mirror of
https://github.com/jxxghp/MoviePilot.git
synced 2026-11-01 21:07:01 +08:00
274 lines
12 KiB
Python
274 lines
12 KiB
Python
"""AMLL TTML 的主唱提取、时轴转换和外部 XML 边界测试。"""
|
||
|
||
import pytest
|
||
import yaml
|
||
|
||
from app.domain.context import MusicLyrics
|
||
from app.modules.amll.lyrics import parse_lyrics
|
||
|
||
|
||
def _document(body: str, *, head: str = "", root_attributes: str = "") -> str:
|
||
"""用合成歌词构造符合 AMLL 命名空间的测试文档，避免访问真实歌词源。"""
|
||
return (
|
||
'<tt xmlns="http://www.w3.org/ns/ttml" '
|
||
'xmlns:ttm="http://www.w3.org/ns/ttml#metadata" '
|
||
'xmlns:tts="http://www.w3.org/ns/ttml#styling" '
|
||
'xmlns:itunes="http://itunes.apple.com/lyric-ttml-extensions" '
|
||
f'xml:lang="en" {root_attributes}>'
|
||
f'<head>{head}</head><body dur="05:00.000"><div>{body}</div></body></tt>'
|
||
)
|
||
|
||
|
||
def _parse(content: str):
|
||
"""固定已经由来源匹配层确认的元数据，直接测试纯格式转换入口。"""
|
||
return parse_lyrics(
|
||
content, title="Example Track", artist="Example Artist",
|
||
provider_id="123", match_score=95,
|
||
)
|
||
|
||
|
||
def test_word_ttml_preserves_times_spaces_and_removes_auxiliary_tracks() -> None:
|
||
"""逐词歌词保留主唱绝对时轴和英文空格，不混入三类辅助声音或头部翻译。"""
|
||
lyrics = _parse(_document(
|
||
'<p begin="1.250" end="3.000">'
|
||
'<span ttm:role="x-bg" begin="1" end="2">Background first</span>'
|
||
'<span begin="1.250" end="2">Example</span> '
|
||
'<span begin="2" end="3">line</span>'
|
||
'<span ttm:role="x-translation">翻译示例</span>'
|
||
'<span ttm:role="x-roman">Romanized example</span>'
|
||
'<span ttm:role="x-bg"><span begin="2" end="3">Backing</span></span>'
|
||
'</p>',
|
||
head='<metadata><translation><p begin="1" end="3">Sidecar</p>'
|
||
'</translation></metadata>',
|
||
))
|
||
|
||
assert lyrics is not None
|
||
assert lyrics.provider == "amll"
|
||
assert lyrics.provider_id == "123"
|
||
assert lyrics.match_score == 95
|
||
assert lyrics.language == "en"
|
||
assert lyrics.synced_lyrics == "[00:01.25]Example line"
|
||
assert lyrics.plain_lyrics == "Example line"
|
||
assert lyrics.quality_rank == 4
|
||
payload = yaml.safe_load(lyrics.lyricsfile)
|
||
assert payload["version"] == "1.0"
|
||
assert payload["lines"] == [{
|
||
"start_ms": 1250, "end_ms": 3000, "text": "Example line",
|
||
"words": [
|
||
{"start_ms": 1250, "end_ms": 2000, "text": "Example "},
|
||
{"start_ms": 2000, "end_ms": 3000, "text": "line"},
|
||
],
|
||
}]
|
||
assert "duration" not in payload["metadata"]
|
||
|
||
|
||
@pytest.mark.parametrize(("begin", "end", "expected"), [
|
||
("1.001", "2.501", 1001),
|
||
("01:02.345", "01:03.456", 62345),
|
||
("1:02:03.456", "1:02:04.567", 3723456),
|
||
("120", "121", 120000),
|
||
("0", "0.5", 0),
|
||
(" 1.250 ", "2", 1250),
|
||
])
|
||
