From 945930d75e60d57bc17e8de0d7988c36a84cf284 Mon Sep 17 00:00:00 2001
From: cxfksword <718792+cxfksword@users.noreply.github.com>
Date: Sat, 25 May 2024 16:00:19 +0800
Subject: [PATCH] fix: episode number not correctly identified. #82
---
AnitomySharp/Keyword.cs | 4 +-
AnitomySharp/ParserHelper.cs | 31 +++++++++++--
AnitomySharp/ParserNumber.cs | 29 +++++++++++--
.../ParseNameTest.cs | 43 +++++++++++++------
Jellyfin.Plugin.MetaShark/Core/NameParser.cs | 22 ++++++++--
.../Core/StringExtension.cs | 5 +++
.../Providers/EpisodeProvider.cs | 10 ++---
7 files changed, 113 insertions(+), 31 deletions(-)
diff --git a/AnitomySharp/Keyword.cs b/AnitomySharp/Keyword.cs
index a0adcae..f5bab41 100644
--- a/AnitomySharp/Keyword.cs
+++ b/AnitomySharp/Keyword.cs
@@ -67,7 +67,7 @@ static KeywordManager()
"番外編", "總集編","DRAMA",
"映像特典","特典","特典アニメ",
// 特典 Special 剩下的各种类型可以全部命名成 SP，对于较特殊意义的特典也可以自定义命名
- "SPECIAL", "SPECIALS", "SP", "SPs",
+ "SPECIAL", "SPECIALS", "SP", "SPs", "特報",
// 真人特典 Interview/Talk/Stage... 目前我们对于节目、采访、舞台活动、制作等三次元画面的长视频，一概怼成 IV。
"IV",
// 音乐视频 Music Video
@@ -85,7 +85,7 @@ static KeywordManager()
// 无字 OP/ED Non-Credit Opening/Ending
"ED", "ENDING", "NCED", "NCOP", "OP", "OPENING",
// 预告 Preview 预告下一话内容 注意编号表示其预告的是第几话的内容而不是跟在哪一话后面
- "PREVIEW", "YOKOKU",
+ "PREVIEW", "YOKOKU", "予告",
// 菜单 Menu BD/DVD 播放选择菜单
"MENU",
// 广告 Commercial Message 电视放送广告，时长一般在 7s/15s/30s/45s/... 左右
diff --git a/AnitomySharp/ParserHelper.cs b/AnitomySharp/ParserHelper.cs
index 904666e..1b6f1f5 100644
--- a/AnitomySharp/ParserHelper.cs
+++ b/AnitomySharp/ParserHelper.cs
@@ -119,6 +119,29 @@ public static string GetNumberFromOrdinal(string str)
if (string.IsNullOrEmpty(str)) return "";
return Ordinals.TryGetValue(str, out var foundString) ? foundString : "";
}
+ ///
+ /// 转换原始值中的全角数字
+ /// １２３４５６７８９０
+ ///
+ ///
+ ///
+ public static string GetNumberFromFullWidth(string str)
+ {
+ string output = str;
+ for (int i = 0; i < str.Length; i++)
+ {
+ if (char.IsDigit(str[i]))
+ {
+ int fullwidthDigit = (int)str[i];
+ if (fullwidthDigit >= 65296 && fullwidthDigit <= 65305)
+ {
+ int halfwidthDigit = fullwidthDigit - 65248;
+ output = output.Replace(str[i], (char)halfwidthDigit);
+ }
+ }
+ }
+ return output;
+ }
///
/// Returns the index of the first digit in the str; -1 otherwise.
