ytdl-nightly/youtube_dl/extractor/tvigle.py

120 lines
4.1 KiB
Python
Raw Normal View History

2016-10-02 13:39:18 +02:00
# coding: utf-8
2014-03-02 13:59:34 +01:00
from __future__ import unicode_literals
import re
2014-03-02 13:59:34 +01:00
from .common import InfoExtractor
from ..utils import (
2015-06-03 16:52:47 +02:00
ExtractorError,
2014-09-03 14:59:36 +02:00
float_or_none,
2015-05-30 23:00:13 +02:00
int_or_none,
2014-12-08 17:03:02 +01:00
parse_age_limit,
2014-03-02 13:59:34 +01:00
)
class TvigleIE(InfoExtractor):
IE_NAME = 'tvigle'
IE_DESC = 'Интернет-телевидение Tvigle.ru'
_VALID_URL = r'https?://(?:www\.)?(?:tvigle\.ru/(?:[^/]+/)+(?P<display_id>[^/]+)/$|cloud\.tvigle\.ru/video/(?P<id>\d+))'
2017-02-25 17:27:14 +01:00
_GEO_BYPASS = False
_GEO_COUNTRIES = ['RU']
_TESTS = [
{
2014-12-08 17:03:02 +01:00
'url': 'http://www.tvigle.ru/video/sokrat/',
'md5': '36514aed3657d4f70b4b2cef8eb520cd',
'info_dict': {
2014-12-08 17:03:02 +01:00
'id': '1848932',
'display_id': 'sokrat',
'ext': 'flv',
'title': 'Сократ',
2015-05-30 23:00:13 +02:00
'description': 'md5:d6b92ffb7217b4b8ebad2e7665253c17',
2014-12-08 17:03:02 +01:00
'duration': 6586,
2015-05-30 23:00:13 +02:00
'age_limit': 12,
},
2015-06-03 16:53:54 +02:00
'skip': 'georestricted',
},
{
2014-09-03 14:59:36 +02:00
'url': 'http://www.tvigle.ru/video/vladimir-vysotskii/vedushchii-teleprogrammy-60-minut-ssha-o-vladimire-vysotskom/',
2015-05-30 23:00:13 +02:00
'md5': 'e7efe5350dd5011d0de6550b53c3ba7b',
'info_dict': {
2014-09-03 14:59:36 +02:00
'id': '5142516',
2015-05-30 23:00:13 +02:00
'ext': 'flv',
'title': 'Ведущий телепрограммы «60 минут» (США) о Владимире Высоцком',
'description': 'md5:027f7dc872948f14c96d19b4178428a4',
2014-09-03 14:59:36 +02:00
'duration': 186.080,
'age_limit': 0,
},
2015-06-03 16:53:54 +02:00
'skip': 'georestricted',
}, {
'url': 'https://cloud.tvigle.ru/video/5267604/',
'only_matching': True,
}
]
2014-03-02 13:59:34 +01:00
def _real_extract(self, url):
mobj = re.match(self._VALID_URL, url)
video_id = mobj.group('id')
display_id = mobj.group('display_id')
2014-03-02 13:59:34 +01:00
if not video_id:
webpage = self._download_webpage(url, display_id)
video_id = self._html_search_regex(
2016-04-20 19:52:41 +02:00
(r'<div[^>]+class=["\']player["\'][^>]+id=["\'](\d+)',
r'var\s+cloudId\s*=\s*["\'](\d+)',
r'class="video-preview current_playing" id="(\d+)"'),
webpage, 'video id')
2014-03-02 13:59:34 +01:00
2014-09-03 14:59:36 +02:00
video_data = self._download_json(
'http://cloud.tvigle.ru/api/play/video/%s/' % video_id, display_id)
2014-03-02 13:59:34 +01:00
2014-09-03 14:59:36 +02:00
item = video_data['playlist']['items'][0]
2015-06-03 16:52:47 +02:00
videos = item.get('videos')
error_message = item.get('errorMessage')
if not videos and error_message:
2017-02-25 17:27:14 +01:00
if item.get('isGeoBlocked') is True:
self.raise_geo_restricted(
msg=error_message, countries=self._GEO_COUNTRIES)
else:
raise ExtractorError(
'%s returned error: %s' % (self.IE_NAME, error_message),
expected=True)
2015-06-03 16:52:47 +02:00
2014-09-03 14:59:36 +02:00
title = item['title']
2015-05-30 23:01:41 +02:00
description = item.get('description')
thumbnail = item.get('thumbnail')
2014-12-08 17:03:02 +01:00
duration = float_or_none(item.get('durationMilliseconds'), 1000)
age_limit = parse_age_limit(item.get('ageRestrictions'))
2014-03-02 13:59:34 +01:00
2014-09-03 14:59:36 +02:00
formats = []
for vcodec, fmts in item['videos'].items():
2016-04-21 18:15:20 +02:00
if vcodec == 'hls':
continue
2015-05-30 23:00:13 +02:00
for format_id, video_url in fmts.items():
if format_id == 'm3u8':
continue
height = self._search_regex(
r'^(\d+)[pP]$', format_id, 'height', default=None)
2014-09-03 14:59:36 +02:00
formats.append({
'url': video_url,
2015-05-30 23:00:13 +02:00
'format_id': '%s-%s' % (vcodec, format_id),
2014-09-03 14:59:36 +02:00
'vcodec': vcodec,
2015-05-30 23:00:13 +02:00
'height': int_or_none(height),
2015-05-30 23:01:41 +02:00
'filesize': int_or_none(item.get('video_files_size', {}).get(vcodec, {}).get(format_id)),
2014-09-03 14:59:36 +02:00
})
2014-03-02 13:59:34 +01:00
self._sort_formats(formats)
return {
'id': video_id,
2014-09-03 14:59:36 +02:00
'display_id': display_id,
2014-03-02 13:59:34 +01:00
'title': title,
'description': description,
'thumbnail': thumbnail,
2014-09-03 14:59:36 +02:00
'duration': duration,
'age_limit': age_limit,
2014-03-02 13:59:34 +01:00
'formats': formats,
2014-11-23 20:41:03 +01:00
}