version 1.2.4

~ Improved ~ Sports parsing: MLB, NHL, Formula X, Indycar, Moto X ~ Slovenian language ~ Bluray source ~ XXX type title parsing ~ Ringtone flag + Added + HR flag + Vertical flag + Test case
pr0pz · Aug 25, 2023 · cb693e8 · cb693e8
1 parent e458082
commit cb693e8
Show file tree

Hide file tree

Showing 5 changed files with 102 additions and 39 deletions.
diff --git a/ReleaseParser.php b/ReleaseParser.php
@@ -9,7 +9,7 @@
  *
  * @package ReleaseParser
  * @author Wellington Estevo
- * @version 1.2.3
+ * @version 1.2.4
  */
 
 class ReleaseParser extends ReleasePatterns
@@ -1119,25 +1119,63 @@ private function parseTitle()
 
 			// XXX
 			case 'xxx':
+				$matches = [];
 
-				// Setup regex pattern
-				$regex_pattern = !empty( $this->get( 'date' ) ) ? self::REGEX_TITLE_XXX_DATE : self::REGEX_TITLE_XXX;
-				$regex_used = !empty( $this->get( 'date' ) ) ? 'REGEX_TITLE_XXX_DATE' : 'REGEX_TITLE_XXX';
+				// Some XXX releases with episode
+				if ( !empty( $this->get( 'episode' ) ) )
+				{
+					// Check for episode
+					$regex_pattern = self::REGEX_TITLE_TV;
+					$regex_used = 'REGEX_TITLE_TV';
 
-				// Search and replace pattern in regex pattern for better matching
-				$regex_pattern = $this->cleanupPattern( $this->release, $regex_pattern, [ 'flags', 'year', 'language', 'source', 'regex_date', 'regex_date_monthname' ] );
+					// Match title
+					\preg_match( $regex_pattern, $release_name_cleaned, $matches );
+
+					// Check for matches with regex title tv
+					if ( !empty( $matches ) )
+					{
+						$title = $matches[1];
 
-				// Match title
-				\preg_match( $regex_pattern, $release_name_cleaned, $matches );
+						// Build pattern and try to get episode title
+						// So search and replace needed data to match properly.
+						$regex_pattern = self::REGEX_TITLE_TV_EPISODE;
+						$regex_used .= ' + REGEX_TITLE_TV_EPISODE';
 
-				if ( !empty( $matches ) )
+						// Search and replace pattern in regex pattern for better matching
+						//$regex_pattern = $this->cleanupPattern( $this->release, $regex_pattern, [ 'flags', 'format', 'language', 'resolution', 'source' ] );
+						$release_name_cleaned = $this->cleanup( $release_name_cleaned, [ 'audio', 'flags', 'format', 'language', 'resolution', 'source' ] );
+
+						// Match episode title
+						\preg_match( $regex_pattern, $release_name_cleaned, $matches );
+
+						$title_extra = !empty( $matches[1] ) && $matches[1] !== '.' ? $matches[1] : '';
+
+						break;
+					}
+				}
+
+				// Default pattern
+				if ( empty( $title ) )
 				{
-					// 1st Match = Publisher, Website, etc.
-					$title = $matches[1];
-					// 2nd Match = Specific release name (movie/episode/model name, etc.)
-					$title_extra = !empty( $matches[2] ) ? $matches[2] : '';
+					// Setup regex pattern
+					$regex_pattern = !empty( $this->get( 'date' ) ) ? self::REGEX_TITLE_XXX_DATE : self::REGEX_TITLE_XXX;
+					$regex_used = !empty( $this->get( 'date' ) ) ? 'REGEX_TITLE_XXX_DATE' : 'REGEX_TITLE_XXX';
 
-					break;
+					// Search and replace pattern in regex pattern for better matching
+					$regex_pattern = $this->cleanupPattern( $this->release, $regex_pattern, [ 'flags', 'year', 'language', 'source', 'regex_date', 'regex_date_monthname' ] );
+
+					// Match title
+					\preg_match( $regex_pattern, $release_name_cleaned, $matches );
+
+					if ( !empty( $matches ) )
+					{
+						// 1st Match = Publisher, Website, etc.
+						$title = $matches[1];
+						// 2nd Match = Specific release name (movie/episode/model name, etc.)
+						$title_extra = !empty( $matches[2] ) ? $matches[2] : '';
+
+						break;
+					}
 				}
 
 				// Jump to default if no title found
@@ -1340,7 +1378,9 @@ private function parseAttribute( array $attribute, string $type = '' )
 						$attr_key === 'V2' ||
 						$attr_key === 'V3' ||
 						$attr_key === 'Cover' ||
-						$attr_key === 'Docu'
+						$attr_key === 'Docu' ||
+						$attr_key === 'HR' ||
+						$attr_key === 'Vertical'
 					)
 					{
 						$pattern = $this->cleanupPattern( $this->release, $pattern, [ 'flags', 'format', 'source', 'language', 'resolution' ] );

diff --git a/ReleaseParserTest.php b/ReleaseParserTest.php
@@ -10,7 +10,7 @@
  * 
  * @package ReleaseParser
  * @author Wellington Estevo
- * @version 1.2.3
+ * @version 1.2.4
  */
 
