Update spider.py

release
FongMi 2 years ago
parent 8913b35404
commit 840dd928db
  1. 7
      chaquo/src/main/python/base/spider.py

@ -80,6 +80,13 @@ class Spider(metaclass=ABCMeta):
path = os.path.join(os.path.join(cache_dir, 'py'), f'{name}.py')
return SourceFileLoader(name, path).load_module()
def regStr(self, src, reg, group=1):
m = re.search(reg, src)
src = ''
if m:
src = m.group(group)
return src
def removeHtmlTags(self, src):
clean = re.compile('<.*?>')
return re.sub(clean, '', src)

Loading…
Cancel
Save