[cleanup] Fix infodict returned fields (#8906)

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

diff --git a/yt_dlp/extractor/archiveorg.py b/yt_dlp/extractor/archiveorg.py
index a0b26ac5a05e52e7e79beef45b10ab926f13a318..41f3a4ff2709e5a3468f0bf28cba98163e35d4b3 100644 (file)
--- a/yt_dlp/extractor/archiveorg.py
+++ b/yt_dlp/extractor/archiveorg.py
@@ -31,6 +31,7 @@
     unified_timestamp,
     url_or_none,
     urlhandle_detect_ext,
+    variadic,
 )
 
 
@@ -49,10 +50,9 @@ class ArchiveOrgIE(InfoExtractor):
             'release_date': '19681210',
             'timestamp': 1268695290,
             'upload_date': '20100315',
-            'creator': 'SRI International',
+            'creators': ['SRI International'],
             'uploader': 'laura@archive.org',
             'thumbnail': r're:https://archive\.org/download/.*\.jpg',
-            'release_year': 1968,
             'display_id': 'XD300-23_68HighlightsAResearchCntAugHumanIntellect.cdr',
             'track': 'XD300-23 68HighlightsAResearchCntAugHumanIntellect',
 
@@ -110,7 +110,7 @@ class ArchiveOrgIE(InfoExtractor):
             'title': 'Turning',
             'ext': 'flac',
             'track': 'Turning',
-            'creator': 'Grateful Dead',
+            'creators': ['Grateful Dead'],
             'display_id': 'gd1977-05-08d01t01.flac',
             'track_number': 1,
             'album': '1977-05-08 - Barton Hall - Cornell University',
@@ -130,11 +130,10 @@ class ArchiveOrgIE(InfoExtractor):
             'location': 'Barton Hall - Cornell University',
             'duration': 438.68,
             'track': 'Deal',
-            'creator': 'Grateful Dead',
+            'creators': ['Grateful Dead'],
             'album': '1977-05-08 - Barton Hall - Cornell University',
             'release_date': '19770508',
             'display_id': 'gd1977-05-08d01t07.flac',
-            'release_year': 1977,
             'track_number': 7,
         },
     }, {
@@ -169,7 +168,7 @@ class ArchiveOrgIE(InfoExtractor):
             'upload_date': '20160610',
             'description': 'md5:f70956a156645a658a0dc9513d9e78b7',
             'uploader': 'dimitrios@archive.org',
-            'creator': ['British Broadcasting Corporation', 'Time-Life Films'],
+            'creators': ['British Broadcasting Corporation', 'Time-Life Films'],
             'timestamp': 1465594947,
         },
         'playlist': [
@@ -259,7 +258,7 @@ def _real_extract(self, url):
             'title': m['title'],
             'description': clean_html(m.get('description')),
             'uploader': dict_get(m, ['uploader', 'adder']),
-            'creator': m.get('creator'),
+            'creators': traverse_obj(m, ('creator', {variadic}, {lambda x: x[0] and list(x)})),
             'license': m.get('licenseurl'),
             'release_date': unified_strdate(m.get('date')),
             'timestamp': unified_timestamp(dict_get(m, ['publicdate', 'addeddate'])),
@@ -274,7 +273,7 @@ def _real_extract(self, url):
                     'title': f.get('title') or f['name'],
                     'display_id': f['name'],
                     'description': clean_html(f.get('description')),
-                    'creator': f.get('creator'),
+                    'creators': traverse_obj(f, ('creator', {variadic}, {lambda x: x[0] and list(x)})),
                     'duration': parse_duration(f.get('length')),
                     'track_number': int_or_none(f.get('track')),
                     'album': f.get('album'),
@@ -302,7 +301,7 @@ def _real_extract(self, url):
             is_logged_in = bool(self._get_cookies('https://archive.org').get('logged-in-sig'))
             if extension in KNOWN_EXTENSIONS and (not f.get('private') or is_logged_in):
                 entry['formats'].append({
-                    'url': 'https://archive.org/download/' + identifier + '/' + f['name'],
+                    'url': 'https://archive.org/download/' + identifier + '/' + urllib.parse.quote(f['name']),
                     'format': f.get('format'),
                     'width': int_or_none(f.get('width')),
                     'height': int_or_none(f.get('height')),