Parse another additional Chinese anime release format

This commit is contained in:
Mark McDowall 2022-05-29 10:15:48 -07:00
parent 2ea7b477cb
commit 93fc9abae9
2 changed files with 10 additions and 9 deletions

View File

@ -26,7 +26,6 @@ namespace NzbDrone.Core.Test.ParserTests
[TestCase("[星空字幕组] / Anime Series Title [05][1080p][]", "Anime Series Title", "", 5)] [TestCase("[星空字幕组] / Anime Series Title [05][1080p][]", "Anime Series Title", "", 5)]
[TestCase("【DHR动研字幕组】[多田君不恋爱_Anime Series Title][13][][720P][MP4]", "Anime Series Title", "DHR动研字幕组", 13)] [TestCase("【DHR动研字幕组】[多田君不恋爱_Anime Series Title][13][][720P][MP4]", "Anime Series Title", "DHR动研字幕组", 13)]
[TestCase("【动漫国字幕组】★01月新番[Anime Series Title][01][1080P][][MP4]", "Anime Series Title", "", 1)] [TestCase("【动漫国字幕组】★01月新番[Anime Series Title][01][1080P][][MP4]", "Anime Series Title", "", 1)]
[TestCase("[OPFans楓雪動漫][ONE PIECE ][1008][][1080P][MKV][]", "ONE PIECE", "OPFans", 1008)]
public void should_parse_chinese_anime_releases(string postTitle, string title, string subgroup, int absoluteEpisodeNumber) public void should_parse_chinese_anime_releases(string postTitle, string title, string subgroup, int absoluteEpisodeNumber)
{ {
postTitle = XmlCleaner.ReplaceUnicode(postTitle); postTitle = XmlCleaner.ReplaceUnicode(postTitle);
@ -45,6 +44,8 @@ namespace NzbDrone.Core.Test.ParserTests
[TestCase("[Lilith-Raws] II 簿 - Grace note- / Anime-Series Title - 04 [BiliBili][WEB-DL][1080p][AVC AAC][CHT][MKV]", "Anime-Series Title", "Lilith-Raws", 4)] [TestCase("[Lilith-Raws] II 簿 - Grace note- / Anime-Series Title - 04 [BiliBili][WEB-DL][1080p][AVC AAC][CHT][MKV]", "Anime-Series Title", "Lilith-Raws", 4)]
[TestCase("[NC-Raws] / Anime-Series Title - 07 [B-Global][WEB-DL][1080p][AVC AAC][CHS_CHT_ENG_TH_SRT][MKV]", "Anime-Series Title", "NC-Raws", 7)] [TestCase("[NC-Raws] / Anime-Series Title - 07 [B-Global][WEB-DL][1080p][AVC AAC][CHS_CHT_ENG_TH_SRT][MKV]", "Anime-Series Title", "NC-Raws", 7)]
[TestCase("[NC-Raws] ANIME-SERIES TITLE / Anime-Series Title - 07 [Baha][WEB-DL][1080p][AVC AAC][CHT][MP4]", "Anime-Series Title", "NC-Raws", 7)] [TestCase("[NC-Raws] ANIME-SERIES TITLE / Anime-Series Title - 07 [Baha][WEB-DL][1080p][AVC AAC][CHT][MP4]", "Anime-Series Title", "NC-Raws", 7)]
[TestCase("[OPFans楓雪動漫][ANIME SERIES ][1008][][1080P][MKV][]", "ANIME SERIES", "OPFans", 1008)]
[TestCase("[Skymoon-Raws][Anime Series ][1008][ViuTV][WEB-RIP][CHT][SRTx2][1080p][MKV]", "Anime Series", "Skymoon-Raws", 1008)]
public void should_parse_unbracketed_chinese_anime_releases(string postTitle, string title, string subgroup, int absoluteEpisodeNumber) public void should_parse_unbracketed_chinese_anime_releases(string postTitle, string title, string subgroup, int absoluteEpisodeNumber)
{ {
postTitle = XmlCleaner.ReplaceUnicode(postTitle); postTitle = XmlCleaner.ReplaceUnicode(postTitle);

