Merge branch 'atomicdryad-pr-twitter'

This commit is contained in:
Yen Chi Hsuan 2015-10-18 18:44:21 +08:00
commit 09ff81316e
2 changed files with 90 additions and 15 deletions

View File

@ -690,7 +690,7 @@ from .twitch import (
TwitchBookmarksIE,
TwitchStreamIE,
)
from .twitter import TwitterCardIE
from .twitter import TwitterCardIE, TwitterIE
from .ubu import UbuIE
from .udemy import (
UdemyIE,

View File

@ -1,3 +1,4 @@
# coding: utf-8
from __future__ import unicode_literals
import re
@ -6,23 +7,38 @@ from .common import InfoExtractor
from ..compat import compat_urllib_request
from ..utils import (
float_or_none,
unescapeHTML,
xpath_text,
remove_end,
)
class TwitterCardIE(InfoExtractor):
IE_NAME = 'twitter:card'
_VALID_URL = r'https?://(?:www\.)?twitter\.com/i/cards/tfw/v1/(?P<id>\d+)'
_TEST = {
'url': 'https://twitter.com/i/cards/tfw/v1/560070183650213889',
'md5': 'a74f50b310c83170319ba16de6955192',
'info_dict': {
'id': '560070183650213889',
'ext': 'mp4',
'title': 'TwitterCard',
'thumbnail': 're:^https?://.*\.jpg$',
'duration': 30.033,
_TESTS = [
{
'url': 'https://twitter.com/i/cards/tfw/v1/560070183650213889',
'md5': '7d2f6b4d2eb841a7ccc893d479bfceb4',
'info_dict': {
'id': '560070183650213889',
'ext': 'mp4',
'title': 'TwitterCard',
'thumbnail': 're:^https?://.*\.jpg$',
'duration': 30.033,
}
},
}
{
'url': 'https://twitter.com/i/cards/tfw/v1/623160978427936768',
'md5': '7ee2a553b63d1bccba97fbed97d9e1c8',
'info_dict': {
'id': '623160978427936768',
'ext': 'mp4',
'title': 'TwitterCard',
'thumbnail': 're:^https?://.*\.jpg',
'duration': 80.155,
},
}
]
def _real_extract(self, url):
video_id = self._match_id(url)
@ -40,10 +56,18 @@ class TwitterCardIE(InfoExtractor):
request.add_header('User-Agent', user_agent)
webpage = self._download_webpage(request, video_id)
config = self._parse_json(
unescapeHTML(self._search_regex(
r'data-player-config="([^"]+)"', webpage, 'data player config')),
config = self._parse_json(self._html_search_regex(
r'data-player-config="([^"]+)"', webpage, 'data player config'),
video_id)
if 'playlist' not 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({
'url': video_url,
})
break # same video regardless of UA
continue
video_url = config['playlist'][0]['source']
@ -70,3 +94,54 @@ class TwitterCardIE(InfoExtractor):
'duration': duration,
'formats': formats,
}
class TwitterIE(InfoExtractor):
IE_NAME = 'twitter'
_VALID_URL = r'https?://(?:www\.|m\.|mobile\.)?twitter\.com/(?P<user_id>[^/]+)/status/(?P<id>\d+)'
_TEMPLATE_URL = 'https://twitter.com/%s/status/%s'
_TEST = {
'url': 'https://twitter.com/freethenipple/status/643211948184596480',
'md5': '31cd83a116fc41f99ae3d909d4caf6a0',
'info_dict': {
'id': '643211948184596480',
'ext': 'mp4',
'title': 'FREE THE NIPPLE - FTN supporters on Hollywood Blvd today!',
'thumbnail': 're:^https?://.*\.jpg',
'duration': 12.922,
'description': 'FREE THE NIPPLE on Twitter: "FTN supporters on Hollywood Blvd today! http://t.co/c7jHH749xJ"',
'uploader': 'FREE THE NIPPLE',
'uploader_id': 'freethenipple',
},
}
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
user_id = mobj.group('user_id')
twid = mobj.group('id')
webpage = self._download_webpage(self._TEMPLATE_URL % (user_id, twid), twid)
username = remove_end(self._og_search_title(webpage), ' on Twitter')
title = self._og_search_description(webpage).strip('').replace('\n', ' ')
# strip 'https -_t.co_BJYgOjSeGA' junk from filenames
mobj = re.match(r'“(.*)\s+(https?://[^ ]+)”', title)
title, short_url = mobj.groups()
card_id = self._search_regex(
r'["\']/i/cards/tfw/v1/(\d+)', webpage, 'twitter card url')
card_url = 'https://twitter.com/i/cards/tfw/v1/' + card_id
return {
'_type': 'url_transparent',
'ie_key': 'TwitterCard',
'uploader_id': user_id,
'uploader': username,
'url': card_url,
'webpage_url': url,
'description': '%s on Twitter: "%s %s"' % (username, title, short_url),
'title': username + ' - ' + title,
}