aboutsummaryrefslogtreecommitdiff
path: root/youtube_dl/extractor/vrv.py
blob: 921e9e172496975a2b7ff78d7f08871b6c27f8f7 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
# coding: utf-8
from __future__ import unicode_literals

import base64
import json
import hashlib
import hmac
import random
import string
import time

from .common import InfoExtractor
from ..compat import (
    compat_urllib_parse_urlencode,
    compat_urllib_parse,
)
from ..utils import (
    float_or_none,
    int_or_none,
)


class VRVBaseIE(InfoExtractor):
    _API_DOMAIN = None
    _API_PARAMS = {}
    _CMS_SIGNING = {}

    def _call_api(self, path, video_id, note, data=None):
        base_url = self._API_DOMAIN + '/core/' + path
        encoded_query = compat_urllib_parse_urlencode({
            'oauth_consumer_key': self._API_PARAMS['oAuthKey'],
            'oauth_nonce': ''.join([random.choice(string.ascii_letters) for _ in range(32)]),
            'oauth_signature_method': 'HMAC-SHA1',
            'oauth_timestamp': int(time.time()),
            'oauth_version': '1.0',
        })
        headers = self.geo_verification_headers()
        if data:
            data = json.dumps(data).encode()
            headers['Content-Type'] = 'application/json'
        method = 'POST' if data else 'GET'
        base_string = '&'.join([method, compat_urllib_parse.quote(base_url, ''), compat_urllib_parse.quote(encoded_query, '')])
        oauth_signature = base64.b64encode(hmac.new(
            (self._API_PARAMS['oAuthSecret'] + '&').encode('ascii'),
            base_string.encode(), hashlib.sha1).digest()).decode()
        encoded_query += '&oauth_signature=' + compat_urllib_parse.quote(oauth_signature, '')
        return self._download_json(
            '?'.join([base_url, encoded_query]), video_id,
            note='Downloading %s JSON metadata' % note, headers=headers, data=data)

    def _call_cms(self, path, video_id, note):
        if not self._CMS_SIGNING:
            self._CMS_SIGNING = self._call_api('index', video_id, 'CMS Signing')['cms_signing']
        return self._download_json(
            self._API_DOMAIN + path, video_id, query=self._CMS_SIGNING,
            note='Downloading %s JSON metadata' % note, headers=self.geo_verification_headers())

    def _set_api_params(self, webpage, video_id):
        if not self._API_PARAMS:
            self._API_PARAMS = self._parse_json(self._search_regex(
                r'window\.__APP_CONFIG__\s*=\s*({.+?})</script>',
                webpage, 'api config'), video_id)['cxApiParams']
            self._API_DOMAIN = self._API_PARAMS.get('apiDomain', 'https://api.vrv.co')

    def _get_cms_resource(self, resource_key, video_id):
        return self._call_api(
            'cms_resource', video_id, 'resource path', data={
                'resource_key': resource_key,
            })['__links__']['cms_resource']['href']


class VRVIE(VRVBaseIE):
    IE_NAME = 'vrv'
    _VALID_URL = r'https?://(?:www\.)?vrv\.co/watch/(?P<id>[A-Z0-9]+)'
    _TESTS = [{
        'url': 'https://vrv.co/watch/GR9PNZ396/Hidden-America-with-Jonah-Ray:BOSTON-WHERE-THE-PAST-IS-THE-PRESENT',
        'info_dict': {
            'id': 'GR9PNZ396',
            'ext': 'mp4',
            'title': 'BOSTON: WHERE THE PAST IS THE PRESENT',
            'description': 'md5:4ec8844ac262ca2df9e67c0983c6b83f',
            'uploader_id': 'seeso',
        },
        'params': {
            # m3u8 download
            'skip_download': True,
        },
    }]

    def _extract_vrv_formats(self, url, video_id, stream_format, audio_lang, hardsub_lang):
        if not url or stream_format not in ('hls', 'dash'):
            return []
        stream_id = hardsub_lang or audio_lang
        format_id = '%s-%s' % (stream_format, stream_id)
        if stream_format == 'hls':
            adaptive_formats = self._extract_m3u8_formats(
                url, video_id, 'mp4', m3u8_id=format_id,
                note='Downloading %s m3u8 information' % stream_id,
                fatal=False)
        elif stream_format == 'dash':
            adaptive_formats = self._extract_mpd_formats(
                url, video_id, mpd_id=format_id,
                note='Downloading %s MPD information' % stream_id,
                fatal=False)
        if audio_lang:
            for f in adaptive_formats:
                if f.get('acodec') != 'none':
                    f['language'] = audio_lang
        return adaptive_formats

