2011-01-16 18:03:08 +00:00
|
|
|
/*
|
2015-04-13 07:36:54 +00:00
|
|
|
* This file is part of mpv.
|
2011-01-16 18:03:08 +00:00
|
|
|
*
|
Relicense some non-MPlayer source files to LGPL 2.1 or later
This covers source files which were added in mplayer2 and mpv times
only, and where all code is covered by LGPL relicensing agreements.
There are probably more files to which this applies, but I'm being
conservative here.
A file named ao_sdl.c exists in MPlayer too, but the mpv one is a
complete rewrite, and was added some time after the original ao_sdl.c
was removed. The same applies to vo_sdl.c, for which the SDL2 API is
radically different in addition (MPlayer supports SDL 1.2 only).
common.c contains only code written by me. But common.h is a strange
case: although it originally was named mp_common.h and exists in MPlayer
too, by now it contains only definitions written by uau and me. The
exceptions are the CONTROL_ defines - thus not changing the license of
common.h yet.
codec_tags.c contained once large tables generated from MPlayer's
codecs.conf, but all of these tables were removed.
From demux_playlist.c I'm removing a code fragment from someone who was
not asked; this probably could be done later (see commit 15dccc37).
misc.c is a bit complicated to reason about (it was split off mplayer.c
and thus contains random functions out of this file), but actually all
functions have been added post-MPlayer. Except get_relative_time(),
which was written by uau, but looks similar to 3 different versions of
something similar in each of the Unix/win32/OSX timer source files. I'm
not sure what that means in regards to copyright, so I've just moved it
into another still-GPL source file for now.
screenshot.c once had some minor parts of MPlayer's vf_screenshot.c, but
they're all gone.
2016-01-19 17:36:06 +00:00
|
|
|
* mpv is free software; you can redistribute it and/or
|
|
|
|
* modify it under the terms of the GNU Lesser General Public
|
|
|
|
* License as published by the Free Software Foundation; either
|
|
|
|
* version 2.1 of the License, or (at your option) any later version.
|
2011-01-16 18:03:08 +00:00
|
|
|
*
|
2015-04-13 07:36:54 +00:00
|
|
|
* mpv is distributed in the hope that it will be useful,
|
2011-01-16 18:03:08 +00:00
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
Relicense some non-MPlayer source files to LGPL 2.1 or later
This covers source files which were added in mplayer2 and mpv times
only, and where all code is covered by LGPL relicensing agreements.
There are probably more files to which this applies, but I'm being
conservative here.
A file named ao_sdl.c exists in MPlayer too, but the mpv one is a
complete rewrite, and was added some time after the original ao_sdl.c
was removed. The same applies to vo_sdl.c, for which the SDL2 API is
radically different in addition (MPlayer supports SDL 1.2 only).
common.c contains only code written by me. But common.h is a strange
case: although it originally was named mp_common.h and exists in MPlayer
too, by now it contains only definitions written by uau and me. The
exceptions are the CONTROL_ defines - thus not changing the license of
common.h yet.
codec_tags.c contained once large tables generated from MPlayer's
codecs.conf, but all of these tables were removed.
From demux_playlist.c I'm removing a code fragment from someone who was
not asked; this probably could be done later (see commit 15dccc37).
misc.c is a bit complicated to reason about (it was split off mplayer.c
and thus contains random functions out of this file), but actually all
functions have been added post-MPlayer. Except get_relative_time(),
which was written by uau, but looks similar to 3 different versions of
something similar in each of the Unix/win32/OSX timer source files. I'm
not sure what that means in regards to copyright, so I've just moved it
into another still-GPL source file for now.
screenshot.c once had some minor parts of MPlayer's vf_screenshot.c, but
they're all gone.
2016-01-19 17:36:06 +00:00
|
|
|
* GNU Lesser General Public License for more details.
|
2011-01-16 18:03:08 +00:00
|
|
|
*
|
Relicense some non-MPlayer source files to LGPL 2.1 or later
This covers source files which were added in mplayer2 and mpv times
only, and where all code is covered by LGPL relicensing agreements.
There are probably more files to which this applies, but I'm being
conservative here.
A file named ao_sdl.c exists in MPlayer too, but the mpv one is a
complete rewrite, and was added some time after the original ao_sdl.c
was removed. The same applies to vo_sdl.c, for which the SDL2 API is
radically different in addition (MPlayer supports SDL 1.2 only).
common.c contains only code written by me. But common.h is a strange
case: although it originally was named mp_common.h and exists in MPlayer
too, by now it contains only definitions written by uau and me. The
exceptions are the CONTROL_ defines - thus not changing the license of
common.h yet.
codec_tags.c contained once large tables generated from MPlayer's
codecs.conf, but all of these tables were removed.
From demux_playlist.c I'm removing a code fragment from someone who was
not asked; this probably could be done later (see commit 15dccc37).
misc.c is a bit complicated to reason about (it was split off mplayer.c
and thus contains random functions out of this file), but actually all
functions have been added post-MPlayer. Except get_relative_time(),
which was written by uau, but looks similar to 3 different versions of
something similar in each of the Unix/win32/OSX timer source files. I'm
not sure what that means in regards to copyright, so I've just moved it
into another still-GPL source file for now.
screenshot.c once had some minor parts of MPlayer's vf_screenshot.c, but
they're all gone.
2016-01-19 17:36:06 +00:00
|
|
|
* You should have received a copy of the GNU Lesser General Public
|
|
|
|
* License along with mpv. If not, see <http://www.gnu.org/licenses/>.
|
2011-01-16 18:03:08 +00:00
|
|
|
*/
|
|
|
|
|
|
|
|
#include <stdlib.h>
|
|
|
|
#include <stdbool.h>
|
sub: add subtitle charset conversion
This code was once part of subreader.c, then traveled to libass, and now
made its way back to the fork of the fork of the original code, MPlayer.
It works pretty much the same as subreader.c, except that we have to
concatenate some packets to do auto-detection. This is rather annoying,
but for all we know the actual source file could be a binary format.
Unlike subreader.c, the iconv context is reopened on each packet. This
is simpler, and with respect to multibyte encodings, more robust.
Reopening is probably not a very fast, but I suspect subtitle charset
conversion is not an operation that happens often or has to be fast.
Also, this auto-detection is disabled for microdvd - this is the only
format we know that has binary data in its packets, but is actually
decoded to text. FFmpeg doesn't really allow us to solve this properly,
because a) the input packets can be binary, and b) the output will be
checked whether it's UTF-8, and if it's not, the output is thrown away
and an error message is printed. We could just recode the decoded
subtitles before sd_ass if it weren't for that.
2013-06-23 20:15:04 +00:00
|
|
|
#include <string.h>
|
2013-07-07 21:54:11 +00:00
|
|
|
#include <math.h>
|
2011-01-16 18:03:08 +00:00
|
|
|
#include <assert.h>
|
2024-01-21 23:14:39 +00:00
|
|
|
#include <limits.h>
|
2011-01-16 18:03:08 +00:00
|
|
|
|
2013-06-11 19:39:54 +00:00
|
|
|
#include "demux/demux.h"
|
2013-06-01 17:54:18 +00:00
|
|
|
#include "sd.h"
|
|
|
|
#include "dec_sub.h"
|
2017-12-29 16:19:25 +00:00
|
|
|
#include "options/m_config.h"
|
2013-12-17 01:02:25 +00:00
|
|
|
#include "options/options.h"
|
2013-12-21 18:06:37 +00:00
|
|
|
#include "common/global.h"
|
2013-12-17 01:39:45 +00:00
|
|
|
#include "common/msg.h"
|
2017-02-07 16:05:17 +00:00
|
|
|
#include "common/recorder.h"
|
sub: remove only user of demux_read_packet()
There are 3 packet reading functions in the demux API, which all
function completely differently. One of them, demux_read_packet(), has
only 1 caller, which is in dec_sub.c. Change this caller to use
demux_read_packet_async() instead. Since it really wants to do a
blocking call, setup some proper waiting. This uses mp_dispatch_queue,
because even though it's overkill, it needs the least code.
In practice, waiting actually never happens. This code is only called on
code paths where everything is already read into memory (libavformat's
subtitle demuxers simply behave this way). It's still a bit of a
"coincidence", so implement it properly anyway.
If suubtitle decoder init fails, we still need to unset the demuxer
wakeup callback. Add a sub_destroy() call to the failure path. This also
happens to fix a missed pthread_mutex_destroy() call (in practice this
was a nop, or a memory leak on BSDs).
2019-05-17 16:52:22 +00:00
|
|
|
#include "misc/dispatch.h"
|
2014-01-31 18:50:25 +00:00
|
|
|
#include "osdep/threads.h"
|
2011-01-16 18:03:08 +00:00
|
|
|
|
|
|
|
extern const struct sd_functions sd_ass;
|
2012-08-16 15:21:21 +00:00
|
|
|
extern const struct sd_functions sd_lavc;
|
sub: add sd_spu.c to wrap spudec, cleanup mplayer.c
This unifies the subtitle rendering path. Now all subtitle rendering
goes through sd_ass.c/sd_lavc.c/sd_spu.c.
Before that commit, the spudec.h functions were used directly in
mplayer.c, which introduced many special cases. Add sd_spu.c, which is
just a small wrapper connecting the new subtitle render API with the
dusty old vobsub decoder in spudec.c.
One detail that changes is that we always pass the palette as extra
data, instead of passing the libdvdread palette as pointer to spudec
directly. This is a bit roundabout, but actually makes the code simpler
and more elegant: the difference between DVD and non-DVD dvdsubs is
reduced.
Ideally, we would just delete spudec.c and use libavcodec's DVD sub
decoder. However, DVD playback with demux_mpg produces packets
incompatible to lavc. There are incompatibilities the other way around
as well: packets from libavformat's vobsub demuxer are incompatible to
spudec.c. So we define a new subtitle codec name for demux_mpg subs,
"dvd_subtitle_mpg", which only sd_spu can decode.
There is actually code in spudec.c to "assemble" fragments into complete
packets, but using the whole spudec.c is easier than trying to move this
code into demux_mpg to fix subtitle packets.
As additional complication, Libav 9.x can't decode DVD subs correctly,
so use sd_spu in that case as well.
2013-04-28 23:13:22 +00:00
|
|
|
|
2014-06-10 21:56:05 +00:00
|
|
|
static const struct sd_functions *const sd_list[] = {
|
2015-12-18 00:54:14 +00:00
|
|
|
&sd_lavc,
|
sub: add sd_spu.c to wrap spudec, cleanup mplayer.c
This unifies the subtitle rendering path. Now all subtitle rendering
goes through sd_ass.c/sd_lavc.c/sd_spu.c.
Before that commit, the spudec.h functions were used directly in
mplayer.c, which introduced many special cases. Add sd_spu.c, which is
just a small wrapper connecting the new subtitle render API with the
dusty old vobsub decoder in spudec.c.
One detail that changes is that we always pass the palette as extra
data, instead of passing the libdvdread palette as pointer to spudec
directly. This is a bit roundabout, but actually makes the code simpler
and more elegant: the difference between DVD and non-DVD dvdsubs is
reduced.
Ideally, we would just delete spudec.c and use libavcodec's DVD sub
decoder. However, DVD playback with demux_mpg produces packets
incompatible to lavc. There are incompatibilities the other way around
as well: packets from libavformat's vobsub demuxer are incompatible to
spudec.c. So we define a new subtitle codec name for demux_mpg subs,
"dvd_subtitle_mpg", which only sd_spu can decode.
There is actually code in spudec.c to "assemble" fragments into complete
packets, but using the whole spudec.c is easier than trying to move this
code into demux_mpg to fix subtitle packets.
As additional complication, Libav 9.x can't decode DVD subs correctly,
so use sd_spu in that case as well.
2013-04-28 23:13:22 +00:00
|
|
|
&sd_ass,
|
|
|
|
NULL
|
|
|
|
};
|
2011-01-16 18:03:08 +00:00
|
|
|
|
2013-06-01 17:44:12 +00:00
|
|
|
struct dec_sub {
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex lock;
|
sub: uglify sub decoder with locking
The plan is to make the whole OSD thread-safe, and we start with this.
We just put locks on all entry points (fortunately, dec_sub.c and all
sd_*.c decoders are very closed off, and only the entry points in
dec_sub.h let you access it). I think this is pretty ugly, but at least
it's very simple.
There's a special case with sub_get_bitmaps(): this function returns
pointers to decoder data (specifically, libass images). There's no way
to synchronize this internally, so expose sub_lock/sub_unlock functions.
To make things simpler, and especially because the lock is sort-of
exposed to the outside world, make the locks recursive. Although the
only case where this is actually needed (although trivial) is
sub_set_extradata().
One corner case are ASS subtitles: for some reason, we keep a single
ASS_Renderer instance for subtitles around (probably to avoid rescanning
fonts with ordered chapters), and this ASS_Renderer instance is not
synchronized. Also, demux_libass.c loads ASS_Track objects, which are
directly passed to sd_ass.c. These things are not synchronized (and
would be hard to synchronize), and basically we're out of luck. But I
think for now, accesses happen reasonably serialized, so there is no
actual problem yet, even if we start to access OSD from other threads.
2014-01-17 22:13:09 +00:00
|
|
|
|
2013-12-21 18:06:37 +00:00
|
|
|
struct mp_log *log;
|
2016-02-15 19:26:01 +00:00
|
|
|
struct mpv_global *global;
|
2017-12-29 16:19:25 +00:00
|
|
|
struct mp_subtitle_opts *opts;
|
2023-12-13 21:36:58 +00:00
|
|
|
struct mp_subtitle_shared_opts *shared_opts;
|
2017-12-29 16:19:25 +00:00
|
|
|
struct m_config_cache *opts_cache;
|
2023-12-13 21:36:58 +00:00
|
|
|
struct m_config_cache *shared_opts_cache;
|
2013-06-01 17:44:12 +00:00
|
|
|
|
2017-02-07 16:05:17 +00:00
|
|
|
struct mp_recorder_sink *recorder_sink;
|
|
|
|
|
2016-03-03 17:48:56 +00:00
|
|
|
struct attachment_list *attachments;
|
2016-02-15 19:26:01 +00:00
|
|
|
|
2015-12-16 22:54:25 +00:00
|
|
|
struct sh_stream *sh;
|
Implement backwards playback
See manpage additions. This is a huge hack. You can bet there are shit
tons of bugs. It's literally forcing square pegs into round holes.
Hopefully, the manpage wall of text makes it clear enough that the whole
shit can easily crash and burn. (Although it shouldn't literally crash.
That would be a bug. It possibly _could_ start a fire by entering some
sort of endless loop, not a literal one, just something where it tries
to do work without making progress.)
(Some obvious bugs I simply ignored for this initial version, but
there's a number of potential bugs I can't even imagine. Normal playback
should remain completely unaffected, though.)
How this works is also described in the manpage. Basically, we demux in
reverse, then we decode in reverse, then we render in reverse.
The decoding part is the simplest: just reorder the decoder output. This
weirdly integrates with the timeline/ordered chapter code, which also
has special requirements on feeding the packets to the decoder in a
non-straightforward way (it doesn't conflict, although a bugmessmass
breaks correct slicing of segments, so EDL/ordered chapter playback is
broken in backward direction).
Backward demuxing is pretty involved. In theory, it could be much
easier: simply iterating the usual demuxer output backward. But this
just doesn't fit into our code, so there's a cthulhu nightmare of shit.
To be specific, each stream (audio, video) is reversed separately. At
least this means we can do backward playback within cached content (for
example, you could play backwards in a live stream; on that note, it
disables prefetching, which would lead to losing new live video, but
this could be avoided).
The fuckmess also meant that I didn't bother trying to support
subtitles. Subtitles are a problem because they're "sparse" streams.
They need to be "passively" demuxed: you don't try to read a subtitle
packet, you demux audio and video, and then look whether there was a
subtitle packet. This means to get subtitles for a time range, you need
to know that you demuxed video and audio over this range, which becomes
pretty messy when you demux audio and video backwards separately.
Backward display is the most weird (and potentially buggy) part. To
avoid that we need to touch a LOT of timing code, we negate all
timestamps. The basic idea is that due to the navigation, all
comparisons and subtractions of timestamps keep working, and you don't
need to touch every single of them to "reverse" them.
E.g.:
bool before = pts_a < pts_b;
would need to be:
bool before = forward
? pts_a < pts_b
: pts_a > pts_b;
or:
bool before = pts_a * dir < pts_b * dir;
or if you, as it's implemented now, just do this after decoding:
pts_a *= dir;
pts_b *= dir;
and then in the normal timing/renderer code:
bool before = pts_a < pts_b;
Consequently, we don't need many changes in the latter code. But some
assumptions inhererently true for forward playback may have been broken
anyway. What is mainly needed is fixing places where values are passed
between positive and negative "domains". For example, seeking and
timestamp user display always uses positive timestamps. The main mess is
that it's not obvious which domain a given variable should or does use.
