youtube_dl/extractor/tvigle.py

   1 # encoding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import (
   8     float_or_none,
   9     int_or_none,
  10     parse_age_limit,
  11 )
  12
  13
  14 class TvigleIE(InfoExtractor):
  15     IE_NAME = 'tvigle'
  16     IE_DESC = 'Интернет-телевидение Tvigle.ru'
  17     _VALID_URL = r'https?://(?:www\.)?(?:tvigle\.ru/(?:[^/]+/)+(?P<display_id>[^/]+)/$|cloud\.tvigle\.ru/video/(?P<id>\d+))'
  18
  19     _TESTS = [
  20         {
  21             'url': 'http://www.tvigle.ru/video/sokrat/',
  22             'md5': '36514aed3657d4f70b4b2cef8eb520cd',
  23             'info_dict': {
  24                 'id': '1848932',
  25                 'display_id': 'sokrat',
  26                 'ext': 'flv',
  27                 'title': 'Сократ',
  28                 'description': 'md5:d6b92ffb7217b4b8ebad2e7665253c17',
  29                 'duration': 6586,
  30                 'age_limit': 12,
  31             },
  32         },
  33         {
  34             'url': 'http://www.tvigle.ru/video/vladimir-vysotskii/vedushchii-teleprogrammy-60-minut-ssha-o-vladimire-vysotskom/',
  35             'md5': 'e7efe5350dd5011d0de6550b53c3ba7b',
  36             'info_dict': {
  37                 'id': '5142516',
  38                 'ext': 'flv',
  39                 'title': 'Ведущий телепрограммы «60 минут» (США) о Владимире Высоцком',
  40                 'description': 'md5:027f7dc872948f14c96d19b4178428a4',
  41                 'duration': 186.080,
  42                 'age_limit': 0,
  43             },
  44         }, {
  45             'url': 'https://cloud.tvigle.ru/video/5267604/',
  46             'only_matching': True,
  47         }
  48     ]
  49
  50     def _real_extract(self, url):
  51         mobj = re.match(self._VALID_URL, url)
  52         video_id = mobj.group('id')
  53         display_id = mobj.group('display_id')
  54
  55         if not video_id:
  56             webpage = self._download_webpage(url, display_id)
  57             video_id = self._html_search_regex(
  58                 r'class="video-preview current_playing" id="(\d+)">',
  59                 webpage, 'video id')
  60
  61         video_data = self._download_json(
  62             'http://cloud.tvigle.ru/api/play/video/%s/' % video_id, display_id)
  63
  64         item = video_data['playlist']['items'][0]
  65
  66         title = item['title']
  67         description = item.get('description')
  68         thumbnail = item.get('thumbnail')
  69         duration = float_or_none(item.get('durationMilliseconds'), 1000)
  70         age_limit = parse_age_limit(item.get('ageRestrictions'))
  71
  72         formats = []
  73         for vcodec, fmts in item['videos'].items():
  74             for format_id, video_url in fmts.items():
  75                 if format_id == 'm3u8':
  76                     formats.extend(self._extract_m3u8_formats(
  77                         video_url, video_id, 'mp4', m3u8_id=vcodec))
  78                     continue
  79                 height = self._search_regex(
  80                     r'^(\d+)[pP]$', format_id, 'height', default=None)
  81                 formats.append({
  82                     'url': video_url,
  83                     'format_id': '%s-%s' % (vcodec, format_id),
  84                     'vcodec': vcodec,
  85                     'height': int_or_none(height),
  86                     'filesize': int_or_none(item.get('video_files_size', {}).get(vcodec, {}).get(format_id)),
  87                 })
  88         self._sort_formats(formats)
  89
  90         return {
  91             'id': video_id,
  92             'display_id': display_id,
  93             'title': title,
  94             'description': description,
  95             'thumbnail': thumbnail,
  96             'duration': duration,
  97             'age_limit': age_limit,
  98             'formats': formats,
  99         }