]>
Commit | Line | Data |
---|---|---|
f47754f0 S |
1 | # coding: utf-8 |
2 | from __future__ import unicode_literals | |
3 | ||
db2058f6 RA |
4 | import base64 |
5 | import hashlib | |
6 | import json | |
f47754f0 | 7 | import random |
26394d02 | 8 | import re |
f47754f0 S |
9 | |
10 | from .common import InfoExtractor | |
e5187493 | 11 | from ..compat import ( |
e5187493 RA |
12 | compat_str, |
13 | ) | |
8c25f81b | 14 | from ..utils import ( |
63c3ee4f | 15 | HEADRequest, |
8c25f81b PH |
16 | ExtractorError, |
17 | float_or_none, | |
26394d02 | 18 | orderedSet, |
e7d34c03 | 19 | str_or_none, |
b89378a6 | 20 | try_get, |
8c25f81b | 21 | ) |
f47754f0 S |
22 | |
23 | ||
24 | class GloboIE(InfoExtractor): | |
25042f73 | 25 | _VALID_URL = r'(?:globo:|https?://.+?\.globo\.com/(?:[^/]+/)*(?:v/(?:[^/]+/)?|videos/))(?P<id>\d{7,})' |
d81ffc3a | 26 | _NETRC_MACHINE = 'globo' |
ad607563 | 27 | _TESTS = [{ |
ad607563 | 28 | 'url': 'http://g1.globo.com/carros/autoesporte/videos/t/exclusivos-do-g1/v/mercedes-benz-gla-passa-por-teste-de-colisao-na-europa/3607726/', |
ad607563 S |
29 | 'info_dict': { |
30 | 'id': '3607726', | |
31 | 'ext': 'mp4', | |
32 | 'title': 'Mercedes-Benz GLA passa por teste de colisão na Europa', | |
33 | 'duration': 103.204, | |
b89378a6 AG |
34 | 'uploader': 'G1', |
35 | 'uploader_id': '2015', | |
36 | }, | |
37 | 'params': { | |
38 | 'skip_download': True, | |
264cd00f | 39 | }, |
ad607563 | 40 | }, { |
264cd00f | 41 | 'url': 'http://globoplay.globo.com/v/4581987/', |
ad607563 | 42 | 'info_dict': { |
264cd00f | 43 | 'id': '4581987', |
ad607563 | 44 | 'ext': 'mp4', |
264cd00f S |
45 | 'title': 'Acidentes de trânsito estão entre as maiores causas de queda de energia em SP', |
46 | 'duration': 137.973, | |
47 | 'uploader': 'Rede Globo', | |
e7d34c03 | 48 | 'uploader_id': '196', |
264cd00f | 49 | }, |
b89378a6 AG |
50 | 'params': { |
51 | 'skip_download': True, | |
52 | }, | |
264cd00f S |
53 | }, { |
54 | 'url': 'http://canalbrasil.globo.com/programas/sangue-latino/videos/3928201.html', | |
55 | 'only_matching': True, | |
56 | }, { | |
57 | 'url': 'http://globosatplay.globo.com/globonews/v/4472924/', | |
58 | 'only_matching': True, | |
59 | }, { | |
60 | 'url': 'http://globotv.globo.com/t/programa/v/clipe-sexo-e-as-negas-adeus/3836166/', | |
61 | 'only_matching': True, | |
62 | }, { | |
63 | 'url': 'http://globotv.globo.com/canal-brasil/sangue-latino/t/todos-os-videos/v/ator-e-diretor-argentino-ricado-darin-fala-sobre-utopias-e-suas-perdas/3928201/', | |
64 | 'only_matching': True, | |
5d501a09 S |
65 | }, { |
66 | 'url': 'http://canaloff.globo.com/programas/desejar-profundo/videos/4518560.html', | |
67 | 'only_matching': True, | |
26394d02 S |
68 | }, { |
69 | 'url': 'globo:3607726', | |
70 | 'only_matching': True, | |
63c3ee4f B |
71 | }, { |
72 | 'url': 'https://globoplay.globo.com/v/10248083/', | |
73 | 'info_dict': { | |
74 | 'id': '10248083', | |
75 | 'ext': 'mp4', | |
76 | 'title': 'Melhores momentos: Equador 1 x 1 Brasil pelas Eliminatórias da Copa do Mundo 2022', | |
77 | 'duration': 530.964, | |
78 | 'uploader': 'SporTV', | |
79 | 'uploader_id': '698', | |
80 | }, | |
81 | 'params': { | |
82 | 'skip_download': True, | |
83 | }, | |
ad607563 | 84 | }] |
f47754f0 | 85 | |
f47754f0 S |
86 | def _real_extract(self, url): |
87 | video_id = self._match_id(url) | |
88 | ||
63c3ee4f B |
89 | self._request_webpage( |
90 | HEADRequest('https://globo-ab.globo.com/v2/selected-alternatives?experiments=player-isolated-experiment-02&skipImpressions=true'), | |
91 | video_id, 'Getting cookies') | |
92 | ||
f47754f0 | 93 | video = self._download_json( |
db2058f6 RA |
94 | 'http://api.globovideos.com/videos/%s/playlist' % video_id, |
