Compare commits

..
226 Commits
Author SHA1 Message Date
wm4 ca8dc9147b Release 0.3.9 2014-04-30 11:47:07 +02:00
wm4 df8b99490c demux: add hack to get initial ICY info to be printed
Newly added metadata (such as the ICY title, sent some seconds after
opening the stream) simply wasn't printed.

This problem doesn't exist in git master.

Fixes #753.
2014-04-30 11:46:40 +02:00
wm4 c78142a973 TOOLS/umpv: make it work with Python 3 (again)
Apparently, the 3rd (2nd) parameter to string.translate() function was
removed.

Also, make_abs() had a mistake - not sure how this passed testing.
2014-04-30 11:45:32 +02:00
wm4 a8e6caf56b TOOLS/umpv: don't mangle URLs
This attempted to prefix the current directory to URLs, because it
didn't recognize them as already absolute paths.
2014-04-30 11:45:32 +02:00
wm4 e6cca1e43b TOOLS/umpv: allow passing 0 files
Some might use that to just create an empty window with --force-window
--idle (for whatever reasons).
2014-04-30 11:45:32 +02:00
wm4 d7538d3597 TOOLS/umpv: allow passing options
But only via a special environment variable.
2014-04-30 11:45:32 +02:00
wm4 f3eb5b143c TOOLS/umpv: make it work with Python 3
Apparently they removed octal literals, and made them invalid syntax.
2014-04-30 11:45:32 +02:00
Stefano Pigozzi 9993fd8888 build: use -pthread when available
Our code currently tries to link -lpthread and adds stuff like -D_REENTRANT
based on the target platform.

GCC actually supports to just pass a -pthread compiler and linker flag that
will automatically enable threading and define the correct symbols for the
platform, so let's try to just use that as our first choice.

clang also supports -pthread but it must be used only as a compiler flag,
so we also take care of that scenario with this commit.
2014-04-30 11:45:31 +02:00
wm4 d9c3a51d87 terminal: always use SA_RESTART with sigaction()
One problem is that for example stdio functions won't restart syscalls
manually, and instead treat EINTR as an error. So passing SA_RESTART is
the only sane thing to do, unless you have special requirements, which
we don't.
2014-04-30 11:45:31 +02:00
wm4 619fa761af demux_mkv: enable parsing for VP9
VP9 packets can contain 2 frames in some video packets (from which 1
frame is invisible). Due to a design mismatch between libvpx and the
libavcodec vp9 decoder, libvpx can take the "full" packets, but lavc vp9
can not. The consequence is that we have to split the packets if we want
to feed them to the lavc codec.

This is not entirely correct yet: timestamp handling is missing.
--demuxer=lavf and ffmpeg native utilities have the same problem. We can
fix this only once the ffmpeg VP9 parser is fixed.
2014-04-30 11:45:31 +02:00
wm4 717d729dcb osd: make code C99
This used an unnamed union, which is allowed in GNU C and C11, but not
C99. This broke the build with some older compilers.

Replaces pull request #744.
2014-04-30 11:45:31 +02:00
Alexander Preisinger 4eb5b62e5a Revert "wayland: print waylands display erros"
This reverts commit 6e34b0ec1f.

There has always been an error message "proxy already has a listener" and
I couldn't reproduce where it is comming from until now. The display interface
already has a listener and we can't overwrite it. Now remove the code and avoid
this error message.

Conflicts:
	video/out/wayland_common.c
2014-04-30 11:45:31 +02:00
wm4 b9e32d0d3d manpage: fix grammar 2014-04-30 11:45:31 +02:00
wm4 28e2b95f58 manpage: forced->used 2014-04-30 11:45:31 +02:00
wm4 491f633a20 stream: use uninterruptible sleep on reconnecting
This is the only function which actually used the time argument of
stream_check_interrupt(). Considering that the whole player freezes
anyway, this is not worth the complication.

Also generally reduce the maximum wait time due to timeout. Introduce
exponential backoff, which makes the first reconnect retries faster, but
still waits up to 500ms in the later retries.
2014-04-30 11:45:31 +02:00
wm4 eb5e45855a manpage: improve --osd-level description 2014-04-30 11:45:31 +02:00
wm4 59e55bb0e1 demux_mkv: enable parsing for mp3
For some reason, some files appear to have broken mp3 packets, or at
least in a form that libavcodec can't deal with. The audio in the sample
file in question could not be decoded using libavcodec.

The problematic file had variable packet sizes, and the libavcodec
decoder kept printing "mp3: Header missing" for each packet it was fed.
Remuxing with mkvmerge fixes the problem. The mp3 data is probably not
VBR, and remuxing resulted in fixed-size mp3 frames. So I don't know why
the sample file was muxed this way - it might just be incorrect.

The sample file had "libmkv 0.6.4" as MuxingApp (although I could not
get mkvinfo to print this element, maybe the file uses an incorrect
element ID), and "HandBrake 0.9.4" as WritingApp.

Note that the libmpg123 decoder does not have any issues with it. It's
probably more robust, because libmpg123 was made to decode whole mp3
files, not just single frames.

Fixes issue #742.
2014-04-30 11:45:31 +02:00
Stefano Pigozzi 9f46be7373 ao_coreaudio: log even more info in verbose mode
This logs more info that can be used for debugging purposes, in particular
it prints all the AudioChannelDescription in the descriptions array.
2014-04-30 11:45:31 +02:00
Stefano Pigozzi 3623fb38ee ao_coreaudio: add verbose output of detected channel layouts
This can be useful for debugging purposes.
2014-04-30 11:45:31 +02:00
wm4 1ad9e322e3 cache: remove redundant log prefix 2014-04-30 11:45:31 +02:00
Kevin Mitchell cad63810c5 man: minor typo why not to use portaudio 2014-04-30 11:45:31 +02:00
wm4 da17c64c76 vf: remove autoinserted filters on reconfig
When using rotation with hw decoding, and the VO does not support
rotation, vf_rotate is attempted to be inserted. This will go wrong, and
after that it can't recover because a vf_scale filter was autoinserted.
Just removing all autoinserted filters before reconfig fixes this.
2014-04-30 11:45:31 +02:00
wm4 57f32f7ca8 vf: add function to remove a filter from the chain 2014-04-30 11:45:31 +02:00
wm4 3d85137495 encode: don't resume playback when encoding
The previous commit doesn't handle additionally loaded config files,
such as the playback resume mechanism.
2014-04-30 11:45:31 +02:00
wm4 4aff0076a0 manpage: add notes when to use/not use certain AO and VO drivers
Conflicts:
	DOCS/man/en/vo.rst
2014-04-30 11:45:31 +02:00
James Ross-Gowan 70f8ea228d encode: disable keep-open when encoding 2014-04-30 11:45:31 +02:00
wm4 3196f4cdd7 audio: preallocate audio buffers on resize
This avoids too many realloc() calls if the caller is appending to an
audo buffer. This case is actually quite noticeable when using something
that buffers a large amount of audio.
2014-04-30 11:45:31 +02:00
wm4 4fa0853c34 Release 0.3.8 2014-04-17 22:58:58 +02:00
wm4 6598b55c35 af_lavcac3enc: detach on any passthrough format, not just ac3 2014-04-17 22:58:48 +02:00
wm4 848d7979e5 README: mention minimum FFmpeg/Libav versions 2014-04-17 22:58:48 +02:00
Kevin Mitchell 13301daf6a talloc README: more human readible api reference 2014-04-17 22:58:48 +02:00
Ricardo Vieira 763a38f889 wayland: don't schedule resize if going fullscreen
mpv was resizing to the same size before it went to fullscreen, we don't need to schedule a resize because the compositor will send a configure event with the new dimensions and thats when we should do it.
2014-04-17 22:58:48 +02:00
James Ross-Gowan ed1e4de452 manpage: fix --vf=scale options 2014-04-17 22:58:48 +02:00
wm4 0e2fe55fde encode: write 2-pass stats only per-packet
The stats were retrieved and written on every encode call, instead of
every encode call that actually returned a packet. ffmpeg.c also does it
this way, so it must be "more correct". Fixes 2-pass encoding.
2014-04-17 22:58:48 +02:00
wm4 22cfb48b52 encode: add a missing \n to a log call
Conflicts:
	audio/out/ao_lavc.c
2014-04-17 22:58:48 +02:00
wm4 4a07120f83 terminal-unix: reject overlong termcap strings
Our own tables have size for only 8 chars, so these sequences must be
rejected. It seems strings of length 8 are still ok, because the code
uses memcmp and not strcmp, so still allow these.

Based on mplayer-svn commit r37129.
2014-04-17 22:58:48 +02:00
wm4 d06ff691ed vo_vdpau: add an additional check for timestamp robustness
This might be a good idea in order to prevent queuing a frame too far in
the future (causing apparent freezing of the video display), or dropping
an infinite number of frames (also apparent as freezing).

I think at this point this is most of what we can do if the vdpau time
source is unreliable (like with Mesa). There are still inherent race
conditions which can't be fixed.
2014-04-17 22:58:48 +02:00
wm4 34cee2a77d vo_vdpau: document what WRAP_ADD does
This wasn't necessarily clear.
2014-04-17 22:58:47 +02:00
wm4 8f5eb7b780 vo_vdpau: simplify previous vsync timestamp calculation
The strange thing about this code was the shift parameter of the
prev_vs2 function. The parameter is used to handle timestamps before the
last vsync, since the % operator handles negative values incorrectly.
Most callers set shift to 0, and _usually_ pass a timestamp after the
last vsync. One caller sets it to 16, and can pass a timestamp before
the last timestamp.

The mystery is why prev_vs2 doesn't just compensate for the % operator
semantics in the most simple way: if the result of the operator is
negative, add the divisor to it. Instead, it adds a huge value to it
(how huge is influenced by shift). If shift is 0, the result of the
function will not be aligned to vsyncs.

I have no idea why it was written in this way. Were there concerns about
certain numeric overflows that could happen in the calculations? But I
can't think of any (the difference between ts and vc->recent_vsync_time
is usually not that huge). Or is there something more clever about it,
which is important for the timing code? I can't think of anything
either.

So scrap it and simplify it.
2014-04-17 22:58:47 +02:00
wm4 b991c072f6 vo_vdpau: simplify time management and make it more robust
vo_vdpau used a somewhat complicated and fragile mechanism to convert
the vdpau time to internal mpv time. This was fragile as in it couldn't
deal well with Mesa's (apparently) random timestamps, which can change
the base offset in multiple situations. It can happen when moving the
mpv window to a different screen, and somehow it also happens when
pausing the player.

It seems this mechanism to synchronize the vdpau time is not actually
needed. There are only 2 places where sync_vdptime() is used (i.e.
returning the current vdpau time interpolated by system time).

The first call is for determining the PTS used to queue a frame. This
also uses convert_to_vdptime(). It's easily replaced by querying the
time directly, and adding the wait time to it (rel_pts_ns in the patch).

The second call is pretty odd: it updates the vdpau time a second time
in the same function. From what I can see, this can matter only if
update_presentation_queue_status() is very slow. I'm not sure what to
make out of this, because the call merely queries the presentation
queue. Just assume it isn't slow, and that we don't have to update the
time.

Another potential issue with this is that we call VdpPresentationQueueGetTime()
every frame now, instead of every 5 seconds and interpolating the other
calls via system time. More over, this is per video frame (which can be
portantially dropped, and not per actually displayed frame. Assume this
doesn't matter.

This simplifies the code, and should make it more robust on Mesa. But
note that what Mesa does is obviously insane - this is one situation
where you really need a stable time source. There are still plenty of
race condition windows where things can go wrong, although this commit
should drastically reduce the possibility of this.

In my tests, everything worked well. But I have no access to a Mesa
system with vdpau, so it needs testing by others.

See github issues #520, #694, #695.
2014-04-17 22:58:47 +02:00
wm4 3c2777994b vo_vdpau: more debugging output
Might help to debug certain problems with Mesa.
2014-04-17 22:58:47 +02:00
Kevin Mitchell 80741f27f3 remove redundant/outdated line from encodeing.rst:"what works" 2014-04-17 22:58:47 +02:00
wm4 3d18ab77b3 DOCS/encoding: make "What does not work yet" a proper RST section 2014-04-17 22:58:47 +02:00
wm4 8d2baaa318 ass_mp: don't use --subcp for --ass-styles
The --ass-styles option is implemented by calling ass_read_styles().
This function can take a codepage (so libass will use iconv to convert
it). This was implemented before our --subcp option was changed, and
this code was not updated. Now libass fails opening iconv, because
--subcp is not always (and not by default) a valid iconv codepage.

Just always pass NULL, which means the file passed to --ass-styles must
be in UTF-8. The --ass-styles option is a fringe option anyway (and will
destroy your subtitles), so having codepage support for it isn't
important at all.
2014-04-17 22:58:46 +02:00
wm4 2758bc462d Release 0.3.7 2014-03-30 17:56:54 +02:00
wm4 2687b43dd0 vf_eq: don't malloc priv struct
There wasn't any reason for this. In fact, it's a memory leak. The
proper priv struct is already allocated vf.c and the option parser.
2014-03-30 17:55:12 +02:00
Sebastian Morr bb40daf367 player: add missing "-" to options in workaround notice on A/V desynchronization
While it technically works, using GNU-style options seems cleaner nowadays.

Signed-off-by: wm4 <wm4@nowhere>
2014-03-30 17:55:07 +02:00
wm4 cc1f926441 player: dvdnav: fix start time when entering and leaving menu
Unfortunately, quite a hack, because we have check the nav state
outside of discnav.c.
2014-03-30 17:54:50 +02:00
wm4 67f653324a mf: fix operation with --cache
demux_mf.c explicitly checks for the stream type to check whether images
are opened via pattern (mf://..., i.e. stream_mf.c) or directly. Of
course the stream type is not set to STREAMTYPE_MF if the stream is
wrapped through the cache, so it tried to open the pattern directly as
file, which failed.

Fix this by disabling caching for mf://. The cache doesn't make sense
here anyway, because each file is opened and closed every frame (perhaps
to avoid memory bloat).
2014-03-30 17:54:33 +02:00
wm4 6e2c29b786 player: fix unchecked access for chapter metadata
It's possible that MPContext has a chapter list, but the demuxer
doesn't. In this case, accesing the chapter-metadata property would
lead to invalid accesses.

(This fixes the out of bound access, but in theory, the returned data
can still be incorrect, since MPContext chapters don't need to map
directly to demuxer chapters.)
2014-03-30 17:54:30 +02:00
wm4 e565217f22 stream_cdda: print cd text header only if there are any cd text fields 2014-03-30 17:54:18 +02:00
wm4 144c67efaf stream_cdda: remove unused stuff
This cd_info_t struct was practically unused. The only thing it did was
storing the track name of the form "Track %d" in a very roundabout way.
Remove it. (It made more sense when there was still CDDB support.)
2014-03-30 17:54:15 +02:00
wm4 96d7e80ddf stream_cdda: fix track time accuracy
Don't use an integer division to get the time, since that would round on
second boundaries. Also round up the time by sector size. Seeking rounds
down due to alignment constraints, but if we round up the time, we can
make it land on the exact destination sector.

This fixes that the track change code printed the previous track when
seeking by chapter.
2014-03-30 17:54:10 +02:00
wm4 dde72c5c54 stream_cdda: report track times
Report the time for each chapter (tracks are treated as chapters). This
allows us to get rid of the "old" chapter mechanism, and also behaves
better with the frontend.

This makes assumptions about the audio formats, but that format is
hardcoded anyway in the rawaudio demuxer defaults (and always was).
2014-03-30 17:54:05 +02:00
wm4 577d03da0c dvdnav: fix minor memory leak
This was usually handled at the end of the switch statement, so if
something returns from the function before that, the event has to be
freed explicitly.
2014-03-30 17:53:46 +02:00
xylosper fc510f0da8 dvdnav: make MP_NAV_EVENT_RESET_ALL handled properly
dvdnav.c did not handle event in regular sequence. Usually this
does not make any trouble except around MP_NAV_EVENT_RESET_ALL.
Those events should be handled in regular sequence. If they're
mixed, it can make wrong result.
For instance, MP_NAV_EVENT_HIGHLIGHT right after
MP_NAV_EVENT_RESET_ALL should not be ignored but it might be
because MP_NAV_EVENT_RESET_ALL makes the demuxer reloaded and osd
hidden.
2014-03-30 17:53:05 +02:00
xylosper 8d5775a8d2 dvdnav: prevent reallocation of fake highlight buffer
Since subsizes were never updated, the bitmap buffer was always
reallocated whenever mp_nav_get_highlight() called.
2014-03-30 17:52:43 +02:00
xylosper cf8ff2fd87 stream_bluray: fix for significant memory leak
It's obvious but, since STREAM_CTRL_GET_TIME_LENGTH is called
frequently, the amount of leaked memory here is quite big.
2014-03-30 17:51:45 +02:00
wm4 852c924a60 manpage: clarify what to pass to --hwdec-codecs option 2014-03-30 17:51:28 +02:00
wm4 1685d198b9 vd_lavc: ridiculous workaround for Libav 9 compatibility
This "sometimes" crashed when seeking. The fault apparently lies in
libavcodec: the decoder returns an unreferenced frame! This is
completely insane, but somehow I'm apparently still expected to
work this around. As a reaction, I will drop Libav 9 support in the
next commit. (While this commit will go into release/0.3.)
2014-03-30 17:50:11 +02:00
wm4 1e2d28f567 DOCS/coding-style: add another requirement for commit messages 2014-03-30 17:50:02 +02:00
wm4 89973ebca1 options: fix off-by-1 error in option help output 2014-03-30 17:49:57 +02:00
wm4 3f4e528f73 manpage: af: minor syntax improvements
Conflicts:
	DOCS/man/en/af.rst
2014-03-30 17:49:19 +02:00
Ryan Goulden 6f7c9f516e build: fix unicode handling of the version string
Don't pass unicode types to waf ENV.
As per https://code.google.com/p/waf/issues/detail?id=1420
This directly fixes the "CFVersion" key in the .app bundle plist.
2014-03-30 17:47:57 +02:00
James Ross-Gowan 33e0074603 player/main: Fix Cygwin build
Xlib.h (included from x11_common.h) defines a macro 'Status' as 'int'.
This messed up a bunch of definitions in windows.h and broke the build
in Cygwin. Including windows.h first seems to solve the problem.

This commit also removes the definition of an unused flag.

Conflicts:
	player/main.c
2014-03-30 17:47:44 +02:00
wm4 b6ed94fa7c DOCS/coding-style: add an example and another rule 2014-03-30 17:42:40 +02:00
wm4 a419703549 README: split list of dependencies
And make some additions.

I'm not really sure why we list dependencies that are not ours, but that
of external libraries. Still might be useful, though.
2014-03-30 17:42:30 +02:00
wm4 5171da86e3 DOCS: add coding style
I've tried not to be too detailed (because it's not a reference, just
some guidelines), but it still got relatively long. Also contains
conventions for sending patches.
2014-03-30 17:42:11 +02:00
James Ross-Gowan 7bd670949c w32_common: prevent MOUSE_BTN0 sticking after drag
The window doesn't recieve a WM_LBUTTONUP message after it's dragged,
probably because it's swallowed by the modal loop. To stop the button
from sticking, release it manually when the drag is complete.
2014-03-30 17:42:07 +02:00
James Ross-Gowan 421b2be348 w32_common: capture mouse input
Mouse buttons can get stuck down if the button is pressed inside the
video window and released outside. Avoid this by capturing mouse input
when a button is pressed.
2014-03-30 17:42:02 +02:00
Diogo Franco b659476ed9 options.rst: correct broken URL 2014-03-30 17:39:31 +02:00
wm4 689c3f47a8 Release 0.3.6 2014-03-11 00:23:11 +01:00
wm4 677d4615ba Revert "player: simplify audio reset when seeking"
This reverts commit 75dd3ec210.

This broke seeking with ordered chapters in some situations. While
the reverted commit was perfectly fine for playback of normal files,
it overlooked that in the ordered chapters case switching segments
actually reinitialized the audio chain completely, including the
decoder. And decoders still read packets on initialization. We can
restore the original commit as soon as decoders stop doing this.
2014-03-11 00:22:33 +01:00
wm4 5f788a7812 ao_alsa: fix return value for volume operations with spdif
This AO pretended to support volume operations when in spdif passthrough
mode, but actually did nothing. This is wrong: at least the GET
operations must write their argument. Signal that volume is unsupported
instead.

This was probably a hack to prevent insertion of volume filters or so,
but it didn't work anyway, while recovering after failed volume filter
insertion does work, so this is not needed at all.
2014-03-11 00:21:45 +01:00
foo86 dba25e13c0 ao_alsa: check ALSA PCM state before pause and resume
It is possible to have ao->reset() called between ao->pause() and
ao->resume() when seeking during the pause. If the underlying PCM
supports pausing, resuming an already reset PCM will produce an error.
Avoid that by explicitly checking PCM state before calling
snd_pcm_pause().

Signed-off-by: wm4 <wm4@nowhere>
2014-03-11 00:21:32 +01:00
Hans-Kristian Arntzen 623b97e45a ao_rsound: pass correct data type to rsd_set_param()
Signed-off-by: wm4 <wm4@nowhere>
2014-03-11 00:20:56 +01:00
Hans-Kristian Arntzen 7a645ba62f build: fix compilation with MinGW-w64
References to WinMM/OLE/UUID were missing.

Signed-off-by: wm4 <wm4@nowhere>
2014-03-11 00:20:53 +01:00
wm4 63afe3eb98 lua: add license header
Oops.
2014-03-11 00:20:26 +01:00
wm4 72c70eb05b ao_jack: fix termination on the end of file
The player didn't quit when the end of a file was reached. The reason
for this is that jack reported a constant audio delay even when all
audio was done playing. Whether that was recognized as EOF by the player
depended whether the exact value was higher or lower than the player's
threshhold for what it considers no more audio.

get_delay() should return amount of time it takes until the last sample
written to the audio buffer reaches the speaker. Therefore, we have to
track the estimated time when the last sample is done, and subtract it
from the calculated latency. Basically, the latency is the only amount
of time left in the delay, and it should go towards 0 as audio reaches
ths speakers.

I'm not sure if this is correct, but at least it solves the problem. One
suspicious thing is that we use system time to estimate the end of the
audio time. Maybe using jack_frame_time() would be more correct. But
apart from this, there doesn't seem to be a better way to handle this.
2014-03-11 00:19:56 +01:00
wm4 df0ec55499 build: update waf
<lachs0r> it fixes issues with recent python3
2014-03-11 00:19:43 +01:00
wm4 542bcdfc9d command: fix null pointer dereference in idle mode
Pressing 'h' in idle mode -> crash.
2014-03-11 00:19:26 +01:00
Andrey Morozov 218667b5a2 vdpau: change the error message when video too large 2014-03-11 00:16:46 +01:00
wm4 13a0e34aec example.conf: change subtitle codepage example to use UTF-8 fallback
This should be pretty safe compared to forcing the codepage.
2014-03-11 00:16:20 +01:00
Yaser Alraddadi 76b7a1a713 example.conf: subtitle encoding
Signed-off-by: wm4 <wm4@nowhere>

Closes #580.
2014-03-11 00:16:18 +01:00
wm4 6d6eee2e70 manpage: fix yadif example in one case 2014-03-11 00:15:53 +01:00
wm4 89f20cd2f3 input: check for abort cmd in multi-commands
MP_CMD_COMMAND_LIST commands (used to implement key bindings with
multiple commands) were not checked for abort commands. Implement it.

Remove the remarks about multi-commands being special from the manpage.
Seek coalescing is handled differently now, and the issue with abort
commands is fixed with this commit.
2014-03-11 00:14:30 +01:00
James Ross-Gowan cce2b675ac input: ignore modifiers when releasing keys
This prevents keys from getting stuck if the modifier is released before
the base key.
2014-03-11 00:14:22 +01:00
wm4 c1ed4dc242 video: add rounding to aspect ratio calculations
Small errors are unavoidable, but truncation can cause anamorphic video
to be off by 1 or 2 pixels.
2014-03-11 00:13:57 +01:00
wm4 712657a894 player: fix start time if timeline is used (ordered chapters, EDL)
When timeline was used, and the --start option was not used, the initial
seek (needed to switch to the first timeline segment) seeked to -1 due
to an oversight.

Conflicts:
	player/loadfile.c
2014-03-11 00:13:39 +01:00
wm4 496956de3d edl: fix offset of user-visible chapters
Basically, chapter marks and chapter seek-points were incorrect, while
the rest worked.
2014-03-11 00:12:25 +01:00
James Ross-Gowan 483e703793 w32_common: don't set small icon
Windows will automatically choose the correct icon size if this field is
unset.
2014-03-11 00:12:00 +01:00
elevengu 717cdeb95d win32: restore support for exe directory as config directory
Same rationale as b2c2fe7a but updated to work with path-win.c

Signed-off-by: wm4 <wm4@nowhere>

Merges/closes #543.
2014-03-11 00:10:20 +01:00
wm4 315f3a9218 crosscompile-mingw: improve instructions for MXE 2014-03-11 00:10:17 +01:00
xylosper d857722618 sd_lavc: handle subtitles with no subtitle resolution set
Set subtitle resolution to video resolution when avctx->width and
avctx->height are zero.

This can happen with broken vobsubs that have no size set in their
.idx file (or Matroska extradata). At least with the test file provided
in issue #551, using the video resolution as fallback instead of what
guess_resolution() does is better.

Note that these files clearly are broken. It seems this particular
file was created by trying to use ffmpeg to transcode DVB subtitles
to vobsub, and ffmpeg "forgot" to set the subtitle resolution in the
destination file. On the other hand, ffmpeg DVB and PGS decoders set
the resolution on the first subtitle packet (or somewhere close), so
it's not really clear what to do here.

Closes #551.

Signed-off-by: wm4 <wm4@nowhere>

Patch by xylosper, rewritten commit message by wm4.
2014-03-11 00:09:52 +01:00
Diogo Franco (Kovensky) 6d9d597b10 Parse::Matroska::Reader: get rid of bigrat
It seems that it was causing issues with certain perl setups (such as
the one on issue #549). It also turns out that it was not behaving correctly
(not all constants were being promoted to big nums as they should), so we
use explicit objects to derive the constants.

There were also precedence issues. I wonder if this even worked right to
begin with.

The 'double' path (8-byte floats) is untested, as I couldn't easily find
a file with such a field.

Closes #549.
2014-03-11 00:08:57 +01:00
Diogo Franco (Kovensky) c4d905b567 build: Add -U__STRICT_ANSI__ to CFLAGS on Cygwin
Cygwin's libc (newlib) doesn't obey a lot of unix feature test macros,
including _GNU_SOURCE; as a result, a lot of functions and defines get
masked out -- important defines such as M_PI and strcasecmp. Work around
it by undefining __STRICT_ANSI__ on cygwin systems.

This will still cause compilation issues on any non-cygwin system that
uses newlib, but hopefully nobody does that, or if they do, they will
find this commit message and know to add -U__STRICT_ANSI__ to their
CFLAGS. Hopefully.
2014-03-11 00:08:53 +01:00
wm4 e4c69376fc sub: handle vobsub-in-mp4
The mplayer decoder (spudec.c) actually handled this. There was explicit
code for binary palettes (16 32 bit values), and the subtitle resolution
was handled by video resolution coincidentally matching the subtitle
resolution.

Whoever puts vobsub into mp4 should be punished.

Fixes the sample gundam_sample.mp4, closes github issue #547.
2014-03-11 00:08:29 +01:00
wm4 842689bbf6 sd_lavc: use mp_lavc_set_extradata()
This includes the magical input padding required by libavcodec, which we
possibly didn't do before this commit.
2014-03-11 00:07:57 +01:00
wm4 5ab2aa84e0 Factor out setting AVCodecContext extradata 2014-03-11 00:07:51 +01:00
Stefano Pigozzi 229d398ba9 cocoa: fix deadlock during initialization [2]
Fixup commit for 20fa191ad.
2014-03-11 00:04:15 +01:00
wm4 5f1b155307 build: bump libmpg123 version
The minimum required version was bumped in the old configure script, but
for the waf build system is was somehow forgotten or overlooked.
Probably happened while the waf build system was developed in a separate
branch.

Closes #546.
2014-03-11 00:04:10 +01:00
Stefano Pigozzi 241cbe5065 cocoa: fix deadlock during initialization
Thanks to @wm4 for catching the bug.

Fixes #405
2014-03-11 00:04:07 +01:00
wm4 76d771a6fa Release 0.3.5 2014-02-12 22:39:52 +01:00
nand 19caba0794 vo_opengl: make :srgb decompand the BT.709 values correctly
This is the same issue as addressed by 257d9f1, except this time for
the :srgb option as well. (257d9f1 only addressed :icc-profile)

The conditions of the srgb_compand mix() call are also flipped to
prevent an off-by-one error.
2014-02-12 22:34:44 +01:00
wm4 391546c52f command: fix metadata property
This crashed when retrieving the raw property value. Oops.
2014-02-12 22:34:28 +01:00
Alexander Preisinger a8af6a7852 wayland/shm: prevent the window from flying away
With the new xdg_shell the problem will be no gone by itself.
2014-02-12 22:34:09 +01:00
wm4 960aca3386 manpage: document --aspect special values
Use of these is "discouraged", but they're there to select these special
cases with the "aspect" property. They really should use some sort of
choice option type, but since it would be some work to make these work
with float values, the simple and dumb alternative was picked.
2014-02-12 22:33:54 +01:00
wm4 e72a327d74 video: fix --no-aspect
This also affects the --aspect option and the "aspect" property.
2014-02-12 22:33:52 +01:00
nand 87f955f3e9 vo_opengl: fix typo in gamma function's parameter
The correct value is 0.081, not 0.18. The scale factor also needed
slight adjustment due to the order of operations.
2014-02-12 22:33:32 +01:00
Jonas Zetterberg 5536ebe82f vo_vdpau: Discard zero timestamps
Some drivers do not supply timestamps, use old timestamp in these conditions.
2014-02-12 22:33:29 +01:00
Jonas Zetterberg ef8b37ca2a vo_vdpau: Ensure presentation time is within bounds
When a time sync happens the last sync time is the minimum time that can be
used for presentation.
2014-02-12 22:33:27 +01:00
wm4 d4a5513847 demux: fix some newlines in output messages 2014-02-12 22:33:09 +01:00
wm4 d42212cd67 player: fix an assert when reinitializing audio in some cases
This sometimes happened when changing playback speed (= reinitializing
audio) after seeking of playback start. The assertion in audio.c:441 was
triggered, because buffer_playable_samples wasn't reset correctly when
the audio buffer was cleared or shortened. The assertion is correct and
should hold up any time.
2014-02-12 22:33:03 +01:00
nand 674cc1d3a9 vo_opengl: use exactly the values defined by BT.709 for CMS
I could not see any difference whatsoever, but for usage with a 3DLUT
there's zero performance difference so we might as well follow the spec to
the letter.
2014-02-12 22:32:56 +01:00
wm4 c47bb9ca42 gl_x11: don't require a X visual for modern GL context creation
Legacy GL context creation (glXCreateContext) explicitly requires a X
visual, while the modern one (glXCreateContextAttribsARB) does not for
some reason. So fail only on the legacy code path if we don't find a
visual. Note that vo_x11_config_vo_window() will select a default visual
if a NULL visual is passed to it.
2014-02-12 22:32:43 +01:00
wm4 938dad0b88 gl_x11: always request true color visual
This fixes issue #504. For some reason, glXChooseFBConfig() will return
a fbconfig with no associated visual. (I'm not sure if this allowed.
They don't always have a visual, but since GLX_X_RENDERABLE is set
and GLX_DRAWABLE_TYPE is (implicitly) set to GLX_WINDOW_BIT, why would
there be no visual?)

Even worse, a test program seems to show that a 16 bit fbconfig is
selected (instead of 24/32 bit), which doesn't sound nice at all. Since
there _are_ better fbconfigs available, glXChooseFBConfig() should
normally sort them by quality, and return the better ones first. It's
worth noting that this function should also prefer GLX_TRUE_COLOR
over anything else, although this comes last in the sort order.

Whatever is going on, requesting GLX_X_VISUAL_TYPE with GLX_TRUE_COLOR
seems to fix it.
2014-02-12 22:32:40 +01:00
wm4 82b5361946 umpv: silence mpv terminal output
Now it does basically the same as mkv.desktop. Explanation is included
in the description comment.
2014-02-12 22:32:06 +01:00
wm4 f99f48684b umpv: improve description 2014-02-12 22:32:04 +01:00
wm4 25864a5539 umpv: use --force-window
This makes it show a window even when playing audio-only files.
2014-02-12 22:32:02 +01:00
wm4 524d9585fd audio: fix signedness of AF_FORMAT_S32P
This was marked as unsigned, but it's signed. Found by xylosper.
2014-02-12 22:32:00 +01:00
wm4 95a72baf92 mf: don't limit filename length with PATH_MAX
Use an arbitrary constant instead, which is as good as PATH_MAX.

This helps us to avoid having to think about pull request #523.

Also fix a case where a potentially signed char was passed to isspace().
2014-02-12 22:31:54 +01:00
wm4 48112a74ed player: refresh OSD on track switching
Apparently, at least sub_reload was missing a refresh at all.
2014-02-12 22:31:40 +01:00
Alexander Preisinger 8427cdc717 wayland: change minimum version
Change minimum version to 1.3 and remove the version checking in the source
code.
2014-02-12 22:31:27 +01:00
wm4 b695ca1e27 x11: fix initial VO size
This was done incorrectly in the previous commit: the fallback size used
the window size as requested with the first config call, which is the
size of the hidden window in the vo_opengl case. (That damn hidden
window again...)

Conflicts:
	video/out/x11_common.c
2014-02-12 22:31:21 +01:00
wm4 801feb6420 x11: remove apparently useless code
This code essentially does nothing. As far as I could find out, this
actually used to do something. Then it was removed with commit efe7c39f,
leaving some leftover code that didn't do anything useful. This happened
12 years ago!

Also remove a commented debug printf.
2014-02-12 22:30:34 +01:00
wm4 de6a452039 x11: fix race condition when setting aspect when leaving fullscreen
vo_opengl creates a hidden X11 window to probe the OpenGL context. It
must do that before creating a visible window, because VO creation and
VO config are separate phases.

There's a race condition involving the hidden window: when starting with
--fs, and then leaving fullscreen, the unfullscreened window is
sometimes set to the aspect ratio of the hidden window. I'm not sure why
the window size itself uses the correct size (but corrupted by the wrong
aspect), but that's perhaps because the window manager is free to ignore
the size hint while honoring the aspect, or something equally messed up.

It turns out this happens because x11_common.c thinks the size of the
hidden window is the size of the unfullscreened window. This in turn
happens because vo_x11_update_geometry() reads the size of the hidden
window when called in vo_x11_fullscreen() (called from
vo_x11_config_vo_window()) when mapping the fullscreen window. At that
point, the window could be mapped, but not necessarily. If it's not
mapped, it will get the size of the unfullscreened window... I think.

One could fix this by actively waiting until the window is mapped. Try
to pick a less hacky approach instead, and never read the window size
until MapNotify is received.

vo_x11_create_window() needs a hack, because we'd possibly set the VO's
size to 0, resulting e.g. in vdpau to fail initialization. (It'll print
error messages until a proper resize is received.)

Conflicts:
	video/out/x11_common.c
2014-02-12 22:30:23 +01:00
wm4 4b764b6cc6 old-configure: use -std=c99
See previous commit.
2014-02-12 22:29:50 +01:00
wm4 7b7c31b190 build: switch to -std=c99 for saner float semantics
This fixes a weird bug with aspect ratio handling. It has to do with
float handling: with -std=gnu99, gcc implicitly enables broken non-
standard semantics giving float variables excess precision. This can for
example make this fail in theory: "float a = 0.1; assert(a == a);"

While standard C allows excess precision _within_ expressions, it
requires truncation when storing float values in variables of types
"float" or "double". The "gnu99" mode breaks this. It can be unbroken by
using "c99", or by specifying -fexcess-precision=standard. The former
seems less likely to break compilers other than modern gcc. Note that
-ffloat-store would also fix this, but also makes float expressions less
efficient and less precise for no reason.

The code that mistakenly fails because of this is dec_video.c line 393.
It caused the container aspect to be ignored in some or all situations,
depending how the compiler optimizes. For example, on gcc-4.6 with -Os,
the aspect is always ignored.

In future, we should probably just get rid of storing aspects as floats.

Note: requires mingw-w64 trunk for math.h fixes (M_PI was hidden).
2014-02-12 22:29:21 +01:00
wm4 70a66b461d cache: refuse to seek outside of cache boundaries
Note that this still happens in the stream level, so we can't have
nice highlevel behavior restricting seeking. Instead, if a seek leads
to the demuxer requesting data outside of the cached range, the seek
will simply fail. This might confuse the demuxer, and the resulting
behavior is not necessarily useful.

Note that this also doesn't try to skip data on a forward seek. This
would just freeze the stream with slow unseekable streams.

One nice thing is that stream.h has a separate function for merely
skipping data (separate from seeking forward), which is pretty useful
in this case: we want skipping of data to work, even if we reject
seeking forward by skipping data as too expensive. This probably is
or will be useful for demux_mkv.c.
2014-02-12 22:29:11 +01:00
wm4 844e8f2b51 Release 0.3.4 2014-01-31 19:11:48 +01:00
wm4 3f4ed8920f vo_wayland: silence shadowing warning
No real problem.
2014-01-31 19:10:15 +01:00
wm4 7ad2999b5f quvi: disable subtitle fetching by default
This is slow and unreliable, basically unusable.
2014-01-31 19:08:30 +01:00
Andre D 14fdbe07ac quvi: add option to not fetch subtitles
Signed-off-by: wm4 <wm4@nowhere>
2014-01-31 19:08:26 +01:00
wm4 61ed9a8747 demux_lavf: fix crash with empty files
This used to work; I'm not sure when or why it regressed. When setting
AVProbeData.filename to NULL, libavformat will crash in rtp_probe() by
unconditionally accessing the string.

We used to set the filename to NULL to prevent probing by file extension
when we don't deem it as necessary. Using an empty string also works for
this purpose.
2014-01-31 19:08:12 +01:00
wm4 0801345d68 mp_image: reject too large image sizes
Larger sizes can introduce overflows, depending on the image format. In
the worst case, something larger than 16000x16000 with 8 bytes per pixel
will overflow 31 bits.

Maybe there should be a proper failure path instead of a hard crash, but
not yet. I imagine anything that sets a higher image size than a known
working size should be forced to call a function to check the size (much
like in ffmpeg/libavutil).
2014-01-31 19:07:47 +01:00
wm4 5bd0fe5daf sd_lavc: skip 0 sized sub-bitmaps
Not everything in the OSD path handles 0x0 sized sub-bitmaps well. At
least the code implementing --sub-gray had severe problems with it.

Fix this by skipping such bitmaps.
2014-01-31 19:07:43 +01:00
Alexander Preisinger cab3c5fbb8 wayland/shm: RGB888 as default, change options
RGB565 is one of the fastest and most supported formats on low end consumer
devices, but ffmpeg spams warning when using it. Make it opt-in instead of
opt-out.
2014-01-31 19:07:40 +01:00
Alexander Preisinger 99ceae8cc3 wayland/shm: fix memory leak 2014-01-31 19:07:37 +01:00
Alexander Preisinger ebe04ab692 wayland/shm: remove resize boilerplate
The problem seems to have solved itself. I guess the previous changes to
resizing and commit ba101ab made this possible. Consider me happy for removing
that crap.
2014-01-31 19:07:31 +01:00
James Ross-Gowan 808aa08cc0 w32: use safe DLL search paths everywhere
Windows applications that use LoadLibrary are vulnerable to DLL
preloading attacks if a malicious DLL with the same name as a system DLL
is placed in the current directory. mpv had some code to avoid this in
ao_wasapi.c. This commit just moves it to main.c, since there's no
reason it can't be used process-wide.

This change can affect how plugins are loaded in AviSynth, but it
shouldn't be a problem since MPC-HC also does this and it's a very
popular AviSynth client.
2014-01-31 19:07:09 +01:00
James Ross-Gowan 2065e0ba36 w32: enable heap corruption detection
Enable the terminate-on-corruption feature. This is recommended for new
Windows applications and shouldn't cause a performance hit. It actually
shouldn't change anything for 64-bit builds, since Win64 has this
switched on by default.

See:
http://blogs.msdn.com/b/michael_howard/archive/2008/02/18/faq-about-heapsetinformation-in-windows-vista-and-heap-based-buffer-overruns.aspx
2014-01-31 19:07:07 +01:00
James Ross-Gowan 3c43a022af w32: don't disable the error reporting dialog
Windows users expect this when a program crashes. Without it, the
program just disappears. Also change the SetErrorMode call to use macros
instead of a hardcoded constant.
2014-01-31 19:07:05 +01:00
wm4 02513cd5df sub: fix crash with certain uses of --vf=sub
If, for some reason, the subtitle renderer attempts to render a
subtitle before SD_CTRL_SET_VIDEO_PARAMS was called, it passed a
value calculated from invalid values. This can happen with --vf=sub
and --start. The crash happens if 1. there was a subtitle packet that
falls into the timestamp of the rendered video frame, 2. the playloop
hasn't informed the subtitle decoder about the video resolution yet
(normally unneeded, because that is used for weird corner cases only,
so this code is a bit fuzzy), and 3. something actually requests a
frame to be drawn from the subtitle renderer, like with vf_sub.

The actual crash was due to passing NaN as pixel aspect to libass,
which then created glyphs with ridiculous sizes, involving a few
integer overflows and unchecked mallocs.

The sd_lavc.c and sd_spu.c cases probably don't crash, but I'm not
sure, and it's better fix them anyway.

Not bothering with sd_spu.c, this crap is for compatibility and will
be removed soon.

Note that this would have been no problem, had the code checked whether
SD_CTRL_SET_VIDEO_PARAMS was actually called. This commit adds such a
check (although it basically checks after using the parameters).

Regression since 49caa0a7 and 633fde4a.
2014-01-31 19:06:58 +01:00
wm4 4d986d2a98 demux_lavf: add hack to workaround too unreliable mp3 detection
This generally affects mp3 files that don't have any (or many) mp3
frames in the first 2 MB. 2 MB is the maximum probe size, and
libavformat returns a low probescore even if we give it the full 2 MB.
Trying to probe a larger buffer (or even the full file) doesn't work for
mysterious reasons.

The workaround consists in accepting a very weak probescore if the
format is detected as mp3 and we probed already 2 MB.
2014-01-31 19:06:49 +01:00
wm4 ada6f96811 demux_lavf: refactor format probing hack
Restructure it a bit, so we can use the format hack list even if no mime
type applies. Shouldn't change anything functionally yet. Preparation
for the next commit.
2014-01-31 19:06:44 +01:00
wm4 321b3952b0 vo_wayland: fix confusion of video and window sizes 2014-01-31 19:06:29 +01:00
reimar d49b49ff96 stream_pvr: Fix fd check, -1 indicates invalid, not 0.
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@36677 b3059339-0415-0410-9bf9-f77b7e298cf2
2014-01-31 19:06:20 +01:00
wm4 f088abb2bf audio: fix balance control
Balance controls as used by mixer.c was broken, because af_pan.c stopped
accepting its arguments. We have to allow 0 channels explicitly. Also,
fix null pointer access if the matrix parameter is not used.

Regression from commit 82983970.
2014-01-31 19:06:17 +01:00
wm4 4a3975bf7e waf: try to fix unicode/byte string messup
Some mpv builds identify with e.g. "mpv b'0.3.3' ". The version looks
like str() was called on a Python byte string. I couldn't reproduce it
on my machine (I tried with both Python 2 and 3), so I'm not exactly
sure what's going on here, but I'm hoping this commit does fix it.
2014-01-31 19:05:20 +01:00
wm4 a064143899 manpage: fix af_equalizer syntax
This doesn't need quoting, for some reason.
2014-01-31 19:04:54 +01:00
11rcombs f264d52099 af: fixed out-of-bounds accesses caused by NUM_FMT and co.
Signed-off-by: wm4 <wm4@nowhere>

This merges pull request #496. The problem was that at least the
initialization of the distance[] array accessed af_fmtstr_table[]
entries that were out of bounds. Small cosmetic changes applied to
the original pull request.
2014-01-31 19:04:29 +01:00
reimar 74a3c169b5 cookies.c: cols must (and does) have 7 elements.
Doesn't affect the generated code, but avoids confusion
in both humans and newer Coverity versions.

git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@36623 b3059339-0415-0410-9bf9-f77b7e298cf2
2014-01-31 19:04:15 +01:00
reimar 34d9787fcf vcd_read: Fix sizeof argument.
The struct we need to copy is actually a cdrom_msf0, not cdrom_msf.
Even though the kernel for no good reason reads it in as a
cdrom_msf struct, but only uses the part shared with cdrom_msf0 -
this is probably a kernel bug.

git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@36622 b3059339-0415-0410-9bf9-f77b7e298cf2
2014-01-31 19:04:09 +01:00
wm4 1c0b80d518 mkv.desktop: add two more mime types
Stolen from github issue #484. No idea whether they're needed.
2014-01-31 19:03:57 +01:00
wm4 a9bafeb589 cache: remove debug code
This was accidentally committed.

