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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 from .common import InfoExtractor
   5 from ..compat import compat_str
   6 from ..utils import month_by_name
   7
   8
   9 class FranceInterIE(InfoExtractor):
  10     _VALID_URL = r'https?://(?:www\.)?franceinter\.fr/emissions/(?P<id>[^?#]+)'
  11
  12     _TEST = {
  13         'url': 'https://www.franceinter.fr/emissions/la-marche-de-l-histoire/la-marche-de-l-histoire-18-decembre-2013',
  14         'md5': '4764932e466e6f6c79c317d2e74f6884',
  15         'info_dict': {
  16             'id': 'la-marche-de-l-histoire/la-marche-de-l-histoire-18-decembre-2013',
  17             'ext': 'mp3',
  18             'title': 'L’Histoire dans les jeux vidéo du 18 décembre 2013 - France Inter',
  19             'description': 'md5:7f2ce449894d1e585932273080fb410d',
  20             'upload_date': '20131218',
  21         },
  22     }
  23
  24     def _real_extract(self, url):
  25         video_id = self._match_id(url)
  26
  27         webpage = self._download_webpage(url, video_id)
  28
  29         video_url = self._search_regex(
  30             r'(?s)<div[^>]+class=["\']page-diffusion["\'][^>]*>.*?<button[^>]+data-url=(["\'])(?P<url>(?:(?!\1).)+)\1',
  31             webpage, 'video url', group='url')
  32
  33         title = self._og_search_title(webpage)
  34         description = self._og_search_description(webpage)
  35
  36         upload_date_str = self._search_regex(
  37             r'class=["\']cover-emission-period["\'][^>]*>[^<]+\s+(\d{1,2}\s+[^\s]+\s+\d{4})<',
  38             webpage, 'upload date', fatal=False)
  39         if upload_date_str:
  40             upload_date_list = upload_date_str.split()
  41             upload_date_list.reverse()
  42             upload_date_list[1] = compat_str(month_by_name(upload_date_list[1], lang='fr'))
  43             upload_date = ''.join(upload_date_list)
  44         else:
  45             upload_date = None
  46
  47         return {
  48             'id': video_id,
  49             'title': title,
  50             'description': description,
  51             'upload_date': upload_date,
  52             'formats': [{
  53                 'url': video_url,
  54                 'vcodec': 'none',
  55             }],
  56         }