def decrypt(self,soup,cls_dict,css_dict,pattern='.*',comment=False):
'''
soup:加密的标签
cls_dict,css_dict:解析css文件得到的解密字典。
pattern:解密后的正则匹配模式，会匹配解密后的文本，将符合正则的内容返回
comment:当前标签是否属于点评评论标签。有些不是属于评论内容的标签不用置True
'''
_contents = soup.contents
_ = []
while _contents:
i = _contents.pop(0)
if isinstance(i, Tag):
if i.name in DECRYPT_TAGS:
if i['class'][0] in IGNORED_SPAN_CLASS:
continue
if i['class'][0] == 'item':
i_contents = i.contents
for j in reversed(i_contents):
_contents.insert(0,j)
continue
i = self._get_decrypted(i,cls_dict,css_dict,comment)
elif not isinstance(i, str):
continue
_.append(i)
text = _clean(_)
return from_pattern(pattern, text)
def get_full_phone(phone_str,cityId):
if not phone_str:
return
res = []
_phone =[i.strip() for i in phone_str.split('\xa0') if i]
for i in _phone:
if from_pattern(PATTERN_PHONE,i):
res.append(i)
else:
code = get_city_areacode(cityId)
res.append('-'.join([code,i]))
return res
def decrypt(self, soup, cls_dict, css_dict, pattern='.*', comment=False):
_contents = soup.contents
_ = []
while _contents:
i = _contents.pop(0)
if isinstance(i, Tag):
if i.name in DECRYPT_TAGS:
if i['class'][0] in IGNORED_SPAN_CLASS:
continue
if i['class'][0] == 'item':
i_contents = i.contents
for j in reversed(i_contents):
_contents.insert(0, j)
continue
i = self._get_decrypted(i, cls_dict, css_dict, comment)
elif not isinstance(i, str):
continue
_.append(i)
text = _clean(_)
return from_pattern(pattern, text)
