mirror of
https://github.com/yt-dlp/yt-dlp.git
synced 2024-10-31 07:50:11 +01:00
Add key requested_downloads
in the root info_dict
This commit is contained in:
parent
9c906919ae
commit
f46e2f9d92
@ -30,6 +30,7 @@ def __init__(self, *args, **kwargs):
|
||||
self.msgs = []
|
||||
|
||||
def process_info(self, info_dict):
|
||||
info_dict = info_dict.copy()
|
||||
info_dict.pop('__original_infodict', None)
|
||||
self.downloaded_info_dicts.append(info_dict)
|
||||
|
||||
@ -908,7 +909,7 @@ def process_info(self, info_dict):
|
||||
def _match_entry(self, info_dict, incomplete=False):
|
||||
res = super(FilterYDL, self)._match_entry(info_dict, incomplete)
|
||||
if res is None:
|
||||
self.downloaded_info_dicts.append(info_dict)
|
||||
self.downloaded_info_dicts.append(info_dict.copy())
|
||||
return res
|
||||
|
||||
first = {
|
||||
|
@ -53,7 +53,7 @@ def report_warning(self, message):
|
||||
raise ExtractorError(message)
|
||||
|
||||
def process_info(self, info_dict):
|
||||
self.processed_info_dicts.append(info_dict)
|
||||
self.processed_info_dicts.append(info_dict.copy())
|
||||
return super(YoutubeDL, self).process_info(info_dict)
|
||||
|
||||
|
||||
|
@ -2552,6 +2552,8 @@ def is_wellformed(f):
|
||||
continue
|
||||
break
|
||||
|
||||
info_dict['requested_downloads'] = formats_to_download
|
||||
best_format = formats_to_download[-1] if formats_to_download else {}
|
||||
if not formats_to_download:
|
||||
if not self.params.get('ignore_no_formats_error'):
|
||||
raise ExtractorError('Requested format is not available', expected=True,
|
||||
@ -2565,19 +2567,24 @@ def is_wellformed(f):
|
||||
'[info] %s: Downloading %d format(s): %s' % (
|
||||
info_dict['id'], len(formats_to_download),
|
||||
", ".join([f['format_id'] for f in formats_to_download])))
|
||||
for fmt in formats_to_download:
|
||||
new_info = dict(info_dict)
|
||||
for i, fmt in enumerate(formats_to_download):
|
||||
formats_to_download[i] = new_info = dict(info_dict)
|
||||
# Save a reference to the original info_dict so that it can be modified in process_info if needed
|
||||
new_info['__original_infodict'] = info_dict
|
||||
new_info.update(fmt)
|
||||
new_info['__original_infodict'] = info_dict
|
||||
self.process_info(new_info)
|
||||
new_info.pop('__original_infodict')
|
||||
# Remove copied info
|
||||
for key, val in tuple(new_info.items()):
|
||||
if info_dict.get(key) == val:
|
||||
new_info.pop(key)
|
||||
|
||||
for pp in self._pps['after_video']:
|
||||
info_dict = self.run_pp(pp, info_dict)
|
||||
|
||||
# We update the info dict with the selected best quality format (backwards compatibility)
|
||||
if formats_to_download:
|
||||
info_dict.update(formats_to_download[-1])
|
||||
info_dict.update(best_format)
|
||||
return info_dict
|
||||
|
||||
def process_subtitles(self, video_id, normal_subtitles, automatic_captions):
|
||||
@ -2730,9 +2737,10 @@ def dl(self, name, info, subtitle=False, test=False):
|
||||
return fd.download(name, new_info, subtitle)
|
||||
|
||||
def process_info(self, info_dict):
|
||||
"""Process a single resolved IE result."""