def test_line_ttml_parses_official_timestamp_forms(begin, end, expected) -> None:
|
||
"""秒、分秒和时分秒格式统一为整数毫秒，逐行内容不虚报逐词质量。"""
|
||
lyrics = _parse(_document(f'<p begin="{begin}" end="{end}">Example line</p>'))
|
||
|
||
assert lyrics is not None
|
||
assert lyrics.quality_rank == 3
|
||
payload = yaml.safe_load(lyrics.lyricsfile)
|
||
assert payload["lines"][0]["start_ms"] == expected
|
||
assert "words" not in payload["lines"][0]
|
||
|
||
|
||
def test_indentation_does_not_insert_spaces_between_chinese_words() -> None:
|
||
"""格式化 XML 的换行缩进不应拆开无词间空格的中文歌词。"""
|
||
lyrics = _parse(_document('''
|
||
<p begin="1" end="3">
|
||
<span begin="1" end="2">示</span>
|
||
<span begin="2" end="3">例</span>
|
||
</p>
|
||
'''))
|
||
|
||
assert lyrics is not None
|
||
assert lyrics.plain_lyrics == "示例"
|
||
assert lyrics.quality_rank == 4
|
||
|
||
|
||
def test_word_internal_spaces_survive_formatted_document() -> None:
|
||
"""英文词尾空格放在 span 内时，即使文件格式化也应完整保留。"""
|
||
lyrics = _parse(_document('''
|
||
<p begin="1" end="3">
|
||
<span begin="1" end="2">Example </span>
|
||
<span begin="2" end="3">line</span>
|
||
</p>
|
||
'''))
|
||
|
||
assert lyrics is not None
|
||
assert lyrics.plain_lyrics == "Example line"
|
||
assert yaml.safe_load(lyrics.lyricsfile)["lines"][0]["words"][0]["text"] == "Example "
|
||
|
||
|
||
def test_nested_styles_keep_finest_word_times_and_exclude_ruby_annotations() -> None:
|
||
"""嵌套样式不重复文本，Ruby 注音不混入主唱句子，词级时轴保持最细粒度。"""
|
||
lyrics = _parse(_document(
|
||
'<p begin="10" end="12"><span begin="10" end="12">'
|
||
'<span begin="10" end="11"><span tts:ruby="base">示</span>'
|
||
'<span tts:ruby="textContainer"><span tts:ruby="text">shi</span></span></span>'
|
||
'<span begin="11" end="12">例</span></span></p>'
|
||
))
|
||
|
||
assert lyrics is not None
|
||
assert lyrics.plain_lyrics == "示例"
|
||
assert lyrics.quality_rank == 4
|
||
assert len(yaml.safe_load(lyrics.lyricsfile)["lines"][0]["words"]) == 2
|
||
|
||
|
||
def test_word_times_can_supply_missing_line_time() -> None:
|
||
"""完整逐词数据可以恢复缺失的行范围，不把正文开始时间累加到词时间。"""
|
||
lyrics = _parse(_document(
|
||
'<p><span begin="10" end="11">Example </span>'
|
||
'<span begin="11" end="12">line</span></p>'
|
||
).replace('<div>', '<div begin="10" end="20">'))
|
||
|
||
assert lyrics is not None
|
||
assert lyrics.synced_lyrics == "[00:10.00]Example line"
|
||
assert lyrics.quality_rank == 4
|
||
|
||
|
||
@pytest.mark.parametrize("word", [
|
||
'<span begin="2" end="1">line</span>',
|
||
'<span begin="1" end="5">line</span>',
|
||
'<span begin="1">line</span>',
|
||
'<span begin="1s" end="2s">line</span>',
|
||
'<span>line</span>',
|
||
])
|
||
def test_invalid_or_partial_word_timing_falls_back_to_complete_line(word) -> None:
|
||
"""无效、越界或缺失的部分词时间不得虚报逐词质量，也不能丢失正文。"""
|
||
