tweak(anitomy): update version v0.4.0

AnitomySharp/AnitomySharp.csproj

@@ -4,23 +4,25 @@
         <TargetFramework>net6.0</TargetFramework>
         <GeneratePackageOnBuild>true</GeneratePackageOnBuild>
         <PackageId>AnitomySharp.NET6</PackageId>
-        <Authors>tabratton;senritsu</Authors>
+        <PackageVersion>0.4.0</PackageVersion>
+        <Version>0.4.0</Version>
+        <Authors>tabratton;senritsu;chu-shen</Authors>
         <Description>AnitomySharp is a C# port of Anitomy by erengy, a library for parsing anime video filenames. All credit to erengy for the actual library and logic.
+        This fork of AnitomySharp is inspired by tabratton and senritsu, which adds more custom rules.
         </Description>
         <RepositoryUrl>https://github.com/chu-shen/AnitomySharp.git</RepositoryUrl>
         <RepositoryType>git</RepositoryType>
-        <PackageLicenseFile>LICENSE</PackageLicenseFile>
+        <PackageTags>Anitomy Anime</PackageTags>
         <PackageRequireLicenseAcceptance>true</PackageRequireLicenseAcceptance>
-        <AssemblyVersion>0.3.0</AssemblyVersion>
+        <PackageLicenseFile>LICENSE</PackageLicenseFile>
-        <FileVersion>0.3.0</FileVersion>
+        <PackageReadmeFile>README.md</PackageReadmeFile>
-        <Version>0.3.0</Version>
-        <GenerateDocumentationFile>false</GenerateDocumentationFile>
         <DocumentationFile>AnitomySharp.xml</DocumentationFile>
+        <GenerateDocumentationFile>false</GenerateDocumentationFile>
     </PropertyGroup>
 
     <ItemGroup>
         <None Include="..\LICENSE" Pack="true" Visible="false" PackagePath="" />
-        <PackageReference Include="Microsoft.DocAsCode.App" Version="2.60.0" />
+        <None Include="..\README.md" Pack="true" PackagePath=""/>
     </ItemGroup>
 
 </Project>

AnitomySharp/Keyword.cs

@@ -24,13 +24,13 @@ namespace AnitomySharp
     public static class KeywordManager
     {
         /// <summary>
-        /// 包含所有关键词（大写）的内部关键词元素词典
+        /// 包含所有关键词的内部关键词元素词典，比较器忽略大小写
         /// </summary>
-        private static readonly Dictionary<string, Keyword> Keys = new Dictionary<string, Keyword>();
+        private static readonly Dictionary<string, Keyword> Keys = new Dictionary<string, Keyword>(StringComparer.OrdinalIgnoreCase);
         /// <summary>
-        /// 文件扩展名，无值
+        /// 文件扩展名，无值，比较器忽略大小写
         /// </summary>
-        private static readonly Dictionary<string, Keyword> Extensions = new Dictionary<string, Keyword>();
+        private static readonly Dictionary<string, Keyword> Extensions = new Dictionary<string, Keyword>(StringComparer.OrdinalIgnoreCase);
 
         /// <summary>
         /// ~~一眼真~~
@@ -64,18 +64,19 @@ namespace AnitomySharp
               "GEKIJOUBAN", "MOVIE",
               "OAD", "OAV", "ONA", "OVA",
               "TV",
-              "番外編", "總集編","映像特典","特典","特典アニメ",
+              "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",  
+              "SPECIAL", "SPECIALS", "SP", "SPs",  
               // 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
               "MV"});
 
-            //       add "SP" to ElementAnimeType with optionsUnidentifiable
+            // add "SP" to ElementAnimeType with optionsUnidentifiable
-            //       Add(Element.ElementCategory.ElementAnimeType,
+            // Add(Element.ElementCategory.ElementAnimeType,
-            //         optionsUnidentifiableUnsearchable,
+            //   optionsUnidentifiableUnsearchable,
-            //         new List<string> {"SP"}); // e.g. "Yumeiro Patissiere SP Professional"
+            //   new List<string> { "SP" }); // e.g. "Yumeiro Patissiere SP Professional", but it is widely used to represent special
 
