[newgrounds] Fix metadata extraction (closes #15531)

This commit is contained in:
Sergey M․ 2018-02-09 21:17:02 +07:00
parent 5828489072
commit 9e167e1ee3
No known key found for this signature in database
GPG key ID: 2C393E0F18A9236D

View file

@ -87,19 +87,21 @@ def _real_extract(self, url):
self._check_formats(formats, media_id) self._check_formats(formats, media_id)
self._sort_formats(formats) self._sort_formats(formats)
uploader = self._search_regex( uploader = self._html_search_regex(
r'(?:Author|Writer)\s*<a[^>]+>([^<]+)', webpage, 'uploader', (r'(?s)<h4[^>]*>(.+?)</h4>.*?<em>\s*Author\s*</em>',
r'(?:Author|Writer)\s*<a[^>]+>([^<]+)'), webpage, 'uploader',
fatal=False) fatal=False)
timestamp = unified_timestamp(self._search_regex( timestamp = unified_timestamp(self._html_search_regex(
r'<dt>Uploaded</dt>\s*<dd>([^<]+)', webpage, 'timestamp', (r'<dt>\s*Uploaded\s*</dt>\s*<dd>([^<]+</dd>\s*<dd>[^<]+)',
r'<dt>\s*Uploaded\s*</dt>\s*<dd>([^<]+)'), webpage, 'timestamp',
default=None)) default=None))
duration = parse_duration(self._search_regex( duration = parse_duration(self._search_regex(
r'<dd>Song\s*</dd><dd>.+?</dd><dd>([^<]+)', webpage, 'duration', r'(?s)<dd>\s*Song\s*</dd>\s*<dd>.+?</dd>\s*<dd>([^<]+)', webpage,
default=None)) 'duration', default=None))
filesize_approx = parse_filesize(self._html_search_regex( filesize_approx = parse_filesize(self._html_search_regex(
r'<dd>Song\s*</dd><dd>(.+?)</dd>', webpage, 'filesize', r'(?s)<dd>\s*Song\s*</dd>\s*<dd>(.+?)</dd>', webpage, 'filesize',
default=None)) default=None))
if len(formats) == 1: if len(formats) == 1:
formats[0]['filesize_approx'] = filesize_approx formats[0]['filesize_approx'] = filesize_approx