Handling HLS format errors

[pythonsan]

Hello, I encountered an error when using this tool to process files in hls format with the following message.

(base) ➜   ffmpeg -i video.mp4 -f hls -master_pl_name replay.m3u8 -c:a aac -ab 192k \
  -var_stream_map "a:0,agroup:aud v:0,agroup:aud" \
  -hls_playlist_type vod -hls_list_size 0 -hls_time 10 vs%v/out.m3u8

(base) ➜   audiowmark hls-prepare vs0 vs0prep out.m3u8 video.mp4
AAC Bitrate:  55000 (detected)
Segments:     7
audiowmark: hls: ff_decode failed: Format not recognised.

At the same time, I executed the test script in the source directory and it prompted the same error, is there something wrong with my operation? Or is HLS not supported?

-> test/hls-test.sh
...
...
...
==== audiowmark hls-prepare hls-test-dir.30950/as0 hls-test-dir.30950/as0prep out.m3u8 hls-test-dir.30950/test-input.wav ====
audiowmark: failed to load audio master: hls-test-dir.30950/test-input.wav
./hls-test.sh: failed to run audiowmark hls-prepare hls-test-dir.30950/as0 hls-test-dir.30950/as0prep out.m3u8 hls-test-dir.30950/test-input.wav

MacOS 10.15.7, ffmpeg version 5.0

ffmpeg version 5.0 Copyright (c) 2000-2022 the FFmpeg developers
  built with Apple clang version 12.0.0 (clang-1200.0.32.29)
  configuration: --prefix=/usr/local/Cellar/ffmpeg/5.0 --enable-shared --enable-pthreads --enable-version3 --cc=clang --host-cflags= --host-ldflags= --enable-ffplay --enable-gnutls --enable-gpl --enable-libaom --enable-libbluray 

--enable-libdav1d --enable-libmp3lame --enable-libopus --enable-librav1e --enable-librist --enable-librubberband --enable-libsnappy --enable-libsrt --enable-libtesseract --enable-libtheora --enable-libvidstab --enable-libvmaf 

--enable-libvorbis --enable-libvpx --enable-libwebp --enable-libx264 --enable-libx265 --enable-libxml2 --enable-libxvid --enable-lzma --enable-libfontconfig --enable-libfreetype --enable-frei0r --enable-libass 
--enable-libopencore-amrnb --enable-libopencore-amrwb --enable-libopenjpeg --enable-libspeex --enable-libsoxr --enable-libzmq --enable-libzimg 
--disable-libjack --disable-indev=jack --enable-videotoolbox

audiowmark build from git source - newest
video.mp4 info:

h264 1920x1080 25fps 1024Kbps, aac 44.1Khz, 2ch, 32Kpbs

[swesterfeld]

I can reproduce the problem. This issue seems to be macOS specific, caused by the way audiowmark creates its temporary files and uses them when calling ffmpeg as subprocess. The code is a bit tricky because it tries to avoid leaking the temp file if audiowmark dies by deleting the file after creation, which could be the problem. There is a similar problem here:

#7 (comment)

On Linux everything works fine.

[pythonsan]

I replaced the temporary file generation and discarded the use of PIPE and the error was gone.

So I have a suggestion:
if it is a MACOS system, use the value of the TEMPDIR environment variable as the temporary file storage directory, and store the result data after execp execution of ffmpeg command in the temporary file in the temporary directory, and delete it after using this data. Even if an abnormal program exit occurs at this time, the file is still a temporary file and will be automatically cleared after the system restarts.

[pythonsan]

Of course, the final production environment will most likely be a linux system, so this issue is not a very high priority

[swesterfeld]

if it is a MACOS system, use the value of the TEMPDIR environment variable as the temporary file storage directory

At the moment I am not fully convinced that it is impossible with the macOS API to get something that is equivalent to Linux. We have file descriptors on macOS. We can pass these to sub processes. We have /dev/fd/N. So maybe there is some way to get the same behaviour we have on Linux.

That said, I'd probably merge a PR that does different temp file handling on macOS like you suggest, as long as it is cleanly abstracted (for instance similar to pythons NamedTemporaryFile) and doesn't change behaviour on Linux.