Well, in my tests with a single file, it suddenly started to work when I
did this. I'm honestly surprised that it did, and that I didn't have to
change a single line in the timing code past decoder (just something
minor to make external/cached text subtitles display). I committed it
immediately while avoiding thinking about it. But there really likely
are subtle problems of all sorts.
As far as I'm aware, gstreamer also supports backward playback. When I
looked at this years ago, I couldn't find a way to actually try this,
and I didn't revisit it now. Back then I also read talk slides from the
person who implemented it, and I'm not sure if and which ideas I might
have taken from it. It's possible that the timestamp reversal is
inspired by it, but I didn't check. (I think it claimed that it could
avoid large changes by changing a sign?)
VapourSynth has some sort of reverse function, which provides a backward
view on a video. The function itself is trivial to implement, as
VapourSynth aims to provide random access to video by frame numbers (so
you just request decreasing frame numbers). From what I remember, it
wasn't exactly fluid, but it worked. It's implemented by creating an
index, and seeking to the target on demand, and a bunch of caching. mpv
could use it, but it would either require using VapourSynth as demuxer
and decoder for everything, or replacing the current file every time
something is supposed to be played backwards.
FFmpeg's libavfilter has reversal filters for audio and video. These
require buffering the entire media data of the file, and don't really
fit into mpv's architecture. It could be used by playing a libavfilter
graph that also demuxes, but that's like VapourSynth but worse.
2019-05-18 00:10:51 +00:00
|
|
|
int play_dir;
|
sub: align ytdl-hook secondary subs to the top
29e15e6248 prefixed youtube-dl's subs url with an edl prefix to not
download them until they're selected, which is useful when there are
many sub languages. But this prefix broke the alignment of secondary
subs, which would overlap the primary subs instead of always being
placed at the top. This can be tested with
--sub-file='edl://!no_clip;!delay_open,media_type=sub;secondary_sub.srt'
When a sub is added, sub.c:reinit_sub() is called. This calls in
init_subdec() -> dec_sub.c:sub_create() -> init_decoder() ->
sd_ass:init(). Then reinit_sub() calls
sub_control(track->d_sub, SD_CTRL_SET_TOP, &(bool){!!order}) which sets
sd_ass_priv.on_top = true for secondary subs.
But for EDL subs the real sub is initialized again when in
dec_sub.c:sub_read_packets() is_new_segment() returns true and
update_segment() is called, or when sub_get_bitmaps() calls
update_segment(). update_segment() then calls init_decoder(), which
calls sd_ass:init(), so sd_ass_priv is reinitialized, and its on_top
property is left false. This commit sets it to true again.
For URLs that need to be downloaded it seems that the update_segment()
call that reinitializes sd_ass_priv is always the one in
sub_read_packets(), but with local subs sub_get_bitmaps() is usually
called earlier (though there shouldn't be a reason to use the EDL URL
for local subs), so I added the order parameter to sub_create(), rather
than adding it to all of update_segment(), sub_read_packets() and
sub_get_bitmaps().
Also removes the cast to bool in the other sub_control call, since
sub/sd_ass.c:control already casts arg to bool when cmd is
SD_CTRL_SET_TOP.
2021-08-10 20:11:53 +00:00
|
|
|
int order;
|
2015-12-29 00:35:52 +00:00
|
|
|
double last_pkt_pts;
|
sub: make preloading more robust
Subtitles can be preloaded, which means they're fully read and copied
into ASS_Track. This in turn is mainly for the sake of being able to do
subtitle seeking (when it comes down to it, subtitle seeking is the
cause for most trouble here).
Commit a714f8e92 broke preloaded subtitles which have events with
unknown duration, such as some MicroDVD samples. The event list gets
cleared on every seek, so the property of being preloaded obviously gets
lost.
Fix this by moving most of the preloading logic to dec_sub.c. If the
subtitle list gets cleared, they are not considered preloaded anymore,
and the logic for demuxed subtitles is used.
As another minor thing, preloadeding subtitles did neither disable the
demux stream, nor did it discard packets. Thus you could get queue
overflows in theory (harmless, but annoying). Fix this by explicitly
discarding packets in preloaded mode.
In summary, now the only difference between preloaded and normal
demuxing are:
1. a seek is issued, and all packets are read on start
2. during playback, discard the packets instead of feeding them to the
subtitle decoder
This is still petty annoying. It would be nice if maintaining the
subtitle index (and maybe a subtitle packet cache for instant subtitle
presentation when seeking back) could be maintained in the demuxer
instead. Half of all file formats with interleaved subtitles have
this anyway (mp4, mkv muxed with newer mkvmerge).
2016-03-06 13:50:36 +00:00
|
|
|
bool preload_attempted;
|
2017-12-29 14:39:38 +00:00
|
|
|
double video_fps;
|
|
|
|
double sub_speed;
|
2024-01-21 23:14:39 +00:00
|
|
|
bool sub_visible;
|
2013-06-11 19:39:54 +00:00
|
|
|
|
Rewrite ordered chapters and timeline stuff
This uses a different method to piece segments together. The old
approach basically changes to a new file (with a new start offset) any
time a segment ends. This meant waiting for audio/video end on segment
end, and then changing to the new segment all at once. It had a very
weird impact on the playback core, and some things (like truly gapless
segment transitions, or frame backstepping) just didn't work.
The new approach adds the demux_timeline pseudo-demuxer, which presents
an uniform packet stream from the many segments. This is pretty similar
to how ordered chapters are implemented everywhere else. It also reminds
of the FFmpeg concat pseudo-demuxer.
The "pure" version of this approach doesn't work though. Segments can
actually have different codec configurations (different extradata), and
subtitles are most likely broken too. (Subtitles have multiple corner
cases which break the pure stream-concatenation approach completely.)
To counter this, we do two things:
- Reinit the decoder with each segment. We go as far as allowing
concatenating files with completely different codecs for the sake
of EDL (which also uses the timeline infrastructure). A "lighter"
approach would try to make use of decoder mechanism to update e.g.
the extradata, but that seems fragile.
- Clip decoded data to segment boundaries. This is equivalent to
normal playback core mechanisms like hr-seek, but now the playback
core doesn't need to care about these things.
These two mechanisms are equivalent to what happened in the old
implementation, except they don't happen in the playback core anymore.
In other words, the playback core is completely relieved from timeline
implementation details. (Which honestly is exactly what I'm trying to
do here. I don't think ordered chapter behavior deserves improvement,
even if it's bad - but I want to get it out from the playback core.)
There is code duplication between audio and video decoder common code.
This is awful and could be shareable - but this will happen later.
Note that the audio path has some code to clip audio frames for the
purpose of codec preroll/gapless handling, but it's not shared as
sharing it would cause more pain than it would help.
2016-02-15 20:04:07 +00:00
|
|
|
struct mp_codec_params *codec;
|
|
|
|
double start, end;
|
|
|
|
|
|
|
|
double last_vo_pts;
|
2015-12-18 00:54:14 +00:00
|
|
|
struct sd *sd;
|
Rewrite ordered chapters and timeline stuff
This uses a different method to piece segments together. The old
approach basically changes to a new file (with a new start offset) any
time a segment ends. This meant waiting for audio/video end on segment
end, and then changing to the new segment all at once. It had a very
weird impact on the playback core, and some things (like truly gapless
segment transitions, or frame backstepping) just didn't work.
The new approach adds the demux_timeline pseudo-demuxer, which presents
an uniform packet stream from the many segments. This is pretty similar
to how ordered chapters are implemented everywhere else. It also reminds
of the FFmpeg concat pseudo-demuxer.
The "pure" version of this approach doesn't work though. Segments can
actually have different codec configurations (different extradata), and
subtitles are most likely broken too. (Subtitles have multiple corner
cases which break the pure stream-concatenation approach completely.)
To counter this, we do two things:
- Reinit the decoder with each segment. We go as far as allowing
concatenating files with completely different codecs for the sake
of EDL (which also uses the timeline infrastructure). A "lighter"
approach would try to make use of decoder mechanism to update e.g.
the extradata, but that seems fragile.
- Clip decoded data to segment boundaries. This is equivalent to
normal playback core mechanisms like hr-seek, but now the playback
core doesn't need to care about these things.
These two mechanisms are equivalent to what happened in the old
implementation, except they don't happen in the playback core anymore.
In other words, the playback core is completely relieved from timeline
implementation details. (Which honestly is exactly what I'm trying to
do here. I don't think ordered chapter behavior deserves improvement,
even if it's bad - but I want to get it out from the playback core.)
There is code duplication between audio and video decoder common code.
This is awful and could be shareable - but this will happen later.
Note that the audio path has some code to clip audio frames for the
purpose of codec preroll/gapless handling, but it's not shared as
sharing it would cause more pain than it would help.
2016-02-15 20:04:07 +00:00
|
|
|
|
|
|
|
struct demux_packet *new_segment;
|
2024-01-22 18:55:50 +00:00
|
|
|
struct demux_packet **cached_pkts;
|
2024-01-21 23:14:39 +00:00
|
|
|
int cached_pkt_pos;
|
2024-01-22 18:55:50 +00:00
|
|
|
int num_cached_pkts;
|
2013-06-01 17:44:12 +00:00
|
|
|
};
|
|
|
|
|
2017-12-29 14:39:38 +00:00
|
|
|
static void update_subtitle_speed(struct dec_sub *sub)
|
|
|
|
{
|
2017-12-29 16:19:25 +00:00
|
|
|
struct mp_subtitle_opts *opts = sub->opts;
|
2017-12-29 14:39:38 +00:00
|
|
|
sub->sub_speed = 1.0;
|
|
|
|
|
|
|
|
if (sub->video_fps > 0 && sub->codec->frame_based > 0) {
|
|
|
|
MP_VERBOSE(sub, "Frame based format, dummy FPS: %f, video FPS: %f\n",
|
|
|
|
sub->codec->frame_based, sub->video_fps);
|
|
|
|
sub->sub_speed *= sub->codec->frame_based / sub->video_fps;
|
|
|
|
}
|
|
|
|
|
|
|
|
if (opts->sub_fps && sub->video_fps)
|
|
|
|
sub->sub_speed *= opts->sub_fps / sub->video_fps;
|
|
|
|
|
|
|
|
sub->sub_speed *= opts->sub_speed;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Return the subtitle PTS used for a given video PTS.
|
|
|
|
static double pts_to_subtitle(struct dec_sub *sub, double pts)
|
|
|
|
{
|
2023-12-13 21:36:58 +00:00
|
|
|
struct mp_subtitle_shared_opts *opts = sub->shared_opts;
|
2023-11-23 03:13:57 +00:00
|
|
|
float delay = sub->order < 0 ? 0.0f : opts->sub_delay[sub->order];
|
2017-12-29 14:39:38 +00:00
|
|
|
|
|
|
|
if (pts != MP_NOPTS_VALUE)
|
2023-11-23 03:13:57 +00:00
|
|
|
pts = (pts * sub->play_dir - delay) / sub->sub_speed;
|
2017-12-29 14:39:38 +00:00
|
|
|
|
|
|
|
return pts;
|
|
|
|
}
|
|
|
|
|
|
|
|
static double pts_from_subtitle(struct dec_sub *sub, double pts)
|
|
|
|
{
|
2023-12-13 21:36:58 +00:00
|
|
|
struct mp_subtitle_shared_opts *opts = sub->shared_opts;
|
2023-11-23 03:13:57 +00:00
|
|
|
float delay = sub->order < 0 ? 0.0f : opts->sub_delay[sub->order];
|
2017-12-29 14:39:38 +00:00
|
|
|
|
|
|
|
if (pts != MP_NOPTS_VALUE)
|
2023-11-23 03:13:57 +00:00
|
|
|
pts = (pts * sub->sub_speed + delay) * sub->play_dir;
|
2017-12-29 14:39:38 +00:00
|
|
|
|
|
|
|
return pts;
|
|
|
|
}
|
|
|
|
|
sub: remove only user of demux_read_packet()
There are 3 packet reading functions in the demux API, which all
function completely differently. One of them, demux_read_packet(), has
only 1 caller, which is in dec_sub.c. Change this caller to use
demux_read_packet_async() instead. Since it really wants to do a
blocking call, setup some proper waiting. This uses mp_dispatch_queue,
because even though it's overkill, it needs the least code.
In practice, waiting actually never happens. This code is only called on
code paths where everything is already read into memory (libavformat's
subtitle demuxers simply behave this way). It's still a bit of a
"coincidence", so implement it properly anyway.
If suubtitle decoder init fails, we still need to unset the demuxer
wakeup callback. Add a sub_destroy() call to the failure path. This also
happens to fix a missed pthread_mutex_destroy() call (in practice this
was a nop, or a memory leak on BSDs).
2019-05-17 16:52:22 +00:00
|
|
|
static void wakeup_demux(void *ctx)
|
|
|
|
{
|
|
|
|
struct mp_dispatch_queue *q = ctx;
|
|
|
|
mp_dispatch_interrupt(q);
|
|
|
|
}
|
|
|
|
|
2024-04-04 12:06:55 +00:00
|
|
|
static void destroy_cached_pkts(struct dec_sub *sub)
|
2024-01-22 18:55:50 +00:00
|
|
|
{
|
|
|
|
int index = 0;
|
|
|
|
while (index < sub->num_cached_pkts) {
|
|
|
|
TA_FREEP(&sub->cached_pkts[index]);
|
|
|
|
++index;
|
|
|
|
}
|
2024-01-21 23:14:39 +00:00
|
|
|
sub->cached_pkt_pos = 0;
|
2024-01-22 18:55:50 +00:00
|
|
|
sub->num_cached_pkts = 0;
|
|
|
|
}
|
|
|
|
|
2013-06-01 17:44:12 +00:00
|
|
|
void sub_destroy(struct dec_sub *sub)
|
|
|
|
{
|
|
|
|
if (!sub)
|
|
|
|
return;
|
sub: remove only user of demux_read_packet()
There are 3 packet reading functions in the demux API, which all
function completely differently. One of them, demux_read_packet(), has
only 1 caller, which is in dec_sub.c. Change this caller to use
demux_read_packet_async() instead. Since it really wants to do a
blocking call, setup some proper waiting. This uses mp_dispatch_queue,
because even though it's overkill, it needs the least code.
In practice, waiting actually never happens. This code is only called on
code paths where everything is already read into memory (libavformat's
subtitle demuxers simply behave this way). It's still a bit of a
"coincidence", so implement it properly anyway.
If suubtitle decoder init fails, we still need to unset the demuxer
wakeup callback. Add a sub_destroy() call to the failure path. This also
happens to fix a missed pthread_mutex_destroy() call (in practice this
was a nop, or a memory leak on BSDs).
2019-05-17 16:52:22 +00:00
|
|
|
demux_set_stream_wakeup_cb(sub->sh, NULL, NULL);
|
|
|
|
if (sub->sd) {
|
|
|
|
sub_reset(sub);
|
|
|
|
sub->sd->driver->uninit(sub->sd);
|
|
|
|
}
|
2015-12-27 01:07:01 +00:00
|
|
|
talloc_free(sub->sd);
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_destroy(&sub->lock);
|
2013-06-01 17:44:12 +00:00
|
|
|
talloc_free(sub);
|
|
|
|
}
|
|
|
|
|
2016-02-15 19:26:01 +00:00
|
|
|
static struct sd *init_decoder(struct dec_sub *sub)
|
2013-06-01 17:44:12 +00:00
|
|
|
{
|
2015-12-27 01:07:01 +00:00
|
|
|
for (int n = 0; sd_list[n]; n++) {
|
|
|
|
const struct sd_functions *driver = sd_list[n];
|
2016-02-15 19:26:01 +00:00
|
|
|
struct sd *sd = talloc(NULL, struct sd);
|
|
|
|
*sd = (struct sd){
|
|
|
|
.global = sub->global,
|
|
|
|
.log = mp_log_new(sd, sub->log, driver->name),
|
2015-12-27 01:07:01 +00:00
|
|
|
.opts = sub->opts,
|
2023-12-13 21:36:58 +00:00
|
|
|
.shared_opts = sub->shared_opts,
|
2015-12-27 01:07:01 +00:00
|
|
|
.driver = driver,
|
2023-12-17 19:18:11 +00:00
|
|
|
.order = sub->order,
|
2016-03-03 17:48:56 +00:00
|
|
|
.attachments = sub->attachments,
|
2016-02-15 19:26:01 +00:00
|
|
|
.codec = sub->codec,
|
sub: make preloading more robust
Subtitles can be preloaded, which means they're fully read and copied
into ASS_Track. This in turn is mainly for the sake of being able to do
subtitle seeking (when it comes down to it, subtitle seeking is the
cause for most trouble here).
Commit a714f8e92 broke preloaded subtitles which have events with
unknown duration, such as some MicroDVD samples. The event list gets
cleared on every seek, so the property of being preloaded obviously gets
lost.
