Exemplo n.º 1
0
def start():
    """ 入口函数 """
    keys = globals().get('keywords', list())
    for key in keys:
        raw_companies = tyc_client.search(key)
        cost_time = 2 * raw_companies.__len__()
        log.info('正在处理爬取[%s],大概需要%s秒' % (key, cost_time))
        # company对象
        company = Company()
        for raw_company in raw_companies:
            company.keyword = key
            manager.assembly(company, raw_company)
            # company detail
            raw_company_detail = tyc_client.search_detail(raw_company.get('id'))
            manager.assembly_detail(company, raw_company_detail)
            log.info(company)
            # save(company.__dict__)
            company.clear()
    log.info("completed")
Exemplo n.º 2
0
def start():
    keywords = globals().get('keywords')
    for keyword in keywords:
        raw_companies = qcc_client.search(keyword)
        log.info('正在处理爬取[%s]' % keyword)
        # company对象
        company = Company()
        for raw_company in raw_companies:
            company.keyword = keyword
            # 组装公司信息
            manager.assembly(company, raw_company)
            raw_company_detail = qcc_client.search_detail(raw_company.get('KeyNo'))
            # 补充公司详细信息
            manager.assembly_detail(company, raw_company_detail)
            # 保存到数据库
            # save(company.__dict__)
            log.info(company)
            company.clear()
    log.info('completed')
Exemplo n.º 3
0
def start():
    keywords = globals().get('keywords')
    if keywords:
        for keyword in keywords:
            raw_companies = qcc_client.search(keyword)
            cost_time = 2 * raw_companies.__len__() + 4
            log.info('正在处理爬取[%s],大概需要%s秒' % (keyword, cost_time))
            # company对象
            company = Company()
            for raw_company in raw_companies:
                company.keyword = keyword
                # 组装公司信息
                manager.assembly(company, raw_company)
                raw_company_detail = qcc_client.search_detail(
                    raw_company.get('KeyNo'))
                # 补充公司详细信息
                manager.assembly_detail(company, raw_company_detail)
                # 保存到数据库
                save(company.__dict__)
                # 重置当前对象
                company.clear()

    log.info('completed')