_ 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 c\u00edrculo de hierro de Michelle Bachelet en su regreso a La Moneda',
  20             }
  21         },
  22         {
  23             'url': 'http://www.t13.cl/videos/mundo/tendencias/video-captan-misteriosa-bola-fuego-cielos-bangkok',
  24             'md5': '65d1ae54812c96f4b345dd21d3bb1adc',
  25             'info_dict': {
  26                 'id': 'rOoKv2OMpOw',
  27                 'ext': 'mp4',
  28                 'title': 'Shooting star seen on 7-Sep-2015',
  29                 'description': 'md5:a1cd2e74f6ee6851552c9cf5851d6b06',
  30                 'uploader': 'Porjai Jaturongkhakun',
  31                 'upload_date': '20150906',
  32                 'uploader_id': 'UCnLY_3ezwNcDSC_Wc6suZxw',
  33             },
  34             'add_ie': ['Youtube'],
  35         }
  36     ]
  37
  38     def _real_extract(self, url):
  39         display_id = self._match_id(url)
  40
  41         webpage = self._download_webpage(url, display_id)
  42
  43         setup_js = self._parse_json(
  44             js_to_json(
  45                 self._search_regex(
  46                     r"jwplayer\('player-vivo'\).setup\((\{.*?\})\)",
  47                     webpage,
  48                     'setup code',
  49                     flags=re.DOTALL
  50                 ).replace('\n//', '')
  51             ),
  52             display_id
  53         )
  54         title = setup_js['title']
  55         thumbnail = setup_js.get('image') or setup_js['playlist'][0].get('image')
  56         description = self._html_search_meta(
  57             'description', webpage, 'description')
  58
  59         formats = []
  60         for f in setup_js['playlist'][0]['sources']:
  61             format_url = f['file']
  62             if format_url != '':
  63                 if '.m3u8' in format_url:
  64                     formats.extend(self._extract_m3u8_formats(format_url, display_id))
  65                 else:
  66                     if 'youtube.com' in format_url:
  67                         return self.url_result(format_url, 'Youtube')
  68                     else:
  69                         formats.append({'url': format_url, 'format_id': f.get('label')})
  70
  71         return {
  72             'id': display_id,
  73             'title': title,
  74             'description': description,
  75             'thumbnail': thumbnail,
  76             'formats': formats,
  77         }