]> git.bitcoin.ninja Git - youtube-dl/commitdiff
[lemonde] Add extractor
authorSergey M․ <dstftw@gmail.com>
Tue, 19 Jan 2016 16:09:55 +0000 (22:09 +0600)
committerSergey M․ <dstftw@gmail.com>
Tue, 19 Jan 2016 16:09:55 +0000 (22:09 +0600)
youtube_dl/extractor/__init__.py
youtube_dl/extractor/lemonde.py [new file with mode: 0644]

index 1d838cdd0afdd1be2b0ea8d33271025b8b47f1de..0cc339a370a559ef603333e2a39e633beea1587f 100644 (file)
@@ -333,6 +333,7 @@ from .kuwo import (
 from .la7 import LA7IE
 from .laola1tv import Laola1TvIE
 from .lecture2go import Lecture2GoIE
+from .lemonde import LemondeIE
 from .letv import (
     LetvIE,
     LetvTvIE,
diff --git a/youtube_dl/extractor/lemonde.py b/youtube_dl/extractor/lemonde.py
new file mode 100644 (file)
index 0000000..be66fff
--- /dev/null
@@ -0,0 +1,34 @@
+from __future__ import unicode_literals
+
+from .common import InfoExtractor
+
+
+class LemondeIE(InfoExtractor):
+    _VALID_URL = r'https?://(?:.+?\.)?lemonde\.fr/(?:[^/]+/)*(?P<id>[^/]+)\.html'
+    _TESTS = [{
+        'url': 'http://www.lemonde.fr/police-justice/video/2016/01/19/comprendre-l-affaire-bygmalion-en-cinq-minutes_4849702_1653578.html',
+        'md5': '01fb3c92de4c12c573343d63e163d302',
+        'info_dict': {
+            'id': 'lqm3kl',
+            'ext': 'mp4',
+            'title': "Comprendre l'affaire Bygmalion en 5 minutes",
+            'thumbnail': 're:^https?://.*\.jpg',
+            'duration': 320,
+            'upload_date': '20160119',
+            'timestamp': 1453194778,
+            'uploader_id': '3pmkp',
+        },
+    }, {
+        'url': 'http://redaction.actu.lemonde.fr/societe/video/2016/01/18/calais-debut-des-travaux-de-defrichement-dans-la-jungle_4849233_3224.html',
+        'only_matching': True,
+    }]
+
+    def _real_extract(self, url):
+        display_id = self._match_id(url)
+
+        webpage = self._download_webpage(url, display_id)
+
+        digiteka_url = self._proto_relative_url(self._search_regex(
+            r'url\s*:\s*(["\'])(?P<url>(?:https?://)?//(?:www\.)?(?:digiteka\.net|ultimedia\.com)/deliver/.+?)\1',
+            webpage, 'digiteka url', group='url'))
+        return self.url_result(digiteka_url, 'Digiteka')