]> gitweb @ CieloNegro.org - youtube-dl.git/blob - youtube_dl/extractor/sixplay.py
[sixplay] handle videos with empty assets(closes #20016)
[youtube-dl.git] / youtube_dl / extractor / sixplay.py
1 # coding: utf-8
2 from __future__ import unicode_literals
3
4 import re
5
6 from .common import InfoExtractor
7 from ..compat import (
8     compat_parse_qs,
9     compat_str,
10     compat_urllib_parse_urlparse,
11 )
12 from ..utils import (
13     determine_ext,
14     int_or_none,
15     try_get,
16     qualities,
17 )
18
19
20 class SixPlayIE(InfoExtractor):
21     IE_NAME = '6play'
22     _VALID_URL = r'(?:6play:|https?://(?:www\.)?(?P<domain>6play\.fr|rtlplay\.be|play\.rtl\.hr)/.+?-c_)(?P<id>[0-9]+)'
23     _TESTS = [{
24         'url': 'https://www.6play.fr/minute-par-minute-p_9533/le-but-qui-a-marque-lhistoire-du-football-francais-c_12041051',
25         'md5': '31fcd112637baa0c2ab92c4fcd8baf27',
26         'info_dict': {
27             'id': '12041051',
28             'ext': 'mp4',
29             'title': 'Le but qui a marqué l\'histoire du football français !',
30             'description': 'md5:b59e7e841d646ef1eb42a7868eb6a851',
31         },
32     }, {
33         'url': 'https://www.rtlplay.be/rtl-info-13h-p_8551/les-titres-du-rtlinfo-13h-c_12045869',
34         'only_matching': True,
35     }, {
36         'url': 'https://play.rtl.hr/pj-masks-p_9455/epizoda-34-sezona-1-catboyevo-cudo-na-dva-kotaca-c_11984989',
37         'only_matching': True,
38     }]
39
40     def _real_extract(self, url):
41         domain, video_id = re.search(self._VALID_URL, url).groups()
42         service, consumer_name = {
43             '6play.fr': ('6play', 'm6web'),
44             'rtlplay.be': ('rtlbe_rtl_play', 'rtlbe'),
45             'play.rtl.hr': ('rtlhr_rtl_play', 'rtlhr'),
46         }.get(domain, ('6play', 'm6web'))
47
48         data = self._download_json(
49             'https://pc.middleware.6play.fr/6play/v2/platforms/m6group_web/services/%s/videos/clip_%s' % (service, video_id),
50             video_id, headers={
51                 'x-customer-name': consumer_name
52             }, query={
53                 'csa': 5,
54                 'with': 'clips',
55             })
56
57         clip_data = data['clips'][0]
58         title = clip_data['title']
59
60         urls = []
61         quality_key = qualities(['lq', 'sd', 'hq', 'hd'])
62         formats = []
63         subtitles = {}
64         assets = clip_data.get('assets') or []
65         for asset in assets:
66             asset_url = asset.get('full_physical_path')
67             protocol = asset.get('protocol')
68             if not asset_url or protocol == 'primetime' or asset.get('type') == 'usp_hlsfp_h264' or asset_url in urls:
69                 continue
70             urls.append(asset_url)
71             container = asset.get('video_container')
72             ext = determine_ext(asset_url)
73             if protocol == 'http_subtitle' or ext == 'vtt':
74                 subtitles.setdefault('fr', []).append({'url': asset_url})
75                 continue
76             if container == 'm3u8' or ext == 'm3u8':
77                 if protocol == 'usp':
78                     if compat_parse_qs(compat_urllib_parse_urlparse(asset_url).query).get('token', [None])[0]:
79                         urlh = self._request_webpage(
80                             asset_url, video_id, fatal=False,
81                             headers=self.geo_verification_headers())
82                         if not urlh:
83                             continue
84                         asset_url = urlh.geturl()
85                     for i in range(3, 0, -1):
86                         asset_url = asset_url = asset_url.replace('_sd1/', '_sd%d/' % i)
87                         m3u8_formats = self._extract_m3u8_formats(
88                             asset_url, video_id, 'mp4', 'm3u8_native',
89                             m3u8_id='hls', fatal=False)
90                         formats.extend(m3u8_formats)
91                         formats.extend(self._extract_mpd_formats(
92                             asset_url.replace('.m3u8', '.mpd'),
93                             video_id, mpd_id='dash', fatal=False))
94                         if m3u8_formats:
95                             break
96                 else:
97                     formats.extend(self._extract_m3u8_formats(
98                         asset_url, video_id, 'mp4', 'm3u8_native',
99                         m3u8_id='hls', fatal=False))
100             elif container == 'mp4' or ext == 'mp4':
101                 quality = asset.get('video_quality')
102                 formats.append({
103                     'url': asset_url,
104                     'format_id': quality,
105                     'quality': quality_key(quality),
106                     'ext': ext,
107                 })
108         self._sort_formats(formats)
109
110         def get(getter):
111             for src in (data, clip_data):
112                 v = try_get(src, getter, compat_str)
113                 if v:
114                     return v
115
116         return {
117             'id': video_id,
118             'title': title,
119             'description': get(lambda x: x['description']),
120             'duration': int_or_none(clip_data.get('duration')),
121             'series': get(lambda x: x['program']['title']),
122             'formats': formats,
123             'subtitles': subtitles,
124         }