Discard info_dict from memory if no longer needed

Closes #1399

diff --git a/yt_dlp/YoutubeDL.py b/yt_dlp/YoutubeDL.py
index 85219ac95cae627f95dacb9297d3d6f06e16188b..7e9c0949b2a7cb043a04007a7433e3b5a6e80b8a 100644 (file)
--- a/yt_dlp/YoutubeDL.py
+++ b/yt_dlp/YoutubeDL.py
@@ -319,9 +319,14 @@ class YoutubeDL:
     default_search:    Prepend this string if an input url is not valid.
                        'auto' for elaborate guessing
     encoding:          Use this encoding instead of the system-specified.
-    extract_flat:      Do not resolve URLs, return the immediate result.
-                       Pass in 'in_playlist' to only show this behavior for
-                       playlist items.
+    extract_flat:      Whether to resolve and process url_results further
+                       * False:     Always process (default)
+                       * True:      Never process
+                       * 'in_playlist': Do not process inside playlist/multi_video
+                       * 'discard': Always process, but don't return the result
+                                    from inside playlist/multi_video
+                       * 'discard_in_playlist': Same as "discard", but only for
+                                    playlists (not multi_video)
     wait_for_video:    If given, wait for scheduled streams to become available.
                        The value should be a tuple containing the range
                        (min_secs, max_secs) to wait between retries
@@ -1725,6 +1730,12 @@ def __process_playlist(self, ie_result, download):
         self.to_screen(f'[{ie_result["extractor"]}] Playlist {title}: Downloading {n_entries} videos'
                        f'{format_field(ie_result, "playlist_count", " of %s")}')
 
+        keep_resolved_entries = self.params.get('extract_flat') != 'discard'
+        if self.params.get('extract_flat') == 'discard_in_playlist':
+            keep_resolved_entries = ie_result['_type'] != 'playlist'
+        if keep_resolved_entries:
+            self.write_debug('The information of all playlist entries will be held in memory')
+
         failures = 0
         max_failures = self.params.get('skip_playlist_after_errors') or float('inf')
         for i, (playlist_index, entry) in enumerate(entries):
@@ -1765,7 +1776,8 @@ def __process_playlist(self, ie_result, download):
                 self.report_error(
                     f'Skipping the remaining entries in playlist "{title}" since {failures} items failed extraction')
                 break
-            resolved_entries[i] = (playlist_index, entry_result)
+            if keep_resolved_entries:
+                resolved_entries[i] = (playlist_index, entry_result)
 
         # Update with processed data
         ie_result['requested_entries'], ie_result['entries'] = tuple(zip(*resolved_entries)) or ([], [])

diff --git a/yt_dlp/__init__.py b/yt_dlp/__init__.py
index fd44e1ab9636e18daed58d69e8c526db58f90224..7caf41c60707a97fb9b28e1378fbff732b4dbba8 100644 (file)
--- a/yt_dlp/__init__.py
+++ b/yt_dlp/__init__.py
@@ -688,6 +688,21 @@ def parse_options(argv=None):
         'getformat', 'getid', 'getthumbnail', 'gettitle', 'geturl'
     ))
 
+    playlist_pps = [pp for pp in postprocessors if pp.get('when') == 'playlist']
+    write_playlist_infojson = (opts.writeinfojson and not opts.clean_infojson
+                               and opts.allow_playlist_files and opts.outtmpl.get('pl_infojson') != '')
+    if not any((
+        opts.extract_flat,
+        opts.dump_single_json,
+        opts.forceprint.get('playlist'),
+        opts.print_to_file.get('playlist'),
+        write_playlist_infojson,
+    )):
+        if not playlist_pps:
+            opts.extract_flat = 'discard'
+        elif playlist_pps == [{'key': 'FFmpegConcat', 'only_multi_video': True, 'when': 'playlist'}]:
+            opts.extract_flat = 'discard_in_playlist'
+
     final_ext = (
         opts.recodevideo if opts.recodevideo in FFmpegVideoConvertorPP.SUPPORTED_EXTS
         else opts.remuxvideo if opts.remuxvideo in FFmpegVideoRemuxerPP.SUPPORTED_EXTS