 /**
@@ -438,6 +438,13 @@ function release_parser_test()
 			new ReleaseParser( 'NaughtyAmerica.com_17.10.12.Sloan.Harper.Sean.Lawless.Dirty.Wives.Club.XXX.IMAGESET-FuGLi', 'imgset' ),
 			'Publisher: NaughtyAmerica.com / Title: Sloan Harper Sean Lawless Dirty Wives Club / Group: FuGLi / Year: 2017 / Date: 12.10.2017 / Flags: Imageset, XXX / Type: XXX'
 		],
+		// XXX #2 - XXX with episode and special flags
+		[
+			new ReleaseParser( 'Lustery.E727.Leo.And.Madly.A.Happy.Ending.For.Him.XXX.VERTICAL.HRp.MP4-WRB', 'XXX' ),
+			'Publisher: Lustery / Title: Leo And Madly A Happy Ending For Him / Group: WRB / Episode: 727 / Flags: HR, Vertical, XXX / Format: MP4 / Type: XXX'
+		],
+
+
 
 		// MusicVideo
 		// MusicVideo #1 - Recorded on a live show with special date (like normal music live recording)

diff --git a/ReleasePatterns.php b/ReleasePatterns.php
@@ -6,7 +6,7 @@
  *
  * @package ReleaseParser
  * @author Wellington Estevo
- * @version 1.2.3
+ * @version 1.2.4
  */
 