lyrics = _parse(_document(
|
||
'<p begin="1" end="3"><span begin="1" end="2">Example </span>'
|
||
f'{word}</p>'
|
||
))
|
||
|
||
assert lyrics is not None
|
||
assert lyrics.plain_lyrics == "Example line"
|
||
assert lyrics.quality_rank == 3
|
||
assert "words" not in yaml.safe_load(lyrics.lyricsfile)["lines"][0]
|
||
|
||
|
||
@pytest.mark.parametrize("attributes", [
|
||
'', 'begin="2" end="1"', 'begin="-1" end="2"', 'begin="NaN" end="2"',
|
||
'begin="1s" end="2s"', 'begin="1:2:3:4" end="5"', 'begin="1"',
|
||
])
|
||
def test_any_untimed_line_falls_back_to_plain_without_losing_other_lines(attributes) -> None:
|
||
"""混合无效和有效行时轴时整份降为纯文本，保留完整文本而不伪造零秒时间。"""
|
||
lyrics = _parse(_document(
|
||
f'<p {attributes}>First line</p><p begin="3" end="4">Second line</p>'
|
||
))
|
||
|
||
assert lyrics is not None
|
||
assert lyrics.plain_lyrics == "First line\nSecond line"
|
||
assert lyrics.synced_lyrics is None
|
||
assert lyrics.lyricsfile is None
|
||
assert lyrics.quality_rank == 1
|
||
|
||
|
||
def test_auxiliary_containers_and_empty_lines_are_ignored() -> None:
|
||
"""只有辅助人声和空行的条目无候选，完整辅助容器也不贡献正文。"""
|
||
assert _parse(_document('<p begin="1" end="2"><span ttm:role="x-bg">Backing</span></p>')) is None
|
||
lyrics = _parse(_document(
|
||
'<div ttm:role="x-translation"><p begin="1" end="2">Translation</p></div>'
|
||
'<p begin="1" end="2"> </p><p begin="2" end="3">Main</p>'
|
||
))
|
||
assert lyrics is not None
|
||
assert lyrics.plain_lyrics == "Main"
|
||
|
||
|
||
def test_xml_entities_and_doctypes_are_rejected_without_resolution(tmp_path) -> None:
|
||
"""任何 DTD 都被拒绝，外部文件实体不能泄露内容，网络实体不能触发请求。"""
|
||
secret = tmp_path / "secret.txt"
|
||
secret.write_text("should never appear", encoding="utf-8")
|
||
for source in (secret.as_uri(), "https://example.invalid/secret"):
|
||
document = '<!DOCTYPE tt [<!ENTITY payload SYSTEM "' + source + '">]>'
|
||
document += _document('<p begin="1" end="2">&payload;</p>')
|
||
assert _parse(document) is None
|
||
assert _parse('<!DOCTYPE tt [<!ENTITY value "expanded">]>' + _document(
|
||
'<p begin="1" end="2">&value;</p>'
|
||
)) is None
|
||
|
||
|
||
@pytest.mark.parametrize("content", [
|
||
'', '<tt>', '<html><body><p>Wrong format</p></body></html>',
|
||
'<tt xmlns="http://www.w3.org/ns/ttml"><head/></tt>',
|
||
'<tt xmlns="urn:wrong"><body><p>Wrong namespace</p></body></tt>',
|
||
'\ud800',
|
||
])
|
||
def test_malformed_empty_or_non_ttml_documents_return_no_candidate(content) -> None:
|
||
"""无正文、非 TTML、损坏 XML 和不可编码文本不应产生可写入候选。"""
|
||
assert _parse(content) is None
|
||
|
||
|
||
def test_size_depth_and_node_limits_reject_oversized_documents() -> None:
|
||
"""在转换前限制字节数、深度和节点数，避免外部歌词放大解析成本。"""
|
||
assert _parse(_document('<p>' + 'a' * (1024 * 1024) + '</p>')) is None
