Sirius/youtube-dl
1
0
Fork 0
Code Issues Pull requests Projects Releases Wiki Activity

[utils] Add get_element_by_class

Browse source 
For #9950
This commit is contained in:
Yen Chi Hsuan 2016-07-06 20:02:52 +08:00
parent ab49d7a9fa
commit 84c237fb8a
No known key found for this signature in database
GPG key ID: 3FDDD575826C5C30
2 changed files with 19 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

12
youtube_dl/utils.py
View file

@ -310,9 +310,17 @@ def get_element_by_id(id, html):
return get_element_by_attribute('id', id, html)
def get_element_by_attribute(attribute, value, html):
def get_element_by_class(class_name, html):
return get_element_by_attribute(
'class', r'[^\'"]*\b%s\b[^\'"]*' % re.escape(class_name),
html, escape_value=False)
def get_element_by_attribute(attribute, value, html, escape_value=True):
"""Return the content of the tag with the specified attribute in the passed HTML document"""
value = re.escape(value) if escape_value else value
m = re.search(r'''(?xs)
<([a-zA-Z0-9:._-]+)
(?:\s+[a-zA-Z0-9:._-]+(?:=[a-zA-Z0-9:._-]*|="[^"]*"|='[^']*'))*?
@ -321,7 +329,7 @@ def get_element_by_attribute(attribute, value, html):
\s*>
(?P<content>.*?)
</\1>
''' % (re.escape(attribute), re.escape(value)), html)
''' % (re.escape(attribute), value), html)
if not m:
return None