diff options
| author | remitamine <remitamine@gmail.com> | 2015-09-11 04:44:17 +0100 |
|---|---|---|
| committer | remitamine <remitamine@gmail.com> | 2015-09-11 04:44:17 +0100 |
| commit | 689fb748ee1ba8e61f99d21a3bcb1bc83b708649 (patch) | |
| tree | 368ea9a04adef249dcce3633cb54a9add8eb344c | |
| parent | 1721fef28b89ac4264db978ab7fee3b4dd154056 (diff) | |
[utlis] add extract_attributes for extracting html tags attributes
| -rw-r--r-- | youtube_dl/utils.py | 8 |
1 files changed, 8 insertions, 0 deletions
diff --git a/youtube_dl/utils.py b/youtube_dl/utils.py index 206dd56bc..bcebf9cc5 100644 --- a/youtube_dl/utils.py +++ b/youtube_dl/utils.py @@ -248,6 +248,14 @@ def get_element_by_attribute(attribute, value, html): return unescapeHTML(res) +def extract_attributes(attributes_str, attributes_regex=r'(?s)\s*([^\s=]+)\s*=\s*["\']([^"\']+)["\']'): + attributes = re.findall(attributes_regex, attributes_str) + attributes_dict = {} + if attributes: + attributes_dict = {attribute_name: attribute_value for (attribute_name, attribute_value) in attributes} + return attributes_dict + + def clean_html(html): """Clean an HTML snippet into a readable string""" |