meta:
@@ -784,10 +817,16 @@ class TransferWorkflowOwner(_TransferOwnerBase):
_download_hash = download_hash
# 自动整理预载的媒体信息来自整条下载历史；电影合集内文件年份冲突时逐文件识别。
task_mediainfo = mediainfo or history_music_info
task_mediainfo = (
None
if discard_recording_identity
else mediainfo or history_music_info
)
if not task_mediainfo and isinstance(file_meta, MetaMusic):
# 无标签音频按目录级专辑匹配补齐曲目身份，命中结果带缓存不会逐文件重复请求
# 无标签音频或误带单曲身份的整包按目录级专辑匹配；命中结果带缓存不会逐文件重复请求
file_meta, task_mediainfo = self._match_music_album_context(file_item, file_path, file_meta)
if not task_mediainfo and discard_recording_identity:
task_mediainfo = self._music_info_from_meta(file_meta)
if not manual and task_mediainfo and self._is_movie_year_conflict(file_meta, task_mediainfo):
task_mediainfo = None
tests/test_music_transfer.py
+195 -1
View File
@@ -4,6 +4,7 @@ from unittest.mock import Mock
from jinja2 import Template
from app.application.history import DownloadHistorySnapshot
from app.application.messaging.message import TemplateHelper
from app.application.transfer.execution import (
TransferExecutionCheckpoint,
@@ -11,7 +12,7 @@ from app.application.transfer.execution import (
)
from app.application.transfer.workflow import JobManager, TransferTask
from app.chain.media import MediaChain
from app.chain.transfer import TransferChain
from app.chain.transfer import TransferChain # pylint: disable=no-name-in-module
from app.domain.context import MusicInfo
from app.domain.meta.metamusic import MetaMusic
from app.runtime.config import settings
@@ -247,6 +248,50 @@ def test_restore_music_context_from_download_history():
assert restored_info.album == "Random Access Memories"
def test_restore_music_context_discards_shared_recording_identity(tmp_path, monkeypatch):
"""多音轨批次不得把下载记录中的单曲身份恢复到每个音频文件。"""
meta, info = _music_context()
history = DownloadHistorySnapshot(
id=1,
path=tmp_path.as_posix(),
type=MediaType.MUSIC.value,
title="Random Access Memories",
note={
"music": {
"version": 1,
"meta": meta.to_dict(),
"media": info.to_dict(),
}
},
)
audio_file = tmp_path / "01 - Give Life Back to Music.flac"
audio_file.write_bytes(b"fake-flac")
file_meta = MetaMusic(
org_string=audio_file.name,
title="Give Life Back to Music",
artists=["Daft Punk"],
album="Local Album",
album_artist="Daft Punk",
year=2020,
track_number=1,
)
monkeypatch.setattr(MediaChain, "read_path_meta", Mock(return_value=file_meta))
restored_meta, restored_info = TransferChain._restore_music_download_context(
history,
audio_file,
discard_recording_identity=True,
)
assert restored_meta is not None
assert restored_meta.title == "Give Life Back to Music"
assert restored_meta.album == "Local Album"
assert restored_meta.year == 2020
assert restored_meta.media_source is None
assert restored_meta.media_id is None
assert restored_info is None
def test_download_history_music_type_falls_back_to_versioned_note():
"""旧下载记录缺少独立字段时应从版本化备注恢复实体类型。"""
history = SimpleNamespace(
@@ -673,6 +718,155 @@ def test_automatic_audio_transfer_runs_music_recognition(tmp_path, monkeypatch):
assert preview["items"][0]["type"] == MediaType.MUSIC.value
def test_automatic_multi_track_recording_context_rematches_album(tmp_path, monkeypatch):
"""自动整专不得复用下载时误选的单曲身份，应按目录恢复各音轨身份。"""
source_dir = tmp_path / "徐良 情话"
source_dir.mkdir()
audio_paths = [
source_dir / "01 - 女骑士.flac",
source_dir / "02 - 悲伤的李白.flac",
]
for audio_path in audio_paths:
audio_path.write_bytes(b"fake-flac")
source_items = [
FileItem(
