collect.py 文件源码

def collect_news(html):
    try:
        news_list = []
        now_notice = html.find('div', attrs={'id': 'now_notice_area'})
        for tr in now_notice.findAll('tr'):
            td = tr.findAll('td')
            news = map(text, td[0:2])
            # ???link????????
            links = []
            if td[1].a is not None:
                for a in td[1].findAll('a'):
                    link = a.get('href')
                    links.append(link)
            urls = " ".join(links)
            # URL???
            news.append(urls)
            # ?????????
            s = news[0] + news[1]
            unique_hash = hashlib.sha1(s.encode('utf-8')).hexdigest()
            news.append(unique_hash)
            news_list.append(news)
        else:
            return news_list
    except Exception as e:
        log.exception(e)