             Add(Element.ElementCategory.ElementAnimeType,
               optionsUnidentifiableInvalid,
@@ -84,7 +85,7 @@ namespace AnitomySharp
               // 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",
+              "PREVIEW", "YOKOKU",
               // 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
@@ -92,7 +93,7 @@ namespace AnitomySharp
               // 语音信息
               "MESSAGE",
               // 宣传片/预告片 Promotion Video / Trailer 一般时长在 1~2min 命名参考原盘和 jsum
-              "PV", "Teaser","TRAILER", "DRAMA",
+              "PV", "Teaser","TRAILER",
               // 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。
               "INTERVIEW",
               "EVENT", "TOKUTEN", "LOGO"});
@@ -150,7 +151,7 @@ namespace AnitomySharp
 
             Add(Element.ElementCategory.ElementOther,
               optionsDefault,
-              new List<string> { "REMASTER", "REMASTERED", "UNCUT", "TS", "VFR", "WIDESCREEN", "WS", "SPURSENGINE" });
+              new List<string> { "REMASTER", "REMASTERED", "UNCUT", "TS", "VFR", "WIDESCREEN", "WS", "SPURSENGINE","DISC" });
 
             Add(Element.ElementCategory.ElementReleaseGroup,
               optionsDefault,
@@ -281,6 +282,16 @@ namespace AnitomySharp
 
             return false;
         }
+        /// <summary>
+        /// 判断预处理元素列表中是否包含给定的字符串(<paramref name="keyword"/>)
+        /// </summary>
+        /// <param name="category">元素类别</param>
+        /// <param name="keyword">待判断的字符串</param>
+        /// <returns>`true`表示包含</returns>
+        public static bool ContainsInPeekEntries(Element.ElementCategory category, string keyword)
+        {
+            return PeekEntries.Any(entry => entry.Item1 == category && entry.Item2.Contains(keyword, StringComparer.OrdinalIgnoreCase));
+        }
 
         /// <summary>
         /// Finds a particular <c>keyword</c>. If found sets <c>category</c> and <c>options</c> to the found search result.

AnitomySharp/Parser.cs

@@ -179,6 +179,7 @@ namespace AnitomySharp
         private void SearchForEpisodeNumber()
         {
             var tokens = new List<int>();
+            var allTokens = new List<int>();
             for (var i = 0; i < Tokens.Count; i++)
             {
                 var token = Tokens[i];
@@ -187,6 +188,7 @@ namespace AnitomySharp
                     ParserHelper.IndexOfFirstDigit(token.Content) != -1)
                 {
                     tokens.Add(i);
+                    allTokens.Add(i);
                 }
             }
 
@@ -228,6 +230,12 @@ namespace AnitomySharp
             // "e.g. "[12]", "(2006)"
             if (ParseNumber.SearchForIsolatedNumbers(tokens)) return;
 
+            // e.g. "OVA 3", "OtherToken[Hint05]", "[Web Preview 06]": maybe incorrect, so put the last
+            if (ParseNumber.SearchForSymbolWithEpisode(allTokens)) return;
+
+            // e.g. [13(341)], [13 (341)]
+            if (ParseNumber.SearchForEquivalentNumbersWithBracket(allTokens)) return;
+
             // Consider using the last number as a last resort
             ParseNumber.SearchForLastNumber(tokens);
         }
@@ -235,7 +243,7 @@ namespace AnitomySharp
         /// <summary>
         /// Search for anime title
         /// 
-        /// 搜索动画名
+        /// 搜索动画名 
         /// </summary>
         private void SearchForAnimeTitle()
         {
@@ -283,6 +291,13 @@ namespace AnitomySharp
                     {
                         tokenBegin = tokenBeginWithNoReleaseGroup;
                     }
+                    // 去除纯数字标题
+                    // skip token with only number
+                    if (Regex.Match(Tokens[tokenBegin].Content, ParserNumber.RegexMatchOnlyStart + @"^[0-9]+$" + ParserNumber.RegexMatchOnlyEnd).Success)
+                    {
+                        tokenBegin = tokenBeginWithNoReleaseGroup;
+                    }
+
                     skippedPreviousGroup = true;
                 } while (Token.InListRange(tokenBegin, Tokens));
             }
@@ -398,7 +413,7 @@ namespace AnitomySharp
             {
                 var token = Tokens[i];
                 /** 跳过括号标记类型的标记 */
-                if (token.Category == Token.TokenCategory.Bracket) continue;
+                if (token.Category != Token.TokenCategory.Unknown) continue;
                 var tokenContent = token.Content;
 