|
||
assert _parse(_document('<p>' + '中' * 400000 + '</p>')) is None
|
||
assert _parse(_document('<span>' * 21 + '<p>Deep</p>' + '</span>' * 21)) is None
|
||
assert _parse(_document('<p>Line</p>' * 10001)) is None
|
||
|
||
|
||
def test_existing_xml_declaration_and_escaped_text_are_decoded_safely() -> None:
|
||
"""已解码字符串的声明不会重解释 UTF-8 字节，标准转义字符正常还原。"""
|
||
content = '<?xml version="1.0" encoding="ISO-8859-1"?>' + _document(
|
||
'<p begin="1" end="2">示例 & text</p>'
|
||
)
|
||
lyrics = _parse(content)
|
||
|
||
assert lyrics is not None
|
||
assert lyrics.plain_lyrics == "示例 & text"
|
||
|
||
|
||
def test_missing_confirmed_identity_is_rejected() -> None:
|
||
"""转换器不为缺少已匹配标题或艺人的调用伪造 Lyricsfile 元数据。"""
|
||
content = _document('<p begin="1" end="2">Example line</p>')
|
||
assert parse_lyrics(content, title="", artist="Artist", provider_id="1", match_score=90) is None
|
||
assert parse_lyrics(content, title="Track", artist=" ", provider_id="1", match_score=90) is None
|
||
|
||
|
||
@pytest.mark.parametrize("text", [
|
||
"No", "on", "yes", "Off", "TRUE", "false", "null", "Null", "~",
|
||
"123", "0123", "1.25", "01:02", "2026-09-09",
|
||
])
|
||
def test_lyricsfile_roundtrip_preserves_yaml_scalar_looking_lyrics(text: str) -> None:
|
||
"""歌词词语不能被宿主 YAML 1.1 读者误解为布尔、空值、数字或日期。"""
|
||
content = _document(
|
||
f'<p begin="1" end="2"><span begin="1" end="2">{text}</span></p>'
|
||
)
|
||
lyrics = _parse(content)
|
||
|
||
assert lyrics is not None
|
||
assert lyrics.plain_lyrics == text
|
||
assert lyrics.quality_rank == 4
|
||
# 再经过真实宿主模型读取生成物，覆盖写入旁挂后重新读取的类型合同。
|
||
restored = MusicLyrics(provider="amll", lyricsfile=lyrics.lyricsfile)
|
||
assert restored.plain_lyrics == text
|
||
assert restored.synced_lyrics == f"[00:01.00]{text}"
|
||
payload = yaml.safe_load(lyrics.lyricsfile)
|
||
assert payload["lines"][0]["text"] == text
|
||
assert payload["lines"][0]["words"][0]["text"] == text
|
||
assert payload["lines"][0]["start_ms"] == 1000
|
||
assert isinstance(payload["lines"][0]["start_ms"], int)
|
||
|
||
|
||
def test_lyricsfile_roundtrip_preserves_scalar_looking_identity_and_null_language() -> None:
|
||
"""元数据名称仍为字符串，缺失语言仍为 null，不通过全字段字符串化修复。"""
|
||
content = _document('<p begin="1" end="2">Example line</p>').replace('xml:lang="en"', '')
|
||
lyrics = parse_lyrics(content, title="On", artist="No", provider_id="1", match_score=90)
|
||
|
||
assert lyrics is not None
|
||
restored = MusicLyrics(provider="amll", lyricsfile=lyrics.lyricsfile)
|
||
assert restored.plain_lyrics == "Example line"
|
||
assert restored.language is None
|
||
payload = yaml.safe_load(lyrics.lyricsfile)
|
||
assert payload["version"] == "1.0"
|
||
assert payload["metadata"] == {"title": "On", "artist": "No", "language": None}
|