[cleanup] Update extractor tests (#7718)

Authored by: trainman261

diff --git a/yt_dlp/extractor/aenetworks.py b/yt_dlp/extractor/aenetworks.py
index f049a0fb3cc53bbfd8605ba9f61a18fa0f2aeb01..cc26653c1deee0e684a945a237ab84df8dae8b96 100644 (file)
--- a/yt_dlp/extractor/aenetworks.py
+++ b/yt_dlp/extractor/aenetworks.py
@@ -338,6 +338,7 @@ class BiographyIE(AENetworksBaseIE):
             'skip_download': True,
         },
         'add_ie': ['ThePlatform'],
+        'skip': '404 Not Found',
     }]
 
     def _real_extract(self, url):

diff --git a/yt_dlp/extractor/amcnetworks.py b/yt_dlp/extractor/amcnetworks.py
index c58bc7bfbf8ffbfe5e362e27d43a90e0a712f645..10bd021c55ac20657efb0c271d9a9f88dd1587a5 100644 (file)
--- a/yt_dlp/extractor/amcnetworks.py
+++ b/yt_dlp/extractor/amcnetworks.py
@@ -26,6 +26,7 @@ class AMCNetworksIE(ThePlatformIE):  # XXX: Do not subclass from concrete IE
             # m3u8 download
             'skip_download': True,
         },
+        'skip': '404 Not Found',
     }, {
         'url': 'http://www.bbcamerica.com/shows/the-hunt/full-episodes/season-1/episode-01-the-hardest-challenge',
         'only_matching': True,

diff --git a/yt_dlp/extractor/cbc.py b/yt_dlp/extractor/cbc.py
index 2920b9027d8ef416387467f8f702cec88441b83b..be2d13e442f5cde6f069010b2af875e2e34142e6 100644 (file)
--- a/yt_dlp/extractor/cbc.py
+++ b/yt_dlp/extractor/cbc.py
@@ -66,6 +66,7 @@ class CBCIE(InfoExtractor):
             'uploader': 'CBCC-NEW',
             'timestamp': 255977160,
         },
+        'skip': '404 Not Found',
     }, {
         # multiple iframes
         'url': 'http://www.cbc.ca/natureofthings/blog/birds-eye-view-from-vancouvers-burrard-street-bridge-how-we-got-the-shot',
@@ -97,7 +98,7 @@ class CBCIE(InfoExtractor):
         # multiple CBC.APP.Caffeine.initInstance(...)
         'url': 'http://www.cbc.ca/news/canada/calgary/dog-indoor-exercise-winter-1.3928238',
         'info_dict': {
-            'title': 'Keep Rover active during the deep freeze with doggie pushups and other fun indoor tasks',
+            'title': 'Keep Rover active during the deep freeze with doggie pushups and other fun indoor tasks',  # FIXME
             'id': 'dog-indoor-exercise-winter-1.3928238',
             'description': 'md5:c18552e41726ee95bd75210d1ca9194c',
         },
@@ -476,6 +477,10 @@ class CBCGemPlaylistIE(InfoExtractor):
             'id': 'schitts-creek/s06',
             'title': 'Season 6',
             'description': 'md5:6a92104a56cbeb5818cc47884d4326a2',
+            'series': 'Schitt\'s Creek',
+            'season_number': 6,
+            'season': 'Season 6',
+            'thumbnail': 'https://images.radio-canada.ca/v1/synps-cbc/season/perso/cbc_schitts_creek_season_06_carousel_v03.jpg?impolicy=ott&im=Resize=(_Size_)&quality=75',
         },
     }, {
         'url': 'https://gem.cbc.ca/schitts-creek/s06',