                 // e.g. "2016-17"
@@ -408,13 +423,21 @@ namespace AnitomySharp
                 {
                     tokenContent = tokenContent.Split(match.Groups[2].Value)[0];
                 }
-                // add newtype e.g. "2021 OVA"
+
-                if (token.Category != Token.TokenCategory.Unknown || !StringHelper.IsNumericString(tokenContent) ||
+                if (!StringHelper.IsNumericString(tokenContent))
-                    !(ParseHelper.IsTokenContainAnimeType(i) ^ ParseHelper.IsTokenIsolated(i)))
                 {
                     continue;
                 }
 
+                // e.g. "[2021 OVA]"
+                if(ParseHelper.IsNextTokenContainAnimeType(i)&&!ParseHelper.IsTokenIsolated(i)){}
+
+                // TODO may not be necessary
+                // if (!ParseHelper.IsTokenIsolated(i))
+                // {
+                //     continue;
+                // }
+
                 var number = StringHelper.StringToInt(tokenContent);
 
                 // Anime year
@@ -422,7 +445,7 @@ namespace AnitomySharp
                 {
                     if (Empty(Element.ElementCategory.ElementAnimeYear))
                     {
-                        Elements.Add(new Element(Element.ElementCategory.ElementAnimeYear, token.Content));
+                        Elements.Add(new Element(Element.ElementCategory.ElementAnimeYear, tokenContent));
                         token.Category = Token.TokenCategory.Identifier;
                         continue;
                     }

AnitomySharp/ParserHelper.cs

@@ -235,7 +235,7 @@ namespace AnitomySharp
         /// <summary>
         /// Returns whether or not a token at the current <c>pos</c> is isolated(surrounded by braces).
         /// 
-        /// 判断当前位置标记(token)是否孤立，即是否被括号包裹
+        /// 判断当前位置标记(token)是否孤立，是否被括号包裹
         /// </summary>
         /// <param name="pos"></param>
         /// <returns></returns>
@@ -246,6 +246,20 @@ namespace AnitomySharp
             var nextToken = Token.FindNextToken(_parser.Tokens, pos, Token.TokenFlag.FlagNotDelimiter);
             return IsTokenCategory(nextToken, Token.TokenCategory.Bracket);
         }
+        /// <summary>
+        /// Returns whether or not a token at the current <c>pos</c> is isolated(surrounded by braces, delimiter).
+        /// 
+        /// 判断当前位置标记(token)是否孤立，前面是否为分隔符，后面是否为括号包裹
+        /// </summary>
+        /// <param name="pos"></param>
+        /// <returns></returns>
+        public bool IsTokenIsolatedWithDelimiterAndBracket(int pos)
+        {
+            var prevToken = Token.FindPrevToken(_parser.Tokens, pos, Token.TokenFlag.FlagNone);
+            if (!IsTokenCategory(prevToken, Token.TokenCategory.Delimiter)) return false;
+            var nextToken = Token.FindNextToken(_parser.Tokens, pos, Token.TokenFlag.FlagNotDelimiter);
+            return IsTokenCategory(nextToken, Token.TokenCategory.Bracket);
+        }
 
