def main(): samples = [ '{}/samples/appledaily.html'.format(pkgdir), 'https://tw.news.appledaily.com/local/realtime/20181025/1453825', ] logging.info('-' * 80) for path in samples: ns = NewsSoup(path, mobile=False) logging.info('路徑: {}'.format(path)) logging.info('頻道: {}'.format(ns.channel)) logging.info('標題: {}'.format(ns.title())) logging.info('日期: {}'.format(ns.date().isoformat())) logging.info('記者: {}'.format(ns.author())) logging.info('內文:') logging.info(ns.contents()) logging.info('有效內容率: {:.2f}%'.format(ns.effective_text_rate() * 100)) logging.info('-' * 80)
def soup(path): """ soup(path) """ print('-' * 75) nsoup = NewsSoup(path, mobile=False) print('路徑: {}'.format(path)) print('頻道: {}'.format(nsoup.channel)) print('標題: {}'.format(nsoup.title())) ndt = nsoup.date() if ndt is not None: print('日期: {}'.format(ndt.strftime('%Y-%m-%d %H:%M:%S'))) print('記者: {}'.format(nsoup.author())) print('內文:') print(nsoup.contents()) print('有效內容率: {:.2f}%'.format(nsoup.effective_text_rate() * 100)) print('-' * 75)