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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import json
   5 import re
   6 import time
   7
   8 from .common import InfoExtractor
   9 from ..compat import compat_urlparse
  10 from ..utils import (
  11     USER_AGENTS,
  12     int_or_none,
  13     update_url_query,
  14 )
  15
  16
  17 class DPlayIE(InfoExtractor):
  18     _VALID_URL = r'https?://(?P<domain>it\.dplay\.com|www\.dplay\.(?:dk|se|no))/[^/]+/(?P<id>[^/?#]+)'
  19
  20     _TESTS = [{
  21         # geo restricted, via direct unsigned hls URL
  22         'url': 'http://it.dplay.com/take-me-out/stagione-1-episodio-25/',
  23         'info_dict': {
  24             'id': '1255600',
  25             'display_id': 'stagione-1-episodio-25',
  26             'ext': 'mp4',
  27             'title': 'Episodio 25',
  28             'description': 'md5:cae5f40ad988811b197d2d27a53227eb',
  29             'duration': 2761,
  30             'timestamp': 1454701800,
  31             'upload_date': '20160205',
  32             'creator': 'RTIT',
  33             'series': 'Take me out',
  34             'season_number': 1,
  35             'episode_number': 25,
  36             'age_limit': 0,
  37         },
  38         'expected_warnings': ['Unable to download f4m manifest'],
  39     }, {
  40         # non geo restricted, via secure api, unsigned download hls URL
  41         'url': 'http://www.dplay.se/nugammalt-77-handelser-som-format-sverige/season-1-svensken-lar-sig-njuta-av-livet/',
  42         'info_dict': {
  43             'id': '3172',
  44             'display_id': 'season-1-svensken-lar-sig-njuta-av-livet',
  45             'ext': 'mp4',
  46             'title': 'Svensken lär sig njuta av livet',
  47             'description': 'md5:d3819c9bccffd0fe458ca42451dd50d8',
  48             'duration': 2650,
  49             'timestamp': 1365454320,
  50             'upload_date': '20130408',
  51             'creator': 'Kanal 5 (Home)',
  52             'series': 'Nugammalt - 77 händelser som format Sverige',
  53             'season_number': 1,
  54             'episode_number': 1,
  55             'age_limit': 0,
  56         },
  57     }, {
  58         # geo restricted, via secure api, unsigned download hls URL
  59         'url': 'http://www.dplay.dk/mig-og-min-mor/season-6-episode-12/',
  60         'info_dict': {
  61             'id': '70816',
  62             'display_id': 'season-6-episode-12',
  63             'ext': 'mp4',
  64             'title': 'Episode 12',
  65             'description': 'md5:9c86e51a93f8a4401fc9641ef9894c90',
  66             'duration': 2563,
  67             'timestamp': 1429696800,
  68             'upload_date': '20150422',
  69             'creator': 'Kanal 4 (Home)',
  70             'series': 'Mig og min mor',
  71             'season_number': 6,
  72             'episode_number': 12,
  73             'age_limit': 0,
  74         },
  75     }, {
  76         # geo restricted, via direct unsigned hls URL
  77         'url': 'http://www.dplay.no/pga-tour/season-1-hoydepunkter-18-21-februar/',
  78         'only_matching': True,
  79     }]
  80
  81     def _real_extract(self, url):
  82         mobj = re.match(self._VALID_URL, url)
  83         display_id = mobj.group('id')
  84         domain = mobj.group('domain')
  85
  86         webpage = self._download_webpage(url, display_id)
  87
  88         video_id = self._search_regex(
  89             r'data-video-id=["\'](\d+)', webpage, 'video id')
  90
  91         info = self._download_json(
  92             'http://%s/api/v2/ajax/videos?video_id=%s' % (domain, video_id),
  93             video_id)['data'][0]
  94
  95         title = info['title']
  96
  97         PROTOCOLS = ('hls', 'hds')
  98         formats = []
  99
 100         def extract_formats(protocol, manifest_url):
 101             if protocol == 'hls':
 102                 m3u8_formats = self._extract_m3u8_formats(
 103                     manifest_url, video_id, ext='mp4',
 104                     entry_protocol='m3u8_native', m3u8_id=protocol, fatal=False)
 105                 # Sometimes final URLs inside m3u8 are unsigned, let's fix this
 106                 # ourselves. Also fragments' URLs are only served signed for
 107                 # Safari user agent.
 108                 query = compat_urlparse.parse_qs(compat_urlparse.urlparse(manifest_url).query)
 109                 for m3u8_format in m3u8_formats:
 110                     m3u8_format.update({
 111                         'url': update_url_query(m3u8_format['url'], query),
 112                         'http_headers': {
 113                             'User-Agent': USER_AGENTS['Safari'],
 114                         },
 115                     })
 116                 formats.extend(m3u8_formats)
 117             elif protocol == 'hds':
 118                 formats.extend(self._extract_f4m_formats(
 119                     manifest_url + '&hdcore=3.8.0&plugin=flowplayer-3.8.0.0',
 120                     video_id, f4m_id=protocol, fatal=False))
 121
 122         domain_tld = domain.split('.')[-1]
 123         if domain_tld in ('se', 'dk', 'no'):
 124             for protocol in PROTOCOLS:
 125                 # Providing dsc-geo allows to bypass geo restriction in some cases
 126                 self._set_cookie(
 127                     'secure.dplay.%s' % domain_tld, 'dsc-geo',
 128                     json.dumps({
 129                         'countryCode': domain_tld.upper(),
 130                         'expiry': (time.time() + 20 * 60) * 1000,
 131                     }))
 132                 stream = self._download_json(
 133                     'https://secure.dplay.%s/secure/api/v2/user/authorization/stream/%s?stream_type=%s'
 134                     % (domain_tld, video_id, protocol), video_id,
 135                     'Downloading %s stream JSON' % protocol, fatal=False)
 136                 if stream and stream.get(protocol):
 137                     extract_formats(protocol, stream[protocol])
 138
 139         # The last resort is to try direct unsigned hls/hds URLs from info dictionary.
 140         # Sometimes this does work even when secure API with dsc-geo has failed (e.g.
 141         # http://www.dplay.no/pga-tour/season-1-hoydepunkter-18-21-februar/).
 142         if not formats:
 143             for protocol in PROTOCOLS:
 144                 if info.get(protocol):
 145                     extract_formats(protocol, info[protocol])
 146
 147         self._sort_formats(formats)
 148
 149         subtitles = {}
 150         for lang in ('se', 'sv', 'da', 'nl', 'no'):
 151             for format_id in ('web_vtt', 'vtt', 'srt'):
 152                 subtitle_url = info.get('subtitles_%s_%s' % (lang, format_id))
 153                 if subtitle_url:
 154                     subtitles.setdefault(lang, []).append({'url': subtitle_url})
 155
 156         return {
 157             'id': video_id,
 158             'display_id': display_id,
 159             'title': title,
 160             'description': info.get('video_metadata_longDescription'),
 161             'duration': int_or_none(info.get('video_metadata_length'), scale=1000),
 162             'timestamp': int_or_none(info.get('video_publish_date')),
 163             'creator': info.get('video_metadata_homeChannel'),
 164             'series': info.get('video_metadata_show'),
 165             'season_number': int_or_none(info.get('season')),
 166             'episode_number': int_or_none(info.get('episode')),
 167             'age_limit': int_or_none(info.get('minimum_age')),
 168             'formats': formats,
 169             'subtitles': subtitles,
 170         }