diff --git a/yt_dlp/extractor/cbs.py b/yt_dlp/extractor/cbs.py
index 1c0dbdea9444c370bf8d5b7811ad39ccc9905cf9..d97fbd758c387d570b50064730685e224be072d7 100644 (file)
--- a/yt_dlp/extractor/cbs.py
+++ b/yt_dlp/extractor/cbs.py
@@ -101,6 +101,7 @@ class CBSIE(CBSBaseIE):
             # m3u8 download
             'skip_download': True,
         },
+        'skip': 'Subscription required',
     }, {
         'url': 'https://www.cbs.com/shows/video/sZH1MGgomIosZgxGJ1l263MFq16oMtW1/',
         'info_dict': {
@@ -117,6 +118,7 @@ class CBSIE(CBSBaseIE):
         },
         'expected_warnings': [
             'This content expired on', 'No video formats found', 'Requested format is not available'],
+        'skip': '404 Not Found',
     }, {
         'url': 'http://colbertlateshow.com/video/8GmB0oY0McANFvp2aEffk9jZZZ2YyXxy/the-colbeard/',
         'only_matching': True,

diff --git a/yt_dlp/extractor/cnbc.py b/yt_dlp/extractor/cnbc.py
index 68fd025b7c691acbd40a299740ddc938a47c2534..7d209b6d9012a0ca8ba25ac7f1ddf61bd6722619 100644 (file)
--- a/yt_dlp/extractor/cnbc.py
+++ b/yt_dlp/extractor/cnbc.py
@@ -19,6 +19,7 @@ class CNBCIE(InfoExtractor):
             # m3u8 download
             'skip_download': True,
         },
+        'skip': 'Dead link',
     }
 
     def _real_extract(self, url):
@@ -49,6 +50,7 @@ class CNBCVideoIE(InfoExtractor):
         'params': {
             'skip_download': True,
         },
+        'skip': 'Dead link',
     }
 
     def _real_extract(self, url):

diff --git a/yt_dlp/extractor/corus.py b/yt_dlp/extractor/corus.py
index c03d65310d2b9ca0924d6dc085390ff08e82560b..bcc34ddd8a12f8c942b43ab091309345b91ddd48 100644 (file)
--- a/yt_dlp/extractor/corus.py
+++ b/yt_dlp/extractor/corus.py
@@ -41,7 +41,7 @@ class CorusIE(ThePlatformFeedIE):  # XXX: Do not subclass from concrete IE
                         )
                     '''
     _TESTS = [{
-        'url': 'http://www.hgtv.ca/shows/bryan-inc/videos/movie-night-popcorn-with-bryan-870923331648/',
+        'url': 'https://www.hgtv.ca/video/bryan-inc/movie-night-popcorn-with-bryan/870923331648/',
         'info_dict': {
             'id': '870923331648',
             'ext': 'mp4',
@@ -54,6 +54,7 @@ class CorusIE(ThePlatformFeedIE):  # XXX: Do not subclass from concrete IE
             'skip_download': True,
         },
         'expected_warnings': ['Failed to parse JSON'],
+        # FIXME: yt-dlp wrongly raises for geo restriction
     }, {
         'url': 'http://www.foodnetwork.ca/shows/chopped/video/episode/chocolate-obsession/video.html?v=872683587753',
         'only_matching': True,

