1
0
mirror of https://github.com/mpv-player/mpv synced 2024-12-24 07:42:17 +00:00
Commit Graph

599 Commits

Author SHA1 Message Date
wm4
e970113932 input: remove some unneeded code 2014-10-10 23:05:33 +02:00
wm4
d0f77b5615 input: cosmetics: move code 2014-10-10 23:01:00 +02:00
wm4
c9f45ea93e input: use mpv_node parser for char** command parsers
Minor simplification, also drops some useless stuff.
2014-10-10 22:58:28 +02:00
wm4
63903c27bd input: add a function to parse mpv_node as command
For future client API enhancements.
2014-10-10 22:37:11 +02:00
Stefano Pigozzi
dba2b90d9a libmpv/cocoa: don't start the event monitor
The event monitor is used to get keyboard events when there is no window, but
since it is a global monitor to the current process, we don't want it in a
library setting.
2014-10-09 22:14:41 +02:00
wm4
e294656cb1 client API: rename --input-x11-keyboard to --input-vo-keyboard
Apparently we need this for Cocoa too. (The option was X11 specific in
the hope that only X11 would need this hack.)
2014-10-09 18:28:37 +02:00
wm4
3273db1ef7 client API, X11: change default keyboard input handling again
Commit 64b7811c tried to do the "right thing" with respect to whether
keyboard input should be enabled or not. It turns out that X11 does
something stupid by design. All modern toolkits work around this native
X11 behavior, but embedding breaks these workarounds.

The only way to handle this correctly is the XEmbed protocol. It needs
to be supported by the toolkit, and probably also some mpv support. But
Qt has inconsistent support for it. In Qt 4, a X11 specific embedding
widget was needed. Qt 5.0 doesn't support it at all. Qt 5.1 apparently
supports it via QWindow, but if it really does, I couldn't get it to
work.

So add a hack instead. The new --input-x11-keyboard option controls
whether mpv should enable keyboard input on the X11 window or not. In
the command line player, it's enabled by default, but in libmpv it's
disabled.

This hack has the same problem as all previous embedding had: move the
mouse outside of the window, and you don't get keyboard input anymore.
Likewise, mpv will steal all keyboard input from the parent application
as long as the mouse is inside of the mpv window.

Also see issue #1090.
2014-09-28 20:11:00 +02:00
wm4
cbf7180c90 input: copy options automatically
Originally, all options were copied to ensure that input_ctx remins
thread-safe, even if options are changed asynchronously. But this got
a bit inconsistent. Copy them automatically and reduce some weirdness.
2014-09-27 16:25:29 +02:00
wm4
b4d1494336 input: separate creation and loading of config
Until now, creating the input_ctx was delayed until the command line
and config files were parsed. Separate creation and loading so that
input_ctx is available from start.

This should make it possible to simplify some things. For example,
some complications with Cocoa were apparently only because input_ctx
was available only "later". (Although I'm not sure if this is still
relevant, or if the Cocoa code should even be organized this way.)
2014-09-27 16:01:55 +02:00
wm4
5e6c9963d8 input: explain why we use semaphores
Also switch function names for better self-documentation.
2014-09-20 04:22:37 +02:00
Alexander Preisinger
30175538fe input: add locking for repeat info 2014-09-19 17:48:06 +02:00
Alexander Preisinger
be516022b6 input: add function for setting repeat info
Let us set a different rate and delay.

Needed for the following commit where we set rate and delay reported by weston.
But only if the option native-keyrepeat is set.
2014-09-19 17:36:37 +02:00
wm4
e0b4daf3ad input: use libwaio for pipe input on Windows
Use libwaio to read from pipes (stdin or named pipes) on Windows. This
liberates us from nasty issues, such as pipes (as created by most
programs) not being possible to read in a non-blocking or event-driven
way. Although it would be possible to do that in a somewhat sane way
on Vista+, it's still not easy, and on XP it's especially hard. libwaio
handles these things for us.

