moparisthebest/SickRage
1
0
Fork 0
mirror of https://github.com/moparisthebest/SickRage synced 2024-12-12 11:02:21 -05:00
Code Issues Releases Wiki Activity

Testing new faster method of obtaining show object via regex matching for series name during Name Parsing for results.

Browse Source
This commit is contained in:
echel0n 2014-06-22 11:33:58 -07:00
parent 7defffb4f1
commit b1c9b71846
1 changed files with 35 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

64
sickbeard/name_parser/parser.py
View File

@ -37,7 +37,8 @@ class NameParser(object):
SPORTS_REGEX = 2
ANIME_REGEX = 3
def __init__(self, file_name=True, showObj=None, epObj=None, useIndexers=False, convert=False, naming_pattern=False):
def __init__(self, file_name=True, showObj=None, epObj=None, useIndexers=False, convert=False,
naming_pattern=False):
self.file_name = file_name
self.showList = sickbeard.showList or []
@ -47,17 +48,6 @@ class NameParser(object):
self.convert = convert
self.naming_pattern = naming_pattern
self.regexMode = self.ALL_REGEX
if self.showObj and self.showObj.is_anime:
self.regexMode = self.ANIME_REGEX
elif self.showObj and self.showObj.is_sports:
self.regexMode = self.SPORTS_REGEX
elif self.showObj and not self.showObj.is_anime and not self.showObj.is_sports:
self.regexMode = self.NORMAL_REGEX
self.compiled_regexes = {}
self._compile_regexes(self.regexMode)
def clean_series_name(self, series_name):
"""Cleans up series name by removing any . and _
characters, along with any trailing hyphens.
@ -117,6 +107,36 @@ class NameParser(object):
if not name:
return
if not self.naming_pattern:
# Regex pattern to return the Show / Series Name regardless of the file pattern tossed at it, matched 53 show name examples from regexes.py
show_pattern = '''^(?:(UEFA|MLB|ESPN|WWE|MMA|UFC|TNA|EPL|NASCAR|NBA|NFL|NHL|NRL|PGA|SUPER LEAGUE|FORMULA|FIFA|NETBALL|MOTOG(P)))?(?:[0-9]+)?(?:\[(?:.+?)\][ ._-])?(?P<series_name>.*?)(?:[ ._-])+?(?:Season|Part)?(?:.[eE][0-9][0-9]?)?(?:.?[sS]?[0-9][0-9]?)'''
try:
show_regex = re.compile(show_pattern, re.VERBOSE | re.IGNORECASE)
except re.error, errormsg:
logger.log(u"WARNING: Invalid show series name pattern, %s: [%s]" % (errormsg, show_pattern))
else:
seriesname_match = show_regex.match(name)
seriesname_groups = seriesname_match.groupdict().keys()
if not self.showObj and 'series_name' in seriesname_groups:
# Do we have recognize this show?
series_name = self.clean_series_name(seriesname_match.group('series_name'))
self.showObj = helpers.get_show_by_name(series_name, useIndexer=self.useIndexers)
if not self.showObj:
return
regexMode = self.ALL_REGEX
if self.showObj and self.showObj.is_anime:
regexMode = self.ANIME_REGEX
elif self.showObj and self.showObj.is_sports:
regexMode = self.SPORTS_REGEX
elif self.showObj and not self.showObj.is_anime and not self.showObj.is_sports:
regexMode = self.NORMAL_REGEX
self.compiled_regexes = {}
self._compile_regexes(regexMode)
matches = []
result = None
for (cur_regex_type, cur_regex_name), cur_regex in self.compiled_regexes.items():
@ -137,22 +157,6 @@ class NameParser(object):
result.series_name = self.clean_series_name(result.series_name)
result.score += 1
if not self.showObj and not self.naming_pattern:
self.showObj = helpers.get_show_by_name(result.series_name, useIndexer=self.useIndexers)
if self.showObj:
result.show = self.showObj
if getattr(self.showObj, 'air_by_date', None) and not cur_regex_type == 'normal':
continue
elif getattr(self.showObj, 'sports', None) and not cur_regex_type == 'sports':
continue
elif getattr(self.showObj, 'anime', None) and not cur_regex_type == 'anime':
continue
# don't continue parsing if we don't have a show object by now, try next regex pattern
if not self.showObj and not self.naming_pattern:
continue
if 'season_num' in named_groups:
tmp_season = int(match.group('season_num'))
if not (cur_regex_name == 'bare' and tmp_season in (19, 20)):
@ -555,7 +559,9 @@ class ParseResult(object):
self.episode_numbers = new_episode_numbers
self.season_number = new_season_numbers[0]
logger.log(u"Converted parsed result " + self.original_name + " into " + str(self).decode('utf-8', 'xmlcharrefreplace'), logger.DEBUG)
logger.log(u"Converted parsed result " + self.original_name + " into " + str(self).decode('utf-8',
'xmlcharrefreplace'),
logger.DEBUG)
return self