[IMP] web_editor: improve video embed regexes (youku, dailymotion)

The regex patterns for dailymotion and youku were too permissive and
not specific enough for the way they were intended to be used.

The new regexes should correctly isolate the video ID from the URL
string and cover all embedding needs for the two providers.
There are also new test cases for specific embed URL variations.

For the initiating discussion and PR, see following link:
https://github.com/odoo/odoo/pull/44537#discussion_r753250276

task-2696033

closes odoo/odoo#82369

Signed-off-by: Quentin Smetz (qsm) <qsm@odoo.com>

addons/web_editor/tests/test_tools.py

@@ -14,7 +14,14 @@ class TestVideoUtils(common.BaseCase):
         'vimeo': 'https://vimeo.com/395399735',
         'dailymotion': 'https://www.dailymotion.com/video/x7svr6t',
         'youku': 'https://v.youku.com/v_show/id_XMzY1MjY4.html?spm=a2hzp.8244740.0.0',
-        'instagram': 'https://www.instagram.com/p/B6dXGTxggTG/'
+        'instagram': 'https://www.instagram.com/p/B6dXGTxggTG/',
+        'dailymotion_hub_no_video': 'http://www.dailymotion.com/hub/x9q_Galatasaray',
+        'dailymotion_hub_#video': 'http://www.dailymotion.com/hub/x9q_Galatasaray#video=x2jvvep',
+        'dai.ly': 'https://dai.ly/x578has',
+        'dailymotion_embed': 'https://www.dailymotion.com/embed/video/x578has?autoplay=1',
+        'dailymotion_video_extra': 'https://www.dailymotion.com/video/x2jvvep_hakan-yukur-klip_sport',
+        'player_youku': 'https://player.youku.com/player.php/sid/XMTI5Mjg5NjE4MA==/v.swf',
+        'youku_embed': 'https://player.youku.com/embed/XNTIwMzE1MzUzNg',
     }
 
     def test_player_regexes(self):
@@ -40,9 +47,22 @@ class TestVideoUtils(common.BaseCase):
         #dailymotion
         self.assertEqual('dailymotion', tools.get_video_source_data(TestVideoUtils.urls['dailymotion'])[0])
         self.assertEqual('x7svr6t', tools.get_video_source_data(TestVideoUtils.urls['dailymotion'])[1])
+        self.assertEqual(None, tools.get_video_source_data(TestVideoUtils.urls['dailymotion_hub_no_video']))
+        self.assertEqual('dailymotion', tools.get_video_source_data(TestVideoUtils.urls['dailymotion_hub_#video'])[0])
+        self.assertEqual('x2jvvep', tools.get_video_source_data(TestVideoUtils.urls['dailymotion_hub_#video'])[1])
+        self.assertEqual('dailymotion', tools.get_video_source_data(TestVideoUtils.urls['dai.ly'])[0])
+        self.assertEqual('x578has', tools.get_video_source_data(TestVideoUtils.urls['dai.ly'])[1])
+        self.assertEqual('dailymotion', tools.get_video_source_data(TestVideoUtils.urls['dailymotion_embed'])[0])
+        self.assertEqual('x578has', tools.get_video_source_data(TestVideoUtils.urls['dailymotion_embed'])[1])
+        self.assertEqual('dailymotion', tools.get_video_source_data(TestVideoUtils.urls['dailymotion_video_extra'])[0])
+        self.assertEqual('x2jvvep', tools.get_video_source_data(TestVideoUtils.urls['dailymotion_video_extra'])[1])
         #youku
         self.assertEqual('youku', tools.get_video_source_data(TestVideoUtils.urls['youku'])[0])
         self.assertEqual('XMzY1MjY4', tools.get_video_source_data(TestVideoUtils.urls['youku'])[1])
+        self.assertEqual('youku', tools.get_video_source_data(TestVideoUtils.urls['player_youku'])[0])
+        self.assertEqual('XMTI5Mjg5NjE4MA', tools.get_video_source_data(TestVideoUtils.urls['player_youku'])[1])
+        self.assertEqual('youku', tools.get_video_source_data(TestVideoUtils.urls['youku_embed'])[0])
+        self.assertEqual('XNTIwMzE1MzUzNg', tools.get_video_source_data(TestVideoUtils.urls['youku_embed'])[1])
         #instagram
         self.assertEqual('instagram', tools.get_video_source_data(TestVideoUtils.urls['instagram'])[0])
         self.assertEqual('B6dXGTxggTG', tools.get_video_source_data(TestVideoUtils.urls['instagram'])[1])

addons/web_editor/tools.py

@@ -18,9 +18,9 @@ valid_url_regex = r'^(http://|https://|//)[a-z0-9]+([\-\.]{1}[a-z0-9]+)*\.[a-z]{
 player_regexes = {
     'youtube': r'^(?:(?:https?:)?//)?(?:www\.)?(?:youtu\.be/|youtube(-nocookie)?\.com/(?:embed/|v/|watch\?v=|watch\?.+&v=))((?:\w|-){11})\S*$',
     'vimeo': r'//(player.)?vimeo.com/([a-z]*/)*([0-9]{6,11})[?]?.*',
-    'dailymotion': r'(?:.+dailymotion.com/(video|hub|embed/video|embed)|dai\.ly)/([^_?]+)[^#]*(#video=([^_&]+))?',
+    'dailymotion': r'(https?:\/\/)(www\.)?(dailymotion\.com\/(embed\/video\/|embed\/|video\/|hub\/.*#video=)|dai\.ly\/)(?P<id>[A-Za-z0-9]{6,7})',
     'instagram': r'(?:(.*)instagram.com|instagr\.am)/p/(.[a-zA-Z0-9-_\.]*)',
-    'youku': r'(.*).youku\.com/(v_show/id_|embed/)(.+)',
+    'youku': r'(?:(https?:\/\/)?(v\.youku\.com/v_show/id_|player\.youku\.com/player\.php/sid/|player\.youku\.com/embed/|cloud\.youku\.com/services/sharev\?vid=|video\.tudou\.com/v/)|youku:)(?P<id>[A-Za-z0-9]+)(?:\.html|/v\.swf|)',
 }
 
 
@@ -40,16 +40,13 @@ def get_video_source_data(video_url):
             return ('vimeo', vimeo_match[3], vimeo_match)
         dailymotion_match = re.search(player_regexes['dailymotion'], video_url)
         if dailymotion_match:
-            return ('dailymotion', dailymotion_match[2], dailymotion_match)
+            return ('dailymotion', dailymotion_match.group("id"), dailymotion_match)
         instagram_match = re.search(player_regexes['instagram'], video_url)
         if instagram_match:
             return ('instagram', instagram_match[2], instagram_match)
         youku_match = re.search(player_regexes['youku'], video_url)
         if youku_match:
-            youku_link = youku_match[3]
-            if '.html?' in youku_link:
-                youku_link = youku_link.split('.html?')[0]
-            return ('youku', youku_link, youku_match)
+            return ('youku', youku_match.group("id"), youku_match)
     return None