[utils,franceinter] Add french months' names and fix extraction

[youtube-dl.git] / youtube_dl / extractor / franceinter.py

diff --git a/youtube_dl/extractor/franceinter.py b/youtube_dl/extractor/franceinter.py
index 932a1f161794a8b89487e2d7483f651c0c3b1551..6dad8d712289bc55799dbaac9a649e2b9abbe3ff 100644 (file)
--- a/youtube_dl/extractor/franceinter.py
+++ b/youtube_dl/extractor/franceinter.py
@@ -1,31 +1,53 @@
+# coding: utf-8
 from __future__ import unicode_literals
 
-import re
-
 from .common import InfoExtractor
+from ..utils import (
+    unified_timestamp,
+    month_by_name,
+)
+
+
 class FranceInterIE(InfoExtractor):
-    
-    _VALID_URL=r'http://(?:www\.)?franceinter\.fr/player/reecouter\?play=(?P<id>[0-9]{6})'
-    _TEST={
-           u'url':u'http://www.franceinter.fr/player/reecouter?play=793962',
-           u'file':u'793962.mp3'
-                  
-           }
-           
-   
-        
-   
-    def _real_extract(self,url):
-
-        mobj = re.match(self._VALID_URL, url)
-        video_id = mobj.group('id')
-        
-        webpage=self._download_webpage(url,video_id)
-        
-        title=self._search_regex(u'(?<=<span class="roll_overflow">)(.*)(?=</span></h1>)', webpage, u'title')
-        
-        video_url='http://www.franceinter.fr/'+self._search_regex(u'(?<=&urlAOD=)(.*)(?=&startTime)', webpage, u'video url')
-        
-        return{'id': video_id,u'url': video_url,u'title': title}
-        
-        
\ No newline at end of file
+    _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',
+        'info_dict': {
+            'id': 'la-marche-de-l-histoire/la-marche-de-l-histoire-18-decembre-2013',
+            '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',
+        },
+    }
+
+    def _real_extract(self, url):
+        video_id = self._match_id(url)
+
+        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')
+
+        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)
+
+        return {
+            'id': video_id,
+            'title': title,
+            'description': description,
+            'timestamp': timestamp,
+            'formats': [{
+                'url': video_url,
+                'vcodec': 'none',
+            }],
+        }