how to play file type ".unknown_video" [Solved]

[dpaddy]

youtube-dl successfully downloads (what I suppose is a video) with filetype .unknown_video but mplayer can't play it. How do I get youtube-dl to convert to a playable format, or how do I convert -- after the download -- to a playable format, or how do I get mplayer to play type .unknown_video, or what player other than mplayer will play type .unknown_video?

[alamahant]

Try

Code: Select all

emerge -pv ffmpeg mplayer

[mike155]

What does 'file' tell you for your '.unknown_video' file?

Code: Select all

file <filename>

[dpaddy]

Yikes! I feel stupid. The file foo.unknown_video I downloaded via youtube-dl is only 892 bytes, so I suppose it is at best some sort of pointer to a video rather than the video itself, or at worst youtube-dl is simply not an appropriate tool to snag the video

Anyhow, since you asked:

Code: Select all

# emerge -pv ffmpeg mplayer 
 * Last emerge --sync was 160d 1h 3m 9s ago.

These are the packages that would be merged, in order:

Calculating dependencies... done!
[ebuild   R    ] media-video/ffmpeg-4.4-r1:0/56.58.58::gentoo  USE="X alsa amr amrenc bzip2 cdio cpudetection dav1d doc encode gmp gnutls gpl iconv jack libdrm mp3 network openh264 postproc rubberband theora threads vaapi vpx x264 x265 xvid zlib (-appkit) -bluray -bs2b -chromaprint -chromium -codec2 (-cuda) -debug -fdk -flite -fontconfig -frei0r -fribidi -gcrypt -gme -gsm -hardcoded-tables -iec61883 -ieee1394 -jpeg2k -kvazaar -ladspa -libaom -libaribb24 -libass -libcaca -libilbc -librtmp -libsoxr -libtesseract -libv4l -libxml2 -lv2 -lzma (-mipsdspr1) (-mipsdspr2) (-mipsfpu) (-mmal) -modplug -openal -opencl -opengl -openssl -opus -oss -pic -pulseaudio -rav1e -samba -sdl -snappy -sndio -speex -srt -ssh -static-libs -svg -test -truetype -twolame -v4l -vdpau -vidstab -vorbis -vulkan -webp -zeromq -zimg -zvbi" ABI_X86="(64) -32 (-x32)" CPU_FLAGS_X86="aes avx avx2 fma3 mmx mmxext sse sse2 sse3 sse4_1 sse4_2 ssse3 -3dnow -3dnowext -fma4 -xop" FFTOOLS="aviocat cws2fws ffescape ffeval ffhash fourcc2pixfmt graph2dot ismindex pktdumper qt-faststart sidxindex trasher" VIDEO_CARDS="nvidia" 0 KiB
[ebuild   R    ] media-video/mplayer-1.4-r1::gentoo  USE="X alsa bs2b cdio cdparanoia cpudetection dga dvd dvdnav enca encode iconv ipv6 jack jpeg libass mp3 network opengl osdmenu rtc shm truetype unicode xscreensaver xv xvid yuv4mpeg -a52 -aalib (-altivec) (-aqua) -bidi -bl -bluray -cddb -debug -doc -dts -dv -dvb -faac -faad -fbcon -ftp -ggi -gsm -joystick -ladspa -libcaca -libmpeg2 -lirc -live -lzo -mad -md5sum -mng -nas -openal -oss -png -pnm -pulseaudio -pvr -radio -rar -rtmp -samba -sdl (-selinux) -speex -tga -theora -toolame -tremor -twolame -v4l -vcd -vdpau (-vidix) -vorbis -x264 -xinerama -zoran" CPU_FLAGS_X86="mmx mmxext sse sse2 ssse3 -3dnow -3dnowext" VIDEO_CARDS="-mga" 0 KiB

Total: 2 packages (2 reinstalls), Size of downloads: 0 KiB

Moreover the build of ffmpeg mplayer succeeds, and subsequently

Code: Select all

$ mplayer foo.unknown_video 
MPlayer 1.4-9.3.0 (C) 2000-2019 MPlayer Team

Playing foo.unknown_video.
libavformat version 58.76.100 (external)
libavformat file format detected.
[lavf] stream 0: subtitle (text), -sid 0
LAVF: no audio or video headers found - broken file?
No stream found.


Exiting... (End of file)

Opening foo.unknown_video in an editor (emacs) reveals it to be meta data for a video (including a url) of the type

https:\/\/site\/videos\/directory_1\/directory_2\/directory_3\/video.mp4?validfrom=1644595194&validto=1645602394&rate=50000k&burst=50000k&hash=L3%2BBwTLIFA3UfTEYs2uqZ9LgkBw%3D

Attempting to get the indicated video with wget results in "Connection refused".

I conjecture that the "validfrom" and "validto" are timestamps and that "hash" is a hash of part of what follows "?" in the url. Any ideas concerning how to modify the url (perhaps timestamps and hash) so as to enable wget to succeed?

[alamahant]

It works fine on me
Plz try

Code: Select all

youtube-dl https://www.youtube.com/watch?v=EngW7tLk6R8

What is output?
You will need to enable "opus" in both ffmpeg and mplayer in adittion to any needed formats and codecs.

[dpaddy]

I've recompiled enabling "opus" in both ffmpeg and mplayer .

Sure, youtube-dl works on your url (namely https://www.youtube.com/watch?v=EngW7tLk6R8), but with my_url the behavior is as I've described. More precisely, executing youtube-dl with my_url first prints (to the xterminal)

[1] 24084
[2] 24085

and then immediately displays a command prompt ($) followed by four lines, namely

$ [website] 38971641: Downloading webpage
[website] 38971641: Downloading m3u8 information
[download] Destination: foo.unknown_video
[download] 100% of 892.00B in 00:00

whereupon I get no new command prompt - as I suppose it ($) was already printed before the four lines following it were output. By the way, "website" and "my_url" are place holders for the actual website and url.

