[arte] Improve _VALID_URLs

[youtube-dl] / youtube_dl / extractor / arte.py

diff --git a/youtube_dl/extractor/arte.py b/youtube_dl/extractor/arte.py
index b524866d91b064eabce645e4f998187da4078a20..793da2ee1d8b015a5573d7febea439bb4f5e465a 100644 (file)
--- a/youtube_dl/extractor/arte.py
+++ b/youtube_dl/extractor/arte.py
@@ -63,7 +63,7 @@ class ArteTvIE(InfoExtractor):
 
 class ArteTVPlus7IE(InfoExtractor):
     IE_NAME = 'arte.tv:+7'
-    _VALID_URL = r'https?://(?:www\.)?arte\.tv/guide/(?P<lang>fr|de|en|es)/(?:(?:sendungen|emissions)/)?(?P<id>.*?)/(?P<name>.*?)(\?.*)?'
+    _VALID_URL = r'https?://(?:www\.)?arte\.tv/guide/(?P<lang>fr|de|en|es)/(?:(?:sendungen|emissions)/)?(?P<id>[^/]+)/(?P<name>[^/?#&+])'
 
     @classmethod
     def _extract_url_info(cls, url):
@@ -198,7 +198,7 @@ class ArteTVPlus7IE(InfoExtractor):
 # It also uses the arte_vp_url url from the webpage to extract the information
 class ArteTVCreativeIE(ArteTVPlus7IE):
     IE_NAME = 'arte.tv:creative'
-    _VALID_URL = r'https?://creative\.arte\.tv/(?P<lang>fr|de|en|es)/(?:magazine?/)?(?P<id>[^?#]+)'
+    _VALID_URL = r'https?://creative\.arte\.tv/(?P<lang>fr|de|en|es)/(?:magazine?/)?(?P<id>[^/?#&]+)'
 
     _TESTS = [{
         'url': 'http://creative.arte.tv/de/magazin/agentur-amateur-corporate-design',
@@ -222,7 +222,7 @@ class ArteTVCreativeIE(ArteTVPlus7IE):
 
 class ArteTVFutureIE(ArteTVPlus7IE):
     IE_NAME = 'arte.tv:future'
-    _VALID_URL = r'https?://future\.arte\.tv/(?P<lang>fr|de|en|es)/(?P<id>.+)'
+    _VALID_URL = r'https?://future\.arte\.tv/(?P<lang>fr|de|en|es)/(?P<id>[^/?#&]+)'
 
     _TESTS = [{
         'url': 'http://future.arte.tv/fr/info-sciences/les-ecrevisses-aussi-sont-anxieuses',
@@ -239,7 +239,7 @@ class ArteTVFutureIE(ArteTVPlus7IE):
 
 class ArteTVDDCIE(ArteTVPlus7IE):
     IE_NAME = 'arte.tv:ddc'
-    _VALID_URL = r'https?://ddc\.arte\.tv/(?P<lang>emission|folge)/(?P<id>.+)'
+    _VALID_URL = r'https?://ddc\.arte\.tv/(?P<lang>emission|folge)/(?P<id>[^/?#&]+)'
 
     def _real_extract(self, url):
         video_id, lang = self._extract_url_info(url)
@@ -257,7 +257,7 @@ class ArteTVDDCIE(ArteTVPlus7IE):
 
 class ArteTVConcertIE(ArteTVPlus7IE):
     IE_NAME = 'arte.tv:concert'
-    _VALID_URL = r'https?://concert\.arte\.tv/(?P<lang>fr|de|en|es)/(?P<id>.+)'
+    _VALID_URL = r'https?://concert\.arte\.tv/(?P<lang>fr|de|en|es)/(?P<id>[^/?#&]+)'
 
     _TEST = {
         'url': 'http://concert.arte.tv/de/notwist-im-pariser-konzertclub-divan-du-monde',
@@ -291,7 +291,7 @@ class ArteTVCinemaIE(ArteTVPlus7IE):
 
 class ArteTVMagazineIE(ArteTVPlus7IE):
     IE_NAME = 'arte.tv:magazine'
-    _VALID_URL = r'https?://(?:www\.)?arte\.tv/magazine/(?P<name>.*?)/(?P<lang>fr|de)/(?P<id>.*?)'
+    _VALID_URL = r'https?://(?:www\.)?arte\.tv/magazine/[^/]+/(?P<lang>fr|de|en|es)/(?P<id>[^/?#&]+)'
 
     _TESTS = [{
         'url': 'http://www.arte.tv/magazine/trepalium/fr/entretien-avec-le-realisateur-vincent-lannoo-trepalium',
@@ -303,12 +303,7 @@ class ArteTVMagazineIE(ArteTVPlus7IE):
         },
     }, {
         'url': 'http://www.arte.tv/magazine/metropolis/de/frank-woeste-german-paris-metropolis',
-        'md5': '53185a2b4876383552c037c21aeca8f7',
-        'info_dict': {
-            'id': '066149-000-A',
-            'ext': 'mp4',
-            'title': 'Bonus: Frank Woeste - ein deutscher Jazzmusiker in Paris',
-        },
+        'only_matching': True,
     }]