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 (
   6     ExtractorError,
   7     remove_end,
   8 )
   9
  10
  11 class BioBioChileTVIE(InfoExtractor):
  12     _VALID_URL = r'https?://(?:tv|www)\.biobiochile\.cl/(?:notas|noticias)/(?:[^/]+/)+(?P<id>[^/]+)\.shtml'
  13
  14     _TESTS = [{
  15         'url': 'http://tv.biobiochile.cl/notas/2015/10/21/sobre-camaras-y-camarillas-parlamentarias.shtml',
  16         'md5': '26f51f03cf580265defefb4518faec09',
  17         'info_dict': {
  18             'id': 'sobre-camaras-y-camarillas-parlamentarias',
  19             'ext': 'mp4',
  20             'title': 'Sobre Cámaras y camarillas parlamentarias',
  21             'thumbnail': r're:^https?://.*\.jpg$',
  22             'uploader': 'Fernando Atria',
  23         },
  24         'skip': 'URL expired and redirected to http://www.biobiochile.cl/portada/bbtv/index.html',
  25     }, {
  26         # different uploader layout
  27         'url': 'http://tv.biobiochile.cl/notas/2016/03/18/natalia-valdebenito-repasa-a-diputado-hasbun-paso-a-la-categoria-de-hablar-brutalidades.shtml',
  28         'md5': 'edc2e6b58974c46d5b047dea3c539ff3',
  29         'info_dict': {
  30             'id': 'natalia-valdebenito-repasa-a-diputado-hasbun-paso-a-la-categoria-de-hablar-brutalidades',
  31             'ext': 'mp4',
  32             'title': 'Natalia Valdebenito repasa a diputado Hasbún: Pasó a la categoría de hablar brutalidades',
  33             'thumbnail': r're:^https?://.*\.jpg$',
  34             'uploader': 'Piangella Obrador',
  35         },
  36         'params': {
  37             'skip_download': True,
  38         },
  39         'skip': 'URL expired and redirected to http://www.biobiochile.cl/portada/bbtv/index.html',
  40     }, {
  41         'url': 'http://www.biobiochile.cl/noticias/bbtv/comentarios-bio-bio/2016/07/08/edecanes-del-congreso-figuras-decorativas-que-le-cuestan-muy-caro-a-los-chilenos.shtml',
  42         'info_dict': {
  43             'id': 'b4xd0LK3SK',
  44             'ext': 'mp4',
  45             # TODO: fix url_transparent information overriding
  46             # 'uploader': 'Juan Pablo Echenique',
  47             'title': 'Comentario Oscar Cáceres',
  48         },
  49         'params': {
  50             # empty m3u8 manifest
  51             'skip_download': True,
  52         },
  53     }, {
  54         'url': 'http://tv.biobiochile.cl/notas/2015/10/22/ninos-transexuales-de-quien-es-la-decision.shtml',
  55         'only_matching': True,
  56     }, {
  57         'url': 'http://tv.biobiochile.cl/notas/2015/10/21/exclusivo-hector-pinto-formador-de-chupete-revela-version-del-ex-delantero-albo.shtml',
  58         'only_matching': True,
  59     }]
  60
  61     def _real_extract(self, url):
  62         video_id = self._match_id(url)
  63
  64         webpage = self._download_webpage(url, video_id)
  65
  66         rudo_url = self._search_regex(
  67             r'<iframe[^>]+src=(?P<q1>[\'"])(?P<url>(?:https?:)?//rudo\.video/vod/[0-9a-zA-Z]+)(?P=q1)',
  68             webpage, 'embed URL', None, group='url')
  69         if not rudo_url:
  70             raise ExtractorError('No videos found')
  71
  72         title = remove_end(self._og_search_title(webpage), ' - BioBioChile TV')
  73
  74         thumbnail = self._og_search_thumbnail(webpage)
  75         uploader = self._html_search_regex(
  76             r'<a[^>]+href=["\'](?:https?://(?:busca|www)\.biobiochile\.cl)?/(?:lista/)?(?:author|autor)[^>]+>(.+?)</a>',
  77             webpage, 'uploader', fatal=False)
  78
  79         return {
  80             '_type': 'url_transparent',
  81             'url': rudo_url,
  82             'id': video_id,
  83             'title': title,
  84             'thumbnail': thumbnail,
  85             'uploader': uploader,
  86         }