Fix this by moving most of the preloading logic to dec_sub.c. If the
subtitle list gets cleared, they are not considered preloaded anymore,
and the logic for demuxed subtitles is used.
As another minor thing, preloadeding subtitles did neither disable the
demux stream, nor did it discard packets. Thus you could get queue
overflows in theory (harmless, but annoying). Fix this by explicitly
discarding packets in preloaded mode.
In summary, now the only difference between preloaded and normal
demuxing are:
1. a seek is issued, and all packets are read on start
2. during playback, discard the packets instead of feeding them to the
subtitle decoder
This is still petty annoying. It would be nice if maintaining the
subtitle index (and maybe a subtitle packet cache for instant subtitle
presentation when seeking back) could be maintained in the demuxer
instead. Half of all file formats with interleaved subtitles have
this anyway (mp4, mkv muxed with newer mkvmerge).
2016-03-06 13:50:36 +00:00
|
|
|
.preload_ok = true,
|
2015-12-27 01:07:01 +00:00
|
|
|
};
|
2013-06-01 17:44:55 +00:00
|
|
|
|
2016-02-15 19:26:01 +00:00
|
|
|
if (sd->driver->init(sd) >= 0)
|
|
|
|
return sd;
|
2013-06-01 17:44:55 +00:00
|
|
|
|
2016-02-15 19:26:01 +00:00
|
|
|
talloc_free(sd);
|
2013-06-23 22:47:08 +00:00
|
|
|
}
|
2015-12-18 00:54:14 +00:00
|
|
|
|
2016-02-15 19:26:01 +00:00
|
|
|
MP_ERR(sub, "Could not find subtitle decoder for format '%s'.\n",
|
|
|
|
sub->codec->codec);
|
|
|
|
return NULL;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Thread-safety of the returned object: all functions are thread-safe,
|
|
|
|
// except sub_get_bitmaps() and sub_get_text(). Decoder backends (sd_*)
|
|
|
|
// do not need to acquire locks.
|
2020-02-27 01:24:43 +00:00
|
|
|
// Ownership of attachments goes to the callee, and is released with
|
2016-03-03 17:48:56 +00:00
|
|
|
// talloc_free() (even on failure).
|
2021-05-27 05:56:41 +00:00
|
|
|
struct dec_sub *sub_create(struct mpv_global *global, struct track *track,
|
sub: align ytdl-hook secondary subs to the top
29e15e6248 prefixed youtube-dl's subs url with an edl prefix to not
download them until they're selected, which is useful when there are
many sub languages. But this prefix broke the alignment of secondary
subs, which would overlap the primary subs instead of always being
placed at the top. This can be tested with
--sub-file='edl://!no_clip;!delay_open,media_type=sub;secondary_sub.srt'
When a sub is added, sub.c:reinit_sub() is called. This calls in
init_subdec() -> dec_sub.c:sub_create() -> init_decoder() ->
sd_ass:init(). Then reinit_sub() calls
sub_control(track->d_sub, SD_CTRL_SET_TOP, &(bool){!!order}) which sets
sd_ass_priv.on_top = true for secondary subs.
But for EDL subs the real sub is initialized again when in
dec_sub.c:sub_read_packets() is_new_segment() returns true and
update_segment() is called, or when sub_get_bitmaps() calls
update_segment(). update_segment() then calls init_decoder(), which
calls sd_ass:init(), so sd_ass_priv is reinitialized, and its on_top
property is left false. This commit sets it to true again.
For URLs that need to be downloaded it seems that the update_segment()
call that reinitializes sd_ass_priv is always the one in
sub_read_packets(), but with local subs sub_get_bitmaps() is usually
called earlier (though there shouldn't be a reason to use the EDL URL
for local subs), so I added the order parameter to sub_create(), rather
than adding it to all of update_segment(), sub_read_packets() and
sub_get_bitmaps().
Also removes the cast to bool in the other sub_control call, since
sub/sd_ass.c:control already casts arg to bool when cmd is
SD_CTRL_SET_TOP.
2021-08-10 20:11:53 +00:00
|
|
|
struct attachment_list *attachments, int order)
|
2016-02-15 19:26:01 +00:00
|
|
|
{
|
2021-05-27 05:56:41 +00:00
|
|
|
assert(track->stream && track->stream->type == STREAM_SUB);
|
2016-02-15 19:26:01 +00:00
|
|
|
|
|
|
|
struct dec_sub *sub = talloc(NULL, struct dec_sub);
|
|
|
|
*sub = (struct dec_sub){
|
|
|
|
.log = mp_log_new(sub, global->log, "sub"),
|
|
|
|
.global = global,
|
2017-12-29 16:19:25 +00:00
|
|
|
.opts_cache = m_config_cache_alloc(sub, global, &mp_subtitle_sub_opts),
|
2023-12-13 21:36:58 +00:00
|
|
|
.shared_opts_cache = m_config_cache_alloc(sub, global, &mp_subtitle_shared_sub_opts),
|
2021-05-27 05:56:41 +00:00
|
|
|
.sh = track->stream,
|
|
|
|
.codec = track->stream->codec,
|
2016-03-03 17:48:56 +00:00
|
|
|
.attachments = talloc_steal(sub, attachments),
|
Implement backwards playback
See manpage additions. This is a huge hack. You can bet there are shit
tons of bugs. It's literally forcing square pegs into round holes.
Hopefully, the manpage wall of text makes it clear enough that the whole
shit can easily crash and burn. (Although it shouldn't literally crash.
That would be a bug. It possibly _could_ start a fire by entering some
sort of endless loop, not a literal one, just something where it tries
to do work without making progress.)
(Some obvious bugs I simply ignored for this initial version, but
there's a number of potential bugs I can't even imagine. Normal playback
should remain completely unaffected, though.)
How this works is also described in the manpage. Basically, we demux in
reverse, then we decode in reverse, then we render in reverse.
The decoding part is the simplest: just reorder the decoder output. This
weirdly integrates with the timeline/ordered chapter code, which also
has special requirements on feeding the packets to the decoder in a
non-straightforward way (it doesn't conflict, although a bugmessmass
breaks correct slicing of segments, so EDL/ordered chapter playback is
broken in backward direction).
Backward demuxing is pretty involved. In theory, it could be much
easier: simply iterating the usual demuxer output backward. But this
just doesn't fit into our code, so there's a cthulhu nightmare of shit.
To be specific, each stream (audio, video) is reversed separately. At
least this means we can do backward playback within cached content (for
example, you could play backwards in a live stream; on that note, it
disables prefetching, which would lead to losing new live video, but
this could be avoided).
The fuckmess also meant that I didn't bother trying to support
subtitles. Subtitles are a problem because they're "sparse" streams.
They need to be "passively" demuxed: you don't try to read a subtitle
packet, you demux audio and video, and then look whether there was a
subtitle packet. This means to get subtitles for a time range, you need
to know that you demuxed video and audio over this range, which becomes
pretty messy when you demux audio and video backwards separately.
Backward display is the most weird (and potentially buggy) part. To
avoid that we need to touch a LOT of timing code, we negate all
timestamps. The basic idea is that due to the navigation, all
comparisons and subtractions of timestamps keep working, and you don't
need to touch every single of them to "reverse" them.
E.g.:
bool before = pts_a < pts_b;
would need to be:
bool before = forward
? pts_a < pts_b
: pts_a > pts_b;
or:
bool before = pts_a * dir < pts_b * dir;
or if you, as it's implemented now, just do this after decoding:
pts_a *= dir;
pts_b *= dir;
and then in the normal timing/renderer code:
bool before = pts_a < pts_b;
Consequently, we don't need many changes in the latter code. But some
assumptions inhererently true for forward playback may have been broken
anyway. What is mainly needed is fixing places where values are passed
between positive and negative "domains". For example, seeking and
timestamp user display always uses positive timestamps. The main mess is
that it's not obvious which domain a given variable should or does use.
Well, in my tests with a single file, it suddenly started to work when I
did this. I'm honestly surprised that it did, and that I didn't have to
change a single line in the timing code past decoder (just something
minor to make external/cached text subtitles display). I committed it
immediately while avoiding thinking about it. But there really likely
are subtle problems of all sorts.
As far as I'm aware, gstreamer also supports backward playback. When I
looked at this years ago, I couldn't find a way to actually try this,
and I didn't revisit it now. Back then I also read talk slides from the
person who implemented it, and I'm not sure if and which ideas I might
have taken from it. It's possible that the timestamp reversal is
inspired by it, but I didn't check. (I think it claimed that it could
avoid large changes by changing a sign?)
VapourSynth has some sort of reverse function, which provides a backward
view on a video. The function itself is trivial to implement, as
VapourSynth aims to provide random access to video by frame numbers (so
you just request decreasing frame numbers). From what I remember, it
wasn't exactly fluid, but it worked. It's implemented by creating an
index, and seeking to the target on demand, and a bunch of caching. mpv
could use it, but it would either require using VapourSynth as demuxer
and decoder for everything, or replacing the current file every time
something is supposed to be played backwards.
FFmpeg's libavfilter has reversal filters for audio and video. These
require buffering the entire media data of the file, and don't really
fit into mpv's architecture. It could be used by playing a libavfilter
graph that also demuxes, but that's like VapourSynth but worse.
2019-05-18 00:10:51 +00:00
|
|
|
.play_dir = 1,
|
sub: align ytdl-hook secondary subs to the top
29e15e6248 prefixed youtube-dl's subs url with an edl prefix to not
download them until they're selected, which is useful when there are
many sub languages. But this prefix broke the alignment of secondary
subs, which would overlap the primary subs instead of always being
placed at the top. This can be tested with
--sub-file='edl://!no_clip;!delay_open,media_type=sub;secondary_sub.srt'
When a sub is added, sub.c:reinit_sub() is called. This calls in
init_subdec() -> dec_sub.c:sub_create() -> init_decoder() ->
sd_ass:init(). Then reinit_sub() calls
sub_control(track->d_sub, SD_CTRL_SET_TOP, &(bool){!!order}) which sets
sd_ass_priv.on_top = true for secondary subs.
But for EDL subs the real sub is initialized again when in
dec_sub.c:sub_read_packets() is_new_segment() returns true and
update_segment() is called, or when sub_get_bitmaps() calls
update_segment(). update_segment() then calls init_decoder(), which
calls sd_ass:init(), so sd_ass_priv is reinitialized, and its on_top
property is left false. This commit sets it to true again.
For URLs that need to be downloaded it seems that the update_segment()
call that reinitializes sd_ass_priv is always the one in
sub_read_packets(), but with local subs sub_get_bitmaps() is usually
called earlier (though there shouldn't be a reason to use the EDL URL
for local subs), so I added the order parameter to sub_create(), rather
than adding it to all of update_segment(), sub_read_packets() and
sub_get_bitmaps().
Also removes the cast to bool in the other sub_control call, since
sub/sd_ass.c:control already casts arg to bool when cmd is
SD_CTRL_SET_TOP.
2021-08-10 20:11:53 +00:00
|
|
|
.order = order,
|
2016-02-15 19:26:01 +00:00
|
|
|
.last_pkt_pts = MP_NOPTS_VALUE,
|
Rewrite ordered chapters and timeline stuff
This uses a different method to piece segments together. The old
approach basically changes to a new file (with a new start offset) any
time a segment ends. This meant waiting for audio/video end on segment
end, and then changing to the new segment all at once. It had a very
weird impact on the playback core, and some things (like truly gapless
segment transitions, or frame backstepping) just didn't work.
The new approach adds the demux_timeline pseudo-demuxer, which presents
an uniform packet stream from the many segments. This is pretty similar
to how ordered chapters are implemented everywhere else. It also reminds
of the FFmpeg concat pseudo-demuxer.
The "pure" version of this approach doesn't work though. Segments can
actually have different codec configurations (different extradata), and
subtitles are most likely broken too. (Subtitles have multiple corner
cases which break the pure stream-concatenation approach completely.)
To counter this, we do two things:
- Reinit the decoder with each segment. We go as far as allowing
concatenating files with completely different codecs for the sake
of EDL (which also uses the timeline infrastructure). A "lighter"
approach would try to make use of decoder mechanism to update e.g.
the extradata, but that seems fragile.
- Clip decoded data to segment boundaries. This is equivalent to
normal playback core mechanisms like hr-seek, but now the playback
core doesn't need to care about these things.
These two mechanisms are equivalent to what happened in the old
implementation, except they don't happen in the playback core anymore.
In other words, the playback core is completely relieved from timeline
implementation details. (Which honestly is exactly what I'm trying to
do here. I don't think ordered chapter behavior deserves improvement,
even if it's bad - but I want to get it out from the playback core.)
There is code duplication between audio and video decoder common code.
This is awful and could be shareable - but this will happen later.
Note that the audio path has some code to clip audio frames for the
purpose of codec preroll/gapless handling, but it's not shared as
sharing it would cause more pain than it would help.
2016-02-15 20:04:07 +00:00
|
|
|
.last_vo_pts = MP_NOPTS_VALUE,
|
|
|
|
.start = MP_NOPTS_VALUE,
|
|
|
|
.end = MP_NOPTS_VALUE,
|
2016-02-15 19:26:01 +00:00
|
|
|
};
|
2017-12-29 16:19:25 +00:00
|
|
|
sub->opts = sub->opts_cache->opts;
|
2023-12-13 21:36:58 +00:00
|
|
|
sub->shared_opts = sub->shared_opts_cache->opts;
|
2024-04-04 12:06:55 +00:00
|
|
|
mp_mutex_init(&sub->lock);
|
2016-02-15 19:26:01 +00:00
|
|
|
|
|
|
|
sub->sd = init_decoder(sub);
|
2017-12-29 14:39:38 +00:00
|
|
|
if (sub->sd) {
|
|
|
|
update_subtitle_speed(sub);
|
2016-02-15 19:26:01 +00:00
|
|
|
return sub;
|
2017-12-29 14:39:38 +00:00
|
|
|
}
|
2016-02-15 19:26:01 +00:00
|
|
|
|
sub: remove only user of demux_read_packet()
There are 3 packet reading functions in the demux API, which all
function completely differently. One of them, demux_read_packet(), has
only 1 caller, which is in dec_sub.c. Change this caller to use
demux_read_packet_async() instead. Since it really wants to do a
blocking call, setup some proper waiting. This uses mp_dispatch_queue,
because even though it's overkill, it needs the least code.
In practice, waiting actually never happens. This code is only called on
code paths where everything is already read into memory (libavformat's
subtitle demuxers simply behave this way). It's still a bit of a
"coincidence", so implement it properly anyway.
If suubtitle decoder init fails, we still need to unset the demuxer
wakeup callback. Add a sub_destroy() call to the failure path. This also
happens to fix a missed pthread_mutex_destroy() call (in practice this
was a nop, or a memory leak on BSDs).
2019-05-17 16:52:22 +00:00
|
|
|
sub_destroy(sub);
|
2015-12-27 01:07:01 +00:00
|
|
|
return NULL;
|
2013-06-23 22:47:08 +00:00
|
|
|
}
|
|
|
|
|
Rewrite ordered chapters and timeline stuff
This uses a different method to piece segments together. The old
approach basically changes to a new file (with a new start offset) any
time a segment ends. This meant waiting for audio/video end on segment
end, and then changing to the new segment all at once. It had a very
weird impact on the playback core, and some things (like truly gapless
segment transitions, or frame backstepping) just didn't work.
The new approach adds the demux_timeline pseudo-demuxer, which presents
an uniform packet stream from the many segments. This is pretty similar
to how ordered chapters are implemented everywhere else. It also reminds
of the FFmpeg concat pseudo-demuxer.
The "pure" version of this approach doesn't work though. Segments can
actually have different codec configurations (different extradata), and
subtitles are most likely broken too. (Subtitles have multiple corner
cases which break the pure stream-concatenation approach completely.)
To counter this, we do two things:
- Reinit the decoder with each segment. We go as far as allowing
concatenating files with completely different codecs for the sake
of EDL (which also uses the timeline infrastructure). A "lighter"
approach would try to make use of decoder mechanism to update e.g.
the extradata, but that seems fragile.
- Clip decoded data to segment boundaries. This is equivalent to
normal playback core mechanisms like hr-seek, but now the playback
core doesn't need to care about these things.
These two mechanisms are equivalent to what happened in the old
implementation, except they don't happen in the playback core anymore.
In other words, the playback core is completely relieved from timeline
implementation details. (Which honestly is exactly what I'm trying to
do here. I don't think ordered chapter behavior deserves improvement,
even if it's bad - but I want to get it out from the playback core.)
There is code duplication between audio and video decoder common code.
This is awful and could be shareable - but this will happen later.
Note that the audio path has some code to clip audio frames for the
purpose of codec preroll/gapless handling, but it's not shared as
sharing it would cause more pain than it would help.
