|
|
@ -14,7 +14,13 @@ from ..utils import (
|
|
|
|
)
|
|
|
|
)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class TwitterCardIE(InfoExtractor):
|
|
|
|
class TwitterBaseIE(InfoExtractor):
|
|
|
|
|
|
|
|
def _get_vmap_video_url(self, vmap_url, video_id):
|
|
|
|
|
|
|
|
vmap_data = self._download_xml(vmap_url, video_id)
|
|
|
|
|
|
|
|
return xpath_text(vmap_data, './/MediaFile').strip()
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class TwitterCardIE(TwitterBaseIE):
|
|
|
|
IE_NAME = 'twitter:card'
|
|
|
|
IE_NAME = 'twitter:card'
|
|
|
|
_VALID_URL = r'https?://(?:www\.)?twitter\.com/i/cards/tfw/v1/(?P<id>\d+)'
|
|
|
|
_VALID_URL = r'https?://(?:www\.)?twitter\.com/i/cards/tfw/v1/(?P<id>\d+)'
|
|
|
|
_TESTS = [
|
|
|
|
_TESTS = [
|
|
|
@ -96,10 +102,8 @@ class TwitterCardIE(InfoExtractor):
|
|
|
|
video_id)
|
|
|
|
video_id)
|
|
|
|
if 'playlist' not in config:
|
|
|
|
if 'playlist' not in config:
|
|
|
|
if 'vmapUrl' in config:
|
|
|
|
if 'vmapUrl' in config:
|
|
|
|
vmap_data = self._download_xml(config['vmapUrl'], video_id)
|
|
|
|
|
|
|
|
video_url = xpath_text(vmap_data, './/MediaFile').strip()
|
|
|
|
|
|
|
|
formats.append({
|
|
|
|
formats.append({
|
|
|
|
'url': video_url,
|
|
|
|
'url': self._get_vmap_video_url(config['vmapUrl'], video_id),
|
|
|
|
})
|
|
|
|
})
|
|
|
|
break # same video regardless of UA
|
|
|
|
break # same video regardless of UA
|
|
|
|
continue
|
|
|
|
continue
|
|
|
@ -226,3 +230,32 @@ class TwitterIE(InfoExtractor):
|
|
|
|
return info
|
|
|
|
return info
|
|
|
|
|
|
|
|
|
|
|
|
raise ExtractorError('There\'s not video in this tweet.')
|
|
|
|
raise ExtractorError('There\'s not video in this tweet.')
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
class TwitterAmplifyIE(TwitterBaseIE):
|
|
|
|
|
|
|
|
IE_NAME = 'twitter:amplify'
|
|
|
|
|
|
|
|
_VALID_URL = 'https?://amp\.twimg\.com/v/(?P<id>[0-9a-f\-]{36})'
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
_TEST = {
|
|
|
|
|
|
|
|
'url': 'https://amp.twimg.com/v/0ba0c3c7-0af3-4c0a-bed5-7efd1ffa2951',
|
|
|
|
|
|
|
|
'md5': '7df102d0b9fd7066b86f3159f8e81bf6',
|
|
|
|
|
|
|
|
'info_dict': {
|
|
|
|
|
|
|
|
'id': '0ba0c3c7-0af3-4c0a-bed5-7efd1ffa2951',
|
|
|
|
|
|
|
|
'ext': 'mp4',
|
|
|
|
|
|
|
|
'title': 'Twitter Video',
|
|
|
|
|
|
|
|
},
|
|
|
|
|
|
|
|
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
def _real_extract(self, url):
|
|
|
|
|
|
|
|
video_id = self._match_id(url)
|
|
|
|
|
|
|
|
webpage = self._download_webpage(url, video_id)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
vmap_url = self._html_search_meta(
|
|
|
|
|
|
|
|
'twitter:amplify:vmap', webpage, 'vmap url')
|
|
|
|
|
|
|
|
video_url = self._get_vmap_video_url(vmap_url, video_id)
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
return {
|
|
|
|
|
|
|
|
'id': video_id,
|
|
|
|
|
|
|
|
'title': 'Twitter Video',
|
|
|
|
|
|
|
|
'url': video_url,
|
|
|
|
|
|
|
|
}
|
|
|
|