1
0
Fork 0
mirror of https://github.com/lidarr/Lidarr synced 2024-12-27 01:57:21 +00:00

More episode parsing tests.

Added new regex to find multi-episode files without series titles.
Cleaned up some of the regex.
Normalizing regex will remove all numbers except years that are in the 1900's or 2000's
This commit is contained in:
Mark McDowall 2011-06-13 11:25:40 -07:00
parent ad89618f58
commit e71a55a33c
2 changed files with 37 additions and 22 deletions

View file

@ -17,10 +17,6 @@ public class ParserTest : TestBase
* Unreported.World.Chinas.Lost.Sons.WS.PDTV.XviD-FTP
*/
[Test]
[TestCase("Sonny.With.a.Chance.S02E15", "Sonny.With.a.Chance", 2, 15)]
[TestCase("Two.and.a.Half.Me.103.720p.HDTV.X264-DIMENSION", "Two.and.a.Half.Me", 1, 3)]
@ -137,6 +133,10 @@ public void parsing_our_own_quality_enum()
[TestCase("White.Collar.2x04.2x05.720p.BluRay-FUTV", "White.Collar", 2, new[] { 4, 5 }, 2)]
[TestCase("Desperate.Housewives.S07E22E23.720p.HDTV.X264-DIMENSION", "Desperate.Housewives", 7, new[] { 22,23 }, 2)]
//[Row("The.Kennedys.Part.1.and.Part.2.DSR.XviD-SYS", 1, new[] { 1, 2 })]
[TestCase("S07E22 - 7x23 - And Lots of Security.. [HDTV].mkv", "", 7, new[] { 22, 23 }, 2)]
[TestCase("Desparate Housewives - S07E22 - 7x23 - And Lots of Security.. [HDTV].mkv", "Desparate Housewives", 7, new[] { 22, 23 }, 2)]
[TestCase("Desparate Housewives - S07E22 - S07E23 - And Lots of Security.. [HDTV].mkv", "Desparate Housewives", 7, new[] { 22, 23 }, 2)]
[TestCase("S03E01.S03E02.720p.HDTV.X264-DIMENSION", "", 3, new[] { 1, 2 }, 2)]
public void episode_multipart_parse(string postTitle, string title, int season, int[] episodes, int count)
{
var result = Parser.ParseEpisodeInfo(postTitle);

