示例#1
0
def do_work(argv):
    url = argv.get('url')
    workmanager = argv.get('workmanager')
    max_size = argv.get('max_size')
    fetcher = Fetch(url)
    for resource in fetcher.get_all_resource():
        if len(resourcelist) > max_size:
            break
        if resource not in resourcelist:
            resourcelist.append(resource)
            logger.get_logger.info(resource)

    for href in fetcher.get_all_link():
        if len(resourcelist) > max_size:
            break
        if href not in urllist:
            urllist.append(href)
            workmanager.add_job(do_work,
                                workmanager=workmanager,
                                url=href,
                                max_size=max_size)