[ie/Rule34Video] Extract creators (#9258)

Authored by: gmes78
This commit is contained in:
gmes78 2024-02-22 00:48:49 +00:00 committed by GitHub
parent 28e53d60df
commit 3d9dc2f359
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194

View File

@ -9,7 +9,6 @@ from ..utils import (
get_element_html_by_class, get_element_html_by_class,
get_elements_by_class, get_elements_by_class,
int_or_none, int_or_none,
join_nonempty,
parse_count, parse_count,
parse_duration, parse_duration,
unescapeHTML, unescapeHTML,
@ -57,7 +56,7 @@ class Rule34VideoIE(InfoExtractor):
'comment_count': int, 'comment_count': int,
'timestamp': 1640131200, 'timestamp': 1640131200,
'description': '', 'description': '',
'creator': 'WildeerStudio', 'creators': ['WildeerStudio'],
'upload_date': '20211222', 'upload_date': '20211222',
'uploader': 'CerZule', 'uploader': 'CerZule',
'uploader_url': 'https://rule34video.com/members/36281/', 'uploader_url': 'https://rule34video.com/members/36281/',
@ -81,13 +80,13 @@ class Rule34VideoIE(InfoExtractor):
'quality': quality, 'quality': quality,
}) })
categories, creator, uploader, uploader_url = [None] * 4 categories, creators, uploader, uploader_url = [None] * 4
for col in get_elements_by_class('col', webpage): for col in get_elements_by_class('col', webpage):
label = clean_html(get_element_by_class('label', col)) label = clean_html(get_element_by_class('label', col))
if label == 'Categories:': if label == 'Categories:':
categories = list(map(clean_html, get_elements_by_class('item', col))) categories = list(map(clean_html, get_elements_by_class('item', col)))
elif label == 'Artist:': elif label == 'Artist:':
creator = join_nonempty(*map(clean_html, get_elements_by_class('item', col)), delim=', ') creators = list(map(clean_html, get_elements_by_class('item', col)))
elif label == 'Uploaded By:': elif label == 'Uploaded By:':
uploader = clean_html(get_element_by_class('name', col)) uploader = clean_html(get_element_by_class('name', col))
uploader_url = extract_attributes(get_element_html_by_class('name', col) or '').get('href') uploader_url = extract_attributes(get_element_html_by_class('name', col) or '').get('href')
@ -115,7 +114,7 @@ class Rule34VideoIE(InfoExtractor):
'comment_count': int_or_none(self._search_regex( 'comment_count': int_or_none(self._search_regex(
r'[^(]+\((\d+)\)', get_element_by_attribute('href', '#tab_comments', webpage), 'comment count', fatal=False)), r'[^(]+\((\d+)\)', get_element_by_attribute('href', '#tab_comments', webpage), 'comment count', fatal=False)),
'age_limit': 18, 'age_limit': 18,
'creator': creator, 'creators': creators,
'uploader': uploader, 'uploader': uploader,
'uploader_url': uploader_url, 'uploader_url': uploader_url,
'categories': categories, 'categories': categories,