diff --git a/yt_dlp/extractor/generic.py b/yt_dlp/extractor/generic.py
index 33e71d1c572989c5ef90454ea6c2f457422b71f4..5e1240c13abf6b20a666f656cef57adc1218cad6 100644 (file)
--- a/yt_dlp/extractor/generic.py
+++ b/yt_dlp/extractor/generic.py
@@ -58,6 +58,8 @@ class GenericIE(InfoExtractor):
                 'ext': 'mp4',
                 'title': 'trailer',
                 'upload_date': '20100513',
+                'direct': True,
+                'timestamp': 1273772943.0,
             }
         },
         # Direct link to media delivered compressed (until Accept-Encoding is *)
@@ -101,6 +103,8 @@ class GenericIE(InfoExtractor):
                 'ext': 'webm',
                 'title': '5_Lennart_Poettering_-_Systemd',
                 'upload_date': '20141120',
+                'direct': True,
+                'timestamp': 1416498816.0,
             },
             'expected_warnings': [
                 'URL could be a direct video link, returning it as such.'
@@ -133,6 +137,7 @@ class GenericIE(InfoExtractor):
                     'upload_date': '20201204',
                 },
             }],
+            'skip': 'Dead link',
         },
         # RSS feed with item with description and thumbnails
         {
@@ -145,12 +150,12 @@ class GenericIE(InfoExtractor):
             'playlist': [{
                 'info_dict': {
                     'ext': 'm4a',
-                    'id': 'c1c879525ce2cb640b344507e682c36d',
+                    'id': '818a5d38-01cd-152f-2231-ee479677fa82',
                     'title': 're:Hydrogen!',
                     'description': 're:.*In this episode we are going.*',
                     'timestamp': 1567977776,
                     'upload_date': '20190908',
-                    'duration': 459,
+                    'duration': 423,
                     'thumbnail': r're:^https?://.*\.jpg$',
                     'episode_number': 1,
                     'season_number': 1,
@@ -267,6 +272,7 @@ class GenericIE(InfoExtractor):
             'params': {
                 'skip_download': True,
             },
+            'skip': '404 Not Found',
         },
         # MPD from http://dash-mse-test.appspot.com/media.html
         {
@@ -278,6 +284,7 @@ class GenericIE(InfoExtractor):
                 'title': 'car-20120827-manifest',
                 'formats': 'mincount:9',
                 'upload_date': '20130904',
+                'timestamp': 1378272859.0,
             },
         },
         # m3u8 served with Content-Type: audio/x-mpegURL; charset=utf-8
@@ -318,7 +325,7 @@ class GenericIE(InfoExtractor):
                 'id': 'cmQHVoWB5FY',
                 'ext': 'mp4',
                 'upload_date': '20130224',
-                'uploader_id': 'TheVerge',
+                'uploader_id': '@TheVerge',
                 'description': r're:^Chris Ziegler takes a look at the\.*',
                 'uploader': 'The Verge',
                 'title': 'First Firefox OS phones side-by-side',

diff --git a/yt_dlp/extractor/mediaset.py b/yt_dlp/extractor/mediaset.py
index e3b728dcae20e827e8a3c07710eb5b50a9882099..2d620429827248ed588c5c676455bd7ef523d9f8 100644 (file)
--- a/yt_dlp/extractor/mediaset.py
+++ b/yt_dlp/extractor/mediaset.py
@@ -127,7 +127,8 @@ class MediasetIE(ThePlatformBaseIE):
         },
         'params': {
             'skip_download': True,
-        }
+        },
+        'skip': 'Dead link',
     }, {
         # WittyTV embed
         'url': 'https://www.wittytv.it/mauriziocostanzoshow/ultima-puntata-venerdi-25-novembre/',

diff --git a/yt_dlp/extractor/movieclips.py b/yt_dlp/extractor/movieclips.py
index 4777f440e0121c36f4905065d5336442cfadc60b..f7f2921fdbac68f773cbc5b823233319d4d44317 100644 (file)
--- a/yt_dlp/extractor/movieclips.py
+++ b/yt_dlp/extractor/movieclips.py
@@ -23,6 +23,7 @@ class MovieClipsIE(InfoExtractor):
             'uploader': 'Movieclips',
         },
         'add_ie': ['ThePlatform'],
+        'skip': 'redirects to YouTube',
     }
 
     def _real_extract(self, url):

diff --git a/yt_dlp/extractor/nationalgeographic.py b/yt_dlp/extractor/nationalgeographic.py
index ad525c2589981c4603e83f469fc4470bd9c8ed87..6f046bc29c100b145d0265110080e176774b6976 100644 (file)
--- a/yt_dlp/extractor/nationalgeographic.py
+++ b/yt_dlp/extractor/nationalgeographic.py
@@ -24,6 +24,7 @@ class NationalGeographicVideoIE(InfoExtractor):
                 'uploader': 'NAGS',
             },
             'add_ie': ['ThePlatform'],
+            'skip': 'Redirects to main page',
         },
         {
             'url': 'http://video.nationalgeographic.com/wild/when-sharks-attack/the-real-jaws',
@@ -38,6 +39,7 @@ class NationalGeographicVideoIE(InfoExtractor):
                 'uploader': 'NAGS',
             },
             'add_ie': ['ThePlatform'],
+            'skip': 'Redirects to main page',
         },
     ]
 