2016-02-15 20:04:07 +00:00
|
|
|
// Called locked.
|
|
|
|
static void update_segment(struct dec_sub *sub)
|
|
|
|
{
|
|
|
|
if (sub->new_segment && sub->last_vo_pts != MP_NOPTS_VALUE &&
|
|
|
|
sub->last_vo_pts >= sub->new_segment->start)
|
|
|
|
{
|
2016-03-25 16:40:18 +00:00
|
|
|
MP_VERBOSE(sub, "Switch segment: %f at %f\n", sub->new_segment->start,
|
|
|
|
sub->last_vo_pts);
|
|
|
|
|
Rewrite ordered chapters and timeline stuff
This uses a different method to piece segments together. The old
approach basically changes to a new file (with a new start offset) any
time a segment ends. This meant waiting for audio/video end on segment
end, and then changing to the new segment all at once. It had a very
weird impact on the playback core, and some things (like truly gapless
segment transitions, or frame backstepping) just didn't work.
The new approach adds the demux_timeline pseudo-demuxer, which presents
an uniform packet stream from the many segments. This is pretty similar
to how ordered chapters are implemented everywhere else. It also reminds
of the FFmpeg concat pseudo-demuxer.
The "pure" version of this approach doesn't work though. Segments can
actually have different codec configurations (different extradata), and
subtitles are most likely broken too. (Subtitles have multiple corner
cases which break the pure stream-concatenation approach completely.)
To counter this, we do two things:
- Reinit the decoder with each segment. We go as far as allowing
concatenating files with completely different codecs for the sake
of EDL (which also uses the timeline infrastructure). A "lighter"
approach would try to make use of decoder mechanism to update e.g.
the extradata, but that seems fragile.
- Clip decoded data to segment boundaries. This is equivalent to
normal playback core mechanisms like hr-seek, but now the playback
core doesn't need to care about these things.
These two mechanisms are equivalent to what happened in the old
implementation, except they don't happen in the playback core anymore.
In other words, the playback core is completely relieved from timeline
implementation details. (Which honestly is exactly what I'm trying to
do here. I don't think ordered chapter behavior deserves improvement,
even if it's bad - but I want to get it out from the playback core.)
There is code duplication between audio and video decoder common code.
This is awful and could be shareable - but this will happen later.
Note that the audio path has some code to clip audio frames for the
purpose of codec preroll/gapless handling, but it's not shared as
sharing it would cause more pain than it would help.
2016-02-15 20:04:07 +00:00
|
|
|
sub->codec = sub->new_segment->codec;
|
|
|
|
sub->start = sub->new_segment->start;
|
|
|
|
sub->end = sub->new_segment->end;
|
|
|
|
struct sd *new = init_decoder(sub);
|
|
|
|
if (new) {
|
|
|
|
sub->sd->driver->uninit(sub->sd);
|
|
|
|
talloc_free(sub->sd);
|
|
|
|
sub->sd = new;
|
2017-12-29 14:39:38 +00:00
|
|
|
update_subtitle_speed(sub);
|
Rewrite ordered chapters and timeline stuff
This uses a different method to piece segments together. The old
approach basically changes to a new file (with a new start offset) any
time a segment ends. This meant waiting for audio/video end on segment
end, and then changing to the new segment all at once. It had a very
weird impact on the playback core, and some things (like truly gapless
segment transitions, or frame backstepping) just didn't work.
The new approach adds the demux_timeline pseudo-demuxer, which presents
an uniform packet stream from the many segments. This is pretty similar
to how ordered chapters are implemented everywhere else. It also reminds
of the FFmpeg concat pseudo-demuxer.
The "pure" version of this approach doesn't work though. Segments can
actually have different codec configurations (different extradata), and
subtitles are most likely broken too. (Subtitles have multiple corner
cases which break the pure stream-concatenation approach completely.)
To counter this, we do two things:
- Reinit the decoder with each segment. We go as far as allowing
concatenating files with completely different codecs for the sake
of EDL (which also uses the timeline infrastructure). A "lighter"
approach would try to make use of decoder mechanism to update e.g.
the extradata, but that seems fragile.
- Clip decoded data to segment boundaries. This is equivalent to
normal playback core mechanisms like hr-seek, but now the playback
core doesn't need to care about these things.
These two mechanisms are equivalent to what happened in the old
implementation, except they don't happen in the playback core anymore.
In other words, the playback core is completely relieved from timeline
implementation details. (Which honestly is exactly what I'm trying to
do here. I don't think ordered chapter behavior deserves improvement,
even if it's bad - but I want to get it out from the playback core.)
There is code duplication between audio and video decoder common code.
This is awful and could be shareable - but this will happen later.
Note that the audio path has some code to clip audio frames for the
purpose of codec preroll/gapless handling, but it's not shared as
sharing it would cause more pain than it would help.
2016-02-15 20:04:07 +00:00
|
|
|
} else {
|
|
|
|
// We'll just keep the current decoder, and feed it possibly
|
|
|
|
// invalid data (not our fault if it crashes or something).
|
|
|
|
MP_ERR(sub, "Can't change to new codec.\n");
|
|
|
|
}
|
|
|
|
sub->sd->driver->decode(sub->sd, sub->new_segment);
|
|
|
|
talloc_free(sub->new_segment);
|
|
|
|
sub->new_segment = NULL;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
sub: make preloading more robust
Subtitles can be preloaded, which means they're fully read and copied
into ASS_Track. This in turn is mainly for the sake of being able to do
subtitle seeking (when it comes down to it, subtitle seeking is the
cause for most trouble here).
Commit a714f8e92 broke preloaded subtitles which have events with
unknown duration, such as some MicroDVD samples. The event list gets
cleared on every seek, so the property of being preloaded obviously gets
lost.
Fix this by moving most of the preloading logic to dec_sub.c. If the
subtitle list gets cleared, they are not considered preloaded anymore,
and the logic for demuxed subtitles is used.
As another minor thing, preloadeding subtitles did neither disable the
demux stream, nor did it discard packets. Thus you could get queue
overflows in theory (harmless, but annoying). Fix this by explicitly
discarding packets in preloaded mode.
In summary, now the only difference between preloaded and normal
demuxing are:
1. a seek is issued, and all packets are read on start
2. during playback, discard the packets instead of feeding them to the
subtitle decoder
This is still petty annoying. It would be nice if maintaining the
subtitle index (and maybe a subtitle packet cache for instant subtitle
presentation when seeking back) could be maintained in the demuxer
instead. Half of all file formats with interleaved subtitles have
this anyway (mp4, mkv muxed with newer mkvmerge).
2016-03-06 13:50:36 +00:00
|
|
|
bool sub_can_preload(struct dec_sub *sub)
|
2013-06-11 19:39:54 +00:00
|
|
|
{
|
sub: make preloading more robust
Subtitles can be preloaded, which means they're fully read and copied
into ASS_Track. This in turn is mainly for the sake of being able to do
subtitle seeking (when it comes down to it, subtitle seeking is the
cause for most trouble here).
Commit a714f8e92 broke preloaded subtitles which have events with
unknown duration, such as some MicroDVD samples. The event list gets
cleared on every seek, so the property of being preloaded obviously gets
lost.
Fix this by moving most of the preloading logic to dec_sub.c. If the
subtitle list gets cleared, they are not considered preloaded anymore,
and the logic for demuxed subtitles is used.
As another minor thing, preloadeding subtitles did neither disable the
demux stream, nor did it discard packets. Thus you could get queue
overflows in theory (harmless, but annoying). Fix this by explicitly
discarding packets in preloaded mode.
In summary, now the only difference between preloaded and normal
demuxing are:
1. a seek is issued, and all packets are read on start
2. during playback, discard the packets instead of feeding them to the
subtitle decoder
This is still petty annoying. It would be nice if maintaining the
subtitle index (and maybe a subtitle packet cache for instant subtitle
presentation when seeking back) could be maintained in the demuxer
instead. Half of all file formats with interleaved subtitles have
this anyway (mp4, mkv muxed with newer mkvmerge).
2016-03-06 13:50:36 +00:00
|
|
|
bool r;
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_lock(&sub->lock);
|
sub: make preloading more robust
Subtitles can be preloaded, which means they're fully read and copied
into ASS_Track. This in turn is mainly for the sake of being able to do
subtitle seeking (when it comes down to it, subtitle seeking is the
cause for most trouble here).
Commit a714f8e92 broke preloaded subtitles which have events with
unknown duration, such as some MicroDVD samples. The event list gets
cleared on every seek, so the property of being preloaded obviously gets
lost.
Fix this by moving most of the preloading logic to dec_sub.c. If the
subtitle list gets cleared, they are not considered preloaded anymore,
and the logic for demuxed subtitles is used.
As another minor thing, preloadeding subtitles did neither disable the
demux stream, nor did it discard packets. Thus you could get queue
overflows in theory (harmless, but annoying). Fix this by explicitly
discarding packets in preloaded mode.
In summary, now the only difference between preloaded and normal
demuxing are:
1. a seek is issued, and all packets are read on start
2. during playback, discard the packets instead of feeding them to the
subtitle decoder
This is still petty annoying. It would be nice if maintaining the
subtitle index (and maybe a subtitle packet cache for instant subtitle
presentation when seeking back) could be maintained in the demuxer
instead. Half of all file formats with interleaved subtitles have
this anyway (mp4, mkv muxed with newer mkvmerge).
2016-03-06 13:50:36 +00:00
|
|
|
r = sub->sd->driver->accept_packets_in_advance && !sub->preload_attempted;
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_unlock(&sub->lock);
|
sub: make preloading more robust
Subtitles can be preloaded, which means they're fully read and copied
into ASS_Track. This in turn is mainly for the sake of being able to do
subtitle seeking (when it comes down to it, subtitle seeking is the
cause for most trouble here).
Commit a714f8e92 broke preloaded subtitles which have events with
unknown duration, such as some MicroDVD samples. The event list gets
cleared on every seek, so the property of being preloaded obviously gets
lost.
Fix this by moving most of the preloading logic to dec_sub.c. If the
subtitle list gets cleared, they are not considered preloaded anymore,
and the logic for demuxed subtitles is used.
As another minor thing, preloadeding subtitles did neither disable the
demux stream, nor did it discard packets. Thus you could get queue
overflows in theory (harmless, but annoying). Fix this by explicitly
discarding packets in preloaded mode.
In summary, now the only difference between preloaded and normal
demuxing are:
1. a seek is issued, and all packets are read on start
2. during playback, discard the packets instead of feeding them to the
subtitle decoder
This is still petty annoying. It would be nice if maintaining the
subtitle index (and maybe a subtitle packet cache for instant subtitle
presentation when seeking back) could be maintained in the demuxer
instead. Half of all file formats with interleaved subtitles have
this anyway (mp4, mkv muxed with newer mkvmerge).
2016-03-06 13:50:36 +00:00
|
|
|
return r;
|
|
|
|
}
|
sub: uglify sub decoder with locking
The plan is to make the whole OSD thread-safe, and we start with this.
We just put locks on all entry points (fortunately, dec_sub.c and all
sd_*.c decoders are very closed off, and only the entry points in
dec_sub.h let you access it). I think this is pretty ugly, but at least
it's very simple.
There's a special case with sub_get_bitmaps(): this function returns
pointers to decoder data (specifically, libass images). There's no way
to synchronize this internally, so expose sub_lock/sub_unlock functions.
To make things simpler, and especially because the lock is sort-of
exposed to the outside world, make the locks recursive. Although the
only case where this is actually needed (although trivial) is
sub_set_extradata().
One corner case are ASS subtitles: for some reason, we keep a single
ASS_Renderer instance for subtitles around (probably to avoid rescanning
fonts with ordered chapters), and this ASS_Renderer instance is not
synchronized. Also, demux_libass.c loads ASS_Track objects, which are
directly passed to sd_ass.c. These things are not synchronized (and
would be hard to synchronize), and basically we're out of luck. But I
think for now, accesses happen reasonably serialized, so there is no
actual problem yet, even if we start to access OSD from other threads.
2014-01-17 22:13:09 +00:00
|
|
|
|
sub: make preloading more robust
Subtitles can be preloaded, which means they're fully read and copied
into ASS_Track. This in turn is mainly for the sake of being able to do
subtitle seeking (when it comes down to it, subtitle seeking is the
cause for most trouble here).
Commit a714f8e92 broke preloaded subtitles which have events with
unknown duration, such as some MicroDVD samples. The event list gets
cleared on every seek, so the property of being preloaded obviously gets
lost.
Fix this by moving most of the preloading logic to dec_sub.c. If the
subtitle list gets cleared, they are not considered preloaded anymore,
and the logic for demuxed subtitles is used.
As another minor thing, preloadeding subtitles did neither disable the
demux stream, nor did it discard packets. Thus you could get queue
overflows in theory (harmless, but annoying). Fix this by explicitly
discarding packets in preloaded mode.
In summary, now the only difference between preloaded and normal
demuxing are:
1. a seek is issued, and all packets are read on start
2. during playback, discard the packets instead of feeding them to the
subtitle decoder
This is still petty annoying. It would be nice if maintaining the
subtitle index (and maybe a subtitle packet cache for instant subtitle
presentation when seeking back) could be maintained in the demuxer
instead. Half of all file formats with interleaved subtitles have
this anyway (mp4, mkv muxed with newer mkvmerge).
2016-03-06 13:50:36 +00:00
|
|
|
void sub_preload(struct dec_sub *sub)
|
|
|
|
{
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_lock(&sub->lock);
|
sub: make preloading more robust
Subtitles can be preloaded, which means they're fully read and copied
into ASS_Track. This in turn is mainly for the sake of being able to do
subtitle seeking (when it comes down to it, subtitle seeking is the
cause for most trouble here).
Commit a714f8e92 broke preloaded subtitles which have events with
unknown duration, such as some MicroDVD samples. The event list gets
cleared on every seek, so the property of being preloaded obviously gets
lost.
Fix this by moving most of the preloading logic to dec_sub.c. If the
subtitle list gets cleared, they are not considered preloaded anymore,
and the logic for demuxed subtitles is used.
As another minor thing, preloadeding subtitles did neither disable the
demux stream, nor did it discard packets. Thus you could get queue
overflows in theory (harmless, but annoying). Fix this by explicitly
discarding packets in preloaded mode.
In summary, now the only difference between preloaded and normal
demuxing are:
1. a seek is issued, and all packets are read on start
2. during playback, discard the packets instead of feeding them to the
subtitle decoder
This is still petty annoying. It would be nice if maintaining the
subtitle index (and maybe a subtitle packet cache for instant subtitle
presentation when seeking back) could be maintained in the demuxer
instead. Half of all file formats with interleaved subtitles have
this anyway (mp4, mkv muxed with newer mkvmerge).
2016-03-06 13:50:36 +00:00
|
|
|
|
2020-02-27 01:33:51 +00:00
|
|
|
struct mp_dispatch_queue *demux_waiter = mp_dispatch_create(NULL);
|
|
|
|
demux_set_stream_wakeup_cb(sub->sh, wakeup_demux, demux_waiter);
|
|
|
|
|
sub: make preloading more robust
Subtitles can be preloaded, which means they're fully read and copied
into ASS_Track. This in turn is mainly for the sake of being able to do
subtitle seeking (when it comes down to it, subtitle seeking is the
cause for most trouble here).
Commit a714f8e92 broke preloaded subtitles which have events with
unknown duration, such as some MicroDVD samples. The event list gets
cleared on every seek, so the property of being preloaded obviously gets
lost.
Fix this by moving most of the preloading logic to dec_sub.c. If the
subtitle list gets cleared, they are not considered preloaded anymore,
and the logic for demuxed subtitles is used.
As another minor thing, preloadeding subtitles did neither disable the
demux stream, nor did it discard packets. Thus you could get queue
overflows in theory (harmless, but annoying). Fix this by explicitly
discarding packets in preloaded mode.
In summary, now the only difference between preloaded and normal
demuxing are:
1. a seek is issued, and all packets are read on start
2. during playback, discard the packets instead of feeding them to the
subtitle decoder
This is still petty annoying. It would be nice if maintaining the
subtitle index (and maybe a subtitle packet cache for instant subtitle
presentation when seeking back) could be maintained in the demuxer
instead. Half of all file formats with interleaved subtitles have
this anyway (mp4, mkv muxed with newer mkvmerge).
2016-03-06 13:50:36 +00:00
|
|
|
sub->preload_attempted = true;
|
2013-06-11 19:39:54 +00:00
|
|
|
|
|
|
|
for (;;) {
|
sub: remove only user of demux_read_packet()
There are 3 packet reading functions in the demux API, which all
function completely differently. One of them, demux_read_packet(), has
only 1 caller, which is in dec_sub.c. Change this caller to use
demux_read_packet_async() instead. Since it really wants to do a
blocking call, setup some proper waiting. This uses mp_dispatch_queue,
because even though it's overkill, it needs the least code.
