Add format, screenSize, videoCodec, audioCodec, website and year patt…

…erns

guessit/rules/__init__.py

@@ -7,7 +7,19 @@
 from rebulk import Rebulk
 
 from .episodes import EPISODES
+from .format import FORMAT
+from .video_codec import VIDEO_CODEC
+from .audio_codec import AUDIO_CODEC
+from .screen_size import SCREEN_SIZE
+from .website import WEBSITE
+from .year import YEAR
 
 REBULK = Rebulk()
-REBULK.rebulk(EPISODES)
 
+REBULK.rebulk(EPISODES)
+REBULK.rebulk(FORMAT)
+REBULK.rebulk(VIDEO_CODEC)
+REBULK.rebulk(AUDIO_CODEC)
+REBULK.rebulk(SCREEN_SIZE)
+REBULK.rebulk(WEBSITE)
+REBULK.rebulk(YEAR)

guessit/rules/audio_codec.py

@@ -0,0 +1,22 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+"""
+audioCodec property
+"""
+from rebulk import Rebulk
+
+from .common import dash
+from .common.validators import seps_surround
+
+import regex as re
+
+AUDIO_CODEC = Rebulk().regex_defaults(flags=re.IGNORECASE, abbreviations=[dash])
+AUDIO_CODEC.defaults(name="audioCodec", validator=seps_surround)
+
+AUDIO_CODEC.regex("MP3", "LAME", r"LAME(?:\d)+-(?:\d)+", value="MP3")
+AUDIO_CODEC.regex("DolbyDigital", "DD", value="DD")
+AUDIO_CODEC.regex("AAC", value="AAC")
+AUDIO_CODEC.regex("AC3", value="AC3")
+AUDIO_CODEC.regex("Flac", value="FLAC")
+AUDIO_CODEC.regex("DTS", value="DTS")  # TODO: LeftValidator
+AUDIO_CODEC.regex("True-HD", value="True-HD")

guessit/rules/common/__init__.py

@@ -0,0 +1,9 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+"""
+Common module
+"""
+
+seps = r' [](){},:;!?+*|&=%§-_~#/\.'  # list of tags/words separators
+
+dash = ("-", r"[\W_]?")  # abbreviation used by many rebulk objects.

guessit/rules/common/validators.py

@@ -0,0 +1,16 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+"""
+Validators
+"""
+
+from rebulk.validators import chars_before, chars_after, chars_surround
+
+from . import seps
+
+from functools import partial
+
+
+seps_before = partial(chars_before, seps)
+seps_after = partial(chars_after, seps)
+seps_surround = partial(chars_surround, seps)

guessit/rules/episodes.py

@@ -8,10 +8,9 @@
 
 import regex as re
 
-EPISODES = Rebulk()
+EPISODES = Rebulk().regex_defaults(flags=re.IGNORECASE)
 
 EPISODES.regex(r'(?P<season>\d+)x(?P<episodeNumber>\d+)',
                r'S(?P<season>\d+)[ex](?P<episodeNumber>\d+)',
-               formatter={'season': int, 'episodeNumber': int},
-               flags=re.IGNORECASE,
+               formatter=int,
                children=True)