Move pipe.c to pipe-unix.c, and remove Windows specific things. Also
adjust the input.c code to make this work cleanly.
2014-09-14 16:24:01 +02:00
wm4
08116feec7 input: fix nested commands
Regression from today.
2014-09-13 18:41:34 +02:00
wm4
893f4a0fee input: distinguish playlist navigation and quit commands for abort
Refine the ugly hack from the previous commit, and let the "quit"
command and some others abort playback immediately. For
playlist_next/playlist_prev, still use the old hack, because we can't
know if they would stop playback or not.
2014-09-13 16:47:30 +02:00
wm4
2e91d44e20 stream: redo playback abort handling
This mechanism originates from MPlayer's way of dealing with blocking
network, but it's still useful. On opening and closing, mpv waits for
network synchronously, and also some obscure commands and use-cases can
lead to such blocking. In these situations, the stream is asynchronously
forced to stop by "interrupting" it.

The old design interrupting I/O was a bit broken: polling with a
callback, instead of actively interrupting it. Change the direction of
this. There is no callback anymore, and the player calls
mp_cancel_trigger() to force the stream to return.

libavformat (via stream_lavf.c) has the old broken design, and fixing it
would require fixing libavformat, which won't happen so quickly. So we
have to keep that part. But everything above the stream layer is
prepared for a better design, and more sophisticated methods than
mp_cancel_test() could be easily introduced.

There's still one problem: commands are still run in the central
playback loop, which we assume can block on I/O in the worst case.
That's not a problem yet, because we simply mark some commands as being
able to stop playback of the current file ("quit" etc.), so input.c
could abort playback as soon as such a command is queued. But there are
also commands abort playback only conditionally, and the logic for that
is in the playback core and thus "unreachable". For example,
"playlist_next" aborts playback only if there's a next file. We don't
want it to always abort playback.

As a quite ugly hack, abort playback only if at least 2 abort commands
are queued - this pretty much happens only if the core is frozen and
doesn't react to input.
2014-09-13 16:09:51 +02:00
wm4
2dd819705d input: "quit_watch_later" and "stop" are abort commands
This means they get special handling for asynchronously aborting
playback, even if the player is "stuck".

Also document "stop". It seems somewhat useful for client API users
(although that will be implemented properly only in the following
commits.)
2014-09-13 14:10:10 +02:00
wm4
b0a3d38b4c input: don't autorepeat cycle_values command
Not sure why this was originally added as autorepeated. It makes no
sense, because switching between choices should never autorepeat. (For
the normal "add"/"cycle" commands, autorepeat is usually enabled, but
command.c tries to disable it specifically for choice properties.)
2014-09-13 01:14:07 +02:00
wm4
a3b393d68c input: simplify
Just some minor things. In particular, don't call mp_input_wakeup()
manually, but make it part of queuing commands (as far as possible).
2014-09-13 01:14:07 +02:00
wm4
fc5df2b970 input: fix autorepeat
Mismatching units in timeout calculation.

Also, as a near-cosmetic change, explicitly wake up the core on the
right time. Currently this does nothing, because the core is woken up
anyway - but it will matter with the next commit.
2014-09-13 01:14:07 +02:00
wm4
e9b756c7ad input: remove central select() call
This is now unused. Get rid of it and all surrounding infrastructure,
and replace the remaining "wakeup pipe" with a semaphore.
2014-09-10 03:24:45 +02:00
wm4
ae63702a2c input: remove useless joystick.h/lirc.h include files
These really just waste space.
2014-09-10 00:51:36 +02:00
wm4
e2a093df02 input: use an input thread for joystick 2014-09-10 00:48:59 +02:00
wm4
fe0ca7559c input: use an input thread for lirc 2014-09-10 00:48:45 +02:00
wm4
5be678386b input: add convenience function for running input sources in threads 2014-09-10 00:48:12 +02:00
wm4
256fea7655 input: make some fields internal 2014-09-10 00:48:12 +02:00
wm4
28fc13977e terminal-unix: move to thread
Do terminal input with a thread, instead of using the central select()
loop. This also changes some details how SIGTERM is handled.

Part of my crusade against mp_input_add_fd().
2014-09-10 00:48:12 +02:00
wm4
76f9eede34 input: fix missed wakeups, simplify
mp_input_read_cmd() reset the wakeup flag, but only mp_input_wait()
should be able to do that.
2014-09-09 01:23:09 +02:00
wm4
4ef531f815 input: fix use after free with legacy commands
To handle legacy commands, string replacement is used; the modified
string is returned by parse_cmd_str(), but it also frees all temporary
memory, which includes the replaced string.