@@ -75,6 +77,7 @@ class NationalGeographicTVIE(FOXIE):  # XXX: Do not subclass from concrete IE
         'params': {
             'skip_download': True,
         },
+        'skip': 'Content not available',
     }]
     _HOME_PAGE_URL = 'https://www.nationalgeographic.com/tv/'
     _API_KEY = '238bb0a0c2aba67922c48709ce0c06fd'

diff --git a/yt_dlp/extractor/nbc.py b/yt_dlp/extractor/nbc.py
index b3c28ab55d5bbaf1aa1bd797209cceb91f171f65..666550a4913f98735d8fc70268e5942299256db7 100644 (file)
--- a/yt_dlp/extractor/nbc.py
+++ b/yt_dlp/extractor/nbc.py
@@ -284,7 +284,7 @@ class NBCSportsIE(InfoExtractor):
 
     _TESTS = [{
         # iframe src
-        'url': 'http://www.nbcsports.com//college-basketball/ncaab/tom-izzo-michigan-st-has-so-much-respect-duke',
+        'url': 'https://www.nbcsports.com/watch/nfl/profootballtalk/pft-pm/unpacking-addisons-reckless-driving-citation',
         'info_dict': {
             'id': 'PHJSaFWbrTY9',
             'ext': 'mp4',
@@ -379,7 +379,7 @@ class NBCNewsIE(ThePlatformIE):  # XXX: Do not subclass from concrete IE
     _TESTS = [
         {
             'url': 'http://www.nbcnews.com/watch/nbcnews-com/how-twitter-reacted-to-the-snowden-interview-269389891880',
-            'md5': 'cf4bc9e6ce0130f00f545d80ecedd4bf',
+            'md5': 'fb3dcd2d7b1dd9804305fa2fc95ab610',  # md5 tends to fluctuate
             'info_dict': {
                 'id': '269389891880',
                 'ext': 'mp4',
@@ -387,6 +387,8 @@ class NBCNewsIE(ThePlatformIE):  # XXX: Do not subclass from concrete IE
                 'description': 'md5:65a0bd5d76fe114f3c2727aa3a81fe64',
                 'timestamp': 1401363060,
                 'upload_date': '20140529',
+                'duration': 46.0,
+                'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/140529/p_tweet_snow_140529.jpg',
             },
         },
         {
@@ -402,7 +404,7 @@ class NBCNewsIE(ThePlatformIE):  # XXX: Do not subclass from concrete IE
         },
         {
             'url': 'http://www.nbcnews.com/nightly-news/video/nightly-news-with-brian-williams-full-broadcast-february-4-394064451844',
-            'md5': '8eb831eca25bfa7d25ddd83e85946548',
+            'md5': '40d0e48c68896359c80372306ece0fc3',
             'info_dict': {
                 'id': '394064451844',
                 'ext': 'mp4',
@@ -410,11 +412,13 @@ class NBCNewsIE(ThePlatformIE):  # XXX: Do not subclass from concrete IE
                 'description': 'md5:1c10c1eccbe84a26e5debb4381e2d3c5',
                 'timestamp': 1423104900,
                 'upload_date': '20150205',
+                'duration': 1236.0,
+                'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/__NEW/nn_netcast_150204.jpg',
             },
         },
         {
             'url': 'http://www.nbcnews.com/business/autos/volkswagen-11-million-vehicles-could-have-suspect-software-emissions-scandal-n431456',
-            'md5': '4a8c4cec9e1ded51060bdda36ff0a5c0',
+            'md5': 'ffb59bcf0733dc3c7f0ace907f5e3939',
             'info_dict': {
                 'id': 'n431456',
                 'ext': 'mp4',
@@ -422,11 +426,13 @@ class NBCNewsIE(ThePlatformIE):  # XXX: Do not subclass from concrete IE
                 'description': 'md5:d22d1281a24f22ea0880741bb4dd6301',
                 'upload_date': '20150922',
                 'timestamp': 1442917800,
+                'duration': 37.0,
+                'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/__NEW/x_lon_vwhorn_150922.jpg',
             },
         },
         {
             'url': 'http://www.today.com/video/see-the-aurora-borealis-from-space-in-stunning-new-nasa-video-669831235788',
-            'md5': '118d7ca3f0bea6534f119c68ef539f71',
+            'md5': '693d1fa21d23afcc9b04c66b227ed9ff',
             'info_dict': {
                 'id': '669831235788',
                 'ext': 'mp4',
@@ -434,6 +440,8 @@ class NBCNewsIE(ThePlatformIE):  # XXX: Do not subclass from concrete IE
                 'description': 'md5:74752b7358afb99939c5f8bb2d1d04b1',
                 'upload_date': '20160420',
                 'timestamp': 1461152093,
+                'duration': 69.0,
+                'thumbnail': 'https://media-cldnry.s-nbcnews.com/image/upload/MSNBC/Components/Video/201604/2016-04-20T11-35-09-133Z--1280x720.jpg',
             },
         },
         {
@@ -447,6 +455,7 @@ class NBCNewsIE(ThePlatformIE):  # XXX: Do not subclass from concrete IE
                 'thumbnail': r're:^https?://.*\.jpg$',
                 'timestamp': 1406937606,
                 'upload_date': '20140802',
+                'duration': 940.0,
             },
         },
         {
@@ -535,6 +544,7 @@ class NBCOlympicsIE(InfoExtractor):
             'upload_date': '20160815',
             'uploader': 'NBCU-SPORTS',
         },
+        'skip': '404 Not Found',
     }
 