         /// <summary>
         /// Returns whether or not a token at the current <c>pos+1</c> is ElementAnimeType.
@@ -254,13 +268,37 @@ namespace AnitomySharp
         /// </summary>
         /// <param name="pos"></param>
         /// <returns></returns>
-        public bool IsTokenContainAnimeType(int pos)
+        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);
             return KeywordManager.Contains(Element.ElementCategory.ElementAnimeType, _parser.Tokens[nextToken].Content);
         }
+        /// <summary>
+        /// 判断当前标记(token)的上一个标记的类型是否为ElementAnimeType。如果是，则返回`true`
+        /// </summary>
+        /// <param name="pos"></param>
+        /// <returns></returns>
+        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 (!IsTokenCategory(nextToken, Token.TokenCategory.Bracket)) return false;
+            return KeywordManager.Contains(Element.ElementCategory.ElementAnimeType, _parser.Tokens[prevToken].Content);
+        }
+        /// <summary>
+        /// 判断当前标记(token)的上一个标记的类型是否为ElementAnimeType（在 PeekEntries 中）。如果是，则返回`true`
+        /// </summary>
+        /// <param name="pos"></param>
+        /// <returns></returns>
+        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 (!IsTokenCategory(nextToken, Token.TokenCategory.Bracket)) return false;
+            return KeywordManager.ContainsInPeekEntries(Element.ElementCategory.ElementAnimeType, _parser.Tokens[prevToken].Content);
+        }
 
         /// <summary>
         /// Finds and sets the anime season keyword.

AnitomySharp/ParserNumber.cs

@@ -412,7 +412,7 @@ namespace AnitomySharp
             _parser.Tokens.Insert(foundIdx,
               new Token(options.Identifiable ? Token.TokenCategory.Identifier : Token.TokenCategory.Unknown, token.Enclosed, prefix));
 
-            return true;
+                return true;
 
         }
 
@@ -698,6 +698,50 @@ namespace AnitomySharp
             return false;
         }
 
+        /// <summary>
+        /// 搜索同动画类型同时出现的集数
+        /// </summary>
+        /// <param name="tokens"></param>
+        /// <returns></returns>
+        public bool SearchForSymbolWithEpisode(List<int> tokens)
+        {
+            // Match from back to front
+            for (int i = tokens.Count - 1; i >= 0; i--)
+            {
+                var it = tokens[i];
+
+                // e.g. OVA 3, [Web Preview 06]: Web Preview in PeekEntries
+                if ((_parser.ParseHelper.IsPrevTokenContainAnimeType(it) || _parser.ParseHelper.IsPrevTokenContainAnimeTypeInPeekEntries(it)) && !_parser.ParseHelper.IsTokenIsolated(it))
+                {
+                    SetEpisodeNumber(_parser.Tokens[it].Content, _parser.Tokens[it], false);
+                    return true;
+                }
+                // e.g. OtherToken[Hint05]
+                // it>1: makesure this token is not first one
+                if (it > 1 && _parser.Tokens[it].Enclosed && _parser.ParseHelper.IsTokenIsolated(it))
+                {
+                    var tokenContent = _parser.Tokens[it].Content;
+                    var numberBegin = ParserHelper.IndexOfFirstDigit(tokenContent);
+                    var prefix = StringHelper.SubstringWithCheck(tokenContent, 0, numberBegin);
+                    var number = StringHelper.SubstringWithCheck(tokenContent, numberBegin, tokenContent.Length - numberBegin);
+                    // token should be: alphaNumeric
+                    if (prefix != "" && StringHelper.IsAlphaString(prefix) && StringHelper.IsNumericString(number))
+                    {
+                        SetEpisodeNumber(number, _parser.Tokens[it], true);
+                        return true;
+                    }
+                }
+                // e.g. OtherToken[Disc 01]
+                if (it > 1 && _parser.Tokens[it].Enclosed && _parser.ParseHelper.IsTokenIsolatedWithDelimiterAndBracket(it) && StringHelper.IsNumericString(_parser.Tokens[it].Content))
+                {
+                    SetEpisodeNumber(_parser.Tokens[it].Content, _parser.Tokens[it], true);
+                    return true;
+                }
+            }
+
+            return false;
+        }
+
         /// <summary>
         /// Searches for equivalent number in a list of <c>tokens</c>. e.g. 08(114)
         /// 
@@ -730,10 +774,7 @@ namespace AnitomySharp
                     continue;
                 }
 