@@ -273,7 +296,7 @@ public bool IsNextTokenContainAnimeType(int pos)
var prevToken = Token.FindPrevToken(_parser.Tokens, pos, Token.TokenFlag.FlagNotDelimiter);
if (!IsTokenCategory(prevToken, Token.TokenCategory.Bracket)) return false;
var nextToken = Token.FindNextToken(_parser.Tokens, pos, Token.TokenFlag.FlagNotDelimiter);
- if (nextToken < 0) return false;
+ if (!Token.InListRange(prevToken, _parser.Tokens) || !Token.InListRange(nextToken, _parser.Tokens)) return false;
return KeywordManager.Contains(Element.ElementCategory.ElementAnimeType, _parser.Tokens[nextToken].Content);
}
///
@@ -285,8 +308,8 @@ public bool IsPrevTokenContainAnimeType(int pos)
{
var prevToken = Token.FindPrevToken(_parser.Tokens, pos, Token.TokenFlag.FlagNotDelimiter);
var nextToken = Token.FindNextToken(_parser.Tokens, pos, Token.TokenFlag.FlagNotDelimiter);
+ if(!Token.InListRange(prevToken, _parser.Tokens)||!Token.InListRange(nextToken, _parser.Tokens)) return false;
if (!IsTokenCategory(nextToken, Token.TokenCategory.Bracket)) return false;
- if (prevToken < 0) return false;
return KeywordManager.Contains(Element.ElementCategory.ElementAnimeType, _parser.Tokens[prevToken].Content);
}
///
@@ -298,8 +321,8 @@ public bool IsPrevTokenContainAnimeTypeInPeekEntries(int pos)
{
var prevToken = Token.FindPrevToken(_parser.Tokens, pos, Token.TokenFlag.FlagNotDelimiter);
var nextToken = Token.FindNextToken(_parser.Tokens, pos, Token.TokenFlag.FlagNotDelimiter);
+ if (!Token.InListRange(prevToken, _parser.Tokens) || !Token.InListRange(nextToken, _parser.Tokens)) return false;
if (!IsTokenCategory(nextToken, Token.TokenCategory.Bracket)) return false;
- if (prevToken < 0) return false;
return KeywordManager.ContainsInPeekEntries(Element.ElementCategory.ElementAnimeType, _parser.Tokens[prevToken].Content);
}
@@ -435,4 +458,4 @@ public void BuildElement(Element.ElementCategory category, bool keepDelimiters,
}
}
}
-}
+}
\ No newline at end of file
diff --git a/AnitomySharp/ParserNumber.cs b/AnitomySharp/ParserNumber.cs
index de3174f..b3f5347 100644
--- a/AnitomySharp/ParserNumber.cs
+++ b/AnitomySharp/ParserNumber.cs
@@ -513,7 +513,8 @@ public bool MatchJapaneseCounterPattern(string word, Token token)
return true;
}
- regexPattern = @"([第全]?)([0-9一二三四五六七八九十壱弐参]+)([期章話话巻卷幕夜期発縛])";
+ // 全角数字：\uFF10-\uFF19
+ regexPattern = @"([第全]?)([0-9一二三四五六七八九十壱弐参\uFF10-\uFF19]+)([回集話话幕夜発縛])";
match = Regex.Match(word, RegexMatchOnlyStart + regexPattern + RegexMatchOnlyEnd, RegexOptions.IgnoreCase);
if (match.Success)
{
@@ -522,11 +523,33 @@ public bool MatchJapaneseCounterPattern(string word, Token token)
{
episodeNumber = ParserHelper.GetNumberFromOrdinal(episodeNumber);
}
+ episodeNumber = ParserHelper.GetNumberFromFullWidth(episodeNumber);
+ SetEpisodeNumber(episodeNumber, token, false);
+ return true;
+ }
+ regexPattern = @"([第全]?)([0-9一二三四五六七八九十壱弐参\uFF10-\uFF19]+)([期章巻卷])";
+ match = Regex.Match(word, RegexMatchOnlyStart + regexPattern + RegexMatchOnlyEnd, RegexOptions.IgnoreCase);
+ if (match.Success)
+ {
+ var episodeNumber = match.Groups[2].Value;
+ if (!StringHelper.IsNumericString(episodeNumber))
+ {
+ episodeNumber = ParserHelper.GetNumberFromOrdinal(episodeNumber);
+ }
+ episodeNumber = ParserHelper.GetNumberFromFullWidth(episodeNumber);
SetEpisodeNumber(episodeNumber, token, false);
