Cutlery/Kavita
1
0
Fork 0
mirror of https://github.com/Kareadita/Kavita.git synced 2025-07-09 03:04:19 -04:00
Code Issues Packages Projects Releases Wiki Activity

More regex! Bonus is now a keyword for specials (#220)

Browse Source
This commit is contained in:
Joseph Milazzo 2021-05-16 13:13:19 -05:00 committed by GitHub
parent d1a7a4fd63
commit 2f793af34c
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 33 additions and 11 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

15
API.Tests/Parser/MangaParserTests.cs
View File

@ -53,7 +53,7 @@ namespace API.Tests.Parser
[InlineData("Kedouin Makoto - Corpse Party Musume, Chapter 12 [Dametrans][v2]", "0")] [InlineData("Kedouin Makoto - Corpse Party Musume, Chapter 12 [Dametrans][v2]", "0")]
[InlineData("Vagabond_v03", "3")] [InlineData("Vagabond_v03", "3")]
[InlineData("Mujaki No Rakune Volume 10.cbz", "10")] [InlineData("Mujaki No Rakune Volume 10.cbz", "10")]
[InlineData("Umineko no Naku Koro ni - Episode 3 - Banquet of the Golden Witch #02.cbz", "3")] [InlineData("Umineko no Naku Koro ni - Episode 3 - Banquet of the Golden Witch #02.cbz", "0")]
[InlineData("Volume 12 - Janken Boy is Coming!.cbz", "12")] [InlineData("Volume 12 - Janken Boy is Coming!.cbz", "12")]
[InlineData("[dmntsf.net] One Piece - Digital Colored Comics Vol. 20 Ch. 177 - 30 Million vs 81 Million.cbz", "20")] [InlineData("[dmntsf.net] One Piece - Digital Colored Comics Vol. 20 Ch. 177 - 30 Million vs 81 Million.cbz", "20")]
[InlineData("Gantz.V26.cbz", "26")] [InlineData("Gantz.V26.cbz", "26")]
@ -61,6 +61,9 @@ namespace API.Tests.Parser
[InlineData("[Hidoi]_Amaenaideyo_MS_vol01_chp02.rar", "1")] [InlineData("[Hidoi]_Amaenaideyo_MS_vol01_chp02.rar", "1")]
[InlineData("NEEDLESS_Vol.4_-_Simeon_6_v2_[SugoiSugoi].rar", "4")] [InlineData("NEEDLESS_Vol.4_-_Simeon_6_v2_[SugoiSugoi].rar", "4")]
[InlineData("Okusama wa Shougakusei c003 (v01) [bokuwaNEET]", "1")] [InlineData("Okusama wa Shougakusei c003 (v01) [bokuwaNEET]", "1")]
[InlineData("Sword Art Online Vol 10 - Alicization Running [Yen Press] [LuCaZ] {r2}.epub", "10")]
[InlineData("Noblesse - Episode 406 (52 Pages).7z", "0")]
[InlineData("X-Men v1 #201 (September 2007).cbz", "1")]
public void ParseVolumeTest(string filename, string expected) public void ParseVolumeTest(string filename, string expected)
{ {
Assert.Equal(expected, API.Parser.Parser.ParseVolume(filename)); Assert.Equal(expected, API.Parser.Parser.ParseVolume(filename));
@ -136,7 +139,11 @@ namespace API.Tests.Parser
[InlineData("Okusama wa Shougakusei c003 (v01) [bokuwaNEET]", "Okusama wa Shougakusei")] [InlineData("Okusama wa Shougakusei c003 (v01) [bokuwaNEET]", "Okusama wa Shougakusei")]
[InlineData("VanDread-v01-c001[MD].zip", "VanDread")] [InlineData("VanDread-v01-c001[MD].zip", "VanDread")]
