1
0
Fork 0
mirror of https://github.com/ytdl-org/youtube-dl.git synced 2024-12-05 00:17:43 +00:00
youtube-dl/youtube_dl/extractor/breakcom.py

92 lines
2.9 KiB
Python
Raw Normal View History

2014-02-10 19:48:46 +00:00
from __future__ import unicode_literals
2013-06-23 20:59:51 +00:00
import re
from .common import InfoExtractor
from .youtube import YoutubeIE
2018-07-21 12:08:28 +00:00
from ..utils import (
int_or_none,
url_or_none,
)
2013-06-23 20:59:51 +00:00
class BreakIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?break\.com/video/(?P<display_id>[^/]+?)(?:-(?P<id>\d+))?(?:[/?#&]|$)'
_TESTS = [{
2014-02-10 19:48:46 +00:00
'url': 'http://www.break.com/video/when-girls-act-like-guys-2468056',
'info_dict': {
'id': '2468056',
'ext': 'mp4',
'title': 'When Girls Act Like D-Bags',
2015-08-08 15:58:24 +00:00
'age_limit': 13,
},
}, {
# youtube embed
'url': 'http://www.break.com/video/someone-forgot-boat-brakes-work',
'info_dict': {
'id': 'RrrDLdeL2HQ',
'ext': 'mp4',
'title': 'Whale Watching Boat Crashing Into San Diego Dock',
'description': 'md5:afc1b2772f0a8468be51dd80eb021069',
'upload_date': '20160331',
'uploader': 'Steve Holden',
'uploader_id': 'sdholden07',
},
'params': {
'skip_download': True,
}
}, {
'url': 'http://www.break.com/video/ugc/baby-flex-2773063',
'only_matching': True,
}]
2013-06-23 20:59:51 +00:00
def _real_extract(self, url):
display_id, video_id = re.match(self._VALID_URL, url).groups()
webpage = self._download_webpage(url, display_id)
youtube_url = YoutubeIE._extract_url(webpage)
if youtube_url:
return self.url_result(youtube_url, ie=YoutubeIE.ie_key())
content = self._parse_json(
self._search_regex(
r'(?s)content["\']\s*:\s*(\[.+?\])\s*[,\n]', webpage,
'content'),
display_id)
2014-10-03 12:37:47 +00:00
formats = []
for video in content:
2018-07-21 12:08:28 +00:00
video_url = url_or_none(video.get('url'))
if not video_url:
continue
bitrate = int_or_none(self._search_regex(
r'(\d+)_kbps', video_url, 'tbr', default=None))
2014-10-03 12:37:47 +00:00
formats.append({
'url': video_url,
'format_id': 'http-%d' % bitrate if bitrate else 'http',
'tbr': bitrate,
2014-10-03 12:37:47 +00:00
})
self._sort_formats(formats)
2014-10-03 12:37:47 +00:00
title = self._search_regex(
(r'title["\']\s*:\s*(["\'])(?P<value>(?:(?!\1).)+)\1',
r'<h1[^>]*>(?P<value>[^<]+)'), webpage, 'title', group='value')
2014-10-03 12:37:47 +00:00
def get(key, name):
return int_or_none(self._search_regex(
r'%s["\']\s*:\s*["\'](\d+)' % key, webpage, name,
default=None))
age_limit = get('ratings', 'age limit')
video_id = video_id or get('pid', 'video id') or display_id
2014-10-03 12:37:47 +00:00
2014-02-10 19:48:46 +00:00
return {
'id': video_id,
'display_id': display_id,
'title': title,
'thumbnail': self._og_search_thumbnail(webpage),
'age_limit': age_limit,
2014-10-03 12:37:47 +00:00
'formats': formats,
2014-02-10 19:48:46 +00:00
}