from __future__ import unicode_literals
from .common import InfoExtractor
-from ..utils import (
- unified_timestamp,
- month_by_name,
-)
+from ..utils import month_by_name
class FranceInterIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?franceinter\.fr/emissions/(?P<id>[^?#]+)'
_TEST = {
- 'url': 'https://www.franceinter.fr/emissions/la-marche-de-l-histoire/la-marche-de-l-histoire-18-decembre-2013',
- 'md5': '4764932e466e6f6c79c317d2e74f6884',
+ 'url': 'https://www.franceinter.fr/emissions/affaires-sensibles/affaires-sensibles-07-septembre-2016',
+ 'md5': '9e54d7bdb6fdc02a841007f8a975c094',
'info_dict': {
- 'id': 'la-marche-de-l-histoire/la-marche-de-l-histoire-18-decembre-2013',
+ 'id': 'affaires-sensibles/affaires-sensibles-07-septembre-2016',
'ext': 'mp3',
- 'title': 'L’Histoire dans les jeux vidéo du 18 décembre 2013 - France Inter',
- 'description': 'L’Histoire dans les jeux vidéo du 18 décembre 2013 par Jean Lebrun en replay sur France Inter. Retrouvez l\'émission en réécoute gratuite et abonnez-vous au podcast !',
- 'timestamp': 1387324800,
- 'upload_date': '20131218',
+ 'title': 'Affaire Cahuzac : le contentieux du compte en Suisse',
+ 'description': 'md5:401969c5d318c061f86bda1fa359292b',
+ 'upload_date': '20160907',
},
}
webpage = self._download_webpage(url, video_id)
video_url = self._search_regex(
- r'<button class="replay-button playable" data-is-aod="1" data-url="([^"]+)"', webpage, 'video url')
+ r'(?s)<div[^>]+class=["\']page-diffusion["\'][^>]*>.*?<button[^>]+data-url=(["\'])(?P<url>(?:(?!\1).)+)\1',
+ webpage, 'video url', group='url')
title = self._og_search_title(webpage)
description = self._og_search_description(webpage)
- extractdate = self._search_regex('(\d{2}-([a-zA-Z\s]+)-\d{4}$)', url, 'extractdate', fatal=False)
- extractdate = extractdate.split('-')
- extractdate = extractdate[2] + "," + str(month_by_name(extractdate[1], 'fr')) + "," + extractdate[0]
-
- timestamp = unified_timestamp(extractdate)
+ upload_date_str = self._search_regex(
+ r'class=["\']\s*cover-emission-period\s*["\'][^>]*>[^<]+\s+(\d{1,2}\s+[^\s]+\s+\d{4})<',
+ webpage, 'upload date', fatal=False)
+ if upload_date_str:
+ upload_date_list = upload_date_str.split()
+ upload_date_list.reverse()
+ upload_date_list[1] = '%02d' % (month_by_name(upload_date_list[1], lang='fr') or 0)
+ upload_date_list[2] = '%02d' % int(upload_date_list[2])
+ upload_date = ''.join(upload_date_list)
+ else:
+ upload_date = None
return {
'id': video_id,
'title': title,
'description': description,
- 'timestamp': timestamp,
+ 'upload_date': upload_date,
'formats': [{
'url': video_url,
'vcodec': 'none',