Apparently this cuases trouble for legacy demuxers. demux_mpg stopped
doing PCM audio. (The problem was probably that it read a bunch of
video packets on detection, and then the sparse video hack prevented
audio packets from being read, because it looked like there were no
more audio packets. With sparse video, this normally helps not reading
too many audio packets.)
Since the legacy demuxers do not need this hack, enable it for
demux_lavf and demux_mkv only.
Some additional hacks that were needed to handle legacy demuxers can be
removed, making the code simpler.
Also see commit 4a40eed.
There should be no functional changes, except that way how avoiding
spamming the terminal with the overflow warning is handled changes a
bit.
The removed check for ds->eof looks suspicious, but it should be
redundant now.
The seek bar appeared to be "stuck" to the start of the current chapter.
This is a regression from 630a2b1. This commit assumed that hrseek_pts
would always contain the hrseek target time (when hrseek_active==true).
But this is not always the case: when playing timeline stuff (e.g.
ordered chapters), hrseek framedropping is abused to handle an obscure
corner case, and then hrseek_pts contains something completely unrelated
to the current playback time. See the added comment in mplayer.c and
commit c1232c9.
Fix this by trying something else to get a correct time "during"
hr-seeks. mpctx->restart_playback looks ideal, because it's set while
audio is being synced / audio buffers being filled, so we know that the
audio time is probably bogus while it is set. Let's hope this is
correct.
Do this to reduce conflicts with <linux/input.h>, which contains some
conflicting defines.
This changes the meaning of MP_KEY_DOWN:
KEY_DOWN is renamed to MP_KEY_DOWN (cursor down key)
MP_KEY_DOWN is renamed to MP_KEY_STATE_DOWN (modifier for key down state)
int64_t was accidentally used with "%lld" format specifiers, which is
incorrect (even though long long int is always 64 bits, the type behind
int64_t can be different, e.g. it can be long int on 64 bit platforms).
Commit 4d016a9 changed how demuxers report the codec of each stream.
Some of that was missed in video.c, which is important for legacy
demuxers (demux_mpg was broken by this, which is needed for DVD
playback).
Not sure about the ASF/AVI related change, but this is also a legacy
demuxers only codepath.
The previous name of this filter was misleading, because it doesn’t actually
normalize volume levels. What it does is closer to performing low-quality
dynamic range compression, hence it is now called af_drc.
Commit 4d016a9 added some configure tests using statement_check.
They wrongly used $_ld_tmp, which causes random failure, depending on
whether the previous test using $_ld_tmp was successful. This happened
because I blindly copied the statement_checks from somewhere else.
Fix them.
Instead of putting codec header data into WAVEFORMATEX and
BITMAPINFOHEADER, pass it directly via AVCodecContext. To do this, we
add mp_copy_lav_codec_headers(), which copies the codec header data
from one AVCodecContext to another (originally, the plan was to use
avcodec_copy_context() for this, but it looks like this would turn
decoder initialization into an even worse mess).
Get rid of the silly CodecID <-> codec_tag mapping. This was originally
needed for codecs.conf: codec tags were used to identify codecs, but
libavformat didn't always return useful codec tags (different file
formats can have different, overlapping tag numbers). Since we don't
go through WAVEFORMATEX etc. and pass all header data directly via
AVCodecContext, we can be absolutely sure that the codec tag mapping is
not needed anymore.
Note that this also destroys the "standard" MPlayer method of exporting
codec header data. WAVEFORMATEX and BITMAPINFOHEADER made sure that
other non-libavcodec decoders could be initialized. However, all these
decoders have been removed, so this is just cruft full of old hacks that
are not needed anymore. There's still ad_spdif and ad_mpg123, bu neither
of these need codec header data. Should we ever add non-libavcodec
decoders, better data structures without the past hacks could be added
to export the headers.
Rearrange some code to make it easier readable. Remove some dead code,
and stop printing AVI headers in demux_lavf. (These are not actual AVI
headers, just for internal use.)
There should be no functional changes, other than reducing output in
verbose mode.
Also move the lang field into the general stream header. (SH_COMMON is
an old hack to "share" code between audio/video/sub headers.)
There should be no functional changes, other than not printing stream
info in verbose mode or with slave mode. (The frontend already prints
stream info, and this is just a leftover when individual demuxers did
this, and slave mode remains broken.)
