Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/tvigle.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import (
   8     ExtractorError,
   9     float_or_none,
  10     int_or_none,
  11     parse_age_limit,
  12 )
  13
  14
  15 class TvigleIE(InfoExtractor):
  16     IE_NAME = 'tvigle'
  17     IE_DESC = 'Интернет-телевидение Tvigle.ru'
  18     _VALID_URL = r'https?://(?:www\.)?(?:tvigle\.ru/(?:[^/]+/)+(?P<display_id>[^/]+)/$|cloud\.tvigle\.ru/video/(?P<id>\d+))'
  19
  20     _GEO_BYPASS = False
  21     _GEO_COUNTRIES = ['RU']
  22
  23     _TESTS = [
  24         {
  25             'url': 'http://www.tvigle.ru/video/sokrat/',
  26             'md5': '36514aed3657d4f70b4b2cef8eb520cd',
  27             'info_dict': {
  28                 'id': '1848932',
  29                 'display_id': 'sokrat',
  30                 'ext': 'flv',
  31                 'title': 'Сократ',
  32                 'description': 'md5:d6b92ffb7217b4b8ebad2e7665253c17',
  33                 'duration': 6586,
  34                 'age_limit': 12,
  35             },
  36             'skip': 'georestricted',
  37         },
  38         {
  39             'url': 'http://www.tvigle.ru/video/vladimir-vysotskii/vedushchii-teleprogrammy-60-minut-ssha-o-vladimire-vysotskom/',
  40             'md5': 'e7efe5350dd5011d0de6550b53c3ba7b',
  41             'info_dict': {
  42                 'id': '5142516',
  43                 'ext': 'flv',
  44                 'title': 'Ведущий телепрограммы «60 минут» (США) о Владимире Высоцком',
  45                 'description': 'md5:027f7dc872948f14c96d19b4178428a4',
  46                 'duration': 186.080,
  47                 'age_limit': 0,
  48             },
  49             'skip': 'georestricted',
  50         }, {
  51             'url': 'https://cloud.tvigle.ru/video/5267604/',
  52             'only_matching': True,
  53         }
  54     ]
  55
  56     def _real_extract(self, url):
  57         mobj = re.match(self._VALID_URL, url)
  58         video_id = mobj.group('id')
  59         display_id = mobj.group('display_id')
  60
  61         if not video_id:
  62             webpage = self._download_webpage(url, display_id)
  63             video_id = self._html_search_regex(
  64                 (r'<div[^>]+class=["\']player["\'][^>]+id=["\'](\d+)',
  65                  r'var\s+cloudId\s*=\s*["\'](\d+)',
  66                  r'class="video-preview current_playing" id="(\d+)"'),
  67                 webpage, 'video id')
  68
  69         video_data = self._download_json(
  70             'http://cloud.tvigle.ru/api/play/video/%s/' % video_id, display_id)
  71
  72         item = video_data['playlist']['items'][0]
  73
  74         videos = item.get('videos')
  75
  76         error_message = item.get('errorMessage')
  77         if not videos and error_message:
  78             if item.get('isGeoBlocked') is True:
  79                 self.raise_geo_restricted(
  80                     msg=error_message, countries=self._GEO_COUNTRIES)
  81             else:
  82                 raise ExtractorError(
  83                     '%s returned error: %s' % (self.IE_NAME, error_message),
  84                     expected=True)
  85
  86         title = item['title']
  87         description = item.get('description')
  88         thumbnail = item.get('thumbnail')
  89         duration = float_or_none(item.get('durationMilliseconds'), 1000)
  90         age_limit = parse_age_limit(item.get('ageRestrictions'))
  91
  92         formats = []
  93         for vcodec, fmts in item['videos'].items():
  94             if vcodec == 'hls':
  95                 continue
  96             for format_id, video_url in fmts.items():
  97                 if format_id == 'm3u8':
  98                     continue
  99                 height = self._search_regex(
 100                     r'^(\d+)[pP]$', format_id, 'height', default=None)
 101                 formats.append({
 102                     'url': video_url,
 103                     'format_id': '%s-%s' % (vcodec, format_id),
 104                     'vcodec': vcodec,
 105                     'height': int_or_none(height),
 106                     'filesize': int_or_none(item.get('video_files_size', {}).get(vcodec, {}).get(format_id)),
 107                 })
 108         self._sort_formats(formats)
 109
 110         return {
 111             'id': video_id,
 112             'display_id': display_id,
 113             'title': title,
 114             'description': description,
 115             'thumbnail': thumbnail,
 116             'duration': duration,
 117             'age_limit': age_limit,
 118             'formats': formats,
 119         }