From 4b16a60bf4a5602b9a4c7611cbcbdf5d05a62d7f Mon Sep 17 00:00:00 2001 From: Maxr1998 Date: Sun, 22 Jun 2025 16:36:50 +0200 Subject: [PATCH] Don't parse word time tags if multiple line times are present Word time tags are absolute timestamps in the lyrics, so having multiple line times would generally conflict with them. Thus, they should be just ignored instead, since the format is wrong. The same applies if there is no line time tag, in that case, it wouldn't be known when the first word starts. Both are inconsidered incorrect LRC and thus only parsed on a best-effort basis. --- .../Parser/Lrc/Lines/LrcLyricParserTest.cs | 11 ++++++++ LrcParser/Parser/Lrc/Lines/LrcLyricParser.cs | 25 +++++++++++++++++-- 2 files changed, 34 insertions(+), 2 deletions(-) diff --git a/LrcParser.Tests/Parser/Lrc/Lines/LrcLyricParserTest.cs b/LrcParser.Tests/Parser/Lrc/Lines/LrcLyricParserTest.cs index be730a4..2cdd150 100644 --- a/LrcParser.Tests/Parser/Lrc/Lines/LrcLyricParserTest.cs +++ b/LrcParser.Tests/Parser/Lrc/Lines/LrcLyricParserTest.cs @@ -105,6 +105,17 @@ public void TestDecode(string lyric, LrcLyric expected) TimeTags = [], }, ], + // Don't parse word time tags if multiple line time tags are found, as this is unsupported by LRC. + // Instead, return the unparsed line without the line time and no word time tags. + [ + "[00:17.00][00:18.00] <00:00.00>帰<00:01.00>り<00:02.00>道<00:03.00>は", + new LrcLyric + { + Text = "<00:00.00>帰<00:01.00>り<00:02.00>道<00:03.00>は", + StartTimes = [17000, 18000], + TimeTags = [], + }, + ], }; [TestCaseSource(nameof(testEncodeSource))] diff --git a/LrcParser/Parser/Lrc/Lines/LrcLyricParser.cs b/LrcParser/Parser/Lrc/Lines/LrcLyricParser.cs index 69c6ff2..0123e76 100644 --- a/LrcParser/Parser/Lrc/Lines/LrcLyricParser.cs +++ b/LrcParser/Parser/Lrc/Lines/LrcLyricParser.cs @@ -1,6 +1,7 @@ // Copyright (c) karaoke.dev . Licensed under the MIT Licence. // See the LICENCE file in the repository root for full licence text. +using LrcParser.Model; using LrcParser.Parser.Lines; using LrcParser.Parser.Lrc.Metadata; using LrcParser.Parser.Lrc.Utils; @@ -14,8 +15,28 @@ public override bool CanDecode(string text) public override LrcLyric Decode(string text) { - var (startTimes, lyricText) = LrcStartTimeUtils.SplitLyricAndTimeTag(text); - var (lyric, timeTags) = LrcTimedTextUtils.TimedTextToObject(lyricText); + var (startTimes, rawLyric) = LrcStartTimeUtils.SplitLyricAndTimeTag(text); + + // If there are multiple start times, it is possible that the given word time tags are incompatible with one or more start times. + // For example, if a line starts at both [01:00.00] and [02:00.00], + // word time tags with the values <01:10.00> and <01:20.00> would be incompatible with the second start time. + // As there isn't an official LRC spec, this isn't clearly defined. + // While the format is technically valid, we chose a reasonable behavior of ignoring the word time tags in this case + // and returning the line as-is without parsing the word time tags. + // The same applies to lines that have no start times: + // As there might be lines like `Every <00:07.56> night`, the first word would not have a start time, + // so we chose the same approach of ignoring the word time tags in this case. + if (startTimes.Length is 0 or > 1) + { + return new LrcLyric + { + Text = rawLyric, + StartTimes = startTimes, + TimeTags = new SortedDictionary() + }; + } + + var (lyric, timeTags) = LrcTimedTextUtils.TimedTextToObject(rawLyric); return new LrcLyric {