[InlineData("Momo The Blood Taker - Chapter 027 Violent Emotion.cbz", "Momo The Blood Taker")] [InlineData("Momo The Blood Taker - Chapter 027 Violent Emotion.cbz", "Momo The Blood Taker")]
[InlineData("Kiss x Sis - Ch.15 - The Angst of a 15 Year Old Boy.cbz", "Kiss x Sis")]
[InlineData("Green Worldz - Chapter 112 Final Chapter (End).cbz", "Green Worldz")] [InlineData("Green Worldz - Chapter 112 Final Chapter (End).cbz", "Green Worldz")]
[InlineData("Noblesse - Episode 406 (52 Pages).7z", "Noblesse")]
[InlineData("X-Men v1 #201 (September 2007).cbz", "X-Men")]
[InlineData("Kodoja #001 (March 2016)", "Kodoja")]
public void ParseSeriesTest(string filename, string expected) public void ParseSeriesTest(string filename, string expected)
{ {
Assert.Equal(expected, API.Parser.Parser.ParseSeries(filename)); Assert.Equal(expected, API.Parser.Parser.ParseSeries(filename));
@ -197,7 +204,12 @@ namespace API.Tests.Parser
[InlineData("Kiss x Sis - Ch.00 - Let's Start from 0.cbz", "0")] [InlineData("Kiss x Sis - Ch.00 - Let's Start from 0.cbz", "0")]
[InlineData("[Hidoi]_Amaenaideyo_MS_vol01_chp02.rar", "2")] [InlineData("[Hidoi]_Amaenaideyo_MS_vol01_chp02.rar", "2")]
[InlineData("Okusama wa Shougakusei c003 (v01) [bokuwaNEET]", "3")] [InlineData("Okusama wa Shougakusei c003 (v01) [bokuwaNEET]", "3")]
[InlineData("Kiss x Sis - Ch.15 - The Angst of a 15 Year Old Boy.cbz", "15")]
[InlineData("Tomogui Kyoushitsu - Chapter 006 Game 005 - Fingernails On Right Hand (Part 002).cbz", "6")] [InlineData("Tomogui Kyoushitsu - Chapter 006 Game 005 - Fingernails On Right Hand (Part 002).cbz", "6")]
[InlineData("Noblesse - Episode 406 (52 Pages).7z", "406")]
[InlineData("X-Men v1 #201 (September 2007).cbz", "201")]
[InlineData("Kodoja #001 (March 2016)", "1")]
[InlineData("Noblesse - Episode 429 (74 Pages).7z", "429")]
public void ParseChaptersTest(string filename, string expected) public void ParseChaptersTest(string filename, string expected)
{ {
Assert.Equal(expected, API.Parser.Parser.ParseChapter(filename)); Assert.Equal(expected, API.Parser.Parser.ParseChapter(filename));
@ -226,6 +238,7 @@ namespace API.Tests.Parser
[InlineData("Corpse Party -The Anthology- Sachikos game of love Hysteric Birthday 2U Extra Chapter", true)] [InlineData("Corpse Party -The Anthology- Sachikos game of love Hysteric Birthday 2U Extra Chapter", true)]
[InlineData("Ani-Hina Art Collection.cbz", true)] [InlineData("Ani-Hina Art Collection.cbz", true)]
[InlineData("Gifting The Wonderful World With Blessings! - 3 Side Stories [yuNS][Unknown]", true)] [InlineData("Gifting The Wonderful World With Blessings! - 3 Side Stories [yuNS][Unknown]", true)]
[InlineData("A Town Where You Live - Bonus Chapter.zip", true)]
[InlineData("Yuki Merry - 4-Komga Anthology", true)] [InlineData("Yuki Merry - 4-Komga Anthology", true)]
public void ParseMangaSpecialTest(string input, bool expected) public void ParseMangaSpecialTest(string input, bool expected)
{ {