    def _real_extract(self, url):
        video_id = self._match_id(url)
        webpage = self._download_webpage(
            url, video_id,
            headers=self.geo_verification_headers())
        media_resource = self._parse_json(self._search_regex(
            r'window\.__INITIAL_STATE__\s*=\s*({.+?})</script>',
            webpage, 'inital state'), video_id).get('watch', {}).get('mediaResource') or {}

        video_data = media_resource.get('json')
        if not video_data:
            self._set_api_params(webpage, video_id)
            episode_path = self._get_cms_resource(
                'cms:/episodes/' + video_id, video_id)
            video_data = self._call_cms(episode_path, video_id, 'video')
        title = video_data['title']

        streams_json = media_resource.get('streams', {}).get('json', {})
        if not streams_json:
            self._set_api_params(webpage, video_id)
            streams_path = video_data['__links__']['streams']['href']
            streams_json = self._call_cms(streams_path, video_id, 'streams')

        audio_locale = streams_json.get('audio_locale')
        formats = []
        for stream_type, streams in streams_json.get('streams', {}).items():
            if stream_type in ('adaptive_hls', 'adaptive_dash'):
                for stream in streams.values():
                    formats.extend(self._extract_vrv_formats(
                        stream.get('url'), video_id, stream_type.split('_')[1],
                        audio_locale, stream.get('hardsub_locale')))
        self._sort_formats(formats)

        subtitles = {}
        for subtitle in streams_json.get('subtitles', {}).values():
            subtitle_url = subtitle.get('url')
            if not subtitle_url:
                continue
            subtitles.setdefault(subtitle.get('locale', 'en-US'), []).append({
                'url': subtitle_url,
                'ext': subtitle.get('format', 'ass'),
            })

        thumbnails = []
        for thumbnail in video_data.get('images', {}).get('thumbnails', []):
            thumbnail_url = thumbnail.get('source')
            if not thumbnail_url:
                continue
            thumbnails.append({
                'url': thumbnail_url,
                'width': int_or_none(thumbnail.get('width')),
                'height': int_or_none(thumbnail.get('height')),
            })

        return {
            'id': video_id,
            'title': title,
            'formats': formats,
            'subtitles': subtitles,
            'thumbnails': thumbnails,
            'description': video_data.get('description'),
            'duration': float_or_none(video_data.get('duration_ms'), 1000),
            'uploader_id': video_data.get('channel_id'),
            'series': video_data.get('series_title'),
            'season': video_data.get('season_title'),
            'season_number': int_or_none(video_data.get('season_number')),
            'season_id': video_data.get('season_id'),
            'episode': title,
            'episode_number': int_or_none(video_data.get('episode_number')),
            'episode_id': video_data.get('production_episode_id'),
        }


class VRVSeriesIE(VRVBaseIE):
    IE_NAME = 'vrv:series'
    _VALID_URL = r'https?://(?:www\.)?vrv\.co/series/(?P<id>[A-Z0-9]+)'
    _TEST = {
        'url': 'https://vrv.co/series/G68VXG3G6/The-Perfect-Insider',
        'info_dict': {
            'id': 'G68VXG3G6',
        },
        'playlist_mincount': 11,
    }

    def _real_extract(self, url):
        series_id = self._match_id(url)
        webpage = self._download_webpage(
            url, series_id,
            headers=self.geo_verification_headers())

        self._set_api_params(webpage, series_id)
        seasons_path = self._get_cms_resource(
            'cms:/seasons?series_id=' + series_id, series_id)
        seasons_data = self._call_cms(seasons_path, series_id, 'seasons')

        entries = []
        for season in seasons_data.get('items', []):
            episodes_path = season['__links__']['season/episodes']['href']
            episodes = self._call_cms(episodes_path, series_id, 'episodes')
            for episode in episodes.get('items', []):
                episode_id = episode['id']
                entries.append(self.url_result(
                    'https://vrv.co/watch/' + episode_id,
                    'VRV', episode_id, episode.get('title')))

        return self.playlist_result(entries, series_id)