]> Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/mitele.py
New upstream version 2018.06.18
[youtubedl] / youtube_dl / extractor / mitele.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 import json
5 import uuid
6
7 from .common import InfoExtractor
8 from .ooyala import OoyalaIE
9 from ..compat import (
10 compat_str,
11 compat_urlparse,
12 )
13 from ..utils import (
14 int_or_none,
15 extract_attributes,
16 determine_ext,
17 smuggle_url,
18 parse_duration,
19 )
20
21
22 class MiTeleBaseIE(InfoExtractor):
23 def _get_player_info(self, url, webpage):
24 player_data = extract_attributes(self._search_regex(
25 r'(?s)(<ms-video-player.+?</ms-video-player>)',
26 webpage, 'ms video player'))
27 video_id = player_data['data-media-id']
28 if player_data.get('data-cms-id') == 'ooyala':
29 return self.url_result(
30 'ooyala:%s' % video_id, ie=OoyalaIE.ie_key(), video_id=video_id)
31 config_url = compat_urlparse.urljoin(url, player_data['data-config'])
32 config = self._download_json(
33 config_url, video_id, 'Downloading config JSON')
34 mmc_url = config['services']['mmc']
35
36 duration = None
37 formats = []
38 for m_url in (mmc_url, mmc_url.replace('/flash.json', '/html5.json')):
39 mmc = self._download_json(
40 m_url, video_id, 'Downloading mmc JSON')
41 if not duration:
42 duration = int_or_none(mmc.get('duration'))
43 for location in mmc['locations']:
44 gat = self._proto_relative_url(location.get('gat'), 'http:')
45 gcp = location.get('gcp')
46 ogn = location.get('ogn')
47 if None in (gat, gcp, ogn):
48 continue
49 token_data = {
50 'gcp': gcp,
51 'ogn': ogn,
52 'sta': 0,
53 }
54 media = self._download_json(
55 gat, video_id, data=json.dumps(token_data).encode('utf-8'),
56 headers={
57 'Content-Type': 'application/json;charset=utf-8',
58 'Referer': url,
59 })
60 stream = media.get('stream') or media.get('file')
61 if not stream:
62 continue
63 ext = determine_ext(stream)
64 if ext == 'f4m':
65 formats.extend(self._extract_f4m_formats(
66 stream + '&hdcore=3.2.0&plugin=aasp-3.2.0.77.18',
67 video_id, f4m_id='hds', fatal=False))
68 elif ext == 'm3u8':
69 formats.extend(self._extract_m3u8_formats(
70 stream, video_id, 'mp4', 'm3u8_native',
71 m3u8_id='hls', fatal=False))
72 self._sort_formats(formats)
73
74 return {
75 'id': video_id,
76 'formats': formats,
77 'thumbnail': player_data.get('data-poster') or config.get('poster', {}).get('imageUrl'),
78 'duration': duration,
79 }
80
81
82 class MiTeleIE(InfoExtractor):
83 IE_DESC = 'mitele.es'
84 _VALID_URL = r'https?://(?:www\.)?mitele\.es/(?:[^/]+/)+(?P<id>[^/]+)/player'
85
86 _TESTS = [{
87 'url': 'http://www.mitele.es/programas-tv/diario-de/57b0dfb9c715da65618b4afa/player',
88 'info_dict': {
89 'id': '57b0dfb9c715da65618b4afa',
90 'ext': 'mp4',
91 'title': 'Tor, la web invisible',
92 'description': 'md5:3b6fce7eaa41b2d97358726378d9369f',
93 'series': 'Diario de',
94 'season': 'La redacción',
95 'season_number': 14,
96 'season_id': 'diario_de_t14_11981',
97 'episode': 'Programa 144',
98 'episode_number': 3,
99 'thumbnail': r're:(?i)^https?://.*\.jpg$',
100 'duration': 2913,
101 },
102 'add_ie': ['Ooyala'],
103 }, {
104 # no explicit title
105 'url': 'http://www.mitele.es/programas-tv/cuarto-milenio/57b0de3dc915da14058b4876/player',
106 'info_dict': {
107 'id': '57b0de3dc915da14058b4876',
108 'ext': 'mp4',
109 'title': 'Cuarto Milenio Temporada 6 Programa 226',