     def _real_extract(self, url):
@@ -578,6 +588,7 @@ class NBCOlympicsStreamIE(AdobePassIE):
             'params': {
                 'skip_download': 'm3u8',
             },
+            'skip': 'Livestream',
         }, {
             'note': 'Plain m3u8 source URL',
             'url': 'https://stream.nbcolympics.com/gymnastics-event-finals-mens-floor-pommel-horse-womens-vault-bars',
@@ -589,6 +600,7 @@ class NBCOlympicsStreamIE(AdobePassIE):
             'params': {
                 'skip_download': 'm3u8',
             },
+            'skip': 'Livestream',
         },
     ]
 

diff --git a/yt_dlp/extractor/scrippsnetworks.py b/yt_dlp/extractor/scrippsnetworks.py
index adfd7e5f297e5d53997d9a42f7e3c9fe81fabd8b..7f0bc9645610bf62a1087f294416443156470949 100644 (file)
--- a/yt_dlp/extractor/scrippsnetworks.py
+++ b/yt_dlp/extractor/scrippsnetworks.py
@@ -39,6 +39,7 @@ class ScrippsNetworksWatchIE(AWSIE):
             'skip_download': True,
         },
         'add_ie': [AnvatoIE.ie_key()],
+        'skip': '404 Not Found',
     }]
 
     _SNI_TABLE = {
@@ -113,6 +114,9 @@ class ScrippsNetworksIE(InfoExtractor):
             'timestamp': 1475678834,
             'upload_date': '20161005',
             'uploader': 'SCNI-SCND',
+            'duration': 29.995,
+            'chapters': [{'start_time': 0.0, 'end_time': 29.995, 'title': '<Untitled Chapter 1>'}],
+            'thumbnail': 'https://images.dds.discovery.com/up/tp/Scripps_-_Food_Category_Prod/122/987/0260338_630x355.jpg',
         },
         'add_ie': ['ThePlatform'],
         'expected_warnings': ['No HLS formats found'],

diff --git a/yt_dlp/extractor/syfy.py b/yt_dlp/extractor/syfy.py
index c79d27a0de5cac6efb738d035efb4090ca390091..afcdbf78043a143cd29cfd32c3a2d048366a3f19 100644 (file)
--- a/yt_dlp/extractor/syfy.py
+++ b/yt_dlp/extractor/syfy.py
@@ -23,6 +23,7 @@ class SyfyIE(AdobePassIE):
             'skip_download': True,
         },
         'add_ie': ['ThePlatform'],
+        'skip': 'Redirects to main page',
     }]
 
     def _real_extract(self, url):