In practice, waiting actually never happens. This code is only called on
code paths where everything is already read into memory (libavformat's
subtitle demuxers simply behave this way). It's still a bit of a
"coincidence", so implement it properly anyway.
If suubtitle decoder init fails, we still need to unset the demuxer
wakeup callback. Add a sub_destroy() call to the failure path. This also
happens to fix a missed pthread_mutex_destroy() call (in practice this
was a nop, or a memory leak on BSDs).
2019-05-17 16:52:22 +00:00
|
|
|
struct demux_packet *pkt = NULL;
|
|
|
|
int r = demux_read_packet_async(sub->sh, &pkt);
|
|
|
|
if (r == 0) {
|
2020-02-27 01:33:51 +00:00
|
|
|
mp_dispatch_queue_process(demux_waiter, INFINITY);
|
sub: remove only user of demux_read_packet()
There are 3 packet reading functions in the demux API, which all
function completely differently. One of them, demux_read_packet(), has
only 1 caller, which is in dec_sub.c. Change this caller to use
demux_read_packet_async() instead. Since it really wants to do a
blocking call, setup some proper waiting. This uses mp_dispatch_queue,
because even though it's overkill, it needs the least code.
In practice, waiting actually never happens. This code is only called on
code paths where everything is already read into memory (libavformat's
subtitle demuxers simply behave this way). It's still a bit of a
"coincidence", so implement it properly anyway.
If suubtitle decoder init fails, we still need to unset the demuxer
wakeup callback. Add a sub_destroy() call to the failure path. This also
happens to fix a missed pthread_mutex_destroy() call (in practice this
was a nop, or a memory leak on BSDs).
2019-05-17 16:52:22 +00:00
|
|
|
continue;
|
|
|
|
}
|
2013-06-11 19:39:54 +00:00
|
|
|
if (!pkt)
|
|
|
|
break;
|
2015-12-28 22:31:23 +00:00
|
|
|
sub->sd->driver->decode(sub->sd, pkt);
|
2024-01-22 18:55:50 +00:00
|
|
|
MP_TARRAY_APPEND(sub, sub->cached_pkts, sub->num_cached_pkts, pkt);
|
2013-06-11 19:39:54 +00:00
|
|
|
}
|
|
|
|
|
2020-02-27 01:33:51 +00:00
|
|
|
demux_set_stream_wakeup_cb(sub->sh, NULL, NULL);
|
|
|
|
talloc_free(demux_waiter);
|
|
|
|
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_unlock(&sub->lock);
|
2013-06-11 19:39:54 +00:00
|
|
|
}
|
|
|
|
|
2016-11-09 15:34:45 +00:00
|
|
|
static bool is_new_segment(struct dec_sub *sub, struct demux_packet *p)
|
|
|
|
{
|
demux: get rid of demux_packet.new_segment field
The new_segment field was used to track the decoder data flow handler of
timeline boundaries, which are used for ordered chapters etc. (anything
that sets demuxer_desc.load_timeline). This broke seeking with the
demuxer cache enabled. The demuxer is expected to set the new_segment
field after every seek or segment boundary switch, so the cached packets
basically contained incorrect values for this, and the decoders were not
initialized correctly.
Fix this by getting rid of the flag completely. Let the decoders instead
compare the segment information by content, which is hopefully enough.
(In theory, two segments with same information could perhaps appear in
broken-ish corner cases, or in an attempt to simulate looping, and such.
I preferred the simple solution over others, such as generating unique
and stable segment IDs.)
We still add a "segmented" field to make it explicit whether segments
are used, instead of doing something silly like testing arbitrary other
segment fields for validity.
Cached seeking with timeline stuff is still slightly broken even with
this commit: the seek logic is not aware of the overlap that segments
can have, and the timestamp clamping that needs to be performed in
theory to account for the fact that a packet might contain a frame that
is always clipped off by segment handling. This can be fixed later.
2017-10-24 17:33:01 +00:00
|
|
|
return p->segmented &&
|
2016-11-09 15:34:45 +00:00
|
|
|
(p->start != sub->start || p->end != sub->end || p->codec != sub->codec);
|
|
|
|
}
|
|
|
|
|
2024-01-21 23:14:39 +00:00
|
|
|
static bool is_packet_visible(struct demux_packet *p, double video_pts)
|
2013-04-28 19:12:11 +00:00
|
|
|
{
|
2024-01-21 23:14:39 +00:00
|
|
|
return p && p->pts <= video_pts && (video_pts <= p->pts + p->sub_duration ||
|
|
|
|
p->sub_duration < 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
static bool update_pkt_cache(struct dec_sub *sub, double video_pts)
|
|
|
|
{
|
|
|
|
if (!sub->cached_pkts[sub->cached_pkt_pos])
|
|
|
|
return false;
|
|
|
|
|
|
|
|
struct demux_packet *pkt = sub->cached_pkts[sub->cached_pkt_pos];
|
|
|
|
struct demux_packet *next_pkt = sub->cached_pkt_pos + 1 < sub->num_cached_pkts ?
|
|
|
|
sub->cached_pkts[sub->cached_pkt_pos + 1] : NULL;
|
|
|
|
if (!pkt)
|
|
|
|
return false;
|
|
|
|
|
|
|
|
double pts = video_pts + sub->shared_opts->sub_delay[sub->order];
|
|
|
|
double next_pts = next_pkt ? next_pkt->pts : INT_MAX;
|
|
|
|
double end_pts = pkt->sub_duration >= 0 ? pkt->pts + pkt->sub_duration : INT_MAX;
|
|
|
|
|
|
|
|
if (next_pts < pts || end_pts < pts) {
|
|
|
|
if (sub->cached_pkt_pos + 1 < sub->num_cached_pkts) {
|
|
|
|
TA_FREEP(&sub->cached_pkts[sub->cached_pkt_pos]);
|
2024-01-23 16:10:35 +00:00
|
|
|
pkt = NULL;
|
2024-01-21 23:14:39 +00:00
|
|
|
sub->cached_pkt_pos++;
|
|
|
|
}
|
|
|
|
if (next_pts < pts)
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2024-05-26 18:42:45 +00:00
|
|
|
if (pkt && pkt->animated == 1)
|
2024-01-23 16:10:35 +00:00
|
|
|
return true;
|
|
|
|
|
2024-01-21 23:14:39 +00:00
|
|
|
return false;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Read packets from the demuxer stream passed to sub_create(). Signals if
|
|
|
|
// enough packets were read and if the subtitle state updated in anyway. If
|
|
|
|
// packets_read is false, the player should wait until the demuxer signals new
|
|
|
|
// packets and retry.
|
|
|
|
void sub_read_packets(struct dec_sub *sub, double video_pts, bool force,
|
|
|
|
bool *packets_read, bool *sub_updated)
|
|
|
|
{
|
|
|
|
*packets_read = true;
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_lock(&sub->lock);
|
2017-12-29 14:39:38 +00:00
|
|
|
video_pts = pts_to_subtitle(sub, video_pts);
|
2015-12-29 00:35:52 +00:00
|
|
|
while (1) {
|
|
|
|
bool read_more = true;
|
|
|
|
if (sub->sd->driver->accepts_packet)
|
sub: don't potentially discard too many subtitles on seek
The accepts_packet packet callback is supposed to deal with subtitle
decoders which have only a small queue of current subtitle events (i.e.
sd_lavc.c), in case feeding it too many packets would discard events
that are still needed.
Normally, the number of subtitles that need to be preserved is estimated
by the rendering pts (get_bitmaps() argument). Rendering lags behind
decoding, so normally the rendering pts is smaller than the next video
frame pts, and we simply discard all subtitle events until the rendering
pts.
This breaks down in some annoying corner cases. One of them is seeking
backwards: the VO will still try to render the old PTS during seeks,
which passes a high PTS to the subtitle renderer, which in turn would
discard more subtitles than it should. There is a similar issue with
forward seeks. Add hacks to deal with those issues.
There should be a better way to deal with the essentially unknown
"rendering position", which is made worse by screenshots or rendering
with vf_sub. At the very least, we could handle seeks better, and e.g.
either force the VO not to re-render subs after seeks (ugly), or
introduce seek sequence numbers to distinguish attempts to render
earlier subtitles when a seek is done.
2016-08-14 18:27:37 +00:00
|
|
|
read_more = sub->sd->driver->accepts_packet(sub->sd, video_pts);
|
2015-12-29 00:35:52 +00:00
|
|
|
|
|
|
|
if (!read_more)
|
|
|
|
break;
|
|
|
|
|
2016-03-25 16:40:18 +00:00
|
|
|
if (sub->new_segment && sub->new_segment->start < video_pts) {
|
|
|
|
sub->last_vo_pts = video_pts;
|
|
|
|
update_segment(sub);
|
|
|
|
}
|
|
|
|
|
Rewrite ordered chapters and timeline stuff
This uses a different method to piece segments together. The old
approach basically changes to a new file (with a new start offset) any
time a segment ends. This meant waiting for audio/video end on segment
end, and then changing to the new segment all at once. It had a very
weird impact on the playback core, and some things (like truly gapless
segment transitions, or frame backstepping) just didn't work.
The new approach adds the demux_timeline pseudo-demuxer, which presents
an uniform packet stream from the many segments. This is pretty similar
to how ordered chapters are implemented everywhere else. It also reminds
of the FFmpeg concat pseudo-demuxer.
The "pure" version of this approach doesn't work though. Segments can
actually have different codec configurations (different extradata), and
subtitles are most likely broken too. (Subtitles have multiple corner
cases which break the pure stream-concatenation approach completely.)
To counter this, we do two things:
- Reinit the decoder with each segment. We go as far as allowing
concatenating files with completely different codecs for the sake
of EDL (which also uses the timeline infrastructure). A "lighter"
approach would try to make use of decoder mechanism to update e.g.
the extradata, but that seems fragile.
- Clip decoded data to segment boundaries. This is equivalent to
normal playback core mechanisms like hr-seek, but now the playback
core doesn't need to care about these things.
These two mechanisms are equivalent to what happened in the old
implementation, except they don't happen in the playback core anymore.
In other words, the playback core is completely relieved from timeline
implementation details. (Which honestly is exactly what I'm trying to
do here. I don't think ordered chapter behavior deserves improvement,
even if it's bad - but I want to get it out from the playback core.)
There is code duplication between audio and video decoder common code.
This is awful and could be shareable - but this will happen later.
Note that the audio path has some code to clip audio frames for the
purpose of codec preroll/gapless handling, but it's not shared as
sharing it would cause more pain than it would help.
2016-02-15 20:04:07 +00:00
|
|
|
if (sub->new_segment)
|
|
|
|
break;
|
|
|
|
|
2020-02-27 01:23:58 +00:00
|
|
|
// (Use this mechanism only if sub_delay matters to avoid corner cases.)
|
2023-12-13 21:36:58 +00:00
|
|
|
float delay = sub->order < 0 ? 0.0f : sub->shared_opts->sub_delay[sub->order];
|
2023-11-23 03:13:57 +00:00
|
|
|
double min_pts = delay < 0 || force ? video_pts : MP_NOPTS_VALUE;
|
2020-02-27 01:23:58 +00:00
|
|
|
|
2015-12-29 00:35:52 +00:00
|
|
|
struct demux_packet *pkt;
|
2023-09-28 02:51:49 +00:00
|
|
|
int st = demux_read_packet_async_until(sub->sh, min_pts, &pkt);
|
2015-12-29 00:35:52 +00:00
|
|
|
// Note: "wait" (st==0) happens with non-interleaved streams only, and
|
|
|
|
// then we should stop the playloop until a new enough packet has been
|
2020-02-27 01:23:58 +00:00
|
|
|
// seen (or the subtitle decoder's queue is full). This usually does not
|
|
|
|
// happen for interleaved subtitle streams, which never return "wait"
|
|
|
|
// when reading, unless min_pts is set.
|
2015-12-29 00:35:52 +00:00
|
|
|
if (st <= 0) {
|
2024-01-21 23:14:39 +00:00
|
|
|
*packets_read = st < 0 || (sub->last_pkt_pts != MP_NOPTS_VALUE &&
|
|
|
|
sub->last_pkt_pts > video_pts);
|
2015-12-29 00:35:52 +00:00
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
2017-02-07 16:05:17 +00:00
|
|
|
if (sub->recorder_sink)
|
|
|
|
mp_recorder_feed_packet(sub->recorder_sink, pkt);
|
|
|
|
|
2015-12-29 00:35:52 +00:00
|
|
|
sub->last_pkt_pts = pkt->pts;
|
2024-01-22 18:55:50 +00:00
|
|
|
MP_TARRAY_APPEND(sub, sub->cached_pkts, sub->num_cached_pkts, pkt);
|
Rewrite ordered chapters and timeline stuff
This uses a different method to piece segments together. The old
approach basically changes to a new file (with a new start offset) any
time a segment ends. This meant waiting for audio/video end on segment
end, and then changing to the new segment all at once. It had a very
weird impact on the playback core, and some things (like truly gapless
segment transitions, or frame backstepping) just didn't work.
The new approach adds the demux_timeline pseudo-demuxer, which presents
an uniform packet stream from the many segments. This is pretty similar
to how ordered chapters are implemented everywhere else. It also reminds
of the FFmpeg concat pseudo-demuxer.
The "pure" version of this approach doesn't work though. Segments can
actually have different codec configurations (different extradata), and
subtitles are most likely broken too. (Subtitles have multiple corner
cases which break the pure stream-concatenation approach completely.)
To counter this, we do two things:
- Reinit the decoder with each segment. We go as far as allowing
concatenating files with completely different codecs for the sake
of EDL (which also uses the timeline infrastructure). A "lighter"
approach would try to make use of decoder mechanism to update e.g.
the extradata, but that seems fragile.
- Clip decoded data to segment boundaries. This is equivalent to
normal playback core mechanisms like hr-seek, but now the playback
core doesn't need to care about these things.
These two mechanisms are equivalent to what happened in the old
implementation, except they don't happen in the playback core anymore.
In other words, the playback core is completely relieved from timeline
implementation details. (Which honestly is exactly what I'm trying to
do here. I don't think ordered chapter behavior deserves improvement,
even if it's bad - but I want to get it out from the playback core.)
There is code duplication between audio and video decoder common code.
This is awful and could be shareable - but this will happen later.
Note that the audio path has some code to clip audio frames for the
purpose of codec preroll/gapless handling, but it's not shared as
sharing it would cause more pain than it would help.
2016-02-15 20:04:07 +00:00
|
|
|
|
2016-11-09 15:34:45 +00:00
|
|
|
if (is_new_segment(sub, pkt)) {
|
2023-11-06 16:12:23 +00:00
|
|
|
sub->new_segment = demux_copy_packet(pkt);
|
Rewrite ordered chapters and timeline stuff
This uses a different method to piece segments together. The old
approach basically changes to a new file (with a new start offset) any
time a segment ends. This meant waiting for audio/video end on segment
end, and then changing to the new segment all at once. It had a very
weird impact on the playback core, and some things (like truly gapless
segment transitions, or frame backstepping) just didn't work.
The new approach adds the demux_timeline pseudo-demuxer, which presents
an uniform packet stream from the many segments. This is pretty similar
to how ordered chapters are implemented everywhere else. It also reminds
of the FFmpeg concat pseudo-demuxer.
The "pure" version of this approach doesn't work though. Segments can
actually have different codec configurations (different extradata), and
subtitles are most likely broken too. (Subtitles have multiple corner
cases which break the pure stream-concatenation approach completely.)
To counter this, we do two things:
- Reinit the decoder with each segment. We go as far as allowing
concatenating files with completely different codecs for the sake
of EDL (which also uses the timeline infrastructure). A "lighter"
approach would try to make use of decoder mechanism to update e.g.
the extradata, but that seems fragile.
- Clip decoded data to segment boundaries. This is equivalent to
normal playback core mechanisms like hr-seek, but now the playback
core doesn't need to care about these things.
These two mechanisms are equivalent to what happened in the old
implementation, except they don't happen in the playback core anymore.
In other words, the playback core is completely relieved from timeline
implementation details. (Which honestly is exactly what I'm trying to
do here. I don't think ordered chapter behavior deserves improvement,
even if it's bad - but I want to get it out from the playback core.)
There is code duplication between audio and video decoder common code.
This is awful and could be shareable - but this will happen later.
Note that the audio path has some code to clip audio frames for the
purpose of codec preroll/gapless handling, but it's not shared as
sharing it would cause more pain than it would help.
2016-02-15 20:04:07 +00:00
|
|
|
// Note that this can be delayed to a much later point in time.
|
|
|
|
update_segment(sub);
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
|
sub: make preloading more robust
Subtitles can be preloaded, which means they're fully read and copied
into ASS_Track. This in turn is mainly for the sake of being able to do
subtitle seeking (when it comes down to it, subtitle seeking is the
cause for most trouble here).
Commit a714f8e92 broke preloaded subtitles which have events with
unknown duration, such as some MicroDVD samples. The event list gets
cleared on every seek, so the property of being preloaded obviously gets
lost.
