_ Git - youtube-dl/blob - youtube_dl/extractor/tele13.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import js_to_json
   8
   9
  10 class Tele13IE(InfoExtractor):
  11     _VALID_URL = r'^http://(?:www\.)?t13\.cl/videos(?:/[^/]+)+/(?P<id>[\w-]+)'
  12     _TESTS = [
  13         {
  14             'url': 'http://www.t13.cl/videos/actualidad/el-circulo-de-hierro-de-michelle-bachelet-en-su-regreso-a-la-moneda',
  15             'md5': '4cb1fa38adcad8fea88487a078831755',
  16             'info_dict': {
  17                 'id': 'el-circulo-de-hierro-de-michelle-bachelet-en-su-regreso-a-la-moneda',
  18                 'ext': 'mp4',
  19                 'title': 'El círculo de hierro de Michelle Bachelet en su regreso a La Moneda',
  20             },
  21             'params': {
  22                 # HTTP Error 404: Not Found
  23                 'skip_download': True,
  24             },
  25         },
  26         {
  27             'url': 'http://www.t13.cl/videos/mundo/tendencias/video-captan-misteriosa-bola-fuego-cielos-bangkok',
  28             'md5': '65d1ae54812c96f4b345dd21d3bb1adc',
  29             'info_dict': {
  30                 'id': 'rOoKv2OMpOw',
  31                 'ext': 'mp4',
  32                 'title': 'Shooting star seen on 7-Sep-2015',
  33                 'description': 'md5:a1cd2e74f6ee6851552c9cf5851d6b06',
  34                 'uploader': 'Porjai Jaturongkhakun',
  35                 'upload_date': '20150906',
  36                 'uploader_id': 'UCnLY_3ezwNcDSC_Wc6suZxw',
  37             },
  38             'add_ie': ['Youtube'],
  39         }
  40     ]
  41
  42     def _real_extract(self, url):
  43         display_id = self._match_id(url)
  44
  45         webpage = self._download_webpage(url, display_id)
  46
  47         setup_js = self._parse_json(
  48             js_to_json(
  49                 self._search_regex(
  50                     r"jwplayer\('player-vivo'\).setup\((\{.*?\})\)",
  51                     webpage,
  52                     'setup code',
  53                     flags=re.DOTALL
  54                 ).replace('\n//', '')
  55             ),
  56             display_id
  57         )
  58         title = setup_js['title']
  59         thumbnail = setup_js.get('image') or setup_js['playlist'][0].get('image')
  60         description = self._html_search_meta(
  61             'description', webpage, 'description')
  62
  63         formats = []
  64         for f in setup_js['playlist'][0]['sources']:
  65             format_url = f['file']
  66             if format_url != '':
  67                 if '.m3u8' in format_url:
  68                     formats.extend(self._extract_m3u8_formats(format_url, display_id))
  69                 else:
  70                     if 'youtube.com' in format_url:
  71                         return self.url_result(format_url, 'Youtube')
  72                     else:
  73                         formats.append({'url': format_url, 'format_id': f.get('label')})
  74
  75         return {
  76             'id': display_id,
  77             'title': title,
  78             'description': description,
  79             'thumbnail': thumbnail,
  80             'formats': formats,
  81         }