Curiously enough, executing

youtube-dl my_url ; ls

results in the output of "ls" being sent to the xterminal before the four lines mentioned above. I suppose youtube-dl creates and puts two jobs in the background, evidenced by the output

[1] 24084
[2] 24085

and those jobs are responsible for the four lines of output mentioned above (which took place after "ls" had finished).

Anyhow, any ideas concerning how to modify the url (perhaps timestamps and hash) so as to enable wget to succeed?

[grknight]

I suggest quoting the URL on the command-line so the shell does not interpret any characters, e.g.

Code: Select all

youtube-dl 'https://foo.example/play?vid=abc123' ; ls

[dpaddy]

Quoting the URL on the command-line does make a difference, thanks for that. However, the resulting file contains (admittedly different) meta data, mplayer behaves as described above, and wget of the url in the metadata fails.

[alamahant]

but with my_url the behavior is as I've described

do you mind sharing your url?

[dpaddy]

I would rather not.

Anyhow, I conjecture that the time stamps are the problem so if I could make a bash script that executes wget soon enough then I might succeed.

Unfortunately experience has shown that I cannot write working bash scripts (except for trivial ones) to save my life I will later open a question about bash scripting in an attempt to get beyond my deep and profound confusion concerning scripting.

Thanks all!

[figueroa]

Add --restrict-filenames option, example:

Code: Select all

youtube-dl --restrict-filenames 'https://www.youtube.com/watch?v=i-x5SQmajnM'

The example is a "small" mp4 file. It should just work. Always quote the URL.

[Chiitoo]

Code: Select all

# emerge -pv ffmpeg mplayer 
 * Last emerge --sync was 160d 1h 3m 9s ago.

Is 'youtube-dl' more than 160 days old now as well? :]

While it does seem like 'youtube-dl' has picked up a lot more contributions as of late, a lot of places switched to the more active (at the time anyblue) 'yt-dlp', and I'd suggest giving that a go as well.

That, and I can find some '.unknown_video' issues on the youtube-dl issue tracker, so looking into those might be worthwhile as well (some of the replies from the developer at the time were simply along the lines of "unsupported source" or so).

[dpaddy]

Through trial and error I managed to get a bash script (with checking throughout to ensure against injection attacks) to fetch the metadata (via youtube-dl) and immediately execute wget on the file pointed to within the metadata (it pulled down a mp4 which mplayer played with no problem). Worked like a charm!

I would imagine that different sites may return different metadata, but perhaps this thread could nevertheless be of some use to others.

I appreciate all comments contributed. Thanks!

[Hu]

Code: Select all

# emerge -pv ffmpeg mplayer 
 * Last emerge --sync was 160d 1h 3m 9s ago.

As hinted at elsewhere, you should sync your tree and apply pending updates.

Opening foo.unknown_video in an editor (emacs)

Check whether emacs has a mode for playing video?

reveals it to be meta data for a video (including a url) of the type

https:\/\/site\/videos\/directory_1\/directory_2\/directory_3\/video.mp4?validfrom=1644595194&validto=1645602394&rate=50000k&burst=50000k&hash=L3%2BBwTLIFA3UfTEYs2uqZ9LgkBw%3D

Attempting to get the indicated video with wget results in "Connection refused".

That is weird. Even if the URL were to a path which the server refused to return, Connection refused should not happen. That should be reserved for "The server process is not running, so the peer kernel refused to accept the TCP connection."

I conjecture that the "validfrom" and "validto" are timestamps

You don't recognize the current time when printed as a Unix epoch? In this case:

Code: Select all

$ date -u -d @1644595194
Fri Feb 11 15:59:54 UTC 2022
$ date -u -d @1645602394
Wed Feb 23 07:46:34 UTC 2022

Any ideas concerning how to modify the url (perhaps timestamps and hash) so as to enable wget to succeed?

Assuming you recounted the wget error correctly, you need to point wget to a server that is actually listening. The path is irrelevant if the server will not listen to you. The full output of wget would be helpful here. Anonymize the URL and hostname if you must, but show all the other status data.

Quoting the URL on the command-line does make a difference, thanks for that.

Your shown URL, even anonymized, has significant shell metacharacters. Specifically, &, which was backgrounding the command. You ran the equivalent of:

Code: Select all

youtube-dl https:\/\/site\/videos\/directory_1\/directory_2\/directory_3\/video.mp4?validfrom=1644595194 &		# Placed into background.  Note the badly truncated URL.
validto=1645602394 &	# Backgrounded assignment.  This is useless, but probably harmless.
rate=50000k &	# Same.
burst=50000k &	# Same.
hash=L3%2BBwTLIFA3UfTEYs2uqZ9LgkBw%3D	# Foreground assignment.  Useless, but probably harmless.

I would rather not.

As you wish. However, it would be much easier for us to help you if we had a working example. It need not be the specific video you are trying to fetch, but having it come from the same server software would help.

Anyhow, I conjecture that the time stamps are the problem so if I could make a bash script that executes wget soon enough then I might succeed.

Maybe, but based on how I decode them, they appear to be valid for days at a time, so even at human reaction speeds, you should have gotten correct results when doing this by hand.

Through trial and error I managed to get a bash script

If you post the script, we could review it for you.

I would imagine that different sites may return different metadata, but perhaps this thread could nevertheless be of some use to others.

The thread would be of more use if you share the script, so that the next person does not need to reimplement it.