<rcombs> `mp_sleep_us(100000);` <-- are you drunk?
2014-01-31 19:03:17 +01:00
wm4 3a33db3d92 cache: reduce message spam
Output only 1 message every 5 seconds at most.
2014-01-31 19:02:45 +01:00
wm4 d276f753f7 Release 0.3.3 2014-01-15 21:03:02 +01:00
Grzegorz Blach c06000eef8 Detect Lua on FreeBSD 2014-01-15 20:57:17 +01:00
wm4 1bc1cd2305 player: avoid stalling when starting a network stream
Starting a network stream could stall by executing uncacheable stream
control requests (STREAM_CTRL_GET_LANG and STREAM_CTRL_GET_DVD_INFO).
Being uncacheable means the player has to wait until the cache is done
reading the current block of data. These requests can't be cached
because they're too complicated, so the only way to avoid them is
special casing the DVD and Bluray streams (which are the only things
which need these requests), and not doing them in other cases.

(This is kind of inelegant, but so is the rest of the DVD/BD code.)
2014-01-15 20:57:09 +01:00
wm4 a735a2abd7 demux_mkv: improve robustness by explicitly checking for level 1 elements
Matroska makes it pretty hard to resync correctly on broken files:
random data returns "valid" EBML IDs with a high probability, and when
trying to skip them it's likely that you skip a random amount of data
(instead of considering the element length invalid).

Improve upon this by skipping known level 1 elements only. Consider
everything else invalid and call the resync code. This might result in
annoying behavior when Matroska adds new level 1 elements, although it
won't be particularly harmful. Matroska doesn't really allow us to do
better (even mkvtoolnix explicitly checks for known level 1 elements).

Since we now don't always want to combine EBML element skipping and
resyncing, remove ebml_read_skip_or_resync_cluster(), and make
ebml_read_skip() more tolerant against skipping broken elements.

Also, don't resync when reading sub-elements, and instead do resyncing
when reading them results in an error.
2014-01-15 20:56:16 +01:00
wm4 112c1abe31 demux_mkv: avoid skipping too much data in corrupted files
Until now, corrupted files were detected if the size of an element (that
should be skipped) was larger than the remaining file. This still could
skip larger regions of the file itself if the broken size happened to be
within the file.

Change it so that it's never allowed to skip outside the parent's
element.
2014-01-15 20:56:13 +01:00
wm4 362b35b631 manpage: document --term-osd=force
Apparently this was forgotten when it was first added, or maybe it's an
arrifact from the rst conversion.

Conflicts:
	DOCS/man/en/options.rst
2014-01-15 20:54:32 +01:00
wm4 49cf0af294 player: don't block terminal OSD while seeking
Seeking usually show the status on OSD. In terminal OSD mode, no status
is shown, because there is already a separate status line.
Unfortunately, the mechanism for showing the status was still active,
which forced showing no message while the code for showing seek status
was active.
2014-01-15 20:53:44 +01:00
wm4 1b26874710 player: mention subtitles in av_desync_help_text
Insane .ass subtitle scripts can cause severe slowdown (depending on the
speed of the machine, or the insanity of the script), so mention how to
test without subtitles. This is mainly to make the user aware that
subtitle rendering can be a problem. For longwinded explanation, there
isn't enough space.
2014-01-15 20:53:41 +01:00
Stefano Pigozzi 8fe636da7a cocoa: fix window placement on secondary screens
For a long time the cocoa backend set the xinerama_x/y and used dx/dy from the
VO instance. This somewhat worked with some workarounds but wasn't really
what was supposed to be happening. Moreover 27e4360, which touched this
workaround introduced a regression.

New code doesn't set the xinerama_x/y values so that dx/dy are offsets in the
current screen (not a virtual screen composed of all the screens). The screen
reference detected during VOCTRL_UPDATE_SCREENINFO is also passed down to the
window initialization code.

Fixes #472
2014-01-15 20:53:19 +01:00
Stefano Pigozzi 812f149585 cocoa: refactor init window positioning code 2014-01-15 20:53:14 +01:00
wm4 635424c8f0 demux_lavf: add hack for MicroDVD for assuming frame based timing
MicroDVD files _can_ contain real timestamps instead of frame timestamps
if they declare a FPS. But this seems to be rare, so ignore that if the
FPS happens to match with the libavformat microdvd parser's default FPS.

This might actually break files that declare 23.976 FPS, but the video
file is not 23.976 FPS, but the chance that this happens is probably
very low, and the commit fixes the more common breakage with 25 FPS
video.
2014-01-15 20:51:50 +01:00
wm4 2319918721 sub: fix frame based subtitle timestamp handling
Subtitle formats with frame based timing require using the video FPS to
compute proper subtitle timestamps. But it looks like the calculation to
do that was inversed.
2014-01-15 20:51:47 +01:00
Alexander Preisinger 9fabf85a44 wayland: properly empty output list 2014-01-15 20:51:39 +01:00
Alexander Preisinger 54fd5d0979 wayland/shm: don't crash if initialization failed 2014-01-15 20:51:37 +01:00
wm4 d141d281ca wayland: fix crash when initialization fails
On X11, if no wayland compositor is running, wl_list_init() will never
be called. This will cause destroy_display() to segfault when trying to
iterate over the list.
2014-01-15 20:51:32 +01:00
Alexander Preisinger ebc3d74b04 wayland/shm: tone down warnings
Those warnings are printed far too often and actually aren't usefull at all.
2014-01-15 20:51:19 +01:00
Alexander Preisinger a63fab8921 wayland: fix memory leaks
There are still some leaks from wayland-cursor stuff, but there is no way to
free the memory as user of the cursor library.

Conflicts:
	video/out/wayland_common.c
2014-01-15 20:50:52 +01:00
wm4 4ee8d8d3f9 player: fix setting smaller timeout on Windows systems
On Windows, we don't have proper input event wakeup handling, so we
need to lower the playloop timeout in order to react fast to input.

Closes #387.
2014-01-15 20:50:02 +01:00
Alexander Preisinger 574a2484c1 wayland: cleanup registry_handle_global
The wl_registry object is already passed as a parameter. No need to create
a temporary variable.
2014-01-15 20:49:57 +01:00
Alexander Preisinger f5b92bee5e wayland: remove set_user_data from seat_listener
The user_data is passed on add_listener and can later be changed with
set_user_data. But because we don't want to change it later and because it is
the same object remove the set_user_data call.

This might be a copy&paste leftover from the initial draft for the wayland
backend.
2014-01-15 20:49:55 +01:00
Alexander Preisinger ca3d7144db wayland: use static consistently
Declare everything that is only needed inside wayland_common.c as static.
2014-01-15 20:49:52 +01:00
wm4 86e7c3defb terminal: don't initialize termcap etc. if stdout is not a terminal
Otherwise, it seems one of the term* libraries will write escape
sequences to stdout, for whatever reason.
2014-01-15 20:49:42 +01:00
Alexander Preisinger 0fe1385d80 wayland: fix fullscreen & resizing for good
I added enough logic to never set ontop or fullscreen twitce.
This commit keeps also the size of the video if multiple videos are played.
If the ratio differs the width will be kept at the same size and only the
height changes.
2014-01-15 20:49:32 +01:00
wm4 f4e9371d0f manpage: mention how to get a list of codecs for use with --hwdec-codecs 2014-01-15 20:49:26 +01:00
wm4 909185d49e wayland: silence annoying libwayland-client message
libwayland-client contains the following code [1]:

	runtime_dir = getenv("XDG_RUNTIME_DIR");
	if (!runtime_dir) {
		fprintf(stderr,
			"error: XDG_RUNTIME_DIR not set in the environment.\n");

This means this message will unconditionally and unavoidably be printed
if XDG_RUNTIME_DIR is not set. Since mpv is a terminal program, and we
want to avoid unnecessary output, work it around by not attempting to
use wayland if this environment variable is not set.

[1] http://cgit.freedesktop.org/wayland/wayland/tree/src/wayland-client.c#n636
    (cd0dccd01e16fa404e03974d30ded3aebdb1c4bc)
2014-01-15 20:49:22 +01:00
wm4 c18c324c6f wayland: fix some memory leaks on initialization failure
This commonly happens when initializing vo_opengl on a X11-only system.

Unfortunately, most wl_*_destroy() functions appear not to accept NULL
pointers, making partial deinitialization a pain: you have to add your
own NULL checks everywhere to avoid crashes.

xkb.context is uninitialized separately, because you can initialize it
just fine, even if the rest of input initialization fails.
2014-01-15 20:49:19 +01:00
wm4 6b20c24413 Release 0.3.2 2014-01-06 20:32:24 +01:00
Stefano Pigozzi 5c42d1eb92 cocoa: don't reset window size when the video size doesn't change
Fixes #459
2014-01-06 20:27:55 +01:00
Martin Herkt e34132f8d5 w32_common: don't force topmost on fullscreen
Fixes #457, #444.
2014-01-06 20:27:53 +01:00
Martin Herkt bd7ff23f96 Windows: use roaming AppData instead of local
Whatever. Fixes #458.
2014-01-06 20:27:47 +01:00
wm4 e6c184e2b1 build: don't depend on both libavresample and libswresample
When both libavresample and libswresample were detected, the script
enabled both at the same time. This is not supported; although nothing
bad happened apparently. Make the dependencies both mutually exclusive.
2014-01-06 20:23:10 +01:00
wm4 33c4920967 player: always ise [statusline] for terminal OSD
This was inconsistent: the actual statusline used [statusline] as
message prefix, while other parts of the terminal OSD used [cplayer]
(and MSGL_STATUS). This commit makes it consistent.
2014-01-06 20:23:06 +01:00
wm4 0c36f655e1 player: print an error message if run command fails
Note that we can't use mp_msg, because it's not async-signal safe (we
might be running other threads while forking, so only functions
specified to be async-signal safe can be called, and this doesn't
include stdio; mp_msg acquires a mutex too).

Also, always print a \n before running the program to flush the status
line. The effect is that a program running successfully as well as the
error message will effectively start on a new line.
2014-01-06 20:23:03 +01:00
wm4 cd31697dd8 manpage: fix typo 2014-01-06 20:22:57 +01:00
wm4 c497b82583 player: don't select subtitles added from quvi by default
Quvi subtitles are considered external subtitles (simply because they're
separate from the audio/video stream), but for the sake of subtitle
auto-selection, they should not be considered external.

Change this so that quvi subtitles are treated like muxed subtitles
(with default flag never set). This means subtitles won't be selected by
default, unless explicitly requested with --sid or --slang.
2014-01-06 20:22:53 +01:00
Alexander Preisinger 6f5dfb5102 Revert "wayland/egl: use redraw callback"
Because of this commit there were problems displaying the frmase in their right
order.

This reverts commit 96e75d234a.

Conflicts:
	video/out/gl_wayland.c
	video/out/wayland_common.h
2014-01-06 20:22:43 +01:00
Alexander Preisinger 59f970d492 wayland: remove workaround
The changes in the vo_wayland_ontop function have no effect on the workaround.

Somehow the problem just disappeared. I guess it is because of the new control
function in gl_wayland.c where the resize happens immediatly after the event
dispatch/flush.
2014-01-06 20:22:41 +01:00
wm4 29c74b42b4 demux_subreader: reject file if not opened by --sub
demux_subreader.c contains the old MPlayer subtitle parser, and I have
absolutely no confidence in this (very crappy) code. There might be
one or two security risks associated with running that code on
arbitrary input.
2014-01-06 20:22:26 +01:00
Alexander Preisinger 4b2a3e50e2 wayland/egl: misc fixes and cleanups 2014-01-06 20:22:11 +01:00
Alexander Preisinger 35b01c9604 wayland/egl: use redraw callback
This solves the issue where we would not receive any frame events. The
difference to my earlier tests is that now it looks like eglSwapBuffers uses
it's own event queue or something similiar along the lines. Becaues the
performance is the same as without any redraw callback.
2014-01-06 20:22:08 +01:00
Alexander Preisinger 09b95257fe wayland/egl: add egl_context to the wayland state
At the moment there are visual glitches when we resize the window. This happens
because in wayland there a special function for resizing EGL windows.

To prevent the glitches move the egl_context to the wayland state in
wayland_common.h and add a new control function to gl_wayland.c to wrap the
vo_wayland_control function to check for resize events.

With the new control wrapper the glitches are gone and the resizing is fluid.
2014-01-06 20:22:06 +01:00
Alexander Preisinger 0cf2df38d1 wayland: Remove nonsense comment and add warning
The reason a segmentation happend here was because we couldn't get the
requested minor version. The major version is enough for differentiating
between OpenGL 3 and OpenGL 2. If it fails there is still a fallback to any
version available.

Also add a warning if we use the fallback.
2014-01-06 20:22:03 +01:00
Alexander Preisinger 4326fdfe2a wayland: move workaround to fullscreen code
Because of this workaround there was a bug in the wayland vo.
Now it works on both vos (opengl and shm) as it should.
2014-01-06 20:21:55 +01:00
Martin Herkt 9755850b3d manpage: mention --pause 2014-01-06 20:21:45 +01:00
wm4 96b50e770b dvdnav: set correct stride for fake highlights
This didn't really matter, because the fake highlight rectangle is in a
single color, and every pixel has the same color value.
2014-01-06 20:21:43 +01:00
wm4 9f6551160a stream_pvr: fix crash on initialization
Fallout from the mp_msg conversions.
2014-01-06 20:21:39 +01:00
wm4 89b854165e audio: check for overflows 2014-01-06 20:21:37 +01:00
wm4 0beece4397 ta: check overflow in array realloc macros 2014-01-06 20:21:19 +01:00
Stefano Pigozzi 8abdf1f56e vda: fix build on OS X 10.7
Looks like on 10.8 OpenGL.h recursively includes CGLIOSurface.h. That is not
the case for 10.7 so the build was broken on that version of OS X.
2014-01-06 20:21:17 +01:00
Stefano Pigozzi 5a81f7678e build: fix cocoa configure check on OS X 10.7
It failed because the 10.7 SDK doesn't natively support array and dictionary
subscripting.
2014-01-06 20:21:14 +01:00
Alexander Preisinger e26286dfdb wayland/opengl: fix garbage borders
It seems mpv draws garbage in those regions. Now we calculate the aspect and
let weston draw the black borders.
2014-01-06 20:21:11 +01:00
Alexander Preisinger 389232e5a2 wayland/shm: only commit new frames
Only attach and commit new buffers. This also increases performance when
paused.
2014-01-06 20:21:09 +01:00
Alexander Preisinger c5de0b0a6c wayland: don't change set fullscreen twice
Newest weston chrashes if we call set_fullscreen twice. This is a major bug
I which I should probably report.
2014-01-06 20:21:07 +01:00
Martin Herkt f9bdfc0782 ao_alsa: Unbreak pause/resume
Well that was dumb.
2014-01-06 20:20:57 +01:00
Martin Herkt 5c9a7e24f6 ao_alsa: Fix PCM resume after suspend
Fixes #324
2014-01-06 20:20:55 +01:00
wm4 b56f61d2ab stream: always respect sector_size, fixes cdda://
Streams like CDDA have special requirements in what quantities data can
be read: you can only read a sector at once, not more and not less. The
stream_peek() function didn't respect that and set less (used internal
buffer size of 2048 bytes, instead of CD sector size of 2352 bytes), so
no data was read and EOF was accidentally set, making playback with
cdda:// fail.

This is a regression since commit 9a723f, but that commit merely exposed
the issue (the redundant seek would clear the EOF flag).
2014-01-06 20:20:48 +01:00
Stefano Pigozzi 82ab3a8160 build: fix typo 2014-01-01 21:50:42 +01:00
wm4 e371f2ebd3 Release 0.3.1 2014-01-01 21:03:51 +01:00
Stefano Pigozzi 29b68d69c0 build: make configure fail if both __atomic and __sync are not available 2014-01-01 20:57:27 +01:00
wm4 6c770eb6df manpage: clarifications about bitmap subtitles and --secondary-sid 2014-01-01 20:57:25 +01:00
wm4 965f336917 options: make --msglevel=help print something helpful 2014-01-01 20:57:22 +01:00
wm4 47fae324c7 manpage: fix description on the slave mode msglevel 2014-01-01 20:57:18 +01:00
wm4 6ab0dbf2ed player: load encoding-profiles.conf before the main config
Otherwise one can't add profiles based on the encoding profiles.
2014-01-01 20:57:16 +01:00
Miro Hrončok c002196ffe mpv.desktop: Add Czech translations 2014-01-01 20:57:14 +01:00
wm4 3c8fc4588d player: fix DVD playback
Crashed in case of lazily added subtitle streams, which add tracks with
track->stream set to NULL.

Fixes gituhub issue #439.
2014-01-01 20:57:12 +01:00
Alessandro Ghedini 416015f059 compat: use __atomic operations instead of __sync, when present
Fixes #434
Fixes #437
2014-01-01 20:57:09 +01:00
Alessandro Ghedini 773db053aa build: check for libatomic and __atomic operations
Add check in old-configure as well. Reformat the check to use a maximum of 80
columns in the wscript.
Signed-off-by: Stefano Pigozzi <stefano.pigozzi@gmail.com>
2014-01-01 20:57:06 +01:00
wm4 a8c09d003f manpage: fix vf_noise parameter name 2014-01-01 20:57:02 +01:00
wm4 1284a6b6b0 video: fix --brightness etc. options
They were set before the VO was intitialized, which silently failed.
2013-12-29 22:17:50 +01:00
wm4 78c35e42ea Release 0.3.0 2013-12-29 18:06:41 +01:00
1376 changed files with 123758 additions and 277705 deletions

