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