Use codec names instead of FourCCs to identify codecs. Rewrite how
codecs are selected and initialized. Now each decoder exports a list
of decoders (and the codec it supports) via add_decoders(). The order
matters, and the first decoder for a given decoder is preferred over
the other decoders. E.g. all ad_mpg123 decoders are preferred over
ad_lavc, because it comes first in the mpcodecs_ad_drivers array.
Likewise, decoders within ad_lavc that are enumerated first by
libavcodec (using av_codec_next()) are preferred. (This is actually
critical to select h264 software decoding by default instead of vdpau.
libavcodec and ffmpeg/avconv use the same method to select decoders by
default, so we hope this is sane.)
The codec names follow libavcodec's codec names as defined by
AVCodecDescriptor.name (see libavcodec/codec_desc.c). Some decoders
have names different from the canonical codec name. The AVCodecDescriptor
API is relatively new, so we need a compatibility layer for older
libavcodec versions for codec names that are referenced internally,
and which are different from the decoder name. (Add a configure check
for that, because checking versions is getting way too messy.)
demux/codec_tags.c is generated from the former codecs.conf (minus
"special" decoders like vdpau, and excluding the mappings that are the
same as the mappings libavformat's exported RIFF tables). It contains
all the mappings from FourCCs to codec name. This is needed for
demux_mkv, demux_mpg, demux_avi and demux_asf. demux_lavf will set the
codec as determined by libavformat, while the other demuxers have to do
this on their own, using the mp_set_audio/video_codec_from_tag()
functions. Note that the sh_audio/video->format members don't uniquely
identify the codec anymore, and sh->codec takes over this role.
Replace the --ac/--vc/--afm/--vfm with new --vd/--ad options, which
provide cover the functionality of the removed switched.
Note: there's no CODECS_FLAG_FLIP flag anymore. This means some obscure
container/video combinations (e.g. the sample Film_200_zygo_pro.mov)
are played flipped. ffplay/avplay doesn't handle this properly either,
so we don't care and blame ffmeg/libav instead.
Simplify --no-config and make it a normal flag option, and doesn't take
an argument anymore. You can get the same behavior by using --no-config
and then --include to explicitly load a certain config file.
Make --no-config work for input.conf as well. Make it so that
--input:conf=file still works in this case. As a technically unrelated
change, the file argument now works as one would expect, instead of
making it relatively to "~/.mpv/". This makes for simpler code and
easier to understand option semantics. We can also print better error
messages.
Doesn't have much of a purpose for normal playback. You can get
milliseconds display with --osd-fractions. It's also possible to build
a custom status line with --status-msg.
This gives more space on the status line and, in my opinion, is a bit
less annoying.
OPT_MAKE_FLAGS() used to emit two options (one with "no" prefixed),
but that has been long removed by special casing flag options in the
option parser. OPT_FLAG_ON() used to imply that there's no "no-"
prefixed option, but this hasn't been the case for a while either.
(Conceptually, it has been replaced by OPT_FLAG_STORE().)
Remove OPT_FLAG_OFF, which was unused.
There were two option syntax variations:
"old": -opt value
"new": --opt=value
"-opt=value" was invalid, and "--opt value" meant "--opt=" followed by
a separate option "value" (i.e. interpreted as filename). There isn't
really any reason to do this. The "old" syntax used to be ambiguous
(you had to call the option parser to know whether the following
argument is an option value or a new option), but that has been removed.
Further, using "=" in the option string is always unambiguous.
Since the distinction between the two option variants is confusing,
just remove the difference and allow "--opt value" and "-opt=value".
To make this easier, do some other cleanups as well (e.g. avoid having
to do a manual lookup of the option just to check for M_OPT_PRE_PARSE,
which somehow ended up with finally getting rid of the m_config.mode
member).
Error reporting is still a mess, and we opt for reporting too many
rather than too few errors to the user.
There shouldn't be many user-visible changes. The --framedrop and
--term-osd options now always require parameters.
The --mute option is intentionally made ambiguous: it works like a flag
option, but a value can be passed to it explicitly ("--mute=auto"). If
the interpretation of the option is ambiguous (like "--mute auto"), the
second string is interpreted as separate option or filename. (Normal
flag options are actually ambiguous in this way too.)
Now setting a value with "=" is not required anymore in config files.
This should work analogous to command line arguments. Putting an entry
"opt=value" into the config file is like "--opt=value" on the command
line, and "opt" is like "--opt=" and "--opt".