View File

@ -22,6 +22,9 @@ namespace NzbDrone.Core.Parser
// Korean series without season number, replace with S01Exxx and remove airdate // Korean series without season number, replace with S01Exxx and remove airdate
new RegexReplace(@"\.E(\d{2,4})\.\d{6}\.(.*-NEXT)$", ".S01E$1.$2", RegexOptions.Compiled), new RegexReplace(@"\.E(\d{2,4})\.\d{6}\.(.*-NEXT)$", ".S01E$1.$2", RegexOptions.Compiled),
// Some Chinese anime releases contain both English and Chinese titles, remove the Chinese title and replace with normal anime pattern
new RegexReplace(@"^\[(?:(?<subgroup>[^\]]+?)(?:[\u4E00-\u9FCC]+)?)\]\[(?<title>[^\]]+?)(?:\s(?<chinesetitle>[\u4E00-\u9FCC][^\]]*?))\]\[(?:(?:[\u4E00-\u9FCC]+?)?(?<episode>\d{1,4})(?:[\u4E00-\u9FCC]+?)?)\]", "[${subgroup}] ${title} - ${episode} - ", RegexOptions.Compiled),
// Chinese LoliHouse/ZERO/Lilith-Raws releases don't use the expected brackets, normalize using brackets // Chinese LoliHouse/ZERO/Lilith-Raws releases don't use the expected brackets, normalize using brackets
new RegexReplace(@"^\[(?<subgroup>[^\]]*?(?:LoliHouse|ZERO|Lilith-Raws)[^\]]*?)\](?<title>[^\[\]]+?)(?: - (?<episode>[0-9-]+)\s*|\[第?(?<episode>[0-9]+(?:-[0-9]+)?)话?(?:END|完)?\])\[", "[${subgroup}][${title}][${episode}][", RegexOptions.Compiled), new RegexReplace(@"^\[(?<subgroup>[^\]]*?(?:LoliHouse|ZERO|Lilith-Raws)[^\]]*?)\](?<title>[^\[\]]+?)(?: - (?<episode>[0-9-]+)\s*|\[第?(?<episode>[0-9]+(?:-[0-9]+)?)话?(?:END|完)?\])\[", "[${subgroup}][${title}][${episode}][", RegexOptions.Compiled),
@ -29,10 +32,7 @@ namespace NzbDrone.Core.Parser
new RegexReplace(@"^\[(?<subgroup>[^\]]+)\](?:\s?★[^\[ -]+\s?)?\[?(?:(?<chinesetitle>[^\]]*?[\u4E00-\u9FCC][^\]]*?)(?:\]\[|\s*[_/·]\s*))?(?<title>[^\]]+?)\]?(?:\[\d{4}\])?\[第?(?<episode>[0-9]+(?:-[0-9]+)?)话?(?:END|完)?\]", "[${subgroup}] ${title} - ${episode} ", RegexOptions.Compiled), new RegexReplace(@"^\[(?<subgroup>[^\]]+)\](?:\s?★[^\[ -]+\s?)?\[?(?:(?<chinesetitle>[^\]]*?[\u4E00-\u9FCC][^\]]*?)(?:\]\[|\s*[_/·]\s*))?(?<title>[^\]]+?)\]?(?:\[\d{4}\])?\[第?(?<episode>[0-9]+(?:-[0-9]+)?)话?(?:END|完)?\]", "[${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 // 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), new RegexReplace(@"^\[(?<subgroup>[^\]]+)\](?:\s)(?:(?<chinesetitle>[^\]]*?[\u4E00-\u9FCC][^\]]*?)(?:\s/\s))(?<title>[^\]]+?)(?:[- ]+)(?<episode>[0-9]+(?:-[0-9]+)?)话?(?:END|完)?", "[${subgroup}] ${title} - ${episode} ", RegexOptions.Compiled)
// Some Chinese anime releases contain both English and Chinese titles, remove the Chinese title and replace with normal anime pattern