No files matched your search

-26
View File
@@ -1,26 +0,0 @@
# To use this config on you editor, follow the instructions at:
# http://editorconfig.org
root = true
[*]
charset = utf-8
insert_final_newline = true
indent_style = space
indent_size = 4
max_line_length = 80
trim_trailing_whitespace = true
[.git/COMMIT*]
max_line_length = 72
[{osdep/dirent-win.h,TOOLS/osxbundle/mpv.app/Contents/PkgInfo,TOOLS/osxbundle/mpv.app/Contents/Resources/Assets.car}]
charset = unset
insert_final_newline = unset
indent_style = unset
indent_size = unset
max_line_length = unset
trim_trailing_whitespace = unset
[menu.conf]
indent_style = tab
-7
View File
@@ -1,7 +0,0 @@
{
"Format": "github-actions",
"Disable": {
"IndentSize": true,
"MaxLineLength": true
}
}
@@ -1,160 +0,0 @@
name: "Report a Linux Issue"
description: "Create a report for a runtime related Linux Issue"
labels: ["os:linux"]
type: Bug
body:
- type: textarea
id: mpv_information
attributes:
label: "mpv Information"
placeholder: |
mpv v0.38.0-96-gf776be6af5 Copyright © 2000-2024 mpv/MPlayer/mplayer2 projects
built on May 5 2024 18:37:43
libplacebo version: v7.349.0 (v6.338.0-120-g7b294350)
FFmpeg version: 6.1.1
FFmpeg library versions:
libavutil 58.29.100
libavcodec 60.31.102
libavformat 60.16.100
libswscale 7.5.100
libavfilter 9.12.100
libswresample 4.12.100
description: |
Provide the output of `mpv --version`.
render: bash
validations:
required: false
- type: textarea
id: other_information
attributes:
label: "Other Information"
description: >
Provide following information if applicable.
- Linux Version: `cat /etc/os-release | grep "NAME"`
- Kernel Version: `uname -a`
- GPU Model: `lspci -nn | grep VGA` or `lshw -C display -numeric`
- Mesa/GPU Driver Version: `glxinfo -B | grep "OpenGL version string"` or `vulkaninfo | grep driverInfo`
- Window Manager and Version
- Source of the mpv binary
- Latest version of mpv you know without the problem
- The time when the problem started to happen (after updating mpv, system, driver, etc.)
- Possible screenshot or video of visual glitches
If you're not using the git master branch or the [latest release](https://github.com/mpv-player/mpv/releases/latest), please update.
Older versions are not supported. If your distribution does not provide the latest version, please report the issue to them.
Packages and builds can be found [here](https://mpv.io/installation/).
value: |-
- Linux version:
- Kernel Version:
- GPU Model:
- Mesa/GPU Driver Version:
- Window Manager and Version:
- Source of mpv:
- Latest known working version:
- Issue started after the following happened:
render: text
validations:
required: true
- type: textarea
id: reproduction_steps
attributes:
label: "Reproduction Steps"
description: >
Try to reproduce your issue with `--no-config` first. If it isn't reproducible
with `--no-config` try to find out which option or script causes your issue.
If the issue is render performance-related, try to reproduce it with `--no-config
--profile=fast`, which is designed to work on lower-end mobile devices.
For decoding performance-related issues, additionally try to enable hardware
decoding with `--hwdec=yes` or the `Ctrl+h` shortcut.
Describe the reproduction steps as precise as possible. It's very likely that
the bug you experience wasn't reproduced by the developer because the workflow
differs from your own.
validations:
required: true
- type: textarea
id: expected_behavior
attributes:
label: "Expected Behavior"
placeholder: "What were you expecting?"
validations:
required: true
- type: textarea
id: actual_behavior
attributes:
label: "Actual Behavior"
placeholder: "What happened instead?"
validations:
required: true
- type: textarea
id: log_file
attributes:
label: "Log File"
placeholder: "Drag and drop log file here (Don't paste content directly)"
description: >
Make a log file with `--gpu-debug --log-file=output.txt`, which is our preferred way, and
attach it to the issue.
If you have trouble producing a log file, you can alternatively use `--gpu-debug --msg-time -v -v`,
save the terminal output to a file, and attach it to the issue.
In the case of a crash, please provide a [backtrace](https://wiki.archlinux.org/title/Debugging/Getting_traces).
Without the log file, this issue will be closed for ignoring the issue template.
Disclaimer: The log file may contain identifiable information, such as your username.
Review the log file before sharing and redact any sensitive data if necessary.
validations:
required: true
- type: textarea
id: sample_files
attributes:
label: "Sample Files"
placeholder: "Drag and drop sample files here"
description: >
Please provide sample files for a minimal reproduction case, if the issue can't be
reproduced without it or it is helpful to demonstrate the issue.
Preferably files should be attached directly within the text field below. Due to GitHub's
upload restriction, you may need to archive the sample file first and upload it as `.zip`
file. If the sample file is too big to attach directly, use a solution that is easy to
access with possible long-term retention. Possible services include, but are not limited
to, the following:
- [0x0.st](https://0x0.st/) (no long-term retention)
- [Dropbox](https://www.dropbox.com)
- [Google Drive](https://www.google.com/drive/)
- [iCLoud Drive](https://www.icloud.com/iclouddrive/)
- [MediaFire](https://www.mediafire.com/)
- [OneDrive](https://onedrive.live.com/)
- type: checkboxes
attributes:
label: "I carefully read all instruction and confirm that I did the following:"
options:
- label: "I tested and confirmed that the issue exists with the [latest release version](https://github.com/mpv-player/mpv/releases/latest) or newer."
required: true
- label: "I provided all required information including system and mpv version."
required: true
- label: >-
I produced the log file with the exact same set of files, parameters, and conditions
used in "Reproduction Steps", with the addition of `--log-file=output.txt`.
required: true
- label: >-
I produced the log file while the behaviors described in "Actual Behavior" were
actively observed.
required: true
- label: "I attached the full, untruncated log file."
required: true
- label: "I attached the backtrace in the case of a crash."
required: true
@@ -1,153 +0,0 @@
name: "Report a macOS Issue"
description: "Create a report for a runtime related macOS Issue"
labels: ["os:mac"]
type: Bug
body:
- type: textarea
id: mpv_information
attributes:
label: "mpv Information"
placeholder: |
mpv v0.38.0-96-gf776be6af5 Copyright © 2000-2024 mpv/MPlayer/mplayer2 projects
built on May 5 2024 18:37:43
libplacebo version: v7.349.0 (v6.338.0-120-g7b294350)
FFmpeg version: 6.1.1
FFmpeg library versions:
libavutil 58.29.100
libavcodec 60.31.102
libavformat 60.16.100
libswscale 7.5.100
libavfilter 9.12.100
libswresample 4.12.100
description: |
Provide the output of `mpv --version`.
render: bash
validations:
required: false
- type: textarea
id: other_info
attributes:
label: "Other Information"
description: >
- macOS Version: `system_profiler SPSoftwareDataType | sed -rn 's/^ *System Version: (.*$)/\1/p'`
- Source of the mpv binary or bundle
- Latest version of mpv you know without the problem
- The time when the problem started to happen (after updating mpv, system, driver, etc.)
- Possible screenshot or video of visual glitches
If you're not using git master or the latest release, update.
Nightly builds from master can be found [here](https://github.com/mpv-player/mpv/actions/workflows/build.yml?query=branch%3Amaster+event%3Apush),
click on the latest workflow run, scroll down to the Artifacts section and download the
appropriate build. Packages and builds can be found [here](https://mpv.io/installation/).
value: |-
- macOS version:
- Source of mpv:
- Latest known working version:
- Issue started after the following happened:
render: text
validations:
required: true
- type: textarea
id: reproduction_steps
attributes:
label: "Reproduction Steps"
description: >
Try to reproduce your issue with `--no-config` first. If it isn't reproducible
with `--no-config` try to find out which option or script causes your issue.
If the issue is render performance-related, try to reproduce it with `--no-config
--profile=fast`, which is designed to work on lower-end mobile devices.
For decoding performance-related issues, additionally try to enable hardware
decoding with `--hwdec=yes` or the `Ctrl+h` shortcut.
Describe the reproduction steps as precise as possible. It's very likely that
the bug you experience wasn't reproduced by the developer because the workflow
differs from your own.
validations:
required: true
- type: textarea
id: expected_behavior
attributes:
label: "Expected Behavior"
placeholder: "What were you expecting?"
validations:
required: true
- type: textarea
id: actual_behavior
attributes:
label: "Actual Behavior"
placeholder: "What happened instead?"
validations:
required: true
- type: textarea
id: log_file
attributes:
label: "Log File"
placeholder: "Drag and drop log file here (Don't paste content directly)"
description: >
Make a log file with `--gpu-debug --log-file=output.txt`, which is our preferred way, and
attach it to the issue. If you use the Bundle, a default log is created for your last run
at `~/Library/Logs/mpv.log`. You can jump to that file via the `Help > Show log File…` menu.
If you have trouble producing a log file, you can alternatively use `--gpu-debug --msg-time -v -v`,
save the terminal output to a file, and attach it to the issue.
In the case of a crash, please provide the macOS Crash Report or create a [backtrace](https://developer.apple.com/library/archive/documentation/General/Conceptual/lldb-guide/chapters/C5-Examining-The-Call-Stack.html)
yourself.
Without the log file, this issue will be closed for ignoring the issue template.
Disclaimer: The log file may contain identifiable information, such as your username.
Review the log file before sharing and redact any sensitive data if necessary.
validations:
required: true
- type: textarea
id: sample_files
attributes:
label: "Sample Files"
placeholder: "Drag and drop sample files here"
description: >
Please provide sample files for a minimal reproduction case, if the issue can't be
reproduced without it or it is helpful to demonstrate the issue.
Preferably files should be attached directly within the text field below. Due to GitHub's
upload restriction, you may need to archive the sample file first and upload it as `.zip`
file. If the sample file is too big to attach directly, use a solution that is easy to
access with possible long-term retention. Possible services include, but are not limited
to, the following:
- [0x0.st](https://0x0.st/) (no long-term retention)
- [Dropbox](https://www.dropbox.com)
- [Google Drive](https://www.google.com/drive/)
- [iCLoud Drive](https://www.icloud.com/iclouddrive/)
- [MediaFire](https://www.mediafire.com/)
- [OneDrive](https://onedrive.live.com/)
- type: checkboxes
attributes:
label: "I carefully read all instruction and confirm that I did the following:"
options:
- label: "I tested and confirmed that the issue exists with the [latest release version](https://github.com/mpv-player/mpv/releases/latest) or newer."
required: true
- label: "I provided all required information including system and mpv version."
required: true
- label: >-
I produced the log file with the exact same set of files, parameters, and conditions
used in "Reproduction Steps", with the addition of `--log-file=output.txt`.
required: true
- label: >-
I produced the log file while the behaviors described in "Actual Behavior" were
actively observed.
required: true
- label: "I attached the full, untruncated log file."
required: true
- label: "I attached the backtrace in the case of a crash."
required: true
@@ -1,154 +0,0 @@
name: "Report a Windows Issue"
description: "Create a report for a runtime related Windows Issue"
labels: ["os:win"]
type: Bug
body:
- type: textarea
id: mpv_information
attributes:
label: "mpv Information"
placeholder: |
mpv v0.38.0-96-gf776be6af5 Copyright © 2000-2024 mpv/MPlayer/mplayer2 projects
built on May 5 2024 18:37:43
libplacebo version: v7.349.0 (v6.338.0-120-g7b294350)
FFmpeg version: 6.1.1
FFmpeg library versions:
libavutil 58.29.100
libavcodec 60.31.102
libavformat 60.16.100
libswscale 7.5.100
libavfilter 9.12.100
libswresample 4.12.100
description: |
Provide the output of `mpv --version`.
render: bash
validations:
required: false
- type: textarea
id: other_information
attributes:
label: "Other Information"
description: >
- Windows Version: `cmd /c ver`
- GPU model, driver and version
- Source of the mpv binary
- Latest version of mpv you know without the problem
- The time when the problem started to happen (after updating mpv, system, driver, etc.)
- Possible screenshot or video of visual glitches
If you're not using git master or the latest release, update.
Nightly builds from master can be found [here](https://github.com/mpv-player/mpv/actions/workflows/build.yml?query=branch%3Amaster+event%3Apush),
click on the latest workflow run, scroll down to the Artifacts section and download the
appropriate build. Packages and builds can be found [here](https://mpv.io/installation/).
value: |-
- Windows version:
- GPU model, driver and version:
- Source of mpv:
- Latest known working version:
- Issue started after the following happened:
render: text
validations:
required: true
- type: textarea
id: reproduction_steps
attributes:
label: "Reproduction Steps"
description: >
Try to reproduce your issue with `--no-config` first. If it isn't reproducible
with `--no-config` try to find out which option or script causes your issue.
If the issue is render performance-related, try to reproduce it with `--no-config
--profile=fast`, which is designed to work on lower-end mobile devices.
For decoding performance-related issues, additionally try to enable hardware
decoding with `--hwdec=yes` or the `Ctrl+h` shortcut.
Describe the reproduction steps as precise as possible. It's very likely that
the bug you experience wasn't reproduced by the developer because the workflow
differs from your own.
validations:
required: true
- type: textarea
id: expected_behavior
attributes:
label: "Expected Behavior"
placeholder: "What were you expecting?"
validations:
required: true
- type: textarea
id: actual_behavior
attributes:
label: "Actual Behavior"
placeholder: "What happened instead?"
validations:
required: true
- type: textarea
id: log_file
attributes:
label: "Log File"
placeholder: "Drag and drop log file here (Don't paste content directly)"
description: >
Make a log file with `--gpu-debug --log-file=output.txt`, which is our preferred way, and
attach it to the issue.
If you have trouble producing a log file, you can alternatively use `--gpu-debug --msg-time -v -v`,
save the terminal output to a file, and attach it to the issue.
In the case of a crash, please provide a backtrace, attach the appropriate `.dmp` file
from `%LOCALAPPDATA%\CrashDumps`.
Without the log file, this issue will be closed for ignoring the issue template.
Disclaimer: The log file may contain identifiable information, such as your username.
Review the log file before sharing and redact any sensitive data if necessary.
validations:
required: true
- type: textarea
id: sample_files
attributes:
label: "Sample Files"
placeholder: "Drag and drop sample files here"
description: >
Please provide sample files for a minimal reproduction case, if the issue can't be
reproduced without it or it is helpful to demonstrate the issue.
Preferably files should be attached directly within the text field below. Due to GitHub's
upload restriction, you may need to archive the sample file first and upload it as `.zip`
file. If the sample file is too big to attach directly, use a solution that is easy to
access with possible long-term retention. Possible services include, but are not limited
to, the following:
- [0x0.st](https://0x0.st/) (no long-term retention)
- [Dropbox](https://www.dropbox.com)
- [Google Drive](https://www.google.com/drive/)
- [iCLoud Drive](https://www.icloud.com/iclouddrive/)
- [MediaFire](https://www.mediafire.com/)
- [OneDrive](https://onedrive.live.com/)
- type: checkboxes
attributes:
label: "I carefully read all instruction and confirm that I did the following:"
options:
- label: "I tested and confirmed that the issue exists with the [latest release version](https://github.com/mpv-player/mpv/releases/latest) or newer."
required: true
- label: "I provided all required information including system and mpv version."
required: true
- label: >-
I produced the log file with the exact same set of files, parameters, and conditions
used in "Reproduction Steps", with the addition of `--log-file=output.txt`.
required: true
- label: >-
I produced the log file while the behaviors described in "Actual Behavior" were
actively observed.
required: true
- label: "I attached the full, untruncated log file."
required: true
- label: "I attached the backtrace in the case of a crash."
required: true
-151
View File
@@ -1,151 +0,0 @@
name: "Report an Issue on other platforms"
description: "Create a report for a runtime related Issue on other platforms"
labels: []
type: Bug
body:
- type: textarea
id: mpv_information
attributes:
label: "mpv Information"
placeholder: |
mpv v0.38.0-96-gf776be6af5 Copyright © 2000-2024 mpv/MPlayer/mplayer2 projects
built on May 5 2024 18:37:43
libplacebo version: v7.349.0 (v6.338.0-120-g7b294350)
FFmpeg version: 6.1.1
FFmpeg library versions:
libavutil 58.29.100
libavcodec 60.31.102
libavformat 60.16.100
libswscale 7.5.100
libavfilter 9.12.100
libswresample 4.12.100
description: |
Provide the output of `mpv --version`.
render: bash
validations:
required: false
- type: textarea
id: important_information
attributes:
label: "Important Information"
description: >
- Platform and Version
- GPU model, driver and version
- Source of the mpv binary
- Latest version of mpv you know without the problem
- The time when the problem started to happen (after updating mpv, system, driver, etc.)
- Possible screenshot or video of visual glitches
If you're not using git master or the latest release, update.
Packages and builds can be found [here](https://mpv.io/installation/).
value: |-
- Platform version:
- GPU model, driver and version:
- Source of mpv:
- Latest known working version:
- Issue started after the following happened:
render: text
validations:
required: true
- type: textarea
id: reproduction_steps
attributes:
label: "Reproduction Steps"
description: >
Try to reproduce your issue with `--no-config` first. If it isn't reproducible
with `--no-config` try to find out which option or script causes your issue.
If the issue is render performance-related, try to reproduce it with `--no-config
--profile=fast`, which is designed to work on lower-end mobile devices.
For decoding performance-related issues, additionally try to enable hardware
decoding with `--hwdec=yes` or the `Ctrl+h` shortcut.
Describe the reproduction steps as precise as possible. It's very likely that
the bug you experience wasn't reproduced by the developer because the workflow
differs from your own.
validations:
required: true
- type: textarea
id: expected_behavior
attributes:
label: "Expected Behavior"
placeholder: "What were you expecting?"
validations:
required: true
- type: textarea
id: actual_behavior
attributes:
label: "Actual Behavior"
placeholder: "What happened instead?"
validations:
required: true
- type: textarea
id: log_file
attributes:
label: "Log File"
placeholder: "Drag and drop log file here (Don't paste content directly)"
description: >
Make a log file with `--gpu-debug --log-file=output.txt`, which is our preferred way, and
attach it to the issue.
If you have trouble producing a log file, you can alternatively use `--gpu-debug --msg-time -v -v`,
save the terminal output to a file, and attach it to the issue.
In the case of a crash, please provide a backtrace.
Without the log file, this issue will be closed for ignoring the issue template.
Disclaimer: The log file may contain identifiable information, such as your username.
Review the log file before sharing and redact any sensitive data if necessary.
validations:
required: true
- type: textarea
id: sample_files
attributes:
label: "Sample Files"
placeholder: "Drag and drop sample files here"
description: >
Please provide sample files for a minimal reproduction case, if the issue can't be
reproduced without it or it is helpful to demonstrate the issue.
Preferably files should be attached directly within the text field below. Due to GitHub's
upload restriction, you may need to archive the sample file first and upload it as `.zip`
file. If the sample file is too big to attach directly, use a solution that is easy to
access with possible long-term retention. Possible services include, but are not limited
to, the following:
- [0x0.st](https://0x0.st/) (no long-term retention)
- [Dropbox](https://www.dropbox.com)
- [Google Drive](https://www.google.com/drive/)
- [iCLoud Drive](https://www.icloud.com/iclouddrive/)
- [MediaFire](https://www.mediafire.com/)
- [OneDrive](https://onedrive.live.com/)
- type: checkboxes
attributes:
label: "I carefully read all instruction and confirm that I did the following:"
options:
- label: "I tested and confirmed that the issue exists with the [latest release version](https://github.com/mpv-player/mpv/releases/latest) or newer."
required: true
- label: "I provided all required information including system and mpv version."
required: true
- label: >-
I produced the log file with the exact same set of files, parameters, and conditions
used in "Reproduction Steps", with the addition of `--log-file=output.txt`.
required: true
- label: >-
I produced the log file while the behaviors described in "Actual Behavior" were
actively observed.
required: true
- label: "I attached the full, untruncated log file."
required: true
- label: "I attached the backtrace in the case of a crash."
required: true
@@ -1,76 +0,0 @@
name: "Report a build Problem"
description: "Create a report for a build related Issue"
labels: ["core:meson"]
type: Bug
body:
- type: textarea
id: important_information
attributes:
label: "Important Information"
description: >
- mpv version
- Platform and Version
- Meson version
If you're not using git master or the latest release, update.
Releases are listed [here](https://github.com/mpv-player/mpv/releases).
value: |-
- mpv version:
- Platform version:
- Meson version:
render: text
validations:
required: true
- type: textarea
id: reproduction_steps
attributes:
label: "Reproduction Steps"
description: >
Describe the reproduction steps as precise as possible. It's very likely that
the bug you experience wasn't reproduced by the developer because the workflow
differs from your own.
validations:
required: true
- type: textarea
id: expected_behavior
attributes:
label: "Expected Behavior"
placeholder: "What were you expecting?"
validations:
required: true
- type: textarea
id: actual_behavior
attributes:
label: "Actual Behavior"
placeholder: "What happened instead?"
validations:
required: true
- type: textarea
id: log_file
attributes:
label: "Log File"
placeholder: "Drag and drop log file here (Don't paste content directly)"
description: >
Attach meson-logs/meson-log.txt or meson setup build output to the issue.
Without the log file, this issue will be closed for ignoring the issue template.
validations:
required: true
- type: checkboxes
attributes:
label: "I carefully read all instruction and confirm that I did the following:"
options:
- label: "I tested and confirmed that the issue exists with the [latest release version](https://github.com/mpv-player/mpv/releases/latest) or newer."
required: true
- label: "I provided all required information including system and mpv version."
required: true
- label: >-
I produced the log file with the exact same set of parameters, and conditions used in
"Reproduction Steps".
required: true
- label: >-
I produced the log file while the behaviors described in "Actual Behavior" were
actively observed.
required: true
- label: "I attached the full, untruncated log file."
required: true
@@ -1,60 +0,0 @@
name: "Request a new Feature"
description: "Create a request for a new feature"
labels: ["meta:feature-request"]
type: Feature
body:
- type: textarea
id: expected_behavior
attributes:
label: "Expected behavior of the wanted feature"
description: >
Before requesting a new feature, make sure it hasn't been
[requested yet](https://github.com/mpv-player/mpv/labels/meta%3Afeature-request),
and isn't already supported by [mpv](https://mpv.io/manual/master/),
[ffmpeg](https://ffmpeg.org/documentation.html), or existing
[mpv scripts](https://github.com/mpv-player/mpv/wiki/User-Scripts).
If you are not sure about this, ask on
[mpv Discussions](https://github.com/mpv-player/mpv/discussions) and
[mpv IRC channel](https://mpv.io/community/) first before opening a feature request.
validations:
required: true
- type: textarea
id: alternative_behavior
attributes:
label: "Alternative behavior of the wanted feature"
- type: textarea
id: log_file
attributes:
label: "Log File"
description: >
Even if you think it's not necessary at first, it might help us later to find
possible issues. Make a log file with `--log-file=output.txt`, which is our preferred way,
and attach it to the issue.
If you have trouble producing a log file, you can alternatively use `-v -v`, save the
terminal output to a file, and attach it to the issue.
- type: textarea
id: sample_files
attributes:
label: "Sample Files"
placeholder: "Drag and drop sample files here"
description: >
Please provide sample files in the case you request a format/codec/container dependent
feature, and it could help us test and implement it.
Preferably files should be attached directly within the text field below. Due to GitHub's
upload restriction, you may need to archive the sample file first and upload it as `.zip`
file. If the sample file is too big to attach directly, use a solution that is easy to
access with possible long-term retention. Possible services include, but are not limited
to, the following:
- [0x0.st](https://0x0.st/) (no long-term retention)
- [Dropbox](https://www.dropbox.com)
- [Google Drive](https://www.google.com/drive/)
- [iCLoud Drive](https://www.icloud.com/iclouddrive/)
- [MediaFire](https://www.mediafire.com/)
- [OneDrive](https://onedrive.live.com/)
-12
View File
@@ -1,12 +0,0 @@
blank_issues_enabled: false
contact_links:
- name: mpv Discussions
url: https://github.com/mpv-player/mpv/discussions
about: If you have questions or want to discuss things use our GitHub Discussions
- name: mpv IRC channel
url: https://mpv.io/community/
about: Feel free to ask questions here irc://irc.libera.chat/mpv
- name: mpv IRC developer channel
url: https://mpv.io/community/
about: Ask questions related to the development of mpv here
irc://irc.libera.chat/mpv-devel
-5
View File
@@ -1,5 +0,0 @@
Read this before you submit this pull request:
https://github.com/mpv-player/mpv/blob/master/DOCS/contribute.md
Reading this link and following the rules will get your pull request reviewed
and merged faster. Nobody wants lazy pull requests.
-8
View File
@@ -1,8 +0,0 @@
version: 2
updates:
- package-ecosystem: "github-actions"
directory: "/"
schedule:
interval: "weekly"
commit-message:
prefix: "ci"
-830
View File
@@ -1,830 +0,0 @@
name: build
on:
push:
branches:
- master
- ci
- 'release/**'
paths-ignore:
- 'DOCS/**'
- 'TOOLS/lua/**'
- '.editorconfig'
- '.github/ISSUE_TEMPLATE/**'
- '.github/PULL_REQUEST_TEMPLATE'
- '.gitignore'
- 'Copyright'
- 'README.md'
- 'RELEASE_NOTES'
pull_request:
branches: [master]
paths-ignore:
- 'DOCS/**'
- 'TOOLS/lua/**'
- '.editorconfig'
- '.github/ISSUE_TEMPLATE/**'
- '.github/PULL_REQUEST_TEMPLATE'
- '.gitignore'
- 'Copyright'
- 'README.md'
- 'RELEASE_NOTES'
concurrency:
group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
cancel-in-progress: true
jobs:
mingw:
runs-on: ubuntu-24.04
env:
CCACHE_BASEDIR: ${{ github.workspace }}
CCACHE_DIR: ${{ github.workspace }}/.ccache
CCACHE_MAXSIZE: 500M
DOWNLOAD_CACHE: ${{ github.workspace }}/.ccache/downloads
TARGET: ${{ matrix.arch }}-w64-mingw32
RUST_TARGET: ${{ matrix.arch }}-pc-windows-gnu
strategy:
fail-fast: false
matrix:
include:
- arch: i686
wine: wine32
- arch: x86_64
wine: wine64
steps:
- uses: actions/checkout@v7
- name: Get time
id: get_time
run: echo "timestamp=`date +%s%N`" >> $GITHUB_OUTPUT
- name: Restore cache
uses: actions/cache/restore@v6
with:
path: ${{ env.CCACHE_DIR }}
key: ${{ env.TARGET }}-${{ steps.get_time.outputs.timestamp }}
restore-keys: ${{ env.TARGET }}-
- name: Install dependencies
run: |
rustup target add ${{ env.RUST_TARGET }}
sudo dpkg --add-architecture i386
sudo apt-get update
sudo apt-get install --no-install-recommends -y ccache g++-mingw-w64 gcc-multilib python3-pip nasm ninja-build pkgconf ${{ matrix.wine }} wine
python -m pip install meson
- name: Install Meson Wraps
run: |
mkdir subprojects -p
meson wrap install mujs
meson subprojects download
- name: Build libraries
run: |
./ci/build-mingw64.sh
- name: Build with meson
id: build
run: |
./ci/build-mingw64.sh meson pack
- name: Print meson log
if: ${{ failure() && steps.build.outcome == 'failure' }}
run: |
cat ./mingw_build/meson-logs/meson-log.txt
- name: Functional test
run: |
cd artifact && wine ./mpv.com -v --no-config
env:
WINEDEBUG: '+loaddll'
- name: Run meson tests
id: tests
run: |
meson test -C mingw_build
- name: Print meson test log
if: ${{ failure() && steps.tests.outcome == 'failure' }}
run: |
cat ./mingw_build/meson-logs/testlog.txt
- name: Artifact name
id: artifact
run: |
VERSION_FILE=$(find . -path "*/common/version.h" -type f | head -n 1)
VER=$(grep VERSION "$VERSION_FILE" | cut -d'"' -f2)
PR=${{ github.event.pull_request.number && format('-pr{0}', github.event.pull_request.number) || '' }}
echo "name=mpv-$VER$PR-${{ github.run_id }}-$TARGET" >> "$GITHUB_OUTPUT"
- uses: actions/upload-artifact@v7
with:
name: ${{ steps.artifact.outputs.name }}
path: artifact/*
- name: Save Cache
uses: actions/cache/save@v6
if: always()
with:
path: ${{ env.CCACHE_DIR }}
key: ${{ env.TARGET }}-${{ steps.get_time.outputs.timestamp }}
win32:
runs-on: ${{ matrix.os }}
strategy:
fail-fast: false
matrix:
include:
- target: x86_64-pc-windows-msvc
arch: "x64"
os: "windows-2025-vs2026"
- target: aarch64-pc-windows-msvc
arch: "arm64"
os: "windows-11-arm"
env:
CC: "ccache clang"
CXX: "ccache clang++"
CC_LD: "lld-link"
CXX_LD: "lld-link"
RUST_LD: "lld-link"
WINDRES: "llvm-rc"
CCACHE_BASEDIR: ${{ github.workspace }}
CCACHE_DIR: "${{ github.workspace }}\\.ccache"
CCACHE_MAXSIZE: 500M
steps:
- name: Disable autocrlf
run: |
git config --global core.autocrlf false
git config --global core.eol lf
- uses: actions/checkout@v7
- name: Get time
id: get_time
run: |
"timestamp=$((Get-Date).Ticks)" >> $env:GITHUB_OUTPUT
- name: Restore cache
uses: actions/cache/restore@v6
with:
path: ${{ env.CCACHE_DIR }}
key: ${{ matrix.target }}-${{ steps.get_time.outputs.timestamp }}
restore-keys: ${{ matrix.target }}-
- name: Install Winget on windows-11-arm
if: matrix.os == 'windows-11-arm'
shell: pwsh
run: |
Install-Module -Name Microsoft.WinGet.Client -Repository PSGallery -Force
Repair-WinGetPackageManager -Latest -Force
- name: Install dependencies
run: |
python -m pip install meson
winget install --accept-source-agreements --accept-package-agreements Ccache.Ccache NASM.NASM
echo "$env:LOCALAPPDATA\Microsoft\WinGet\Links" >> $env:GITHUB_PATH
# For cargo -C
rustup default nightly
- name: Update Meson WrapDB
run: |
meson wrap update-db
# Explicitly download wraps, as nested projects may have older versions of them.
meson wrap install expat
meson wrap install harfbuzz
meson wrap install libpng
meson wrap install zlib
- name: Find Visual Studio
run: |
$vswhere = "${env:ProgramFiles(x86)}\Microsoft Visual Studio\Installer\vswhere.exe"
$vsPath = & $vswhere -latest -products * -requires Microsoft.VisualStudio.Component.VC.Tools.x86.x64 -property installationPath
if (-not $vsPath) { throw "Could not find a Visual Studio installation" }
"VS=$vsPath"
"VS=$vsPath" >> $env:GITHUB_ENV
- name: Build
id: build
run: |
$env:PATH = ($env:PATH -split ';' | Where-Object { $_ -ne 'C:\Program Files\LLVM\bin' -and `
$_ -ne 'C:\Program Files\CMake\bin' -and `
$_ -ne 'C:\Strawberry\c\bin' }) -join ';'
$env:PATH += ';C:\Program Files\NASM'
Import-Module "$env:VS\Common7\Tools\Microsoft.VisualStudio.DevShell.dll"
Enter-VsDevShell -VsInstallPath $env:VS -SkipAutomaticLocation -DevCmdArguments "-arch=${{ matrix.arch }} -host_arch=${{ matrix.arch }}"
./ci/build-win32.ps1
- name: Print build log
if: ${{ failure() && steps.build.outcome == 'failure' }}
run: |
cat ./build/meson-logs/meson-log.txt
- name: Run meson tests
id: tests
run: |
$env:PATH = ($env:PATH -split ';' | Where-Object { $_ -ne 'C:\Program Files\LLVM\bin' -and `
$_ -ne 'C:\Program Files\CMake\bin' -and `
$_ -ne 'C:\Strawberry\c\bin' }) -join ';'
$env:PATH += ';C:\Program Files\NASM'
Import-Module "$env:VS\Common7\Tools\Microsoft.VisualStudio.DevShell.dll"
Enter-VsDevShell -VsInstallPath $env:VS -SkipAutomaticLocation -DevCmdArguments "-arch=${{ matrix.arch }} -host_arch=${{ matrix.arch }}"
meson test -C build -t 2
- name: Print meson test log
if: ${{ failure() && steps.tests.outcome == 'failure' }}
run: |
cat ./build/meson-logs/testlog.txt
- name: Artifact name
id: artifact
run: |
$VERSION_FILE = gci . -r -Filter "version.h" | ? { $_.FullName -match "[/\\]common[/\\]version\.h$" } | select -First 1 -exp FullName
$VER = (sls $VERSION_FILE -Pattern 'VERSION').Line -replace '.*"([^"]+)".*', '$1'
$PR = if ("${{ github.event.pull_request.number }}") { "-pr${{ github.event.pull_request.number }}" } else { "" }
$name = "mpv-$VER$PR-${{ github.run_id }}-${{ matrix.target }}"
"name=$name" >> $env:GITHUB_OUTPUT
- uses: actions/upload-artifact@v7
with:
name: ${{ steps.artifact.outputs.name }}
path: |
build/mpv.???
build/mpv-*.bat
build/vulkan-*.dll
!build/mpv.lib
!build/mpv.pdb
- uses: actions/upload-artifact@v7
with:
name: ${{ steps.artifact.outputs.name }}-pdb
path: |
build/mpv.pdb
- name: Save Cache
uses: actions/cache/save@v6
if: always()
with:
path: ${{ env.CCACHE_DIR }}
key: ${{ matrix.target }}-${{ steps.get_time.outputs.timestamp }}
macos:
runs-on: ${{ matrix.os }}
strategy:
matrix:
include:
- os: "macos-14"
arch: "arm"
xcode: "Xcode_15.0"
- os: "macos-15"
arch: "arm"
- os: "macos-15-intel"
- os: "macos-26"
arch: "arm"
- os: "macos-26"
arch: "test"
swift: "-target arm64-apple-macosx10.15"
target: "10.15"
steps:
- uses: actions/checkout@v7
- name: Remove stray upstream python binary symlinks under /usr/local
run: |
find /usr/local/bin -lname '*/Library/Frameworks/Python.framework/*' -delete -print
brew unlink python && brew link --overwrite python
- name: Unlink pkg-config if the CI runner happens to have it already installed
run: |
find "$(brew config | grep HOMEBREW_PREFIX | cut -f 2- -d ' ')/bin" -lname '*/pkg-config@0*/*' -print -exec brew unlink pkg-config@0.29.2 \; -quit
- name: Change Xcode version
if: ${{ matrix.xcode != '' }}
run: |
sudo xcode-select -s /Applications/${{ matrix.xcode }}.app
- name: Install dependencies
run: |
if [[ -n "${{ matrix.homebrew }}" ]]; then
export HOMEBREW_NO_AUTO_UPDATE=1
cd "$(brew --repo homebrew/core)"
git checkout --force ${{ matrix.homebrew }}
else
brew update
fi
if brew list ffmpeg &> /dev/null; then brew unlink ffmpeg; fi
brew install -q autoconf automake curl pkgconf libtool python freetype fribidi little-cms2 \
luajit libass ffmpeg-full meson rust uchardet mujs libplacebo molten-vk vulkan-loader vulkan-headers \
libarchive libbluray libcaca libcdio-paranoia libdvdnav rubberband zimg
brew link ffmpeg-full
- name: Build with meson
id: build
run: |
./ci/build-macos.sh
env:
TRAVIS_OS_NAME: "${{ matrix.os }}"
SWIFT_FLAGS: "${{ matrix.swift }}"
MACOSX_DEPLOYMENT_TARGET: "${{ matrix.target }}"
- name: Create App Bundle
if: ${{ matrix.arch != 'test' }}
run: |
meson compile -C build macos-bundle
tar -czvf mpv.tar.gz -C build mpv.app
- name: Print meson log
if: ${{ failure() && steps.build.outcome == 'failure' }}
run: |
cat ./build/meson-logs/meson-log.txt
- name: Run meson tests
id: tests
run: |
meson test -C build
- name: Print meson test log
if: ${{ failure() && steps.tests.outcome == 'failure' }}
run: |
cat ./build/meson-logs/testlog.txt
- name: Artifact name
id: artifact
run: |
VERSION_FILE=$(find . -path "*/common/version.h" -type f | head -n 1)
VER=$(grep VERSION "$VERSION_FILE" | cut -d'"' -f2)
TARGET=${{ matrix.os }}${{ matrix.arch != '' && format('-{0}', matrix.arch) || '' }}
PR=${{ github.event.pull_request.number && format('-pr{0}', github.event.pull_request.number) || '' }}
echo "name=mpv-$VER$PR-${{ github.run_id }}-$TARGET" >> "$GITHUB_OUTPUT"
- uses: actions/upload-artifact@v7
if: ${{ matrix.arch != 'test' }}
with:
name: ${{ steps.artifact.outputs.name }}
path: mpv.tar.gz
linux:
runs-on: ubuntu-latest
container:
image: "registry.opensuse.org/home/mia/images/images/mpv-ci:stable-deps"
env:
CC: "${{ matrix.config.cc }}"
CXX: "${{ matrix.config.cxx }}"
CC_LD: "${{ matrix.config.ld }}"
CXX_LD: "${{ matrix.config.ld }}"
strategy:
matrix:
config:
- cc: "gcc"
cxx: "g++"
args: "-Db_ndebug=true"
- cc: "clang"
cxx: "clang++"
ld: "lld"
args: "-Db_lto=true -Db_lto_mode=thin"
steps:
- uses: actions/checkout@v7
- name: Build with meson
id: build
run: |
./ci/build-tumbleweed.sh ${{ matrix.config.args }}
- name: Print meson log
if: ${{ failure() && steps.build.outcome == 'failure' }}
run: |
cat ./build/meson-logs/meson-log.txt
- name: Run meson tests
id: tests
run: |
meson test -C build
- name: Print meson test log
if: ${{ failure() && steps.tests.outcome == 'failure' }}
run: |
cat ./build/meson-logs/testlog.txt
linux-fuzz:
runs-on: ubuntu-latest
container:
image: "registry.opensuse.org/home/mia/images/images/mpv-ci:stable-deps"
env:
CC: "clang"
CXX: "clang++"
steps:
- uses: actions/checkout@v7
- name: Build with meson
id: build
run: |
meson setup build \
--werror \
-Dfuzzers=true -Dlibmpv=true -Dcplayer=false
meson compile -C build fuzzers
- name: Print meson log
if: ${{ failure() && steps.build.outcome == 'failure' }}
run: |
cat ./build/meson-logs/meson-log.txt
linux-ffmpeg-6-1:
runs-on: ubuntu-latest
container:
# We want to test ffmpeg 6.1.2, which Alpine 3.21 includes.
# Also test compatibility with Wayland Protocols 1.38 and Wayland libs 1.23.
# Musl support is an added bonus.
image: alpine:3.21
steps:
- uses: actions/checkout@v7
- name: Install dependencies
run: |
apk update
apk add \
binutils \
curl-dev \
ffmpeg-dev \
gcc \
git \
g++ \
libass-dev \
lua5.2-dev \
pulseaudio-dev \
libplacebo-dev \
libx11-dev \
libxfixes-dev \
libxkbcommon-dev \
libxrandr-dev \
libxpresent-dev \
libxscrnsaver-dev \
meson \
musl-dev \
rust \
samurai \
wayland-dev \
wayland-protocols
- name: Build with meson
id: build
run: |
./ci/build-linux-old.sh
- name: Print meson log
if: ${{ failure() && steps.build.outcome == 'failure' }}
run: |
cat ./build/meson-logs/meson-log.txt
- name: Run meson tests
id: tests
run: |
meson test -C build
- name: Print meson test log
if: ${{ failure() && steps.tests.outcome == 'failure' }}
run: |
cat ./build/meson-logs/testlog.txt
openbsd:
runs-on: ubuntu-latest # until https://github.com/actions/runner/issues/385
timeout-minutes: 15 # avoid any weirdness with the VM
defaults:
run:
shell: cpa.sh {0} --sync-files none
steps:
- uses: actions/checkout@v7
- name: Start OpenBSD VM
uses: cross-platform-actions/action@v1.4.0
with:
operating_system: openbsd
version: '7.8'
sync_files: runner-to-vm
- name: Install dependencies
run: |
sudo pkg_add -U \
cmake \
curl \
ffmpeg \
git \
libarchive \
libbluray \
libcaca \
libcdio-paranoia \
libdvdnav \
libiconv \
libplacebo \
libv4l \
libxkbcommon \
luajit \
meson \
openal \
pkgconf \
pulseaudio \
python3 \
rubberband \
rust \
sdl2 \
shaderc \
spirv-cross \
spirv-headers \
uchardet \
vulkan-loader \
vulkan-headers \
zimg
- name: Build with meson
id: build
run: |
./ci/build-openbsd.sh
- name: Print meson log
if: ${{ failure() && steps.build.outcome == 'failure' }}
run: |
cat ./build/meson-logs/meson-log.txt
- name: Run meson tests
id: tests
run: |
meson test -C build
- name: Print meson test log
if: ${{ failure() && steps.tests.outcome == 'failure' }}
run: |
cat ./build/meson-logs/testlog.txt
freebsd:
runs-on: ubuntu-latest # until https://github.com/actions/runner/issues/385
timeout-minutes: 15 # avoid any weirdness with the VM
defaults:
run:
shell: cpa.sh {0} --sync-files none
steps:
- uses: actions/checkout@v7
- name: Start FreeBSD VM
uses: cross-platform-actions/action@v1.4.0
with:
operating_system: freebsd
version: '15.1'
sync_files: runner-to-vm
- name: Install dependencies
run: |
sudo pkg update
sudo pkg install -y \
alsa-lib \
cmake \
curl \
evdev-proto \
ffmpeg \
git \
iconv \
jackit \
libarchive \
libbluray \
libcaca \
libcdio-paranoia \
libdvdnav \
libdisplay-info \
libplacebo \
libXinerama \
libxkbcommon \
libXpresent \
libXv \
luajit \
meson \
mujs \
openal-soft \
pipewire \
pkgconf \
pulseaudio \
python3 \
rubberband \
rust \
sekrit-twc-zimg \
sdl2 \
sndio \
uchardet \
v4l_compat \
vulkan-headers \
wayland-protocols
- name: Build with meson
id: build
run: |
./ci/build-freebsd.sh
- name: Print meson log
if: ${{ failure() && steps.build.outcome == 'failure' }}
run: |
cat ./build/meson-logs/meson-log.txt
- name: Run meson tests
id: tests
run: |
meson test -C build
- name: Print meson test log
if: ${{ failure() && steps.tests.outcome == 'failure' }}
run: |
cat ./build/meson-logs/testlog.txt
msys2:
runs-on: ${{ matrix.os }}
strategy:
fail-fast: false
matrix:
include:
- sys: "clang64"
os: "windows-2025-vs2026"
auto-var-init: "uninitialized"
- sys: "clang64"
os: "windows-2025-vs2026"
asan: true
- sys: "clang64"
os: "windows-2025-vs2026"
auto-var-init: "zero"
- sys: "clang64"
os: "windows-2025-vs2026"
auto-var-init: "pattern"
- sys: "mingw64"
os: "windows-2025-vs2026"
- sys: "ucrt64"
os: "windows-2025-vs2026"
- sys: "clangarm64"
os: "windows-11-arm"
defaults:
run:
shell: msys2 {0}
steps:
- name: Disable autocrlf
shell: pwsh
run: |
git config --global core.autocrlf false