95 | video_id)['videos'][0] | |
a06916d9 | 96 | if not self.get_param('allow_unplayable_formats') and video.get('encrypted') is True: |
88acdbc2 | 97 | self.report_drm(video_id) |
f47754f0 S |
98 | |
99 | title = video['title'] | |
f47754f0 S |
100 | |
101 | formats = [] | |
b89378a6 | 102 | security = self._download_json( |
63c3ee4f | 103 | 'https://playback.video.globo.com/v2/video-session', video_id, 'Downloading security hash for %s' % video_id, |
b89378a6 AG |
104 | headers={'content-type': 'application/json'}, data=json.dumps({ |
105 | "player_type": "desktop", | |
106 | "video_id": video_id, | |
107 | "quality": "max", | |
108 | "content_protection": "widevine", | |
109 | "vsid": "581b986b-4c40-71f0-5a58-803e579d5fa2", | |
110 | "tz": "-3.0:00" | |
111 | }).encode()) | |
112 | ||
63c3ee4f B |
113 | self._request_webpage(HEADRequest(security['sources'][0]['url_template']), video_id, 'Getting locksession cookie') |
114 | ||
115 | security_hash = security['sources'][0]['token'] | |
b89378a6 AG |
116 | if not security_hash: |
117 | message = security.get('message') | |
118 | if message: | |
119 | raise ExtractorError( | |
120 | '%s returned error: %s' % (self.IE_NAME, message), expected=True) | |
121 | ||
122 | hash_code = security_hash[:2] | |
123 | padding = '%010d' % random.randint(1, 10000000000) | |
124 | if hash_code in ('04', '14'): | |
125 | received_time = security_hash[3:13] | |
126 | received_md5 = security_hash[24:] | |
127 | hash_prefix = security_hash[:23] | |
128 | elif hash_code in ('02', '12', '03', '13'): | |
129 | received_time = security_hash[2:12] | |
130 | received_md5 = security_hash[22:] | |
131 | padding += '1' | |
132 | hash_prefix = '05' + security_hash[:22] | |
133 | ||
134 | padded_sign_time = compat_str(int(received_time) + 86400) + padding | |
135 | md5_data = (received_md5 + padded_sign_time + '0xAC10FD').encode() | |
136 | signed_md5 = base64.urlsafe_b64encode(hashlib.md5(md5_data).digest()).decode().strip('=') | |
137 | signed_hash = hash_prefix + padded_sign_time + signed_md5 | |
63c3ee4f | 138 | source = security['sources'][0]['url_parts'] |
b89378a6 AG |
139 | resource_url = source['scheme'] + '://' + source['domain'] + source['path'] |
140 | signed_url = '%s?h=%s&k=html5&a=%s' % (resource_url, signed_hash, 'F' if video.get('subscriber_only') else 'A') | |
141 | ||
3f047fc4 F |
142 | fmts, subtitles = self._extract_m3u8_formats_and_subtitles( |
143 | signed_url, video_id, 'mp4', entry_protocol='m3u8_native', m3u8_id='hls', fatal=False) | |
144 | formats.extend(fmts) | |
b89378a6 AG |
145 | self._sort_formats(formats) |
146 | ||
f47754f0 | 147 | for resource in video['resources']: |
b89378a6 | 148 | if resource.get('type') == 'subtitle': |
30eb05cb | 149 | subtitles.setdefault(resource.get('language') or 'por', []).append({ |
b89378a6 | 150 | 'url': resource.get('url'), |
30eb05cb | 151 | }) |
b89378a6 AG |
152 | subs = try_get(security, lambda x: x['source']['subtitles'], expected_type=dict) or {} |
153 | for sub_lang, sub_url in subs.items(): | |
154 | if sub_url: | |
155 | subtitles.setdefault(sub_lang or 'por', []).append({ | |
156 | 'url': sub_url, | |
db2058f6 | 157 | }) |
b89378a6 AG |
158 | subs = try_get(security, lambda x: x['source']['subtitles_webvtt'], expected_type=dict) or {} |
159 | for sub_lang, sub_url in subs.items(): | |
160 | if sub_url: | |
161 | subtitles.setdefault(sub_lang or 'por', []).append({ | |
162 | 'url': sub_url, | |
8c72beb2 | 163 | }) |
f47754f0 | 164 | |
fffccaaf | 165 | duration = float_or_none(video.get('duration'), 1000) |
fffccaaf | 166 | uploader = video.get('channel') |
e7d34c03 | 167 | uploader_id = str_or_none(video.get('channel_id')) |
fffccaaf | 168 | |
