]> jfr.im git - yt-dlp.git/blame - youtube_dl/extractor/franceculture.py
Revert "pull changes from remote master (#190)" (#193)
[yt-dlp.git] / youtube_dl / extractor / franceculture.py
CommitLineData
cce929ea
PH
1# coding: utf-8
2from __future__ import unicode_literals
3
cce929ea 4from .common import InfoExtractor
23d9ded6
PH
5from ..utils import (
6 determine_ext,
31a12140
RA
7 extract_attributes,
8 int_or_none,
23d9ded6 9)
cce929ea
PH
10
11
12class FranceCultureIE(InfoExtractor):
82997dad 13 _VALID_URL = r'https?://(?:www\.)?franceculture\.fr/emissions/(?:[^/]+/)*(?P<id>[^/?#&]+)'
cce929ea 14 _TEST = {
82997dad 15 'url': 'http://www.franceculture.fr/emissions/carnet-nomade/rendez-vous-au-pays-des-geeks',
cce929ea 16 'info_dict': {
82997dad
S
17 'id': 'rendez-vous-au-pays-des-geeks',
18 'display_id': 'rendez-vous-au-pays-des-geeks',
cce929ea
PH
19 'ext': 'mp3',
20 'title': 'Rendez-vous au pays des geeks',
ec85ded8 21 'thumbnail': r're:^https?://.*\.jpg$',
cce929ea 22 'upload_date': '20140301',
31a12140 23 'timestamp': 1393642916,
82997dad 24 'vcodec': 'none',
cce929ea
PH
25 }
26 }
27
82997dad
S
28 def _real_extract(self, url):
29 display_id = self._match_id(url)
23d9ded6 30
82997dad 31 webpage = self._download_webpage(url, display_id)
cce929ea 32
31a12140 33 video_data = extract_attributes(self._search_regex(
19a107f2 34 r'(?s)<div[^>]+class="[^"]*?(?:title-zone-diffusion|heading-zone-(?:wrapper|player-button))[^"]*?"[^>]*>.*?(<button[^>]+data-asset-source="[^"]+"[^>]+>)',
31a12140 35 webpage, 'video data'))
ecf17d16 36
31a12140
RA
37 video_url = video_data['data-asset-source']
38 title = video_data.get('data-asset-title') or self._og_search_title(webpage)
cce929ea 39
31a12140
RA
40 description = self._html_search_regex(
41 r'(?s)<div[^>]+class="intro"[^>]*>.*?<h2>(.+?)</h2>',
42 webpage, 'description', default=None)
82997dad 43 thumbnail = self._search_regex(
31a12140 44 r'(?s)<figure[^>]+itemtype="https://schema.org/ImageObject"[^>]*>.*?<img[^>]+(?:data-dejavu-)?src="([^"]+)"',
82997dad 45 webpage, 'thumbnail', fatal=False)
23d9ded6 46 uploader = self._html_search_regex(
31a12140 47 r'(?s)<span class="author">(.*?)</span>',
23d9ded6 48 webpage, 'uploader', default=None)
31a12140 49 ext = determine_ext(video_url.lower())
cce929ea
PH
50
51 return {
82997dad
S
52 'id': display_id,
53 'display_id': display_id,
cce929ea 54 'url': video_url,
cce929ea 55 'title': title,
31a12140 56 'description': description,
cce929ea 57 'thumbnail': thumbnail,
31a12140
RA
58 'ext': ext,
59 'vcodec': 'none' if ext == 'mp3' else None,
82997dad 60 'uploader': uploader,
31a12140
RA
61 'timestamp': int_or_none(video_data.get('data-asset-created-date')),
62 'duration': int_or_none(video_data.get('data-duration')),
cce929ea 63 }