git config --global core.eol lf
- uses: actions/checkout@v7
- uses: msys2/setup-msys2@v2
with:
msystem: ${{ matrix.sys }}
update: true
install: git
pacboy: >-
amf-headers:p
angleproject:p
ca-certificates:p
cc:p
cppwinrt:p
curl:p
ffmpeg:p
lcms2:p
libarchive:p
libass:p
libcdio-paranoia:p
libdvdnav:p
libjpeg-turbo:p
libplacebo:p
lua51:p
meson:p
ninja:p
pkgconf:p
python:p
rst2pdf:p
rubberband:p
rust:p
shaderc:p
spirv-cross:p
uchardet:p
vapoursynth:p
vulkan-devel:p
- name: Build with meson
id: build
run: |
./ci/build-msys2.sh meson
env:
SYS: ${{ matrix.sys }}
ASAN: ${{ matrix.asan }}
AUTO_VAR_INIT: ${{ matrix.auto-var-init }}
- name: Print meson log
if: ${{ failure() && steps.build.outcome == 'failure' }}
run: |
cat ./build/meson-logs/meson-log.txt
- name: Run meson tests
id: tests
run: |
meson test -C build
env:
GGML_NO_BACKTRACE: 1
- name: Print meson test log
if: ${{ failure() && steps.tests.outcome == 'failure' }}
run: |
cat ./build/meson-logs/testlog.txt
publish:
name: Publish git release
needs: [mingw, win32, macos]
if: github.event_name == 'push' && github.ref == 'refs/heads/master' &&
github.repository == 'mpv-player/mpv'
runs-on: ubuntu-latest
permissions:
contents: write
actions: read
concurrency:
group: publish-git
cancel-in-progress: false
steps:
- uses: actions/checkout@v7
with:
fetch-depth: 0
- name: Publish to git release
env:
GH_REPO: ${{ github.repository }}
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
run: |
set -eu
tag=git-release
base_url="https://github.com/$GH_REPO/releases/download/$tag"
artifacts=$(gh api "repos/{owner}/{repo}/actions/runs/${{ github.run_id }}/artifacts")
# Attach every build artifact unchanged; the notes link to them with
# labels derived from the verbose names.
mkdir -p release
jq -r '.artifacts[] | "\(.id)\t\(.name)"' <<< "$artifacts" |
while IFS=$'\t' read -r id name; do
gh api "repos/{owner}/{repo}/actions/artifacts/$id/zip" > "release/$name.zip"
done
section() {
jq -r --arg url "$base_url" --arg re "$1" '
def pretty:
(if test("i686") then "i686"
elif test("aarch64") then "arm64"
elif test("-arm$") then "arm64"
elif test("intel") then "x86_64"
else "x86_64" end) as $arch
| [ (if test("mingw") then "GCC" elif test("msvc") then "Clang" else empty end),
(if test("macos") then "macOS " + capture("macos-(?<v>[0-9]+)").v else empty end),
(if test("-pdb") then "debug symbols" else empty end) ] as $q
| $arch + (if ($q | length) > 0 then " (" + ($q | join(", ")) + ")" else "" end);
# primary builds first (x86_64, arm64, i686; Clang before GCC), debug symbols last
def rank:
[ (if test("-pdb") then 1 else 0 end),
(if test("x86_64|intel") then 0 elif test("i686") then 2 else 1 end),
(if test("mingw") then 1 else 0 end),
(if test("macos") then (capture("macos-(?<v>[0-9]+)").v | tonumber) else 0 end) ];
[ .artifacts[] | select(.name | test($re)) ] | sort_by(.name | rank)
| map("* [\(.name | pretty)](\($url)/\(.name).zip)") | join("\n")
' <<< "$artifacts"
}
# Match the version string baked into the build (see common/meson.build).
version=$(git describe --abbrev=9 --tags --dirty --match "v0.*")
notes=$(cat <<EOF
Automated development build of the latest \`master\` branch.
- Version: \`$version\`
- Commit: ${{ github.sha }}
- Built: $(date -u '+%Y-%m-%d %H:%M:%S UTC')
### Windows
$(section 'w64|msvc')
### macOS
$(section 'macos')
> [!WARNING]
> These are untested development builds of the very latest \`master\`.
> They are **not** official releases. For stable builds see the
> [latest release](https://github.com/$GH_REPO/releases/latest).
EOF
)
# Create the release once (which also creates the tag), then only edit
# it in place so watchers aren't notified on every master push. Stale
# assets are dropped first since their names embed the run id.
if gh release view "$tag" >/dev/null 2>&1; then
gh release view "$tag" --json assets --jq '.assets[].name' |
while read -r asset; do
gh release delete-asset "$tag" "$asset" --yes
done
gh release edit "$tag" \
--prerelease \
--latest=false \
--title "mpv development build" \
--notes "$notes"
gh release upload "$tag" release/*
else
gh release create "$tag" \
--target "$GITHUB_SHA" \
--prerelease \
--latest=false \
--title "mpv development build" \
--notes "$notes" \
release/*
fi
-66
View File
@@ -1,66 +0,0 @@
name: Cleanup caches
on:
workflow_run:
workflows: [build]
types: [completed]
concurrency:
group: ${{ github.workflow }}
cancel-in-progress: true
permissions:
actions: write
jobs:
cache:
runs-on: ubuntu-latest
steps:
- run: |
set -o pipefail
gh cache list -L 1000 --json id,key,ref -S last_accessed_at -O desc > caches.json
# Resolve the state of every pull request that owns a cache.
for pr in $(jq -r '.[].ref | select(test("^refs/pull/[0-9]+/merge$")) |
split("/")[2]' caches.json | sort -u); do
jq -n --arg pr "$pr" --arg state "$(gh pr view $pr --json state --jq .state)" \
'{($pr): $state}'
done | jq -s 'add // {}' > prs.json
jq -r --slurpfile prs prs.json --arg default "$DEFAULT_REF" '
$prs[0] as $state |
# Group caches by kind, dropping the timestamp or content hash
# that every save appends to the key.
def family: if startswith("msys2-") then "msys2"
elif test("-[0-9]+$") then "ccache"
else "other" end;
def group: if family == "msys2" then sub("-(files|state):.*$"; "")
else sub("-[0-9]+$"; "") end;
[ .[] | . + { family: (.key | family),
group: (.key | group),
pr: (.ref | if test("^refs/pull/[0-9]+/merge$")
then split("/")[2] else null end) } ]
| map(select(.family != "other")) as $all
# Caches of merged or closed pull requests, and msys2 caches
# outside the default branch, which every ref restores from.
| ( $all | map(select(
(.pr != null and $state[.pr] != "OPEN") or
(.family == "msys2" and .ref != $default)
)) ) as $dead
# Of the rest only the newest cache of each kind per ref is used.
| ( ($all - $dead) | group_by([.ref, .group]) | map(.[1:]) | add // [] ) as $superseded
| ($dead + $superseded)[] | "\(.id)\t\(.ref)\t\(.key)"
' caches.json |
while IFS=$'\t' read -r id ref key; do
echo "deleting $ref $key"
gh cache delete "$id"
done
env:
GH_REPO: ${{ github.repository }}
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
DEFAULT_REF: refs/heads/${{ github.event.repository.default_branch }}
-55
View File
@@ -1,55 +0,0 @@
name: comment
on:
workflow_run:
workflows: [build]
types: [completed]
concurrency:
group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
cancel-in-progress: true
permissions:
actions: read
contents: read
pull-requests: write
jobs:
pr_comment:
if: github.event.workflow_run.event == 'pull_request' &&
github.event.workflow_run.conclusion == 'success'
runs-on: ubuntu-latest
steps:
- env:
GH_REPO: ${{ github.repository }}
GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
run: |
pr=$(gh pr list --state open --json headRefOid,number \
--jq '.[] | select(.headRefOid == "${{ github.event.workflow_run.head_sha }}") | .number')
if [ -z "$pr" ]; then
echo "No matching open pull request found"
exit 0
fi
artifacts=$(gh api repos/{owner}/{repo}/actions/runs/${{ github.event.workflow_run.id }}/artifacts --jq '.artifacts')
if [ "$(echo "$artifacts" | jq 'length')" -eq 0 ]; then
echo "No artifacts found"
exit 0
fi
body=$(echo "$artifacts" | jq -r '
def link: "https://nightly.link/${{ github.repository }}/actions/artifacts/\(.id).zip";
def entry: "* [\(.name)](\(link))";
"Download the artifacts for this pull request:\n<details><summary>Windows</summary>\n\n" +
(sort_by(.name) | map(select(.name | test("w64|msvc")) | entry) | join("\n")) +
"\n</details>\n<details><summary>macOS</summary>\n\n" +
(sort_by(.name) | map(select(.name | test("macos")) | entry) | join("\n")) +
"\n</details>"
')
comment_id=$(gh issue view $pr --json comments \
--jq '.comments[] | select(.author.login == "github-actions") | .id')
if [ -n "$comment_id" ]; then
edit=--edit-last
fi
gh pr comment $pr --body "$body" $edit
-31
View File
@@ -1,31 +0,0 @@
name: docs
on:
push:
branches:
- master
- ci
- 'release/**'
paths:
- 'DOCS/**'
pull_request:
branches: [master]
paths:
- 'DOCS/**'
concurrency:
group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
cancel-in-progress: true
jobs:
check-docs:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v7
- name: Docs
run: |
sudo apt-get update
sudo apt-get install --no-install-recommends -y python3-docutils rst2pdf
./TOOLS/docutils-wrapper.py rst2man --strip-elements-with-class=contents --halt=2 ./DOCS/man/mpv.rst mpv.1
./TOOLS/docutils-wrapper.py rst2html --halt=2 ./DOCS/man/mpv.rst mpv.html
./TOOLS/docutils-wrapper.py rst2pdf -c -b 1 --repeat-table-rows ./DOCS/man/mpv.rst -o mpv.pdf
-61
View File
@@ -1,61 +0,0 @@
name: Pull Request Fuzzing
on:
pull_request:
branches: [master]
paths-ignore:
- 'DOCS/**'
- 'TOOLS/lua/**'
- '.editorconfig'
- '.github/ISSUE_TEMPLATE/**'
- '.github/PULL_REQUEST_TEMPLATE'
- '.gitignore'
- 'Copyright'
- 'README.md'
- 'RELEASE_NOTES'
concurrency:
group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
cancel-in-progress: true
permissions: {}
jobs:
Fuzzing:
runs-on: ubuntu-latest
permissions:
security-events: write
strategy:
fail-fast: false
matrix:
sanitizer: [address, undefined, memory]
steps:
- name: Build Fuzzers (${{ matrix.sanitizer }})
id: build
uses: google/oss-fuzz/infra/cifuzz/actions/build_fuzzers@master
with:
oss-fuzz-project-name: 'mpv'
language: c
sanitizer: ${{ matrix.sanitizer }}
- name: Run Fuzzers (${{ matrix.sanitizer }})
uses: google/oss-fuzz/infra/cifuzz/actions/run_fuzzers@master
with:
oss-fuzz-project-name: 'mpv'
language: c
fuzz-seconds: 1200
sanitizer: ${{ matrix.sanitizer }}
report-ooms: true
report-timeouts: true
output-sarif: true
- name: Upload Crash
uses: actions/upload-artifact@v7
if: failure() && steps.build.outcome == 'success'
with:
name: ${{ matrix.sanitizer }}-artifacts
path: ./out/artifacts
- name: Upload Sarif
if: always() && steps.build.outcome == 'success'
uses: github/codeql-action/upload-sarif@v4
with:
sarif_file: cifuzz-sarif/results.sarif
checkout_path: cifuzz-sarif
-75
View File
@@ -1,75 +0,0 @@
name: lint
on:
push:
branches:
- master
- ci
- 'release/**'
pull_request:
branches: [master]
concurrency:
group: ${{ github.workflow }}-${{ github.head_ref || github.run_id }}
cancel-in-progress: true
jobs:
commit-msg:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v7
with:
fetch-depth: 50
- name: Lint
run: |
./ci/lint-commit-msg.py
swift:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v7
- uses: norio-nomura/action-swiftlint@3.2.1
lua:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v7
- uses: lunarmodules/luacheck@v1
ruff:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v7
- name: Check
id: check
uses: astral-sh/ruff-action@v3
with:
src: ". TOOLS/umpv"
- name: Suggested changes
if: ${{ failure() && steps.check.outcome == 'failure' }}
run: ruff check . TOOLS/umpv --diff
mypy:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v7
- uses: python/mypy@master
with:
install_project_dependencies: no
options: '--check-untyped-defs'
paths: ". TOOLS/umpv"
editorconfig:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v7
- uses: editorconfig-checker/action-editorconfig-checker@main
- run: editorconfig-checker -config .editorconfig-checker.json
pre-commit:
runs-on: ubuntu-latest
steps:
- uses: actions/checkout@v7
- uses: pre-commit/action@v3.0.1
+36 -7
View File
@@ -1,7 +1,36 @@
/subprojects/*
!/subprojects/packagefiles
!/subprojects/*.wrap
/build/
/dist/
.hvigor/
.DS_Store
*.o
*.a
*.d
*.exe
*.pyc
.depend
/config.h
/config.mak
/config.log
/mpv
/mpv.app
/version.h
/input/input_conf.h
/tags
/TAGS
/video/out/gl_video_shaders.h
/video/out/x11_icon.inc
/demux/ebml_defs.c
/demux/ebml_types.h
/sub/osd_font.h
/player/lua/*.inc
/DOCS/man/*/mpv.1
/DOCS/man/*/mpv.aux
/DOCS/man/*/mpv.log
/DOCS/man/*/mpv.out
/DOCS/man/*/mpv.pdf
/DOCS/man/*/mpv.toc
/waf
/build
/.waf*
/.lock-waf_*
/old_build
/Makefile
-110
View File
@@ -1,110 +0,0 @@
local mp_globals = {
mp = {
fields = {
command = {},
commandv = {},
command_native = {},
command_native_async = {},
add_hook = {},
abort_async_command = {},
del_property = {},
get_property = {},
get_property_osd = {},
get_property_bool = {},
get_property_number = {},
get_property_native = {},
set_property = {},
set_property_bool = {},
set_property_number = {},
set_property_native = {},
get_time = {},
add_key_binding = {},
add_forced_key_binding = {},
remove_key_binding = {},
register_event = {},
unregister_event = {},
observe_property = {},
unobserve_property = {},
add_timeout = {},
add_periodic_timer = {},
get_opt = {},
get_script_name = {},
get_script_directory = {},
osd_message = {},
get_wakeup_pipe = {},
get_next_timeout = {},
dispatch_events = {},
register_idle = {},
unregister_idle = {},
enable_messages = {},
register_script_message = {},
unregister_script_message = {},
create_osd_overlay = {},
get_osd_size = {},
msg = {
fields = {
fatal = {},
error = {},
warn = {},
info = {},
verbose = {},
debug = {},
trace = {},
}
},
-- Not documented
-- TODO: Document or remove them
disable_key_bindings = {},
enable_key_bindings = {},
find_config_file = {},
flush_keybindings = {},
format_time = {},
get_mouse_pos = {},
set_key_bindings = {},
set_mouse_area = {},
set_osd_ass = {},
}
},
exit = {},
unpack = {},
}
local mp_internal = {
mp = {
fields = {
-- Internal
-- TODO: Move to mp_internal module
ARRAY = { fields = { info = {}, type = {} }},
MAP = { fields = { info = {}, type = {} }},
UNKNOWN_TYPE = { fields = { info = {}, type = {} }},
_legacy_overlay = { fields = { res_x = {}, res_y = {}, data = {}, update = {} }},
cancel_timer = {},
get_osd_margins = {},
input_define_section = {},
input_disable_section = {},
input_enable_section = {},
input_set_section_mouse_area = {},
keep_running = {},
log = {},
raw_abort_async_command = {},
raw_command_native_async = {},
raw_hook_add = {},
raw_hook_continue = {},
raw_observe_property = {},
raw_unobserve_property = {},
raw_wait_event = {},
request_event = {},
script_name = {},
use_suspend = {},
wait_event = {},
}
}
}
std = "min+mp"
stds = { mp = { read_globals = mp_globals } }
-- mp_internal seems to be merged with mp for other files too...
files["player/lua/defaults.lua"] = { globals = mp_internal }
files["player/lua/auto_profiles.lua"] = { globals = { "p", "get" } }
files["player/lua/select.lua"] = { globals = { "p", "get" } }
max_line_length = 100
-23
View File
@@ -1,23 +0,0 @@
repos:
- repo: https://github.com/pre-commit/pre-commit-hooks
rev: v5.0.0
hooks:
- id: check-case-conflict
- id: check-executables-have-shebangs
- id: check-illegal-windows-names
- id: check-json
- id: check-shebang-scripts-are-executable
- id: check-xml
exclude: ^TOOLS/mpv-osd-symbols.sfdir/font.props$
- id: check-yaml
- id: end-of-file-fixer
exclude: ^TOOLS/osxbundle/mpv.app/Contents/PkgInfo$
- id: fix-byte-order-marker
- id: mixed-line-ending
- id: trailing-whitespace
- repo: https://github.com/codespell-project/codespell
rev: v2.3.0
hooks:
- id: codespell
args: ["--ignore-words-list", "ALS,als,datas,DNE,enew,HDA,numer,ontop,optionA,Paeth,pathc,inout,requestor"]
exclude: ^(etc/menu.conf|misc/language.c)$
-10
View File
@@ -1,10 +0,0 @@
disabled_rules:
- identifier_name
- line_length
- large_tuple
- void_function_in_ternary
- function_body_length
- cyclomatic_complexity
- file_length
- type_body_length
strict: true
+30
View File
@@ -0,0 +1,30 @@
language: c
env:
- LIBAV=libav-stable
- LIBAV=libav-git
- LIBAV=ffmpeg-stable
- LIBAV=ffmpeg-git
compiler:
- clang
branches:
only:
- master
- ci
before_install: ./travis-deps libass-stable $LIBAV
script:
- ./bootstrap.py
- ./waf configure
- ./waf build
notifications-policy: &notifications-policy
on_success: change
on_failure: always
notifications:
email: false
irc:
channels:
- "irc.freenode.org#mpv-player-dev"
<<: *notifications-policy
+12 -71
View File
@@ -1,78 +1,19 @@
mpv is a fork of mplayer2, which is a fork of MPlayer.
mpv as a whole is licensed under the GNU General Public License GPL version 2
or later (called GPLv2+ in this document, see LICENSE.GPL for full license
text) by default. The mpv program is licensed the GNU Lesser General Public
License LGPL version 2 or later (LGPLv2.1+ in this document, see LICENSE.LGPL
for full license text) if built without using any GPL only files.
The -Dgpl=false configure switch is provided as a convenience for excluding
the GPL only files listed below from the build process.
However, do note that the build system is provided "as is" and using the
-Dgpl=false configure switch does not in itself create a LGPLv2.1+ license
grant.
Additionally, some non-program parts, such as the documentation and some config
files listed below, are still built and licensed under their respective licenses
even when -Dgpl=false is used. Please consult GPLv2+ and LGPLv2.1+ license texts
for further details on how to distribute these files.
Most source files are LGPLv2.1+ or GPLv2+, but some files are available under
more liberal licenses, such as BSD, MIT, ISC, and possibly others. Look at the
copyright header of each source file, and grep the sources for "Copyright" if
you need to know details. C source files without Copyright notice are usually
licensed as LGPLv2.1+. Also see the list of files with specific licenses below
(not all files can have a standard license header).
All new contributions must be LGPLv2.1+ licensed. Using a more liberal license
compatible to LGPLv2.1+ is also ok.
Changes done to GPL code must come with the implicit/explicit agreement that the
project can relicense the changes to LGPLv2.1+ at a later point without asking
the contributor. This is a safeguard for making potential relicensing of
remaining GPL code to LGPLv2.1+ easier.
See DOCS/contribute.md for binding rules wrt. licensing for contributions.
mpv as a whole is licensed as GPL version 2 or later (see LICENSE). Most source
files are GPLv2+, but some files are available under a more liberal license,
such as LGPLv2+, BSD, MIT, ISC, and possibly others. Look at the copyright
header of each source file, and grep the sources for "Copyright" if you need
to know details. Files without Copyright notice are licensed as LGPLv2+.
For information about authors and contributors, consult the git log, which
contains the complete SVN and CVS history as well.
mpv can be built as LGPLv2.1+ without using any GPL only files. To add a
LGPL mode to mpv, MPlayer code had to be relicensed from GPLv2+ to LGPLv2.1+ by
asking the MPlayer authors for permission. Since permission could not be
obtained from everyone, LGPL mode disables the following features, some of them
quite central:
- Linux X11 video output
- BSD audio output via OSS
- NVIDIA/Linux hardware decoding (vdpau, although nvdec usually works)
- minor features: jack, DVD, CDDA, DVB, CACA, legacy direct3d VO
Some of these will be fixed in the future. The intended use for LGPL mode is
with libmpv, and currently it's not recommended to build mpv CLI in LGPL mode
at all.
Note that mplayer2 as a whole is licensed under GPLv3+. This is because it uses
a copy of talloc (part of Samba), which is LGPLv3+, and the next compatible
license for this mix is GPLv3+.
The following files are still GPL only (-Dgpl=false disables them):
audio/out/ao_jack.c will stay GPL
audio/out/ao_oss.c will stay GPL
stream/dvb* must stay GPL
stream/stream_cdda.c unknown
stream/stream_dvb.* must stay GPL
stream/stream_dvdnav.c unknown
video/out/vo_caca.c unknown
video/out/vo_direct3d.c unknown
video/out/vo_vaapi.c probably impossible (some company's code)
video/out/vo_vdpau.c probably impossible (nVidia's code)
video/out/vo_x11.c probably impossible
video/out/vo_xv.c probably impossible
video/out/x11_common.* probably impossible
video/vdpau.c hard (GPL-only parts must be ifdefed)
video/vdpau.h unknown
video/vdpau_mixer.* actual code must be rewritten
DOCS/man/ GPLv2+
etc/mplayer-input.conf unknown license, probably GPLv2+
etc/mpv.desktop unknown license, probably GPLv2+
etc/restore-old-bindings.conf unknown license, probably GPLv2+
None of the cases listed above affect the final binary if it's built as
LGPL. Linked libraries still can affect the final license (for example if
FFmpeg was built as GPL).
MPlayer as a whole is licensed under GPLv2 (incompatible to GPLv3!), because
some files are licensed to GPLv2 (and _not_ any later version of the license).
In particular, this affects the file libmpdemux/demux_ty_osd.c. It is disabled
under mplayer2, and has been removed from mpv.
-22
View File
@@ -1,22 +0,0 @@
# Phase 6 Benchmark 方法
## 执行
```sh
python3 benchmark/runner/run_emulator.py startup_local --repeat 20
python3 benchmark/runner/run_emulator.py seek_local --repeat 50
python3 benchmark/runner/run_emulator.py stress_lifecycle --repeat 100
python3 benchmark/runner/run_emulator.py play_pause --repeat 200
```
runner 通过 hdc/uitest 驱动 `Playback Lab`,每次 load 使用内置 `sample.mp4`,记录 timeline、错误、状态和进程采样。startup 的首轮标为 cold-ish,后续为 warm;OS page cache 未被强制清空。seek 使用加载后立即触发的原生 `+10s`,避免短样片在输入框多次 keyEvent 期间自然结束。
每个场景输出逐次 JSON 和 min/max/avg/median/P90/P95。百分位采用排序后的 nearest-rank;负时间线值是 unavailable,不参与统计。错误单独保留,不从成功样本中删除。
## 采样边界
- VmRSS、VmSize、Threads、fd count 和原始 CPU jiffies 来自 `/proc`;PSS、精确 native heap、可靠 CPU permille 未被模拟器公开,输出 null/原始值。
- host wall time 包含 hdc 往返,只用于发现自动化超时,不作为启动性能结论。
- `firstAudioCallbackMs` 是 Playback Lab 诊断轮询观察到 callback_count>0 的时间;要取得 callback 内精确时间需独立 tracing,因此不把它当作硬件时钟证据。
- `stress_lifecycle` 覆盖 load/stop;release/recreate 需要新 XComponent surface,另行作为实体设备/手工场景。
- HTTPS、HLS、鉴权、网络字幕必须使用受控 fixture;本轮没有把公网失败当作性能数据。
-7
View File
@@ -1,7 +0,0 @@
# Buffering Model
buffering 是 MPV `cache-buffering` 属性与网络状态的边沿诊断,不是 UI 的瞬时文案。进入时记录 `bufferingEnterCount` 和起始 monotonic 时间;退出时累加 `bufferingTotalMs`,并更新 `bufferingMaxSingleMs`。重复的同值属性通知不会重复计数。
网络快照同时保留 protocol、host、HTTP status、redirect count、cache duration/bytes、current bitrate、range support 和最近 seek result。`networkWaitMs` 只计算已观察到的 network-open 到 first-byte/file-loaded 间隔;file:// 输入为 -1。
本地短片通常在首帧后很快 EOF,因此 buffering 计数为 0 不代表网络流在真实设备上不会重缓冲。受控 HTTP Range fixture 已在 Phase 5 验证;HTTPS/HLS、鉴权和弱网重缓冲应在实体设备用固定服务端脚本复测。
-97
View File
@@ -1,97 +0,0 @@
# Phase 1 Build Issues
## BUILD-001
Component: FFmpeg configure
Command: `scripts/build-ffmpeg-ohos.sh`
Original Error: `Unknown option "--disable-postproc".`
File: `scripts/build-ffmpeg-ohos.sh`
Root Cause: 当前 FFmpeg 已移除独立 postproc 配置项,Phase 0 示例参数过期。
Platform Difference: 无;属于当前 upstream 配置接口变化。
Fix: 删除不存在的 `--disable-postproc`,以当前 `configure --help` 为准。
Files Changed: `scripts/build-ffmpeg-ohos.sh`
Patch Required: Build script only
Result: configure 继续执行。
Regression Risk: Low
## BUILD-002
Component: FFmpeg compiler/linker probe
Command: `scripts/build-ffmpeg-ohos.sh`
Original Error: `ld: library 'System' not found`,随后 `C compiler test failed`。
File: `scripts/build-ffmpeg-ohos.sh`、`ffbuild/config.log`
Root Cause: `--sysroot` 不会自动改变 Clang target;链接探针仍按 macOS 目标寻找 `libSystem`。
Platform Difference: macOS arm64 与 OHOS arm64 架构相同,但 ABI、对象格式和系统库不同。
Fix: 同时向 CFLAGS 和 LDFLAGS 显式传入 `--target=aarch64-linux-ohos`。
Files Changed: `scripts/build-ffmpeg-ohos.sh`
Patch Required: Build script only
Result: FFmpeg configure、build、install 成功,产物为 AArch64 ELF。
Regression Risk: Low
## BUILD-003
Component: FreeType zlib resolution
Command: `scripts/build-font-stack-ohos.sh`
Original Error: `ERROR: Subproject zlib is buildable: NO`,Meson 尝试下载 WrapDB zlib。
File: `scripts/build-font-stack-ohos.sh`
Root Cause: FreeType 的 `zlib=external` 表示构建 fallback subproject,不表示使用外部系统库。
Platform Difference: SDK 已提供 target `libz.so`,不应另行下载 target zlib。
Fix: 使用当前 FreeType 定义的 `zlib=system`,通过隔离 `.pc` 指向 SDK AArch64 zlib。
Files Changed: `scripts/build-font-stack-ohos.sh`
Patch Required: Build script only
Result: FreeType 2.14.3 静态库构建成功。
Regression Risk: Low
## BUILD-004
Component: Meson cross-file pkg-config resolution
Command: `scripts/build-font-stack-ohos.sh`
Original Error: HarfBuzz 报 `Dependency 'freetype2' is required but not found`,包装器被解析到 HarfBuzz 源码目录。
File: `toolchains/ohos-api24-arm64.ini`
Root Cause: `@GLOBAL_SOURCE_ROOT@` 指向当前被配置项目的 source root,不固定指向 Velix-MPV-Core。
Platform Difference: 多个独立 Meson 源码树共享一个 target prefix。
Fix: 使用以 Cross File 自身为基准的 `@DIRNAME@/../` 定位包装器与 prefix。
Files Changed: `toolchains/ohos-api24-arm64.ini`
Patch Required: Cross File only
Result: FreeType、HarfBuzz、libass 和 MPV 均只从 OHOS prefix 解析依赖。
Regression Risk: Low
-64
View File
@@ -1,64 +0,0 @@
# HarmonyOS Native Build Strategy
## Scope
Phase 1 只构建 HarmonyOS API 24 arm64 `libmpv`。不包含播放器 UI、XComponent 集成、`ao_harmony`、MPV OHCodec backend、性能优化、HDR 或真机运行声明。
## Toolchain
| Item | Value |
|---|---|
| SDK | `/Applications/DevEco-Studio.app/Contents/sdk/default/openharmony` |
| Version | 6.1.1.125 / API 24 |
| Target | `aarch64-linux-ohos` |
| Sysroot | `<SDK>/native/sysroot` |
| Compiler | OHOS Clang 15.0.4 |
| Meson | 1.9.0,仓库隔离 venv |
| Ninja | SDK 1.12.0 |
| pkgconf | 3.0.6,仓库隔离 host tool |
| Cross File | `toolchains/ohos-api24-arm64.ini` |
| Prefix | `build/ohos-arm64/prefix` |
`scripts/setup-ohos-env.sh` 清空 `CPATH`、`C_INCLUDE_PATH`、`CPLUS_INCLUDE_PATH`、`LIBRARY_PATH`、`CMAKE_PREFIX_PATH`,并将 pkg-config 限制到 OHOS prefix。target 编译和链接必须同时带 `--target` 与 `--sysroot`。
## Reproducible Build
```sh
scripts/build-all-ohos.sh
```
顺序固定为:host tools → pinned sources → C/C++ probes → SDK zlib metadata → OpenSSL → FFmpeg → FreeType/FriBidi/HarfBuzz/libass → libplacebo → MPV → 全量 ABI/污染/符号校验。
各步骤也可单独执行:
```sh
scripts/verify-toolchain.sh
scripts/build-openssl-ohos.sh
scripts/build-ffmpeg-ohos.sh
scripts/build-font-stack-ohos.sh
scripts/build-libplacebo-ohos.sh
scripts/build-mpv-ohos.sh
scripts/verify-ohos-prefix.sh
```
所有下载源码按 `scripts/fetch-phase1-sources.sh` 的 tag/commit 固定。构建目录可删除后重建,禁止从 Homebrew 或 macOS prefix 提供 target library。
## MPV Configuration
MPV 使用 `-Dauto_features=disabled`,再显式启用 `libmpv`、GPL、`plain-gl`、OpenGL、vector、zlib。CLI、测试、脚本语言、桌面后端、平台硬件加速 backend 和 Vulkan 均关闭。实际 feature 列表以 `DOCS/MPV_BUILD_REPORT.md` 和 `build/ohos-arm64/logs/mpv.log` 为准。
MPV 源码仅增加明确的 `host_machine.system() == 'ohos'` 检测与 `ohos` feature。POSIX 路径继续复用现有 threads、timer、path、poll、IPC、subprocess 实现;这些能力当前只能标记 compile verified。
## Output and Gates
主产物:`build/ohos-arm64/prefix/lib/libmpv.so.2.5.0`。
`scripts/verify-ohos-prefix.sh` 必须同时证明:
- 所有 prefix 静态库成员为 AArch64 ELF;
- `libmpv` 为 AArch64 ELF shared object;
- 动态依赖仅来自 HarmonyOS SDK allowlist(启用 `harmony-audio` 时包含 `libohaudio.so`);
- `.pc`/CMake/Ninja target metadata 无 Homebrew、`/usr/local`、`.dylib` 或 Apple framework;
- Client API 与 OpenGL Render API 必需符号存在。
真机行为、CA store、XComponent/EGL 生命周期及 OHCodec buffer 语义留到后续阶段验证。
-11
View File
@@ -1,11 +0,0 @@
# Core API versioning
当前冻结版本:`VELIX_CORE_API_VERSION = 0x000100`(v0)。
- `MAJOR` 变化表示 ABI/API 不兼容,必须升级主 App 适配层。
- `MINOR` 变化只增加尾部字段或能力,旧调用方保持可用。
- `PATCH` 变化只修复实现,不改变公开结构布局。
版本宏、`velix_core_get_api_version()`、`velix_core_get_version()` 和 `velix_core_get_build_info()` 位于 `include/velix/core.h`,C ABI 使用 opaque handle 与固定宽度整数。变更 API 时同步更新 `metadata/build-info.json`、此文档与集成文档,并在发布前检查动态符号。
v0 不把 `mpv_handle`、`mpv_node`、FFmpeg、OHCodec、EGL、OHAudio 或 OHNativeWindow 类型带出 SDK。
-27
View File
@@ -1,27 +0,0 @@
# Velix Core SDK
Phase 7 SDK v0 以 `include/velix/core.h` 与 `libvelix_core_binding.so` 为产品边界,主 App 不编译 MPV/FFmpeg 源码。
## 目录
```text
velix-core/
├── include/velix/core.h
├── include/mpv/harmony_playback.h
├── lib/arm64-v8a/libvelix_core_binding.so
├── lib/arm64-v8a/libmpv.so.2
├── lib/arm64-v8a/libc++_shared.so
├── metadata/build-info.json
└── licenses/
```
`libvelix_core_binding.so` 仅提供产品生命周期、控制、状态、媒体、音轨、字幕、指标与 build-info NAPI;XComponent 的 OHNativeWindow/EGL/mpv render context 仍由 Core 内部实现。主 App 只能看到 `PlaybackEngine`/`PlaybackController` 的业务类型。
默认 profile 为 software decode,OHCodec GL mapper 保持关闭。SDK 产物位于 `dist/velix-core`,已在 `.gitignore` 中忽略,禁止把 34MB `libmpv` 二进制提交到 Git。
构建与 staging:
```sh
./scripts/package-velix-core-sdk.sh --profile default
../Velix/scripts/build-velix-app.sh --release-core --build-mode debug
```
-81
View File
@@ -1,81 +0,0 @@
# FFmpeg Strategy
## 决策
Phase 1 使用独立构建的 **官方 FFmpeg upstream**,不把源码揉进 MPV,也暂不建立 `Velix-FFmpeg-Harmony`。
研究基线:FFmpeg master `3acec0a1af2dda0a0838689b8b8649e7deb080a0`(2026-08-28)。该 commit 已提供 OpenHarmony `ohcodec` hwcontext/pixel format 以及 H.264/HEVC decoder/encoder,显著降低自行 fork 的必要性。
## MPV 依赖约束
MPV 当前 Meson 最低要求:
| Library | 最低版本 |
|---|---:|
| libavcodec | 60.31.102 |
| libavfilter | 9.12.100 |
| libavformat | 60.16.100 |
| libavutil | 58.29.100 |
| libswresample | 4.12.100 |
| libswscale | 7.5.100 |
必须同时生成 pkg-config 文件供 MPV cross build 使用。
## 两种方案评估
### A. 官方 FFmpeg 独立构建(采用)
- 独立源码目录与安装前缀。
- arm64 OHOS toolchain 交叉编译。
- Phase 1 首先保留 software decode、demux、HTTP/HLS、字幕相关基础;OHCodec 可编译但不要求首次 MPV 构建即启用硬解。
- TLS 使用 FFmpeg 已支持并经过许可审核的第三方 backend(优先 OpenSSL);Harmony `libnet_ssl.so` 不能直接替代。
- 优点:上游同步清晰、已有 OHCodec、MPV 仓库 patch 最小。
### B. Velix-FFmpeg-Harmony fork(暂不采用)
只有满足以下任一条件才触发:
1. 官方 OHCodec 在 API 24 上存在阻断编译/运行问题,且修复无法及时 upstream。
2. 需要 OHNativeBuffer/GPU import 或 HDR metadata 等实质性 FFmpeg API 扩展。
3. 网络/TLS 或 Harmony 生命周期需要不可在外部构建脚本解决的源码 patch。
建立后必须同样保留完整历史、配置 `upstream`、小 patch 分支,并单独维护许可证审计。
## Phase 1 建议配置轮廓
实际参数必须先由交叉编译探针和 FFmpeg `configure --help` 固化;以下是预计命令,不是已成功结果:
```sh
./configure \
--prefix="$FFMPEG_PREFIX" \
--target-os=linux \
--arch=aarch64 \
--cc="$OHOS_CLANG --target=aarch64-linux-ohos --sysroot=$OHOS_SYSROOT" \
--cxx="$OHOS_CLANGXX --target=aarch64-linux-ohos --sysroot=$OHOS_SYSROOT" \
--ar="$OHOS_LLVM_AR" \
--strip="$OHOS_LLVM_STRIP" \
--enable-cross-compile \
--enable-pic \
--enable-shared \
--disable-programs \
--disable-doc \
--enable-network \
--enable-openssl
```
注意:FFmpeg 当前 configure 未发现独立 `target-os=ohos` 结论,是否应使用 `linux` 必须由真实配置日志决定;编译器会同时定义 `__OHOS__` 与 `__linux__`。`--enable-ohcodec` 在首次软件解码 build 可暂缓,第二个受控构建再开启并验证其五个 Native media 库。
## 网络与 TLS
- HTTP/HLS 主要由 libavformat protocol/demuxer 提供,不新增 MPV 自定义网络栈。
- socket/DNS 可直接使用 OHOS libc。
- HTTPS 需要 FFmpeg 支持的 TLS backend。首选独立交叉构建 OpenSSL,并固定 CA bundle/系统证书策略;证书路径、SNI、代理、IPv6、重定向、Range 请求必须真机验证。
- 禁止将“SDK 有 `libnet_ssl.so`”写成“FFmpeg HTTPS 已支持”。
## 验收与 fork 决策点
1. `configure` 识别 arm64 OHOS、pthreads、network、dlopen。
2. 所需 FFmpeg libraries 完整生成,pkg-config 不泄漏宿主机路径。
3. software decode + HTTP/HLS 单测通过。
4. `--enable-ohcodec` 能链接 API 24 SDK,并枚举 `h264_oh` / `hevc_oh`。
5. 任何源码 patch 先形成最小 diff、根因、upstream 可接受性评估,再决定是否 fork。
-102
View File
@@ -1,102 +0,0 @@
# HarmonyOS OHAudio Backend
状态:Phase 3 CODE COMPLETE / DEVICE VALIDATION PENDING。
代码、OHOS arm64 交叉编译和模拟器 H.264/AAC smoke test 已完成。实体设备上的真实输出路由、延迟、A/V sync 精度、蓝牙/耳机、多声道和长时间稳定性仍未宣称完成。
## Architecture
```text
MPV decoder/filter
↓
MPV AO buffer (audio/out/buffer.c)
↓ pull callback
audio/out/ao_harmony.c
↓
OH_AudioRenderer write callback
↓
HarmonyOS audio device
```
`ao_harmony` 是 MPV pull AO。PCM 始终在 native 层流动;ArkTS 只负责播放控制和诊断展示,不参与持续 PCM 搬运。
实现文件:
- `audio/out/ao_harmony.c`:AO、OHAudio renderer、callback、时钟和统计。
- `audio/out/ao_harmony.h`:Playback Lab 使用的只读诊断 ABI。
- `audio/out/ao.c`、`meson.build`、`meson.options`:driver 注册和 `harmony-audio` feature。
## OHAudio API
实现以当前 API 24 SDK headers 为准:
- `ohaudio/native_audiostreambuilder.h`
- `ohaudio/native_audiorenderer.h`
- `ohaudio/native_audiostream_base.h`
- `multimedia/native_audio_channel_layout.h`
- 链接库:`libohaudio.so`
Builder 设置 `SamplingRate`、`ChannelCount`、`ChannelLayout`、`SampleFormat`、`EncodingType(RAW)`、`LatencyMode(NORMAL)`、`RendererInfo(USAGE_MOVIE)` 和 write/device/error callbacks,然后 `GenerateRenderer` 并销毁 Builder。Renderer 使用 `Start`、`Pause`、`Stop`、`Flush`、`Release`、`SetVolume`、`GetFrameSizeInCallback`、`GetLatency` 和 `GetAudioTimestampInfo`。
## MPV AO / callback model
`write` 保持 NULL,`start` 触发 OHAudio。OHAudio 回调只计算 frame 数、调用 `ao_read_data(..., pad_silence=true, blocking=false)`,不足部分填静音并更新轻量原子计数;不访问网络、不调用 N-API/ArkTS、不分配大对象、不逐回调打印日志。回调首包仅在 debug 日志中记录一次。
MPV AO buffer 是 producer,OHAudio callback 是 consumer。`ao_read_data` 返回的 PCM 只在当前 callback 内有效,系统提供的 `audioData` 由 OHAudio 所有;没有 AVFrame 或 PCM 指针跨生命周期保存。
## Format / sample rate / channel layout
- 仅接受 PCM;MPV float 优先映射 `AUDIOSTREAM_SAMPLE_F32LE`,创建失败时回退 `AF_FORMAT_S16` / `AUDIOSTREAM_SAMPLE_S16LE`。
- 采样率使用 MPV 请求值,不强制改成 48 kHz;设备拒绝时返回真实 renderer 创建错误,由 MPV 上层重新协商/重采样。
- 单声道映射 `CH_LAYOUT_MONO`,其余首版限制为双声道 `CH_LAYOUT_STEREO`;超过 2 声道先通过 MPV channel map downmix 到 stereo。没有声称 5.1/7.1/Atmos/Dolby 已支持,实体设备能力仍待验证。
模拟器实际日志:请求 48 kHz、5.1、floatp,经 MPV downmix 后创建 48 kHz、2 声道、float32 renderer。
## Pause / resume / stop / seek / flush
- pause:`OH_AudioRenderer_Pause`。
- resume/start:`OH_AudioRenderer_Start`。
- stop:MPV 的 stop/EOF 通过 AO reset/uninit 路径停止 renderer;不把 pause、stop、release 混为 destroy。
- seek/reset:先禁止新 callback,`Stop`,等待 inflight callback 清零,再 `Flush`;清空本地提交/播放 epoch,增加 `audioFlushCount` 与 `seekAudioResetCount`,下一次 `start` 重新开始。
- release:禁止 callback → Stop → 等待 callback → 停止并 join timestamp 线程 → Release renderer;回调不会持有释放后的 player/AO 指针。
所有 OHAudio 返回值都检查;失败通过 MPV AO error/reload 路径处理,使用 `AUDIO_BUILDER_FAILED`、`AUDIO_RENDERER_CREATE_FAILED`、`AUDIO_RENDERER_START_FAILED`、`AUDIO_FLUSH_FAILED`、`AUDIO_RELEASE_ERROR`、`AUDIO_UNSUPPORTED_FORMAT` 等明确日志。
## A/V clock and delay
timestamp 线程每 20 ms 调用 `OH_AudioRenderer_GetAudioTimestampInfo`。`submittedFrames` 在 callback 完成填充后递增;`playedFrames` 使用 renderer frame position 减去本次 Flush 的 timestamp origin,避免设备在 Flush 后继续返回旧 epoch。延迟计算为:
```text
pending = max(submittedFrames - playedFrames, 0)
estimatedDelayMs = rendererLatencyMs + pending * 1000 / sampleRate
```
`audioTimestamp` 保留 OHAudio 的 monotonic timestamp。MPV 播放倍速仍由 MPV 音频滤镜/时钟负责,不叠加 renderer speed。算法已编译并在模拟器观察到稳定计数;绝对 A/V 偏差与设备路由连续性必须实体机测量。
## Volume and underrun
Renderer volume 固定为 unity,最终音量由 MPV 软件 gain 处理,避免双重增益。API 24 当前没有可用的系统 underrun counter,因此不伪造 `systemUnderrun`;callback 返回帧不足时只增加 Velix-side `bufferStarvation`。
## Playback Lab diagnostics
Playback Lab 保持原有 video-only 启动路径,新增 `Audio On/Off`(切换 MPV `aid=auto/no`)和以下 native 状态:`audioBackend`、`sampleRate`、`sampleFormat`、`channels`、`channelLayout`、`submittedFrames`、`playedFrames`、`estimatedAudioDelayMs`、`audioBufferStarvation`、`audioFlushCount`、`seekAudioResetCount`、`audioState`。返回值使用 ArkTS Number,避免 BigInt 让 Inspector 无法序列化。
## Validation
- Audio-enabled:OHOS arm64 `libmpv.so.2.5.0` clean build,DT_NEEDED 新增 `libohaudio.so`,Playback Lab HAP 构建成功。
- Video-only:使用 `-Dharmony-audio=disabled` 的独立 Meson build 编译成功。
- Emulator smoke:H.264/AAC MP4 上确认 renderer create、start、callback PCM consumed、pause、resume、seek/flush、stop、release;示例观察到 `starvation=0`,模拟器日志无新增崩溃。
- 未验证:实体设备、真实 latency/A-V 偏差、蓝牙/有线/扬声器切换、系统 audio focus/interruption、5.1/7.1、30 分钟压力。
## Known limitations and physical TODO
1. 当前仅实现 OHAudio renderer;没有 OHCodec/hwdec、Vulkan、音频焦点业务和设备专属路由策略。
2. `systemUnderrun` 需要 SDK 提供真实接口后再接入;当前仅有 `bufferStarvation`。
3. 多声道真实 capability、采样率拒绝后的设备行为、timestamp 在路由切换后的连续性需要实体设备矩阵。
4. 上层应用负责电话/音频焦点策略;Core 只接受 pause/resume 或 AO reload。
Phase 4 不在本次变更中启动。未来 OHCodec 评估至少需要单独验证 surface/buffer 同步、stride/format、fence 与生命周期、零拷贝/DRM 约束、不同设备实现差异五类风险。
## License
本次仅修改 Velix-MPV-Core 内的 MPV 集成代码和 Playback Lab;MPV 原有 GPL/LGPL 许可保持不变。HarmonyOS SDK 头文件/系统库按 SDK 自身许可使用,没有复制第三方实现源码。
-59
View File
@@ -1,59 +0,0 @@
# HarmonyOS Audio Research
> 状态:IMPLEMENTED(代码、OHOS arm64 编译、Playback Lab 模拟器音频 smoke test 已完成);实体设备输出路由、真实延迟、A/V sync 精度、多声道与长稳仍为 NEED PHYSICAL DEVICE VERIFY。
## 1. 推荐方案
新增 MPV 音频输出 `ao_harmony`,直接接 HarmonyOS OHAudio:
```text
MPV audio pipeline -> native PCM ring/buffer -> OHAudio write callback -> AudioRenderer
```
禁止 PCM 逐块跨 N-API/ArkTS。ArkTS 只发送播放器级控制和生命周期事件。
## 2. SDK 依据
- headers:`ohaudio/native_audiostreambuilder.h`、`native_audiorenderer.h`、`native_audiostream_base.h`、`multimedia/native_audio_channel_layout.h`
- library:`libohaudio.so`
- SDK 依据:当前 API 24 headers;builder 基础接口从 API 10 起,独立 write callback 从 API 12 起,`GetAudioTimestampInfo` 从 API 15 起,`GetLatency` 从 API 23 起。
- builder:采样率、声道数、sample format、encoding、latency mode、channel layout、usage、write callback、callback frame size
- renderer:Start、Pause、Stop、Flush、Release、`GetAudioTimestampInfo`、`SetVolume`、`GetLatency`;API 24 没有可用的系统 underrun counter,Velix 只记录 callback starvation。
API 24 还提供 advanced callback 等新接口,但首版使用最低且稳定的 callback API,避免无必要抬高最低版本。
## 3. 与 MPV AO 的映射
| MPV / 播放需求 | OHAudio 映射 | 设计 |
|---|---|---|
| PCM 格式协商 | `SetSamplingRate`、`SetChannelCount/Layout`、`SetSampleFormat` | `init()` 将 MPV `ao->format` 映射为 OHAudio;不支持时返回 unsupported,让 MPV 重采样 |
| 数据输出 | renderer write callback | 回调只读取 MPV AO buffer,禁止阻塞、分配大对象或调用 ArkTS |
| Pause/Resume | `OH_AudioRenderer_Pause/Start` | 与 MPV `control()` 状态串行化 |
| Stop/Seek | `Stop/Flush/Start` | Seek 时停止 callback、清空 OHAudio、重建 timestamp epoch,避免旧 PCM 泄漏 |
| Volume | `OH_AudioRenderer_SetVolume` | 系统 renderer 固定 unity;最终增益由 MPV 软件音量处理,避免重复增益 |
| Speed | MPV 音频滤镜为主 | 不直接用 Renderer speed 改变 MPV 时钟;系统 speed 仅后续专项验证 |
| A/V Sync | `GetAudioTimestampInfo` | 驱动 delay 由 callback 已提交帧与 timestamp epoch 内的硬件播放位置计算;路由切换仍需实体设备测试 |
| 低延迟 | `SetLatencyMode` | 视频播放默认 normal;fast 仅在设备支持和稳定性测试后启用 |
| 多声道 | channel layout + capability | 先 stereo;5.1/7.1 按设备能力协商,失败回退重混音,不 fake success |
## 4. MPV 接入位置
- 新文件建议:`audio/out/ao_harmony.c`。
- 注册位置:`meson.build` 的 audio outputs 区域和 `audio/out/ao.c` driver list。
- 参考结构:`ao_aaudio.c` 的 callback/pull 模式及状态控制;不能复制其 Android API。
- OHAudio callback 与 MPV AO buffer 的线程安全遵循 `audio/out/buffer.c` 现有模型;回调只消费系统请求的 frame 数。
## 5. 格式策略
首版候选优先级:float32(若稳定支持)→ signed 16-bit。采样率优先内容原始值,设备拒绝则让 MPV/libswresample 转为 48 kHz。声道首版 2.0,随后验证 5.1/7.1。所有实际支持值通过 builder 创建结果与设备行为确认,不从枚举存在推断硬件支持。
## 6. 风险与验证
1. write callback 的实时线程约束和最大执行时间。
2. `GetTimestamp` 在暂停、Flush、蓝牙/有线切换后的连续性。
3. underrun、首包预填充与首音频时间。
4. Surface 暂停但音频继续、应用后台、音频焦点中断。
5. 播放速度与 MPV 时钟双重调整风险。
6. 多声道 layout 枚举和实际设备 route 不一致。
代码验收:PCM 正常播放;pause/resume/seek/flush/release 生命周期已在模拟器通过;starvation 与 timestamp metrics 已接入。1 小时稳定性、真实 underrun、A/V sync 偏差、路由与多声道仍需实体设备,不宣称低延迟或多声道完成。
-65
View File
@@ -1,65 +0,0 @@
# HarmonyOS Native Capability Research
## 1. 取证基线
- SDK:`/Applications/DevEco-Studio.app/Contents/sdk/default/openharmony`
- Native / Toolchains:`6.1.1.125`,API 24
- 目标:`aarch64-linux-ohos`
- 编译器:OHOS Clang 15.0.4
- 最小头文件探针已通过:pthread、dlopen、mmap、epoll、poll、socket、DNS、EGL、GLES3、Vulkan、XComponent、OHAudio、Native AVCodec 可被同一 C translation unit 包含。
- 编译器预定义:`__OHOS__`、`__linux__`、`__unix__`、`__aarch64__`、`__LP64__`。EGL 平台头使用 `OHOS_PLATFORM` 选择 `OHNativeWindow *`,构建时必须显式定义该宏。
以下“最低 API”来自本机 API 24 SDK 头文件的 `@since` / availability;运行行为仍需 Phase 1/2 真机验证。
## 2. 能力矩阵
| 能力 | C/C++ API / header | library | 最低 API | 生命周期与限制 | 结论 |
|---|---|---|---:|---|---|
| ArkUI Surface | `OH_NativeXComponent_RegisterCallback`;`OnSurfaceCreated/Changed/Destroyed`;`ace/xcomponent/native_interface_xcomponent.h` | `libace_ndk.z.so` | 8 | `window` 只在组件 Surface 生命周期内有效;销毁前停止渲染并释放 EGL/Vulkan surface | 可用 |
| Native Window | `OHNativeWindow`;`native_window/external_window.h` | `libnative_window.so` | 8 起,部分新 buffer API 23 | XComponent 回调交付 window;不得替换成 Android `ANativeWindow` | 可用 |
| Native Buffer | `OH_NativeBuffer_*`;`native_buffer/native_buffer.h` | `libnative_buffer.so` | 11 起,具体函数不同 | 用于后续互操作;格式、usage、fence 必须真机查询 | 可用,Phase 0 不承诺零拷贝 |
| OpenGL ES | `<GLES2/gl2.h>`、`<GLES3/gl3.h>` | `libGLESv2.so` / `libGLESv3.so` | SDK 基础能力 | 实际 GLES 版本、扩展、10-bit/HDR 格式按设备查询 | 可用 |
| EGL | `eglGetDisplay`、`eglCreateWindowSurface`、`eglSwapBuffers`;`<EGL/egl.h>` | `libEGL.so` | SDK 基础能力 | `EGLNativeWindowType` 在 `OHOS_PLATFORM` 下是 `OHNativeWindow *`;上下文线程归属必须固定 | 可用,首选渲染入口 |
| Vulkan | `VK_OHOS_surface`、`vkCreateSurfaceOHOS`;`<vulkan/vulkan.h>` / `vulkan_ohos.h` | `libvulkan.so` | 10 | Surface 绑定 `OHNativeWindow`;扩展和 format 支持需设备枚举 | 可用,但不是 libmpv 公共 Render API |
| VSync | `OH_NativeVSync_*`;`native_vsync/native_vsync.h` | `libnative_vsync.so` | 9 | 回调与 Surface 生命周期协调 | 可用,首版可先由 EGL swap/MPV 调度 |
| Native Audio | `OH_AudioStreamBuilder_*`、`OH_AudioRenderer_*`;`ohaudio/*.h` | `libohaudio.so` | 10 | builder 生成 renderer 后销毁;renderer 使用写回调拉取 PCM;Start/Pause/Stop/Flush/Release 有明确状态机 | 可用,需新增 AO |
| Native Video Decoder | `OH_VideoDecoder_*`;`multimedia/player_framework/native_avcodec_videodecoder.h` | `libnative_media_vdec.so`、`libnative_media_codecbase.so`、`libnative_media_core.so`、`libnative_window.so` | 9;buffer 新接口 11 | Create→Configure→SetSurface→Prepare→Start;Seek 需要 Flush 后重启;输出可直接 RenderOutputBuffer 到 Surface | 可用 |
| Codec Capability | `OH_AVCodec_GetCapability*`、`OH_AVCapability_*`;`native_avcapability.h` | Native media libs | 10 | 必须在目标设备查询硬件/软件、尺寸、帧率、profile/level;SDK 头不能证明某机型支持 4K/10-bit/HDR | 可用,能力值待真机 |
| POSIX file | `open/read/write/stat`;`fcntl.h`、`unistd.h`、`sys/stat.h` | libc | 基础 | 受应用沙箱与授权 URI 限制;ArkUI picker URI/FD 需桥接到 native | 可用但路径权限需集成层处理 |
| Packaged rawfile | `OH_ResourceManager_*` / `rawfile/*.h` | `librawfile.z.so` | 8 起 | 只适合包内资源,不是通用媒体文件系统 | 可用,不替代 POSIX file |
| pthread | `pthread_create`、mutex/cond;`pthread.h` | libc | 基础 | MPV `threads-posix.c` 可优先复用;线程名函数需 Meson 实测 | 可复用 |
| socket | BSD sockets;`sys/socket.h`、`netinet/in.h` | libc | 基础 | 需要网络权限;移动网络切换、IPv6、代理需真机测试 | 可复用 |
| DNS | `getaddrinfo/freeaddrinfo`;`netdb.h` | libc | 基础 | DNS/网络切换错误传播由 FFmpeg 网络层处理 | 可复用 |
| Native HTTP | `OH_Http_*`;`network/netstack/net_http.h` | `libnet_http.so` | 20 | 与 FFmpeg AVIO 不直接兼容;不作为首版 MPV 网络栈 | API 24 可用但不采用 |
| TLS certificate verification | `OH_NetStack_CertVerification` 等;`network/netstack/net_ssl/net_ssl_c.h` | `libnet_ssl.so` | 11 | 这是证书链验证接口,不是 FFmpeg TLS transport backend;首版需交叉构建 FFmpeg 支持的 TLS 库 | 系统验证能力可用,MPV 路径需第三方 TLS |
| Shared library | `dlopen/dlsym/dlclose`;`dlfcn.h` | `libdl.a` / libc | 基础 | 插件路径和符号可见性受沙箱约束 | 编译可用,运行待测 |
| Timer | `clock_gettime`、`nanosleep`;`time.h` | libc | 基础 | MPV `timer-linux.c` 实际只依赖这两个 POSIX API;源码应改名/条件化以免平台语义错误 | 可复用实现,需整理检测 |
| Memory mapping | `mmap/munmap`;`sys/mman.h` | libc | 基础 | 文件映射权限及大文件压力需真机测试 | 编译可用 |
| poll / epoll | `poll`、`epoll_*`;`poll.h`、`sys/epoll.h` | libc | 基础 | 头文件存在且探针通过;运行语义、FD 类型兼容性仍需测 | 可用,优先 POSIX poll |
| System information | `uname`、`sysconf`、`getauxval`;`sys/utsname.h`、`unistd.h`、`sys/auxv.h`;`OH_GetDeviceType` (`deviceinfo.h`) | libc / `libdeviceinfo_ndk.z.so` | API 10 起(DeviceInfo) | 只用于诊断与保守能力探测,不按设备名硬编码功能 | 可用 |
| CPU detection | AArch64 编译宏、`getauxval`;FFmpeg CPU detection | libc | 基础 | arm64 基线已有 NEON;不得未经实测强开额外 ISA | 可复用 |
| Lifecycle | XComponent Surface callbacks + N-API/ArkTS 页面生命周期桥接 | `libace_ndk.z.so` / `libace_napi.z.so` | 8 | MPV Core 仅接受 surface attach/detach、pause/resume 等播放器级事件;业务生命周期留在宿主 | 需薄桥接 |
