-
Notifications
You must be signed in to change notification settings - Fork 72
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
implement looping, demuxer options #420
base: master
Are you sure you want to change the base?
implement looping, demuxer options #420
Conversation
Pretty straightforward, follows the same syntax as the output muxer options.
So -stream_loop doesn't work when provided as a demuxer option, as it's implemented in https://github.com/FFmpeg/FFMpeg/blob/master/fftools/ffmpeg_demux.c and we handle our own demuxing. Thankfully, this seemed to be relatively easy to port over using the preexisting work that we have for handling discontinuity. Lightly tested.
I love to see improvements like this, thanks @iameli ! The demuxer changes LGTM, it might be nice to have a test case exercising that. One scenario where I think it would be needed to specify the demuxer is for m3u8 files that don't have a standard extension - see the ffmpeg code (hit this during the ffmpeg 7 upgrade; see the diff for the TestTranscoder_FomatOptions testcase) The looping changes look good too but I am curious how it would behave with non-seekable inputs like most non-HTTP network protocols, pipes, etc. Other than that it would be nice to have tests for those too: check that the output is as expected etc for both seekable and non-seekable inputs. For non-seekable it would still be good to codify the behavior even if it is explicitly not supported. |
Thanks for taking a look @j0sh! I'm having trouble getting that test working, I've tried a couple environments...
|
Is this the most recent ffmpeg build / debug executable from
install_ffmpeg.sh ? The 7.0 upgrade resulted in a bunch of timestamp
adjustments to the CLI output and this looks like what the old version had
…On Sat, Aug 31, 2024 at 12:15 PM iameli-streams ***@***.***> wrote:
Thanks for taking a look @j0sh <https://github.com/j0sh>!
I'm having trouble getting that test working, I've tried a couple
environments...
***@***.***:/home/iameli/code/lpms# go test -coverprofile cover.out -run ^TestTranscoder_FormatOptions$ ./ffmpeg/... 2>&1 | tee test.log
? github.com/livepeer/lpms/ffmpeg/proto [no test files]
ERROR: encoder.c:220] Unable to guess output format : Invalid argument
ERROR: transcoder.c:238] Unable to open output : Invalid argument
E0831 12:15:07.179209 37135 ffmpeg.go:1015] Transcoder Return : Invalid argument
--- FAIL: TestTranscoder_FormatOptions (0.51s)
ffmpeg_test.go:1637: Invalid argument
ffmpeg_test.go:35: + ffprobe -loglevel warning -show_format test.flv
+ grep format_name=flv
test.flv: No such file or directory
ffmpeg_test.go:35: + ffprobe -loglevel warning -show_entries format=format_name,duration actually_mpegts.flv
+ ffprobe -loglevel warning -show_entries format=format_name,duration actually_mp4.flv
+ cat
+ diff -u actually_mpegts.out test.out
+ diff -u actually_mp4.out expected_mp4.out
--- actually_mp4.out 2024-08-31 12:15:07.459846342 -0700
+++ expected_mp4.out 2024-08-31 12:15:07.463846331 -0700
@@ -1,4 +1,4 @@
[FORMAT]
format_name=mov,mp4,m4a,3gp,3g2,mj2
-duration=8.033000
+duration=8.032667
[/FORMAT]
ffmpeg_test.go:35: + ffprobe -loglevel warning -show_entries format=format_name,duration actually_mpegts_2.flv
+ ffprobe -loglevel warning -show_entries format=format_name,duration actually_mp4_2.flv
+ diff -u actually_mpegts_2.out test.out
+ diff -u actually_mp4_2.out expected_mp4.out
--- actually_mp4_2.out 2024-08-31 12:15:07.570846055 -0700
+++ expected_mp4.out 2024-08-31 12:15:07.463846331 -0700
@@ -1,4 +1,4 @@
[FORMAT]
format_name=mov,mp4,m4a,3gp,3g2,mj2
-duration=8.033000
+duration=8.032667
[/FORMAT]
FAIL
github.com/livepeer/lpms/ffmpeg coverage: 39.3% of statements
FAIL github.com/livepeer/lpms/ffmpeg 0.520s
FAIL
***@***.***:/home/iameli/code/lpms# cat test.log
? github.com/livepeer/lpms/ffmpeg/proto [no test files]
ERROR: encoder.c:220] Unable to guess output format : Invalid argument
ERROR: transcoder.c:238] Unable to open output : Invalid argument
E0831 12:15:07.179209 37135 ffmpeg.go:1015] Transcoder Return : Invalid argument
--- FAIL: TestTranscoder_FormatOptions (0.51s)
ffmpeg_test.go:1637: Invalid argument
ffmpeg_test.go:35: + ffprobe -loglevel warning -show_format test.flv
+ grep format_name=flv
test.flv: No such file or directory
ffmpeg_test.go:35: + ffprobe -loglevel warning -show_entries format=format_name,duration actually_mpegts.flv
+ ffprobe -loglevel warning -show_entries format=format_name,duration actually_mp4.flv
+ cat
+ diff -u actually_mpegts.out test.out
+ diff -u actually_mp4.out expected_mp4.out
--- actually_mp4.out 2024-08-31 12:15:07.459846342 -0700
+++ expected_mp4.out 2024-08-31 12:15:07.463846331 -0700
@@ -1,4 +1,4 @@
[FORMAT]
format_name=mov,mp4,m4a,3gp,3g2,mj2
-duration=8.033000
+duration=8.032667
[/FORMAT]
ffmpeg_test.go:35: + ffprobe -loglevel warning -show_entries format=format_name,duration actually_mpegts_2.flv
+ ffprobe -loglevel warning -show_entries format=format_name,duration actually_mp4_2.flv
+ diff -u actually_mpegts_2.out test.out
+ diff -u actually_mp4_2.out expected_mp4.out
--- actually_mp4_2.out 2024-08-31 12:15:07.570846055 -0700
+++ expected_mp4.out 2024-08-31 12:15:07.463846331 -0700
@@ -1,4 +1,4 @@
[FORMAT]
format_name=mov,mp4,m4a,3gp,3g2,mj2
-duration=8.033000
+duration=8.032667
[/FORMAT]
FAIL
github.com/livepeer/lpms/ffmpeg coverage: 39.3% of statements
FAIL github.com/livepeer/lpms/ffmpeg 0.520s
FAIL
***@***.***:/home/iameli/code/lpms#
—
Reply to this email directly, view it on GitHub
<#420 (comment)>, or
unsubscribe
<https://github.com/notifications/unsubscribe-auth/AACHNHR4MBZXAVBHI5WXGDDZUII6VAVCNFSM6AAAAABNJDA5TWVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMZDGMRTGAZDAOJRGM>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
So -stream_loop doesn't work when provided as a demuxer option, as it's implemented in https://github.com/FFmpeg/FFMpeg/blob/master/fftools/ffmpeg_demux.c and we handle our own demuxing. Thankfully, this seemed to be relatively easy to port over using the preexisting work that we have for handling discontinuity. Lightly tested.
Demuxer options: Pretty straightforward, follows the same syntax as the output muxer options.
My use case: I've already got a livestream being sliced up into a bunch of C2PA-signed MP4 files. But now I need to turn them back into an MKV stream. To do this I'm using a hack on top of ffmpeg's
concat
muxer, looks like this on the command line:And that concat file looks like this:
And those
latest.mp4
files get 301 redirected to the correct MP4 files. Works great!