holerite.py 文件源码

def parse_salaries(self, response):
        """
        The values about person salary is in another table
        in another page, that function grab all the table headers
        and values and assign to the entity[entity_id]
        The id was passed in the response.meta
        """

        item = VereadorItem()
        item['name'] = response.meta['name']
        item['id'] = response.meta['entity_id']
        item['mesano'] = response.meta['mesano']

        for salary in response.xpath('//*[@id="holerite"]').extract():
            selector = Selector(text=salary)
            table = selector.xpath('//tr[@class="holerite_valor"]/td/text()').extract()
            item["salary_gross"] = table[0]
            item["salary_liquid"] = selector.xpath('//tr[@class="holerite_valor"]/td/strong/text()').extract_first()
            return item