|
||||
"""Process a single resolved IE result. (Modified it in-place)"""
|
||||
|
||||
assert info_dict.get('_type', 'video') == 'video'
|
||||
original_infodict = info_dict
|
||||
|
||||
max_downloads = self.params.get('max_downloads')
|
||||
if max_downloads is not None:
|
||||
@ -2857,8 +2865,16 @@ def _write_link_file(link_type):
|
||||
for link_type, should_write in write_links.items()):
|
||||
return
|
||||
|
||||
def replace_info_dict(new_info):
|
||||
nonlocal info_dict
|
||||
if new_info == info_dict:
|
||||
return
|
||||
info_dict.clear()
|
||||
info_dict.update(new_info)
|
||||
|
||||
try:
|
||||
info_dict, files_to_move = self.pre_process(info_dict, 'before_dl', files_to_move)
|
||||
new_info, files_to_move = self.pre_process(info_dict, 'before_dl', files_to_move)
|
||||
replace_info_dict(new_info)
|
||||
except PostProcessingError as err:
|
||||
self.report_error('Preprocessing: %s' % str(err))
|
||||
return
|
||||
@ -2868,7 +2884,7 @@ def _write_link_file(link_type):
|
||||
info_dict['filepath'] = temp_filename
|
||||
info_dict['__finaldir'] = os.path.dirname(os.path.abspath(encodeFilename(full_filename)))
|
||||
info_dict['__files_to_move'] = files_to_move
|
||||
info_dict = self.run_pp(MoveFilesAfterDownloadPP(self, False), info_dict)
|
||||
replace_info_dict(self.run_pp(MoveFilesAfterDownloadPP(self, False), info_dict))
|
||||
else:
|
||||
# Download
|
||||
info_dict.setdefault('__postprocessors', [])
|
||||
@ -3087,7 +3103,7 @@ def ffmpeg_fixup(cndn, msg, cls):
|
||||
|
||||
fixup()
|
||||
try:
|
||||
info_dict = self.post_process(dl_filename, info_dict, files_to_move)
|
||||
replace_info_dict(self.post_process(dl_filename, info_dict, files_to_move))
|
||||
except PostProcessingError as err:
|
||||
self.report_error('Postprocessing: %s' % str(err))
|
||||
return
|
||||
@ -3101,6 +3117,7 @@ def ffmpeg_fixup(cndn, msg, cls):
|
||||
|
||||
if must_record_download_archive or self.params.get('force_write_download_archive', False):
|
||||
self.record_download_archive(info_dict)
|
||||
assert info_dict is original_infodict
|
||||
max_downloads = self.params.get('max_downloads')
|
||||
if max_downloads is not None and self._num_downloads >= int(max_downloads):
|
||||
raise MaxDownloadsReached()
|
||||
@ -3170,8 +3187,8 @@ def sanitize_info(info_dict, remove_private_keys=False):
|
||||
keep_keys = ['_type'] # Always keep this to facilitate load-info-json
|
||||
if remove_private_keys:
|
||||
remove_keys |= {
|
||||
'requested_formats', 'requested_subtitles', 'requested_entries', 'entries',
|
||||
'filepath', 'infojson_filename', 'original_url', 'playlist_autonumber',
|
||||
'requested_downloads', 'requested_formats', 'requested_subtitles', 'requested_entries',
|
||||
'entries', 'filepath', 'infojson_filename', 'original_url', 'playlist_autonumber',
|
||||
}
|
||||
reject = lambda k, v: k not in keep_keys and (
|
||||
k.startswith('_') or k in remove_keys or v is None)
|
||||
@ -3250,13 +3267,12 @@ def pre_process(self, ie_info, key='pre_process', files_to_move=None):
|
||||
info = self.run_pp(pp, info)
|
||||
return info, info.pop('__files_to_move', None)
|
||||
|
||||
def post_process(self, filename, ie_info, files_to_move=None):
|
||||
def post_process(self, filename, info, files_to_move=None):
|
||||
"""Run all the postprocessors on the given file."""
|
||||
info = dict(ie_info)
|
||||
info['filepath'] = filename
|
||||
info['__files_to_move'] = files_to_move or {}
|
||||
|
||||
for pp in ie_info.get('__postprocessors', []) + self._pps['post_process']:
|
||||
for pp in info.get('__postprocessors', []) + self._pps['post_process']:
|
||||
info = self.run_pp(pp, info)
|
||||
info = self.run_pp(MoveFilesAfterDownloadPP(self), info)
|
||||
del info['__files_to_move']
|
||||
|
@ -19,6 +19,9 @@ def run(self, info):
|
||||
elif info.get('filepath'): # PP was called after download (default)
|
||||
filepath = info.get('filepath')
|
||||
self.to_screen(f'Post-processed {filepath!r} with {self._kwargs}')
|
||||
elif info.get('requested_downloads'): # PP was called after_video
|
||||
filepaths = [f.get('filepath') for f in info.get('requested_downloads')]
|
||||
self.to_screen(f'Post-processed {filepaths!r} with {self._kwargs}')
|
||||
else: # PP was called before actual download
|
||||
filepath = info.get('_filename')
|
||||
self.to_screen(f'Pre-processed {filepath!r} with {self._kwargs}')
|
||||
|
Loading…
Reference in New Issue
Block a user