diff --git a/yt_dlp/extractor/theplatform.py b/yt_dlp/extractor/theplatform.py
index 99caeb5f99ab4af689cfc0e5c9ac51fd0a855136..433ce8427c5dec84fda2447b841c2026e4f0a6c1 100644 (file)
--- a/yt_dlp/extractor/theplatform.py
+++ b/yt_dlp/extractor/theplatform.py
@@ -167,7 +167,7 @@ class ThePlatformIE(ThePlatformBaseIE, AdobePassIE):
             # rtmp download
             'skip_download': True,
         },
-        'skip': '404 Not Found',
+        'skip': 'CNet no longer uses ThePlatform',
     }, {
         'url': 'https://player.theplatform.com/p/D6x-PC/pulse_preview/embed/select/media/yMBg9E8KFxZD',
         'info_dict': {
@@ -177,7 +177,7 @@ class ThePlatformIE(ThePlatformBaseIE, AdobePassIE):
             'title': 'HIGHLIGHTS: USA bag first ever series Cup win',
             'uploader': 'EGSM',
         },
-        'skip': '404 Not Found',
+        'skip': 'Dead link',
     }, {
         'url': 'http://player.theplatform.com/p/NnzsPC/widget/select/media/4Y0TlYUr_ZT7',
         'only_matching': True,
@@ -195,7 +195,7 @@ class ThePlatformIE(ThePlatformBaseIE, AdobePassIE):
             'upload_date': '20150701',
             'uploader': 'NBCU-NEWS',
         },
-        'skip': '404 Not Found',
+        'skip': 'Error: Player PID "nbcNewsOffsite" is disabled',
     }, {
         # From http://www.nbc.com/the-blacklist/video/sir-crispin-crandall/2928790?onid=137781#vc137781=1
         # geo-restricted (US), HLS encrypted with AES-128

diff --git a/yt_dlp/extractor/theweatherchannel.py b/yt_dlp/extractor/theweatherchannel.py
index 682e4335d2c55fa1d6ab76b8ee5244257b887645..d1921e4f9a835215902125978ce0dbc94ad11914 100644 (file)
--- a/yt_dlp/extractor/theweatherchannel.py
+++ b/yt_dlp/extractor/theweatherchannel.py
@@ -11,17 +11,19 @@
 class TheWeatherChannelIE(ThePlatformIE):  # XXX: Do not subclass from concrete IE
     _VALID_URL = r'https?://(?:www\.)?weather\.com(?P<asset_name>(?:/(?P<locale>[a-z]{2}-[A-Z]{2}))?/(?:[^/]+/)*video/(?P<id>[^/?#]+))'
     _TESTS = [{
-        'url': 'https://weather.com/series/great-outdoors/video/ice-climber-is-in-for-a-shock',
-        'md5': 'c4cbe74c9c17c5676b704b950b73dd92',
+        'url': 'https://weather.com/storms/hurricane/video/invest-95l-in-atlantic-has-a-medium-chance-of-development',
+        'md5': '68f0cf616435683f27ce36bd9c927394',
         'info_dict': {
-            'id': 'cc82397e-cc3f-4d11-9390-a785add090e8',
+            'id': '81acef2d-ee8c-4545-ba83-bff3cc80db97',
             'ext': 'mp4',
-            'title': 'Ice Climber Is In For A Shock',
-            'description': 'md5:55606ce1378d4c72e6545e160c9d9695',
-            'uploader': 'TWC - Digital (No Distro)',
-            'uploader_id': '6ccd5455-16bb-46f2-9c57-ff858bb9f62c',
-            'upload_date': '20160720',
-            'timestamp': 1469018835,
+            'title': 'Invest 95L In Atlantic Has A Medium Chance Of Development',
+            'description': 'md5:0de720fd5f0d0e32207bd4c270fff824',
+            'uploader': 'TWC - Digital',
+            'uploader_id': 'b5a999e0-9e04-11e1-9ee2-001d092f5a10',
+            'upload_date': '20230721',
+            'timestamp': 1689967343,
+            'display_id': 'invest-95l-in-atlantic-has-a-medium-chance-of-development',
+            'duration': 34.0,
         }
     }, {
         'url': 'https://weather.com/en-CA/international/videos/video/unidentified-object-falls-from-sky-in-india',