| Power | 系统 running lock / 后台任务 API主要由应用层管理 | 系统服务 | 版本依 API | Core 不自行申请长期后台权限;向宿主暴露播放状态,由宿主管理屏幕常亮/后台播放 | 不建 MPV backend |
## 3. 直接可复用的 MPV/POSIX 区域
- `osdep/threads-posix.c`:pthread 基础满足。
- `osdep/path-unix.c`、`osdep/language-posix.c`:大部分 POSIX 路径/locale 逻辑可编译,沙箱路径语义需测试。
- `osdep/timer-linux.c` 的实现仅用 `clock_gettime`/`nanosleep`,能力可复用,但构建选择需改为 `ohos` 明确分支或重命名为 POSIX timer。
- socket、DNS、poll、mmap、dlopen 头文件均存在,先通过 feature detection,不因 `__linux__` 自动宣称全部 Linux 特性可用。
- demux、字幕、滤镜、同步、命令和 libmpv client API 与平台无关,可直接复用。
## 4. 必须新增或调整的边界
1. Meson 增加 `host_machine.system() == 'ohos'` / `features['harmony']`,生成 `HAVE_HARMONY`,防止把 OHOS 的 `__linux__` 等同桌面 Linux。
2. 新增 OHAudio AO,不能复用 Android AAudio/OpenSL ES。
3. 若构建 MPV 自管 VO,新增 Harmony EGL/Vulkan context;首版 Playback Lab 可由宿主持有 EGL context,使用 libmpv OpenGL Render API,减少 upstream patch。
4. 硬解需要 MPV 识别 FFmpeg 的 `AV_PIX_FMT_OHCODEC` 与 `AV_HWDEVICE_TYPE_OHCODEC`;该工作晚于软件解码首帧。
5. 日志可先复用 stderr;实验室集成阶段增加 HiLog sink,不能让日志 backend 阻塞首次构建。
## 5. 未验证项
- 各目标设备具体 Vulkan/GLES 扩展、HDR/10-bit swapchain、codec profile/level、最大分辨率/码率。
- `epoll`、`dlopen`、mmap 大文件在应用沙箱内的运行行为。
- 后台播放、音频焦点、路由切换、蓝牙延迟和 Surface 重建行为。
- Native HTTP/TLS 与系统代理/证书策略;首版按 FFmpeg + 独立 TLS 依赖验证。
-7
View File
@@ -1,7 +0,0 @@
# Harmony 字体策略
第一优先级使用系统字体;应用字体只允许由产品明确提供的开源字体;libass 使用 mpv 已配置的字体 provider,不新增商业字体包。当前 OHOS 字体栈按 `scripts/build-font-stack-ohos.sh` 构建:FreeType + HarfBuzz + FriBidi + libass,libass 明确关闭 fontconfig、DirectWrite、CoreText 和 iconv provider,因此不依赖 Linux `fonts.conf`。
Playback Lab 初始化时将 mpv `sub-fonts-dir` 指向 `/system/fonts`;模拟器已确认该目录存在,`/system/etc/fontconfig` 与 `/system/etc/fonts` 不存在。libass 的 `ass_set_fonts_dir` 会从该目录加载字体,`sub-font-provider=auto` 在当前无系统 provider 时仍可使用目录字体和字幕内嵌字体。中文、英文、日文、韩文的缺字由目录字体顺序、libass fallback 与内嵌字体共同决定;应用若需补充字体,应通过 Resource Manager/应用沙箱提供明确的开源字体目录并在产品层映射到 `sub-fonts-dir`,不修改核心、不提交商业字体。
Playback Lab 只验证字幕渲染和缺字表现,不把字体文件或用户凭证提交到仓库。设备侧若系统字体目录、provider 行为或 fallback 顺序不同,应记录实际字体、脚本和缺字结果。
-64
View File
@@ -1,64 +0,0 @@
# HarmonyOS Hardware Decode Research
## 1. 关键发现
截至 FFmpeg 官方 master `3acec0a1af2dda0a0838689b8b8649e7deb080a0`(2026-08-28),upstream 已包含 OpenHarmony Codec:
- configure:`--enable-ohcodec`
- hw device:`AV_HWDEVICE_TYPE_OHCODEC`
- pixel format:`AV_PIX_FMT_OHCODEC`
- public device context:`AVOHCodecDeviceContext.native_window`
- decoder/encoder:`h264_oh`、`hevc_oh`
- SDK linkage:`native_media_vdec`、`native_media_venc`、`native_media_codecbase`、`native_media_core`、`native_window`
其 surface 路径把 `OHNativeWindow *` 放入 hw device,FFmpeg 配置 `AV_PIXEL_FORMAT_SURFACE_FORMAT`,并在 AVFrame release 时调用 `OH_VideoDecoder_RenderOutputBuffer()`。因此“不存在 FFmpeg Harmony hwcontext”已不是事实;当前缺口主要转移到 MPV 对 OHCODEC 的识别、Surface 所有权和呈现同步。
## 2. Codec 矩阵
| Codec | Harmony Native Support | FFmpeg Support | MPV Support | Surface Decode | Pixel Format | Copy | Zero Copy Feasibility | 10-bit / HDR | Known Limitation | Complexity |
|---|---|---|---|---|---|---|---|---|---|---|
| H.264 | Native decoder API 存在;具体 profile/4K 由设备 capability 查询 | `h264_oh` 已在 upstream master | 无 OHCODEC 映射/backend | 已实现 | `AV_PIX_FMT_OHCODEC` / surface format | 解码器到 Surface 可避免 CPU frame;是否经过系统内部 copy 不可见 | 高,但直出播放 Surface 与 MPV shader/字幕合成冲突 | 设备查询;H.264 10-bit 通常不作为首要目标,不能猜 | FFmpeg 实现新、MPV 未接入、字幕/OSD 合成路径未解决 | 高 |
| HEVC | Native decoder API 存在;profile/level/10-bit 待设备查询 | `hevc_oh` 已在 upstream master | 无 OHCODEC 映射/backend | 已实现 | 同上 | 同上 | 高 | Main10/HDR 必须按真机 capability + 实际流验证 | HDR metadata 传递和 Surface colorspace 未证明 | 高 |
| AV1 | SDK API 可按 MIME 查询,不代表设备必有硬件 | 当前 OHCodec backend 未声明 `av1_oh` | 无 | FFmpeg OH 路径无 | 无 | 未定 | 未定 | 未定 | upstream FFmpeg 尚未接 OH AV1 | 很高 |
| VP9 | SDK API可查询,不代表设备必有硬件 | 当前 OHCodec backend 未声明 `vp9_oh` | 无 | FFmpeg OH 路径无 | 无 | 未定 | 未定 | 未定 | upstream FFmpeg 尚未接 OH VP9 | 很高 |
## 3. MPV 缺口
1. `video/fmt-conversion.c` 没有 `IMGFMT_OHCODEC <-> AV_PIX_FMT_OHCODEC`。
2. `video/decode/vd_lavc.c` 尚无 OHCODEC device/backend 选择。
3. GPU interop driver 列表没有 Harmony/OHNativeBuffer 路径。
4. libmpv OpenGL Render API的目标 Surface与 OHCodec 直接输出 Surface存在所有权冲突:同一个播放画面若让 decoder 直接呈现,MPV 的 shader、OSD、ASS 字幕、色彩管理无法自然叠加。
5. FFmpeg 当前 OHCodec surface frame 在 AVFrame release 时触发呈现,呈现时钟、drop/late frame 与 MPV VO 调度需要逐帧语义验证。
## 4. 推荐路线
### Stage 1:软件解码基线
先以 FFmpeg software decode + libmpv GLES Render API 跑通首帧、Seek、字幕、网络和稳定性。该基线用于证明其后优化收益。
### Stage 2:FFmpeg OHCodec 可用性验证
独立验证官方 `h264_oh` / `hevc_oh`:创建 decoder、Flush/Seek、Surface destroy/recreate、4K、高码率、Main10;记录设备 capability。此阶段不修改 MPV 大量代码。
### Stage 3:选择合成架构
- 路线 3A,直接 Surface decode:最快接通、copy 少,但 MPV GPU shader/ASS/OSD 不能直接叠加。只适合验证硬解吞吐,不直接定为正式架构。
- 路线 3B,OHNativeBuffer/GPU import:decoder 输出可导入 GLES/Vulkan,再由 MPV/libplacebo 合成。最符合最终产品,但 FFmpeg 当前 OHCodec API是否暴露可导入 buffer 尚需源码和真机证明。
- 路线 3C,双 Surface/system overlay:视频与字幕分层,由系统合成。同步、HDR、截图和滤镜语义复杂,仅作备选。
正式路线优先研究 3B;在其不可行证据充分前,不宣称零拷贝。
## 5. Patch 边界
如进入实现,优先局限在:
- `video/fmt-conversion.c` / image format 定义
- `video/decode/vd_lavc.c` 的 hw device 选择
- `video/out/hwdec/hwdec_ohcodec.c` 或明确命名的 Harmony interop driver
- `meson.build` Harmony feature gate
任何不支持格式明确返回 unsupported;不得以 CPU 黑帧、fake success 或注释掉检查伪装完成。
## 6. 当前判定
**不创建 Velix-FFmpeg-Harmony。** 官方 FFmpeg 已有关键 OHCodec 基础,Phase 1 应固定官方 commit 并先编译;只有真实 blocker 证明需要 patch 且无法快速 upstream 后,才建立独立 FFmpeg fork。AV1、VP9、buffer import、HDR metadata 仍为 `NEED VERIFY`。
-74
View File
@@ -1,74 +0,0 @@
# HarmonyOS Rendering Research
## 1. 结论
第一阶段推荐 **OpenGL ES + EGL + libmpv OpenGL Render API**:ArkUI XComponent 仅交付 `OHNativeWindow *`,Native 层创建 EGL window surface/context,并在 GL 线程调用 `mpv_render_context_render()` 渲染到默认 FBO,随后 `eglSwapBuffers()`。视频帧不经过 ArkTS/N-API。
```text
XComponent Surface
-> OHNativeWindow
-> EGLSurface + GLES context(Native 渲染线程)
-> libmpv OpenGL Render API
-> GPU render -> eglSwapBuffers -> Display
```
依据:本机 API 24 SDK 明确把 `EGLNativeWindowType` 定义为 `struct NativeWindow *`;MPV 公共 Render API 当前只公开 `MPV_RENDER_API_TYPE_OPENGL` 与软件渲染,未公开 Vulkan Render API。
## 2. 方案 A:OpenGL ES / EGL
### 调用边界
1. `OH_NativeXComponent_RegisterCallback()` 接收 Surface 创建/变化/销毁。
2. `OnSurfaceCreated` 的 `void *window` 按 SDK 契约转换为 `OHNativeWindow *`。
3. Native GL 线程执行 `eglGetDisplay`、`eglInitialize`、`eglChooseConfig`、`eglCreateWindowSurface`、`eglCreateContext`。
4. 用 `MPV_RENDER_PARAM_API_TYPE=MPV_RENDER_API_TYPE_OPENGL` 和 `mpv_opengl_init_params` 创建 render context。
5. update callback 只唤醒渲染线程;该线程 make-current,调用 `mpv_render_context_render()`,目标为默认 FBO 0 和当前 Surface 尺寸,再 swap。
6. Surface changed 更新 viewport/FBO 尺寸;Surface destroyed 先停止提交帧,再释放 render context 的 surface 相关状态和 EGLSurface。
### 评估
| 维度 | 结论 |
|---|---|
| 实现复杂度 | 中;MPV 已有完整 OpenGL Render API,Harmony 只负责 EGL/XComponent 生命周期 |
| MPV 支持度 | 高;公共 API 稳定,`include/mpv/render.h`、`render_gl.h` 为直接依据 |
| HarmonyOS 支持度 | 高;API 24 提供 EGL、GLES2/3、OHNativeWindow |
| Surface 生命周期 | 必须单线程串行化 create/change/destroy,销毁后禁止 swap |
| Copy 次数 | 软件解码时 FFmpeg frame 上传 GPU,至少一次 CPU→GPU;无逐帧 N-API copy |
| 4K | 渲染路径可行,性能取决于上传、shader、带宽;需 Benchmark |
| 10-bit | 软件帧可由 MPV shader 处理;Surface 色深/输出格式待设备 EGLConfig 验证 |
| HDR | MPV 色彩处理可复用,但 Harmony Surface 色域、transfer、HDR metadata 通路尚未证明 |
| 风险 | EGL context 线程归属、Surface 快速重建、默认 FBO 方向、vsync/帧节奏 |
### MPV 修改范围
- Phase 1 构建 `libmpv` 时启用 `plain-gl`,无需先新增完整 `vo_harmony`。
- Playback Lab 的 Harmony glue 独立放在后续 `playback_lab/native/`,不把 ArkTS 业务代码放进 MPV core。
- 只有确认必须让 MPV 自管 window/context 时,才新增 `video/out/opengl/context_harmony.c` 并在 Meson 中以 `features['harmony']` 注册。
## 3. 方案 B:Vulkan
### 可行基础
API 24 提供 `VK_OHOS_surface`、`VkSurfaceCreateInfoOHOS`、`vkCreateSurfaceOHOS()` 与 `libvulkan.so`,可从 `OHNativeWindow *` 创建 `VkSurfaceKHR`。MPV 内部 `vo_gpu_next` / libplacebo 支持 Vulkan,但 libmpv 公共 Render API 没有 Vulkan 类型。
| 维度 | 结论 |
|---|---|
| 实现复杂度 | 高;需要新增/扩展 MPV Vulkan context 或设计受支持的嵌入路径 |
| MPV 支持度 | 内部高、公共嵌入 API 低 |
| HarmonyOS 支持度 | API 10+ 有 surface 扩展,设备扩展需枚举 |
| Surface 生命周期 | 需重建 swapchain,处理 out-of-date/suboptimal、旋转与尺寸变化 |
| Copy 次数 | 软件解码仍需上传;未来 native buffer/Vulkan interop 有降 copy 潜力 |
| 4K | 理论适合,未实测不能优于 GLES |
| 10-bit/HDR | Vulkan format/colorspace 路径潜力更强,但 Harmony 设备与系统合成支持待证 |
| 风险 | libplacebo/Vulkan 版本、OHOS surface 接入、swapchain、同步/fence、驱动差异 |
## 4. 禁止路径
`MPV -> CPU frame -> Native Buffer -> ArkTS -> ArkUI Image` 不进入正式架构。它会增加逐帧跨 N-API、CPU copy、GC/调度与背压风险,也无法作为 4K 主路径。
## 5. Phase 门禁
- Phase 1:只证明 `libmpv` arm64 编译,OpenGL Render API 符号存在。
- Phase 2:Playback Lab 证明 XComponent→EGL→首帧;记录 Surface 重建和旋转。
- Phase 3:软件解码基线后再接 OHCodec;分别测 CPU upload 与 surface decode。
- Vulkan 只有在 GLES 基线与瓶颈证据齐备后启动,不以“理论更快”替代 Benchmark。
-7
View File
@@ -1,7 +0,0 @@
# OHCodec HDR pipeline notes
本阶段只保留 metadata 研究,不声明 HDR 完成。
FFmpeg upstream OHCodec stream-change 路径会把存在的 matrix、primaries、transfer、range 和 SAR 元数据复制到 `AVCodecContext`;MPV 现有 color-management pipeline 继续消费这些 metadata。设备是否输出 10-bit、支持 Main10 或 HDR surface,必须由 API24 capability 和真实设备 frame metadata 证明。
当前实现不添加 HDR transfer、tone mapping、Dolby Vision 或 10-bit surface 特判。Playback Lab 的 capability/diagnostics 只显示真实查询结果;没有设备数据时显示 `DEVICE TEST REQUIRED`,不得显示 `HDR YES`。
-79
View File
@@ -1,79 +0,0 @@
# HarmonyOS First Native Build Report
## Result
Phase 1 已在 DevEco Studio OpenHarmony SDK 6.1.1.125 / API 24 下生成真实 arm64 `libmpv`。本报告只证明编译、链接和静态检查,不声明真机运行语义已验证。
| 项目 | 结果 |
|---|---|
| Origin | `https://git.leee.xin/leeexin/Velix-MPV-Core.git` |
| origin/harmony-port | 以本次 Phase 1 提交推送后的 commit 为准 |
| MPV | `182fa6ca49f455cadb884858f386e2f00540aeb7` / 0.41.0-UNKNOWN |
| FFmpeg | `3acec0a1af2dda0a0838689b8b8649e7deb080a0` |
| libmpv | `build/ohos-arm64/prefix/lib/libmpv.so.2.5.0` |
| Size | 34,314,184 bytes |
| SHA-256 | `80e4c6fda86d9822543615a97045fff3641e9125b23a4f736fb8511fe5a47790` |
| Architecture / ABI | ELF64, little-endian, AArch64, System V ABI, shared object, SONAME `libmpv.so.2` |
| Toolchain | OHOS Clang/LLD 15.0.4, target `aarch64-linux-ohos`, musl sysroot |
| MPV / build files | 7:`ao_harmony.c/.h`、AO 注册、Meson feature、OHOS build script、依赖 allowlist;第三方依赖源码修改 0 |
| Build issues | 4,全部关闭 |
## Features
MPV enabled:`clone`、`ffmpeg`、`gl`、`glibc-thread-name`、`glob`、`glob-posix`、`gpl`、`harmony-audio`、`libass`、`libdl`、`libplacebo`、`linux-fstatfs`、`memrchr`、`ohos`、`posix`、`posix-shm`、`ppoll`、`pthread-condattr-setclock`、`vector`、`vt.h`、`zlib`、`libmpv shared`、`plain-gl`。
FFmpeg enabled:软件解码、demux、network、HTTP/HTTPS/TCP/TLS、OpenSSL 3.6.4、SDK zlib、OHCodec H.264/HEVC decoder/encoder。FFmpeg GPL/version3/nonfree 均为 0。
Disabled:mpv CLI、tests、fuzzers、Lua/JavaScript、libcurl、桌面音视频后端、X11、Wayland、DRM/GBM、Cocoa、Windows backend、Android-specific backend、Vulkan、VAAPI、VDPAU、CUDA。libplacebo 的 Vulkan/OpenGL 自有 backend 关闭;MPV 自身 OpenGL Render API 保留。
## Verification
- 最小 C/C++ 探针均为 OHOS AArch64 ELF;C++ 探针依赖 SDK `libc++_shared.so`。
- prefix 内 13 个静态库的全部可见 archive object 均通过 AArch64 ELF 检查。
- `libmpv.so.2.5.0` 动态依赖为 `libc.so`、`libz.so`、`libnative_media_vdec.so`、`libnative_media_venc.so`、`libnative_media_codecbase.so`、`libnative_media_core.so`、`libnative_window.so`、`libohaudio.so`;均来自 HarmonyOS SDK allowlist。
- target metadata 未发现 Homebrew、`/usr/local`、`.dylib`、Apple framework 或 Mach-O 污染。
- 已验证 `mpv_create`、`mpv_initialize`、`mpv_command`、`mpv_set_property`、`mpv_get_property`、`mpv_render_context_create`、`mpv_render_context_render`、`mpv_render_context_set_update_callback`、`mpv_render_context_free`。
- `scripts/build-all-ohos.sh` 已完成本次 Phase 3 全链路构建与最终校验(`verified_archives=13`)。
## Warnings
剩余 18 条构建 warning:FFmpeg 5 条 upstream fallthrough/unused-function;FriBidi 5 条 enum-conversion;HarfBuzz 1 条仅开发者需要 Ragel;libass/MPV 4 条非阻塞 CMake compiler-state 探测;libplacebo 3 条无 graphics backend/name-prefix 提示。均不影响当前 target 产物,未通过关闭警告掩盖问题。
## Dependency Graph
实际依赖图见 `DOCS/PHASE1_DEPENDENCY_MATRIX.md`。所有第三方库均静态进入 `libmpv`,SDK zlib、OHCodec 与 libc 保持系统动态依赖。
## License
仓库与 `libmpv` 当前按 MPV GPL-2.0-or-later 路线管理。FFmpeg 为 LGPL 配置且未启用 GPL/nonfree;OpenSSL 为 Apache-2.0;静态 LGPL 组件仍需在正式发布时履行可重新链接、源码与 notices 义务。本文件不是最终发行 notice。
## Phase 2 Risks
1. POSIX、DNS、HTTP/HTTPS 与动态库加载目前仅 compile/link verified,尚未真机 runtime verified。
2. XComponent / OHNativeWindow 生命周期、线程归属、窗口销毁与 EGL context loss 尚未验证。
3. MPV plain-gl 与 HarmonyOS EGL/GLES 的实际格式、扩展、同步和首帧路径尚未联调。
4. FFmpeg OHCodec 已编译,但 MPV hwdec 选择、surface/output buffer 生命周期及零拷贝均未验证。
5. OpenSSL CA 路径与 HAP 内依赖打包、SDK系统库版本兼容仍需真机验证。
## Phase 2 Entry Condition
已具备 XComponent + OHNativeWindow + EGL + libmpv OpenGL Render API 真机首帧测试的编译侧条件:SDK API 可用、libmpv 可链接、OpenGL Render API symbols 齐全。尚不代表首帧已经成功;Phase 2 必须在真机建立窗口和 EGL context 后验证。
## FFmpeg-Harmony export prefix
Phase 4.5 的 FFmpeg fork 独立维护于
`https://git.leee.xin/leeexin/Velix-FFmpeg-Harmony.git`,分支
`harmony-ohcodec-export`,当前分支 tip `f3a036c695`(跨 origin 重定向丢弃敏感凭证;代码修正 commit
`aacbe7d578`)。MPV 不包含 FFmpeg
源码,也不读取其私有头文件;需要切换时执行:
```sh
./scripts/build-mpv-ohos-ffmpeg-harmony.sh
```
脚本默认消费 `../Velix-FFmpeg-Harmony/build/ohos-arm64/prefix`,也可用
`FFMPEG_PREFIX=/path/to/prefix MPV_BUILD_ROOT=/path/to/mpv-build MPV_PREFIX=/path/to/mpv-prefix` 覆盖。切换脚本默认把 MPV 安装到
`build/ohos-arm64/prefix-mpv-ffmpeg-harmony`,避免覆盖默认 prefix。默认
`build-mpv-ohos.sh` 与 `build-mpv-ohos-software.sh` 的原有前缀和
`harmony-ohcodec-gl=disabled` 行为不变。本次切换构建已通过 AArch64 ELF、符号
和动态依赖校验,但没有进入 GPU mapper 或设备硬解验证。
-20
View File
@@ -1,20 +0,0 @@
# MPV 与 OHCodec 缺口分析
扫描基线:MPV 当前 `harmony-port` 工作树,FFmpeg OHCodec 头文件来自锁定 commit `3acec0a1af2dda0a0838689b8b8649e7deb080a0`。
| MPV location | 当前行为 | OHCodec requirement | 本阶段改动/结论 | 风险与 upstreamability |
|---|---|---|---|---|
| `video/img_format.h` / `video/fmt-conversion.c` | 只有通用 hardware format,没有 OHCodec 专用 ID。 | `AV_PIX_FMT_OHCODEC` 必须成为 MPV hardware image format。 | 新增 `IMGFMT_OHCODEC ↔ AV_PIX_FMT_OHCODEC` 单一映射。 | 纯 MPV 架构改动,可 upstream;依赖 FFmpeg enum。 |
| `video/decode/vd_lavc.c:add_all_hwdec_methods()` | 动态枚举 FFmpeg HW config,并使用 `av_hwdevice_get_type_name()` 生成 method。 | 能识别 `AV_HWDEVICE_TYPE_OHCODEC`、`h264_oh`、`hevc_oh`。 | 不改 decoder core;FFmpeg config 会自然生成 `hwdec=ohcodec`,且仅 H.264/HEVC 由 adapter 放行。 | upstreamable;真实 capability 仍需设备验证。 |
| `video/hwdec.c:hwcontext_fns[]` | 未注册 OHCodec copy-device callback;未知类型会尝试空 device。 | copy path 必须不能伪造 null-surface 成功。 | 注册 `hwcontext_fns_ohcodec`,明确拒绝无独占窗口的 copy path,并触发 software fallback。 | 纯 MPV 保护逻辑。 |
| `video/out/gpu/hwdec.c` | 只注册已有 GPU interop/overlay drivers。 | Direct Surface 需要独立 backend,不能复用 MediaCodec/DRM。 | 注册 `ohcodec-direct-surface-experimental`,仅在 host 提供 `ohos_direct_surface` native resource 时创建设备。 | 当前 Playback Lab 不提供该 resource,故默认不会抢占 EGL Surface。 |
| `video/out/gpu/video.c` / `gpu_next` | mapper 或 overlay 二选一;overlay 会绕过视频 GL 合成但保留 OSD。 | Direct Surface 不能与同一 EGL window 双重呈现。 | OHCodec 只使用 overlay probe;没有独占窗口时 init 失败。 | Direct Surface 不是最终 ASS/shader/color-management 产品路径。 |
| `video/mp_image.c:mp_image_from_av_frame()` | 复制 AVFrame buffers/data;无 OHCodec 特判。 | frame 需要保留 `buf[0]` 生命周期和硬件格式。 | 现有 generic path 已满足:`data[3]` 作为 opaque plane,`buf[0]` 持有 release。 | 不能把 opaque plane 当 CPU pixels 或 fd。 |
| `video/out/gpu/ra.h` native resources | 可由 host 注入平台资源。 | Surface ownership 需要明确边界。 | 约定 `ohos_direct_surface` 为独占、已由 host 管理的 `OHNativeWindow*`。 | 当前没有 host 注入,设备验证前不能宣称 direct active。 |
| Diagnostics / Playback Lab | 原来只显示 software audio/video 状态。 | 必须显示 requested/actual decoder、hw state、fallback/capability。 | 新增导出统计 API、日志识别和 Playback Lab 面板字段。 | actual decoder 在设备上需以 FFmpeg 日志和真实 frame 验证。 |
## 选择路径
`hwdec=ohcodec` → FFmpeg OHCodec method → MPV 请求 `IMGFMT_OHCODEC` → GPU backend 请求 OHCodec device → adapter 查询 capability → 检查独占 Surface → 创建 `AVHWDeviceContext` → Direct Surface overlay。
任何一步失败均返回负值并保留 software decoder。GPU interop 不走猜测性的 opaque-pointer reinterpretation。
-13
View File
@@ -1,13 +0,0 @@
# 网络播放
网络播放复用 mpv/FFmpeg 的单一网络栈,不在产品中创建第二 HTTP client。支持范围是 mpv 已提供的 HTTP、HTTPS、HLS、重定向、Range seek、chunked 和大范围 seek 能力。
每次 load 先清空 http-header-fields,再设置本次请求的 headers;Cookie 作为本次请求的 Cookie header 发送,未启用 mpv 持久化 cookie 文件。请求可设置 user-agent、referrer、network-timeout、语言偏好和轨道选择。
网络诊断只报告实际可取得的 mpv 属性:协议、host、networkState、cache duration/bytes、demuxer、container、当前 bitrate,以及从实际日志识别到的 HTTP status/redirect count;`rangeSupported` 仅在观察到 206 响应时为 1,`seekResult` 记录最近一次 seek 的 `ok` 或脱敏错误。没有对应属性或日志时返回 -1/0/空值,不做推断。
HTTP MP4 已在 Playback Lab 模拟器通过宿主受控 Range server 验证:请求日志出现 `bytes=0-`、后续大范围 byte range,播放结束且无核心错误。跨 origin 重定向的敏感头过滤由 FFmpeg-Harmony 固定分支提供,实际 HTTPS/HLS/鉴权资源仍需受控网络源继续验收。
HLS Master/Variant 的当前媒体分辨率来自 video-params/width/height,码率来自 demux-bitrate;当前 MPV 没有独立稳定的“variant 名称”属性,因此不新增伪造字段。
DNS、TCP connect、connection reset、timeout 和 FFmpeg open failure 统一归入 NETWORK,并保留脱敏原始消息;TLS handshake/certificate 归入 TLS,HTTP status 归入 HTTP,demux/decoder 错误不伪装成网络错误。
-15
View File
@@ -1,15 +0,0 @@
# 网络安全
## Header 与会话隔离
每次 load 都覆盖 http-header-fields,不共享上一个 session 的 header、Cookie 或 URL。禁止把认证信息写入仓库、默认日志或诊断快照。Playback Lab 输入框可以编辑 header,但诊断默认只显示字段名和掩码值。
## 脱敏规则
Authorization、Proxy-Authorization、Cookie、X-Api-Key、X-Emby-Token、X-Auth-Token、X-Access-Token 及 token/api-key/signature 等认证头,以及 URL 中的 token、access_token、auth、key、api_key、apikey、auth_token、signature、x-amz-signature、sig、expires 参数在错误文本和日志中替换为 [REDACTED]。原始 mpv 日志不得直接写入 OH 日志。
## TLS、重定向与凭证
默认保持 tls-verify=yes。mpv 网络后端负责 TLS、重定向、Range 和 HLS;本层不复制 HTTP 行为。固定的 FFmpeg-Harmony 分支 `f3a036c695` 已在 `libavformat/http.c` 实现跨 origin(协议/host/port 任一变化)重定向时丢弃自定义 Authorization/Cookie/Token/API-key/Referer 等敏感头及 cookie 状态,同源重定向仍保留请求头;因此不会把凭证无条件转发到不同 host。Referer、User-Agent、header 仅属于当前加载请求。
实现依据:mpv 官方手册的 user-agent、referrer、cookies、http-header-fields、network-timeout、HLS 与 sub-add 选项说明。
-34
View File
@@ -1,34 +0,0 @@
# OHCodec GPU interop 设计与当前 blocker
## 目标
将 OHCodec hardware frame 交给 MPV OpenGL composition,同时保留 ASS、OSD、shader 和 color management。该路径必须是显式实验开关 `harmony-ohcodec-gl`,不能成为默认 hwdec。
## API24 合法链路(仅设计验证)
SDK 头文件提供:
1. `OH_AVBuffer_GetNativeBuffer(OH_AVBuffer*)`,返回一个需要 `OH_NativeBuffer_Unreference()` 的 `OH_NativeBuffer*`。
2. `OH_NativeWindow_CreateNativeWindowBufferFromNativeBuffer()`,把 NativeBuffer 转成 `OHNativeWindowBuffer*`,并要求随后销毁 wrapper。
3. `EGL_OHOS_image_native_buffer` / `EGL_NATIVE_BUFFER_OHOS`、`eglCreateImageKHR()` 和 `glEGLImageTargetTexture2DOES()`。
这证明 API 类型是 HarmonyOS 原生接口,不应使用 Android `AHardwareBuffer`、fd 猜测或指针重解释。未来 mapper 的资源生命周期应为:
`OH_AVBuffer` frame ref → `OH_NativeBuffer_Get/Reference` → `OHNativeWindowBuffer` wrapper → EGLImage → external texture → unmap 时反向销毁。
## OHCODEC-BLOCKER-001
当前 upstream `libavcodec/ohdec.c` 只有两种输出:
- 有 OHCodec device:强制 `output_to_window=true`、配置 `AV_PIXEL_FORMAT_SURFACE_FORMAT`、`SetSurface(window)`;
- 无 device:CPU NV12 buffer,复制到 software frame。
hardware frame 的 `OH_AVBuffer*` 只存在于带 Surface 的分支,且自定义 `AVBuffer` release 会调用 `OH_VideoDecoder_RenderOutputBuffer()`。因此 MPV 不能在不修改 FFmpeg 的前提下请求“NativeBuffer hardware output but no decoder Surface present”模式。若 mapper 同时导入该 buffer 到 EGL,decoder release 仍可能向另一 Surface present,导致两个 owner、未定义 fence/queue ownership 和 seek/recreate race。
结论:本阶段不提交伪 interop mapper,不调用 `eglFinish/glFinish` 充当同步,不宣称零拷贝/GPU interop/HDR 完成。`harmony-ohcodec-gl` 作为 disabled-by-default feature 保留 API 探针入口;启用它前必须先解决该 upstream contract blocker,并补齐 output buffer、fence、crop/stride 和 surface ownership 的真实设备验证。
## Direct Surface ownership
`ohos_direct_surface` 是 adapter 专用 native resource。只有 host 明确注入一个不等于 Playback Lab EGL window 的独占窗口时,`ohcodec-direct-surface-experimental` 才会取得引用并创建设备;当前 Playback Lab 不注入此 resource,所以 decoder 不会和 EGL/libmpv 抢同一 Surface。
Surface destroy/recreate 必须停止 render/decoder、等待所有 AVFrame release、递增 generation,再以新窗口重新创建 OHCodec device。当前只完成 MPV adapter 侧的 ownership guard,实体设备上的 fence、present count、surface recreate 仍为 `DEVICE TEST REQUIRED`。
-42
View File
@@ -1,42 +0,0 @@
# MPV OHCodec 集成说明
## 已实现路径
`hwdec=ohcodec` 由 MPV/FFmpeg 的动态 hardware-decoder 枚举进入独立
`AV_HWDEVICE_TYPE_OHCODEC` backend。`IMGFMT_OHCODEC` 与
`AV_PIX_FMT_OHCODEC` 使用单一映射。当前 adapter 只允许 H.264 和 HEVC,先
查询 API 24 OHCodec hardware capability,再检查 host 注入的独占
`ohos_direct_surface`,最后创建带 `OHNativeWindow*` 的 FFmpeg hwdevice。
Direct Surface backend 的名称是
`ohcodec-direct-surface-experimental`。它只负责证明 device/decoder/surface
plumbing,不与 Playback Lab 的 EGL window 共享呈现权;没有独占窗口时明确失败,
MPV 保留 software decoder fallback。
## 状态与回退
适配器记录 `HWDEC_DISABLED`、`HWDEC_REQUESTED`、`HWDEC_CAPABILITY_CHECK`、
`HWDEC_DEVICE_CREATED`、`HWDEC_DECODER_CREATED`、`HWDEC_DIRECT_SURFACE`、
`HWDEC_GPU_INTEROP`、`HWDEC_ACTIVE`、`HWDEC_FALLBACK`、`HWDEC_ERROR`。
Playback Lab 通过 libmpv 诊断 API 与属性显示 requested/actual decoder、active
状态、backend、capability、fallback reason 及 hardware/software frame 计数。
capability 查询失败、codec/profile/level 不支持、surface 不可用、device/decoder
创建失败或后续 interop 失败都必须返回失败并回到 software path;不把空 device、
opaque pointer 或 CPU readback 当成成功的硬件帧。
## Surface 生命周期约束
`ohos_direct_surface` 的引用由 adapter 持有,host 负责提供一个不等于
Playback Lab EGL window 的独占窗口。Surface 销毁或重建时,host 必须先停止
decoder/render、等待 AVFrame release,再销毁旧 device 并用新窗口重建;当前代
码只完成 ownership guard,generation/fence/present 的实体设备验证仍待做。
## GPU interop 边界
`harmony-ohcodec-gl` 默认关闭,仅编译 API probe:API24 可探测
`OH_AVBuffer_GetNativeBuffer`、NativeWindow wrapper、EGL native-buffer image
和 GLES external texture 扩展。但 FFmpeg upstream 的硬件分支是 surface-first,
没有“不 present 到 decoder Surface、直接导出可导入 NativeBuffer”的契约;见
`DOCS/OHCODEC_GPU_INTEROP_DESIGN.md` 中的 `OHCODEC-BLOCKER-001`。因此当前不
提交 mapper、不宣称 zero-copy,也不宣称 HDR。
-28
View File
@@ -1,28 +0,0 @@
# FFmpeg OHCodec upstream analysis
基线:FFmpeg `3acec0a1af2dda0a0838689b8b8649e7deb080a0`。本阶段没有修改 FFmpeg。
## 已核对的源码事实
| 位置 | 事实 | 对 MPV 的结论 |
|---|---|---|
| `libavutil/hwcontext.h` | 注册 `AV_HWDEVICE_TYPE_OHCODEC`。 | 设备类型可作为独立 MPV hwdec method,不能伪装成 MediaCodec/VAAPI。 |
| `libavutil/pixfmt.h` | 注册 `AV_PIX_FMT_OHCODEC`,并标记为 OpenHarmony hardware decoding。 | MPV 需要自己的 `IMGFMT_OHCODEC` 映射。 |
| `libavutil/hwcontext_oh.h` | `AVOHCodecDeviceContext` 只有 `void *native_window`,注释为 `OHNativeWindow`。 | FFmpeg 公共设备上下文只承载 Surface 指针,没有 device handle、fence 或 NativeBuffer export 字段。 |
| `libavutil/hwcontext_oh.c:27-45` | `oh_device_create()` 只拒绝非空 device string;初始化不创建窗口、不取得引用,也不做 capability 检查。 | MPV 不能把 `av_hwdevice_ctx_create(..., NULL, ...)` 当成可呈现设备;Surface ownership 必须在 MPV/host adapter 中完成。 |
| `libavcodec/ohdec.c:150-197` | OHCodec device 被发现后读取 `native_window`,设置 `output_to_window=true`,使用 `AV_PIXEL_FORMAT_SURFACE_FORMAT`,并调用 `OH_VideoDecoder_SetSurface(window)`。 | upstream 的硬件输出是 Direct Surface 语义。空窗口不是合法的 GPU interop 设备。 |
| `libavcodec/ohdec.c:219-289` | stream change 读取 width/height/stride/slice height/pixel format;Surface 输出时设置 `avctx->pix_fmt=AV_PIX_FMT_OHCODEC`,并复制颜色元数据与 SAR。 | 尺寸/stride/pixel format 可进入 MPV diagnostics;不能由 enum 推导设备能力或 HDR 支持。 |
| `libavcodec/ohdec.c:432-451` | AVFrame buffer release 时读取 buffer attr;非 discard 调 `OH_VideoDecoder_RenderOutputBuffer()`,否则 `FreeOutputBuffer()`;随后释放 decoder ref。 | AVFrame 生命周期会触发 Surface present/release。Surface 的 decoder ownership 必须持续到最后一个 frame 引用释放。 |
| `libavcodec/ohdec.c:453-486` | hardware frame 的 `format=AV_PIX_FMT_OHCODEC`;`buf[0]` 持有 `OH_AVBuffer*` 的自定义 release;`data[3]=buf[0]->data`。 | 当前 FFmpeg 没有 `AVHWFramesContext`、transfer/map/export 回调。`data[3]` 是 upstream 当前内部布局,不应被重解释成 fd/Android hardware buffer。 |
| `libavcodec/ohdec.c:489-532` | 非 Surface 分支才取得 CPU 地址并复制到 software frame。 | 当前 upstream 没有“NativeBuffer output、但不直接 present Surface”的公共 decoder 配置路径。 |
| `libavcodec/ohdec.c` decoder close/flush | decoder 可能因硬件 frame 引用延迟释放;flush 会清空队列并重启 decoder。 | seek/flush 必须让 MPV 丢弃旧 generation 的 frame;Surface recreate 需要完整 decoder recreate。 |
## 关键问题结论
1. `AV_PIX_FMT_OHCODEC` frame 确实携带 `OH_AVBuffer*` 指针(`data[3]`),但 upstream 没有把它作为稳定的 MPV interop contract 暴露,也没有 non-surface output 模式。
2. API24 SDK 的 `OH_AVBuffer_GetNativeBuffer()` 可以合法取得 `OH_NativeBuffer*`,但要导入 EGL 还需要转换为 `OHNativeWindowBuffer*`;FFmpeg 当前仍会在 frame release 时调用 `RenderOutputBuffer`,因此不能安全地同时让 decoder Surface 和 libmpv EGL surface 认为自己拥有呈现权。
3. 因此本阶段在 MPV adapter 层实现了独立的 `ohcodec-direct-surface-experimental` 探针;GPU interop 保留 feature gate 与设计文档,明确记录 upstream blocker,不修改 FFmpeg。
## 变更边界
本仓库只新增 MPV image-format mapping、OHCodec hwdevice/Surface adapter、capability/fallback diagnostics 和 Playback Lab diagnostics。没有复制或修改 FFmpeg 源码。
+160
View File
@@ -0,0 +1,160 @@
1. Input layer, supported devices, methods:
- plain file, with seeking
- STDIN, without seeking backward
- network streaming (currently plain wget-like HTTP and MMS (.asx))
- VCD (Video CD) track, by direct CDROM device access (not requires mounting disc)
- DVD titles using .IFO structure, by direct DVD device access (not requires mounting disc)
- DVD titles using menu navigation (experimental/alpha, not yet finished!!!)
- CDDA - raw audio from audio CD-ROM discs (using cdparanoia libs)
- RTP streaming (mpeg-ps over multicast only)
- LIVE555 streaming - support SDP/RTSP (using the LIVE555 libs)
- SMB - file access over samba (experimental)
2. Demuxer/parser layer, supported file/media formats:
- MPEG streams (ES,PES,PS. no TS support yet)
note: mpeg demuxer silently ignore non-mpeg content, and find mpeg packets
in arbitrary streams. it means you can play directly VCD images (for example
CDRwin's .BIN files) without extracting mpeg files first (with tools like vcdgear)
It accepts all PES variants, including files created by VDR.
Note: VOB (video object) is simple mpeg stream, but it usually has 01BD
packets which may contain subtitles and non-mpeg audio. Usually found on DVD discs.
Headers: mpeg streams has no global header. each frame sequence (also called GOP,
group of pictures) contains an sequence header, it describes that block.
In normal mpeg 1/2 content there are groups of 12-15 frames (24/30 fps).
It means you can freely seek in mpeg streams, and even can cut it to
small parts with standard file tools (dd, cut) without destroying it.
Codecs: video is always mpeg video (mpeg1, mpeg2 or mpeg4).
audio is usually mpeg audio (any layer allowed, but it's layer 2 in most files)
but 01BD packets may contain AC3, DTS or LPCM too.
FPS: mpeg2 content allows variable framerate, in form of delayed frames.
It's mostly used for playback 24fps content at 29.97/30 fps (NTSC) rate.
(so called Telecine or 3:2 pulldown effect)
It means you see 30 frames per second, but there are only 24 different
pictures and some of them are shown longer to fill 30 frame time.
If you encode such files with mencoder, using -ofps 24 or -ofps 23.976
is recommended.
- AVI streams.
Two kind of RIFF AVI files exists:
1. interleaved: audio and video content is interleaved. it's faster and
requires only 1 reading thread, so it's recommended (and mostly used).
2. non-interleaved: audio and video aren't interleaved, i mean first come
whole video followed by whole audio. it requires 2 reading process or
1 reading with lots of seeking. very bad for network or cdrom.
3. badly interleaved streams: mplayer detects interleaving at startup and
enables -ni option if it finds non-interleaved content. but sometimes
the stream seems to be interleaved, but with bad sync so it should be
played as non-interleaved otherwise you get a-v desync or buffer overflow.
MPlayer supports 2 kind of timing for AVI files:
- bps-based: it is based on bitrate/samplerate of video/audio stream.
this method is used by most players, including avifile and wmp.
files with broken headers, and files created with VBR audio but not
vbr-compliant encoder will result a-v desync with this method.
(mostly at seeking).
- interleaving-based: note: it can't be used togethwer with -ni
it doesn't use bitrate stuff of header, it uses the relative position
of interleaved audio and video chunks. makes some badly encoded file
with vbr audio playable.
Headers: AVI files has a mandatory header at the begin of the file,
describing video parameters (resolution, fps) and codecs. Optionally
they have an INDEX block at the end of the file. It's optional, but
most files has such block, because it's REQUIRED for seeking.
Btw usually it can be rebuilt from file content, mplayer does it with
the -idx switch. MPlayer can recreate broken index blocks using -forceidx.
As AVI files needs index for random access, broken files with no index
are usually unplayable.
Of course, cutting/joining AVI files needs special programs.
Codecs: any audio and video codecs allowed, but I note that VBR audio is
not well supported by most players. The file format makes it possible to
use VBR audio, but most players expect CBR audio and fails with VBR,
as VBR is unusual, and Microsoft's AVI specs only describe CBR audio.
I also note, that most AVI encoders/multiplexers create bad files if
using VBR audio. only 2 exception (known by me): NaNDub and MEncoder.
FPS: only constant framerate allowed, but it's possible to skip frames.
- ASF streams:
ASF (active streaming format) comes from Microsoft. they developed two
variant of ASF, v1.0 and v2.0. v1.0 is used by their media tools (wmp and
wme) and v2.0 is published and patented :). of course, they differ,
no compatibility at all. (it's just a legality game)
MPlayer supports only v1.0, as nobody ever seen v2.0 files :)
Note, that .ASF files are nowdays come with extension .WMA or .WMV.
UPDATE: MS recently released the ASF v1.0 specs too, but it has some
restrictions making it illegal to read by us :)
Headers: Stream headers (codecs parameters) can be everywhere (in theory),
but all files i've seen had it at the beginning of the file.
Asf uses fixed packet size, so it is seekable without any INDEX block,
and broken files are playable well.
Codecs: video is mostly microsoft's mpeg4 variants: MP42, MP43 (aka DivX),
WMV1 and WMV2. but any codecs allowed.
audio is usually wma or voxware, sometimes mp3, but any codecs allowed.
FPS: no fixed fps, every video frame has an exact timestamp instead.
I've got stream with up to 3 sec frame display times.
- QuickTime / MOV files:
They come from Mac users, usually with .mov or .qt extension, but as
MPEG Group chose quicktime as recommended file format for MPEG4,
sometimes you meet quicktime files with .mpg or .mp4 extension.
At first look, it's a mixture of ASF and AVI.
It requires INDEX block for random access and seeking, and even for
playback, like AVI, but uses timestamps instead of constant framerate
and has more flexible stream options (including network stuff) like ASF.
Headers: header can be placed at the beginning or at the end of file.
About half of my files have it at the beginning, others have it at the end.
Broken files are only playable if they have header at the beginning!
Codecs: any codecs allowed, both CBR and VBR.
Note: most new mov files use Sorenson video and QDesign Music audio,
they are patented, closed, secret, (TM)-ed etc formats, only Apple's
quicktime player is able to playback these files (on win/mac only).
- VIVO files:
They are funny streams. They have a human-readable ascii header at
the beginning, followed by interleaved audio and video chunks.
It has no index block, has no fixed packetsize or sync bytes, and most
files even has no keyframes, so forget seeking!
Video is standard h.263 (in vivo/2.0 files it's modified, non-standard
h.263), audio is either standard g.723 or Vivo Siren codec.
Note, that microsoft licensed vivo stuff, and included in their netshow
v2.0 program, so there are VfW/ACM codecs for vivo video and audio.
- RealMedia files:
A mixture of AVI and ASF features. It has mandatory headers at the
beginning and an optional INDEX (missing in most files).
The file is constructed of variable size chunks, with small header
telling the stream ID, timestamp, flags (keyframe...) and size.
But it has some features found in ASF files:
The video is actually double-muxed, the video chunks are really
appended fragments of the video frame. RV30+ supports B frames, so
you have to parse some bits of the first fragment to get the real PTS.
The audio frames are fixed size (CBR) but using the same scrambling
(out-of-order interleaving) as in the ASF files.
Codecs: Audio is either COOK(er), SIPR(o), ATRAC3 or DNET.
The DNET is actually a byte-swapped low-bitrate Dolby AC3 variant :)
Video is RV10 (h263 variant), RV20 (rp G2), RV30 (rp v8) or RV40 (rp v9).
FPS: variable, just like in ASF.
Note, that similarity of real and asf has some background - they worked
together on the (never finished/used) ASF v2 spec for some time.
- GIF files:
The GIF format is a common format for web graphics that supports
animation. These are read through libungif or compatible library.
Variable frame delays are supported, but seeking is not supported.
Seeking will be supported once an index of gif frames can be built.
+228
View File
@@ -0,0 +1,228 @@
So, I'll describe how this stuff works.
The main modules:
1. stream.c: this is the input layer, this reads the input media (file, stdin,
vcd, dvd, network etc). what it has to know: appropriate buffering by
sector, seek, skip functions, reading by bytes, or blocks with any size.
The stream_t (stream.h) structure describes the input stream, file/device.
There is a stream cache layer (cache2.c), it's a wrapper for the stream
API. It does fork(), then emulates stream driver in the parent process,
and stream user in the child process, while proxying between them using
preallocated big memory chunk for FIFO buffer.
2. demuxer.c: this does the demultiplexing (separating) of the input to
audio, video or dvdsub channels, and their reading by buffered packages.
The demuxer.c is basically a framework, which is the same for all the
input formats, and there are parsers for each of them (mpeg-es,
mpeg-ps, avi, avi-ni, asf), these are in the demux_*.c files.
The structure is the demuxer_t. There is only one demuxer.
2.a. demux_packet_t, that is DP.
Contains one chunk (avi) or packet (asf,mpg). They are stored in memory as
in linked list, cause of their different size.
2.b. demuxer stream, that is DS.
Struct: demux_stream_t
Every channel (a/v/s) has one. This contains the packets for the stream
(see 2.a). For now, there can be 3 for each demuxer :
- audio (d_audio)
- video (d_video)
- DVD subtitle (d_dvdsub)
2.c. stream header. There are 2 types (for now): sh_audio_t and sh_video_t
This contains every parameter essential for decoding, such as input/output
buffers, chosen codec, fps, etc. There are each for every stream in
the file. At least one for video, if sound is present then another,
but if there are more, then there'll be one structure for each.
These are filled according to the header (avi/asf), or demux_mpg.c
does it (mpg) if it founds a new stream. If a new stream is found,
the ====> Found audio/video stream: <id> messages is displayed.
The chosen stream header and its demuxer are connected together
(ds->sh and sh->ds) to simplify the usage. So it's enough to pass the
ds or the sh, depending on the function.
For example: we have an asf file, 6 streams inside it, 1 audio, 5
video. During the reading of the header, 6 sh structs are created, 1
audio and 5 video. When it starts reading the packet, it chooses the
stream for the first found audio & video packet, and sets the sh
pointers of d_audio and d_video according to them. So later it reads
only these streams. Of course the user can force choosing a specific
stream with
-vid and -aid switches.
A good example for this is the DVD, where the english stream is not
always the first, so every VOB has different language :)
That's when we have to use for example the -aid 128 switch.
Now, how this reading works?
- demuxer.c/demux_read_data() is called, it gets how many bytes,
and where (memory address), would we like to read, and from which
DS. The codecs call this.
- this checks if the given DS's buffer contains something, if so, it
reads from there as much as needed. If there isn't enough, it calls
ds_fill_buffer(), which:
- checks if the given DS has buffered packages (DP's), if so, it moves
the oldest to the buffer, and reads on. If the list is empty, it
calls demux_fill_buffer() :
- this calls the parser for the input format, which reads the file
onward, and moves the found packages to their buffers.
Well it we'd like an audio package, but only a bunch of video
packages are available, then sooner or later the:
DEMUXER: Too many (%d in %d bytes) audio packets in the buffer
error shows up.
2.d. video.c: this file/function handle the reading and assembling of the
video frames. each call to video_read_frame() should read and return a
single video frame, and it's duration in seconds (float).
The implementation is splitted to 2 big parts - reading from mpeg-like
streams and reading from one-frame-per-chunk files (avi, asf, mov).
Then it calculates duration, either from fixed FPS value, or from the
PTS difference between and after reading the frame.
2.e. other utility functions: there are some useful code there, like
AVI muxer, or mp3 header parser, but leave them for now.
So everything is ok 'till now. It can be found in libmpdemux/ library.
It should compile outside of mplayer tree, you just have to implement few
simple functions, like mp_msg() to print messages, etc.
See libmpdemux/test.c for example.
See also formats.txt, for description of common media file formats and their
implementation details in libmpdemux.
Now, go on:
3. mplayer.c - ooh, he's the boss :)
Its main purpose is connecting the other modules, and maintaining A/V
sync.
The given stream's actual position is in the 'timer' field of the
corresponding stream header (sh_audio / sh_video).
The structure of the playing loop :
while(not EOF) {
fill audio buffer (read & decode audio) + increase a_frame
read & decode a single video frame + increase v_frame
sleep (wait until a_frame>=v_frame)
display the frame
apply A-V PTS correction to a_frame
handle events (keys,lirc etc) -> pause,seek,...
}
When playing (a/v), it increases the variables by the duration of the
played a/v.
- with audio this is played bytes / sh_audio->o_bps
Note: i_bps = number of compressed bytes for one second of audio
o_bps = number of uncompressed bytes for one second of audio
(this is = bps*samplerate*channels)
- with video this is usually == 1.0/fps, but I have to note that
fps doesn't really matters at video, for example asf doesn't have that,
instead there is "duration" and it can change per frame.
MPEG2 has "repeat_count" which delays the frame by 1-2.5 ...
Maybe only AVI and MPEG1 has fixed fps.
So everything works right until the audio and video are in perfect
synchronity, since the audio goes, it gives the timing, and if the
time of a frame passed, the next frame is displayed.
But what if these two aren't synchronized in the input file?
PTS correction kicks in. The input demuxers read the PTS (presentation
timestamp) of the packages, and with it we can see if the streams
are synchronized. Then MPlayer can correct the a_frame, within
a given maximal bounder (see -mc option). The summary of the
corrections can be found in c_total .
Of course this is not everything, several things suck.
For example the soundcards delay, which has to be corrected by
MPlayer! The audio delay is the sum of all these:
- bytes read since the last timestamp:
t1 = d_audio->pts_bytes/sh_audio->i_bps
- if Win32/ACM then the bytes stored in audio input buffer
t2 = a_in_buffer_len/sh_audio->i_bps
- uncompressed bytes in audio out buffer
t3 = a_buffer_len/sh_audio->o_bps
- not yet played bytes stored in the soundcard's (or DMA's) buffer
t4 = get_audio_delay()/sh_audio->o_bps
From this we can calculate what PTS we need for the just played
audio, then after we compare this with the video's PTS, we have
the difference!
Life didn't get simpler with AVI. There's the "official" timing
method, the BPS-based, so the header contains how many compressed
audio bytes or chunks belong to one second of frames.
In the AVI stream header there are 2 important fields, the
dwSampleSize, and dwRate/dwScale pairs:
- If the dwSampleSize is 0, then it's VBR stream, so its bitrate
isn't constant. It means that 1 chunk stores 1 sample, and
dwRate/dwScale gives the chunks/sec value.
- If the dwSampleSize is >0, then it's constant bitrate, and the
time can be measured this way: time = (bytepos/dwSampleSize) /
(dwRate/dwScale) (so the sample's number is divided with the
samplerate). Now the audio can be handled as a stream, which can
be cut to chunks, but can be one chunk also.
The other method can be used only for interleaved files: from
the order of the chunks, a timestamp (PTS) value can be calculated.
The PTS of the video chunks are simple: chunk number * fps
The audio is the same as the previous video chunk was.
We have to pay attention to the so called "audio preload", that is,
there is a delay between the audio and video streams. This is
usually 0.5-1.0 sec, but can be totally different.
The exact value was measured until now, but now the demux_avi.c
handles it: at the audio chunk after the first video, it calculates
the A/V difference, and take this as a measure for audio preload.
3.a. audio playback:
Some words on audio playback:
Not the playing is hard, but:
1. knowing when to write into the buffer, without blocking
2. knowing how much was played of what we wrote into
The first is needed for audio decoding, and to keep the buffer
full (so the audio will never skip). And the second is needed for
correct timing, because some soundcards delay even 3-7 seconds,
which can't be forgotten about.
To solve this, the OSS gives several possibilities:
- ioctl(SNDCTL_DSP_GETODELAY): tells how many unplayed bytes are in
the soundcard's buffer -> perfect for timing, but not all drivers
support it :(
- ioctl(SNDCTL_DSP_GETOSPACE): tells how much can we write into the
soundcard's buffer, without blocking. If the driver doesn't
support GETODELAY, we can use this to know how much the delay is.
- select(): should tell if we can write into the buffer without
blocking. Unfortunately it doesn't say how much we could :((
Also, doesn't/badly works with some drivers.
Only used if none of the above works.
4. Codecs. Consists of libmpcodecs/* and separate files or libs,
for example libmpeg2, loader, mp3lib.
mplayer.c doesn't call them directly, but through the dec_audio.c and
dec_video.c files, so the mplayer.c doesn't have to know anything about
the codecs.
libmpcodecs contains wrapper for every codecs, some of them include the
codec function implementation, some calls functions from other files
included with mplayer, some calls optional external libraries.
file naming convention in libmpcodecs:
ad_*.c - audio decoder (called through dec_audio.c)
