]>
Commit | Line | Data |
---|---|---|
99afb3dd | 1 | from .common import InfoExtractor |
c5f51551 | 2 | from ..compat import compat_str |
86916dae | 3 | from ..utils import ( |
7c60c33e | 4 | ExtractorError, |
57ce8a6d | 5 | int_or_none, |
7c60c33e | 6 | try_get, |
7 | unified_strdate, | |
86916dae | 8 | ) |
99afb3dd JMF |
9 | |
10 | ||
11 | class WatIE(InfoExtractor): | |
c5f51551 | 12 | _VALID_URL = r'(?:wat:|https?://(?:www\.)?wat\.tv/video/.*-)(?P<id>[0-9a-z]+)' |
99afb3dd | 13 | IE_NAME = 'wat.tv' |
c28df247 S |
14 | _TESTS = [ |
15 | { | |
16 | 'url': 'http://www.wat.tv/video/soupe-figues-l-orange-aux-epices-6z1uz_2hvf7_.html', | |
c28df247 S |
17 | 'info_dict': { |
18 | 'id': '11713067', | |
c28df247 S |
19 | 'ext': 'mp4', |
20 | 'title': 'Soupe de figues à l\'orange et aux épices', | |
21 | 'description': 'Retrouvez l\'émission "Petits plats en équilibre", diffusée le 18 août 2014.', | |
22 | 'upload_date': '20140819', | |
23 | 'duration': 120, | |
24 | }, | |
0adf213d RA |
25 | 'params': { |
26 | # m3u8 download | |
27 | 'skip_download': True, | |
28 | }, | |
29 | 'expected_warnings': ['HTTP Error 404'], | |
7c60c33e | 30 | 'skip': 'This content is no longer available', |
c28df247 S |
31 | }, |
32 | { | |
33 | 'url': 'http://www.wat.tv/video/gregory-lemarchal-voix-ange-6z1v7_6ygkj_.html', | |
0adf213d | 34 | 'md5': 'b16574df2c3cd1a36ca0098f2a791925', |
c28df247 S |
35 | 'info_dict': { |
36 | 'id': '11713075', | |
c28df247 S |
37 | 'ext': 'mp4', |
38 | 'title': 'Grégory Lemarchal, une voix d\'ange depuis 10 ans (1/3)', | |
c28df247 | 39 | 'upload_date': '20140816', |
c28df247 | 40 | }, |
57ce8a6d | 41 | 'expected_warnings': ["Ce contenu n'est pas disponible pour l'instant."], |
7c60c33e | 42 | 'skip': 'This content is no longer available', |
fa800269 | 43 | }, |
c28df247 | 44 | ] |
7c60c33e | 45 | _GEO_BYPASS = False |
e7916255 | 46 | |
99afb3dd | 47 | def _real_extract(self, url): |
c5f51551 | 48 | video_id = self._match_id(url) |
49 | video_id = video_id if video_id.isdigit() and len(video_id) > 6 else compat_str(int(video_id, 36)) | |
8244288d | 50 | |
c5f51551 | 51 | # 'contentv4' is used in the website, but it also returns the related |
52 | # videos, we don't need them | |
7c60c33e | 53 | # video_data = self._download_json( |
54 | # 'http://www.wat.tv/interface/contentv4s/' + video_id, video_id) | |
57ce8a6d | 55 | video_data = self._download_json( |
7c60c33e | 56 | 'https://mediainfo.tf1.fr/mediainfocombo/' + video_id, |
25f0e68f | 57 | video_id, query={'context': 'MYTF1', 'pver': '4020003'}) |
57ce8a6d | 58 | video_info = video_data['media'] |
a54bda3a | 59 | |
86916dae S |
60 | error_desc = video_info.get('error_desc') |
61 | if error_desc: | |
7c60c33e | 62 | if video_info.get('error_code') == 'GEOBLOCKED': |
63 | self.raise_geo_restricted(error_desc, video_info.get('geoList')) | |
64 | raise ExtractorError(error_desc, expected=True) | |
99afb3dd | 65 | |
7c60c33e | 66 | title = video_info['title'] |
8244288d | 67 | |
7c60c33e | 68 | formats = [] |
ec4f374c | 69 | subtitles = {} |
c5f51551 | 70 | |
7c60c33e | 71 | def extract_formats(manifest_urls): |
72 | for f, f_url in manifest_urls.items(): | |
73 | if not f_url: | |
74 | continue | |
75 | if f in ('dash', 'mpd'): | |
ec4f374c | 76 | fmts, subs = self._extract_mpd_formats_and_subtitles( |
7c60c33e | 77 | f_url.replace('://das-q1.tf1.fr/', '://das-q1-ssl.tf1.fr/'), |
ec4f374c | 78 | video_id, mpd_id='dash', fatal=False) |
7c60c33e | 79 | elif f == 'hls': |
ec4f374c | 80 | fmts, subs = self._extract_m3u8_formats_and_subtitles( |
7c60c33e | 81 | f_url, video_id, 'mp4', |
ec4f374c F |
82 | 'm3u8_native', m3u8_id='hls', fatal=False) |
83 | else: | |
84 | continue | |
85 | formats.extend(fmts) | |
86 | self._merge_subtitles(subs, target=subtitles) | |
c5f51551 | 87 | |
7c60c33e | 88 | delivery = video_data.get('delivery') or {} |
89 | extract_formats({delivery.get('format'): delivery.get('url')}) | |
90 | if not formats: | |
91 | if delivery.get('drm'): | |
88acdbc2 | 92 | self.report_drm(video_id) |
7c60c33e | 93 | manifest_urls = self._download_json( |
94 | 'http://www.wat.tv/get/webhtml/' + video_id, video_id, fatal=False) | |
95 | if manifest_urls: | |
96 | extract_formats(manifest_urls) | |
948cd5b7 | 97 | |
e7916255 | 98 | return { |
c5f51551 | 99 | 'id': video_id, |
57ce8a6d | 100 | 'title': title, |
7c60c33e | 101 | 'thumbnail': video_info.get('preview'), |
102 | 'upload_date': unified_strdate(try_get( | |
103 | video_data, lambda x: x['mediametrie']['chapters'][0]['estatS4'])), | |
104 | 'duration': int_or_none(video_info.get('duration')), | |
a54bda3a | 105 | 'formats': formats, |
ec4f374c | 106 | 'subtitles': subtitles, |
e7916255 | 107 | } |