storage="local",
path=audio_path.as_posix(),
name=audio_path.name,
basename=audio_path.stem,
type="file",
extension="flac",
size=audio_path.stat().st_size,
)
for audio_path in audio_paths
]
source_item = FileItem(
storage="local",
path=source_dir.as_posix(),
name=source_dir.name,
type="dir",
)
saved_recording = MusicInfo(
media_source="musicbrainz",
media_id="wrong-recording",
music_type="recording",
title="情话",
artists=["徐良", "孙羽幽"],
album="北京巷弄",
album_artist="徐良",
year=2013,
cover_url="https://example.com/wrong-cover.jpg",
)
saved_meta = MetaMusic.from_music_info(saved_recording)
history = DownloadHistorySnapshot(
id=1,
path=source_dir.as_posix(),
type=MediaType.MUSIC.value,
title="情话",
note={
"music": {
"version": 1,
"meta": saved_meta.to_dict(),
"media": saved_recording.to_dict(),
}
},
music_type="recording",
downloader="qbittorrent",
download_hash="hash-1",
)
file_metas = {
audio_paths[0]: MetaMusic(
org_string=audio_paths[0].name,
title="女骑士",
artists=["徐良"],
album="情话",
album_artist="徐良",
year=2013,
track_number=1,
total_tracks=12,
),
audio_paths[1]: MetaMusic(
org_string=audio_paths[1].name,
title="悲伤的李白",
artists=["徐良"],
album="情话",
album_artist="徐良",
year=2013,
track_number=2,
total_tracks=12,
),
}
matched_tracks = {
str(path.resolve()): MusicInfo(
media_source="musicbrainz",
media_id=f"recording-{index}",
music_type="recording",
title=file_metas[path].title,
artists=["徐良"],
album="情话",
album_artist="徐良",
album_id="correct-release-group",
year=2013,
track_number=index,
total_tracks=12,
cover_url="https://example.com/correct-cover.jpg",
)
for index, path in enumerate(audio_paths, start=1)
}
chain = TransferChain()
monkeypatch.setattr(
chain,
"_TransferChain__get_trans_fileitems",
Mock(return_value=[(item, False) for item in source_items]),
)
monkeypatch.setattr(chain, "_resolve_download_history", Mock(return_value=history))
monkeypatch.setattr(
MediaChain,
"read_path_meta",
Mock(side_effect=lambda path: file_metas[Path(path)]),
)
album_match = Mock(return_value=matched_tracks)
monkeypatch.setattr(MediaChain, "recognize_music_album_directory", album_match)
captured_tasks = []
def execute(task, **_kwargs):
captured_tasks.append(task)
target_dir = tmp_path / "library" / "徐良" / "情话 (2013)"
target_item = target_dir / task.fileitem.name
return TransferInfo(
success=True,
fileitem=task.fileitem,
target_item=FileItem(storage="local", path=target_item.as_posix(), type="file"),
target_diritem=FileItem(storage="local", path=target_dir.as_posix(), type="dir"),
)
monkeypatch.setattr(chain, "_plan_checkpoint_and_execute", execute)
state, preview = chain.do_transfer(
fileitem=source_item,
mediainfo=saved_recording,
mtype=MediaType.MUSIC,
target_directory=TransferDirectoryConf(
library_path=(tmp_path / "library").as_posix(),
library_storage="local",
),
force=True,
preview=True,
)
assert state is True
assert preview["summary"] == {"total": 2, "success": 2, "failed": 0}
assert [task.mediainfo.media_id for task in captured_tasks] == [
"recording-1",
"recording-2",
]
assert {task.mediainfo.album_id for task in captured_tasks} == {"correct-release-group"}
assert {task.mediainfo.cover_url for task in captured_tasks} == {
"https://example.com/correct-cover.jpg"
}
assert album_match.call_count == 2
def test_explicit_music_batch_excludes_video_from_mixed_directory(tmp_path, monkeypatch):
"""明确音乐上下文时只规划音频主文件，混合目录中的视频不得套用音乐身份。"""
audio_path = tmp_path / "08 - Get Lucky.flac"