vd_*.c - video decoder (called through dec_video.c)
ve_*.c - video encoder (used by mencoder)
vf_*.c - video filter (see option -vf)
On this topic, see also:
libmpcodecs.txt - The structure of the codec-filter path, with explanation
dr-methods.txt - Direct rendering, MPI buffer management for video codecs
codecs.conf.txt - How to write/edit codec configuration file (codecs.conf)
codec-devel.txt - Mike's hints about codec development - a bit OUTDATED
hwac3.txt - about SP/DIF audio passthrough
5. libvo: this displays the frame.
for details on this, read libvo.txt
6. libao2: this control audio playing
6.a audio plugins
for details on this, read libao2.txt
+145
View File
@@ -0,0 +1,145 @@
mails by A'rpi and Marcus Blomenkamp <Marcus.Blomenkamp@epost.de>
describing how this ac3-passtrough hack work under linux and mplayer...
-----------------------------------------------------------------------
Hi,
> I received the following patch from Steven Brookes <stevenjb@mda.co.uk>.
> He is working on fixing the digital audio output of the dxr3 driver and
> told me he fixed some bugs in mplayer along the way. I don't know shit
> about hwac3 output so all I did was to make sure the patch applied
> against latest cvs.
> This is from his e-mail to me:
>
> "Secondly there is a patch to dec_audio.c and
> ac3-iec958 to fix the -ac hwac3 codec stuff and to use liba52 to sync it.
> Seems to work for everything I've thrown at and maintains sync for all audio
> types through the DXR3."
patch applied (with some comments added and an unwanted change (in software
a52 decoder) removed)
now i understand how this whole hwac3 mess work.
it's very very tricky. it virtually decodes ac3 to LPCM packets, but really
it keeps the original compressed data padded by zeros. this way it's
constant bitrate, and sync is calculated just like for stereo PCM.
(so it bypass LPCM-capable media converters...)
so, every ac3 frame is translated to 6144 byte long tricky LPCM packet.
6144 = 4*(6*256) = 4 * samples_per_ac3_frame = LPCM size of uncompressed ac3
frame.
i wanna know if it works for sblive and other ac3-capable cards too?
(i can't test it, lack of ac3 decoder)
A'rpi / Astral & ESP-team
-----------------------------------------------------------------------
Hi folks.
I spend some time fiddling with ac3 passthrough in mplayer. The
traditional way of setting the output format to AFMT_AC3 was no ideal
solution since not all digital io cards/drivers supported this format or
honoured it to set the spdif non-audio bit. To make it short, it only
worked with oss sblive driver IIRC.
Inspired by alsa's ac3dec program I found an alternative way by
inspecting to which format the alsa device had been set. Suprise: it was
simple 16bit_le 2_channel pcm. So setting the non-audio bit doesn't
necessarily mean the point. The only important thing seems to be
bit-identical output at the correct samplerate. Modern AV-Receivers seem
to be quite tolerant/compatible.
So I changed the output format of hwac3 from
AFMT_AC3 channels=1
to
AFMT_S16_LE channels=2
and corrected the absolute time calculation. That was all to get it
running for me.
-----------------------------------------------------------------------
Hi there.
Perhaps I can clear up some mystification about AC3 passthrough in
general and mplayer in special:
To get the external decoder solution working, it must be fed with data
which is bitidentical to the chunks in the source ac3 file (compressed
data is very picky about bit errors). Additionally - or better to say
'historically' - the non-audio bit should be set in the spdif status
fields to prevent old spdif hardware from reproducing ugly scratchy
noise. Note: for current decoders (probably those with DTS capability)
this safety bit isn't needed anymore. At least I can state that for my
Sherwood RVD-6095RDS. I think it is due to DTS because DTS sound can
reside on a ordinary AudioCD and an ordinary AudioCD-Player will always
have it's audio-bit set.
The sample format of the data must be 2channel 16bit (little endian
IIRC). Samplerates are 48kHz - although my receiver also accepts
44100Hz. I do not know if this is due to an over-compatability of my
receiver or if 44100 is also possible in the ac3 specs. For safety's
sake lets keep this at 48000Hz. AC3 data chunks are inserted into the
stream every 0x1600 bytes (don't bite me on that, look into
'ac3-iec958.c': 'ac3_iec958_build_burst').
To come back to the problem: data must be played bit-identically through
the soundcard at the correct samplerate and should optionally have it's
non-audio bit set. There are two ways to accomplish this:
1) Some OSS guy invented the format AFMT_AC3. Soundcard drivers
implementing this format should therefore adjust it's mixers and
switches to produce the desired output. Unfortunately some soundcard
drivers do not support this format correctly and most do not even
support it at all (including ALSA).
2) The alternative approach currently in mplayer CVS is to simply set
the output format to 48kHz16bitLE and rely on the user to have the
soundcard mixers adjusted properly.
I do have two soundcards with digital IO facilities (CMI8738 and
Trident4DWaveNX based) plus the mentioned decoder. I'm currently running
Linux-2.4.17. Following configurations are happily running here:
1. Trident with ALSA drivers (OSS does not support Hoontech's dig. IO)
2. CMI with ALSA drivers
3. CMI with OSS drivers
For Linux I'd suggest using ALSA because of it's cleaner architecture
and more consitent user interface. Not to mention that it'll be the
standard sound support in Linux soon.
For those who want to stick to OSS drivers: The CMI8738 drivers works
out-of-the-box, if the PCM/Wave mixer is set to 100%.
For ALSA I'd suggest using its OSS emulation. More on that later.
ALSA-0.9 invented the idea of cards, devices and dubdevices. You can
reach the digital interface of all supported cards consitently by using
the device 'hw:x,2' (x counting from 0 is the number of your soundcard).
So most people would end up at 'hw:0,2'. This device can only be opened
in sample formats and rates which are directly supported in hardware
hence no samplerate conversion is done keeping the stream as-is. However
most consumer soundcards do not support 44kHz so it would definitively
be a bad idea to use this as your standard device if you wanted to
listen to some mp3s (most of them are 44kHz due to CD source). Here the
OSS comes to play again. You can configure which OSS device (/dev/dsp
and /dev/adsp) uses which ALSA device. So I'd suggest pointing the
standard '/dev/dsp' to standard 'hw:0,0' which suports mixing and
samplerate conversion. No further reconfiguration would be needed for
your sound apps. For movies I'd point '/dev/adsp' to 'hw:0,2' and
configure mplayer to use adsp instead of dsp. The samplerate constrain
is no big deal here since movies usually are in 48Khz anyway. The
configuration in '/etc/modules.conf' is no big deal also:
alias snd-card-0 snd-card-cmipci # insert your card here
alias snd-card-1 snd-pcm-oss # load OSS emulation
options snd-pcm-oss snd_dsp_map=0 snd_adsp_map=2 # do the mapping
This works flawlessly in combination with alsa's native
SysVrc-init-script 'alsasound'. Be sure to disable any distribution
dependent script (e.g. Mandrake-8.1 has an 'alsa' script which depends
on ALSA-0.5).
Sorry for you *BSD'lers out there. I have no grasp on sound support there.
HTH Marcus
+56
View File
@@ -0,0 +1,56 @@
6. libao2: this control audio playing
As in libvo (see 5.) also here are some drivers, based on the same API:
static int control(int cmd, int arg);
This is for reading/setting driver-specific and other special parameters.
Not really used for now.
static int init(int rate,int channels,int format,int flags);
The init of driver, opens device, sets sample rate, channels, sample format
parameters.
Sample format: usually AFMT_S16_LE or AFMT_U8, for more definitions see
dec_audio.c and linux/soundcards.h files!
static void uninit(void);
Guess what.
Ok I help: closes the device, not (yet) called when exit.
static void reset(void);
Resets device. To be exact, it's for deleting buffers' contents,
so after reset() the previously received stuff won't be output.
(called if pause or seek)
static int get_space(void);
Returns how many bytes can be written into the audio buffer without
blocking (making caller process wait). MPlayer occasionally checks the
remaining space and tries to fill the buffer with play() if there's free
space. The buffer size used should be sane; a buffer that is too small
could run empty before MPlayer tries filling it again (normally once per
video frame), a buffer that is too big would force MPlayer decode the file
far ahead trying to find enough audio data to fill it.
static int play(void* data,int len,int flags);
Plays a bit of audio, which is received throught the "data" memory area, with
a size of "len". It has to copy the data, because they can be overwritten
after the call is made. Doesn't have to use all the bytes; it has to
return the number of bytes used used (copied to buffer). If
flags|AOPLAY_FINAL_CHUNK is true then this is the last audio in the file.
The purpose of this flag is to tell aos that round down the audio played
from "len" to a multiple of some chunksize that this "len" should not be
rounded down to 0 or the data will never be played (as MPlayer will never
call play() with a larger len).
static float get_delay(void);
Returns how long time it will take to play the data currently in the
output buffer. Be exact, if possible, since the whole timing depends
on this! In the worst case, return the maximum delay.
!!! Because the video is synchronized to the audio (card), it's very important
!!! that the get_delay function is correctly implemented!
static void audio_pause(void);
Pause playing but do not delete buffered data if possible.
static void audio_resume(void);
Continue playing after audio_pause().
+173
View File
@@ -0,0 +1,173 @@
libvo --- the library to handle video output by A'rpi, 2002.04
============================================
Note: before start on this, read colorspaces.txt !
The constants for different pixelformats are defined in img_format.h,
their usage is mandatory.
WARNING: Please keep in mind that some of this information may be out-dated,
so if you are working on a new vo, consider submitting preliminary patches
very early on. Currently vo_gl is one of the more up-to-date VOs to use
as reference if you are unsure about something and do not want to ask on the
list.
vo_vdpau and vo_direct3d may be a good choice, too, they use different
approaches closer to the sometimes convoluted way DirectX works.
Each vo driver _has_ to implement these:
preinit():
init the video system (to support querying for supported formats)
uninit():
Uninit the whole system, this is on the same "level" as preinit.
control():
Current controls (VOCTRL_QUERY_FORMAT must be implemented,
VOCTRL_DRAW_IMAGE, VOCTRL_FULLSCREEN, VOCTRL_UPDATE_SCREENINFO
should be implemented):
VOCTRL_QUERY_FORMAT - queries if a given pixelformat is supported.
It also returns various flags decsirbing the capabilities
of the driver with teh given mode. for the flags, see
file vfcaps.h !
the most important flags, every driver must properly report
these:
0x1 - supported (with or without conversion)
0x2 - supported without conversion (define 0x1 too!)
0x100 - driver/hardware handles timing (blocking)
also SET sw/hw scaling and osd support flags, and flip,
and accept_stride if you implement VOCTRL_DRAW_IMAGE (see bellow)
NOTE: VOCTRL_QUERY_FORMAT may be called _before_ first config()
but is always called between preinit() and uninit()
VOCTRL_GET_IMAGE
libmpcodecs Direct Rendering interface
You need to update mpi (mp_image.h) structure, for example,
look at vo_x11, vo_sdl, vo_xv or mga_common.
VOCTRL_DRAW_IMAGE
replacement for the current draw_slice/draw_frame way of
passing video frames. by implementing SET_IMAGE, you'll get
image in mp_image struct instead of by calling draw_*.
unless you return VO_TRUE for VOCTRL_DRAW_IMAGE call, the
old-style draw_* functils will be called!
Note: draw_slice is still mandatory, for per-slice rendering!
VOCTRL_RESET - reset the video device
This is sent on seeking and similar and is useful if you are
using a device which prebuffers frames that need to flush them
before refilling audio/video buffers.
VOCTRL_PAUSE
VOCTRL_RESUME
VOCTRL_GUISUPPORT
return true only if driver supports co-operation with
MPlayer's GUI (not yet used by GUI)
VOCTRL_SET_EQUALIZER
set the video equalizer to the given values
two arguments are provided: item and value
item is a string, the possible values are (currently):
brightness, contrast, saturation, hue
VOCTRL_GET_EQUALIZER
get the current video equalizer values
two arguments are provided: item and value
item is a string, the possible values are (currently):
brightness, contrast, saturation, hue
VOCTRL_ONTOP
Makes the player window stay-on-top. Only supported (currently)
by drivers which use X11, except SDL, as well as directx and
gl2 under Windows.
VOCTRL_BORDER
Makes the player window borderless.
VOCTRL_FULLSCREEN
Switch from and to fullscreen mode
VOCTRL_GET_PANSCAN
VOCTRL_SET_PANSCAN
Needed to implement pan-scan support ('w' and 'e' keys during
playback in fullscreen mode)
VOCTRL_START_SLICE
Called before the first draw_slice of each frame, useful if
you need to do some set-up work.
VOCTRL_DRAW_EOSD
Required for EOSD (ASS subtitle) support. Provides all
information necessary to draw the EOSD for the current video
frame.
VOCTRL_GET_EOSD_RES
Required for EOSD (ASS subtitle) support. Informs the ASS
renderer about the properties of the drawing area (size,
borders).
VOCTRL_SET_DEINTERLACE
VOCTRL_GET_DEINTERLACE
Get or set deinterlacing status for VOs that support some kind
of deinterlacing.
VOCTRL_UPDATE_SCREENINFO
Should set the xinerama_x, xinerama_y, vo_screenwidth and
vo_screenheight appropriately for the currently used
monitor and -xineramascreen option.
Usually should simply call the w32_update_xinerama_info or
update_xinerama_info function.
By supporting this, the VO also requests the newer API
that sets vo_dx, vo_dy etc. appropriately before config()
is called.
config():
Set up the video system. You get the dimensions and flags.
width, height: size of the source image
d_width, d_height: wanted scaled/display size (it's a hint)
Flags:
0x01 - force fullscreen (-fs)
0x02 - allow mode switching (-vm)
0x04 - allow software scaling (-zoom)
0x08 - flipping (-flip)
They're defined as VOFLAG_* (see libvo/video_out.h)
IMPORTANT NOTE: config() may be called 0 (zero), 1 or more (2,3...)
times between preinit() and uninit() calls. You MUST handle it, and
you shouldn't crash at second config() call or at uninit() without
any config() call! To make your life easier, vo_config_count is
set to the number of previous config() call, counted from preinit().
It's set by the caller (vf_vo.c), you don't have to increase it!
So, you can check for vo_config_count>0 in uninit() when freeing
resources allocated in config() to avoid crash!
You should implement VOCTRL_UPDATE_SCREENINFO so that vo_dx, vo_dy,
vo_dwidth and vo_dheight are already pre-set to values that take
aspect and -geometry into account. It is also necessary to properly
support multi-monitor setups (if based on x11_common, w32_common).
draw_slice(): this displays YV12 pictures (3 planes, one full sized that
contains brightness (Y), and 2 quarter-sized which the colour-info
(U,V). MPEG codecs (libmpeg2, opendivx) use this. This doesn't have
to display the whole frame, only update small parts of it.
If this is not supported, it must be signaled in QUERY_FORMAT with
VOCAP_NOSLICES.
draw_frame(): this is the older interface, this displays only complete
frames, and can do only packed format (YUY2, RGB/BGR).
Win32 codecs use this (DivX, Indeo, etc).
If you implement VOCTRL_DRAW_IMAGE, you do not need to implement draw_frame.
draw_osd(): this displays subtitles and OSD.
It's a bit tricky to use it, since it's a callback-style stuff.
It should call vo_draw_text() with screen dimension and your
draw_alpha implementation for the pixelformat (function pointer).
The vo_draw_text() checks the characters to draw, and calls
draw_alpha() for each. As a help, osd.c contains draw_alpha for
each pixelformats, use this if possible!
Note that if you do not draw directly onto the video you should
use vo_draw_text_ext() which allows you to specify the border
values etc. needed to draw DVD menu highlights at the correct place.
If you do not want to implement this, you can still use -vf
expand=osd=1 to draw the OSD, or even implement code to insert
this filter automatically.
Make sure you set VFCAP_OSD depending on whether you implemented it
or not.
NOTE: This one will be obsolete soon! But it's still useful when
you want to do tricks, like rendering osd _after_ hardware scaling
(tdfxfb) or render subtitles under of the image (vo_mpegpes, sdl)
NOTE2: above NOTE is probably wrong, there are currently no plans to
obsolete draw_osd, though there is the more advanced EOSD support for
ASS subtitles.
flip_page(): this is called after each frame, this displays the buffer for
real. This is 'swapbuffers' when doublebuffering.
Try to do as little work here as possible, since that affect jitter/
A-V sync.
+100
View File
@@ -0,0 +1,100 @@
TITLE=Weird Al Yankovic - The Saga Begins (clip)
# just to be informative (maybe future
# use?)
FILE=14233792,7972c637f619e01854ecb453b7339e3f # size,md5
AUTHOR=Gabucino
TYPE=VIDEO # maybe AUDIO in future (for displaying
# mp3 lyrics while playing or whatever)
FORMAT=TIME # this file contains measures based on
# TIME, not frames.
# if FORMAT is a number, then it's frame based, i.e.:
# FORMAT=25
NOTE=MPlayerSub test file v0.1
# first number : wait this much after previous subtitle disappeared
# second number : display the current subtitle for this many seconds
15 3
A long, long time ago...
0 3
in a galaxy far away...
0 3
Naboo was under an attack.
1 2.5
And I thought me and
Qui-Gon Jinn could
0 2.5
talk the Federation into
0 4
...maybe cutting them a
little slack.
2 3
But their response, it
didn't thrill us,
0 3
They locked the doors,
and tried to kill us.
0 2.5
We escaped from that gas,
0 3.5
then met Jar-jar and
Boss-Nass.
1 6
We took a bongo from the
scene and we went to
Theed to see the Queen.
0 5
We all wound' up on
Tatooine.
0 6
That's where, we've found
this boy.
0 4
Oh my, my this here
Anakin guy,
0 5
maybe Vader someday
later now he's just
a small fry.
0 4
And he left his home and
kissed his mommy goodbye,
0 5
singing "Soon I'm gonna be
a Jedi!"
6 6
Did you know this junkyard
slave isn't even old enough
to shave,
0 3
but he can use the Force,
they say.
1 6
Ahh, do you see him hitting
on the queen though he's
just nine and she's fourteen
0 6
yeah, he's probably gonna
marry her, someday!
+19
View File
@@ -0,0 +1,19 @@
software scaler filter docs
-ssf cgb=<float> chrominance gaussian blur
useful for videos with some chroma (color) noise (TV captures, VHS, ...)
-ssf lgb=<float> -> luminance gaussian blur
hmm, mostly special effect, could be used as lame noise reducer though
-ssf chs=<int> -> shift chrominance horizontally
-ssf cvs=<int> -> shift chrominance vertically
useful if the luma & chroma are not aligned exactly
-ssf cs=<float> -> chrominance sharpen
-ssf ls=<float> -> luminance sharpen
simple sharpening filter
note: these will only work if the swscaler is used at all (use -vf scale)
example: mplayer foobar.avi -vf scale -ssf lgb=3.0
+68
View File
@@ -0,0 +1,68 @@
artifact types:
---------------------------------------------------------------------
ringing
wave or noise like patterns around sharp edges
bad: sinc, lanczos (with high filter length)
acceptable: lanczos (with low filter length), cubic, spline
ok: area, (fast)blinear, gauss, point
blur
loss of detail / high frequency
bad: gauss with high variance
acceptable: (fast)bilinear, area
ok: others
aliasing (only downscale)
straight lines look like staircases
areas of high detail look worse
regular patterns turn into a funny looking mess (moire patterns)
bad: fast bilinear, point, gauss with low variance
acceptable: area
ok: others
blocky (upscale only)
looks like the image is made of rectangular blocks like a mosaic
bad: point
ok: others
recommendations: (based upon my personal opinion many ppl might disagree ...)
---------------------------------------------------------------------
the recommended scalers for upscaling:
fast_bilinear, point if speed is important
cubic, spline, lanczos if quality is important
the recommended scalers for downscaling:
fast_bilinear, point if speed is important
gauss, bilinear if quality is important
cubic, spline, lanczos if a sharper picture is important
note: when encoding at a limited amount of bits (not constant quantizer)
then a slightly blurred input might look better after decoding than a slightly
sharpened one, especially for lower bitrates
sws numbers / names / artifact types
-------------------------------------------------------------------------
r ringing
b blurry
a aliasing (downscale only)
m mosaic (blocky) (upscale only)
-sws 0 fast bilinear bA
-sws 1 bilinear b
-sws 2 bicubic high sharpness r
low sharpness b
-sws 3 experimental ????
-sws 4 nearest neighbour AM
-sws 5 area ba
-sws 6 luma bicubic / chroma bilinear r
-sws 7 gauss low sharpness B
high sharpness AM
-sws 8 sinc R
-sws 9 lanczos long filter R
short filter b
-sws 10 bicubic spline r
Notes:
area upscale is identical to bilinear
-18
View File
@@ -1,18 +0,0 @@
# Phase 6 性能改动与证据
## 已采用
1. **事件驱动渲染更新**:`render.cpp` 不再每 16ms 无条件调用 `mpv_render_context_update`,只在 mpv callback/窗口事件唤醒时更新;为避免个别会话漏回调,最多每 250ms 做一次安全探测。5 次 startup 的 update 累计从优化前 617 降到最终实现 364,且 20 次 startup 未复现永久 BUFFERING。
2. **核心快照合并**:`PlaybackEngine` 对 100ms 内的重复 `get_snapshot/get_tracks/get_media_info/get_network_info` 做合并,动作仍立即提交到 MPV,状态最多延迟一个合并窗口。200 次 play/pause 末次面板显示 `coreSyncSkipped=239`;无 error,状态为 PAUSED。
3. **NAPI 诊断轮询降频**:Playback Lab 面板由 250ms 调整为 500ms;播放控制、音频 callback 和渲染线程不依赖该轮询。这样减少纯 UI 诊断序列化,不改变 MPV 时钟。旧/新间隔在 `Index.ets` 可审计,运行中由 `getStateCalls`、`serializeUsTotal` 记录。
4. **session id 原子读取**:渲染线程不再为每次唤醒获取 `PlaybackEngine` 快照互斥锁,session 更新仍由核心快照写入 release/acquire atomic;首帧 session 隔离 smoke 和 stress 均通过。
## 未采用
- 不做主动丢帧、降低音频采样率或缩小视频分辨率;这些会改变播放质量。
- 不做 GPU mapper、OHCodec 直出或 FFmpeg-Harmony 源码改动;实体设备 capability 和 fence 证据不足。
- 不把 PSS、heap、CPU permille、2 小时长跑结果伪装成已测数据。
## 回归与开销
默认构建、软件构建和 FFmpeg-Harmony 构建均通过。默认 HAP 在模拟器上完成 startup/seek/audio 状态面板、load/stop 100 次和 play/pause 100 次。release/profile 构建未启用持续 tracing;诊断仅累加轻量计数和受控序列化时间。
-20
View File
@@ -1,20 +0,0 @@
# Phase 6 性能指标
所有时间均为单调时钟相对当前 load session 的毫秒数。`-1` 表示该事件没有发生或无法可靠观察;runner 会把负值转换为 `null`,不参与百分位统计。
| 类别 | 字段/来源 | 语义 | 模拟器可用性 |
| --- | --- | --- | --- |
| 启动 | `timeline.startupTotalMs` / `firstSwapMs` | load 到首个成功 EGL swap | 可用 |
| 启动分段 | create、initialize、load、START_FILE、FILE_LOADED、decoder reconfig、render context | 各阶段首次事件 | 可用,缺失阶段为 -1 |
| 网络 | `firstByteMs`、`networkWaitMs`、HTTP status、206/range | 只记录观察到的请求/日志证据 | HTTP fixture 可用,HTTPS/HLS pending |
| seek | `seekCommandLatencyMs`、`seekRecoveryMs`、`seekFirstSwapMs` | seek request/command 到恢复首帧/首 swap | 相对 `+10s` 可用 |
| buffering | enter/exit 次数、累计时长、单次最大时长 | cache-buffering 状态边沿 | 字段可用;本地短片通常为 0 |
| 音频 | renderer start、首 callback、submitted/played、delay、starvation、flush/reset | 首 callback 时间由诊断轮询观察,不冒充 callback 内精确时间 | 模拟器字段可用,设备路由 pending |
| 视频 | dropped、late、mistimed、decoder frames、render count | MPV stats 或 renderer 计数 | 计数接口可用,部分 MPV 值可能为 -1 |
| 内存 | `/proc/<pid>/status` VmRSS/VmSize | 进程 RSS/虚拟地址空间 | 可用;PSS/精确 heap 为 null |
| CPU | `/proc/<pid>/stat` utime+stime | 原始 jiffies | 原始值可用;未取得可靠 tick/采样归一化,不输出 permille |
| 线程/资源 | Threads、`/proc/<pid>/fd` | 进程线程和 fd 快照 | 可用;fd=0 视为不可用 |
| NAPI | getState 次数、序列化微秒、估算 payload | NAPI 状态导出成本 | 可用;payload 是上界估算,不是分配追踪 |
| 渲染诊断 | render update/frame update、core sync/skipped | 优化前后计数及合并效果 | 可用 |
NAPI 返回使用 Number 以保持 Inspector 可序列化;大整数仅用于内部原子计数。任何未被设备或 API 真实提供的数据都保持 unavailable,不由模型推断。
-40
View File
@@ -1,40 +0,0 @@
# Phase 1 Dependency Matrix
基线:MPV `182fa6ca49f455cadb884858f386e2f00540aeb7`。状态来自当前 `meson.build`、实际 Meson 配置与最终产物,不沿用 Phase 0 假设。
| Dependency | Version / Commit | Required / Optional | Build System | License | Link | OHOS Status | pkg-config | Blocking libmpv |
|---|---|---|---|---|---|---|---|---|
| libavcodec | 63.8.101 / FFmpeg `3acec0a1` | Required | configure + Make | LGPL-2.1+ | Static | Built AArch64 | `libavcodec` | Yes |
| libavfilter | 12.3.101 / FFmpeg `3acec0a1` | Required | configure + Make | LGPL-2.1+ | Static | Built AArch64 | `libavfilter` | Yes |
| libavformat | 63.6.100 / FFmpeg `3acec0a1` | Required | configure + Make | LGPL-2.1+ | Static | Built AArch64 | `libavformat` | Yes |
| libavutil | 61.6.100 / FFmpeg `3acec0a1` | Required | configure + Make | LGPL-2.1+ | Static | Built AArch64 | `libavutil` | Yes |
| libswresample | 7.2.100 / FFmpeg `3acec0a1` | Required | configure + Make | LGPL-2.1+ | Static | Built AArch64 | `libswresample` | Yes |
| libswscale | 10.2.100 / FFmpeg `3acec0a1` | Required | configure + Make | LGPL-2.1+ | Static | Built AArch64 | `libswscale` | Yes |
| libplacebo | 7.360.1 / `cee9b076` | Required | Meson | LGPL-2.1+ | Static | Built AArch64; graphics backends off | `libplacebo` | Yes |
| libass | 0.17.5 / `4a05d812` | Required | Meson | ISC | Static | Built AArch64 | `libass` | Yes |
| FreeType | 2.14.3 / `0a0221a1` | libass transitive | Meson | FTL/GPL-2.0 dual | Static | Built AArch64 | `freetype2` | Yes via libass |
| HarfBuzz | 14.4.0 / `36cb489c` | libass transitive | Meson | MIT | Static | Built AArch64 | `harfbuzz` | Yes via libass |
| FriBidi | 1.0.16 / `68162bab` | libass transitive | Meson | LGPL-2.1+ | Static | Built AArch64 | `fribidi` | Yes via libass |
| OpenSSL | 3.6.4 / `d3c1b116` | HTTPS transitive | Configure + Make | Apache-2.0 | Static | Built AArch64 | `openssl` | Yes for HTTPS |
| zlib | SDK 1.2.13 | MPV optional; enabled | SDK | Zlib | Dynamic | SDK AArch64 `libz.so` | generated `zlib` | No |
| threads / dl / m / rt | SDK API 24 | Platform required | SDK | SDK system | Dynamic/system | Compile verified | Meson compiler/system | Yes |
| EGL / GLES | SDK API 24 | Phase 2 integration | SDK | SDK system | Not linked | Headers/libs available | No target `.pc` used | No |
| libcurl, libarchive, libbluray, libavdevice | Optional | Disabled | Various | Various | None | Not built | Not queried or skipped | No |
| CD/DVD, Lua, MuJS, uchardet, VapourSynth, zimg, rubberband | Optional | Disabled | Various | Various | None | Not built | Skipped | No |
| ALSA, AudioUnit, CoreAudio, JACK, OpenAL, AAudio, OpenSLES, OSS, PipeWire, Pulse, sndio, WASAPI | Optional AO | Disabled | Various | Various | None | Not applicable | Skipped | No |
| X11, Wayland, DRM/GBM, Cocoa, D3D11, Vulkan, VAAPI, VDPAU, CUDA, Android Media NDK | Optional VO/hwaccel | Disabled | Various | Various | None | Not applicable | Skipped | No |
实际依赖图:
```text
OpenSSL ───────────────┐
SDK zlib ──────────────┼─> FFmpeg ───────────────┐
└─> FreeType ─> HarfBuzz ─┤
FriBidi ──────────────────────────────┐ │
FreeType + HarfBuzz + FriBidi ───────> libass ───┤
libplacebo ──────────────────────────────────────┼─> libmpv.so
SDK libc/pthread/dl/m/rt/z ──────────────────────┘
SDK native_media_* + native_window <─ FFmpeg OHCodec
```
所有 `.pc` 仅由 `scripts/pkg-config-ohos.sh` 从 OHOS prefix 解析。最终校验未发现 `/opt/homebrew`、`/usr/local`、Mach-O 或 `.dylib` target 输入。
-67
View File
@@ -1,67 +0,0 @@
# Phase 2 Playback Lab 报告
日期:2026-08-29
分支:`harmony-port`
## 范围
`playback_lab/` 是独立 HarmonyOS 验证工程,仅链接 Phase 1 生成的 arm64 `libmpv.so.2`,不复用 Velix 主应用的页面或业务逻辑。当前只验证 GLES/EGL、软件解码和本地 MP4 首帧,不接入音频或硬解。
## 构建与产物
环境:DevEco Studio SDK API 24,HarmonyOS emulator 6.1.0.126,`aarch64-linux-ohos`。构建命令:
```sh
./scripts/stage-mpv.sh
DEVECO_SDK_HOME=/Applications/DevEco-Studio.app/Contents/sdk \
JAVA_HOME=/Applications/DevEco-Studio.app/Contents/jbr/Contents/Home \
PATH="$JAVA_HOME/bin:$PATH" \
/Applications/DevEco-Studio.app/Contents/tools/hvigor/bin/hvigorw \
--no-daemon assembleHap --mode module -p product=default
```
结果:`BUILD SUCCESSFUL`。HAP:`playback_lab/entry/build/default/outputs/default/entry-default-unsigned.hap`,约 31 MiB,SHA-256 `877568c34a02a4d61b8c4ed7721dd2e6486cb675cd1a58a8eaf44c9b00de05f5`。包内包含 `libentry.so`、`libmpv.so.2` 和 `libc++_shared.so`;未提交 34 MiB 的 staged native 库。
构建仅保留两项预期警告:NAPI 动态模块在当前 SDK 尚未完成类型校验,以及调试 HAP 未配置正式签名;两者不影响本轮模拟器安装验证。
## 模拟器验证
目标:`127.0.0.1:5555`,虚拟设备进程配置为 Pura 90 Pro Max。
| 项目 | 结果 | 证据 |
| --- | --- | --- |
| HAP 安装、启动 | 通过 | `hdc install -r`、`aa start` 成功 |
| XComponent / OHNativeWindow | 通过 | 页面显示 `surface=READY`;Native 回调无崩溃 |
| EGLDisplay / EGLConfig / EGLContext / WindowSurface / makeCurrent | 通过 | 页面显示 `egl=READY`;日志 `EGL vendor=Express ... renderer=Mali-G77` |
| libmpv 初始化与 Render API | 通过 | 页面显示 `mpv=READY`;日志 `libmpv initialized` |
| 本地 H.264 MP4 软件解码 | 通过 | 本地 `sample.mp4` 首帧及持续播放成功 |
| Pause / Resume / Stop / Release | 通过 | UI 状态分别出现 `PAUSED`、`PLAYING`、`STOPPED`、`RELEASED` |
| Seek | 通过 | 有效播放中 `-10s/+10s` 不报错;跳转按 duration 做边界裁剪 |
| 释放后重新进入 | 通过 | `RELEASED` 后重新启动回到 `PLAYER_CREATED`,Surface/EGL/mpv 再次 READY |
| BigInt 状态序列化 | 通过 | `renderCount` 改为 Number,Inspector 不再出现 BigInt 序列化错误 |
一次本地播放快照:`position=2.00`、`duration=5.05`、`fps=29.97`、`dropped=0`、`firstFrameMs=622`、`renderCount=52`。另一轮快照首帧为 117 ms;`firstFrameMs` 从 `load()` 调用开始计时,并在第一次 `eglSwapBuffers` 后记录。
## First Frame Timeline
1. T0:XComponent 创建并取得回调传入的 `OHNativeWindow`,增加 NativeObject 引用。
2. T1:专用渲染线程创建 EGLDisplay、RGBA8 EGLConfig、ES3 EGLContext 和 EGLWindowSurface,执行 `eglMakeCurrent`。
3. T2:同一渲染线程创建 mpv OpenGL Render Context,注册 update callback。
4. T3:`loadfile` 完成本地文件打开,MPV 软件解码产生可渲染帧。
5. T4:渲染线程执行 `mpv_render_context_render`、`eglSwapBuffers`,记录 `firstFrameMs` 和 `renderCount`。
## 尚未完成的真机验收
本轮只有模拟器,没有接入 HarmonyOS 实体手机,因此 Phase 2 不能标记为最终 COMPLETE。以下项目保持待验证:实体机 HAP 安装、实体机本地 1080P H.264/AAC、HTTP MP4、HTTPS MP4、HLS、20 次 Surface recreate、20 次进出、50 次 Play/Pause、50 次 Seek、横竖屏 20 次、30 分钟稳定性、CPU/RAM 曲线。
## 风险状态
| 风险 | 当前状态 |
| --- | --- |
| RISK-01 POSIX / DNS / network runtime | 未验证,网络链路留待实体机 |
| RISK-02 XComponent / OHNativeWindow lifecycle | 模拟器已验证;实体机压力测试待做 |
| RISK-03 EGL context / Surface / plain-gl | 模拟器已验证;旋转与重建待做 |
| RISK-04 OHCodec buffer lifecycle | 保持开放;Phase 2 明确禁用硬解 |
| RISK-05 CA / HAP packaging | 未签名调试 HAP 安装通过;正式签名待接入 |
新增 MPV upstream patch:0;未修改第三方源码。进入 Phase 3(`ao_harmony + OHAudio`)的前置条件是实体机首帧、控制和压力矩阵完成。
-47
View File
@@ -1,47 +0,0 @@
# Phase 4 状态
## 结论
**Phase 4 CODE COMPLETE / DEVICE VALIDATION PENDING**
GPU interop:**BLOCKED**(见 `OHCODEC-BLOCKER-001`)。
代码完成了 MPV 识别 FFmpeg OHCodec、H.264/HEVC decoder selection、能力检查、
direct-surface 实验接入、软件回退、诊断和构建矩阵。当前没有实体设备,因此
不能写成 hardware decode complete、zero-copy complete 或 HDR complete。
## 验证矩阵
| 项目 | 状态 |
| --- | --- |
| `AV_PIX_FMT_OHCODEC` → `IMGFMT_OHCODEC` | COMPILE VERIFIED |
| `AV_HWDEVICE_TYPE_OHCODEC` 独立 backend | COMPILE VERIFIED |
| H.264 / HEVC selection guard | COMPILE VERIFIED;DEVICE TEST REQUIRED |
| API24 OHCodec capability 查询 | COMPILE VERIFIED;设备返回值待验 |
| Direct Surface plumbing | COMPILE VERIFIED;需独占窗口 |
| GPU interop mapper | BLOCKED:上游无可导入 non-surface output contract |
| Software decode / OHAudio | CLEAN BUILD VERIFIED |
| OHCodec-enabled / software / GL-probe | AArch64 BUILD VERIFIED |
| Playback Lab HAP | BUILD VERIFIED;unsigned |
| Emulator | APP START + EGL SMOKE VERIFIED;不是硬解验证 |
| Physical device | PENDING |
## 产物与审计
- `build/ohos-arm64/prefix/lib/libmpv.so.2.5.0` 为 AArch64 ELF,保留 OHCodec、
OHAudio 及诊断符号。
- `scripts/build-mpv-ohos-software.sh` 使用独立 build 目录关闭 OHCodec,避免
软件回退配置与硬解构建互相污染。
- FFmpeg 保持锁定 upstream commit,未复制 SDK/第三方实现;许可证审计见
`LICENSE_AUDIT.md`。
- 当前 blocker、NativeBuffer 生命周期、EGL/fence/crop/stride 待验证项见
`DOCS/OHCODEC_GPU_INTEROP_DESIGN.md` 与 `DOCS/HDR_PIPELINE_NOTES.md`。
- Phase 4.5 FFmpeg export contract 独立维护于
`../Velix-FFmpeg-Harmony`;MPV 只提供
`scripts/build-mpv-ohos-ffmpeg-harmony.sh` 前缀切换,未接入 mapper。
## 下一阶段最大风险
实体设备上的 codec capability、surface recreate、output buffer ownership、
acquire/release fence、crop/stride/color metadata,以及如何在不修改 FFmpeg
surface contract 的前提下把硬件帧交给 MPV GPU composition。
-18
View File
@@ -1,18 +0,0 @@
# Phase 5 状态
状态:CODE COMPLETE / HTTP Range emulator verified / device and remaining network validation pending。
已完成:
- 默认软件解码路径保持 hwdec=no,未启用 harmony-ohcodec-gl。
- 新增 include/mpv/harmony_playback.h 稳定 C API;公共结构不包含 mpv_handle、mpv_node、AVFrame。
- 请求字段、会话 generation、线程安全控制、稳定状态/错误、轨道、媒体信息和网络缓存诊断已实现。
- Playback Lab 已接入请求对象、脱敏 header 编辑器、速度/音量/静音、外部字幕、轨道/媒体/网络面板。
- 文档、测试矩阵和构建矩阵已补齐。
- default/software/FFmpeg-Harmony 三套 AArch64 libmpv 构建、稳定 API 符号校验与 Playback Lab Native/ArkTS/PackageHap 编译已通过;使用 DevEco 自带 JBR 的 `--no-daemon` 流程生成 unsigned HAP,并由 devecocli 安装模拟器。
- 模拟器已启动 Playback Lab,完成本地 H.264/AAC MP4 首帧、软件解码、播放结束状态和渲染指标 smoke;通过宿主受控 HTTP Range server 完成 HTTP MP4 播放与 Range 请求验证;HTTPS/HLS/网络字幕/鉴权资源仍需真实网络源。
- 当前构建日志显示 iconv feature disabled,因此字幕适配器只承诺 UTF-8;GBK/Big5/Shift-JIS 会返回稳定 SUBTITLE 错误。
尚待真实环境确认:HTTPS/HLS、重定向 header 继承、鉴权资源、网络字幕和物理设备音频/视频输出。不可从本地编译结果推断这些运行时结论。
许可:新增代码沿用 MPV 的 LGPL/ISC 边界,不引入商业字体或新的第三方网络库;公共头文件保持独立 ABI。Phase 6 最大风险是将 Provider 业务请求正确转换为通用请求,以及在实体设备上确认网络、音频路由和字幕字体 provider 行为。
-31
View File
@@ -1,31 +0,0 @@
# Phase 5 测试矩阵
| 场景 | 方式 | 结果 |
|---|---|---|
| 本地 MP4 H264/AAC | Playback Lab | 模拟器安装启动并完成首帧/播放/结束 smoke;网络与实体机待确认 |
| 本地 MKV H264/AAC/SRT | Playback Lab | API/UI 编译覆盖;设备运行待确认 |
| 本地 MKV HEVC/AAC/ASS(software) | software mpv | AArch64 构建通过;设备运行待确认 |
| HTTP MP4 | 宿主受控 Range server + Playback Lab | 模拟器通过;观察到 `bytes=0-`、后续 byte range,请求完成且 state=ENDED;`rangeSupported=1` |
| HTTPS MP4 | 受控测试源 | TLS 校验代码就绪;网络环境待确认 |
| HLS Master | 受控测试源 | 复用 mpv/FFmpeg;网络环境待确认 |
| HLS Variant/chunked | 受控测试源 | 复用 mpv/FFmpeg;网络环境待确认 |
| 多音轨 MKV | Playback Lab | Track API/UI 已接入;设备运行待确认 |
| 内封 SRT | Playback Lab | Track API/UI 已接入;设备运行待确认 |
| 内封 ASS/SSA | Playback Lab | libass 路径保留原样式;设备运行待确认 |
| 外挂 SRT | Playback Lab | URL 与本地 file 路径入口已接入;设备运行待确认 |
| 外挂 ASS/SSA | Playback Lab | URL 与本地 file 路径入口已接入;设备运行待确认 |
| 网络字幕/WebVTT | Playback Lab | API/UI 已接入;网络环境待确认 |
| Authorization header | 受控测试源 | 脱敏与隔离代码覆盖;服务端待确认 |
| Cookie header | 受控测试源 | 单会话 Cookie header;服务端待确认 |
| Referer/User-Agent | 受控测试源 | 请求字段与 header 均支持;服务端待确认 |
| Range seek / 大 seek | 宿主受控 Range server + Playback Lab | 模拟器通过 HTTP Range 请求;最近 seek 结果由 `seekResult` 暴露;HTTPS 大 seek 待继续验收 |
| Redirect | FFmpeg-Harmony `f3a036c695` + 受控测试源 | 跨 origin 丢弃 Authorization/Cookie/Token/API-key/Referer;同源保留;受控重定向运行验证待补 |
| 403 / 404 / 500 | 受控测试源 | 稳定 HTTP 错误码已接入;网络环境待确认 |
| Broken media / reset / timeout / TLS | 受控测试源 | DEMUX/NETWORK/TLS 分类已接入;网络环境待确认 |
| 默认 FFmpeg(OHCodec enabled) | scripts/build-mpv-ohos.sh | AArch64 ELF 与符号校验通过 |
| software MPV(OHCodec disabled) | scripts/build-mpv-ohos-software.sh | AArch64 ELF 校验通过 |
| FFmpeg-Harmony | scripts/build-mpv-ohos-ffmpeg-harmony.sh | AArch64 ELF 与符号校验通过 |
| audio-enabled MPV | 默认脚本 `harmony-audio=enabled` | 与默认构建一并通过 |
| Playback Lab HAP | hvigor assembleHap --no-daemon | CompileArkTS/Native/PackageHap 通过;无正式签名,devecocli 已安装模拟器 |
产品禁止以 UI 定时器伪造 buffering;只有 mpv 事件/属性可改变对应诊断字段。
-33
View File
@@ -1,33 +0,0 @@
# Phase 6 状态
状态:**CODE COMPLETE / EMULATOR BENCHMARKED / DEVICE BENCHMARK PENDING**。
Phase 6 交付范围是可重复的 Playback Lab 性能时间线、资源采样、NAPI 诊断、四项低风险优化和模拟器回归。没有修改 GPU mapper,也没有修改 `Velix-FFmpeg-Harmony`。
## 已完成
- 启动、seek、buffering、音频、渲染、NAPI 和资源指标已接入稳定快照;未能可靠取得的 PSS、精确 heap、归一化 CPU 均明确返回 `null` 或文档标记为 unavailable。
- `benchmark/runner/run_emulator.py` 支持 startup、seek、load/stop stress、play/pause stress,输出重复统计(min/max/avg/median/P90/P95)和逐次资源采样。
- 渲染循环改为 mpv update callback 驱动,并保留 250ms 安全探测;快照同步增加 100ms 合并窗口;session id 读取改为 atomic;诊断轮询从 250ms 调整为 500ms。
- default、software、FFmpeg-Harmony 三套 AArch64 libmpv 均通过符号和 OHOS ELF 校验;Playback Lab HAP 编译、安装和模拟器启动通过。
## 模拟器结果
测试设备:`127.0.0.1:5555`,本地 H.264/AAC MP4,软件解码,`ao_harmony`。
| 场景 | 结果文件 | 关键结果 |
| --- | --- | --- |
| startup_local ×20 | `benchmark/results/startup_local-1787999362.json` | 首帧 P50 506ms,P95 1150ms;20/20 暴露时间线,无错误 |
| seek_local ×50 | `benchmark/results/seek_local-1787999663.json` | 恢复 P50 277ms,P95 631ms;49/50 有恢复时间线,1 次 `SEEK` 命令错误 |
| play_pause ×200 | `benchmark/results/play_pause-1788000475.json` | 无错误;末次 RSS 213228KB、线程 39、fd 1 |
| load/stop ×100 | `benchmark/results/stress_lifecycle-1788000229.json` | 进程保持存活;RSS/线程/fd 采样完成,PSS 不可用 |
优化前后计数基线:同等 5 次 startup 中,渲染 update 累计由 617 降至最终版本 364;NAPI/核心同步计数通过面板快照记录。200 次 play/pause 无错误,末次 `coreSyncSkipped=239`,说明合并窗口实际生效。
最终 5 次 smoke 还观察到 `firstAudioCallbackMs` P50 256ms、`audioRendererStartMs` P50 720ms;这些值来自 500ms 诊断轮询,不能替代设备音频 trace。file:// 场景的 `networkWaitMs` 保持 -1,buffering 计数按真实状态边沿记录。
## 尚待实体设备
实体设备上的真实 PSS/heap、CPU 利用率、显示合成延迟、音频路由/A-V 偏差、OHCodec capability、长时间(2h)稳定性、HTTPS/HLS/鉴权/网络字幕仍未验证。Phase 7 建议先建立实体设备矩阵和受控 HTTPS/HLS fixture,再评估硬解直出和端到端 A/V 指标。
许可边界沿用 MPV LGPL/ISC;本阶段未引入新的第三方依赖或商业资源。
-13
View File
@@ -1,13 +0,0 @@
# PlaybackEngine
`PlaybackEngine` 是主 App 的唯一播放抽象,`MpvPlaybackEngine` 是 Core binding 的适配器,`PlaybackController` 负责页面状态刷新与生命周期编排。
业务层只使用:
- `PlaybackSource { kind, url, headers }`
- `PlaybackState { state, position, duration, media, tracks, metrics }`
- play/pause/stop/release、seek、音量/倍速、音轨/字幕操作
provider 接口暂不绑定 Emby、Jellyfin、百度网盘等服务。HTTP(S) 源支持有限 headers;复杂 header 编辑器、登录与播放列表不在 v0 范围。
默认解码策略为软件解码;`requestedDecoder=software` 是主 App 的固定默认值,OHCodec GL mapper 不在本阶段启用。
-13
View File
@@ -1,13 +0,0 @@
# Playback 公共 API
mp_harmony_playback.h 是上层唯一稳定契约,提供 PlaybackRequest、PlaybackState、Track、MediaInfo、PlaybackMetrics 和 PlaybackError 模型别名。PlaybackRequest 只携带 URL、标题、起始位置、User-Agent、Referer、Cookie、任意 HTTP headers、字幕源、语言偏好、轨道选择、速度和超时。空值或负值按头文件注释表示未设置。
公开控制命令:Play、Pause、TogglePause、SeekAbsolute、SeekRelative、SeekPercent、Stop、SetSpeed、SetVolume、Mute、SetAudioEnabled、SelectAudioTrack、SelectSubtitleTrack、DisableSubtitle、AddSubtitle、RemoveSubtitle、SetSubtitleDelay、SetSubtitleScale、SetSubtitlePosition、SetAssOverride、GetTracks、GetMediaInfo、GetNetworkInfo、GetPlaybackState。
所有控制入口串行化;请求中的字符串在调用返回前复制到 mpv 所需的临时节点,调用者可以立即释放输入内存。session id 随每次 load 增长,旧 playlist event 不得改变新会话状态。
Playback Lab 的 C++ 外壳只做 NAPI/渲染适配,所有请求、控制和状态快照均委托上述 C API;渲染器使用私有句柄桥接,不进入公共 ABI。
当前 OHOS 构建未启用 iconv,外挂字幕 encoding 只接受 UTF-8/UTF8;其他编码通过稳定 SUBTITLE 错误拒绝。
起播 `start_position_ms` 在 `MPV_EVENT_FILE_LOADED` 后执行 exact seek,避免异步 load 竞态;`title` 映射到 `force-media-title`。MediaInfo 同时保留历史 `width`/`height`/`color_space` 字段,并提供 `video_width`/`video_height`/`color_matrix` 别名。NetworkInfo 的 `range_supported` 为 -1/0/1(未知/未观察到 206/观察到 206),`seek_result` 为最近一次 seek 的 `ok` 或脱敏错误。
-14
View File
@@ -1,14 +0,0 @@
# Player lifecycle
产品调用顺序:
```text
create → attachSurface → prepare(source) → play
↘ pause / seekTo / seekBy / tracks / subtitles
stop → prepare 或 release
detachSurface → release
```
`onForeground`、`onBackground`、`onSurfaceAvailable`、`onSurfaceDestroyed` 是显式适配接口。当前后台策略只释放/暂停 Surface 相关资源,不承诺后台播放、PiP 或通知栏控制。Surface 与 player 生命周期解耦:Surface 销毁不会自动销毁 Core player,新的 Surface 可重新 attach。
页面状态以 500ms 轮询读取,错误通过 `errorCode` 与 `lastError` 映射到 UI;首帧、缓冲、seek、音频与 renderer 指标仅在“显示调试统计”打开时显示。
-70
View File
@@ -1,70 +0,0 @@
# HarmonyOS Porting Plan
## 1. 源码扫描结论
MPV 核心播放、demux、filter、subtitle、A/V sync、command、client API 基本跨平台。平台耦合集中在 `meson.build`、`osdep/`、`audio/out/`、`video/out/`、`video/decode/` 与 hwdec interop。当前仓库搜索不到 Harmony/OHOS 代码。
MPV 当前检测逻辑 `posix = not win32`;非 Darwin POSIX 自动选择 `osdep/timer-linux.c`。OHOS 编译器同时定义 `__OHOS__` 与 `__linux__`,因此所有 Linux feature 必须由 Meson probe 或显式 Harmony gate 决定,不能只看 `__linux__`。
## 2. 文件级移植矩阵
| MPV source / file | 当前依赖 API / 行为 | HarmonyOS 替代或复用 | Patch | 风险 | 阶段 | 工作量 |
|---|---|---|---:|---|---|---|
| `meson.build` OS detection | win32/darwin/posix/android/linux 分支 | 增加 `ohos = host_machine.system() == 'ohos'`、`features['harmony']`、config `HAVE_HARMONY` | 是 | 高 | P1-01 | 1–2d |
| `meson.build` platform sources | 非 Darwin POSIX选 Linux timer;Android 单独 EGL/media | 为 OHOS 明确选择 POSIX可复用源、Harmony AO/可选 context;不进入 Android 分支 | 是 | 高 | P1-01 | 1d |
| `osdep/threads-posix.c` | pthread、cond、mutex、TLS | 直接使用 SDK `pthread.h`;线程名由 feature probe | 预计否/小 | 中 | P1-02 | 0.5–1d |
| `osdep/timer-linux.c` | `clock_gettime`、`nanosleep` | API均可用;可先明确选该实现,后续重命名/拆成 POSIX timer 以便 upstream | 小 | 低 | P1-02 | 0.5d |
| `osdep/path-unix.c` | POSIX paths、HOME/XDG 等桌面约定 | 文件 API复用;应用 data/cache 路径最终由宿主注入,不能依赖 HOME | 小/后续 | 中 | P1-03 | 1–2d |
| `osdep/language-posix.c` | locale/environment | 先复用;设备语言由宿主 API注入作为后续增强 | 预计否 | 低 | P1-03 | 0.5d |
| `osdep/subprocess-posix.c` | fork/exec/process | Harmony app 不需要 subprocess;若 probe失败明确选 `subprocess-dummy.c` 并返回 unsupported | 小 | 中 | P1-03 | 0.5d |
| `osdep/terminal-unix.c` | tty/terminal | libmpv build 不需要交互 terminal;cplayer关闭 | 配置 | 低 | P1-01 | 0.2d |
| `input/ipc-unix.c` | Unix domain socket | 首版 libmpv 可关闭/保留按 probe;产品 API不暴露 IPC | 小/配置 | 低 | P1-03 | 0.5d |
| `stream/` + FFmpeg AVIO | POSIX socket、DNS、FFmpeg HTTP/TLS | socket/DNS复用;HTTP/HLS由 FFmpeg;TLS 使用交叉构建 OpenSSL | MPV 否 | 高 | P1-04 | 3–5d |
| `demux/demux_lavf.c` | libavformat | 直接复用官方 FFmpeg OHOS build | 否 | 中 | P1-04 | 1d 验证 |
| `sub/`、`sub/osd_libass.c` | libass、字体栈 | 直接复用;字体发现/应用字体目录需配置验证 | 小/配置 | 中 | P1-05 | 2–3d |
| `include/mpv/client.h` | libmpv C API | 直接复用仅作内部依赖;最终 Velix API再封装 | 否 | 低 | P1 | 0.5d |
| `include/mpv/render*.h` | OpenGL / software public render API | 直接使用 OpenGL API;宿主提供 GLES proc resolver/FBO | 否 | 中 | P2-01 | 2–4d lab glue |
| `video/out/vo_libmpv.c` | libmpv render bridge | 直接复用 | 预计否 | 中 | P2-01 | 1d 验证 |
| `video/out/opengl/libmpv_gl.c` | OpenGL Render API | 直接复用 GLES/EGL context;验证所需 GL symbols | 小/预计否 | 中 | P2-01 | 2d |
| `video/out/opengl/context*.c` | Android/X11/Wayland/DRM/Windows context | 首版 libmpv plain-gl 不新增;如 MPV自管 context,再新增 `context_harmony.c` | 后续 | 中 | P2-02 | 3–5d |
| `video/out/vulkan/context.c` | libplacebo/Vulkan context registry | 若采用 Vulkan新增 `context_harmony.c`,调用 `vkCreateSurfaceOHOS` | 后续是 | 高 | P4 | 7–12d |
| `audio/out/ao.c` | AO driver registry | 注册 `audio_out_harmony` | 是 | 低 | P2-03 | 0.5d |
| `audio/out/ao_aaudio.c` | Android AAudio callback模型 | 仅作结构参考;实际新增 `ao_harmony.c` 使用 OHAudio | 新文件 | 高 | P2-03 | 5–8d |
