From 9a96dee49dbca348b14edf3272bed34ac62d0807 Mon Sep 17 00:00:00 2001 From: Mark McDowall Date: Tue, 13 Dec 2022 23:32:58 -0800 Subject: [PATCH] New: Parsing of releases extra slash in Chinese anime titles Closes #5299 --- .../ParserTests/UnicodeReleaseParserFixture.cs | 1 + src/NzbDrone.Core/Parser/Parser.cs | 2 +- 2 files changed, 2 insertions(+), 1 deletion(-) diff --git a/src/NzbDrone.Core.Test/ParserTests/UnicodeReleaseParserFixture.cs b/src/NzbDrone.Core.Test/ParserTests/UnicodeReleaseParserFixture.cs index 7509f1bcf..dd57fe402 100644 --- a/src/NzbDrone.Core.Test/ParserTests/UnicodeReleaseParserFixture.cs +++ b/src/NzbDrone.Core.Test/ParserTests/UnicodeReleaseParserFixture.cs @@ -28,6 +28,7 @@ namespace NzbDrone.Core.Test.ParserTests [TestCase("【动漫国字幕组】★01月新番[Anime Series Title～！][01][1080P][简体][MP4]", "Anime Series Title～！", "动漫国字幕组", 1)] [TestCase("[风车字幕组][名侦探柯南][857][米花町反复变化之谜（前篇）][简体][MP4][1080P]", "名侦探柯南", "风车字幕组", 857)] [TestCase("[风车字幕组][名侦探柯南][857集][米花町反复变化之谜（前篇）][简体][MP4][1080P]", "名侦探柯南", "风车字幕组", 857)] + [TestCase("【喵萌奶茶屋】★10月新番★[后宫之乌/后宫の乌/Series Title][07][1080p][简日双语][招募翻译校对]", "Series Title", "喵萌奶茶屋", 7)] public void should_parse_chinese_anime_releases(string postTitle, string title, string subgroup, int absoluteEpisodeNumber) { postTitle = XmlCleaner.ReplaceUnicode(postTitle); diff --git a/src/NzbDrone.Core/Parser/Parser.cs b/src/NzbDrone.Core/Parser/Parser.cs index 1d536bb2f..6defb7944 100644 --- a/src/NzbDrone.Core/Parser/Parser.cs +++ b/src/NzbDrone.Core/Parser/Parser.cs @@ -29,7 +29,7 @@ namespace NzbDrone.Core.Parser new RegexReplace(@"^\[(?[^\]]*?(?:LoliHouse|ZERO|Lilith-Raws)[^\]]*?)\](?[^\[\]]+?)(?: - (?<episode>[0-9-]+)\s*|\[第?(?<episode>[0-9]+(?:-[0-9]+)?)话?(?:END|完)?\])\[", "[${subgroup}][${title}][${episode}][", RegexOptions.Compiled), // Most Chinese anime releases contain additional brackets/separators for chinese and non-chinese titles, remove junk and replace with normal anime pattern - new RegexReplace(@"^\[(?<subgroup>[^\]]+)\](?:\s?★[^\[ -]+\s?)?\[?(?:(?<chinesetitle>(?=[^\]]*?[\u4E00-\u9FCC])[^\]]*?)(?:\]\[|\s*[_/·]\s*))?(?<title>[^\]]+?)\]?(?:\[\d{4}\])?\[第?(?<episode>[0-9]+(?:-[0-9]+)?)(?:话|集)?(?: ?END|完| ?Fin)?\]", "[${subgroup}] ${title} - ${episode} ", RegexOptions.Compiled), + new RegexReplace(@"^\[(?<subgroup>[^\]]+)\](?:\s?★[^\[ -]+\s?)?\[?(?:(?<chinesetitle>(?=[^\]]*?[\u4E00-\u9FCC])[^\]]*?)(?:\]\[|\s*[_/·]\s*)){0,2}(?<title>[^\]]+?)\]?(?:\[\d{4}\])?\[第?(?<episode>[0-9]+(?:-[0-9]+)?)(?:话|集)?(?: ?END|完| ?Fin)?\]", "[${subgroup}] ${title} - ${episode} ", RegexOptions.Compiled), // Some Chinese anime releases contain both Chinese and English titles, remove the Chinese title and replace with normal anime pattern new RegexReplace(@"^\[(?<subgroup>[^\]]+)\](?:\s)(?:(?<chinesetitle>(?=[^\]]*?[\u4E00-\u9FCC])[^\]]*?)(?:\s/\s))(?<title>[^\]]+?)(?:[- ]+)(?<episode>[0-9]+(?:-[0-9]+)?)话?(?:END|完)?", "[${subgroup}] ${title} - ${episode} ", RegexOptions.Compiled),