Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/rtlnl.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import (
   6     int_or_none,
   7     parse_duration,
   8 )
   9
  10
  11 class RtlNlIE(InfoExtractor):
  12     IE_NAME = 'rtl.nl'
  13     IE_DESC = 'rtl.nl and rtlxl.nl'
  14     _VALID_URL = r'''(?x)
  15         https?://(?:(?:www|static)\.)?
  16         (?:
  17             rtlxl\.nl/[^\#]*\#!/[^/]+/|
  18             rtl\.nl/(?:(?:system/videoplayer/(?:[^/]+/)+(?:video_)?embed\.html|embed)\b.+?\buuid=|video/)
  19         )
  20         (?P<id>[0-9a-f-]+)'''
  21
  22     _TESTS = [{
  23         'url': 'http://www.rtlxl.nl/#!/rtl-nieuws-132237/82b1aad1-4a14-3d7b-b554-b0aed1b2c416',
  24         'md5': '473d1946c1fdd050b2c0161a4b13c373',
  25         'info_dict': {
  26             'id': '82b1aad1-4a14-3d7b-b554-b0aed1b2c416',
  27             'ext': 'mp4',
  28             'title': 'RTL Nieuws',
  29             'description': 'md5:d41d8cd98f00b204e9800998ecf8427e',
  30             'timestamp': 1461951000,
  31             'upload_date': '20160429',
  32             'duration': 1167.96,
  33         },
  34     }, {
  35         # best format avaialble a3t
  36         'url': 'http://www.rtl.nl/system/videoplayer/derden/rtlnieuws/video_embed.html#uuid=84ae5571-ac25-4225-ae0c-ef8d9efb2aed/autoplay=false',
  37         'md5': 'dea7474214af1271d91ef332fb8be7ea',
  38         'info_dict': {
  39             'id': '84ae5571-ac25-4225-ae0c-ef8d9efb2aed',
  40             'ext': 'mp4',
  41             'timestamp': 1424039400,
  42             'title': 'RTL Nieuws - Nieuwe beelden Kopenhagen: chaos direct na aanslag',
  43             'thumbnail': r're:^https?://screenshots\.rtl\.nl/(?:[^/]+/)*sz=[0-9]+x[0-9]+/uuid=84ae5571-ac25-4225-ae0c-ef8d9efb2aed$',
  44             'upload_date': '20150215',
  45             'description': 'Er zijn nieuwe beelden vrijgegeven die vlak na de aanslag in Kopenhagen zijn gemaakt. Op de video is goed te zien hoe omstanders zich bekommeren om één van de slachtoffers, terwijl de eerste agenten ter plaatse komen.',
  46         }
  47     }, {
  48         # empty synopsis and missing episodes (see https://github.com/ytdl-org/youtube-dl/issues/6275)
  49         # best format available nettv
  50         'url': 'http://www.rtl.nl/system/videoplayer/derden/rtlnieuws/video_embed.html#uuid=f536aac0-1dc3-4314-920e-3bd1c5b3811a/autoplay=false',
  51         'info_dict': {
  52             'id': 'f536aac0-1dc3-4314-920e-3bd1c5b3811a',
  53             'ext': 'mp4',
  54             'title': 'RTL Nieuws - Meer beelden van overval juwelier',
  55             'thumbnail': r're:^https?://screenshots\.rtl\.nl/(?:[^/]+/)*sz=[0-9]+x[0-9]+/uuid=f536aac0-1dc3-4314-920e-3bd1c5b3811a$',
  56             'timestamp': 1437233400,
  57             'upload_date': '20150718',
  58             'duration': 30.474,
  59         },
  60         'params': {
  61             'skip_download': True,
  62         },
  63     }, {
  64         # encrypted m3u8 streams, georestricted
  65         'url': 'http://www.rtlxl.nl/#!/afl-2-257632/52a74543-c504-4cde-8aa8-ec66fe8d68a7',
  66         'only_matching': True,
  67     }, {
  68         'url': 'http://www.rtl.nl/system/videoplayer/derden/embed.html#!/uuid=bb0353b0-d6a4-1dad-90e9-18fe75b8d1f0',
  69         'only_matching': True,
  70     }, {
  71         'url': 'http://rtlxl.nl/?_ga=1.204735956.572365465.1466978370#!/rtl-nieuws-132237/3c487912-023b-49ac-903e-2c5d79f8410f',
  72         'only_matching': True,
  73     }, {
  74         'url': 'https://www.rtl.nl/video/c603c9c2-601d-4b5e-8175-64f1e942dc7d/',
  75         'only_matching': True,
  76     }, {
  77         'url': 'https://static.rtl.nl/embed/?uuid=1a2970fc-5c0b-43ff-9fdc-927e39e6d1bc&autoplay=false&publicatiepunt=rtlnieuwsnl',
  78         'only_matching': True,
  79     }]
  80
  81     def _real_extract(self, url):
  82         uuid = self._match_id(url)
  83         info = self._download_json(
  84             'http://www.rtl.nl/system/s4m/vfd/version=2/uuid=%s/fmt=adaptive/' % uuid,
  85             uuid)
  86
  87         material = info['material'][0]
  88         title = info['abstracts'][0]['name']
  89         subtitle = material.get('title')
  90         if subtitle:
  91             title += ' - %s' % subtitle
  92         description = material.get('synopsis')
  93
  94         meta = info.get('meta', {})
  95
  96         videopath = material['videopath']
  97         m3u8_url = meta.get('videohost', 'http://manifest.us.rtl.nl') + videopath
  98
  99         formats = self._extract_m3u8_formats(
 100             m3u8_url, uuid, 'mp4', m3u8_id='hls', fatal=False)
 101         self._sort_formats(formats)
 102
 103         thumbnails = []
 104
 105         for p in ('poster_base_url', '"thumb_base_url"'):
 106             if not meta.get(p):
 107                 continue
 108
 109             thumbnails.append({
 110                 'url': self._proto_relative_url(meta[p] + uuid),
 111                 'width': int_or_none(self._search_regex(
 112                     r'/sz=([0-9]+)', meta[p], 'thumbnail width', fatal=False)),
 113                 'height': int_or_none(self._search_regex(
 114                     r'/sz=[0-9]+x([0-9]+)',
 115                     meta[p], 'thumbnail height', fatal=False))
 116             })
 117
 118         return {
 119             'id': uuid,
 120             'title': title,
 121             'formats': formats,
 122             'timestamp': material['original_date'],
 123             'description': description,
 124             'duration': parse_duration(material.get('duration')),
 125             'thumbnails': thumbnails,
 126         }