| `audio/out/buffer.c` | MPV AO buffering | 直接复用,由 OHAudio callback消费 | 预计否 | 中 | P2-03 | 1–2d 验证 |
| `video/fmt-conversion.c` | FFmpeg pixfmt↔MPV imgfmt | 增加 `AV_PIX_FMT_OHCODEC` 映射前先确认 MPV image semantics | 后续是 | 高 | P3-02 | 1–2d |
| `video/decode/vd_lavc.c` | hw device选择、frames | 加入 `AV_HWDEVICE_TYPE_OHCODEC` 并传递 native window/device ctx | 后续是 | 高 | P3-02 | 3–6d |
| `video/out/gpu/hwdec.c` | GPU interop registry | 注册 Harmony interop driver,仅在真实 buffer import 方案明确后 | 后续是 | 极高 | P3-03 | 7–15d |
| `video/out/hwdec/` | VAAPI/DRM/VT/Android interop | 新增隔离的 `hwdec_ohcodec.c`,不散落 OHOS ifdef | 后续是 | 极高 | P3-03 | 10–20d |
| `video/out/opengl/egl_helpers.c` | 通用 EGL config/context helpers | 优先复用;验证 GLES profile、10-bit config、context loss | 可能小 | 中 | P2-01 | 2d |
| `misc/io_utils.c` / poll wrapper | poll/FD helper | POSIX poll 复用;epoll不作为无证据优化 | 预计否 | 低 | P1-02 | 0.5d |
| logging (`common/msg.c`) | stderr/terminal | 首编译复用;lab阶段加 HiLog adapter,不能替换核心 logger | 后续小 | 低 | P2 | 1–2d |
## 3. 分阶段执行
### P1:First Build
1. P1-01:Meson 识别 OHOS,构建只产 libmpv。
2. P1-02:逐项编译 POSIX 基础层;每个 blocker 单独记录。
3. P1-03:明确 subprocess/IPC/path 不支持或可复用边界。
4. P1-04:构建 FFmpeg software decode + network + TLS。
5. P1-05:构建 libass/font stack、libplacebo,链接 libmpv。
门禁:AArch64 libmpv 真实产物、导出符号、依赖清单、无 Android API、无宿主机库。
### P2:Playback Lab 基线
XComponent + EGL + libmpv OpenGL Render API;新增 OHAudio AO;URL/local file、play/pause/seek/stop/release、日志和基础 metrics。无正式产品 UI。
### P3:硬件解码
先验证 FFmpeg `h264_oh` / `hevc_oh`,再决定 direct surface 或 GPU import。所有优化必须对比软件解码基线。
### P4:定点性能与 Vulkan
只有 Benchmark 证明 GLES/上传或合成是瓶颈时,研究 Vulkan/HDR/zero-copy;每个 patch 必须有前后数据。
## 4. 可直接复用与必须新增
可直接复用:libmpv client/render API、player core、demux、software decode glue、filters、A/V sync、subtitle/libass、playlist/property/command、POSIX threads/timer/socket/DNS/poll/mmap 的已验证子集、OpenGL renderer。
必须新增:Harmony Meson detection、OHAudio AO、Playback Lab 的 XComponent/EGL native glue。硬解阶段预计新增 OHCodec image/device/interoperability backend。Vulkan context 不是首版必需。
-35
View File
@@ -1,35 +0,0 @@
# Risk Assessment
## Top 10
| # | 风险 | 可能性 / 影响 | 证据 | 缓解与验收 |
|---:|---|---|---|---|
| 1 | MPV 无 OHOS 平台检测,OHOS 被泛 POSIX/Linux 逻辑误分类 | 高 / 高 | 当前 `meson.build` 无 harmony/ohos;编译器同时定义 `__linux__` | 显式 `features['harmony']`;每个 Linux feature 用 compile/link probe |
| 2 | 交叉依赖链复杂且 pkg-config 污染宿主库 | 高 / 高 | 本机无 Meson/Ninja/pkg-config;MPV 强依赖 FFmpeg/libplacebo/libass | 单一 staging prefix、清空默认 `PKG_CONFIG_PATH`、产物 readelf 审计 |
| 3 | HTTPS/TLS 无可直接复用的 FFmpeg Harmony backend | 高 / 高 | SDK有 `libnet_ssl.so`,FFmpeg不使用它 | 交叉构建 OpenSSL;真机证书/SNI/Range/HLS 测试 |
| 4 | Surface 生命周期与 GL 线程竞态导致黑屏/崩溃 | 高 / 高 | XComponent异步 create/change/destroy,EGL context 有线程归属 | 单一 render thread 状态机;销毁屏障;旋转/后台压力测试 |
| 5 | OHAudio callback 与 MPV buffer/clock 语义不匹配 | 中高 / 高 | MPV无 Harmony AO | realtime-safe callback;timestamp/delay 校准;seek/route switch 测试 |
| 6 | FFmpeg OHCodec虽已 upstream,但 MPV完全不识别 OHCODEC | 高 / 高 | MPV无 `AV_PIX_FMT_OHCODEC` / `AV_HWDEVICE_TYPE_OHCODEC` | 软件基线后小步接入;先独立验证 FFmpeg backend |
| 7 | Direct Surface decode 绕过 MPV shader/ASS/色彩管理 | 高 / 高 | FFmpeg OHCodec向 window直接 render | 优先研究 OHNativeBuffer GPU import;direct surface仅作验证路径 |
| 8 | 4K/10-bit/HDR 能力因机型/驱动不同,SDK存在不等于设备支持 | 高 / 高 | capability API要求运行时查询 | 保存设备矩阵、profile/level/format/extension;不做泛化承诺 |
| 9 | GPL/LGPL 与静态链接/FFmpeg选项污染商业发布 | 中高 / 极高 | MPV默认 GPLv2+;开关不自动产生 LGPL授权 | 默认 GPL 管理;保存完整 build manifest;发布前法务审计 |
| 10 | arm64 OHOS ABI/符号/打包规则与普通 Linux不同 | 中 / 高 | target 是 `aarch64-linux-ohos`,系统库命名含 `.z.so` | 小探针→依赖→libmpv;检查 SONAME、NEEDED、符号可见性和 HAP 装载 |
## 其他风险
- 字体发现和字体包体积影响 ASS 首帧及中日韩字幕正确性。
- 网络切换、代理、IPv6、慢速 Range、HLS discontinuity 对稳定性影响大。
- MPV master 与 FFmpeg master 同时变化,需锁 commit 并记录组合。
- API 24 SDK header含更低 API接口,但实际目标机系统版本仍要匹配。
- Vulkan headers/loader存在不代表驱动质量和 HDR swapchain 可用。
## 决策门禁
| 决策 | 必需证据 |
|---|---|
| 进入 Phase 1 | Phase 0 文档完成;origin/upstream/branch明确;依赖与 license 策略确认 |
| 开始 Playback Lab | libmpv arm64真实产物及 build report |
| 启用 OHCodec | software decode 基线 + 官方 FFmpeg OHCodec 独立验证 |
| 宣称 low-copy/zero-copy | 帧路径、buffer ownership、GPU import、同步与 copy 测量证据 |
| 采用 Vulkan | GLES benchmark证明瓶颈且 Vulkan原型在目标机显著改善 |
| 商业发布 | 实际链接清单、许可证正文、源码义务、法务审核全部完成 |
-20
View File
@@ -1,20 +0,0 @@
# SDK packaging
使用 `scripts/package-velix-core-sdk.sh` 生成 staging 目录。脚本按 profile 选择已构建的 `libmpv`,编译 binding,复制公开 headers、运行库、许可证,并生成包含 API 版本、commit、profile、尺寸和 SHA-256 的 `metadata/build-info.json`。
支持 profile:
```sh
./scripts/package-velix-core-sdk.sh --profile default
./scripts/package-velix-core-sdk.sh --profile software
./scripts/package-velix-core-sdk.sh --profile ffmpeg-harmony
```
ABI 检查重点:
- `libvelix_core_binding.so` 只导出 3 个 `velix_core_get_*` ABI 符号;NAPI 模块描述由系统注册段加载,不把内部 C++ 符号作为 ABI;
- `libmpv.so.2` 保持自身运行时导出,不将其头文件类型传给主 App;
- 所有动态依赖必须存在于 `lib/arm64-v8a` 或 OpenHarmony 系统库中;
- staging 与构建目录不应进入 Git,发布包必须保留 LGPL/GPL/第三方声明。
验证:`./scripts/verify-velix-core-sdk.sh`。
-15
View File
@@ -1,15 +0,0 @@
# Seek Pipeline
```text
seek request
-> mpv seek command
-> SEEK event / demux recovery
-> VIDEO_RECONFIG
-> PLAYBACK_RESTART
-> first video frame
-> first render / EGL swap
```
`seekCommandLatencyMs` 是 API request 到 command 提交的耗时;`seekRecoveryMs` 是 request 到 `PLAYBACK_RESTART` 的耗时;`seekFirstVisibleFrameMs` 是 request 到首个成功 swap 的耗时。若 MPV 没有产生对应事件,字段为 -1。每次新 seek 会重置上一次 seek 字段,不把旧会话值带入新会话。
模拟器 runner 使用 `+10s` 相对 seek,50 次中 49 次产生 recovery,1 次返回稳定 `SEEK / error running command`;该失败保留在 JSON,不被隐藏。精确/百分比 seek 入口仍由公共 API 暴露,但短样片不足以让输入框自动化可靠覆盖,需更长受控 fixture。
-19
View File
@@ -1,19 +0,0 @@
# Startup Pipeline
```text
session setup
-> mpv_create / mpv_initialize
-> load command
-> START_FILE
-> network open / first byte (network only)
-> FILE_LOADED (demux ready)
-> VIDEO_RECONFIG / AUDIO_RECONFIG
-> render context ready
-> first video frame available
-> first render begin/end
-> first EGL swap (startupTotalMs)
```
时间线在 core 内以 `clock_gettime(CLOCK_MONOTONIC)` 记录,所有阶段只写首次发生值。文件输入没有网络阶段时保持 -1;网络阶段只在真实 MPV log/event 证据出现时写入。`startupTotalMs` 在首个成功 `eglSwapBuffers` 写入,`firstFrameMs` 保留外壳兼容字段并与首 swap 对齐。
音频 renderer start 和首 callback 由 OHAudio stats poll 观察;这两个字段可能晚于真实 callback,因此用于阶段诊断,不替代设备级 audio trace。
-7
View File
@@ -1,7 +0,0 @@
# 字幕管线
内部字幕和外部字幕统一由 mpv/libass 管线处理。外部源通过 sub-add 加入当前 session,可选中或移除,不销毁播放 session;sub-remove 仅移除指定轨道。language/title 通过 sub-add 元数据传入,forced/default 映射为 mpv 的同名 track flags。
目标格式:SRT、ASS/SSA、WebVTT;ASS 默认保留原样式与 libass override 语义。上层可开关、选择字幕、调整 delay、scale、position、override(具体控件按实际可用属性开放)。
当前 OHOS 构建未启用 iconv,因此只接受 UTF-8/UTF8;GBK、Big5、Shift-JIS 等输入明确返回稳定 SUBTITLE 错误,不静默走 Latin-1 fallback。启用并验证 iconv 后再扩展编码白名单。Playback Lab 同时提供网络 URL 与本地 file 路径入口。
-5
View File
@@ -1,5 +0,0 @@
# Track 模型
上层只接收固定 Track 字段:id、type、title、language、codec、default、forced、selected、external。mpv_node 只在实现层解析,不跨 NAPI/C API 边界。
音频通过 aid 选择,字幕通过 sid 选择或设置为 no。语言偏好使用 mpv 的 alang/slang,可接受 zh、zho、chi、en、eng、ja、jpn 等媒体元数据值。不可用的字段保持空值或 false。
-10
View File
@@ -1,10 +0,0 @@
# Velix 主 App 集成
主 App 工程为 `../Velix`,当前是独立的 Stage Model/API 24 工程,目录本身没有 Git 元数据。集成脚本提供两种依赖模式:
- `scripts/sync-velix-core-sdk.sh --local`:从相邻 Core 重新构建并 staging,适合本地开发。
- `scripts/sync-velix-core-sdk.sh --release --sdk PATH`:消费已经发布的 SDK 目录,适合 release 验证。
`scripts/build-velix-app.sh` 会先同步 SDK,再执行 `devecocli build`。主 App 只将 `libvelix_core_binding.so`、`libmpv.so.2` 与 `libc++_shared.so` 放进 `entry/libs/arm64-v8a`,不引入 Core/FFmpeg 源码。
ArkTS 入口为 `domain/playback/PlaybackEngine.ets` → `MpvPlaybackEngine.ets` → native binding;页面为 `pages/PlayerPage.ets`。XComponent 的 `libraryname` 为 `velix_core_binding`,Surface 回调由 Core binding 注册,页面不重复实现 EGL renderer。
-290
View File
@@ -1,290 +0,0 @@
Introduction
============
This file lists all changes that can cause compatibility issues when using
mpv through the client API (libmpv and ``client.h``). Since the client API
interfaces to input handling (commands, properties) as well as command line
options, you should also look at ``interface-changes.rst``.
Normally, changes to the C API that are incompatible to previous iterations
receive a major version bump (i.e. the first version number is increased),
while C API additions bump the minor version (i.e. the second number is
increased). Changes to properties/commands/options may also lead to a minor
version bump, in particular if they are incompatible.
The version number is the same as used for MPV_CLIENT_API_VERSION (see
``client.h`` how to convert between major/minor version numbers and the flat
32 bit integer).
Also, read the section ``Compatibility`` in ``client.h``, and compatibility.rst.
Options, commands, properties
=============================
Changes to these are not listed here, but in ``interface-changes.rst``. (Before
client API version 1.17, they were listed here partially.)
This listing includes changes to the bare C API and behavior only, not what
you can access with them.
API changes
===========
::
--- mpv 0.40.0 ---
2.5 - Deprecate MPV_RENDER_PARAM_AMBIENT_LIGHT. no replacement.
--- mpv 0.39.0 ---
2.4 - mpv_render_param with the MPV_RENDER_PARAM_ICC_PROFILE argument no
longer has incorrect assumptions about memory allocation and can be
correctly used.
--- mpv 0.38.0 ---
2.3 - partially revert the changes from API version 1.27, remove libmpv as
the default VO and move it to the bottom of the auto-probing order.
This restores the prior behavior on all platforms other than macOS,
but still auto selects libmpv/cocoa-cb on macOS if it was built with
support for cocoa-cb.
--- mpv 0.37.0 ---
2.2 - add mpv_time_ns()
--- mpv 0.36.0 ---
2.1 - add mpv_del_property()
--- mpv 0.35.0 ---
2.0 - remove headers/functions of the obsolete opengl_cb API
- remove mpv_opengl_init_params.extra_exts field
- remove deprecated mpv_detach_destroy. Use mpv_destroy instead.
- remove obsolete mpv_suspend and mpv_resume
- remove deprecated SCRIPT_INPUT_DISPATCH, PAUSE and UNPAUSE, TRACKS_CHANGED
TRACK_SWITCHED, METADATA_UPDATE, CHAPTER_CHANGE events
--- mpv 0.33.0 ---
1.109 - add MPV_RENDER_API_TYPE_SW and related (software rendering API)
- deactivate the opengl_cb API (always fails to initialize now)
The opengl_cb API was deprecated over 2 years ago. Use the render API
instead.
1.108 - Deprecate MPV_EVENT_IDLE
- add mpv_event_start_file
- add the following fields to mpv_event_end_file: playlist_entry_id,
playlist_insert_id, playlist_insert_num_entries
- add mpv_event_to_node()
- add mpv_client_id()
1.107 - Remove the deprecated qthelper.hpp. This was obviously not part of the
libmpv API, only an "additionally" provided helper, thus this is not
considered an API change. If you are maintaining a project that relies
on this header, you can simply download this file and adjust the
include statement to use it instead:
https://raw.githubusercontent.com/mpv-player/mpv/v0.32.0/libmpv/qthelper.hpp
It is a good idea to write better wrappers for your use, though.
--- mpv 0.31.0 ---
1.107 - Deprecate MPV_EVENT_TICK
--- mpv 0.30.0 ---
1.106 - Add cancel_fn to mpv_stream_cb_info
1.105 - Fix deadlock problems with MPV_RENDER_PARAM_ADVANCED_CONTROL and if
the "vd-lavc-dr" option is enabled (which it is by default).
There were no actual API changes.
API users on older API versions and mpv releases should set
"vd-lavc-dr" to "no" to avoid these issues.
API users must still adhere to the tricky rules documented in render.h
to avoid other deadlocks.
1.104 - Deprecate struct mpv_opengl_drm_params. Replaced by mpv_opengl_drm_params_v2
- Deprecate MPV_RENDER_PARAM_DRM_DISPLAY. Replaced by MPV_RENDER_PARAM_DRM_DISPLAY_V2.
1.103 - redo handling of async commands
- add mpv_event_command and make it possible to return values from
commands issued with mpv_command_async() or mpv_command_node_async()
- add mpv_abort_async_command()
1.102 - rename struct mpv_opengl_drm_osd_size to mpv_opengl_drm_draw_surface_size
- rename MPV_RENDER_PARAM_DRM_OSD_SIZE to MPV_RENDER_PARAM_DRM_DRAW_SURFACE_SIZE
--- mpv 0.29.0 ---
1.101 - add MPV_RENDER_PARAM_ADVANCED_CONTROL and related API
- add MPV_RENDER_PARAM_NEXT_FRAME_INFO and related symbols
- add MPV_RENDER_PARAM_BLOCK_FOR_TARGET_TIME
- add MPV_RENDER_PARAM_SKIP_RENDERING
- add mpv_render_context_get_info()
1.100 - bump API number to avoid confusion with mpv release versions
- actually apply the GL_MP_MPGetNativeDisplay change for the new render
API. This also means compatibility for anything but x11 and wayland
through the old opengl-cb GL_MP_MPGetNativeDisplay method is now
unsupported.
- deprecate mpv_get_wakeup_pipe(). It's complex, but easy to replace
using normal API (just set a wakeup callback to a function which
writes to a pipe).
- add a 1st class hook API, which replaces the hacky mpv_command()
based one. The old API is deprecated and will be removed soon. The
old API was never meant to be stable, while the new API is.
1.29 - the behavior of mpv_terminate_destroy() and mpv_detach_destroy()
changes subtly (see documentation in the header file). In particular,
mpv_detach_destroy() will not leave the player running in all
situations anymore (it gets closer to refcounting).
- rename mpv_detach_destroy() to mpv_destroy() (the old function will
remain valid as deprecated alias)
- add mpv_create_weak_client(), which makes use of above changes
- MPV_EVENT_SHUTDOWN is now returned exactly once if a mpv_handle
should terminate, instead of spamming the event queue with this event
1.28 - deprecate the render opengl_cb API, and replace it with render.h
and render_gl.h. The goal is allowing support for APIs other than
OpenGL. The old API is emulated with the new API.
Likewise, the "opengl-cb" VO is renamed to "libmpv".
mpv_get_sub_api() is deprecated along the opengl_cb API.
The new API is relatively similar, but not the same. The rough
equivalents are:
mpv_opengl_cb_init_gl => mpv_render_context_create
mpv_opengl_cb_set_update_callback => mpv_render_context_set_update_callback
mpv_opengl_cb_draw => mpv_render_context_render
mpv_opengl_cb_report_flip => mpv_render_context_report_swap
mpv_opengl_cb_uninit_gl => mpv_render_context_free
The VO opengl-cb is also renamed to "libmpv".
Also, the GL_MP_MPGetNativeDisplay pseudo extension is not used by the
render API anymore, and the old opengl-cb API only handles the "x11"
and "wl" names anymore. Support for everything else has been removed.
The new render API uses proper API parameters, e.g. for X11 you pass
MPV_RENDER_PARAM_X11_DISPLAY directly.
- deprecate the qthelper.hpp header file. This provided some C++ helper
utility functions for Qt with use of libmpv. There is no reason to
keep this in the mpv git repository, nor to make it part of the libmpv
API. If you're using this header, you can safely copy it into your
project - it uses only libmpv public API. Alternatively, it could be
maintained in a separate repository by interested parties.
1.27 - make opengl-cb the default VO. This causes a subtle behavior change
if the API user called mpv_opengl_cb_init_gl(), but does not set
the "vo" option. Before, it would still have used another VO (like
on the CLI, e.g. vo=gpu). Now it'll behave as if vo=opengl-cb was
used.
--- mpv 0.28.0 ---
1.26 - remove glMPGetNativeDisplay("drm") support
- add mpv_opengl_cb_window_pos and mpv_opengl_cb_drm_params and
support via glMPGetNativeDisplay() for using it
- make --stop-playback-on-init-failure=no the default in libmpv (just
like in mpv CLI)
--- mpv 0.27.0 ---
1.25 - remove setting "no-" options via mpv_set_option*(). (See corresponding
deprecation in 0.23.0.)
--- mpv 0.25.0 ---
1.24 - add a MPV_ENABLE_DEPRECATED preprocessor symbol, which can be defined
by the user to exclude deprecated API symbols from the C headers
--- mpv 0.23.0 ---
1.24 - the deprecated mpv_suspend() and mpv_resume() APIs now do nothing.
--- mpv 0.22.0 ---
1.23 - deprecate setting "no-" options via mpv_set_option*(). For example,
instead of "no-video=" you should set "video=no".
- do not override the SIGPIPE signal handler anymore. This was done as
workaround for the FFmpeg TLS code, which has been fixed long ago.
- deprecate mpv_suspend() and mpv_resume(). They will be stubbed out
in mpv 0.23.0.
- make mpv_set_property() work to some degree before mpv_initialize().
It can now be used instead of mpv_set_option().
- semi-deprecate mpv_set_option()/mpv_set_option_string(). You should
use mpv_set_property() instead. There are some deprecated properties
which conflict with some options (see client.h remarks on
mpv_set_option()), for which mpv_set_option() might still be required.
In future mpv releases, the conflicting deprecated options/properties
will be removed, and mpv_set_option() will internally translate API
calls to mpv_set_property().
- qthelper.hpp: deprecate get_property_variant, set_property_variant,
set_option_variant, command_variant, and replace them with
get_property, set_property, command.
--- mpv 0.19.0 ---
1.22 - add stream_cb API for custom protocols
--- mpv 0.18.1 ---
---- - remove "status" log level from mpv_request_log_messages() docs. This
is 100% equivalent to "v". The behavior is still the same, thus no
actual API change.
--- mpv 0.18.0 ---
1.21 - mpv_set_property() changes behavior with MPV_FORMAT_NODE. Before this
change it rejected mpv_nodes with format==MPV_FORMAT_STRING if the
property was not a string or did not have special mechanisms in place
the function failed. Now it always invokes the option string parser,
and mpv_node with a basic data type works exactly as if the function
is invoked with that type directly. This new behavior is equivalent
to mpv_set_option().
This also affects the mp.set_property_native() Lua function.
- generally, setting choice options/properties with "yes"/"no" options
can now be set as MPV_FORMAT_FLAG
- reading a choice property as MPV_FORMAT_NODE will now return a
MPV_FORMAT_FLAG value if the choice is "yes" (true) or "no" (false)
This implicitly affects Lua and JSON IPC interfaces as well.
- big changes to vo-cmdline on vo_opengl and vo_opengl_hq (but not
vo_opengl_cb): options are now normally not reset, but applied on top
of the current options. The special undocumented value "-" still
works, but now resets all options to before any vo-cmdline command
has been called.
--- mpv 0.12.0 ---
1.20 - deprecate "GL_MP_D3D_interfaces"/"glMPGetD3DInterface", and introduce
"GL_MP_MPGetNativeDisplay"/"glMPGetNativeDisplay" (this is a
backwards-compatible rename)
--- mpv 0.11.0 ---
--- mpv 0.10.0 ---
1.19 - add "GL_MP_D3D_interfaces" pseudo extension to make it possible to
use DXVA2 in OpenGL fullscreen mode in some situations
- mpv_request_log_messages() now accepts "terminal-default" as parameter
1.18 - add MPV_END_FILE_REASON_REDIRECT, and change behavior of
MPV_EVENT_END_FILE accordingly
- a bunch of interface-changes.rst changes
1.17 - mpv_initialize() now blocks SIGPIPE (details see client.h)
--- mpv 0.9.0 ---
1.16 - add mpv_opengl_cb_report_flip()
- introduce mpv_opengl_cb_draw() and deprecate mpv_opengl_cb_render()
- add MPV_FORMAT_BYTE_ARRAY
1.15 - mpv_initialize() will now load config files. This requires setting
the "config" and "config-dir" options. In particular, it will load
mpv.conf.
- minor backwards-compatible change to the "seek" and "screenshot"
commands (new flag syntax, old additional args deprecated)
--- mpv 0.8.0 ---
1.14 - add mpv_wait_async_requests()
- the --msg-level option changes its native type from a flat string to
a key-value list (setting/reading the option as string still works)
1.13 - add MPV_EVENT_QUEUE_OVERFLOW
1.12 - add class Handle to qthelper.hpp
- improve opengl_cb.h API uninitialization behavior, and fix the qml
example
- add mpv_create_client() function
1.11 - add OpenGL rendering interop API - allows an application to combine
its own and mpv's OpenGL rendering
Warning: this API is not stable yet - anything in opengl_cb.h might
be changed in completely incompatible ways in minor API bumps
--- mpv 0.7.0 ---
1.10 - deprecate/disable everything directly related to script_dispatch
(most likely affects nobody)
1.9 - add enum mpv_end_file_reason for mpv_event_end_file.reason
- add MPV_END_FILE_REASON_ERROR and the mpv_event_end_file.error field
for slightly better error reporting on playback failure
- add --stop-playback-on-init-failure option, and make it the default
behavior for libmpv only
- add qthelper.hpp set_option_variant()
- mark the following events as deprecated:
MPV_EVENT_TRACKS_CHANGED
MPV_EVENT_TRACK_SWITCHED
MPV_EVENT_PAUSE
MPV_EVENT_UNPAUSE
MPV_EVENT_METADATA_UPDATE
MPV_EVENT_CHAPTER_CHANGE
They are handled better with mpv_observe_property() as mentioned in
the documentation comments. They are not removed and still work.
1.8 - add qthelper.hpp
1.7 - add mpv_command_node(), mpv_command_node_async()
1.6 - modify "core-idle" property behavior
- MPV_EVENT_LOG_MESSAGE now always sends complete lines
- introduce numeric log levels (mpv_log_level)
--- mpv 0.6.0 ---
1.5 - change in X11 and "--wid" behavior again. The previous change didn't
work as expected, and now the behavior can be explicitly controlled
with the "input-x11-keyboard" option. This is only a temporary
measure until XEmbed is implemented and confirmed working.
Note: in 1.6, "input-x11-keyboard" was renamed to "input-vo-keyboard",
although the old option name still works.
1.4 - subtle change in X11 and "--wid" behavior
(this change was added to 0.5.2, and broke some things, see #1090)
--- mpv 0.5.0 ---
1.3 - add MPV_MAKE_VERSION()
1.2 - remove "stream-time-pos" property (no replacement)
1.1 - remap dvdnav:// to dvd://
- add "--cache-file", "--cache-file-size"
- add "--colormatrix-primaries" (and property)
- add "primaries" sub-field to image format properties
- add "playback-time" property
- extend the "--start" option; a leading "+", which was previously
insignificant is now significant
- add "cache-free" and "cache-used" properties
- macOS: the "coreaudio" AO spdif code is split into a separate AO
--- mpv 0.4.0 ---
1.0 - the API is declared stable
+139
View File
@@ -0,0 +1,139 @@
Coding style
============
mpv uses C99 with K&R formatting, with some exceptions.
General formatting
------------------
- Use the K&R indent style.
- Use 4 spaces of indentation, never use tabs (except in Makefiles).
- Add a single space between keywords and binary operators. There are some other
cases where spaces should be added. Example:
```C
if ((a * b) > c) {
// code
some_function(a, b, c);
}
```
- Break lines on 80 columns. There is a hard limit of 85 columns. You may ignore
this limit if there's a strong case that not breaking the line will increase
readability. Going over 85 columns might provoke endless discussions about
whether such a limit is needed or not, so avoid it.
- If the body of an if/for/while statement has more than 1 physical lines, then
always add braces, even if they're technically redundant.
Bad:
```C
if (a)
// do something if b
if (b)
do_something();
```
Good:
```C
if (a) {
// do something if b
if (b)
do_something();
}
```
- If the body of an if statement uses braces, the else branch should also
use braces (and reverse).
Example:
```C
if (a) {
// do something
something();
something_else();
} else {
one_line();
}
```
- If an if condition spans multiple physical lines, then put the opening brace
for the if body on the next physical line. (Also, preferably always add a
brace, even if technically none is needed.)
Example:
```C
if (very_long_condition_a &&
very_long_condition_b)
{
code();
}
```
- Remove any trailing whitespace.
- If the file you're editing uses formatting different from from what is
described here, it's probably an old file from times when nobody followed a
consistent style. You're free to use the existing style, or the new style, or
to send a patch to reformat the file to the new style before making functional
changes.
General coding
--------------
- Use C99. Also freely make use of C99 features if it's appropriate, such as
stdbool.h.
- Don't use GNU-only features. In some cases they may be warranted, if they
are optional (such as attributes enabling printf-like format string checks).
But in general, standard C99 should be used.
- The same applies to libc functions. We have to be Windows-compatible too. Use
functions guaranteed by C99 or POSIX only, unless your use is guarded by a
configure check.
- Prefer fusing declaration and initialization, rather than putting declarations
on the top of a block. Obvious data flow is more important than avoiding
mixing declarations and statements, which is just a C90 artifact.
- tech-overview.txt might help to get an overview how mpv is structured.
- If you add features that require intrusive changes, discuss them on the dev
channel first. There might be a better way to add a feature and it can avoid
wasted work.
Sending patches
---------------
- Make a github pull request, or send a link to a plaintext patch created with
``git format-patch``. diffs posted as pastebins (especially if the http link
returns HTML) just cause extra work for everyone, because they lack commit
message and authorship information.
- Write informative commit messages. Use present tense to describe the
situation with the patch applied, and past tense for the situation before
the change.
- The subject line (the first line in a commit message) should contain a
prefix identifying the sub system, followed by a short description what
impact this commit has. This subject line shouldn't be longer than 72
characters, because it messes up the output of many git tools otherwise.
For example, you fixed a crash in af_volume.c:
Bad: ``fixed the bug (wtf?)``
Good: ``af_volume: fix crash due to null pointer access``
Having a prefix gives context, and is especially useful when trying to find
a specific change by looking at the history, or when running ``git blame``.
- The body of the commit message (everything else after the subject line) should
be as informative as possible and contain everything that isn't obvious. Don't
hesitate to dump as much information as you can - it doesn't cost you
anything. Put some effort into it. If someone finds a bug months or years
later, and finds that it's caused by your commit (even though your commit was
supposed to fix another bug), it would be bad if there wasn't enough
information to test the original bug. The old bug might be reintroduced while
fixing the new bug.
The commit message should be wrapped on 72 characters per line, because git
tools usually do not break text automatically. On the other hand, you do not
need to break text that would be unnatural to break (like data for test cases,
or long URLs).
Important: put an empty line between the subject line and the commit message.
If this is missing, it will break display in common git tools.
- Try to separate cosmetic and functional changes. It's ok to make a few
additional cosmetic changes in the same file you're working on, but don't do
something like reformatting a whole file and hiding an actual functional
change in the same commit.
-195
View File
@@ -1,195 +0,0 @@
CLI and API compatibility policy
================================
Human users and API users rely on the mpv/libmpv/scripting/IPC interface not
breaking their use case. On the other hand, active development occasionally
requires breaking things, such as removing old options or changing options in
a way that may break.
This document lists rules when, what, and how incompatible changes can be made.
It's interesting both for mpv developers, who want to change user-visible parts,
and mpv users, who want to know what is guaranteed to remain stable.
Any of the rules below may be overridden by statements in more specific
documentation (for example, if the manpage says that a particular option may be
removed any time, it means that, and the option probably won't even go through
deprecation).
Additions
---------
Additions are basically always allowed. API users etc. are supposed to deal with
the possibility that for example new API functions are added. Some parts of the
API may document how they are extended.
Options, commands, properties, events, hooks (command interface)
----------------------------------------------------------------
All of these are important for interfacing both with end users and API users
(which include Lua scripts, libmpv, and the JSON IPC). As such, they constitute
a large part of the user interface and APIs.
Certain options and commands may have documented default values. These default
values are considered a part of the API since scripts may already rely on these
documented behaviors. Changing these defaults are considered incompatible
changes and must be documented. Undocumented default values do not subject to
this requirement, and it is recommended to discourage such usage in the related
documentations if there is a need to frequently change such defaults.
All incompatible changes to this must be declared in interface-changes.rst,
which include the types of changes, the impact of these changes, and suggested
actions to address such impact so that the incompatibility is alleviated.
(This may also list compatible additions, but it's not a requirement.)
Degrees of importance and compatibility preservation
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Critical and central parts of the command interface have the strictest
requirements. It may not be reasonable to break them, and other means to achieve
some change have to be found. For example, the "seek" command is a bit of a
mess, but since changing it would likely affect almost every user, it may be
impossible to break at least the commonly used syntax. If changed anyway, there
should be a deprecation period of at least 1 year, during which the command
still works, and possibly a warning should remain even after this.
Important/often used parts must be deprecated for at least 2 releases before
they can be broken. There must be at least 1 release where the deprecated
functionality still works, and a replacement is available (if technically
reasonable). For example, a feature deprecated in mpv 0.30.0 may be removed in
mpv 0.32.0. Minor releases do not count towards this.
Under extraordinary circumstances, such as missed incompatible changes that are
already included in a release, critical security fixes, or a severe shortage of
developer time to address the known incompatible changes, important/often used
parts may be broken immediately, but the change must be extensively documented:
all of the related documentations (including manpage, interface-changes.rst,
etc., retrospectively modified if applicable) must clearly state the following:
the fact that the change is a breaking change; the version when the breaking
change happened; and the reason, impact, and suggested remedy actions.
Less useful parts can be broken immediately, but must come with some sort of
removal warning.
Parts for debugging and testing may be removed any time, potentially even
without any sort of documentation.
Currently, the importance of a part is not documented and not even well-defined,
which is probably a mistake.
Renaming or removing options
^^^^^^^^^^^^^^^^^^^^^^^^^^^^
Typically, renaming an option can be done in a compatible way with OPT_REPLACED.
You may need to check whether the corresponding properties still work (including
messy details like observing properties for changes).
OPT_REMOVED can be used to inform the user of alternatives or reasons for the
removal, which is better than an option not found error. Likewise,
m_option.deprecation_message should be set to something helpful.
Both OPT_REPLACED and OPT_REMOVED can remain in the code for a long time, since
they're unintrusive and hopefully make incompatible changes less painful for
users.
Scripting APIs
--------------
This affects internal scripting APIs (currently Lua and JavaScript).
Vaguely the same rules as with the command interface apply. Since there is a
large number of scripts, an effort should be made to provide compatibility
for old scripts, but it does not need to be stronger than that of the command
interface.
Undocumented parts of the scripting APIs are _not_ guaranteed for compatibility.
This applies especially for internals. Languages like Lua do not have strict
access control (nor does the mpv code try to emulate any), so if a script
accesses private parts, and breaks on the next mpv release, it's not mpv's
problem.
JSON IPC
--------
The JSON IPC is a thin protocol wrapping the libmpv API and the command
interface. Compatibility-wise, it's about the same as the scripting APIs.
The JSON protocol commands should remain as compatible as possible, and it
should probably accept the current way commands are delimited (line breaks)
forever.
The protocol may accept non-standard JSON extensions, but only standard JSON
(possibly with restrictions) is guaranteed for compatibility. Clients which want
to remain compatible should not use any extensions.
CLI
---
Things such as default key bindings do not necessarily require compatibility.
However, the release notes should be extremely clear on changes to "important"
key bindings. Bindings which restore the old behavior should be added to
restore-old-bindings.conf and restore-osc-bindings.conf.
Some option parsing is CLI-only and not available from libmpv or scripting. No
compatibility guarantees come with them. However, the rules which mpv uses to
distinguish between options and filenames must remain consistent (if the
non-deprecated options syntax is used).
Terminal and log output
-----------------------
There are no compatibility guarantees for the terminal output, or the text
logged via ``MPV_EVENT_LOG_MESSAGE`` and similar APIs. In particular, scripts
invoking mpv CLI are extremely discouraged from trying to parse text output,
and should use other mechanisms such as the JSON IPC.
Protocols, filters, demuxers, etc.
----------------------------------
Which of these are present is generally not guaranteed, and can even depend
on compile time settings.
The filter list and their sub-options are considered part of the
command-interface.
libmpv C API
------------
The libmpv client API (such as ``<mpv/client.h>``) mostly gives access to
the command interface. The API itself (if looked at as a component separate
from the command interface) is intended to be extremely stable.
All API changes are documented in client-api-changes.rst.
API compatibility
^^^^^^^^^^^^^^^^^
The API is *always* compatible. Incompatible changes are only allowed on major
API version changes (see ``MPV_CLIENT_API_VERSION``). A major version change is
an extremely rare event, which means usually no API symbols are ever removed.
Essentially removing API functions by making them always return an error, or
making it do nothing is allowed in cases where it is unlikely to break most
clients, but requires a deprecation period of 2 releases. (This has happened to
``mpv_suspend()`` for example.)
API symbols can be deprecated. This should be clearly marked in the doxygen
with ``@deprecated``, and if possible, the affected API symbols should not be
visible if the API user defines ``MPV_ENABLE_DEPRECATED`` to 0.
ABI compatibility
^^^^^^^^^^^^^^^^^
The ABI must never be broken, except on major API version changes. For example,
constants don't change their values.
Structs are tricky. If a struct can be allocated by a user (such as ``mpv_node``),
no fields can be added. (Unless it's an union, and the addition does not change
the offset or alignment of any of the fields or the struct itself. This has
happened to ``mpv_node`` in the past.) If a struct is allocated by libmpv only,
new fields can be appended to the end (for example ``mpv_event``).
The ABI is only backward compatible. This means if a host application is linked
to an older libmpv, and libmpv is updated to a newer version, it will still
work (as in not causing any undefined behavior).
Forward compatibility (an application would work with an older libmpv than it
was linked to) is not required.
-284
View File
@@ -1,284 +0,0 @@
# Compiling for Windows
Compiling for Windows is supported using GNU-like compilers (GCC/Clang). Clang
is compatible with both the ``w64-windows-gnu`` [MinGW-w64](https://www.mingw-w64.org/)
and ``pc-windows-msvc`` [Windows SDK](https://developer.microsoft.com/windows/downloads/windows-sdk)
targets. It supports the production of both 32-bit and 64-bit binaries and is
suitable for building on Windows as well as cross-compiling from Linux and Cygwin.
Although it is possible to build a complete MinGW-w64 toolchain yourself, there
are build environments and scripts available to simplify the process, such as
MSYS2 on Windows or a packaged toolchain provided by your favorite Linux
distribution. Note that MinGW-w64 environments included in Linux distributions
can vary in versions. As a general guideline, mpv only supports the MinGW-w64
toolchain version included in the latest Ubuntu LTS release.
mpv employs Meson for building, and the process is the same as any standard Meson
compilation.
For the most up-to-date reference on build scripts, you can refer to
[build.yml](https://github.com/mpv-player/mpv/blob/master/.github/workflows/build.yml),
which builds and tests all supported configurations: ``MinGW-w64``, ``Windows SDK``,
and ``MSYS2`` builds.
## Cross-compilation
When cross-compiling, it is recommended to use a Meson ``--cross-file`` to set up the
cross-compiling environment. For a basic example, please refer to
[Cross-compilation](https://mesonbuild.com/Cross-compilation.html).
Alternatively, consider using [mpv-winbuild-cmake](https://github.com/shinchiro/mpv-winbuild-cmake),
which bootstraps a MinGW-w64 toolchain and builds mpv along with its dependencies.
### Example with Meson
1. Create ``cross-file.txt`` with definitions for your toolchain and target platform.
Refer to [x86_64-w64-mingw32.txt](https://mesonbuild.com/Cross-compilation.html)
as a directly usable example.
- **Important**: Beware of pkg-config usage. By default, it uses build machine
files for dependency detection, even when ``--cross-file`` is used. It must
be configured correctly. Refer to ``pkg_config_libdir`` and ``sys_root``
in the [documentation](https://mesonbuild.com/Cross-compilation.html#defining-the-environment)
for proper setup. **In this example pkg-config is not used/required.**
2. Initialize subprojects. This step is optional; other methods are also
available to provide the necessary dependencies.
``` bash
# Update the subprojects database from Meson's WrapDB.
meson wrap update-db
# Explicitly download wraps as nested projects may have older versions of them.
meson wrap install expat
meson wrap install harfbuzz
meson wrap install libpng
meson wrap install zlib
# Add wraps for mpv's required dependencies
mkdir -p subprojects
cat <<EOF > subprojects/libplacebo.wrap
[wrap-git]
url = https://github.com/haasn/libplacebo
revision = master
depth = 1
clone-recursive = true
EOF
cat <<EOF > subprojects/libass.wrap
[wrap-git]
revision = master
url = https://github.com/libass/libass
depth = 1
EOF
# For FFmpeg, use Meson's build system port; alternatively, you can compile
# the upstream version yourself. See https://trac.ffmpeg.org/wiki/CompilationGuide
cat <<EOF > subprojects/ffmpeg.wrap
[wrap-git]
url = https://gitlab.freedesktop.org/gstreamer/meson-ports/ffmpeg.git
revision = meson-8.1
depth = 1
[provide]
libavcodec = libavcodec_dep
libavdevice = libavdevice_dep
libavfilter = libavfilter_dep
libavformat = libavformat_dep
libavutil = libavutil_dep
libswresample = libswresample_dep
libswscale = libswscale_dep
EOF
```
3. Build
``` bash
meson setup -Ddefault_library=static -Dprefer_static=true \
-Dc_link_args='-static' -Dcpp_link_args='-static' \
--cross-file cross-file.txt build
ninja -C build mpv.exe mpv.com
```
This will produce fully portabiel, statically linked, ``mpv.exe`` and ``mpv.com``
binaries.
#### Building libmpv
- To also build ``libmpv``, execute the following commands:
``` bash
# For a static library
meson configure build -Dlibmpv=true -Ddefault_library=static
ninja -C build libmpv.a
# For a shared library
meson configure build -Dlibmpv=true -Ddefault_library=shared
ninja -C build libmpv-2.dll
```
- Depending on the use case, you might want to use ``-Dgpl=false`` and review
similar options for subprojects.
- If any of Meson's subprojects are not linked statically, you might need to
specify options like the following example for ffmpeg:
``-Dffmpeg:default_library=static``.
#### Enabling more mpv features
The process above produces a basic mpv build. You can enable additional features.
Check the Meson
[WrapDB packages](https://mesonbuild.com/Wrapdb-projects.html) for available
dependencies or by providing them through other means.
Enhancing mpv with more features is as simple as adding more arguments to the
Meson setup command, for example:
``` bash
-Dlua=enabled -Djavascript=enabled -Dlcms2=enabled -Dlibplacebo:lcms=enabled
```
they will be automatically downloaded and built by Meson.
## Native Compilation with Clang (Windows SDK Build)
1. Install [Build Tools for Visual Studio](https://visualstudio.microsoft.com/downloads/#build-tools-for-visual-studio-2022)
or the full [Visual Studio](https://visualstudio.microsoft.com/downloads/#visual-studio-community-2022):
- From the installer, select the necessary components:
- Clang compiler for Windows
- C++ CMake tools for Windows
- Windows SDK
- Activate the developer shell. The target and host architecture must match
(e.g. `amd64` for x86_64, `arm64` for ARM64):
```
& "<Visual Studio Path>\Common7\Tools\Launch-VsDevShell.ps1" -Arch amd64 -HostArch amd64 -SkipAutomaticLocation | Out-Null
```
2. Install the remaining build tools that are not part of the Visual Studio installer:
- [Meson](https://mesonbuild.com/Getting-meson.html), the build system.
- [NASM](https://www.nasm.us/), required to assemble the hand-written assembly
in FFmpeg, dav1d and aom. Make sure it is in `PATH`.
- [Rust](https://www.rust-lang.org/tools/install) (via `rustup`), required to
build mpv's Rust components and the `subrandr` subproject. A **nightly**
toolchain is needed, as the build relies on Cargo's unstable `-C` option:
```powershell
rustup default nightly
```
3. The following build script utilizes the Meson subprojects system to build mpv and its dependencies.
