[ie/matchtv] Fix extractor (#10190)

[yt-dlp.git] / yt_dlp / extractor / bfmtv.py

diff --git a/yt_dlp/extractor/bfmtv.py b/yt_dlp/extractor/bfmtv.py
index 48526e38b64c97646e4794d790e7bf31a4e64322..87f011783bfef6a75f79e05b4a86c8b1b51666f8 100644 (file)
--- a/yt_dlp/extractor/bfmtv.py
+++ b/yt_dlp/extractor/bfmtv.py
@@ -5,9 +5,9 @@
 
 
 class BFMTVBaseIE(InfoExtractor):
-    _VALID_URL_BASE = r'https?://(?:www\.)?bfmtv\.com/'
+    _VALID_URL_BASE = r'https?://(?:www\.|rmc\.)?bfmtv\.com/'
     _VALID_URL_TMPL = _VALID_URL_BASE + r'(?:[^/]+/)*[^/?&#]+_%s[A-Z]-(?P<id>\d{12})\.html'
-    _VIDEO_BLOCK_REGEX = r'(<div[^>]+class="video_block"[^>]*>)'
+    _VIDEO_BLOCK_REGEX = r'(<div[^>]+class="video_block[^"]*"[^>]*>)'
     BRIGHTCOVE_URL_TEMPLATE = 'http://players.brightcove.net/%s/%s_default/index.html?videoId=%s'
 
     def _brightcove_url_result(self, video_id, video_block):
@@ -31,6 +31,9 @@ class BFMTVIE(BFMTVBaseIE):
             'uploader_id': '876450610001',
             'upload_date': '20201002',
             'timestamp': 1601629620,
+            'duration': 44.757,
+            'tags': ['bfmactu', 'politique'],
+            'thumbnail': 'https://cf-images.eu-west-1.prod.boltdns.net/v1/static/876450610001/5041f4c1-bc48-4af8-a256-1b8300ad8ef0/cf2f9114-e8e2-4494-82b4-ab794ea4bc7d/1920x1080/match/image.jpg',
         },
     }]
 
@@ -42,7 +45,7 @@ def _real_extract(self, url):
         return self._brightcove_url_result(video_block['videoid'], video_block)
 
 
-class BFMTVLiveIE(BFMTVIE):
+class BFMTVLiveIE(BFMTVIE):  # XXX: Do not subclass from concrete IE
     IE_NAME = 'bfmtv:live'
     _VALID_URL = BFMTVBaseIE._VALID_URL_BASE + '(?P<id>(?:[^/]+/)?en-direct)'
     _TESTS = [{
@@ -52,8 +55,11 @@ class BFMTVLiveIE(BFMTVIE):
             'ext': 'mp4',
             'title': r're:^le direct BFMTV WEB \d{4}-\d{2}-\d{2} \d{2}:\d{2}$',
             'uploader_id': '876450610001',
-            'upload_date': '20171018',
-            'timestamp': 1508329950,
+            'upload_date': '20220926',
+            'timestamp': 1664207191,
+            'live_status': 'is_live',
+            'thumbnail': r're:https://.+/image\.jpg',
+            'tags': [],
         },
         'params': {
             'skip_download': True,
@@ -81,6 +87,19 @@ class BFMTVArticleIE(BFMTVBaseIE):
     }, {
         'url': 'https://www.bfmtv.com/sante/covid-19-oui-le-vaccin-de-pfizer-distribue-en-france-a-bien-ete-teste-sur-des-personnes-agees_AN-202101060275.html',
         'only_matching': True,
+    }, {
+        'url': 'https://rmc.bfmtv.com/actualites/societe/transports/ce-n-est-plus-tout-rentable-le-bioethanol-e85-depasse-1eu-le-litre-des-automobilistes-regrettent_AV-202301100268.html',
+        'info_dict': {
+            'id': '6318445464112',
+            'ext': 'mp4',
+            'title': 'Le plein de bioéthanol fait de plus en plus mal à la pompe',
+            'uploader_id': '876630703001',
+            'upload_date': '20230110',
+            'timestamp': 1673341692,
+            'duration': 109.269,
+            'tags': ['rmc', 'show', 'apolline de malherbe', 'info', 'talk', 'matinale', 'radio'],
+            'thumbnail': 'https://cf-images.eu-west-1.prod.boltdns.net/v1/static/876630703001/5bef74b8-9d5e-4480-a21f-60c2e2480c46/96c88b74-f9db-45e1-8040-e199c5da216c/1920x1080/match/image.jpg',
+        },
     }]
 
     def _real_extract(self, url):