def parse_season(self, response): print '\n\n\n', 'qweeeeeeeeeeeeeeee', '\n\n\n' sl = SeasonLoader(item=SeasonItem(), response=response) seasons = 'qwe' print '\n\n\n', seasons, '\n\n\n' # seasons_key = response.url # print '\n\n\n\n\n', response.url, '\n\n\n\n\n' sl.add_value('year', response.url) wins = response.xpath( '//*[@id="info_box"]/p[2]/text()[1]' ).extract_first() # .strip().replace('-', ',').split(',')[0] losses = response.xpath( '//*[@id="info_box"]/p[2]/text()[1]' ).extract_first() # .strip().replace('-', ',').split(',')[1] attendance = response.xpath( '//*[@id="info_box"]/p[4]/text()[2]' ).extract_first() # .split()[0] coaches = [] for coach in response.xpath( '//*[@id="info_box"]/p[2]/span[2]/following-sibling::a/text()' ).extract(): coaches.append(coach) sl.add_value('wins', wins) sl.add_value('losses', losses) sl.add_value('attendance', attendance) sl.add_value('coaches', coaches) seasons['one_season'] = sl.load_item() # print sl.load_item() # yield sl.load_item() yield seasons
def parse_season(self, response): l = SeasonLoader(item=SeasonItem(), response=response) l.add_value('year', response.url) wins = response.xpath( '//*[@id="info_box"]/p[2]/text()[1]' ).extract_first() # .strip().replace('-', ',').split(',')[0] losses = response.xpath( '//*[@id="info_box"]/p[2]/text()[1]' ).extract_first() # .strip().replace('-', ',').split(',')[1] attendance = response.xpath( '//*[@id="info_box"]/p[4]/text()[2]' ).extract_first() # .split()[0] coaches = [] for coach in response.xpath( '//*[@id="info_box"]/p[2]/span[2]/following-sibling::a/text()' ).extract(): coaches.append(coach) l.add_value('wins', wins) l.add_value('losses', losses) l.add_value('attendance', attendance) l.add_value('coaches', coaches) yield l.load_item()