def get_items(self, stories):
pattern = re.compile(
r'<div class="msgBox".*?<div class="userName".*?title="(.*?)" gender='
+ '.*?<div class="msgCnt">(.*?)</div>' +
'.*?<div class="pubInfo.*?from="\\d*">(.*?)</a>', re.S)
for story in stories:
story, quotation = tencent_util.depart_quotation(story)
items = re.findall(pattern, story)
if not len(items):
continue
item = items[0]
print('作者：', item[0])
self.document.add_heading('', 0)
tencent_util.add_author(self.document, item[0])
print('内容：', item[1])
content_valid = ''.join(c for c in item[1]
if ord(c) >= 32) # delete control chars
tencent_util.add_content(self.document, content_valid)
tencent_util.add_quotation(self.document, quotation)
tencent_util.add_picture(self.document, story)
tencent_util.add_video(self.document, story)
print('时间：', item[2])
tencent_util.add_time(self.document, item[2])
tencent_util.add_location(self.document, story)
def get_items(self):
for story in self.stories:
story, quotation = tencent_util.depart_quotation(
self.browser, story)
author = story.find_element_by_class_name(
'userName').find_element_by_tag_name('a').get_attribute(
'title')
print('作者：', author)
content_html = story.find_element_by_class_name(
'msgCnt').get_attribute('innerHTML')
content_valid = ''.join(c for c in content_html
if ord(c) >= 32) # delete control chars
content = story.find_element_by_class_name('msgCnt').text
print('内容：', content)
time = story.find_element_by_class_name('time').text
print('时间：', time)
print(
'----------------------------------------------------------------------------------'
)
self.document.add_heading('', 0)
tencent_util.add_author(self.document, author)
tencent_util.add_content(self.document, content_valid)
tencent_util.add_quotation(self.document, quotation)
tencent_util.add_picture(self.document, story)
tencent_util.add_video(self.document, story)
tencent_util.add_time(self.document, time)
tencent_util.add_location(self.document, story)