[test:download] Validate and sort info dict fields (#12299)

Authored by: pzhlkj6612, bashonly

Co-authored-by: bashonly <88596187+bashonly@users.noreply.github.com>

test/helper.py

@@ -237,6 +237,20 @@ def sanitize_got_info_dict(got_dict):
 
 
 def expect_info_dict(self, got_dict, expected_dict):
+    ALLOWED_KEYS_SORT_ORDER = (
+        # NB: Keep in sync with the docstring of extractor/common.py
+        'id', 'ext', 'direct', 'display_id', 'title', 'alt_title', 'description', 'media_type',
+        'uploader', 'uploader_id', 'uploader_url', 'channel', 'channel_id', 'channel_url', 'channel_is_verified',
+        'channel_follower_count', 'comment_count', 'view_count', 'concurrent_view_count',
+        'like_count', 'dislike_count', 'repost_count', 'average_rating', 'age_limit', 'duration', 'thumbnail', 'heatmap',
+        'chapters', 'chapter', 'chapter_number', 'chapter_id', 'start_time', 'end_time', 'section_start', 'section_end',
+        'categories', 'tags', 'cast', 'composers', 'artists', 'album_artists', 'creators', 'genres',
+        'track', 'track_number', 'track_id', 'album', 'album_type', 'disc_number',
+        'series', 'series_id', 'season', 'season_number', 'season_id', 'episode', 'episode_number', 'episode_id',
+        'timestamp', 'upload_date', 'release_timestamp', 'release_date', 'release_year', 'modified_timestamp', 'modified_date',
+        'playable_in_embed', 'availability', 'live_status', 'location', 'license', '_old_archive_ids',
+    )
+
     expect_dict(self, got_dict, expected_dict)
     # Check for the presence of mandatory fields
     if got_dict.get('_type') not in ('playlist', 'multi_video'):
@@ -252,7 +266,13 @@ def expect_info_dict(self, got_dict, expected_dict):
 
     test_info_dict = sanitize_got_info_dict(got_dict)
 
-    missing_keys = set(test_info_dict.keys()) - set(expected_dict.keys())
+    # Check for invalid/misspelled field names being returned by the extractor
+    invalid_keys = sorted(test_info_dict.keys() - ALLOWED_KEYS_SORT_ORDER)
+    self.assertFalse(invalid_keys, f'Invalid fields returned by the extractor: {", ".join(invalid_keys)}')
+
+    missing_keys = sorted(
+        test_info_dict.keys() - expected_dict.keys(),
+        key=lambda x: ALLOWED_KEYS_SORT_ORDER.index(x))
     if missing_keys:
         def _repr(v):
             if isinstance(v, str):