--get-filename broken for webm videos with extracting audio option

[Clevero]

I think it's a bug, so i write this issue. I also did not find any issues related to this.

Getting the filename of the file seems to be broken when your video is in webm format and you want just the audio.

Example with "youtube-dl $URL -x --get-filename":
(url -> proposed filename -> saved file)

//got the right extension
https://www.youtube.com/watch?v=YSAqTdc-Y2g -> *.m4a -> *.m4a

//.opus file was written to disk instead of .webm file
https://www.youtube.com/watch?v=2TCDiK7GpNM -> *.webm -> *.opus

I found this bug (?) in a script written by me. It downloads the audio with "-x" but if the written filename is different as the proposed filename with "-x --get-filename" the script cannot copy the file to an another destination, cause it doesn't exists.

[jaimeMF]

The --get-filename just prints the file where the audio is downloaded, before any conversion.

[Clevero]

so, the first video is downloaded in two slices, one just audio and one just video and then they are converted together?

and at .webm the whole video ist downloaded?

okay, mhm. and how can I get the file after the conversion?

Tried to look up the best audio format with "-F" it shows me .webm (audio only is the best audio only format) but .opus is at the end saved to disk.

should I replace the given extension here from .webm to .opus, cause at any time at any video, webm (audio only) is converted to .opus?

[yan12125]

how can I get the file after the conversion

Currently there's no generic approach.

should I replace the given extension here from .webm to .opus, cause at any time at any video, webm (audio only) is converted to .opus?

Just to remind: WebM may contain Vorbis audio, too.

[ghost]

I've found a side effect to this which looks like an actual bug: resuming a playlist download with -x will skip existing .m4a files correctly, but always redownloads the Opus-format .webm files even if they're already on disk.

[yan12125]

After thinking again I believe it's impossible to get reliable filename without actual downloading. FFmpegExtractAudioPP uses ffprobe to determine the output file extension. This semantic needs real files.

If anyone else have a good idea, feel free to post it.

[yan12125]

Let's continue the discussion in #5710.

[yan12125]

Sorry, they are not the same bug.

[dcht00]

This is still a common problem.

For a yt download that turns out to be a .webm,

--get-filename incorectly gives:

XYZ.mp4

However when loading:

[youtube] Downloading webpage
[youtube] Downloading video info webpage
[...]
[download] Destination: XYZ.webm <--------- CORRECT
[download] Download completed

[byteshiva]

The below command would give the filename after download.

youtube-dl -x --quiet https://www.youtube.com/watch?v=VGYLCXJhJnI -f bestaudio --exec 'echo {}'