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

   1 from __future__ import unicode_literals
   2
   3 import re
   4
   5 from .common import InfoExtractor
   6 from ..utils import (
   7     ExtractorError,
   8     js_to_json,
   9 )
  10
  11
  12 class ABCIE(InfoExtractor):
  13     IE_NAME = 'abc.net.au'
  14     _VALID_URL = r'http://www\.abc\.net\.au/news/[^/]+/[^/]+/(?P<id>\d+)'
  15
  16     _TESTS = [{
  17         'url': 'http://www.abc.net.au/news/2014-11-05/australia-to-staff-ebola-treatment-centre-in-sierra-leone/5868334',
  18         'md5': 'cb3dd03b18455a661071ee1e28344d9f',
  19         'info_dict': {
  20             'id': '5868334',
  21             'ext': 'mp4',
  22             'title': 'Australia to help staff Ebola treatment centre in Sierra Leone',
  23             'description': 'md5:809ad29c67a05f54eb41f2a105693a67',
  24         },
  25     }, {
  26         'url': 'http://www.abc.net.au/news/2015-08-17/warren-entsch-introduces-same-sex-marriage-bill/6702326',
  27         'md5': 'db2a5369238b51f9811ad815b69dc086',
  28         'info_dict': {
  29             'id': 'NvqvPeNZsHU',
  30             'ext': 'mp4',
  31             'upload_date': '20150816',
  32             'uploader': 'ABC News (Australia)',
  33             'description': 'Government backbencher Warren Entsch introduces a cross-party sponsored bill to legalise same-sex marriage, saying the bill is designed to promote "an inclusive Australia, not a divided one.". Read more here: http://ab.co/1Mwc6ef',
  34             'uploader_id': 'NewsOnABC',
  35             'title': 'Marriage Equality: Warren Entsch introduces same sex marriage bill',
  36         },
  37         'add_ie': ['Youtube'],
  38     }]
  39
  40     def _real_extract(self, url):
  41         video_id = self._match_id(url)
  42         webpage = self._download_webpage(url, video_id)
  43
  44         mobj = re.search(
  45             r'inline(?P<type>Video|YouTube)Data\.push\((?P<json_data>[^)]+)\);',
  46             webpage)
  47         if mobj is None:
  48             raise ExtractorError('Unable to extract video urls')
  49
  50         urls_info = self._parse_json(
  51             mobj.group('json_data'), video_id, transform_source=js_to_json)
  52
  53         if not isinstance(urls_info, list):
  54             urls_info = [urls_info]
  55
  56         if mobj.group('type') == 'YouTube':
  57             return self.playlist_result([
  58                 self.url_result(url_info['url']) for url_info in urls_info])
  59
  60         formats = [{
  61             'url': url_info['url'],
  62             'width': int(url_info['width']),
  63             'height': int(url_info['height']),
  64             'tbr': int(url_info['bitrate']),
  65             'filesize': int(url_info['filesize']),
  66         } for url_info in urls_info]
  67         self._sort_formats(formats)
  68
  69         return {
  70             'id': video_id,
  71             'title': self._og_search_title(webpage),
  72             'formats': formats,
  73             'description': self._og_search_description(webpage),
  74             'thumbnail': self._og_search_thumbnail(webpage),
  75         }