-                var list = new List<Token>
+                var list = new List<Token> { _parser.Tokens[it], _parser.Tokens[nextToken] };
-        {
-          _parser.Tokens[it], _parser.Tokens[nextToken]
-        };
 
                 list.Sort((o1, o2) => StringHelper.StringToInt(o1.Content) - StringHelper.StringToInt(o2.Content));
                 SetEpisodeNumber(list[0].Content, list[0], false);
@@ -743,6 +784,50 @@ namespace AnitomySharp
 
             return false;
         }
+        /// <summary>
+        /// Searches for equivalent number in a list of <c>tokens</c>. e.g. 08(114)
+        /// 
+        /// 匹配自带等效集数的数字，常见于分割放送，匹配括号包裹的数字
+        /// </summary>
+        /// <param name="tokens">the list of tokens</param>
+        /// <returns>true if an equivalent number was found</returns>
+        public bool SearchForEquivalentNumbersWithBracket(List<int> tokens)
+        {
+            foreach (var it in tokens)
+            {
+                // Find the first enclosed, non-delimiter token
+                var nextToken = Token.FindNextToken(_parser.Tokens, it, Token.TokenFlag.FlagNotDelimiter);
+                if (!Token.InListRange(nextToken, _parser.Tokens) || !(_parser.Tokens[it].Content.Contains("(") || _parser.Tokens[nextToken].Content.Contains(")")))
+                {
+                    continue;
+                }
+
+                // e.g. [13(341)]
+                if (it > 1 && _parser.Tokens[it].Enclosed && _parser.ParseHelper.IsTokenIsolated(it))
+                {
+                    string[] episodes = _parser.Tokens[it].Content.Split(new string[] { "(", ")" }, StringSplitOptions.RemoveEmptyEntries);
+                    if (StringHelper.IsNumericString(episodes[0]) && StringHelper.IsNumericString(episodes[1]))
+                    {
+                        SetEpisodeNumber(episodes[0], _parser.Tokens[it], false);
+                        SetAlternativeEpisodeNumber(episodes[1], _parser.Tokens[it]);
+                        return true;
+                    }
+                }
+
+                // e.g. [13 (341)]
+                if (it > 1 && _parser.Tokens[nextToken].Enclosed && _parser.ParseHelper.IsTokenIsolatedWithDelimiterAndBracket(nextToken))
+                {
+                    string episode = _parser.Tokens[nextToken].Content.Replace("(", "").Replace(")", "");
+                    if (StringHelper.IsNumericString(_parser.Tokens[it].Content) && StringHelper.IsNumericString(episode))
+                    {
+                        SetEpisodeNumber(_parser.Tokens[it].Content, _parser.Tokens[it], true);
+                        SetAlternativeEpisodeNumber(episode, _parser.Tokens[nextToken]);
+                        return true;
+                    }
+                }
+            }
+            return false;
+        }
 
         /// <summary>
         /// Searches for the last number token in a list of <c>tokens</c>

AnitomySharp/StringHelper.cs

@@ -120,6 +120,17 @@ namespace AnitomySharp
         {
             return str.All(char.IsDigit);
         }
+        /// <summary>
+        /// Returns whether or not the <c>str</c> is a alpha string.
+        /// 
+        /// 判断字符串是否全字母
+        /// </summary>
+        /// <param name="str"></param>
+        /// <returns></returns>
+        public static bool IsAlphaString(string str)
+        {
+            return str.All(char.IsLetter);
+        }
 
         /// <summary>
         /// Returns the int value of the <c>str</c>; 0 otherwise.