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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..utils import remove_end
   6
   7
   8 class BioBioChileTVIE(InfoExtractor):
   9     _VALID_URL = r'https?://tv\.biobiochile\.cl/notas/(?:[^/]+/)+(?P<id>[^/]+)\.shtml'
  10
  11     _TESTS = [{
  12         'url': 'http://tv.biobiochile.cl/notas/2015/10/21/sobre-camaras-y-camarillas-parlamentarias.shtml',
  13         'md5': '26f51f03cf580265defefb4518faec09',
  14         'info_dict': {
  15             'id': 'sobre-camaras-y-camarillas-parlamentarias',
  16             'ext': 'mp4',
  17             'title': 'Sobre Cámaras y camarillas parlamentarias',
  18             'thumbnail': 're:^https?://.*\.jpg$',
  19             'uploader': 'Fernando Atria',
  20         },
  21     }, {
  22         # different uploader layout
  23         'url': 'http://tv.biobiochile.cl/notas/2016/03/18/natalia-valdebenito-repasa-a-diputado-hasbun-paso-a-la-categoria-de-hablar-brutalidades.shtml',
  24         'md5': 'edc2e6b58974c46d5b047dea3c539ff3',
  25         'info_dict': {
  26             'id': 'natalia-valdebenito-repasa-a-diputado-hasbun-paso-a-la-categoria-de-hablar-brutalidades',
  27             'ext': 'mp4',
  28             'title': 'Natalia Valdebenito repasa a diputado Hasbún: Pasó a la categoría de hablar brutalidades',
  29             'thumbnail': 're:^https?://.*\.jpg$',
  30             'uploader': 'Piangella Obrador',
  31         },
  32         'params': {
  33             'skip_download': True,
  34         },
  35     }, {
  36         'url': 'http://tv.biobiochile.cl/notas/2015/10/22/ninos-transexuales-de-quien-es-la-decision.shtml',
  37         'only_matching': True,
  38     }, {
  39         'url': 'http://tv.biobiochile.cl/notas/2015/10/21/exclusivo-hector-pinto-formador-de-chupete-revela-version-del-ex-delantero-albo.shtml',
  40         'only_matching': True,
  41     }]
  42
  43     def _real_extract(self, url):
  44         video_id = self._match_id(url)
  45
  46         webpage = self._download_webpage(url, video_id)
  47
  48         title = remove_end(self._og_search_title(webpage), ' - BioBioChile TV')
  49
  50         file_url = self._search_regex(
  51             r'loadFWPlayerVideo\([^,]+,\s*(["\'])(?P<url>.+?)\1',
  52             webpage, 'file url', group='url')
  53
  54         base_url = self._search_regex(
  55             r'file\s*:\s*(["\'])(?P<url>.+?)\1\s*\+\s*fileURL', webpage,
  56             'base url', default='http://unlimited2-cl.digitalproserver.com/bbtv/',
  57             group='url')
  58
  59         formats = self._extract_m3u8_formats(
  60             '%s%s/playlist.m3u8' % (base_url, file_url), video_id, 'mp4',
  61             entry_protocol='m3u8_native', m3u8_id='hls', fatal=False)
  62         f = {
  63             'url': '%s%s' % (base_url, file_url),
  64             'format_id': 'http',
  65             'protocol': 'http',
  66             'preference': 1,
  67         }
  68         if formats:
  69             f_copy = formats[-1].copy()
  70             f_copy.update(f)
  71             f = f_copy
  72         formats.append(f)
  73         self._sort_formats(formats)
  74
  75         thumbnail = self._og_search_thumbnail(webpage)
  76         uploader = self._html_search_regex(
  77             r'<a[^>]+href=["\']https?://busca\.biobiochile\.cl/author[^>]+>(.+?)</a>',
  78             webpage, 'uploader', fatal=False)
  79
  80         return {
  81             'id': video_id,
  82             'title': title,
  83             'thumbnail': thumbnail,
  84             'uploader': uploader,
  85             'formats': formats,
  86         }