guessit/rules/format.py

@@ -0,0 +1,30 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+"""
+format property
+"""
+from rebulk import Rebulk
+
+from .common import dash
+
+import regex as re
+from .common.validators import seps_surround
+
+FORMAT = Rebulk().regex_defaults(flags=re.IGNORECASE, abbreviations=[dash])
+FORMAT.defaults(name="format", validator=seps_surround)
+
+FORMAT.regex("VHS", "VHS-Rip", value="VHS")
+FORMAT.regex("CAM", "CAM-Rip", "HD-CAM", value="Cam")
+FORMAT.regex("TELESYNC", "TS", "HD-TS", value="Telesync")
+FORMAT.regex("WORKPRINT", "WP", value="Workprint")
+FORMAT.regex("TELECINE", "TC", value="Telecine")
+FORMAT.regex("PPV", "PPV-Rip", value="PPV")  # Pay Per View
+FORMAT.regex("SD-TV", "SD-TV-Rip", "Rip-SD-TV", "TV-Rip", "Rip-TV", value="TV")  # TV is too common to allow matching
+FORMAT.regex("DVB-Rip", "DVB", "PD-TV", value="DVB")
+FORMAT.regex("DVD", "DVD-Rip", "VIDEO-TS", "DVD-R", "DVD-9", "DVD-5", value="DVD")
+FORMAT.regex("HD-TV", "TV-RIP-HD", "HD-TV-RIP", "HD-RIP", value="HDTV")
+FORMAT.regex("VOD", "VOD-Rip", value="VOD")
+FORMAT.regex("WEB-Rip", value="WEBRip")
+FORMAT.regex("WEB-DL", "WEB-HD", "WEB", value="WEB-DL")
+FORMAT.regex("HD-DVD-Rip", "HD-DVD", value="HD-DVD")
+FORMAT.regex("Blu-ray(?:-Rip)?", "B[DR]", "B[DR]-Rip", "BD[59]", "BD25", "BD50", value="BluRay")

guessit/rules/screen_size.py

@@ -0,0 +1,25 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+"""
+screenSize property
+"""
+from rebulk import Rebulk
+
+import regex as re
+from .common.validators import seps_surround
+
+SCREEN_SIZE = Rebulk().regex_defaults(flags=re.IGNORECASE)
+SCREEN_SIZE.defaults(name="screenSize", validator=seps_surround)
+
+SCREEN_SIZE.regex(r"(?:\d{3,}(?:\\|\/|x|\*))?360(?:i|p?x?)", value="360p")
+SCREEN_SIZE.regex(r"(?:\d{3,}(?:\\|\/|x|\*))?368(?:i|p?x?)", value="368p")
+SCREEN_SIZE.regex(r"(?:\d{3,}(?:\\|\/|x|\*))?480(?:i|p?x?)", value="480p")
+SCREEN_SIZE.regex(r"(?:\d{3,}(?:\\|\/|x|\*))?576(?:i|p?x?)", value="576p")
+SCREEN_SIZE.regex(r"(?:\d{3,}(?:\\|\/|x|\*))?720(?:i|p?x?)", value="720p")
+SCREEN_SIZE.regex(r"(?:\d{3,}(?:\\|\/|x|\*))?900(?:i|p?x?)", value="900p")
+SCREEN_SIZE.regex(r"(?:\d{3,}(?:\\|\/|x|\*))?1080i", value="1080i")
+SCREEN_SIZE.regex(r"(?:\d{3,}(?:\\|\/|x|\*))?1080p?x?", value="1080p")
+SCREEN_SIZE.regex(r"(?:\d{3,4}(?:\\|\/|x|\*))?2160(?:i|p?x?)", value="4K")
+
+# TODO: implement validators from guessit 1
+# validator=ChainedValidator(DefaultValidator(), OnlyOneValidator()))

guessit/rules/video_codec.py

@@ -0,0 +1,21 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+"""
+videoCodec property
+"""
+from rebulk import Rebulk
+
+from .common import dash
+from .common.validators import seps_surround
+
+import regex as re
+
+VIDEO_CODEC = Rebulk().regex_defaults(flags=re.IGNORECASE, abbreviations=[dash])
+VIDEO_CODEC.defaults(name="videoCodec", validator=seps_surround)
+
+VIDEO_CODEC.regex(r"Rv\d{2}", value="Real")
+VIDEO_CODEC.regex("Mpeg2", value="Mpeg2")
+VIDEO_CODEC.regex("DVDivX", "DivX", value="DivX")
+VIDEO_CODEC.regex("XviD", value="XviD")
+VIDEO_CODEC.regex("[hx]-264(?:-AVC)?", "MPEG-4(?:-AVC)", value="h264")
+VIDEO_CODEC.regex("[hx]-265(?:-HEVC)?", "HEVC", value="h265")