return true;
}
- regexPattern = @"(vol|EPISODE|ACT|scene|ep|volume|screen|voice|case|menu|rail|round|game|page|collection|cage|office|doll|Princess)([ \.\-_])([0-9]+)";
+ regexPattern = @"(EPISODE|ACT|scene|ep|screen|voice|case|menu|rail|round|game|page|collection|cage|office|doll|Princess)([ \.\-_])([0-9]+)";
+ match = Regex.Match(word, RegexMatchOnlyStart + regexPattern + RegexMatchOnlyEnd, RegexOptions.IgnoreCase);
+ if (match.Success)
+ {
+ var episodeNumber = match.Groups[3].Value;
+ SetEpisodeNumber(episodeNumber, token, false);
+ return true;
+ }
+ regexPattern = @"(vol|volume)([ \.\-_])([0-9]+)";
match = Regex.Match(word, RegexMatchOnlyStart + regexPattern + RegexMatchOnlyEnd, RegexOptions.IgnoreCase);
if (match.Success)
{
@@ -874,4 +897,4 @@ public bool SearchForLastNumber(List tokens)
return false;
}
}
-}
+}
\ No newline at end of file
diff --git a/Jellyfin.Plugin.MetaShark.Test/ParseNameTest.cs b/Jellyfin.Plugin.MetaShark.Test/ParseNameTest.cs
index c8bd63a..83583f6 100644
--- a/Jellyfin.Plugin.MetaShark.Test/ParseNameTest.cs
+++ b/Jellyfin.Plugin.MetaShark.Test/ParseNameTest.cs
@@ -167,23 +167,36 @@ public void TestTVSeriesParse()
[TestMethod]
public void TestEposideParse()
{
+ // 普通数字
+ var fileName = "03.mp4";
+ var parseResult = NameParser.ParseEpisode(fileName);
+ Assert.AreEqual(parseResult.Name, "03");
+ Assert.AreEqual(parseResult.ParentIndexNumber, null);
+ Assert.AreEqual(parseResult.IndexNumber, 3);
+
+ fileName = "03 4K.mp4";
+ parseResult = NameParser.ParseEpisode(fileName);
+ Assert.AreEqual(parseResult.Name, "03");
+ Assert.AreEqual(parseResult.ParentIndexNumber, null);
+ Assert.AreEqual(parseResult.IndexNumber, 3);
+
// 混合中英文
- var fileName = "新世界.New.World.2013.BluRay.1080p.x265.10bit.MNHD-FRDS";
- var parseResult = NameParser.Parse(fileName);
+ fileName = "新世界.New.World.2013.BluRay.1080p.x265.10bit.MNHD-FRDS";
+ parseResult = NameParser.ParseEpisode(fileName);
Assert.AreEqual(parseResult.ChineseName, "新世界");
Assert.AreEqual(parseResult.Name, "New World");
Assert.AreEqual(parseResult.Year, 2013);
// 只英文 S01E01
fileName = "She-Hulk.Attorney.At.Law.S01E01.1080p.WEBRip.x265-RARBG";
- parseResult = NameParser.Parse(fileName);
+ parseResult = NameParser.ParseEpisode(fileName);
Assert.AreEqual(parseResult.Name, "She-Hulk Attorney At Law");
Assert.AreEqual(parseResult.ParentIndexNumber, 1);
Assert.AreEqual(parseResult.IndexNumber, 1);
// 测试 SXXEPXX 格式
fileName = "神探狄仁杰2 Detective.Dee.Ⅱ.S02EP02.2006.2160p.WEB-DL.x264.AAC-HQC";
- parseResult = NameParser.Parse(fileName);
+ parseResult = NameParser.ParseEpisode(fileName);
Assert.AreEqual(parseResult.ChineseName, "神探狄仁杰2");
Assert.AreEqual(parseResult.Name, "Detective Dee Ⅱ");
Assert.AreEqual(parseResult.ParentIndexNumber, 2);
@@ -192,26 +205,26 @@ public void TestEposideParse()
// 日文
fileName = "プロポーズ大作戦Ep05_x264.mp4";
- parseResult = NameParser.Parse(fileName);
+ parseResult = NameParser.ParseEpisode(fileName);
Assert.AreEqual(parseResult.Name, "プロポーズ大作戦Ep05");
Assert.AreEqual(parseResult.ParentIndexNumber, null);
Assert.AreEqual(parseResult.IndexNumber, 5);
fileName = "[01] [ANK-Raws] あっちこっち 01 (BDrip 1920x1080 HEVC-YUV420P10 FLAC)";
- parseResult = NameParser.Parse(fileName);
+ parseResult = NameParser.ParseEpisode(fileName);
Assert.AreEqual(parseResult.Name, "あっちこっち 01");
