mirror of
https://gitlab.com/dstftw/youtube-dl.git
synced 2020-11-16 09:42:26 +00:00
[revison3] fix video_id for --download-archive
This commit is contained in:
parent
51ce91174b
commit
2629a3802c
|
@ -19,7 +19,7 @@ class Revision3IE(InfoExtractor):
|
|||
'url': 'http://www.revision3.com/technobuffalo/5-google-predictions-for-2016',
|
||||
'md5': 'd94a72d85d0a829766de4deb8daaf7df',
|
||||
'info_dict': {
|
||||
'id': '73034',
|
||||
'id': '71089',
|
||||
'display_id': 'technobuffalo/5-google-predictions-for-2016',
|
||||
'ext': 'webm',
|
||||
'title': '5 Google Predictions for 2016',
|
||||
|
@ -42,7 +42,7 @@ class Revision3IE(InfoExtractor):
|
|||
}, {
|
||||
'url': 'https://testtube.com/dnews/5-weird-ways-plants-can-eat-animals?utm_source=FB&utm_medium=DNews&utm_campaign=DNewsSocial',
|
||||
'info_dict': {
|
||||
'id': '60163',
|
||||
'id': '58227',
|
||||
'display_id': 'dnews/5-weird-ways-plants-can-eat-animals',
|
||||
'duration': 275,
|
||||
'ext': 'webm',
|
||||
|
@ -56,7 +56,7 @@ class Revision3IE(InfoExtractor):
|
|||
}, {
|
||||
'url': 'http://testtube.com/tt-editors-picks/the-israel-palestine-conflict-explained-in-ten-min',
|
||||
'info_dict': {
|
||||
'id': '73573',
|
||||
'id': '71618',
|
||||
'ext': 'mp4',
|
||||
'display_id': 'tt-editors-picks/the-israel-palestine-conflict-explained-in-ten-min',
|
||||
'title': 'The Israel-Palestine Conflict Explained in Ten Minutes',
|
||||
|
@ -81,6 +81,7 @@ class Revision3IE(InfoExtractor):
|
|||
|
||||
def _real_extract(self, url):
|
||||
domain, display_id = re.match(self._VALID_URL, url).groups()
|
||||
site = domain.split('.')[0]
|
||||
page_info = self._download_json(
|
||||
self._PAGE_DATA_TEMPLATE % (domain, display_id, domain), display_id)
|
||||
|
||||
|
@ -88,6 +89,7 @@ class Revision3IE(InfoExtractor):
|
|||
page_type = page_data['type']
|
||||
if page_type in ('episode', 'embed'):
|
||||
show_data = page_data['show']['data']
|
||||
page_id = compat_str(page_data['id'])
|
||||
video_id = compat_str(page_data['video']['data']['id'])
|
||||
|
||||
preference = qualities(['mini', 'small', 'medium', 'large'])
|
||||
|
@ -98,7 +100,7 @@ class Revision3IE(InfoExtractor):
|
|||
} for image_id, image_url in page_data.get('images', {}).items()]
|
||||
|
||||
info = {
|
||||
'id': video_id,
|
||||
'id': page_id,
|
||||
'display_id': display_id,
|
||||
'title': unescapeHTML(page_data['name']),
|
||||
'description': unescapeHTML(page_data.get('summary')),
|
||||
|
@ -107,6 +109,7 @@ class Revision3IE(InfoExtractor):
|
|||
'uploader': show_data.get('name'),
|
||||
'uploader_id': show_data.get('slug'),
|
||||
'thumbnails': thumbnails,
|
||||
'extractor_key': site,
|
||||
}
|
||||
|
||||
if page_type == 'embed':
|
||||
|
@ -153,8 +156,13 @@ class Revision3IE(InfoExtractor):
|
|||
entries = []
|
||||
page_num = 1
|
||||
while True:
|
||||
entries.extend([self.url_result(
|
||||
'http://%s%s' % (domain, episode['path'])) for episode in episodes_data])
|
||||
entries.extend([{
|
||||
'_type': 'url',
|
||||
'url': 'http://%s%s' % (domain, episode['path']),
|
||||
'id': compat_str(episode['id']),
|
||||
'ie_key': 'Revision3',
|
||||
'extractor_key': site,
|
||||
} for episode in episodes_data])
|
||||
processed_episodes += len(episodes_data)
|
||||
if processed_episodes == num_episodes:
|
||||
break
|
||||
|
|
Loading…
Reference in a new issue