bacow/AV_Data_Capture
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

madou.py:fix get title for MD0140-2

Browse Source
This commit is contained in:
lededev
2022-03-06 17:29:15 +08:00
parent 788fc4a97c
commit 31da166931
1 changed files with 5 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files

12
WebCrawler/madou.py
View File

@@ -16,12 +16,10 @@ def getActorPhoto(html):
def getTitle(html, number): # 获取标题 def getTitle(html, number): # 获取标题
title = str(html.xpath('//h1[@class="article-title"]/text()')[0]) # <title>MD0140-2 / 家有性事EP2 爱在身边-麻豆社</title>
try: browser_title = str(html.xpath("/html/head/title/text()")[0])
result = str(re.split(r'[/|／|-]', title)[1]) browser_title = browser_title[browser_title.find('/ ') + 2:].strip()
return result.strip() return browser_title[:browser_title.find('-麻豆社')].strip()
except:
return title.replace(number.upper(), '').strip()
def getStudio(html): # 获取厂商 已修改 def getStudio(html): # 获取厂商 已修改
@@ -161,5 +159,5 @@ def main(number):
if __name__ == '__main__': if __name__ == '__main__':
print(main('MD0094'))
print(main('MD0222')) print(main('MD0222'))
print(main('MD0140-2'))