def get_subtitle(self, id=None): subtitle = Subtitle() subtitle.description = CleanText('.//fieldset/span[@itemprop="description"]')(self.doc) if id: subtitle.id = id else: regexp = re.compile('https://www.opensubtitles.org/en/subtitles/(?P<id>\d+)/.*$') result = regexp.match(self.url) subtitle.id = result.groupdict()['id'] subtitle.name = CleanText('.//div//div//h2')(self.doc) subtitle.url = self.url return subtitle
def get_subtitle(self, id=None): subtitle = Subtitle() subtitle.description = CleanText( './/fieldset/span[@itemprop="description"]')(self.doc) if id: subtitle.id = id else: regexp = re.compile( 'https://www.opensubtitles.org/en/subtitles/(?P<id>\d+)/.*$') result = regexp.match(self.url) subtitle.id = result.groupdict()['id'] subtitle.name = CleanText('.//div//div//h2')(self.doc) subtitle.url = self.url return subtitle
def iter_subtitles(self): season = '' series_name = CleanText('//div[has-class("msg")]//h1//span[@itemprop="name"]')(self.doc) # A regexp to recover the sub id from url regexp = re.compile('.*/imdbid-(?P<episode_id>\d+)$') for sub in self.doc.xpath('//table[@id="search_results"]//tbody//tr[not(contains(@class,"head"))]'): if not Attr('.', 'class', default=None)(sub): season = CleanText('.//td[1]')(sub) else: subtitle = Subtitle() episode = CleanText('.//td[1]')(sub) subtitle.name = '%s - %s - Episode %s' % (series_name, season, episode) url = Link('.//td[3]//a')(sub) subtitle.url = self.browser.absurl(url) result = regexp.match(url) subtitle.id = result.groupdict()['episode_id'] yield subtitle
def iter_subtitles(self): season = '' series_name = CleanText( '//div[has-class("msg")]//h1//span[@itemprop="name"]')(self.doc) # A regexp to recover the sub id from url regexp = re.compile('.*/imdbid-(?P<episode_id>\d+)$') for sub in self.doc.xpath( '//table[@id="search_results"]//tbody//tr[not(contains(@class,"head"))]' ): if not Attr('.', 'class', default=None)(sub): season = CleanText('.//td[1]')(sub) else: subtitle = Subtitle() episode = CleanText('.//td[1]')(sub) subtitle.name = '%s - %s - Episode %s' % (series_name, season, episode) url = Link('.//td[3]//a')(sub) subtitle.url = self.browser.absurl(url) result = regexp.match(url) subtitle.id = result.groupdict()['episode_id'] yield subtitle