110 'description': 'md5:5ff132013f0cd968ffbf1f5f3538a65f',
111 'series': 'Cuarto Milenio',
112 'season': 'Temporada 6',
113 'season_number': 6,
114 'season_id': 'cuarto_milenio_t06_12715',
115 'episode': 'Programa 226',
116 'episode_number': 24,
117 'thumbnail': r're:(?i)^https?://.*\.jpg$',
118 'duration': 7313,
119 },
120 'params': {
121 'skip_download': True,
122 },
123 'add_ie': ['Ooyala'],
124 }, {
125 'url': 'http://www.mitele.es/series-online/la-que-se-avecina/57aac5c1c915da951a8b45ed/player',
126 'only_matching': True,
127 }]
128
129 def _real_extract(self, url):
130 video_id = self._match_id(url)
131 webpage = self._download_webpage(url, video_id)
132
133 gigya_url = self._search_regex(
134 r'<gigya-api>[^>]*</gigya-api>[^>]*<script\s+src="([^"]*)">[^>]*</script>',
135 webpage, 'gigya', default=None)
136 gigya_sc = self._download_webpage(
137 compat_urlparse.urljoin('http://www.mitele.es/', gigya_url),
138 video_id, 'Downloading gigya script')
139
140 # Get a appKey/uuid for getting the session key
141 appKey = self._search_regex(
142 r'constant\s*\(\s*["\']_appGridApplicationKey["\']\s*,\s*["\']([0-9a-f]+)',
143 gigya_sc, 'appKey')
144
145 session_json = self._download_json(
146 'https://appgrid-api.cloud.accedo.tv/session',
147 video_id, 'Downloading session keys', query={
148 'appKey': appKey,
149 'uuid': compat_str(uuid.uuid4()),
150 })
151
152 paths = self._download_json(
153 'https://appgrid-api.cloud.accedo.tv/metadata/general_configuration,%20web_configuration',
154 video_id, 'Downloading paths JSON',
155 query={'sessionKey': compat_str(session_json['sessionKey'])})
156
157 ooyala_s = paths['general_configuration']['api_configuration']['ooyala_search']
158 source = self._download_json(
159 'http://%s%s%s/docs/%s' % (
160 ooyala_s['base_url'], ooyala_s['full_path'],
161 ooyala_s['provider_id'], video_id),
162 video_id, 'Downloading data JSON', query={
163 'include_titles': 'Series,Season',
164 'product_name': 'test',
165 'format': 'full',
166 })['hits']['hits'][0]['_source']
167
168 embedCode = source['offers'][0]['embed_codes'][0]
169 titles = source['localizable_titles'][0]
170
171 title = titles.get('title_medium') or titles['title_long']
172
173 description = titles.get('summary_long') or titles.get('summary_medium')
174
175 def get(key1, key2):
176 value1 = source.get(key1)
177 if not value1 or not isinstance(value1, list):
178 return
179 if not isinstance(value1[0], dict):
180 return
181 return value1[0].get(key2)
182
183 series = get('localizable_titles_series', 'title_medium')
184
185 season = get('localizable_titles_season', 'title_medium')
186 season_number = int_or_none(source.get('season_number'))
187 season_id = source.get('season_id')
188
189 episode = titles.get('title_sort_name')
190 episode_number = int_or_none(source.get('episode_number'))
191
192 duration = parse_duration(get('videos', 'duration'))
193
194 return {
195 '_type': 'url_transparent',
196 # for some reason only HLS is supported
197 'url': smuggle_url('ooyala:' + embedCode, {'supportedformats': 'm3u8,dash'}),
198 'id': video_id,
199 'title': title,
200 'description': description,
201 'series': series,
202 'season': season,
203 'season_number': season_number,
204 'season_id': season_id,
205 'episode': episode,
206 'episode_number': episode_number,
207 'duration': duration,
208 'thumbnail': get('images', 'url'),
209 }