1
0
mirror of https://github.com/l1ving/youtube-dl synced 2024-12-29 10:05:35 +08:00
youtube-dl/youtube_dl/extractor/veoh.py

122 lines
4.4 KiB
Python

from __future__ import unicode_literals
import re
import json
from .common import InfoExtractor
from ..utils import (
compat_urllib_request,
int_or_none,
)
class VeohIE(InfoExtractor):
_VALID_URL = r'http://(?:www\.)?veoh\.com/(?:watch|iphone/#_Watch)/(?P<id>(?:v|yapi-)[\da-zA-Z]+)'
_TESTS = [
{
'url': 'http://www.veoh.com/watch/v56314296nk7Zdmz3',
'md5': '620e68e6a3cff80086df3348426c9ca3',
'info_dict': {
'id': '56314296',
'ext': 'mp4',
'title': 'Straight Backs Are Stronger',
'uploader': 'LUMOback',
'description': 'At LUMOback, we believe straight backs are stronger. The LUMOback Posture & Movement Sensor: It gently vibrates when you slouch, inspiring improved posture and mobility. Use the app to track your data and improve your posture over time. ',
},
},
{
'url': 'http://www.veoh.com/watch/v27701988pbTc4wzN?h1=Chile+workers+cover+up+to+avoid+skin+damage',
'md5': '4a6ff84b87d536a6a71e6aa6c0ad07fa',
'info_dict': {
'id': '27701988',
'ext': 'mp4',
'title': 'Chile workers cover up to avoid skin damage',
'description': 'md5:2bd151625a60a32822873efc246ba20d',
'uploader': 'afp-news',
'duration': 123,
},
},
{
'url': 'http://www.veoh.com/watch/v69525809F6Nc4frX',
'md5': '4fde7b9e33577bab2f2f8f260e30e979',
'note': 'Embedded ooyala video',
'info_dict': {
'id': '69525809',
'ext': 'mp4',
'title': 'Doctors Alter Plan For Preteen\'s Weight Loss Surgery',
'description': 'md5:f5a11c51f8fb51d2315bca0937526891',
'uploader': 'newsy-videos',
},
},
]
def _extract_formats(self, source):
formats = []
link = source.get('aowPermalink')
if link:
formats.append({
'url': link,
'ext': 'mp4',
'format_id': 'aow',
})
link = source.get('fullPreviewHashLowPath')
if link:
formats.append({
'url': link,
'format_id': 'low',
})
link = source.get('fullPreviewHashHighPath')
if link:
formats.append({
'url': link,
'format_id': 'high',
})
return formats
def _extract_video(self, source):
return {
'id': source.get('videoId'),
'title': source.get('title'),
'description': source.get('description'),
'thumbnail': source.get('highResImage') or source.get('medResImage'),
'uploader': source.get('username'),
'duration': int_or_none(source.get('length')),
'view_count': int_or_none(source.get('views')),
'age_limit': 18 if source.get('isMature') == 'true' or source.get('isSexy') == 'true' else 0,
'formats': self._extract_formats(source),
}
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
if video_id.startswith('v'):
rsp = self._download_xml(
r'http://www.veoh.com/api/findByPermalink?permalink=%s' % video_id, video_id, 'Downloading video XML')
if rsp.get('stat') == 'ok':
return self._extract_video(rsp.find('./videoList/video'))
webpage = self._download_webpage(url, video_id)
age_limit = 0
if 'class="adultwarning-container"' in webpage:
self.report_age_confirmation()
age_limit = 18
request = compat_urllib_request.Request(url)
request.add_header('Cookie', 'confirmedAdult=true')
webpage = self._download_webpage(request, video_id)
m_youtube = re.search(r'http://www\.youtube\.com/v/(.*?)(\&|"|\?)', webpage)
if m_youtube is not None:
youtube_id = m_youtube.group(1)
self.to_screen('%s: detected Youtube video.' % video_id)
return self.url_result(youtube_id, 'Youtube')
info = json.loads(
self._search_regex(r'videoDetailsJSON = \'({.*?})\';', webpage, 'info').replace('\\\'', '\''))
video = self._extract_video(info)
video['age_limit'] = age_limit
return video