Normally, all flag options can be negated by prepending a "no-", for
example "--no-opt" becomes "--opt=no". Some flag options can't actually
be negated, so add a CONF_TYPE_STORE option type to disallow the "no-"
fallback.
Do the same for choice options. Remove the explicit "no-" prefixed
options, add "no" as choice.
Move the handling of automatic "no-" options from parser-mpcmd.c to
m_config.c, and use it in m_config_set_option/m_config_parse_option.
This makes these options available in the config file. It also
simplifies sub-option parsing, because it doesn't need to handle "no-"
anymore.
mpv -ao help and mpv -vo help shouldn't show the encoding outputs (named
"lavc" on both cases). Also make it impossible to select these manually
when not encoding.
Basically, move vo_opengl above the other VOs (except vo_vdpau). This
changes preferences on Windows and Linux.
Move vo_opengl_old further down and make it the last fallback (before
vo_x11).
vo_caca is crap (no pun intended), and should never be autoprobed.
On Linux, ao_portaudio has weird freezing issues (possibly specific to
the ALSA backend, though). Also ao_dsound is more likely to get multi-
channel audio output right, and ao_portaudio probably mangles these.
This partially reverts earlier decisions, when I thought it would
always be better to prefer the audio volume filter over the AO's,
because the AO's relies on the underlying audio-API, which could
be broken or exhibit unusual behavior (like it happened with ao_dsound).
However, since the audio buffer can be quite large (500 ms), and we
don't attempt to flush & refilter the audio on volume changes, always
prefer AO volume control (as long as the AO mixer doesn't control the
system mixer).
Also document what the mixer.c related AO fields mean (hopefully not
too brief).
This could write .edl files in MPlayer's format. Support for playing
these files has been removed from mplayer2 quite a while ago. (mplayer2
can play its own, "new" .edl format, but does not support writing it.)
Since this is a rather obscure functionality, and it's not really clear
how it should behave (e.g. what should it do if a new file is played),
and wasn't all that great to begin with (what if you made a mistake?
the "edl_mark" command sucks for editing), get rid of it.
Suggestions how to reimplement this in a nicer way are welcome. If it's
just about retrieving timecodes, this in input.conf will do:
KEY print_text "position: ${=time-pos}"
It appears git submodule handling recently changed, changing the .git
directory to a regular file containing submodule specific information.
This means that version.sh would generate "#define VERSION "git-UNKNOWN""
if the checkout is a submodule, because it explicitly checks for a .git
directory using test -d. Change it to -e so that this case is handled
correctly.
This fixes a problem that happened with syncplay.pl [1] when ad_mpg123
was in use, and get_current_time() returning a bogus time position.
This only happens during seeking; the reported time is correct after the
seek is done.
The audio PTS as returned by playing_audio_pts() is simply bogus during
hr-seek. With ad_ffmpeg, it was actually set to MP_NOPTS_VALUE during
seeking, so get_current_time() did a fallback to the video PTS. However,
ad_mpg123 is different and explicitly decodes some audio when resetting
on seek (reasons why it does this unknown and uninvestigated; apparently
it's to reinit libmpg123). As a result, the audio PTS was set to the
start position of the seek (or something similar), which could be very
different from the seek target time.
This confused syncplay. It got the bogus time because it spams the
player with read commands to the "time-pos" property, so this corner
case was hit.
Fix this by making get_current_time() return the seek target time if
hr-seek is active. This should make behavior the same as before commit
3f949cf "mplayer: prefer audio PTS over video PTS for status line".
[1] http://syncplay.pl
When doing a framestep while there is no more video, nothing happened,
and audio continued to play. When advancing to the next file, the player
was paused. Fix it so that it always pauses (except on very low frame
rate video, which is yet another corner case).
We also change the meaning of framestepping a bit: in audio only mode,
framstepping unpauses for a single playloop iteration. This is probably
not useful at all, but makes the code a bit more simpler/uniform.
Just like the previous commit, this matters most for audio files with
cover art, for which this special case is the normal case.
mpctx->delay is used to control audio/video sync. If more audio than
video has been played, it grows larger, meaning A/V desync is happening.
This logic is a bit broken when video has ended, and audio is still
playing. In that case, it tries to read additional video frames from the
video decoder (because even if you don't feed new packets to the
decoder, it could still return delayed frames). For that, the code to
determine whether frames should be dropped is invoked
(check_framedrop()). This function detects that video is behind audio (mpctx-
>delay growing big),
and attempts to issue a framedrop.
