Пример #1
0
    def __init__(self, **kwargs):
        self.logined = False
        super(UserInfoSpider, self).__init__(**kwargs)

        self.spider_index = kwargs['index']  # the spider index
        self.username = kwargs['username']
        self.passwd = kwargs['passwd']
        # 设置登录类,对应每一个spider有一个帐号
        self.login = WeiboLogin(username=self.username, passwd=self.passwd)
        # 设置对应该spider的pipeline
        self.pipelines = ['user-info']

        # 数据库相关
        log.msg('Connecting the MongoDB...', log.INFO)
        self.client = MongoClient()
        self.db = self.client[DB_NAME]
        self.collection = self.db[USER_INFO_COLLECTION_NAME]

        log.msg('Loading uid list...')

        # 开始登录微博
        log.msg('Star to login account: %s' % self.username, log.INFO)
        login_url = self.login.login()
        if login_url:
            self.start_urls.append(login_url)
        else:
            print 'Log in error.'
            log.msg('Log in error. ', log.ERROR)