Fix this by moving most of the preloading logic to dec_sub.c. If the
subtitle list gets cleared, they are not considered preloaded anymore,
and the logic for demuxed subtitles is used.
As another minor thing, preloadeding subtitles did neither disable the
demux stream, nor did it discard packets. Thus you could get queue
overflows in theory (harmless, but annoying). Fix this by explicitly
discarding packets in preloaded mode.
In summary, now the only difference between preloaded and normal
demuxing are:
1. a seek is issued, and all packets are read on start
2. during playback, discard the packets instead of feeding them to the
subtitle decoder
This is still petty annoying. It would be nice if maintaining the
subtitle index (and maybe a subtitle packet cache for instant subtitle
presentation when seeking back) could be maintained in the demuxer
instead. Half of all file formats with interleaved subtitles have
this anyway (mp4, mkv muxed with newer mkvmerge).
2016-03-06 13:50:36 +00:00
|
|
|
if (!(sub->preload_attempted && sub->sd->preload_ok))
|
|
|
|
sub->sd->driver->decode(sub->sd, pkt);
|
2015-12-29 00:35:52 +00:00
|
|
|
}
|
2024-01-21 23:14:39 +00:00
|
|
|
if (sub->cached_pkts && sub->num_cached_pkts) {
|
|
|
|
bool visible = is_packet_visible(sub->cached_pkts[sub->cached_pkt_pos], video_pts);
|
|
|
|
*sub_updated = update_pkt_cache(sub, video_pts) || sub->sub_visible != visible;
|
|
|
|
sub->sub_visible = visible;
|
|
|
|
}
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_unlock(&sub->lock);
|
2013-06-01 17:44:55 +00:00
|
|
|
}
|
|
|
|
|
2024-01-22 18:55:50 +00:00
|
|
|
// Redecode all cached packets if needed.
|
2023-10-20 17:07:31 +00:00
|
|
|
// Used with UPDATE_SUB_HARD and UPDATE_SUB_FILT.
|
|
|
|
void sub_redecode_cached_packets(struct dec_sub *sub)
|
|
|
|
{
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_lock(&sub->lock);
|
2024-01-21 23:14:39 +00:00
|
|
|
int index = sub->cached_pkt_pos;
|
2024-01-22 18:55:50 +00:00
|
|
|
while (index < sub->num_cached_pkts) {
|
|
|
|
sub->sd->driver->decode(sub->sd, sub->cached_pkts[index]);
|
|
|
|
++index;
|
|
|
|
}
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_unlock(&sub->lock);
|
2023-10-20 17:07:31 +00:00
|
|
|
}
|
|
|
|
|
video: make OSD/subtitle bitmaps refcounted (sort of)
Making OSD/subtitle bitmaps refcounted was planend a longer time ago,
e.g. the sub_bitmaps.packed field (which refcounts the subtitle bitmap
data) was added in 2016. But nothing benefited much from it, because
struct sub_bitmaps was usually stack allocated, and there was this weird
callback stuff through osd_draw().
Make it possible to get actually refcounted subtitle bitmaps on the OSD
API level. For this, we just copy all subtitle data other than the
bitmaps with sub_bitmaps_copy(). At first, I had planned some fancy
refcount shit, but when that was a big mess and hard to debug and just
boiled to emulating malloc(), I made it a full allocation+copy. This
affects mostly the parts array. With crazy ASS subtitles, this parts
array can get pretty big (thousands of elements or more), in which case
the extra alloc/copy could become performance relevant. But then again
this is just pure bullshit, and I see no need to care. In practice, this
extra work most likely gets drowned out by libass murdering a single
core (while mpv is waiting for it) anyway. So fuck it.
I just wanted this so draw_bmp.c requires only a single call to render
everything. VOs also can benefit from this, because the weird callback
shit isn't necessary anymore (simpler code), but I haven't done anything
about it yet. In general I'd hope this will work towards simplifying the
OSD layer, which is prerequisite for making actual further improvements.
I haven't tested some cases such as the "overlay-add" command. Maybe it
crashes now? Who knows, who cares.
In addition, it might be worthwhile to reduce the code duplication
between all the things that output subtitle bitmaps (with repacking,
image allocation, etc.), but that's orthogonal.
2020-04-26 21:34:32 +00:00
|
|
|
// Unref sub_bitmaps.rc to free the result. May return NULL.
|
|
|
|
struct sub_bitmaps *sub_get_bitmaps(struct dec_sub *sub, struct mp_osd_res dim,
|
|
|
|
int format, double pts)
|
2012-08-25 18:22:39 +00:00
|
|
|
{
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_lock(&sub->lock);
|
video: make OSD/subtitle bitmaps refcounted (sort of)
Making OSD/subtitle bitmaps refcounted was planend a longer time ago,
e.g. the sub_bitmaps.packed field (which refcounts the subtitle bitmap
data) was added in 2016. But nothing benefited much from it, because
struct sub_bitmaps was usually stack allocated, and there was this weird
callback stuff through osd_draw().
Make it possible to get actually refcounted subtitle bitmaps on the OSD
API level. For this, we just copy all subtitle data other than the
bitmaps with sub_bitmaps_copy(). At first, I had planned some fancy
refcount shit, but when that was a big mess and hard to debug and just
boiled to emulating malloc(), I made it a full allocation+copy. This
affects mostly the parts array. With crazy ASS subtitles, this parts
array can get pretty big (thousands of elements or more), in which case
the extra alloc/copy could become performance relevant. But then again
this is just pure bullshit, and I see no need to care. In practice, this
extra work most likely gets drowned out by libass murdering a single
core (while mpv is waiting for it) anyway. So fuck it.
I just wanted this so draw_bmp.c requires only a single call to render
everything. VOs also can benefit from this, because the weird callback
shit isn't necessary anymore (simpler code), but I haven't done anything
about it yet. In general I'd hope this will work towards simplifying the
OSD layer, which is prerequisite for making actual further improvements.
I haven't tested some cases such as the "overlay-add" command. Maybe it
crashes now? Who knows, who cares.
In addition, it might be worthwhile to reduce the code duplication
between all the things that output subtitle bitmaps (with repacking,
image allocation, etc.), but that's orthogonal.
2020-04-26 21:34:32 +00:00
|
|
|
|
2017-12-29 14:39:38 +00:00
|
|
|
pts = pts_to_subtitle(sub, pts);
|
|
|
|
|
Rewrite ordered chapters and timeline stuff
This uses a different method to piece segments together. The old
approach basically changes to a new file (with a new start offset) any
time a segment ends. This meant waiting for audio/video end on segment
end, and then changing to the new segment all at once. It had a very
weird impact on the playback core, and some things (like truly gapless
segment transitions, or frame backstepping) just didn't work.
The new approach adds the demux_timeline pseudo-demuxer, which presents
an uniform packet stream from the many segments. This is pretty similar
to how ordered chapters are implemented everywhere else. It also reminds
of the FFmpeg concat pseudo-demuxer.
The "pure" version of this approach doesn't work though. Segments can
actually have different codec configurations (different extradata), and
subtitles are most likely broken too. (Subtitles have multiple corner
cases which break the pure stream-concatenation approach completely.)
To counter this, we do two things:
- Reinit the decoder with each segment. We go as far as allowing
concatenating files with completely different codecs for the sake
of EDL (which also uses the timeline infrastructure). A "lighter"
approach would try to make use of decoder mechanism to update e.g.
the extradata, but that seems fragile.
- Clip decoded data to segment boundaries. This is equivalent to
normal playback core mechanisms like hr-seek, but now the playback
core doesn't need to care about these things.
These two mechanisms are equivalent to what happened in the old
implementation, except they don't happen in the playback core anymore.
In other words, the playback core is completely relieved from timeline
implementation details. (Which honestly is exactly what I'm trying to
do here. I don't think ordered chapter behavior deserves improvement,
even if it's bad - but I want to get it out from the playback core.)
There is code duplication between audio and video decoder common code.
This is awful and could be shareable - but this will happen later.
Note that the audio path has some code to clip audio frames for the
purpose of codec preroll/gapless handling, but it's not shared as
sharing it would cause more pain than it would help.
2016-02-15 20:04:07 +00:00
|
|
|
sub->last_vo_pts = pts;
|
|
|
|
update_segment(sub);
|
|
|
|
|
video: make OSD/subtitle bitmaps refcounted (sort of)
Making OSD/subtitle bitmaps refcounted was planend a longer time ago,
e.g. the sub_bitmaps.packed field (which refcounts the subtitle bitmap
data) was added in 2016. But nothing benefited much from it, because
struct sub_bitmaps was usually stack allocated, and there was this weird
callback stuff through osd_draw().
Make it possible to get actually refcounted subtitle bitmaps on the OSD
API level. For this, we just copy all subtitle data other than the
bitmaps with sub_bitmaps_copy(). At first, I had planned some fancy
refcount shit, but when that was a big mess and hard to debug and just
boiled to emulating malloc(), I made it a full allocation+copy. This
affects mostly the parts array. With crazy ASS subtitles, this parts
array can get pretty big (thousands of elements or more), in which case
the extra alloc/copy could become performance relevant. But then again
this is just pure bullshit, and I see no need to care. In practice, this
extra work most likely gets drowned out by libass murdering a single
core (while mpv is waiting for it) anyway. So fuck it.
I just wanted this so draw_bmp.c requires only a single call to render
everything. VOs also can benefit from this, because the weird callback
shit isn't necessary anymore (simpler code), but I haven't done anything
about it yet. In general I'd hope this will work towards simplifying the
OSD layer, which is prerequisite for making actual further improvements.
I haven't tested some cases such as the "overlay-add" command. Maybe it
crashes now? Who knows, who cares.
In addition, it might be worthwhile to reduce the code duplication
between all the things that output subtitle bitmaps (with repacking,
image allocation, etc.), but that's orthogonal.
2020-04-26 21:34:32 +00:00
|
|
|
struct sub_bitmaps *res = NULL;
|
|
|
|
|
|
|
|
if (!(sub->end != MP_NOPTS_VALUE && pts >= sub->end) &&
|
2022-01-19 17:23:52 +00:00
|
|
|
sub->sd->driver->get_bitmaps)
|
2023-12-17 19:18:11 +00:00
|
|
|
res = sub->sd->driver->get_bitmaps(sub->sd, dim, format, pts);
|
2016-02-20 15:23:44 +00:00
|
|
|
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_unlock(&sub->lock);
|
video: make OSD/subtitle bitmaps refcounted (sort of)
Making OSD/subtitle bitmaps refcounted was planend a longer time ago,
e.g. the sub_bitmaps.packed field (which refcounts the subtitle bitmap
data) was added in 2016. But nothing benefited much from it, because
struct sub_bitmaps was usually stack allocated, and there was this weird
callback stuff through osd_draw().
Make it possible to get actually refcounted subtitle bitmaps on the OSD
API level. For this, we just copy all subtitle data other than the
bitmaps with sub_bitmaps_copy(). At first, I had planned some fancy
refcount shit, but when that was a big mess and hard to debug and just
boiled to emulating malloc(), I made it a full allocation+copy. This
affects mostly the parts array. With crazy ASS subtitles, this parts
array can get pretty big (thousands of elements or more), in which case
the extra alloc/copy could become performance relevant. But then again
this is just pure bullshit, and I see no need to care. In practice, this
extra work most likely gets drowned out by libass murdering a single
core (while mpv is waiting for it) anyway. So fuck it.
I just wanted this so draw_bmp.c requires only a single call to render
everything. VOs also can benefit from this, because the weird callback
shit isn't necessary anymore (simpler code), but I haven't done anything
about it yet. In general I'd hope this will work towards simplifying the
OSD layer, which is prerequisite for making actual further improvements.
I haven't tested some cases such as the "overlay-add" command. Maybe it
crashes now? Who knows, who cares.
In addition, it might be worthwhile to reduce the code duplication
between all the things that output subtitle bitmaps (with repacking,
image allocation, etc.), but that's orthogonal.
2020-04-26 21:34:32 +00:00
|
|
|
return res;
|
2013-04-28 19:12:11 +00:00
|
|
|
}
|
|
|
|
|
2020-05-15 14:19:02 +00:00
|
|
|
// The returned string is talloc'ed.
|
2020-05-14 20:14:49 +00:00
|
|
|
char *sub_get_text(struct dec_sub *sub, double pts, enum sd_text_type type)
|
2011-01-16 18:03:08 +00:00
|
|
|
{
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_lock(&sub->lock);
|
2013-06-01 17:44:12 +00:00
|
|
|
char *text = NULL;
|
Rewrite ordered chapters and timeline stuff
This uses a different method to piece segments together. The old
approach basically changes to a new file (with a new start offset) any
time a segment ends. This meant waiting for audio/video end on segment
end, and then changing to the new segment all at once. It had a very
weird impact on the playback core, and some things (like truly gapless
segment transitions, or frame backstepping) just didn't work.
The new approach adds the demux_timeline pseudo-demuxer, which presents
an uniform packet stream from the many segments. This is pretty similar
to how ordered chapters are implemented everywhere else. It also reminds
of the FFmpeg concat pseudo-demuxer.
The "pure" version of this approach doesn't work though. Segments can
actually have different codec configurations (different extradata), and
subtitles are most likely broken too. (Subtitles have multiple corner
cases which break the pure stream-concatenation approach completely.)
To counter this, we do two things:
- Reinit the decoder with each segment. We go as far as allowing
concatenating files with completely different codecs for the sake
of EDL (which also uses the timeline infrastructure). A "lighter"
approach would try to make use of decoder mechanism to update e.g.
the extradata, but that seems fragile.
- Clip decoded data to segment boundaries. This is equivalent to
normal playback core mechanisms like hr-seek, but now the playback
core doesn't need to care about these things.
These two mechanisms are equivalent to what happened in the old
implementation, except they don't happen in the playback core anymore.
In other words, the playback core is completely relieved from timeline
implementation details. (Which honestly is exactly what I'm trying to
do here. I don't think ordered chapter behavior deserves improvement,
even if it's bad - but I want to get it out from the playback core.)
There is code duplication between audio and video decoder common code.
This is awful and could be shareable - but this will happen later.
Note that the audio path has some code to clip audio frames for the
purpose of codec preroll/gapless handling, but it's not shared as
sharing it would cause more pain than it would help.
2016-02-15 20:04:07 +00:00
|
|
|
|
2017-12-29 14:39:38 +00:00
|
|
|
pts = pts_to_subtitle(sub, pts);
|
|
|
|
|
Rewrite ordered chapters and timeline stuff
This uses a different method to piece segments together. The old
approach basically changes to a new file (with a new start offset) any
time a segment ends. This meant waiting for audio/video end on segment
end, and then changing to the new segment all at once. It had a very
weird impact on the playback core, and some things (like truly gapless
segment transitions, or frame backstepping) just didn't work.
The new approach adds the demux_timeline pseudo-demuxer, which presents
an uniform packet stream from the many segments. This is pretty similar
to how ordered chapters are implemented everywhere else. It also reminds
of the FFmpeg concat pseudo-demuxer.
The "pure" version of this approach doesn't work though. Segments can
actually have different codec configurations (different extradata), and
subtitles are most likely broken too. (Subtitles have multiple corner
cases which break the pure stream-concatenation approach completely.)
To counter this, we do two things:
- Reinit the decoder with each segment. We go as far as allowing
concatenating files with completely different codecs for the sake
of EDL (which also uses the timeline infrastructure). A "lighter"
approach would try to make use of decoder mechanism to update e.g.
the extradata, but that seems fragile.
- Clip decoded data to segment boundaries. This is equivalent to
normal playback core mechanisms like hr-seek, but now the playback
core doesn't need to care about these things.
These two mechanisms are equivalent to what happened in the old
implementation, except they don't happen in the playback core anymore.
In other words, the playback core is completely relieved from timeline
implementation details. (Which honestly is exactly what I'm trying to
do here. I don't think ordered chapter behavior deserves improvement,
even if it's bad - but I want to get it out from the playback core.)
There is code duplication between audio and video decoder common code.
This is awful and could be shareable - but this will happen later.
Note that the audio path has some code to clip audio frames for the
purpose of codec preroll/gapless handling, but it's not shared as
sharing it would cause more pain than it would help.
