Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/vzaar.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 (
   8     int_or_none,
   9     float_or_none,
  10 )
  11
  12
  13 class VzaarIE(InfoExtractor):
  14     _VALID_URL = r'https?://(?:(?:www|view)\.)?vzaar\.com/(?:videos/)?(?P<id>\d+)'
  15     _TESTS = [{
  16         'url': 'https://vzaar.com/videos/1152805',
  17         'md5': 'bde5ddfeb104a6c56a93a06b04901dbf',
  18         'info_dict': {
  19             'id': '1152805',
  20             'ext': 'mp4',
  21             'title': 'sample video (public)',
  22         },
  23     }, {
  24         'url': 'https://view.vzaar.com/27272/player',
  25         'md5': '3b50012ac9bbce7f445550d54e0508f2',
  26         'info_dict': {
  27             'id': '27272',
  28             'ext': 'mp3',
  29             'title': 'MP3',
  30         },
  31     }]
  32
  33     @staticmethod
  34     def _extract_urls(webpage):
  35         return re.findall(
  36             r'<iframe[^>]+src=["\']((?:https?:)?//(?:view\.vzaar\.com)/[0-9]+)',
  37             webpage)
  38
  39     def _real_extract(self, url):
  40         video_id = self._match_id(url)
  41         video_data = self._download_json(
  42             'http://view.vzaar.com/v2/%s/video' % video_id, video_id)
  43         source_url = video_data['sourceUrl']
  44
  45         info = {
  46             'id': video_id,
  47             'title': video_data['videoTitle'],
  48             'url': source_url,
  49             'thumbnail': self._proto_relative_url(video_data.get('poster')),
  50             'duration': float_or_none(video_data.get('videoDuration')),
  51         }
  52         if 'audio' in source_url:
  53             info.update({
  54                 'vcodec': 'none',
  55                 'ext': 'mp3',
  56             })
  57         else:
  58             info.update({
  59                 'width': int_or_none(video_data.get('width')),
  60                 'height': int_or_none(video_data.get('height')),
  61                 'ext': 'mp4',
  62             })
  63         return info