guessit/rules/website.py

@@ -0,0 +1,30 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+"""
+Website property.
+"""
+
+from pkg_resources import resource_stream  # @UnresolvedImport
+
+from rebulk import Rebulk
+
+import regex as re
+
+WEBSITE = Rebulk().regex_defaults(flags=re.IGNORECASE)
+WEBSITE.defaults(name="website")
+
+TLDS = [l.strip().decode('utf-8')
+        for l in resource_stream('guessit', 'tlds-alpha-by-domain.txt').readlines()
+        if b'--' not in l][1:]  # All registered domain extension
+
+SAFE_TLDS = ['com', 'org', 'net']  # For sure a website extension
+SAFE_SUBDOMAINS = ['www']  # For sure a website subdomain
+SAFE_PREFIX = ['co', 'com', 'org', 'net']  # Those words before a tlds are sure
+
+WEBSITE.regex(r'(?:[^a-z0-9]|^)((?:\L<safe_subdomains>\.)+(?:[a-z-]+\.)+(?:\L<tlds>))(?:[^a-z0-9]|$)',
+              safe_subdomains=SAFE_SUBDOMAINS, tlds=TLDS, children=True)
+WEBSITE.regex(r'(?:[^a-z0-9]|^)((?:\L<safe_subdomains>\.)*[a-z-]+\.(?:\L<safe_tlds>))(?:[^a-z0-9]|$)',
+              safe_subdomains=SAFE_SUBDOMAINS, safe_tlds=SAFE_TLDS, children=True)
+WEBSITE.regex(r'(?:[^a-z0-9]|^)((?:\L<safe_subdomains>\.)*[a-z-]+\.(?:\L<safe_prefix>\.)+(?:\L<tlds>))(?:[^a-z0-9]|$)',
+              safe_subdomains=SAFE_SUBDOMAINS, safe_prefix=SAFE_PREFIX, tlds=TLDS, children=True)
+

guessit/rules/year.py

@@ -0,0 +1,26 @@
+#!/usr/bin/env python
+# -*- coding: utf-8 -*-
+"""
+year property
+"""
+from rebulk import Rebulk
+
+from .common.validators import seps_surround
+
+YEAR = Rebulk()
+YEAR.defaults(name="year", validator=seps_surround)
+
+
+def validate_year(match):
+    """
+    Check if match is a valid year
+
+    :param match:
+    :type match:
+    :return:
+    :rtype:
+    """
+    return 1920 <= match.value < 2030
+
+
+YEAR.regex(r"\d{4}", formatter=int, validator=validate_year)

guessit/test/movies.yml

@@ -0,0 +1,8 @@
+? Movies/Fear and Loathing in Las Vegas (1998)/Fear.and.Loathing.in.Las.Vegas.720p.HDDVD.DTS.x264-ESiR.mkv
+: #title: Fear and Loathing in Las Vegas
+  year: 1998
+  screenSize: 720p
+  format: HD-DVD
+  audioCodec: DTS
+  videoCodec: h264
+  #releaseGroup: ESiR

guessit/test/rules/audioCodec.yml

@@ -0,0 +1,30 @@
+# Multiple input strings having same expected results can be chained.
+# Use $ marker to check inputs that should not match results.
+
+
+? +MP3
+? +lame
+? +lame3.12
+? +lame3.100
+: audioCodec: MP3
+
+? +DolbyDigital
+? +DD
+? -Dolby Digital
+: audioCodec: DD
+
+? +AAC
+: audioCodec: AAC
+
+? +AC3
+: audioCodec: AC3
+
+? +Flac
+: audioCodec: FLAC
+
+? +DTS
+: audioCodec: DTS
+
+? +True-HD
+? +trueHD
+: audioCodec: True-HD

guessit/test/rules/episodes.yml

@@ -1,17 +1,17 @@
 # Multiple input strings having same expected results can be chained.
 # Use $ marker to check inputs that should not match results.
-? 2x5
-? 2X5
-? 02x05
-? 2X05
-? 02x5
+? +2x5
+? +2X5
+? +02x05
+? +2X05
+? +02x5
 ? S02E05
 ? s02e05
 ? s02e5
 ? s2e05
-? $s03e05
-? $s02e06
-? $3x05
-? $2x06
+? -s03e05
+? -s02e06
+? -3x05
+? -2x06
 : season: 2
   episodeNumber: 5