2016-02-15 20:04:07 +00:00
|
|
|
sub->last_vo_pts = pts;
|
|
|
|
update_segment(sub);
|
|
|
|
|
2018-10-26 14:51:47 +00:00
|
|
|
if (sub->sd->driver->get_text)
|
2020-05-14 20:14:49 +00:00
|
|
|
text = sub->sd->driver->get_text(sub->sd, pts, type);
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_unlock(&sub->lock);
|
2013-06-01 17:44:12 +00:00
|
|
|
return text;
|
2011-01-16 18:03:08 +00:00
|
|
|
}
|
|
|
|
|
2023-08-26 13:17:01 +00:00
|
|
|
char *sub_ass_get_extradata(struct dec_sub *sub)
|
|
|
|
{
|
2024-04-24 13:42:34 +00:00
|
|
|
char *data = NULL;
|
2024-04-07 15:57:18 +00:00
|
|
|
mp_mutex_lock(&sub->lock);
|
2023-08-26 13:17:01 +00:00
|
|
|
if (strcmp(sub->sd->codec->codec, "ass") != 0)
|
2024-04-24 13:42:34 +00:00
|
|
|
goto done;
|
2023-08-26 13:17:01 +00:00
|
|
|
char *extradata = sub->sd->codec->extradata;
|
|
|
|
int extradata_size = sub->sd->codec->extradata_size;
|
2024-04-24 13:42:34 +00:00
|
|
|
data = talloc_strndup(NULL, extradata, extradata_size);
|
|
|
|
done:
|
2024-04-07 15:57:18 +00:00
|
|
|
mp_mutex_unlock(&sub->lock);
|
|
|
|
return data;
|
2023-08-26 13:17:01 +00:00
|
|
|
}
|
|
|
|
|
2019-09-21 18:11:18 +00:00
|
|
|
struct sd_times sub_get_times(struct dec_sub *sub, double pts)
|
|
|
|
{
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_lock(&sub->lock);
|
2019-09-21 18:11:18 +00:00
|
|
|
struct sd_times res = { .start = MP_NOPTS_VALUE, .end = MP_NOPTS_VALUE };
|
|
|
|
|
|
|
|
pts = pts_to_subtitle(sub, pts);
|
|
|
|
|
|
|
|
sub->last_vo_pts = pts;
|
|
|
|
update_segment(sub);
|
|
|
|
|
|
|
|
if (sub->sd->driver->get_times)
|
|
|
|
res = sub->sd->driver->get_times(sub->sd, pts);
|
|
|
|
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_unlock(&sub->lock);
|
2019-09-21 18:11:18 +00:00
|
|
|
return res;
|
|
|
|
}
|
|
|
|
|
2013-06-01 17:44:12 +00:00
|
|
|
void sub_reset(struct dec_sub *sub)
|
2011-01-16 18:03:08 +00:00
|
|
|
{
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_lock(&sub->lock);
|
2015-12-27 01:07:01 +00:00
|
|
|
if (sub->sd->driver->reset)
|
2015-12-18 00:54:14 +00:00
|
|
|
sub->sd->driver->reset(sub->sd);
|
2015-12-29 00:35:52 +00:00
|
|
|
sub->last_pkt_pts = MP_NOPTS_VALUE;
|
Rewrite ordered chapters and timeline stuff
This uses a different method to piece segments together. The old
approach basically changes to a new file (with a new start offset) any
time a segment ends. This meant waiting for audio/video end on segment
end, and then changing to the new segment all at once. It had a very
weird impact on the playback core, and some things (like truly gapless
segment transitions, or frame backstepping) just didn't work.
The new approach adds the demux_timeline pseudo-demuxer, which presents
an uniform packet stream from the many segments. This is pretty similar
to how ordered chapters are implemented everywhere else. It also reminds
of the FFmpeg concat pseudo-demuxer.
The "pure" version of this approach doesn't work though. Segments can
actually have different codec configurations (different extradata), and
subtitles are most likely broken too. (Subtitles have multiple corner
cases which break the pure stream-concatenation approach completely.)
To counter this, we do two things:
- Reinit the decoder with each segment. We go as far as allowing
concatenating files with completely different codecs for the sake
of EDL (which also uses the timeline infrastructure). A "lighter"
approach would try to make use of decoder mechanism to update e.g.
the extradata, but that seems fragile.
- Clip decoded data to segment boundaries. This is equivalent to
normal playback core mechanisms like hr-seek, but now the playback
core doesn't need to care about these things.
These two mechanisms are equivalent to what happened in the old
implementation, except they don't happen in the playback core anymore.
In other words, the playback core is completely relieved from timeline
implementation details. (Which honestly is exactly what I'm trying to
do here. I don't think ordered chapter behavior deserves improvement,
even if it's bad - but I want to get it out from the playback core.)
There is code duplication between audio and video decoder common code.
This is awful and could be shareable - but this will happen later.
Note that the audio path has some code to clip audio frames for the
purpose of codec preroll/gapless handling, but it's not shared as
sharing it would cause more pain than it would help.
2016-02-15 20:04:07 +00:00
|
|
|
sub->last_vo_pts = MP_NOPTS_VALUE;
|
2024-04-04 12:06:55 +00:00
|
|
|
destroy_cached_pkts(sub);
|
2023-10-20 17:07:31 +00:00
|
|
|
TA_FREEP(&sub->new_segment);
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_unlock(&sub->lock);
|
2013-06-01 17:44:55 +00:00
|
|
|
}
|
|
|
|
|
2015-12-26 17:35:36 +00:00
|
|
|
void sub_select(struct dec_sub *sub, bool selected)
|
|
|
|
{
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_lock(&sub->lock);
|
2015-12-27 01:07:01 +00:00
|
|
|
if (sub->sd->driver->select)
|
2015-12-26 17:35:36 +00:00
|
|
|
sub->sd->driver->select(sub->sd, selected);
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_unlock(&sub->lock);
|
2015-12-26 17:35:36 +00:00
|
|
|
}
|
|
|
|
|
2013-06-28 23:34:11 +00:00
|
|
|
int sub_control(struct dec_sub *sub, enum sd_ctrl cmd, void *arg)
|
|
|
|
{
|
sub: uglify sub decoder with locking
The plan is to make the whole OSD thread-safe, and we start with this.
We just put locks on all entry points (fortunately, dec_sub.c and all
sd_*.c decoders are very closed off, and only the entry points in
dec_sub.h let you access it). I think this is pretty ugly, but at least
it's very simple.
There's a special case with sub_get_bitmaps(): this function returns
pointers to decoder data (specifically, libass images). There's no way
to synchronize this internally, so expose sub_lock/sub_unlock functions.
To make things simpler, and especially because the lock is sort-of
exposed to the outside world, make the locks recursive. Although the
only case where this is actually needed (although trivial) is
sub_set_extradata().
One corner case are ASS subtitles: for some reason, we keep a single
ASS_Renderer instance for subtitles around (probably to avoid rescanning
fonts with ordered chapters), and this ASS_Renderer instance is not
synchronized. Also, demux_libass.c loads ASS_Track objects, which are
directly passed to sd_ass.c. These things are not synchronized (and
would be hard to synchronize), and basically we're out of luck. But I
think for now, accesses happen reasonably serialized, so there is no
actual problem yet, even if we start to access OSD from other threads.
2014-01-17 22:13:09 +00:00
|
|
|
int r = CONTROL_UNKNOWN;
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_lock(&sub->lock);
|
sub: make filter_sdh a "proper" filter, allow runtime changes
Until now, filter_sdh was simply a function that was called by sd_ass
directly (if enabled).
I want to add another filter, so it's time to turn this into a somewhat
more general subtitle filtering infrastructure.
I pondered whether to reuse the audio/video filtering stuff - but better
not. Also, since subtitles are horrible and tend to refuse proper
abstraction, it's still messed into sd_ass, instead of working on the
dec_sub.c level. Actually mpv used to have subtitle "filters" and even
made subtitle converters part of it, but it was fairly horrible, so
don't do that again.
In addition, make runtime changes possible. Since this was supposed to
be a quick hack, I just decided to put all subtitle filter options into
a separate option group (=> simpler change notification), to manually
push the change through the playloop (like it was sort of before for OSD
options), and to recreate the sub filter chain completely in every
change. Should be good enough.
One strangeness is that due to prefetching and such, most subtitle
packets (or those some time ahead) are actually done filtering when we
change, so the user still needs to manually seek to actually refresh
everything. And since subtitle data is usually cached in ASS_Track (for
other terrible but user-friendly reasons), we also must clear the
subtitle data, but of course only on seek, since otherwise all subtitles
would just disappear. What a fucking mess, but such is life. We could
trigger a "refresh seek" to make this more automatic, but I don't feel
like it currently.
This is slightly inefficient (lots of allocations and copying), but I
decided that it doesn't matter. Could matter slightly for crazy ASS
subtitles that render with thousands of events.
Not very well tested. Still seems to work, but I didn't have many test
cases.
2020-02-16 00:02:17 +00:00
|
|
|
bool propagate = false;
|
2017-12-29 14:39:38 +00:00
|
|
|
switch (cmd) {
|
|
|
|
case SD_CTRL_SET_VIDEO_DEF_FPS:
|
|
|
|
sub->video_fps = *(double *)arg;
|
|
|
|
update_subtitle_speed(sub);
|
|
|
|
break;
|
|
|
|
case SD_CTRL_SUB_STEP: {
|
|
|
|
double *a = arg;
|
|
|
|
double arg2[2] = {a[0], a[1]};
|
|
|
|
arg2[0] = pts_to_subtitle(sub, arg2[0]);
|
|
|
|
if (sub->sd->driver->control)
|
|
|
|
r = sub->sd->driver->control(sub->sd, cmd, arg2);
|
|
|
|
if (r == CONTROL_OK)
|
|
|
|
a[0] = pts_from_subtitle(sub, arg2[0]);
|
|
|
|
break;
|
2023-03-08 18:05:59 +00:00
|
|
|
}
|
|
|
|
case SD_CTRL_UPDATE_OPTS: {
|
|
|
|
int flags = (uintptr_t)arg;
|
sub: make filter_sdh a "proper" filter, allow runtime changes
Until now, filter_sdh was simply a function that was called by sd_ass
directly (if enabled).
I want to add another filter, so it's time to turn this into a somewhat
more general subtitle filtering infrastructure.
I pondered whether to reuse the audio/video filtering stuff - but better
not. Also, since subtitles are horrible and tend to refuse proper
abstraction, it's still messed into sd_ass, instead of working on the
dec_sub.c level. Actually mpv used to have subtitle "filters" and even
made subtitle converters part of it, but it was fairly horrible, so
don't do that again.
In addition, make runtime changes possible. Since this was supposed to
be a quick hack, I just decided to put all subtitle filter options into
a separate option group (=> simpler change notification), to manually
push the change through the playloop (like it was sort of before for OSD
options), and to recreate the sub filter chain completely in every
change. Should be good enough.
One strangeness is that due to prefetching and such, most subtitle
packets (or those some time ahead) are actually done filtering when we
change, so the user still needs to manually seek to actually refresh
everything. And since subtitle data is usually cached in ASS_Track (for
other terrible but user-friendly reasons), we also must clear the
subtitle data, but of course only on seek, since otherwise all subtitles
would just disappear. What a fucking mess, but such is life. We could
trigger a "refresh seek" to make this more automatic, but I don't feel
like it currently.
This is slightly inefficient (lots of allocations and copying), but I
decided that it doesn't matter. Could matter slightly for crazy ASS
subtitles that render with thousands of events.
Not very well tested. Still seems to work, but I didn't have many test
cases.
2020-02-16 00:02:17 +00:00
|
|
|
if (m_config_cache_update(sub->opts_cache))
|
|
|
|
update_subtitle_speed(sub);
|
2023-12-13 21:36:58 +00:00
|
|
|
m_config_cache_update(sub->shared_opts_cache);
|
sub: make filter_sdh a "proper" filter, allow runtime changes
Until now, filter_sdh was simply a function that was called by sd_ass
directly (if enabled).
I want to add another filter, so it's time to turn this into a somewhat
more general subtitle filtering infrastructure.
I pondered whether to reuse the audio/video filtering stuff - but better
not. Also, since subtitles are horrible and tend to refuse proper
abstraction, it's still messed into sd_ass, instead of working on the
dec_sub.c level. Actually mpv used to have subtitle "filters" and even
made subtitle converters part of it, but it was fairly horrible, so
don't do that again.
In addition, make runtime changes possible. Since this was supposed to
be a quick hack, I just decided to put all subtitle filter options into
a separate option group (=> simpler change notification), to manually
push the change through the playloop (like it was sort of before for OSD
options), and to recreate the sub filter chain completely in every
change. Should be good enough.
One strangeness is that due to prefetching and such, most subtitle
packets (or those some time ahead) are actually done filtering when we
change, so the user still needs to manually seek to actually refresh
everything. And since subtitle data is usually cached in ASS_Track (for
other terrible but user-friendly reasons), we also must clear the
subtitle data, but of course only on seek, since otherwise all subtitles
would just disappear. What a fucking mess, but such is life. We could
trigger a "refresh seek" to make this more automatic, but I don't feel
like it currently.
This is slightly inefficient (lots of allocations and copying), but I
decided that it doesn't matter. Could matter slightly for crazy ASS
subtitles that render with thousands of events.
Not very well tested. Still seems to work, but I didn't have many test
cases.
2020-02-16 00:02:17 +00:00
|
|
|
propagate = true;
|
2023-03-08 18:05:59 +00:00
|
|
|
if (flags & UPDATE_SUB_HARD) {
|
|
|
|
// forget about the previous preload because
|
|
|
|
// UPDATE_SUB_HARD will cause a sub reinit
|
|
|
|
// that clears all preloaded sub packets
|
|
|
|
sub->preload_attempted = false;
|
|
|
|
}
|
sub: make filter_sdh a "proper" filter, allow runtime changes
Until now, filter_sdh was simply a function that was called by sd_ass
directly (if enabled).
I want to add another filter, so it's time to turn this into a somewhat
more general subtitle filtering infrastructure.
I pondered whether to reuse the audio/video filtering stuff - but better
not. Also, since subtitles are horrible and tend to refuse proper
abstraction, it's still messed into sd_ass, instead of working on the
dec_sub.c level. Actually mpv used to have subtitle "filters" and even
made subtitle converters part of it, but it was fairly horrible, so
don't do that again.
In addition, make runtime changes possible. Since this was supposed to
be a quick hack, I just decided to put all subtitle filter options into
a separate option group (=> simpler change notification), to manually
push the change through the playloop (like it was sort of before for OSD
options), and to recreate the sub filter chain completely in every
change. Should be good enough.
One strangeness is that due to prefetching and such, most subtitle
packets (or those some time ahead) are actually done filtering when we
change, so the user still needs to manually seek to actually refresh
everything. And since subtitle data is usually cached in ASS_Track (for
other terrible but user-friendly reasons), we also must clear the
subtitle data, but of course only on seek, since otherwise all subtitles
would just disappear. What a fucking mess, but such is life. We could
trigger a "refresh seek" to make this more automatic, but I don't feel
like it currently.
This is slightly inefficient (lots of allocations and copying), but I
decided that it doesn't matter. Could matter slightly for crazy ASS
subtitles that render with thousands of events.
Not very well tested. Still seems to work, but I didn't have many test
cases.
2020-02-16 00:02:17 +00:00
|
|
|
break;
|
2017-12-29 14:39:38 +00:00
|
|
|
}
|
|
|
|
default:
|
sub: make filter_sdh a "proper" filter, allow runtime changes
Until now, filter_sdh was simply a function that was called by sd_ass
directly (if enabled).
I want to add another filter, so it's time to turn this into a somewhat
more general subtitle filtering infrastructure.
I pondered whether to reuse the audio/video filtering stuff - but better
not. Also, since subtitles are horrible and tend to refuse proper
abstraction, it's still messed into sd_ass, instead of working on the
dec_sub.c level. Actually mpv used to have subtitle "filters" and even
made subtitle converters part of it, but it was fairly horrible, so
don't do that again.
In addition, make runtime changes possible. Since this was supposed to
be a quick hack, I just decided to put all subtitle filter options into
a separate option group (=> simpler change notification), to manually
push the change through the playloop (like it was sort of before for OSD
options), and to recreate the sub filter chain completely in every
change. Should be good enough.
One strangeness is that due to prefetching and such, most subtitle
packets (or those some time ahead) are actually done filtering when we
change, so the user still needs to manually seek to actually refresh
everything. And since subtitle data is usually cached in ASS_Track (for
other terrible but user-friendly reasons), we also must clear the
subtitle data, but of course only on seek, since otherwise all subtitles
would just disappear. What a fucking mess, but such is life. We could
trigger a "refresh seek" to make this more automatic, but I don't feel
like it currently.
This is slightly inefficient (lots of allocations and copying), but I
decided that it doesn't matter. Could matter slightly for crazy ASS
subtitles that render with thousands of events.
Not very well tested. Still seems to work, but I didn't have many test
cases.
2020-02-16 00:02:17 +00:00
|
|
|
propagate = true;
|
2017-12-29 14:39:38 +00:00
|
|
|
}
|
sub: make filter_sdh a "proper" filter, allow runtime changes
Until now, filter_sdh was simply a function that was called by sd_ass
directly (if enabled).
I want to add another filter, so it's time to turn this into a somewhat
more general subtitle filtering infrastructure.
