mirror of https://github.com/yt-dlp/yt-dlp.git
parent
5c333d7496
commit
6c7274ecd2
|
@ -2663,7 +2663,6 @@ class YoutubeDL(object):
|
||||||
os.remove(encodeFilename(file))
|
os.remove(encodeFilename(file))
|
||||||
return None
|
return None
|
||||||
|
|
||||||
self.report_file_already_downloaded(existing_files[0])
|
|
||||||
info_dict['ext'] = os.path.splitext(existing_files[0])[1][1:]
|
info_dict['ext'] = os.path.splitext(existing_files[0])[1][1:]
|
||||||
return existing_files[0]
|
return existing_files[0]
|
||||||
|
|
||||||
|
@ -2718,7 +2717,7 @@ class YoutubeDL(object):
|
||||||
info_dict['protocol'] = _protocols.pop()
|
info_dict['protocol'] = _protocols.pop()
|
||||||
directly_mergable = FFmpegFD.can_merge_formats(info_dict)
|
directly_mergable = FFmpegFD.can_merge_formats(info_dict)
|
||||||
if dl_filename is not None:
|
if dl_filename is not None:
|
||||||
pass
|
self.report_file_already_downloaded(dl_filename)
|
||||||
elif (directly_mergable and get_suitable_downloader(
|
elif (directly_mergable and get_suitable_downloader(
|
||||||
info_dict, self.params, to_stdout=(temp_filename == '-')) == FFmpegFD):
|
info_dict, self.params, to_stdout=(temp_filename == '-')) == FFmpegFD):
|
||||||
info_dict['url'] = '\n'.join(f['url'] for f in requested_formats)
|
info_dict['url'] = '\n'.join(f['url'] for f in requested_formats)
|
||||||
|
@ -2770,9 +2769,13 @@ class YoutubeDL(object):
|
||||||
else:
|
else:
|
||||||
# Just a single file
|
# Just a single file
|
||||||
dl_filename = existing_file(full_filename, temp_filename)
|
dl_filename = existing_file(full_filename, temp_filename)
|
||||||
if dl_filename is None:
|
if dl_filename is None or dl_filename == temp_filename:
|
||||||
|
# dl_filename == temp_filename could mean that the file was partially downloaded with --no-part.
|
||||||
|
# So we should try to resume the download
|
||||||
success, real_download = self.dl(temp_filename, info_dict)
|
success, real_download = self.dl(temp_filename, info_dict)
|
||||||
info_dict['__real_download'] = real_download
|
info_dict['__real_download'] = real_download
|
||||||
|
else:
|
||||||
|
self.report_file_already_downloaded(dl_filename)
|
||||||
|
|
||||||
dl_filename = dl_filename or temp_filename
|
dl_filename = dl_filename or temp_filename
|
||||||
info_dict['__finaldir'] = os.path.dirname(os.path.abspath(encodeFilename(full_filename)))
|
info_dict['__finaldir'] = os.path.dirname(os.path.abspath(encodeFilename(full_filename)))
|
||||||
|
|
Loading…
Reference in New Issue