Improve --clean-infojson

It should not removes fields that may be needed for `--load-infojson`.
Eg: `_ffmpeg_args`, `_has_drm`

test/test_YoutubeDL.py

@@ -661,7 +661,7 @@ def test_add_extra_info(self):
         'duration': 100000,
         'playlist_index': 1,
         'playlist_autonumber': 2,
-        '_last_playlist_index': 100,
+        '__last_playlist_index': 100,
         'n_entries': 10,
         'formats': [{'id': 'id 1'}, {'id': 'id 2'}, {'id': 'id 3'}]
     }

yt_dlp/YoutubeDL.py

@@ -954,7 +954,7 @@ def report_file_delete(self, file_name):
             self.to_screen('Deleting existing file')
 
     def raise_no_formats(self, info, forced=False, *, msg=None):
-        has_drm = info.get('__has_drm')
+        has_drm = info.get('_has_drm')
         ignored, expected = self.params.get('ignore_no_formats_error'), bool(msg)
         msg = msg or has_drm and 'This video is DRM protected' or 'No video formats found!'
         if forced or not ignored:
@@ -1052,7 +1052,7 @@ def prepare_outtmpl(self, outtmpl, info_dict, sanitize=False):
         # For fields playlist_index, playlist_autonumber and autonumber convert all occurrences
         # of %(field)s to %(field)0Nd for backward compatibility
         field_size_compat_map = {
-            'playlist_index': number_of_digits(info_dict.get('_last_playlist_index') or 0),
+            'playlist_index': number_of_digits(info_dict.get('__last_playlist_index') or 0),
             'playlist_autonumber': number_of_digits(info_dict.get('n_entries') or 0),
             'autonumber': self.params.get('autonumber_size') or 5,
         }
@@ -1764,7 +1764,7 @@ def get_entry(i):
                 entry['__x_forwarded_for_ip'] = x_forwarded_for
             extra = {
                 'n_entries': n_entries,
-                '_last_playlist_index': max(playlistitems) if playlistitems else (playlistend or n_entries),
+                '__last_playlist_index': max(playlistitems) if playlistitems else (playlistend or n_entries),
                 'playlist_count': ie_result.get('playlist_count'),
                 'playlist_index': playlist_index,
                 'playlist_autonumber': i,
@@ -2436,10 +2436,11 @@ def sanitize_numeric_fields(info):
         else:
             formats = info_dict['formats']
 
-        info_dict['__has_drm'] = any(f.get('has_drm') for f in formats)
+        # or None ensures --clean-infojson removes it
+        info_dict['_has_drm'] = any(f.get('has_drm') for f in formats) or None
         if not self.params.get('allow_unplayable_formats'):
             formats = [f for f in formats if not f.get('has_drm')]
-            if info_dict['__has_drm'] and all(
+            if info_dict['_has_drm'] and all(
                     f.get('acodec') == f.get('vcodec') == 'none' for f in formats):
                 self.report_warning(
                     'This video is DRM protected and only images are available for download. '
@@ -3266,9 +3267,9 @@ def sanitize_info(info_dict, remove_private_keys=False):
         info_dict.setdefault('_type', 'video')
 
         if remove_private_keys:
-            reject = lambda k, v: v is None or (k.startswith('_') and k != '_type') or k in {
+            reject = lambda k, v: v is None or k.startswith('__') or k in {
                 'requested_downloads', 'requested_formats', 'requested_subtitles', 'requested_entries',
-                'entries', 'filepath', 'infojson_filename', 'original_url', 'playlist_autonumber',
+                'entries', 'filepath', '_filename', 'infojson_filename', 'original_url', 'playlist_autonumber',
             }
         else:
             reject = lambda k, v: False

yt_dlp/downloader/external.py

@@ -20,6 +20,7 @@
     encodeFilename,
     handle_youtubedl_headers,
     remove_end,
+    traverse_obj,
 )
 
 
@@ -363,9 +364,11 @@ def _call_downloader(self, tmpfilename, info_dict):
         if not self.params.get('verbose'):
             args += ['-hide_banner']
 
-        args += info_dict.get('_ffmpeg_args', [])
+        args += traverse_obj(info_dict, ('downloader_options', 'ffmpeg_args'), default=[])
 
-        # This option exists only for compatibility. Extractors should use `_ffmpeg_args` instead
+        # These exists only for compatibility. Extractors should use
+        # info_dict['downloader_options']['ffmpeg_args'] instead
+        args += info_dict.get('_ffmpeg_args')
         seekable = info_dict.get('_seekable')
         if seekable is not None:
             # setting -seekable prevents ffmpeg from guessing if the server

yt_dlp/extractor/common.py

@@ -208,8 +208,10 @@ class InfoExtractor:
                     * no_resume  The server does not support resuming the
                                  (HTTP or RTMP) download. Boolean.
                     * has_drm    The format has DRM and cannot be downloaded. Boolean
-                    * downloader_options  A dictionary of downloader options as
-                                 described in FileDownloader (For internal use only)
+                    * downloader_options  A dictionary of downloader options
+                                 (For internal use only)
+                                 * http_chunk_size Chunk size for HTTP downloads
+                                 * ffmpeg_args     Extra arguments for ffmpeg downloader
                     RTMP formats can also have the additional fields: page_url,
                     app, play_path, tc_url, flash_version, rtmp_live, rtmp_conn,
                     rtmp_protocol, rtmp_real_time

yt_dlp/extractor/nbc.py

@@ -579,7 +579,7 @@ def _real_extract(self, url):
         for f in formats:
             # -http_seekable requires ffmpeg 4.3+ but it doesnt seem possible to
             # download with ffmpeg without this option
-            f['_ffmpeg_args'] = ['-seekable', '0', '-http_seekable', '0', '-icy', '0']
+            f['downloader_options'] = {'ffmpeg_args': ['-seekable', '0', '-http_seekable', '0', '-icy', '0']}
         self._sort_formats(formats)
 
         return {

yt_dlp/extractor/radiko.py

@@ -125,7 +125,7 @@ def _extract_formats(self, video_id, station, is_onair, ft, cursor, auth_token,
                     # Prioritize live radio vs playback based on extractor
                     sf['preference'] = 100 if is_onair else -100
                 if not is_onair and url_attrib['timefree'] == '1' and time_to_skip:
-                    sf['_ffmpeg_args'] = ['-ss', time_to_skip]
+                    sf['downloader_options'] = {'ffmpeg_args': ['-ss', time_to_skip]}
             formats.extend(subformats)
 
         self._sort_formats(formats)

yt_dlp/extractor/turner.py

@@ -141,7 +141,7 @@ def _extract_cvp_info(self, data_src, video_id, path_data={}, ap_data={}, fatal=
                     m3u8_id=format_id or 'hls', fatal=False)
                 if '/secure/' in video_url and '?hdnea=' in video_url:
                     for f in m3u8_formats:
-                        f['_ffmpeg_args'] = ['-seekable', '0']
+                        f['downloader_options'] = {'ffmpeg_args': ['-seekable', '0']}
                 formats.extend(m3u8_formats)
             elif ext == 'f4m':
                 formats.extend(self._extract_f4m_formats(