View file

@ -14,21 +14,36 @@ public static class Parser
private static readonly Regex[] ReportTitleRegex = new[]
{
//Episodes with airdate
new Regex(@"^(?<title>.+?)?\W*(?<airyear>\d{4})\W+(?<airmonth>\d{2})\W+(?<airday>\d{2})\W?(?!\\)",
RegexOptions.IgnoreCase | RegexOptions.Compiled),
new Regex(@"^(?<title>.+?)(?:\WS?(?<season>\d{1,2}(?!\d+))(?:(?:\-|\.|[ex]|\s|\sto\s){1,2}(?<episode>\d{1,2}(?!\d+)))+)+\W?(?!\\)",
//Multi-Part episodes without a title (S01E05.S01E06)
new Regex(@"^(?:\W*S?(?<season>\d{1,2}(?!\d+))(?:(?:\-|\.|[ex]|\s|\sto\s){1,2}(?<episode>\d{1,2}(?!\d+)))+){2,}\W?(?!\\)",
RegexOptions.IgnoreCase | RegexOptions.Compiled),
new Regex(@"^(?<title>.*?)?(?:\W?S?(?<season>\d{1,2}(?!\d+))(?:(?:\-|\.|[ex]|\s|\sto\s){1,2}(?<episode>\d{1,2}(?!\d+)))+)+\W?(?!\\)",
RegexOptions.IgnoreCase | RegexOptions.Compiled),
//Single episodes or multi-episode (S01E05E06, S01E05-06, etc)
new Regex(@"^(?<title>.+?)(?:\W+S?(?<season>\d{1,2}(?!\d+))(?:(?:\-|\.|[ex]|\s|\sto\s){1,2}(?<episode>\d{1,2}(?!\d+)))+)+\W?(?!\\)",
RegexOptions.IgnoreCase | RegexOptions.Compiled),
//No Title - Single episodes or multi-episode (S01E05E06, S01E05-06, etc)
new Regex(@"^(?:\W?S?(?<season>\d{1,2}(?!\d+))(?:(?:\-|\.|[ex]|\s|\sto\s){1,2}(?<episode>\d{1,2}(?!\d+)))+\W*)+\W?(?!\\)",
RegexOptions.IgnoreCase | RegexOptions.Compiled),
//Supports 103/113 naming
new Regex(@"^(?<title>.+?)?\W?(?:\W(?<season>\d+)(?<episode>\d{2}))+\W?(?!\\)",
RegexOptions.IgnoreCase | RegexOptions.Compiled), //Supports 103/113 naming
new Regex(@"^(?<title>.*?)?(?:\W?S?(?<season>\d{1,2}(?!\d+))(?:(?:\-|\.|[ex]|\s|to)+(?<episode>\d+))+)+\W?(?!\\)",
RegexOptions.IgnoreCase | RegexOptions.Compiled),
//Episodes over 99 (3-digits or more)
new Regex(@"^(?<title>.*?)(?:\W?S?(?<season>\d{1,2}(?!\d+))(?:(?:\-|\.|[ex]|\s|to)+(?<episode>\d+))+)+\W?(?!\\)",
RegexOptions.IgnoreCase | RegexOptions.Compiled),
//Supports Season only releases
new Regex(@"^(?<title>.*?)\W(?:S|Season\W?)?(?<season>\d{1,2}(?!\d+))+\W?(?!\\)",
RegexOptions.IgnoreCase | RegexOptions.Compiled) //Supports Season only releases
RegexOptions.IgnoreCase | RegexOptions.Compiled)
};
private static readonly Regex NormalizeRegex = new Regex(@"((^|\W)(a|an|the|and|or|of)($|\W))|\W|\b(?!(?:19\d{2}|20\d{2}))\d+\b",
private static readonly Regex NormalizeRegex = new Regex(@"((^|\W)(a|an|the|and|or|of)($|\W))|\W|(?:(?<=[^0-9]+)|\b)(?!(?:19\d{2}|20\d{2}))\d+(?=[^0-9ip]+|\b)",
RegexOptions.IgnoreCase | RegexOptions.Compiled);
private static readonly Regex SimpleTitleRegex = new Regex(@"480[i|p]|720[i|p]|1080[i|p]|[x|h]264|\\|\/|\<|\>|\?|\*|\:|\|",
@ -203,19 +218,19 @@ internal static Quality ParseQuality(string name)
Logger.Trace("Trying to parse quality for {0}", name);
name = name.Trim();
var normilizedName = NormalizeTitle(name);
var normalizedName = NormalizeTitle(name);
var result = new Quality { QualityType = QualityTypes.Unknown };
result.Proper = normilizedName.Contains("proper");
result.Proper = normalizedName.Contains("proper");
if (normilizedName.Contains("dvd") || normilizedName.Contains("bdrip") || normilizedName.Contains("brrip"))
if (normalizedName.Contains("dvd") || normalizedName.Contains("bdrip") || normalizedName.Contains("brrip"))
{
result.QualityType = QualityTypes.DVD;
return result;
}
if (normilizedName.Contains("xvid") || normilizedName.Contains("divx"))
if (normalizedName.Contains("xvid") || normalizedName.Contains("divx"))
{
if (normilizedName.Contains("bluray"))
if (normalizedName.Contains("bluray"))
{
result.QualityType = QualityTypes.DVD;
return result;
@ -225,15 +240,15 @@ internal static Quality ParseQuality(string name)
return result;
}
if (normilizedName.Contains("bluray"))
if (normalizedName.Contains("bluray"))
{
if (normilizedName.Contains("720p"))
if (normalizedName.Contains("720p"))
{
result.QualityType = QualityTypes.Bluray720p;
return result;
}
if (normilizedName.Contains("1080p"))
if (normalizedName.Contains("1080p"))
{
result.QualityType = QualityTypes.Bluray1080p;
return result;
@ -242,12 +257,12 @@ internal static Quality ParseQuality(string name)
result.QualityType = QualityTypes.Bluray720p;
return result;
}
if (normilizedName.Contains("webdl"))
if (normalizedName.Contains("webdl"))
{
result.QualityType = QualityTypes.WEBDL;
return result;
}
if (normilizedName.Contains("x264") || normilizedName.Contains("h264") || normilizedName.Contains("720p"))
if (normalizedName.Contains("x264") || normalizedName.Contains("h264") || normalizedName.Contains("720p"))
{
result.QualityType = QualityTypes.HDTV;
return result;
@ -284,7 +299,7 @@ internal static Quality ParseQuality(string name)
}
}
if (normilizedName.Contains("sdtv") || (result.QualityType == QualityTypes.Unknown && normilizedName.Contains("hdtv")))
if (normalizedName.Contains("sdtv") || (result.QualityType == QualityTypes.Unknown && normalizedName.Contains("hdtv")))
{
result.QualityType = QualityTypes.SDTV;
return result;