yt_dlp/extractor/reuters.py

   1 import re
   2
   3 from .common import InfoExtractor
   4 from ..utils import (
   5     js_to_json,
   6     int_or_none,
   7     unescapeHTML,
   8 )
   9
  10
  11 class ReutersIE(InfoExtractor):
  12     _VALID_URL = r'https?://(?:www\.)?reuters\.com/.*?\?.*?videoId=(?P<id>[0-9]+)'
  13     _TEST = {
  14         'url': 'http://www.reuters.com/video/2016/05/20/san-francisco-police-chief-resigns?videoId=368575562',
  15         'md5': '8015113643a0b12838f160b0b81cc2ee',
  16         'info_dict': {
  17             'id': '368575562',
  18             'ext': 'mp4',
  19             'title': 'San Francisco police chief resigns',
  20         }
  21     }
  22
  23     def _real_extract(self, url):
  24         video_id = self._match_id(url)
  25         webpage = self._download_webpage(
  26             'http://www.reuters.com/assets/iframe/yovideo?videoId=%s' % video_id, video_id)
  27         video_data = js_to_json(self._search_regex(
  28             r'(?s)Reuters\.yovideo\.drawPlayer\(({.*?})\);',
  29             webpage, 'video data'))
  30
  31         def get_json_value(key, fatal=False):
  32             return self._search_regex(r'"%s"\s*:\s*"([^"]+)"' % key, video_data, key, fatal=fatal)
  33
  34         title = unescapeHTML(get_json_value('title', fatal=True))
  35         mmid, fid = re.search(r',/(\d+)\?f=(\d+)', get_json_value('flv', fatal=True)).groups()
  36
  37         mas_data = self._download_json(
  38             'http://mas-e.cds1.yospace.com/mas/%s/%s?trans=json' % (mmid, fid),
  39             video_id, transform_source=js_to_json)
  40         formats = []
  41         for f in mas_data:
  42             f_url = f.get('url')
  43             if not f_url:
  44                 continue
  45             method = f.get('method')
  46             if method == 'hls':
  47                 formats.extend(self._extract_m3u8_formats(
  48                     f_url, video_id, 'mp4', 'm3u8_native', m3u8_id='hls', fatal=False))
  49             else:
  50                 container = f.get('container')
  51                 ext = '3gp' if method == 'mobile' else container
  52                 formats.append({
  53                     'format_id': ext,
  54                     'url': f_url,
  55                     'ext': ext,
  56                     'container': container if method != 'mobile' else None,
  57                 })
  58
  59         return {
  60             'id': video_id,
  61             'title': title,
  62             'thumbnail': get_json_value('thumb'),
  63             'duration': int_or_none(get_json_value('seconds')),
  64             'formats': formats,
  65         }