To make sure all dependency versions are up-to-date, update the subprojects database from Meson's WrapDB.
Also explicitly download several wraps as some nested projects may pull older versions of them.
```
meson wrap update-db
meson wrap install expat
meson wrap install harfbuzz
meson wrap install libpng
meson wrap install zlib
```
4. Set environment variables so Meson uses the Clang/LLVM toolchain, or pass an
equivalent `--native-file` to Meson.
```powershell
$env:CC = 'clang'
$env:CXX = 'clang++'
$env:CC_LD = 'lld-link'
$env:CXX_LD = 'lld-link'
$env:RUST_LD = 'lld-link'
$env:WINDRES = 'llvm-rc'
```
Note that some dependencies (e.g. LuaJIT) may require `sed` to configure. Fortunately, it is already bundled in
[Git for Windows](https://www.git-scm.com/download/win):
```powershell
$env:PATH += ';<Git Path>\usr\bin'
```
5. Execute [ci\build-win32.ps1](https://github.com/mpv-player/mpv/blob/master/ci/build-win32.ps1). Refer to the script for more details.
This process will produce a fully static ``mpv.exe`` and ``mpv.com``, as well as
a static ``libmpv.a``.
## Native Compilation with MSYS2
For Windows developers seeking a quick setup, MSYS2 is an effective tool for
compiling mpv natively on a Windows machine. The MSYS2 repositories provide
binary packages for most of mpv's dependencies, simplifying the process to
primarily involve building mpv itself.
### Installing MSYS2
1. Follow the installation steps from [MSYS2](https://www.msys2.org/).
2. Initiate one of the [Environments](https://www.msys2.org/docs/environments/),
with the CLANG64 (``clang64.exe``) being the recommended option.
**Note:** This environment is distinct from the MSYS2 shell that opens
automatically after the final installation dialog. You must close that
initial shell and open a new one for the appropriate environment.
### Updating MSYS2
For guidance on updating MSYS2, please refer to the official documentation:
[Updating MSYS2](https://www.msys2.org/docs/updating/).
### Installing mpv Dependencies
``` bash
# Install pacboy and git
pacman -S pactoys git
# Install MSYS2 build dependencies and a MinGW-w64 compiler
pacboy -S python pkgconf cc meson
# Install key dependencies. libass and lcms2 are also included as dependencies
# of ffmpeg.
pacboy -S ffmpeg libjpeg-turbo libplacebo luajit vulkan-headers
```
### Building mpv
To compile and install mpv, execute the following commands.
The binaries will be installed in the directory ``/$MSYSTEM_PREFIX/bin``.
```bash
# Set up the build directory with the desired configuration
meson setup build -Dlibmpv=true --prefix=$MSYSTEM_PREFIX
# Compile
meson compile -C build
# Optionally, install the compiled binaries
meson install -C build
```
## Running mpv
Depending on your build configuration, ``mpv.exe`` may rely on external
libraries. To create a portable package, you will need to include these
dependencies as well. The quickest way to determine which libraries are needed
is to run ``mpv.exe`` and note any error messages that list the required ``.dll``
files. You can find these libraries in the sysroot used for compilation, such as
``/clang64/bin/`` in MSYS2.
## Linking libmpv with MSVC Programs
Building mpv/libmpv directly with the MSVC compiler (cl.exe) is not supported
due to differences in compiler flags. Our build system is designed specifically
for GNU-like compiler drivers. However, you can still build programs in
Visual Studio and link them with libmpv.
When ``libmpv`` is built with the recommended Clang/``lld-link`` toolchain,
``lld-link`` emits an import library (``mpv.lib``) next to the DLL in the build
directory. Add that ``mpv.lib`` to your MSVC project.
To generate an import library from a DLL yourself, first create a ``.def`` file
listing its exports. There are several ways to do this. One is ``gendef``, which
writes a ``.def``, based on DLL.
```bash
gendef libmpv-2.dll
```
Then build the import library from it. ``/name`` must match the DLL filename, and
``/machine`` must match the DLL's architecture (e.g. ``x64`` or ``arm64``):
```bash
lib /def:libmpv-2.def /name:libmpv-2.dll /out:mpv.lib /machine:x64
```
**Note:** Static linking is only feasible with matching compilers. For instance,
if you build with Clang in Visual Studio, static linking is possible.
-314
View File
@@ -1,314 +0,0 @@
How to contribute
=================
General
-------
The main contact for mpv development is IRC, specifically #mpv
and #mpv-devel on Libera.chat. GitHub is used for code review and
long term discussions.
Sending patches
---------------
- Make a GitHub pull request, or send a link to a plaintext patch created with
``git format-patch``.
- Plain diffs posted as pastebins are not acceptable! (Especially if the http
link returns HTML.) They only cause extra work for everyone, because they lack
commit message and authorship information.
- Never send patches to any of the developers email addresses.
- If your changes are not supposed to be merged immediately, mark them as
"[RFC]" in the commit message or the pull request title.
- Be sure to test your changes. If you didn't, please say so in the commit
message and the pull request text.
Copyright of contributions
--------------------------
- The copyright belongs to contributors. The project is a collaborative work. By
sending your changes, you agree to license your contributions according to the
requirements of this project.
- All new code must be LGPLv2.1+ licensed, or come with the implicit agreement
that it will be relicensed to LGPLv2.1+ later (see ``Copyright`` in the
repository root directory).
- 100% compatible licenses are allowed too.
- Changes in files with more liberal licenses (such as BSD, MIT, or ISC) are
assumed to be dual-licensed under LGPLv2.1+ and the license indicated in the
file header.
- You must be either the exclusive author of the patch, or acknowledge all
authors involved in the commit message. If you take 3rd party code, authorship
and copyright must be properly acknowledged. If you're making changes on
behalf of your employer, and the employer owns the copyright, you must mention
this. If the license of the code is not LGPLv2.1+, you must mention this.
- These license statements are legally binding.
- Don't use fake names (something that looks like an actual name, and may be
someone else's name, but is not your legal name). Using a pseudonym is
allowed if it can be used to identify or contact you, even if whatever
account you used to submit the patch dies.
- Do not add your name to the license header. This convention is not used by
this project, and neither copyright law nor any of the used licenses require
it.
Write good commit messages
--------------------------
- Write informative commit messages. Use present tense to describe the
situation with the patch applied, and past tense for the situation before
the change.
- The subject line (the first line in a commit message) must contain a
prefix identifying the sub system, followed by a short description what
impact this commit has. This subject line and the commit message body
must not be longer than 72 characters per line, because it messes up the
output of many git tools otherwise.
For example, you fixed a crash in af_volume.c:
- Bad: ``fixed the bug (wtf?)``
- Good: ``af_volume: fix crash due to null pointer access``
Having a prefix gives context, and is especially useful when trying to find
a specific change by looking at the history, or when running ``git blame``.
Sample prefixes: ``vo_gpu: ...``, ``command: ...``, ``DOCS/input: ...``,
``TOOLS/osxbundle: ...``, ``osc.lua: ...``, etc. You can always check the git
log for commits which modify specific files to see which prefixes are used.
- The first word after the ``:`` is lower case.
- Don't end the subject line with a ``.``.
- Put an empty line between the subject line and the commit message.
If this is missing, it will break display in common git tools.
- The body of the commit message (everything else after the subject line) must
be as informative as possible and contain everything that isn't obvious. Don't
hesitate to dump as much information as you can - it doesn't cost you
anything. Put some effort into it. If someone finds a bug months or years
later, and finds that it's caused by your commit (even though your commit was
supposed to fix another bug), it would be bad if there wasn't enough
information to test the original bug. The old bug might be reintroduced while
fixing the new bug.
The commit message must be wrapped on 72 characters per line, because git
tools usually do not break text automatically. On the other hand, you do not
need to break text that would be unnatural to break (like data for test cases,
or long URLs).
- Another summary of good conventions: https://chris.beams.io/posts/git-commit/
Split changes into multiple commits
-----------------------------------
- Follow git good practices, and split independent changes into several commits.
It's usually OK to put them into a single pull request.
- Try to separate cosmetic and functional changes. It's ok to make a few
additional cosmetic changes in the same file you're working on. But don't do
something like reformatting a whole file, and hiding an actual functional
change in the same commit.
- Splitting changes does _not_ mean that you should make them as fine-grained
as possible. Commits should form logical steps in development. The way you
split changes is important for code review and analyzing bugs.
Always squash fixup commits when making changes to pull requests
----------------------------------------------------------------
- If you make fixup commits to your pull request, you should generally squash
them with "git rebase -i". We prefer to have pull requests in a merge
ready state.
- We don't squash-merge (nor do we use GitHub's feature that does this) because
pull requests with multiple commits are perfectly legitimate, and the only
thing that makes sense in non-trivial cases.
- With complex pull requests, it *may* make sense to keep them separate, but
they should be clearly marked as such. Reviewing commits is generally easier
with fixups squashed.
- Reviewers are encouraged to look at individual commits instead of GitHub's
"changes from all commits" view (which just encourages bad git and review
practices).
Touching user-visible parts may require updating the mpv docs
-------------------------------------------------------------
- Most user-visible things are normally documented in DOCS/man/. If your commit
touches documented behavior, list of sub-options, etc., you need to adjust the
documentation.
- These changes usually go into the same commit that changes the code.
- Changes to command line options (addition/modification/removal) must be
documented in options.rst.
- Changes to input properties or input commands must be documented in input.rst.
- Changes to the libmpv API must be reflected in the libmpv's headers doxygen,
and in client-api-changes.rst.
Interface change policy
-----------------------
- All incompatible changes to the user interface (options, properties, commands)
must be documented by making a new text file with a txt extension containing a
small note in the DOCS/interface-changes directory.
- The name of the file should be brief and related to the commit that makes the
change. The content of the file should begin with the type of the change
(add, remove, deprecate, change, rename, etc.). If the file contains multiple
changes, the change which causes the most serious compatibility issues should
be placed first.
- Grouping multiple related changes in the same file is also OK. Just be sure to
put each separate change on a different line.
- Documenting additions in DOCS/interface-changes is optional but encouraged.
- interface-changes.rst is never directly updated except when making new major
releases.
- See DOCS/interface-changes/example.txt for an example.
Code formatting
---------------
mpv uses C11 with K&R formatting, with some exceptions.
- Use the K&R indent style.
- Use 4 spaces of indentation, never use tabs (except in Makefiles).
- Add a single space between keywords and binary operators. There are some other
cases where spaces must be added. Example:
```C
if ((a * b) > c) {
// code
some_function(a, b, c);
}
```
- Break lines on 80 columns. There is a hard limit of 100 columns. You may ignore
this limit if there's a strong case that not breaking the line will increase
readability. Going over 100 columns might provoke endless discussions about
whether such a limit is needed or not, so avoid it.
- If the body of an if/for/while statement has more than 1 physical lines, then
always add braces, even if they're technically redundant.
Bad:
```C
if (a)
// do something if b
if (b)
do_something();
```
Good:
```C
if (a) {
// do something if b
if (b)
do_something();
}
```
- If the if has an else branch, both branches must use braces, even if they're
technically redundant.
Example:
```C
if (a) {
one_line();
} else {
one_other_line();
}
```
- If an if condition spans multiple physical lines, then put the opening brace
for the if body on the next physical line. (Also, preferably always add a
brace, even if technically none is needed.)
Example:
```C
if (very_long_condition_a &&
very_long_condition_b)
{
code();
} else {
...
}
```
(If the if body is simple enough, this rule can be skipped.)
- Remove any trailing whitespace.
- Do not make stray whitespaces changes.
Header #include statement order
-------------------------------
The order of ``#include`` statements in the source code is not very consistent.
New code must follow the following conventions:
- Put standard includes (``#include <stdlib.h>`` etc.) on the top,
- then after a blank line, add library includes (``#include <zlib.h>`` etc.)
- then after a blank line, add internal includes (``#include "player/core.h"``)
- sort them alphabetically within these sections
General coding
--------------
- Use C11. Also freely make use of C11 features if it's appropriate, but do not
use VLA and complex number types.
- Don't use non-standard language (such as GNU C-only features). In some cases
they may be warranted, if they are optional (such as attributes enabling
printf-like format string checks). "#pragma once" is allowed as an exception.
But in general, standard C11 must be used.
- The same applies to libc functions. We have to be Windows-compatible too. Use
functions guaranteed by C11 or POSIX only, unless your use is guarded by a
configure check. Be mindful of MinGW-specifics since C11 support is not always
guaranteed.
- Prefer fusing declaration and initialization, rather than putting declarations
on the top of a block. Obvious data flow is more important than avoiding
mixing declarations and statements, which is just a C90 artifact.
- If you add features that require intrusive changes, discuss them on the dev
channel first. There might be a better way to add a feature and it can avoid
wasted work.
- Newly added code for any Apple Platform (macOS, iOS, etc), that uses Cocoa or
any other object-oriented Framework or API, should be written in Swift instead
of Objective-C. This also includes complete rewrites/refactors of existing
code. Plain C API usages may stay as C, but can be written in Swift. Existing
Objective-C code can stay as is, though Swift rewrites are welcome.
AI-assisted Contributions
-------------------------
If you have used an AI/LLM for your contribution, you must disclose this in the
PR description. Such contributions are not forbidden, but the submitter takes
full responsibility for the code changes. This includes having full
understanding of the code (knowing what was changed and why) and the ability to
participate in the code review process with human-written responses. This also
includes confirming that the code can be submitted under the same license as the
relevant files (usually LGPLv2 or GPLv2).
Clearly vibe-coded patches will not be considered.
Code of Conduct
---------------
Please note that this project is released with a Contributor Code of Conduct.
By participating in this project you agree to abide by its terms.
The Contributor Code of Conduct can be found here:
https://www.contributor-covenant.org/version/2/0/code_of_conduct/
Rules for git push access
-------------------------
Push access to the main git repository is handed out on an arbitrary basis. If
you got access, the following rules must be followed:
- You are expected to follow the general development rules as outlined in this
whole document.
- You must be present on the IRC dev channel when you push something.
- Anyone can push small fixes: typo corrections, small/obvious/uncontroversial
bug fixes, edits to the user documentation or code comments, and so on.
- You can freely make changes to parts of the code which you maintain. For
larger changes, it's recommended to let others review the changes first.
- You automatically maintain code if you wrote or modified most of it before
(e.g. you made larger changes to it before, did partial or full rewrites, did
major bug fixes, or you're the original author of the code). If there is more
than one maintainer, you may need to come to an agreement with the others how
to handle this to avoid conflict.
- If you make a pull requests (especially if it's to code you maintain), and you
want reviews, explicitly ping the people from which you expect reviews.
- As a maintainer, you can approve pull requests by others to "your" code.
- If you approve or merge 3rd party changes, make sure they follow the general
development rules.
- Changes to user interface and public API must always be approved by the
project leader.
- Seasoned project members are allowed to revert commits that broke the build,
or broke basic functionality in a catastrophic way, and the developer who
broke it is unavailable. (Depending on severity.)
- Adhere to the CoC.
- The project leader is not bound by these rules.
+54
View File
@@ -0,0 +1,54 @@
Cross compiling mpv to Windows is supported with MinGW-w64. This can be used to
produce both 32 bit and 64 bit executables. MinGW-w64 is available from
http://mingw-w64.sourceforge.net.
You have to run mpv's configure with these arguments:
DEST_OS=win32 TARGET=i686-w64-mingw32 ./waf configure
Using mingw-w64-cmake to setup a MinGW-w64 environment is recommended (this will
also build mpv and its dependencies): https://github.com/lachs0r/mingw-w64-cmake
Alternatively, use MXE: http://mxe.cc
With MXE, you have to modify the file settings.mk to target MinGW-w64 (even if
you compile to 32 bit).
Warning: the original MinGW (http://www.mingw.org) is unsupported.
Note that MinGW environments included in Linux distributions are often broken,
outdated and useless, and usually don't use MinGW-w64.
Example with MXE
================
# Download MXE. Note that compiling the required packages requires about 1 GB
# or more!
cd /opt
git clone https://github.com/mxe/mxe mingw
cd mingw
# Edit the MXE target, so that MinGW-w64 for 32 bit targets is built.
echo "MXE_TARGETS := i686-w64-mingw32" > settings.mk
# Build required packages. The following provide a minimum required to build
# mpv. (Not all of the following packages are strictly required.)
make gcc
make ffmpeg
make libass
make jpeg
make pthreads
# Build mpv. The target will be used to automatically select the name of the
# build tools involved (e.g. it will use i686-w64-mingw32-gcc).
git clone https://github.com/mpv-player/mpv.git
cd mpv
export PATH=/opt/mingw/usr/bin/:$PATH
DEST_OS=win32 TARGET=i686-w64-mingw32 ./waf configure
./waf build
# This should work. Note however that MXE’s ffmpeg package might be very old
# in order to avoid breaking e.g. xine-lib, so you might want to update that
# or build it manually as well.
+8 -301
View File
@@ -27,12 +27,13 @@ The general simplified syntax is::
<filename>,<start in seconds>,<length in seconds>
If the start time is omitted, 0 is used. If the length is omitted, the
estimated remaining duration of the source file is used.
estimated duration of the source file is used.
Note::
Usage of relative or absolute paths as well as any protocol prefixes may be
prevented for security reasons.
mpv can't use ordered chapter files or libquvi-resolved URLs in EDL
entries. Usage of relative or absolute paths as well as any protocol
prefixes is prevented for security reasons.
Syntax of mpv EDL files
@@ -51,8 +52,7 @@ The rest of the lines belong to one of these classes:
1) An empty or commented line. A comment starts with ``#``, which must be the
first character in the line. The rest of the line (up until the next line
break) is ignored. An empty line has 0 bytes between two line feed bytes.
2) A header entry if the line starts with ``!``.
3) A segment entry in all other cases.
2) A segment entry in all other cases.
Each segment entry consists of a list of named or unnamed parameters.
Parameters are separated with ``,``. Named parameters consist of a name,
@@ -64,8 +64,8 @@ Syntax::
segment_entry ::= <param> ( <param> ',' )*
param ::= [ <name> '=' ] ( <value> | '%' <number> '%' <valuebytes> )
The ``name`` string can consist of any characters, except ``=%,;\n!``. The
``value`` string can consist of any characters except of ``,;\n!``.
The ``name`` string can consist of any characters, except ``=%,;\n``. The
``value`` string can consist of any characters except of ``,;\n``.
The construct starting with ``%`` allows defining any value with arbitrary
contents inline, where ``number`` is an integer giving the number of bytes in
@@ -78,7 +78,7 @@ The parameter name defines the meaning of the parameter:
2) ``start``, a time value that specifies the start offset into the source file.
3) ``length``, a time value that specifies the length of the segment.
See the section below for the format of timestamps.
(Currently, time values are floating point values in seconds.)
Unnamed parameters carry implicit names. The parameter position determines
which of the parameters listed above is set. For example, the second parameter
@@ -95,299 +95,6 @@ to ``20``, ``param3`` to ``value,escaped``, ``param4`` to ``value2``.
Instead of line breaks, the character ``;`` can be used. Line feed bytes and
``;`` are treated equally.
Header entries start with ``!`` as first character after a line break. Header
entries affect all other file entries in the EDL file. Their format is highly
implementation specific. They should generally follow the file header, and come
before any file entries.
Disabling chapter generation and copying
========================================
By default, chapters from the source ranges are copied to the virtual file's
chapters. Also, a chapter is inserted after each range. This can be disabled
with the ``no_chapters`` header.
Example::
!no_chapters
MP4 DASH
========
This is a header that helps implementing DASH, although it only provides a low
level mechanism.
If this header is set, the given url designates an mp4 init fragment. It's
downloaded, and every URL in the EDL is prefixed with the init fragment on the
byte stream level. This is mostly for use by mpv's internal ytdl support. The
ytdl script will call youtube-dl, which in turn actually processes DASH
manifests. It may work only for this very specific purpose and fail to be
useful in other scenarios. It can be removed or changed in incompatible ways
at any times.
Example::
!mp4_dash,init=url
The ``url`` is encoded as parameter value as defined in the general EDL syntax.
It's expected to point to an "initialization fragment", which will be prefixed
to every entry in the EDL on the byte stream level.
The current implementation will
- ignore stream start times
- use durations as hint for seeking only
- not adjust source timestamps
- open and close segments (i.e. fragments) as needed
- not add segment boundaries as chapter points
- require full compatibility between all segments (same codec etc.)
Another header part of this mechanism is ``no_clip``. This header is similar
to ``mp4_dash``, but does not include on-demand opening/closing of segments,
and does not support init segments. It also exists solely to support internal
ytdl requirements. Using ``no_clip`` with segments is not recommended and
probably breaks. ``mp4_dash`` already implicitly does a variant of ``no_clip``.
The ``mp4_dash`` and ``no_clip`` headers are not part of the core EDL format.
They may be changed or removed at any time, depending on mpv's internal
requirements.
Separate files for tracks
=========================
The special ``new_stream`` header lets you specify separate parts and time
offsets for separate tracks. This can for example be used to source audio and
video track from separate files.
Example::
# mpv EDL v0
video.mkv
!new_stream
audio.mkv
This adds all tracks from both files to the virtual track list. Upon playback,
the tracks will be played at the same time, instead of appending them. The files
can contain more than 1 stream; the apparent effect is the same as if the second
part after the ``!new_stream`` part were in a separate ``.edl`` file and added
with ``--external-file``.
Note that all metadata between the stream sets created by ``new_stream`` is
disjoint. Global metadata is taken from the first part only.
In context of mpv, this is redundant to the ``--audio-file`` and
``--external-file`` options, but (as of this writing) has the advantage that
this will use a unified cache for all streams.
The ``new_stream`` header is not part of the core EDL format. It may be changed
or removed at any time, depending on mpv's internal requirements.
If the first ``!new_stream`` is redundant, it is ignored. This is the same
example as above::
# mpv EDL v0
!new_stream
video.mkv
!new_stream
audio.mkv
Note that ``!new_stream`` must be the first header. Whether the parser accepts
(i.e. ignores) or rejects other headers before that is implementation specific.
Track metadata
==============
The special ``track_meta`` header can set some specific metadata fields of the
current ``!new_stream`` partition. The tags are applied to all tracks within
the partition. It is not possible to set the metadata for individual tracks (the
feature was needed only for single-track media).
It provides following parameters change track metadata:
``lang``
Set the language tag.
``title``
Set the title tag.
``byterate``
Number of bytes per second this stream uses. (Purely informational.)
``program_id``
Assign this track to a program (integer, default: -1 meaning unset).
``index``
The numeric index of the track this should map to (default: -1). This is
the 0-based index of the virtual stream as seen by the player, enumerating
all audio/video/subtitle streams. If nothing matches, this is silently
discarded. The special index -1 (the default) has two meanings: if there
was a previous meta data entry (either ``!track_meta`` or ``!delay_open``
element since the last ``!new_stream``), then this element manipulates
the previous meta data entry. If there was no previous entry, a new meta
data entry that matches all streams is created.
Example::
# mpv EDL v0
!track_meta,lang=bla,title=blabla
file.mkv
!new_stream
!track_meta,title=ducks
sub.srt
If ``file.mkv`` has an audio and a video stream, both will use ``blabla`` as
title. The subtitle stream will use ``ducks`` as title.
The ``track_meta`` header is not part of the core EDL format. It may be changed
or removed at any time, depending on mpv's internal requirements.
Global metadata
===============
The special ``global_tags`` header can set metadata fields (aka tags) of the EDL
file. This metadata is supposed to be informational, much like for example ID3
tags in audio files. Due to lack of separation of different kinds of metadata it
is unspecified what names are allowed, how they are interpreted, and whether
some of them affect playback functionally. (Much of this is unfortunately
inherited from FFmpeg. Another consequence of this is that FFmpeg "normalized"
tags are recognized, or stuff like replaygain tags.)
Example::
!global_tags,title=bla,something_arbitrary=even_more_arbitrary
Any parameter names are allowed. Repeated use of this adds to the tag list. If
``!new_stream`` is used, the location doesn't matter.
May possibly be ignored in some cases, such as delayed media opening.
Delayed media opening
=====================
The special ``delay_open`` header can be used to open the media URL of the
stream only when the track is selected for the first time. This is supposed to
be an optimization to speed up opening of a remote stream if there are many
tracks for whatever reasons.
This has various tricky restrictions, and also will defer failure to open a
stream to "later". By design, it's supposed to be used for single-track streams.
Using multiple segments requires you to specify all offsets and durations (also
it was never tested whether it works at all). Interaction with ``mp4_dash`` may
be strange.
You can describe multiple sub-tracks by using multiple ``delay_open`` headers
before the same source URL. (If there are multiple sub-tracks of the same media
type, then the mapping to the real stream is probably rather arbitrary.) If the
source contains tracks not described, a warning is logged when the delayed
opening happens, and the track is hidden.
This has the following parameters:
``media_type``
Required. Must be set to ``video``, ``audio``, or ``sub``. (Other tracks in
the opened URL are ignored.)
``codec``
The mpv codec name that is expected. Although mpv tries to initialize a
decoder with it currently (and will fail track selection if it does not
initialize successfully), it is not used for decoding - decoding still uses
the information retrieved from opening the actual media information, and may
be a different codec (you should try to avoid this, of course). Defaults to
``null``.
Above also applies for similar fields such as ``w``. These fields are
mostly to help with user track pre-selection.
``flags``
A ``+`` separated list of boolean flags. Currently defined flags:
``default``
Set the default track flag.
``forced``
Set the forced track flag.
Other values are ignored after triggering a warning.
``w``, ``h``
For video codecs: expected video size. See ``codec`` for details.
``fps``
For video codecs: expected video framerate, as integer. (The rate is usually
only crudely reported, and it makes no sense to expect exact values.)
``samplerate``
For audio codecs: expected sample rate, as integer.
The ``delay_open`` header is not part of the core EDL format. It may be changed
or removed at any time, depending on mpv's internal requirements.
Timestamp format
================
Currently, time values are floating point values in seconds.
As an extension, you can set the ``timestamps=chapters`` option. If this option
is set, timestamps have to be integers, and refer to chapter numbers, starting
with 0. The default value for this parameter is ``seconds``, which means the
time is as described in the previous paragraph.
Example::
# mpv EDL v0
file.mkv,2,4,timestamps=chapters
Plays chapter 3 and ends with the start of chapter 7 (4 chapters later).
Implicit chapters
=================
mpv will add one chapter per segment entry to the virtual timeline.
By default, the chapter's titles will match the entries' filenames.
You can override set the ``title`` option to override the chapter title for
that segment.
Example::
# mpv EDL v0
cap.ts,5,240
OP.mkv,0,90,title=Show Opening
The virtual timeline will have two chapters, one called "cap.ts" from 0-240s
and a second one called "Show Opening" from 240-330s.
Entry which defines the track layout
====================================
Normally, you're supposed to put only files with compatible layouts into an EDL
file. However, at least the mpv implementation accepts entries that use
different codecs, or even have a different number of audio/video/subtitle
tracks. In this case, it's not obvious, which virtual tracks the EDL show should
expose when being played.
Currently, mpv will apply an arbitrary heuristic which tracks the EDL file
should expose.
You can set the ``layout`` option to ``this`` to make a specific entry define
the track layout.
Example::
# mpv EDL v0
file_with_2_streams.ts,5,240
file_with_5_streams.mkv,0,90,layout=this
The way the different virtual EDL tracks are associated with the per-segment
ones is highly implementation-defined, and uses a heuristic. If a segment is
missing a track, there will be a "hole", and bad behavior may result. Improving
this is subject to further development (due to being fringe cases, they don't
have a high priority).
If future versions of mpv change this again, this option may be ignored.
Syntax of EDL URIs
==================
+51 -58
View File
@@ -3,50 +3,44 @@ General usage
::
mpv infile --o=outfile [--of=outfileformat] [--ofopts=formatoptions] [--orawts] \
mpv infile -o outfile [-of outfileformat] [-ofopts formatoptions] \
[-ofps outfps | -oautofps] [-oharddup] [-ocopyts | -orawts] [-oneverdrop] \
[(any other mpv options)] \
--ovc=outvideocodec [--ovcopts=outvideocodecoptions] \
--oac=outaudiocodec [--oacopts=outaudiocodecoptions]
-ovc outvideocodec [-ovcopts outvideocodecoptions] \
-oac outaudiocodec [-oacopts outaudiocodecoptions]
Help for these options is provided if giving help as parameter, as in::
mpv --ovc=help
mpv -ovc help
The suboptions of these generally are identical to ffmpeg's (as option parsing
is simply delegated to ffmpeg). The option --ocopyts enables copying timestamps
is simply delegated to ffmpeg). The option -ocopyts enables copying timestamps
from the source as-is, instead of fixing them to match audio playback time
(note: this doesn't work with all output container formats); --orawts even turns
(note: this doesn't work with all output container formats); -orawts even turns
off discontinuity fixing.
Note that if neither --ofps nor --oautofps is specified, VFR encoding is assumed
and the time base is 24000fps. --oautofps sets --ofps to a guessed fps number
Note that if neither -ofps nor -oautofps is specified, VFR encoding is assumed
and the time base is 24000fps. -oautofps sets -ofps to a guessed fps number
from the input video. Note that not all codecs and not all formats support VFR
encoding, and some which do have bugs when a target bitrate is specified - use
--ofps or --oautofps to force CFR encoding in these cases.
-ofps or -oautofps to force CFR encoding in these cases.
Of course, the options can be stored in a profile, like this .config/mpv/mpv.conf
Of course, the options can be stored in a profile, like this .mpv/config
section::
[myencprofile]
vf-add = scale=480:-2
ovc = libx264
ovcopts-add = preset=medium
ovcopts-add = tune=fastdecode
ovcopts-add = preset=medium,tune=fastdecode
ovcopts-add = crf=23
ovcopts-add = maxrate=1500k
ovcopts-add = bufsize=1000k
ovcopts-add = rc_init_occupancy=900k
ovcopts-add = refs=2
ovcopts-add = maxrate=1500k,bufsize=1000k,rc_init_occupancy=900k,refs=2
ovcopts-add = profile=baseline
oac = aac
oacopts-add = b=96k
It's also possible to define default encoding options by putting them into
the section named ``[encoding]``.
One can then encode using this profile using the command::
mpv infile --o=outfile.mp4 --profile=myencprofile
mpv infile -o outfile.mp4 -profile myencprofile
Some example profiles are provided in a file
etc/encoding-profiles.conf; as for this, see below.
@@ -60,33 +54,33 @@ for.
Typical MPEG-4 Part 2 ("ASP", "DivX") encoding, AVI container::
mpv infile --o=outfile.avi \
--vf=fps=25 \
--ovc=mpeg4 --ovcopts=qscale=4 \
--oac=libmp3lame --oacopts=b=128k
mpv infile -o outfile.avi \
-ofps 25 \
-ovc mpeg4 -ovcopts qscale=4 \
-oac libmp3lame -oacopts ab=128k
Note: AVI does not support variable frame rate, so the fps filter must be used.
The frame rate should ideally match the input (25 for PAL, 24000/1001 or
30000/1001 for NTSC)
Note: AVI does not support variable frame rate, so -ofps must be used. The
frame rate should ideally match the input (25 for PAL, 24000/1001 or 30000/1001
for NTSC)
Typical MPEG-4 Part 10 ("AVC", "H.264") encoding, Matroska (MKV) container::
mpv infile --o=outfile.mkv \
--ovc=libx264 --ovcopts=preset=medium,crf=23,profile=baseline \
--oac=libopus --oacopts=qscale=3
mpv infile -o outfile.mkv \
-ovc libx264 -ovcopts preset=medium,crf=23,profile=baseline \
-oac libvorbis -oacopts qscale=3
Typical MPEG-4 Part 10 ("AVC", "H.264") encoding, MPEG-4 (MP4) container::
mpv infile --o=outfile.mp4 \
--ovc=libx264 --ovcopts=preset=medium,crf=23,profile=baseline \
--oac=aac --oacopts=b=128k
mpv infile -o outfile.mp4 \
-ovc libx264 -ovcopts preset=medium,crf=23,profile=baseline \
-oac aac -oacopts ab=128k
Typical VP8 encoding, WebM (restricted Matroska) container::
mpv infile -o outfile.mkv \
--of=webm \
--ovc=libvpx --ovcopts=qmin=6,b=1000000k \
--oac=libopus --oacopts=qscale=3
-of webm \
-ovc libvpx -ovcopts qmin=6,b=1000000k \
-oac libvorbis -oacopts qscale=3
Device targets
@@ -95,29 +89,28 @@ Device targets
As the options for various devices can get complex, profiles can be used.
An example profile file for encoding is provided in
etc/encoding-profiles.conf in the source tree. mpv will read any file
named ``encoding-profiles.conf`` in either the local or system
configuration directories.
etc/encoding-profiles.conf in the source tree. This file is installed and loaded
by default (if libavfilter is enabled at compilation). If you want to modify
it, you can replace and it with your own copy by doing::
Keep in mind that the default profile is the playback one. If you want to add
options that apply only in encoding mode, put them into a ``[encoding]``
section.
mkdir -p ~/.mpv
cp /etc/mpv/encoding-profiles.conf ~/.mpv/encoding-profiles.conf
Refer to the top of that file for more comments - in a nutshell, the following
options are added by it::
--profile=enc-to-dvdpal # DVD-Video PAL, use dvdauthor -v pal+4:3 -a ac3+en
--profile=enc-to-dvdntsc # DVD-Video NTSC, use dvdauthor -v ntsc+4:3 -a ac3+en
--profile=enc-to-bb-9000 # MP4 for Blackberry Bold 9000
--profile=enc-to-nok-6300 # 3GP for Nokia 6300
--profile=enc-to-psp # MP4 for PlayStation Portable
--profile=enc-to-iphone # MP4 for iPhone
--profile=enc-to-iphone-4 # MP4 for iPhone 4 (double res)
--profile=enc-to-iphone-5 # MP4 for iPhone 5 (even larger res)
-profile enc-to-dvdpal DVD-Video PAL, use dvdauthor -v pal+4:3 -a ac3+en
-profile enc-to-dvdntsc DVD-Video NTSC, use dvdauthor -v ntsc+4:3 -a ac3+en
-profile enc-to-bb-9000 MP4 for Blackberry Bold 9000
-profile enc-to-nok-6300 3GP for Nokia 6300
-profile enc-to-psp MP4 for PlayStation Portable
-profile enc-to-iphone MP4 for iPhone
-profile enc-to-iphone-4 MP4 for iPhone 4 (double res)
-profile enc-to-iphone-5 MP4 for iPhone 5 (even larger res)
You can encode using these with a command line like::
mpv infile --o=outfile.mp4 --profile=enc-to-bb-9000
mpv infile -o outfile.mp4 -profile enc-to-bb-9000
Of course, you are free to override options set by these profiles by specifying
them after the -profile option.
@@ -127,20 +120,20 @@ What works
==========
* Encoding at variable frame rate (default)
* Encoding at constant frame rate using --vf=fps=RATE
* 2-pass encoding (specify flags=+pass1 in the first pass's --ovcopts, specify
* Encoding at constant frame rate using -ofps framerate -oharddup
* 2-pass encoding (specify flags=+pass1 in the first pass's -ovcopts, specify
flags=+pass2 in the second pass)
* Hardcoding subtitles using vobsub, ass or srt subtitle rendering (just
configure mpv for the subtitles as usual)
* Hardcoding any other mpv OSD (e.g. time codes, using --osdlevel=3 and
--vf=expand=::::1)
* Hardcoding any other mpv OSD (e.g. time codes, using -osdlevel 3 and -vf
expand=::::1)
* Encoding directly from a DVD, network stream, webcam, or any other source
mpv supports
* Using x264 presets/tunings/profiles (by using profile=, tune=, preset= in the
--ovcopts)
-ovcopts)
* Deinterlacing/Inverse Telecine with any of mpv's filters for that
* Audio file converting: mpv --o=outfile.m4a infile.flac --no-video
--oac=aac --oacopts=b=320k
* Audio file converting: mpv -o outfile.mp3 infile.flac -no-video -oac
libmp3lame -oacopts ab=320k
What does not work yet
======================
File diff suppressed because it is too large. Load diff
-1
View File
@@ -1 +0,0 @@
change `ytdl_hook-all_formats` default to `yes`
@@ -1 +0,0 @@
make `--autocreate-playlist` only activate for regular files
@@ -1 +0,0 @@
make the `metadata/by-key` property writable
-1
View File
@@ -1 +0,0 @@
'command-list' now also returns a default value and available choices where applicable for each command argument.
-1
View File
@@ -1 +0,0 @@
make `mpv_command_node` accept `_name` for command name
@@ -1 +0,0 @@
add `--demuxer-hysteresis-bytes`
-3
View File
@@ -1,3 +0,0 @@
add `--disc-menu` option
add `disc-menu-active` property
add `discnav` command
-2
View File
@@ -1,2 +0,0 @@
add `--input-builtin-drag-and-drop` option
add `dropped-file` property
-2
View File
@@ -1,2 +0,0 @@
add `dvda://` protocol for DVD-Audio playback
add `--dvda-device` and `--dvda-page` options
-1
View File
@@ -1 +0,0 @@
add `vf=format=enhancement-layer` option
@@ -1 +0,0 @@
remove the deprecated `error` field from `end-file`
-1
View File
@@ -1 +0,0 @@
add `env` property that can be used to retrieve environment variables
-1
View File
@@ -1 +0,0 @@
deprecate `--foo`, instead use `--bar=foo` as a replacement
@@ -1 +0,0 @@
the `expand-properties` command modifier also applies to string lists now
-1
View File
@@ -1 +0,0 @@
`track-list/N/ff-index` may be unavailable if no known mapping exists
@@ -1 +0,0 @@
`--geometry` x/y position is now applied only on initial window creation, unless `--force-window-position=yes` is set.
-1
View File
@@ -1 +0,0 @@
add libcurl-based stream backend (http, https, ftp, ftps), with new options `--curl-enabled`, `--curl-http-version`, `--curl-max-redirects`, `--curl-max-retries`, `--curl-connect-timeout`, `--curl-buffer-size` and `--curl-max-request-size`
@@ -1 +0,0 @@
change `--hwdec-extra-frames` default to `auto`, sized from VO frame retention
@@ -1 +0,0 @@
implement `lavfi-tempo` an alternative speed changing audio filter
@@ -1 +0,0 @@
add `libplacebo-version` property
Loaded 100 of 1376 files, more files were not shown because too many files have changed in this diff. Show more