Assert.AreEqual(parseResult.ParentIndexNumber, null);
Assert.AreEqual(parseResult.IndexNumber, 1);
// 只中文
fileName = "齊天大聖 第02集";
- parseResult = NameParser.Parse(fileName);
- Assert.AreEqual(parseResult.Name, "齊天大聖 第02集");
+ parseResult = NameParser.ParseEpisode(fileName);
+ Assert.AreEqual(parseResult.Name, "齊天大聖");
Assert.AreEqual(parseResult.ParentIndexNumber, null);
Assert.AreEqual(parseResult.IndexNumber, 2);
fileName = "齊天大聖 第 02 期";
- parseResult = NameParser.Parse(fileName);
+ parseResult = NameParser.ParseEpisode(fileName);
Assert.AreEqual(parseResult.Name, "齊天大聖");
Assert.AreEqual(parseResult.ParentIndexNumber, null);
Assert.AreEqual(parseResult.IndexNumber, 2);
@@ -219,38 +232,40 @@ public void TestEposideParse()
// anime
fileName = "[YYDM-11FANS][THERMAE_ROMAE][02][BDRIP][720P][X264-10bit_AAC][7FF2269F]";
- parseResult = NameParser.Parse(fileName);
+ parseResult = NameParser.ParseEpisode(fileName);
Assert.AreEqual(parseResult.Name, "THERMAE ROMAE");
Assert.AreEqual(parseResult.ParentIndexNumber, null);
Assert.AreEqual(parseResult.IndexNumber, 2);
// anime带季数
fileName = "[WMSUB][Detective Conan - Zero‘s Tea Time ][S01][E06][BIG5][1080P].mp4";
- parseResult = NameParser.Parse(fileName);
+ parseResult = NameParser.ParseEpisode(fileName);
Assert.AreEqual(parseResult.Name, "Detective Conan - Zero‘s Tea Time");
Assert.AreEqual(parseResult.ParentIndexNumber, 1);
Assert.AreEqual(parseResult.IndexNumber, 6);
fileName = "[KTXP][Machikado_Mazoku_S2][01][BIG5][1080p]";
- parseResult = NameParser.Parse(fileName);
+ parseResult = NameParser.ParseEpisode(fileName);
Assert.AreEqual(parseResult.Name, "Machikado Mazoku");
Assert.AreEqual(parseResult.ParentIndexNumber, null);
Assert.AreEqual(parseResult.IndexNumber, 1);
fileName = "[異域字幕組][她和她的貓 - Everything Flows -][She and Her Cat - Everything Flows -][01][720p][繁體]";
- parseResult = NameParser.Parse(fileName);
+ parseResult = NameParser.ParseEpisode(fileName);
Assert.AreEqual(parseResult.Name, "她和她的貓 - Everything Flows");
Assert.AreEqual(parseResult.ParentIndexNumber, null);
Assert.AreEqual(parseResult.IndexNumber, 1);
// anime特典
fileName = "[KissSub][Steins;Gate][SP][GB_BIG5_JP][BDrip][1080P][HEVC] 边界曲面的缺失之环";
- parseResult = NameParser.Parse(fileName);
+ parseResult = NameParser.ParseEpisode(fileName);
Assert.IsTrue(parseResult.IsSpecial);
Assert.AreEqual(parseResult.Name, "边界曲面的缺失之环");
Assert.AreEqual(parseResult.ParentIndexNumber, null);
Assert.AreEqual(parseResult.IndexNumber, null);
+
+
}
diff --git a/Jellyfin.Plugin.MetaShark/Core/NameParser.cs b/Jellyfin.Plugin.MetaShark/Core/NameParser.cs
index e8bf797..394bf4b 100644
--- a/Jellyfin.Plugin.MetaShark/Core/NameParser.cs
+++ b/Jellyfin.Plugin.MetaShark/Core/NameParser.cs
@@ -108,7 +108,7 @@ public static ParseNameResult Parse(string fileName, bool isEpisode = false)
}
}
- // 假如Anitomy解析不到year，尝试使用jellyfin默认parser，看能不能解析成功
+ // 假如 Anitomy 解析不到 year，尝试使用 jellyfin 默认 parser，看能不能解析成功
if (parseResult.Year == null && !isAnime)
{
var nativeParseResult = ParseMovieByDefault(fileName);
@@ -118,13 +118,22 @@ public static ParseNameResult Parse(string fileName, bool isEpisode = false)
}
}
+ // 假如 Anitomy 解析不到集数，判断 name 是否是数字集号
+ if (parseResult.IndexNumber is null && isEpisode)
+ {
+ if (!string.IsNullOrEmpty(parseResult.Name) && parseResult.Name.IsNumericString())