Reset mpctx->delay if there's no more video.
This fixes the the frame drop display "counting up" on each playloop
iteration when playing audio files with cover art. These files are
basically audio+video files with a single video frame. When playing
these files the the corner case of having run out of video while audio
is still playing is the normal case.
Also reset mpctx->last_av_difference. This is not updated anymore if
video ends (since update_avsync() sets it, but it's not called if
video_left is false). This removes the "stuck" A/V sync value when video
ends. With audio files containing cover art we would display a
meaningless value over the duration of the whole file otherwise.
These were memory leaks in theory, though not in practice (all memory is
free'd on exit anyway). However, it was still annoying when leak
reporting is enabled.
I'm not sure if there was an actual leak in check_autorepeat(), maybe
not.
Add more vf_stereo3d output formats.
Adds high quality green-magenta and yellow-blue dubois
anaglyph 3D output support.
Patch by thomas schorpp, thomas.schorpp gmail
git-svn-id: svn://svn.mplayerhq.hu/mplayer/trunk@35906 b3059339-0415-0410-9bf9-f77b7e298cf2
Conflicts:
video/filter/vf_stereo3d.c
The ffmpeg/libav attached picture hack usually set the PTS of video
packets to AV_NOPTS_VALUE. Set it to 0 to avoid printing a warning by
the filter code.
Explicitly advancing the playlist with input commands ("playlist_next")
didn't jump back to the first file, if the current file was the last on
the playlist and looping was enabled.
Fix this and make the behavior with explicit input and playback EOF the
same.
Also add a minor feature: if looping is enabled, and the current file is
the first on the playlist, going back one entry jumps to the last
playlist entry (without changing loop count).
Fixes#22.
Should be dead code. Stream selection is handled either during
demuxer initialization, or via DEMUXER_CTRL_SWITCH_*.
(If there were actually situations where this code did something, it
was probably broken anyway.)
It appears this is not needed anymore. ffmpeg can handle "chained" ogg
files fine. These can be created with "cat file1.ogg file2.ogg > chained.ogg",
and are similar (or equal) to some internet radio streams. Apparently
ffmpeg used to add new tracks when crossing boundaries in chained files,
and the hack in demux_lavf.c handled this. At some later point, ffmpeg's
ogg demuxer was improved, and stopped adding new tracks as long as the
codec doesn't change.
Since the hack in demux_lavf.c was hardcoded to Vorbis (i.e. only active
if the new and old track were both Vorbis), it's dead code, and we can
remove it. I couldn't find any stream that triggered this hack, or fails
without it.
Firefox had a similar issue, and its bug tracker makes a good reference:
https://bugzilla.mozilla.org/show_bug.cgi?id=455165
NOTE: this doesn't update metadata on track changes anymore.
Allow negative times. Timestamps can be negative, and we actually
display negative time for other reasons too, such as when waiting for
the old audio to drain with gapless audio.)
Avoid overflows with relatively large time values. (We still don't
handle values too large for int64_t.)
Apply setLevel hacks for fullscreen switching with ontop active only in
fullscreen. This keeps the window correctly on top even when losing focus.
Fixes#21
I also reverted NSScreenSaverWindowLevel to NSNormalWindowLevel + 1, so that
the cmd-tab UI is visible
The main problem this commit addresses is when you switched spaced back and
forth with `--ontop` active the video window didn't recive focus again.
Turns out that setting a normal window level just before conceding focus to
other apps / spaces fixes the problem. I think Cocoa is misbehaving here, and
should probably file a radar.
I found 3 related improvements while fixing this:
* fullscreen_window_level is a dead body from an older implementation.
* Use NSScreenSaverWindowLevel for ontop. This should be a really high window
level. Definitely higher than NSNormalWindowLevel + 1.
* The window level was set correctly only when out of fullscreen.
Make the window resizing menu items calculate the new window size based on the
video size and not the current window size.
This only makes a difference when using `--autofit`.
This functionality looked smart but created problems with some kinds of
multi touch events. Moreover some events coming from the windows server – like
hovering a corner for window resize – didn't cause the player to wake up
immediately.
The "correct" non hacky way to implement async event polling with cocoa would
be having the vanilla cocoa event loop driving the player and setting up mpv's
terminal FDs as event sources for the cocoa event loop.
Fixes#20
wm4
Martin
wm4
cehoyos
Stefano Pigozzi