Beispiel #1
0
 def _parse_list_body(self, body):
     # print(body)
     doc = html.fromstring(body)
     items = utils.parse_list_items_1(doc)
     [self._add_article(item) for item in items]
     # print(pprint.pformat(items))
     return items
Beispiel #2
0
 def _parse_list_body(self, body):
     # print(body)
     # sys.exit(0)
     doc = html.fromstring(body)
     items = utils.parse_list_items_1(doc)
     [self._add_article(item) for item in items]
     # print(len(items))
     return items
Beispiel #3
0
    def _parse_list_body(self, body):
        doc = html.fromstring(body)
        first = utils.parse_list_first(doc)
        self._add_article(first)

        columns = utils.parse_list_items_1(doc)
        [self._add_article(column) for column in columns]
        columns.append(first)

        return columns
Beispiel #4
0
 def _parse_list_body(self, body):
     doc = html.fromstring(body)
     first = utils.parse_list_first(doc)
     self._add_article(first)
     # print(first)
     # print(body)
     # sys.exit(0)
     doc = html.fromstring(body)
     items = utils.parse_list_items_1(doc)
     [self._add_article(item) for item in items]
     # print(pprint.pformat(items))
     items.append(first)
     return items
Beispiel #5
0
    def _parse_list_body(self, body):
        '''
<li>
    <p class="tit"><a href="http://news.stcn.com/2020/0228/15689665.shtml" target="_blank" title="养猪改变命运!1530亿身家的河南新首富,风口上的2600亿养猪巨头">养猪改变命运!1530亿身家的河南新首富,风口上的2600亿养猪巨头</a></p>
    <p class="exp"></p>
    <p class="sj">2020-02-28<span>10:11</span></p>
</li>
        '''
        # print(body)
        # sys.exit(0)
        doc = html.fromstring(body)
        items = utils.parse_list_items_1(doc)
        [self._add_article(item) for item in items]
        # print(len(items))
        return items
Beispiel #6
0
    def _parse_list_body(self, body):
        '''
<ul class="news_list">
    <li>
        <p class="tit"><a href="http://news.stcn.com/2020/0225/15682787.shtml" target="_blank" title="新零售商家加快复工复产 菜鸟急调400辆车驰援商家发货">新零售商家加快复工复产 菜鸟急调400辆车驰援商家发货</a></p>
        <p class="exp"></p>
        <p class="sj">2020-02-25<span>20:32</span></p>
    </li>
        '''
        # print(body)
        # sys.exit(0)
        doc = html.fromstring(body)
        items = utils.parse_list_items_1(doc)
        [self._add_article(item) for item in items]
        # print(len(items))
        return items
Beispiel #7
0
    def _parse_list_body(self, body):
        '''
<ul class="news_list">
            <li>
        <p class="tit"><a href="http://news.stcn.com/2020/0224/15657601.shtml" target="_blank" title="高瓴资本宣布成立“高瓴创投” 首期规模100亿">高瓴资本宣布成立“高瓴创投” 首期规模100亿</a></p>
        <p class="exp">高瓴资本在其官方公众号发布“致创业者的一封信”,宣布成立专注于投资早期创业公司的高瓴创投,主要专注于生物医药及医疗器械、软件服务和原发科技创新、消费互联网及科技、新兴消费品牌及服务四大领域的风险投资。</p>
        <p class="sj">2020-02-24<span>12:00</span></p>
    </li>
        '''
        # print(body)
        # with open("hello.html", "w") as f:
        #     f.write(body)
        # sys.exit(0)
        doc = html.fromstring(body)
        items = utils.parse_list_items_1(doc)
        [self._add_article(item) for item in items]
        # print(len(items))
        return items