+ {
+ parseResult.IndexNumber = parseResult.Name.ToInt();
+ }
+ }
+
// 修复纯中文集数/特殊标识集数
if (parseResult.IndexNumber is null)
{
parseResult.IndexNumber = ParseChineseOrSpecialIndexNumber(fileName);
}
- // 解析不到title时，或解析出多个title时，使用默认名
+ // 解析不到 title 时，或解析出多个 title 时，使用默认名
if (string.IsNullOrEmpty(parseResult.Name))
{
parseResult.Name = fileName;
@@ -133,6 +142,11 @@ public static ParseNameResult Parse(string fileName, bool isEpisode = false)
return parseResult;
}
+ public static ParseNameResult ParseEpisode(string fileName)
+ {
+ return Parse(fileName, true);
+ }
+
private static string CleanName(string name)
{
// 电视剧名称后紧跟季信息时，会附加到名称中，需要去掉
@@ -173,9 +187,11 @@ public static ParseNameResult ParseMovieByDefault(string fileName)
///
public static EpisodePathParserResult ParseEpisodeByDefault(string fileName)
{
+ // EpisodePathParser需要路径信息， 这里添加一个分隔符模拟路径
+ var path = Path.DirectorySeparatorChar + fileName;
var nameOptions = new Emby.Naming.Common.NamingOptions();
return new EpisodePathParser(nameOptions)
- .Parse(fileName, false);
+ .Parse(path, false);
}
diff --git a/Jellyfin.Plugin.MetaShark/Core/StringExtension.cs b/Jellyfin.Plugin.MetaShark/Core/StringExtension.cs
index f4cb7c0..21f6bf8 100644
--- a/Jellyfin.Plugin.MetaShark/Core/StringExtension.cs
+++ b/Jellyfin.Plugin.MetaShark/Core/StringExtension.cs
@@ -78,5 +78,10 @@ public static string GetMatchGroup(this string text, Regex reg)
return string.Empty;
}
+
+ public static bool IsNumericString(this string str)
+ {
+ return str.All(char.IsDigit);
+ }
}
}
diff --git a/Jellyfin.Plugin.MetaShark/Providers/EpisodeProvider.cs b/Jellyfin.Plugin.MetaShark/Providers/EpisodeProvider.cs
index 734dc96..b22ad44 100644
--- a/Jellyfin.Plugin.MetaShark/Providers/EpisodeProvider.cs
+++ b/Jellyfin.Plugin.MetaShark/Providers/EpisodeProvider.cs
@@ -135,7 +135,7 @@ public EpisodeInfo FixParseInfo(EpisodeInfo info)
{
// 使用AnitomySharp进行重新解析，解决anime识别错误
var fileName = Path.GetFileNameWithoutExtension(info.Path) ?? info.Name;
- var parseResult = NameParser.Parse(fileName);
+ var parseResult = NameParser.ParseEpisode(fileName);
info.Year = parseResult.Year;
info.Name = parseResult.ChineseName ?? parseResult.Name;
@@ -192,14 +192,14 @@ public EpisodeInfo FixParseInfo(EpisodeInfo info)
// info.ParentIndexNumber = 1;
// }
- // 特典优先使用文件名（特典除了前面特别设置，还有SXX/Season XX等默认的）
+ // 特典优先使用文件名（特典除了前面特别设置，还有 SXX/Season XX 等默认的）
if (info.ParentIndexNumber.HasValue && info.ParentIndexNumber == 0)
{
info.Name = parseResult.SpecialName == info.Name ? fileName : parseResult.SpecialName;
}
- // 大于1000，可能错误解析了分辨率
- if (parseResult.IndexNumber.HasValue && parseResult.IndexNumber < 1000 && info.IndexNumber != parseResult.IndexNumber)
+ // 修正 episode number
+ if (parseResult.IndexNumber.HasValue && info.IndexNumber != parseResult.IndexNumber)
{
this.Log("FixEpisodeNumber by anitomy. old: {0} new: {1}", info.IndexNumber, parseResult.IndexNumber);
info.IndexNumber = parseResult.IndexNumber;
@@ -214,7 +214,7 @@ public EpisodeInfo FixParseInfo(EpisodeInfo info)
{
// 特典或extra视频可能和正片剧集放在同一目录
var fileName = Path.GetFileNameWithoutExtension(info.Path) ?? info.Name;
- var parseResult = NameParser.Parse(fileName);
+ var parseResult = NameParser.ParseEpisode(fileName);
if (parseResult.IsExtra)
{
this.Log($"Found anime extra of [name]: {fileName}");