I pondered whether to reuse the audio/video filtering stuff - but better
not. Also, since subtitles are horrible and tend to refuse proper
abstraction, it's still messed into sd_ass, instead of working on the
dec_sub.c level. Actually mpv used to have subtitle "filters" and even
made subtitle converters part of it, but it was fairly horrible, so
don't do that again.
In addition, make runtime changes possible. Since this was supposed to
be a quick hack, I just decided to put all subtitle filter options into
a separate option group (=> simpler change notification), to manually
push the change through the playloop (like it was sort of before for OSD
options), and to recreate the sub filter chain completely in every
change. Should be good enough.
One strangeness is that due to prefetching and such, most subtitle
packets (or those some time ahead) are actually done filtering when we
change, so the user still needs to manually seek to actually refresh
everything. And since subtitle data is usually cached in ASS_Track (for
other terrible but user-friendly reasons), we also must clear the
subtitle data, but of course only on seek, since otherwise all subtitles
would just disappear. What a fucking mess, but such is life. We could
trigger a "refresh seek" to make this more automatic, but I don't feel
like it currently.
This is slightly inefficient (lots of allocations and copying), but I
decided that it doesn't matter. Could matter slightly for crazy ASS
subtitles that render with thousands of events.
Not very well tested. Still seems to work, but I didn't have many test
cases.
2020-02-16 00:02:17 +00:00
|
|
|
if (propagate && sub->sd->driver->control)
|
|
|
|
r = sub->sd->driver->control(sub->sd, cmd, arg);
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_unlock(&sub->lock);
|
sub: uglify sub decoder with locking
The plan is to make the whole OSD thread-safe, and we start with this.
We just put locks on all entry points (fortunately, dec_sub.c and all
sd_*.c decoders are very closed off, and only the entry points in
dec_sub.h let you access it). I think this is pretty ugly, but at least
it's very simple.
There's a special case with sub_get_bitmaps(): this function returns
pointers to decoder data (specifically, libass images). There's no way
to synchronize this internally, so expose sub_lock/sub_unlock functions.
To make things simpler, and especially because the lock is sort-of
exposed to the outside world, make the locks recursive. Although the
only case where this is actually needed (although trivial) is
sub_set_extradata().
One corner case are ASS subtitles: for some reason, we keep a single
ASS_Renderer instance for subtitles around (probably to avoid rescanning
fonts with ordered chapters), and this ASS_Renderer instance is not
synchronized. Also, demux_libass.c loads ASS_Track objects, which are
directly passed to sd_ass.c. These things are not synchronized (and
would be hard to synchronize), and basically we're out of luck. But I
think for now, accesses happen reasonably serialized, so there is no
actual problem yet, even if we start to access OSD from other threads.
2014-01-17 22:13:09 +00:00
|
|
|
return r;
|
2013-06-28 23:34:11 +00:00
|
|
|
}
|
2017-02-07 16:05:17 +00:00
|
|
|
|
|
|
|
void sub_set_recorder_sink(struct dec_sub *sub, struct mp_recorder_sink *sink)
|
|
|
|
{
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_lock(&sub->lock);
|
2017-02-07 16:05:17 +00:00
|
|
|
sub->recorder_sink = sink;
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_unlock(&sub->lock);
|
2017-02-07 16:05:17 +00:00
|
|
|
}
|
|
|
|
|
Implement backwards playback
See manpage additions. This is a huge hack. You can bet there are shit
tons of bugs. It's literally forcing square pegs into round holes.
Hopefully, the manpage wall of text makes it clear enough that the whole
shit can easily crash and burn. (Although it shouldn't literally crash.
That would be a bug. It possibly _could_ start a fire by entering some
sort of endless loop, not a literal one, just something where it tries
to do work without making progress.)
(Some obvious bugs I simply ignored for this initial version, but
there's a number of potential bugs I can't even imagine. Normal playback
should remain completely unaffected, though.)
How this works is also described in the manpage. Basically, we demux in
reverse, then we decode in reverse, then we render in reverse.
The decoding part is the simplest: just reorder the decoder output. This
weirdly integrates with the timeline/ordered chapter code, which also
has special requirements on feeding the packets to the decoder in a
non-straightforward way (it doesn't conflict, although a bugmessmass
breaks correct slicing of segments, so EDL/ordered chapter playback is
broken in backward direction).
Backward demuxing is pretty involved. In theory, it could be much
easier: simply iterating the usual demuxer output backward. But this
just doesn't fit into our code, so there's a cthulhu nightmare of shit.
To be specific, each stream (audio, video) is reversed separately. At
least this means we can do backward playback within cached content (for
example, you could play backwards in a live stream; on that note, it
disables prefetching, which would lead to losing new live video, but
this could be avoided).
The fuckmess also meant that I didn't bother trying to support
subtitles. Subtitles are a problem because they're "sparse" streams.
They need to be "passively" demuxed: you don't try to read a subtitle
packet, you demux audio and video, and then look whether there was a
subtitle packet. This means to get subtitles for a time range, you need
to know that you demuxed video and audio over this range, which becomes
pretty messy when you demux audio and video backwards separately.
Backward display is the most weird (and potentially buggy) part. To
avoid that we need to touch a LOT of timing code, we negate all
timestamps. The basic idea is that due to the navigation, all
comparisons and subtractions of timestamps keep working, and you don't
need to touch every single of them to "reverse" them.
E.g.:
bool before = pts_a < pts_b;
would need to be:
bool before = forward
? pts_a < pts_b
: pts_a > pts_b;
or:
bool before = pts_a * dir < pts_b * dir;
or if you, as it's implemented now, just do this after decoding:
pts_a *= dir;
pts_b *= dir;
and then in the normal timing/renderer code:
bool before = pts_a < pts_b;
Consequently, we don't need many changes in the latter code. But some
assumptions inhererently true for forward playback may have been broken
anyway. What is mainly needed is fixing places where values are passed
between positive and negative "domains". For example, seeking and
timestamp user display always uses positive timestamps. The main mess is
that it's not obvious which domain a given variable should or does use.
Well, in my tests with a single file, it suddenly started to work when I
did this. I'm honestly surprised that it did, and that I didn't have to
change a single line in the timing code past decoder (just something
minor to make external/cached text subtitles display). I committed it
immediately while avoiding thinking about it. But there really likely
are subtle problems of all sorts.
As far as I'm aware, gstreamer also supports backward playback. When I
looked at this years ago, I couldn't find a way to actually try this,
and I didn't revisit it now. Back then I also read talk slides from the
person who implemented it, and I'm not sure if and which ideas I might
have taken from it. It's possible that the timestamp reversal is
inspired by it, but I didn't check. (I think it claimed that it could
avoid large changes by changing a sign?)
VapourSynth has some sort of reverse function, which provides a backward
view on a video. The function itself is trivial to implement, as
VapourSynth aims to provide random access to video by frame numbers (so
you just request decreasing frame numbers). From what I remember, it
wasn't exactly fluid, but it worked. It's implemented by creating an
index, and seeking to the target on demand, and a bunch of caching. mpv
could use it, but it would either require using VapourSynth as demuxer
and decoder for everything, or replacing the current file every time
something is supposed to be played backwards.
FFmpeg's libavfilter has reversal filters for audio and video. These
require buffering the entire media data of the file, and don't really
fit into mpv's architecture. It could be used by playing a libavfilter
graph that also demuxes, but that's like VapourSynth but worse.
2019-05-18 00:10:51 +00:00
|
|
|
void sub_set_play_dir(struct dec_sub *sub, int dir)
|
|
|
|
{
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_lock(&sub->lock);
|
Implement backwards playback
See manpage additions. This is a huge hack. You can bet there are shit
tons of bugs. It's literally forcing square pegs into round holes.
Hopefully, the manpage wall of text makes it clear enough that the whole
shit can easily crash and burn. (Although it shouldn't literally crash.
That would be a bug. It possibly _could_ start a fire by entering some
sort of endless loop, not a literal one, just something where it tries
to do work without making progress.)
(Some obvious bugs I simply ignored for this initial version, but
there's a number of potential bugs I can't even imagine. Normal playback
should remain completely unaffected, though.)
How this works is also described in the manpage. Basically, we demux in
reverse, then we decode in reverse, then we render in reverse.
The decoding part is the simplest: just reorder the decoder output. This
weirdly integrates with the timeline/ordered chapter code, which also
has special requirements on feeding the packets to the decoder in a
non-straightforward way (it doesn't conflict, although a bugmessmass
breaks correct slicing of segments, so EDL/ordered chapter playback is
broken in backward direction).
Backward demuxing is pretty involved. In theory, it could be much
easier: simply iterating the usual demuxer output backward. But this
just doesn't fit into our code, so there's a cthulhu nightmare of shit.
To be specific, each stream (audio, video) is reversed separately. At
least this means we can do backward playback within cached content (for
example, you could play backwards in a live stream; on that note, it
disables prefetching, which would lead to losing new live video, but
this could be avoided).
The fuckmess also meant that I didn't bother trying to support
subtitles. Subtitles are a problem because they're "sparse" streams.
They need to be "passively" demuxed: you don't try to read a subtitle
packet, you demux audio and video, and then look whether there was a
subtitle packet. This means to get subtitles for a time range, you need
to know that you demuxed video and audio over this range, which becomes
pretty messy when you demux audio and video backwards separately.
Backward display is the most weird (and potentially buggy) part. To
avoid that we need to touch a LOT of timing code, we negate all
timestamps. The basic idea is that due to the navigation, all
comparisons and subtractions of timestamps keep working, and you don't
need to touch every single of them to "reverse" them.
E.g.:
bool before = pts_a < pts_b;
would need to be:
bool before = forward
? pts_a < pts_b
: pts_a > pts_b;
or:
bool before = pts_a * dir < pts_b * dir;
or if you, as it's implemented now, just do this after decoding:
pts_a *= dir;
pts_b *= dir;
and then in the normal timing/renderer code:
bool before = pts_a < pts_b;
Consequently, we don't need many changes in the latter code. But some
assumptions inhererently true for forward playback may have been broken
anyway. What is mainly needed is fixing places where values are passed
between positive and negative "domains". For example, seeking and
timestamp user display always uses positive timestamps. The main mess is
that it's not obvious which domain a given variable should or does use.
Well, in my tests with a single file, it suddenly started to work when I
did this. I'm honestly surprised that it did, and that I didn't have to
change a single line in the timing code past decoder (just something
minor to make external/cached text subtitles display). I committed it
immediately while avoiding thinking about it. But there really likely
are subtle problems of all sorts.
As far as I'm aware, gstreamer also supports backward playback. When I
looked at this years ago, I couldn't find a way to actually try this,
and I didn't revisit it now. Back then I also read talk slides from the
person who implemented it, and I'm not sure if and which ideas I might
have taken from it. It's possible that the timestamp reversal is
inspired by it, but I didn't check. (I think it claimed that it could
avoid large changes by changing a sign?)
VapourSynth has some sort of reverse function, which provides a backward
view on a video. The function itself is trivial to implement, as
VapourSynth aims to provide random access to video by frame numbers (so
you just request decreasing frame numbers). From what I remember, it
wasn't exactly fluid, but it worked. It's implemented by creating an
index, and seeking to the target on demand, and a bunch of caching. mpv
could use it, but it would either require using VapourSynth as demuxer
and decoder for everything, or replacing the current file every time
something is supposed to be played backwards.
FFmpeg's libavfilter has reversal filters for audio and video. These
require buffering the entire media data of the file, and don't really
fit into mpv's architecture. It could be used by playing a libavfilter
graph that also demuxes, but that's like VapourSynth but worse.
2019-05-18 00:10:51 +00:00
|
|
|
sub->play_dir = dir;
|
2023-10-21 02:55:41 +00:00
|
|
|
mp_mutex_unlock(&sub->lock);
|
Implement backwards playback
See manpage additions. This is a huge hack. You can bet there are shit
tons of bugs. It's literally forcing square pegs into round holes.
Hopefully, the manpage wall of text makes it clear enough that the whole
shit can easily crash and burn. (Although it shouldn't literally crash.
That would be a bug. It possibly _could_ start a fire by entering some
sort of endless loop, not a literal one, just something where it tries
to do work without making progress.)
(Some obvious bugs I simply ignored for this initial version, but
there's a number of potential bugs I can't even imagine. Normal playback
should remain completely unaffected, though.)
How this works is also described in the manpage. Basically, we demux in
reverse, then we decode in reverse, then we render in reverse.
The decoding part is the simplest: just reorder the decoder output. This
weirdly integrates with the timeline/ordered chapter code, which also
has special requirements on feeding the packets to the decoder in a
non-straightforward way (it doesn't conflict, although a bugmessmass
breaks correct slicing of segments, so EDL/ordered chapter playback is
broken in backward direction).
Backward demuxing is pretty involved. In theory, it could be much
easier: simply iterating the usual demuxer output backward. But this
just doesn't fit into our code, so there's a cthulhu nightmare of shit.
To be specific, each stream (audio, video) is reversed separately. At
least this means we can do backward playback within cached content (for
example, you could play backwards in a live stream; on that note, it
disables prefetching, which would lead to losing new live video, but
this could be avoided).
The fuckmess also meant that I didn't bother trying to support
subtitles. Subtitles are a problem because they're "sparse" streams.
They need to be "passively" demuxed: you don't try to read a subtitle
packet, you demux audio and video, and then look whether there was a
subtitle packet. This means to get subtitles for a time range, you need
to know that you demuxed video and audio over this range, which becomes
pretty messy when you demux audio and video backwards separately.
Backward display is the most weird (and potentially buggy) part. To
avoid that we need to touch a LOT of timing code, we negate all
timestamps. The basic idea is that due to the navigation, all
comparisons and subtractions of timestamps keep working, and you don't
need to touch every single of them to "reverse" them.
E.g.:
bool before = pts_a < pts_b;
would need to be:
bool before = forward
? pts_a < pts_b
: pts_a > pts_b;
or:
bool before = pts_a * dir < pts_b * dir;
or if you, as it's implemented now, just do this after decoding:
pts_a *= dir;
pts_b *= dir;
and then in the normal timing/renderer code:
bool before = pts_a < pts_b;
Consequently, we don't need many changes in the latter code. But some
assumptions inhererently true for forward playback may have been broken
anyway. What is mainly needed is fixing places where values are passed
between positive and negative "domains". For example, seeking and
timestamp user display always uses positive timestamps. The main mess is
that it's not obvious which domain a given variable should or does use.
Well, in my tests with a single file, it suddenly started to work when I
did this. I'm honestly surprised that it did, and that I didn't have to
change a single line in the timing code past decoder (just something
minor to make external/cached text subtitles display). I committed it
immediately while avoiding thinking about it. But there really likely
are subtle problems of all sorts.
As far as I'm aware, gstreamer also supports backward playback. When I
looked at this years ago, I couldn't find a way to actually try this,
and I didn't revisit it now. Back then I also read talk slides from the
person who implemented it, and I'm not sure if and which ideas I might
have taken from it. It's possible that the timestamp reversal is
inspired by it, but I didn't check. (I think it claimed that it could
avoid large changes by changing a sign?)
VapourSynth has some sort of reverse function, which provides a backward
view on a video. The function itself is trivial to implement, as
VapourSynth aims to provide random access to video by frame numbers (so
you just request decreasing frame numbers). From what I remember, it
wasn't exactly fluid, but it worked. It's implemented by creating an
index, and seeking to the target on demand, and a bunch of caching. mpv
could use it, but it would either require using VapourSynth as demuxer
and decoder for everything, or replacing the current file every time
something is supposed to be played backwards.
FFmpeg's libavfilter has reversal filters for audio and video. These
require buffering the entire media data of the file, and don't really
fit into mpv's architecture. It could be used by playing a libavfilter
graph that also demuxes, but that's like VapourSynth but worse.
2019-05-18 00:10:51 +00:00
|
|
|
}
|
2021-01-13 15:59:08 +00:00
|
|
|
|
2022-01-19 17:23:52 +00:00
|
|
|
bool sub_is_primary_visible(struct dec_sub *sub)
|
|
|
|
{
|
2024-04-04 11:56:21 +00:00
|
|
|
mp_mutex_lock(&sub->lock);
|
|
|
|
bool ret = sub->shared_opts->sub_visibility[0];
|
|
|
|
mp_mutex_unlock(&sub->lock);
|
|
|
|
return ret;
|
2022-01-19 17:23:52 +00:00
|
|
|
}
|
|
|
|
|
2021-01-13 15:59:08 +00:00
|
|
|
bool sub_is_secondary_visible(struct dec_sub *sub)
|
|
|
|
{
|
2024-04-04 11:56:21 +00:00
|
|
|
mp_mutex_lock(&sub->lock);
|
|
|
|
bool ret = sub->shared_opts->sub_visibility[1];
|
|
|
|
mp_mutex_unlock(&sub->lock);
|
|
|
|
return ret;
|
2021-01-13 15:59:08 +00:00
|
|
|
}
|