]> Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/telequebec.py
Update upstream source from tag 'upstream/2019.06.08'
[youtubedl] / youtube_dl / extractor / telequebec.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 from .common import InfoExtractor
5 from ..compat import compat_str
6 from ..utils import (
7 int_or_none,
8 smuggle_url,
9 try_get,
10 )
11
12
13 class TeleQuebecBaseIE(InfoExtractor):
14 @staticmethod
15 def _limelight_result(media_id):
16 return {
17 '_type': 'url_transparent',
18 'url': smuggle_url(
19 'limelight:media:' + media_id, {'geo_countries': ['CA']}),
20 'ie_key': 'LimelightMedia',
21 }
22
23
24 class TeleQuebecIE(TeleQuebecBaseIE):
25 _VALID_URL = r'https?://zonevideo\.telequebec\.tv/media/(?P<id>\d+)'
26 _TESTS = [{
27 # available till 01.01.2023
28 'url': 'http://zonevideo.telequebec.tv/media/37578/un-petit-choc-et-puis-repart/un-chef-a-la-cabane',
29 'info_dict': {
30 'id': '577116881b4b439084e6b1cf4ef8b1b3',
31 'ext': 'mp4',
32 'title': 'Un petit choc et puis repart!',
33 'description': 'md5:b04a7e6b3f74e32d7b294cffe8658374',
34 'upload_date': '20180222',
35 'timestamp': 1519326631,
36 },
37 'params': {
38 'skip_download': True,
39 },
40 }, {
41 # no description
42 'url': 'http://zonevideo.telequebec.tv/media/30261',
43 'only_matching': True,
44 }]
45
46 def _real_extract(self, url):
47 media_id = self._match_id(url)
48
49 media_data = self._download_json(
50 'https://mnmedias.api.telequebec.tv/api/v2/media/' + media_id,
51 media_id)['media']
52
53 info = self._limelight_result(media_data['streamInfo']['sourceId'])
54 info.update({
55 'title': media_data.get('title'),
56 'description': try_get(
57 media_data, lambda x: x['descriptions'][0]['text'], compat_str),
58 'duration': int_or_none(
59 media_data.get('durationInMilliseconds'), 1000),
60 })
61 return info
62
63
64 class TeleQuebecEmissionIE(TeleQuebecBaseIE):
65 _VALID_URL = r'''(?x)
66 https?://
67 (?:
68 [^/]+\.telequebec\.tv/emissions/|
69 (?:www\.)?telequebec\.tv/
70 )
71 (?P<id>[^?#&]+)
72 '''
73 _TESTS = [{
74 'url': 'http://lindicemcsween.telequebec.tv/emissions/100430013/des-soins-esthetiques-a-377-d-interets-annuels-ca-vous-tente',
75 'info_dict': {
76 'id': '66648a6aef914fe3badda25e81a4d50a',
77 'ext': 'mp4',
78 'title': "Des soins esthétiques à 377 % d'intérêts annuels, ça vous tente?",
79 'description': 'md5:369e0d55d0083f1fc9b71ffb640ea014',
80 'upload_date': '20171024',
81 'timestamp': 1508862118,
82 },
83 'params': {
84 'skip_download': True,
85 },
86 }, {
87 'url': 'http://bancpublic.telequebec.tv/emissions/emission-49/31986/jeunes-meres-sous-pression',
88 'only_matching': True,
89 }, {
90 'url': 'http://www.telequebec.tv/masha-et-michka/epi059masha-et-michka-3-053-078',
91 'only_matching': True,
92 }, {
93 'url': 'http://www.telequebec.tv/documentaire/bebes-sur-mesure/',
94 'only_matching': True,
95 }]
96
97 def _real_extract(self, url):
98 display_id = self._match_id(url)
99
100 webpage = self._download_webpage(url, display_id)
101
102 media_id = self._search_regex(
103 r'mediaUID\s*:\s*["\'][Ll]imelight_(?P<id>[a-z0-9]{32})', webpage,
104 'limelight id')
105
106 info = self._limelight_result(media_id)
107 info.update({
108 'title': self._og_search_title(webpage, default=None),
109 'description': self._og_search_description(webpage, default=None),
110 })
111 return info
112
113
114 class TeleQuebecLiveIE(InfoExtractor):
115 _VALID_URL = r'https?://zonevideo\.telequebec\.tv/(?P<id>endirect)'
116 _TEST = {
117 'url': 'http://zonevideo.telequebec.tv/endirect/',
118 'info_dict': {
119 'id': 'endirect',
120 'ext': 'mp4',
121 'title': 're:^Télé-Québec - En direct [0-9]{4}-[0-9]{2}-[0-9]{2} [0-9]{2}:[0-9]{2}$',
122 'is_live': True,
123 },
124 'params': {
125 'skip_download': True,
126 },
127 }
128
129 def _real_extract(self, url):
130 video_id = self._match_id(url)
131
132 m3u8_url = None
133 webpage = self._download_webpage(
134 'https://player.telequebec.tv/Tq_VideoPlayer.js', video_id,
135 fatal=False)
136 if webpage:
137 m3u8_url = self._search_regex(
138 r'm3U8Url\s*:\s*(["\'])(?P<url>(?:(?!\1).)+)\1', webpage,
139 'm3u8 url', default=None, group='url')
140 if not m3u8_url:
141 m3u8_url = 'https://teleqmmd.mmdlive.lldns.net/teleqmmd/f386e3b206814e1f8c8c1c71c0f8e748/manifest.m3u8'
142 formats = self._extract_m3u8_formats(
143 m3u8_url, video_id, 'mp4', m3u8_id='hls')
144 self._sort_formats(formats)
145
146 return {
147 'id': video_id,
148 'title': self._live_title('Télé-Québec - En direct'),
149 'is_live': True,
150 'formats': formats,
151 }