29
API/Parser/Parser.cs
View File

@ -52,10 +52,10 @@ namespace API.Parser
new Regex( new Regex(
@"(?<Series>.*)(\b|_|)(S(?<Volume>\d+))", @"(?<Series>.*)(\b|_|)(S(?<Volume>\d+))",
RegexOptions.IgnoreCase | RegexOptions.Compiled), RegexOptions.IgnoreCase | RegexOptions.Compiled),
// Umineko no Naku Koro ni - Episode 3 - Banquet of the Golden Witch #02.cbz // Umineko no Naku Koro ni - Episode 3 - Banquet of the Golden Witch #02.cbz (this case is too limited, most other cases shouldn't parse out)
new Regex( // new Regex(
@"(?<Series>.*)( |_|-)(?:Episode)(?: |_)(?<Volume>\d+(-\d+)?)", // @"(?<Series>.*)( |_|-)(?:Episode)(?: |_)(?<Volume>\d+(-\d+)?)",
RegexOptions.IgnoreCase | RegexOptions.Compiled), // RegexOptions.IgnoreCase | RegexOptions.Compiled),
}; };
@ -115,6 +115,10 @@ namespace API.Parser
new Regex( new Regex(
@"(?<Series>.*) (?<Chapter>\d+(?:.\d+|-\d+)?) \(\d{4}\)", @"(?<Series>.*) (?<Chapter>\d+(?:.\d+|-\d+)?) \(\d{4}\)",
RegexOptions.IgnoreCase | RegexOptions.Compiled), RegexOptions.IgnoreCase | RegexOptions.Compiled),
// Noblesse - Episode 429 (74 Pages).7z
new Regex(
@"(?<Series>.*)(\s|_)(?:Episode|Ep\.?)(\s|_)(?<Chapter>\d+(?:.\d+|-\d+)?)",
RegexOptions.IgnoreCase | RegexOptions.Compiled),
// Akame ga KILL! ZERO (2016-2019) (Digital) (LuCaZ) // Akame ga KILL! ZERO (2016-2019) (Digital) (LuCaZ)
new Regex( new Regex(
@"(?<Series>.*)\(\d", @"(?<Series>.*)\(\d",
@ -148,10 +152,7 @@ namespace API.Parser
new Regex( new Regex(
@"^(?!Vol\.?)(?<Series>.*)( |_|-)(?<!-)(episode ?)\d+-?\d*", @"^(?!Vol\.?)(?<Series>.*)( |_|-)(?<!-)(episode ?)\d+-?\d*",
RegexOptions.IgnoreCase | RegexOptions.Compiled), RegexOptions.IgnoreCase | RegexOptions.Compiled),
// Baketeriya ch01-05.zip, Akiiro Bousou Biyori - 01.jpg, Beelzebub_172_RHS.zip, Cynthia the Mission 29.rar
new Regex(
@"^(?!Vol\.?)(?<Series>.*)( |_|-)(?<!-)(ch)?\d+-?\d*",
RegexOptions.IgnoreCase | RegexOptions.Compiled),
// Baketeriya ch01-05.zip // Baketeriya ch01-05.zip
new Regex( new Regex(
@"^(?!Vol)(?<Series>.*)ch\d+-?\d?", @"^(?!Vol)(?<Series>.*)ch\d+-?\d?",
@ -164,6 +165,14 @@ namespace API.Parser
new Regex( new Regex(
@"^(?!Vol)(?<Series>.*)(-)\d+-?\d*", // This catches a lot of stuff ^(?!Vol)(?<Series>.*)( |_)(\d+) @"^(?!Vol)(?<Series>.*)(-)\d+-?\d*", // This catches a lot of stuff ^(?!Vol)(?<Series>.*)( |_)(\d+)
RegexOptions.IgnoreCase | RegexOptions.Compiled), RegexOptions.IgnoreCase | RegexOptions.Compiled),
// Kodoja #001 (March 2016)
new Regex(
@"(?<Series>.*)(\s|_|-)#",
RegexOptions.IgnoreCase | RegexOptions.Compiled),
// Baketeriya ch01-05.zip, Akiiro Bousou Biyori - 01.jpg, Beelzebub_172_RHS.zip, Cynthia the Mission 29.rar
new Regex(
@"^(?!Vol\.?)(?<Series>.*)( |_|-)(?<!-)(ch)?\d+-?\d*",
RegexOptions.IgnoreCase | RegexOptions.Compiled),
// [BAA]_Darker_than_Black_c1 (This is very greedy, make sure it's close to last) // [BAA]_Darker_than_Black_c1 (This is very greedy, make sure it's close to last)
new Regex( new Regex(
@"^(?!Vol)(?<Series>.*)( |_|-)(ch?)\d+", @"^(?!Vol)(?<Series>.*)( |_|-)(ch?)\d+",
@ -292,7 +301,7 @@ namespace API.Parser
{ {
// Historys Strongest Disciple Kenichi_v11_c90-98.zip, ...c90.5-100.5 // Historys Strongest Disciple Kenichi_v11_c90-98.zip, ...c90.5-100.5
new Regex( new Regex(
@"(c|ch)(\.? ?)(?<Chapter>(\d+(\.\d)?)-?(\d+(\.\d)?)?)", @"(\b|_)(c|ch)(\.?\s?)(?<Chapter>(\d+(\.\d)?)-?(\d+(\.\d)?)?)",
RegexOptions.IgnoreCase | RegexOptions.Compiled), RegexOptions.IgnoreCase | RegexOptions.Compiled),
// [Suihei Kiki]_Kasumi_Otoko_no_Ko_[Taruby]_v1.1.zip // [Suihei Kiki]_Kasumi_Otoko_no_Ko_[Taruby]_v1.1.zip
new Regex( new Regex(
@ -367,7 +376,7 @@ namespace API.Parser
{ {
// All Keywords, does not account for checking if contains volume/chapter identification. Parser.Parse() will handle. // All Keywords, does not account for checking if contains volume/chapter identification. Parser.Parse() will handle.
new Regex( new Regex(
@"(?<Special>Specials?|OneShot|One\-Shot|Omake|Extra( Chapter)?|Art Collection|Side( |_)Stories|(?<!The\s)Anthology)", @"(?<Special>Specials?|OneShot|One\-Shot|Omake|Extra( Chapter)?|Art Collection|Side( |_)Stories|(?<!The\s)Anthology|Bonus)",
RegexOptions.IgnoreCase | RegexOptions.Compiled), RegexOptions.IgnoreCase | RegexOptions.Compiled),
}; };