 class ReleasePatterns {
@@ -93,7 +93,7 @@ class ReleasePatterns {
 		// Sports
 		'Sports' => 'sport',
 		// XXX Sections
-		'XXX' => [ 'xxx', 'imgset' ],
+		'XXX' => [ 'xxx', 'imgset', 'imageset' ],
 		// Movie Sections
 		'Movie' => [ 'movie', '(?!tv).*26[45]', 'bluray', 'dvdr', 'xvid', 'divx' ],
 	];
@@ -192,8 +192,8 @@ class ReleasePatterns {
 		'MiniDisc' => [ 'md', 'minidisc' ], // Needs to be at the end, since some music releases has MD as source, but normally is MicDubbed for movies, so would wrongfully parse
 		// Misc Fallback
 		'CD' => [ '\d*cdr?\d*', 'cd[._-]?rom' ], // Other CD
-		'DVD' => '(Complete[._-])?\d*dvd[_-]?[r\d]?', // Just normal DVD
-		'Bluray' => [ 'blu[._-]?ray', '\d*bdr' ],
+		'DVD' => '(Complete.)?\d*dvd[_-]?[r\d]?', // Just normal DVD
+		'Bluray' => [ 'bl?u.?r[ae]y', '\d*bdr' ],
 		'RiP' => 'rip', // If no other rip matches
 		// Parse EP source last, it's part of the name if other source given
 		'EP' => 'EP',
@@ -221,12 +221,12 @@ class ReleasePatterns {
 		'CVD' => 'CVD',
 		'CVCD' => 'CVCD', // Compressed Video CD
 		'SVCD' => 'X?SVCD',
-		'VC1' => '(?:Bluray[._-])?VC[._-]?1',
+		'VC1' => '(?:Blu.?ray.)?VC.?1',
 		'WMV' => 'WMV',
 		'MDVDR' => 'MDVDR?',
 		'DVDR' => 'DVD[R\d]',
-		'MBluray' => '(Complete[._-])?MBLURAY',
-		'Bluray' => '(complete[._-]?)?bluray',
+		'MBluray' => '(Complete.?)?MBLURAY',
+		'Bluray' => '(complete.?)?bluray',
 		'MViD' => 'MViD',
 		// Ebook formats
 		'AZW' => 'AZW',
@@ -240,17 +240,17 @@ class ReleasePatterns {
 		'PDF' => 'PDF',
 		// Music formats
 		'DAISY' => 'DAISY', // Audiobook
-		'FLAC' => '(?:WEB[._-]?)?FLAC',
+		'FLAC' => '(?:WEB.?)?FLAC',
 		'KONTAKT' => 'KONTAKT',
 		'MP3' => 'MP3',
 		'WAV' => 'WAV',
 		'3GP' => '3gp',
 		// Software format
-		'ISO' => '(?:Bootable[._-])?ISO',
+		'ISO' => '(?:Bootable.)?ISO',
 		// Font format
 		'CrossPlatform' => 'Cross(?:Format|Platform)',
-		'OpenType' => 'Open[._-]?Type',
-		'TrueType' => 'True[._-]?Type',
+		'OpenType' => 'Open.?Type',
+		'TrueType' => 'True.?Type',
 		// Software/Game format
 		'Java Platform, Micro Edition' => 'j2me(?:v\d*)?',
 		'Java' => 'JAVA',
@@ -426,7 +426,7 @@ class ReleasePatterns {
 		'pt-BR' => [ 'Brazilian(?:.portuguese)?', 'BR' ],
 		'ro' => 'Romanian',
 		'ru' => [ 'Russian', 'RU' ],
-		'sk' => [ 'Slovak', 'SK', 'SLO', 'Slovenian' ],
+		'sk' => [ 'Slovak', 'SK', 'SLO', 'Slovenian', 'SI' ],
 		'sv' => [ 'Swedish', 'SW?E' ],
 		'sw' => 'Swahili',
 		'tg' => 'Tajik',
@@ -493,6 +493,7 @@ class ReleasePatterns {
 		'HLG' => 'HLG', // Hybrid log-gamma (like HDR)
 		'HOTFiX' => 'HOT.?FIX',
 		'HOU' => 'HOU',
+		'HR' => 'HRp?.(%flags%|%format%|%source%|%year%)', // High resolution
 		'HSBS' => 'HS(?:BS)?',
 		'Hybrid' => 'HYBRID',
 		'Imageset' => '(?:Full[._-]?)?(?:IMA?GE?|photo|foto).?SETS?',
@@ -538,7 +539,7 @@ class ReleasePatterns {
 		'RERiP' => 're.?rip',
 		'Restored' => 'RESTORED',
 		'Retail' => 'RETAIL',
-		'Ringtone' => 'rtone',
+		'Ringtone' => 'rtones?',
 		'Samplefix' => 'SAMPLE.?FIX',
 		'SDR' => 'SDR',
 		'Serial' => 'SERIAL(?!.Killer)?', // Software
@@ -569,9 +570,10 @@ class ReleasePatterns {
 		'Untouched' => 'UNTOUCHED',
 		'USK' => 'USK', // German rating system
 		'Update' => '(WITH.)?UPDATE',
-		'V1' => 'V1[._-](%format%|%language%|%source%|%resolution%)',
-		'V2' => 'V2[._-](%format%|%language%|%source%|%resolution%)',
-		'V3' => 'V3[._-](%format%|%language%|%source%|%resolution%)',
+		'V1' => 'V1.(%format%|%language%|%source%|%resolution%)',
+		'V2' => 'V2.(%format%|%language%|%source%|%resolution%)',
+		'V3' => 'V3.(%format%|%language%|%source%|%resolution%)',
+		'Vertical' => 'Vertical.(hrp?|xxx|%format%|%source%|%resolution%)',
 		'VKI' => 'VKI', // Variable Keyframe Intervals
 		'VR' => 'VR', // Virtual reality
 		'VR180' => 'VR180',
@@ -623,17 +625,17 @@ class ReleasePatterns {
 		'caf',				// African
 		'afc.asian',		// Asian
 		// Racing
-		'Formul[ae].?[1234E]',
+		'Formul[ae].?[1234E].\d{4}',
 		'F\d.\d{4}',
 		'Superleague.Formula',	
 		'Nascar.(?:cup|truck|xfinity)',
-		'Indycar',
+		'Indycar\.(series|racing|\d{4})',
 		'Porsche.(Carrera|sprint)',
 		'DTM.(\d{2,4}|spa|lauszitzring|gp\d+|\d+.lauf)',
 		'wrc.(?:fia|\d{4})', // Rallye
 		'Supercars.championship',
 		'W.series.\d{4}',
-		'Moto.?(GP|[123])',
+		'Moto.?(GP|\d).\d{4}',
 		// Cycling
 		'Cycling.(?:volta|giro|tour|strade|paris|criterium|liege|fleche|amstel|la.vuelta)', // International
 		'giro.d.italia', // Italy
@@ -657,9 +659,9 @@ class ReleasePatterns {
 		'wwe.(?:nxt|friday|this|main|monday|wrestlemania)',
 		'aew.(?:collision|dynamite|dark)',
 		// Hockey
-		'NHL',
+		'NHL\.(?:\d{4}|stanley.cup|playoffs)',
 		// Baseball
-		'MLB',
+		'MLB.(?:\d{4}|spring|world.series|pre.?season|playoffs|ws|alcs)',
 		// Boxing
 		'boxing.\d{4}.\d{2}.\d{2}',
 		// Sumo

diff --git a/changelog.txt b/changelog.txt
@@ -1,3 +1,17 @@
+version 1.2.4 ( updated 25.08.2023 )
+~ Improved
+	~ Sports parsing: MLB, NHL, Formula X, Indycar, Moto X
+	~ Slovenian language
+	~ Bluray source
+	~ XXX type title parsing
+	~ Ringtone flag
++ Added
+	+ HR flag
+	+ Vertical flag
+	+ Test case
+
+***
+
 version 1.2.3 ( updated 24.08.2023 )
 ~ Improved
 	~ Screener matching

diff --git a/composer.json b/composer.json
@@ -1,6 +1,6 @@
 {
 	"name": "propz/release-parser",
-	"version": "1.2.3",
+	"version": "1.2.4",
 	"description": "A library for parsing scene release names into human readable data.",
 	"keywords": [
 		"parser",
@@ -16,7 +16,7 @@
 	],
 	"homepage": "https://github.com/pr0pz/scene-release-parser-php#readme",
 	"readme": "README.md",
-	"time": "2023-08-24",
+	"time": "2023-08-25",
 	"license": "MIT",
 	"autoload": {
 		"classmap": [