f47754f0 S |
169 | return { |
170 | 'id': video_id, | |
171 | 'title': title, | |
172 | 'duration': duration, | |
173 | 'uploader': uploader, | |
174 | 'uploader_id': uploader_id, | |
30eb05cb RA |
175 | 'formats': formats, |
176 | 'subtitles': subtitles, | |
5f6a1245 | 177 | } |
ad607563 S |
178 | |
179 | ||
180 | class GloboArticleIE(InfoExtractor): | |
26394d02 | 181 | _VALID_URL = r'https?://.+?\.globo\.com/(?:[^/]+/)*(?P<id>[^/.]+)(?:\.html)?' |
ad607563 S |
182 | |
183 | _VIDEOID_REGEXES = [ | |
184 | r'\bdata-video-id=["\'](\d{7,})', | |
185 | r'\bdata-player-videosids=["\'](\d{7,})', | |
9e5751b9 | 186 | r'\bvideosIDs\s*:\s*["\']?(\d{7,})', |
ad607563 S |
187 | r'\bdata-id=["\'](\d{7,})', |
188 | r'<div[^>]+\bid=["\'](\d{7,})', | |
17b18388 | 189 | r'<bs-player[^>]+\bvideoid=["\'](\d{8,})', |
ad607563 S |
190 | ] |
191 | ||
5d501a09 | 192 | _TESTS = [{ |
ad607563 | 193 | 'url': 'http://g1.globo.com/jornal-nacional/noticia/2014/09/novidade-na-fiscalizacao-de-bagagem-pela-receita-provoca-discussoes.html', |
ad607563 | 194 | 'info_dict': { |
26394d02 S |
195 | 'id': 'novidade-na-fiscalizacao-de-bagagem-pela-receita-provoca-discussoes', |
196 | 'title': 'Novidade na fiscalização de bagagem pela Receita provoca discussões', | |
197 | 'description': 'md5:c3c4b4d4c30c32fce460040b1ac46b12', | |
198 | }, | |
199 | 'playlist_count': 1, | |
200 | }, { | |
201 | 'url': 'http://g1.globo.com/pr/parana/noticia/2016/09/mpf-denuncia-lula-marisa-e-mais-seis-na-operacao-lava-jato.html', | |
202 | 'info_dict': { | |
203 | 'id': 'mpf-denuncia-lula-marisa-e-mais-seis-na-operacao-lava-jato', | |
204 | 'title': "Lula era o 'comandante máximo' do esquema da Lava Jato, diz MPF", | |
205 | 'description': 'md5:8aa7cc8beda4dc71cc8553e00b77c54c', | |
206 | }, | |
207 | 'playlist_count': 6, | |
5d501a09 S |
208 | }, { |
209 | 'url': 'http://gq.globo.com/Prazeres/Poder/noticia/2015/10/all-o-desafio-assista-ao-segundo-capitulo-da-serie.html', | |
210 | 'only_matching': True, | |
211 | }, { | |
212 | 'url': 'http://gshow.globo.com/programas/tv-xuxa/O-Programa/noticia/2014/01/xuxa-e-junno-namoram-muuuito-em-luau-de-zeze-di-camargo-e-luciano.html', | |
213 | 'only_matching': True, | |
9e5751b9 S |
214 | }, { |
215 | 'url': 'http://oglobo.globo.com/rio/a-amizade-entre-um-entregador-de-farmacia-um-piano-19946271', | |
216 | 'only_matching': True, | |
17b18388 B |
217 | }, { |
218 | 'url': 'https://ge.globo.com/video/ta-na-area-como-foi-assistir-ao-jogo-do-palmeiras-que-a-globo-nao-passou-10287094.ghtml', | |
219 | 'info_dict': { | |
220 | 'id': 'ta-na-area-como-foi-assistir-ao-jogo-do-palmeiras-que-a-globo-nao-passou-10287094', | |
221 | 'title': 'Tá na Área: como foi assistir ao jogo do Palmeiras que a Globo não passou', | |
222 | 'description': 'md5:2d089d036c4c9675117d3a56f8c61739', | |
223 | }, | |
224 | 'playlist_count': 1, | |
5d501a09 | 225 | }] |
ad607563 S |
226 | |
227 | @classmethod | |
228 | def suitable(cls, url): | |
229 | return False if GloboIE.suitable(url) else super(GloboArticleIE, cls).suitable(url) | |
230 | ||
231 | def _real_extract(self, url): | |
232 | display_id = self._match_id(url) | |
233 | webpage = self._download_webpage(url, display_id) | |
26394d02 S |
234 | video_ids = [] |
235 | for video_regex in self._VIDEOID_REGEXES: | |
236 | video_ids.extend(re.findall(video_regex, webpage)) | |
237 | entries = [ | |
238 | self.url_result('globo:%s' % video_id, GloboIE.ie_key()) | |
239 | for video_id in orderedSet(video_ids)] | |
17b18388 | 240 | title = self._og_search_title(webpage) |
26394d02 S |
241 | description = self._html_search_meta('description', webpage) |
242 | return self.playlist_result(entries, display_id, title, description) |