def get_old_news_md5(news_md5): col = db_func(col='news_content') for i in col.find({}, {'news_md5': 1}): if 'news_md5' in i.keys(): if news_md5 == i['news_md5'] : return True return False
def get_old_news_links(): col = db_func(col='news_content') old_list = [] for i in col.find({}, {'news_link': 1}): if i['news_link'].find('chaindd'): old_list.append(i['news_link']) return old_list
def update_news_info(links,news_img_dict): for link in links: news = None if link in news_img_dict.keys(): news_img = news_img_dict[link] else: news_img = '' news = news_page_info(link, news_img) if news is not None: col = db_func(col='news_content') col.insert_one(news)
def update_news_info(links,news_img_dict): for link in links: news = None # link_status = http_status(link) # if link_status < 400: if link in news_img_dict.keys(): news_img = news_img_dict[link] else: news_img = '' news = news_page_info(link, news_img) md5_checked = get_old_news_md5(news['news_md5']) if md5_checked is True: continue if news is not None: col = db_func(col='news_content') col.insert_one(news)