new RegexReplace(@"^\[(?:(?<subgroup>[^\]]+?)(?:[\u4E00-\u9FCC]+))\]\[(?<title>[^\]]+?)(?:\s(?<chinesetitle>[\u4E00-\u9FCC][^\]]*?))\]\[(?:(?:[\u4E00-\u9FCC]+?)(?<episode>\d{1,4})(?:[\u4E00-\u9FCC]+?))\]", "[${subgroup}] ${title} - ${episode} - ", RegexOptions.Compiled)
}; };
private static readonly Regex[] ReportTitleRegex = new[] private static readonly Regex[] ReportTitleRegex = new[]
@ -178,6 +178,10 @@ namespace NzbDrone.Core.Parser
new Regex(@"^(?<title>.+?)(?:\W+S(?<season>(?<!\d+)(?:\d{1,2})(?!\d+))\W+(?:(?:Part\W?|(?<!\d+\W+)e)(?<seasonpart>\d{1,2}(?!\d+)))+)", new Regex(@"^(?<title>.+?)(?:\W+S(?<season>(?<!\d+)(?:\d{1,2})(?!\d+))\W+(?:(?:Part\W?|(?<!\d+\W+)e)(?<seasonpart>\d{1,2}(?!\d+)))+)",
RegexOptions.IgnoreCase | RegexOptions.Compiled), RegexOptions.IgnoreCase | RegexOptions.Compiled),
// Anime - 4 digit absolute episode number
new Regex(@"^(?:\[(?<subgroup>.+?)\][-_. ]?)(?<title>.+?)[-_. ]+?(?<absoluteepisode>\d{4}(\.\d{1,2})?(?!\d+))",
RegexOptions.IgnoreCase | RegexOptions.Compiled),
//Anime - Title 4-digit Absolute Episode Number [SubGroup] //Anime - Title 4-digit Absolute Episode Number [SubGroup]
new Regex(@"^(?<title>.+?)[-_. ]+(?<absoluteepisode>(?<!\d+)\d{4}(?!\d+))[-_. ]\[(?<subgroup>.+?)\]", new Regex(@"^(?<title>.+?)[-_. ]+(?<absoluteepisode>(?<!\d+)\d{4}(?!\d+))[-_. ]\[(?<subgroup>.+?)\]",
RegexOptions.IgnoreCase | RegexOptions.Compiled), RegexOptions.IgnoreCase | RegexOptions.Compiled),
@ -256,10 +260,6 @@ namespace NzbDrone.Core.Parser
new Regex(@"^(?:\[(?<subgroup>.+?)\][-_. ]?)?(?<title>.+?)[-_. ]+?(?:Episode[-_. ]+?)(?<absoluteepisode>\d{1}(\.\d{1,2})?(?!\d+))", new Regex(@"^(?:\[(?<subgroup>.+?)\][-_. ]?)?(?<title>.+?)[-_. ]+?(?:Episode[-_. ]+?)(?<absoluteepisode>\d{1}(\.\d{1,2})?(?!\d+))",
RegexOptions.IgnoreCase | RegexOptions.Compiled), RegexOptions.IgnoreCase | RegexOptions.Compiled),
// Anime - 4 digit absolute episode number
new Regex(@"^(?:\[(?<subgroup>.+?)\][-_. ]?)(?<title>.+?)[-_. ]+?(?<absoluteepisode>\d{4}(\.\d{1,2})?(?!\d+))",
RegexOptions.IgnoreCase | RegexOptions.Compiled),
// Anime - Absolute episode number in square brackets // Anime - Absolute episode number in square brackets
new Regex(@"^(?:\[(?<subgroup>.+?)\][-_. ]?)(?<title>.+?)[-_. ]+?\[(?<absoluteepisode>\d{2,3}(\.\d{1,2})?(?!\d+))\]", new Regex(@"^(?:\[(?<subgroup>.+?)\][-_. ]?)(?<title>.+?)[-_. ]+?\[(?<absoluteepisode>\d{2,3}(\.\d{1,2})?(?!\d+))\]",
RegexOptions.IgnoreCase | RegexOptions.Compiled), RegexOptions.IgnoreCase | RegexOptions.Compiled),