New: Improve Chinese language detection

This commit is contained in:
Wu Haotian 2019-11-14 22:02:13 +08:00 committed by Taloth
parent 098f9a2675
commit d45d9e356c
2 changed files with 2 additions and 1 deletions

View File

@ -108,6 +108,7 @@ namespace NzbDrone.Core.Test.ParserTests
[TestCase("[abc] My Series - 01 []")]
[TestCase("[abc] My Series - 01 []")]
[TestCase("[ABC字幕组] My Series - 01 [HDTV]")]
[TestCase("[喵萌奶茶屋&LoliHouse] / Kengan Ashura - 17 [WebRip 1080p HEVC-10bit AAC][]")]
public void should_parse_language_chinese(string postTitle)
{
var result = Parser.Parser.ParseTitle(postTitle);

View File

@ -17,7 +17,7 @@ namespace NzbDrone.Core.Parser
new RegexReplace(@".*?\.(S\d{2}E\d{2,4}\..*)", "$1", RegexOptions.Compiled | RegexOptions.IgnoreCase)
};
private static readonly Regex LanguageRegex = new Regex(@"(?:\W|_)(?<italian>\b(?:ita|italian)\b)|(?<german>german\b|videomann)|(?<flemish>flemish)|(?<greek>greek)|(?<french>(?:\W|_)(?:FR|VOSTFR)(?:\W|_))|(?<russian>\brus\b)|(?<dutch>nl\W?subs?)|(?<hungarian>\b(?:HUNDUB|HUN)\b)|(?<hebrew>\bHebDub\b)|(?<chinese>\[(?:CH[ST]|BIG5|GB)\]|简|繁|字幕)",
private static readonly Regex LanguageRegex = new Regex(@"(?:\W|_)(?<italian>\b(?:ita|italian)\b)|(?<german>german\b|videomann)|(?<flemish>flemish)|(?<greek>greek)|(?<french>(?:\W|_)(?:FR|VOSTFR)(?:\W|_))|(?<russian>\brus\b)|(?<dutch>nl\W?subs?)|(?<hungarian>\b(?:HUNDUB|HUN)\b)|(?<hebrew>\bHebDub\b)|(?<chinese>\[(?:CH[ST]|BIG5|GB)\]|简|繁|字幕)",
RegexOptions.IgnoreCase | RegexOptions.Compiled);
private static readonly Regex CaseSensitiveLanguageRegex = new Regex(@"(?<lithuanian>\bLT\b)|(?<czech>\bCZ\b)",