Closes #1075.
2014-09-08 15:13:11 +02:00
wm4
3b5f28bd0f input: fix exiting with signals
Quitting through SIGTERM etc. was accidentally ignored since commit
f5af5962 from today.
2014-09-08 01:11:32 +02:00
wm4
f5af596237 player: some more input refactoring
Continues commit 348dfd93. Replace other places where input was manually
fetched with common code.

demux_was_interrupted() was a weird function; I'm not entirely sure
about its original purpose, but now we can just replace it with simpler
code as well. One difference is that we always look at the command
queue, rather than just when cache initialization failed. Also, instead
of discarding all but quit/playlist commands (aka abort command), run
all commands. This could possibly lead to unwanted side-effects, like
just ignoring commands that have no effect (consider pressing 'f' for
fullscreen right on start: since the window is not created yet, it would
get discarded). But playlist navigation still works as intended, and
some if not all these problems already existed before that in some
forms, so it should be ok.
2014-09-07 20:44:54 +02:00
shdown
3307af43c5 input: make ar_rate and ar_delay fields of input_ctx signed
ar_rate is set to -1 when autorepeat is disabled; there is no reason
for ar_delay to stay unsigned.
2014-08-30 15:15:37 +02:00
shdown
8ee1bcf1fa input: handle reaching MP_MAX_FDS correctly
Don't dereference fd and increment ictx->num_fds on fail.
2014-08-30 15:15:37 +02:00
wm4
68ff8a0484 Move compat/ and bstr/ directory contents somewhere else
bstr.c doesn't really deserve its own directory, and compat had just
a few files, most of which may as well be in osdep. There isn't really
any justification for these extra directories, so get rid of them.

The compat/libav.h was empty - just delete it. We changed our approach
to API compatibility, and will likely not need it anymore.
2014-08-29 12:31:52 +02:00
Ben Boeckel
7c0a5698eb posix: use STD*_FILENO constants
Rather than "magic" numbers, use meaningful constant names provided by
unistd.h.
2014-08-28 12:03:17 +02:00
wm4
b7f72aa2f4 input: make key bindings like "Shift+X" work (for ASCII)
"Shift+X" didn't actually map any key, as opposed to "Shift+x". This is
because shift usually changes the case of a character, so a plain
printable character like "X" simply can never be combined with shift.

But this is not very intuitive. Always remove the shift code from
printable characters. Also, for ASCII, actually apply the case mapping
to uppercase characters if combined with shift. Doing this for unicode
in general would be nice, but that would require lookup tables. In
general, we don't know anyway what character a key produces when
combined with shift - it could be anything, and depends on the keyboard
layout.
2014-08-26 20:39:28 +02:00
wm4
480febf043 input: make all modifier flags unsigned 2014-08-26 20:39:23 +02:00
wm4
6410c4c8ef input: stdin is 0, not 1
Oops. I can never remember this right.
2014-08-25 21:39:24 +02:00
wm4
740f0f61d8 input: redo how --input-file is handled
Abandon the "old" infrastructure for --input-file (mp_input_add_fd(),
select() loop, non-blocking reads). Replace it with something that
starts a reader thread, using blocking input.

This is for the sake of Windows. Windows is a truly insane operating
system, and there's not even a way to read a pipe in a non-blocking
way, or to wait for new input in an interruptible way (like with
poll()). And unfortunately, some want to use pipe to send input to
mpv. There are probably (slightly) better IPC mechanisms available
on Windows, but for the sake of platform uniformity, make this work
again for now.

On Vista+, CancelIoEx() could probably be used. But there's no way on
XP. Also, that function doesn't work on wine, making development
harder. We could forcibly terminate the thread, which might work, but
is unsafe. So what we do is starting a thread, and if we don't want
the pipe input anymore, we just abandon the thread. The thread might
remain blocked forever, but if we exit the process, the kernel will
forcibly kill it. On Unix, just use poll() to handle this.

Unfortunately the code is pretty crappy, but it's ok, because it's late
and I wanted to stop working on this an hour ago.

Tested on wine; might not work on a real Windows.
2014-08-25 01:00:21 +02:00
wm4
20d88a6dea input: change verbosity of some message levels
For --input-test, print messages on terminal by default.

Raise message level for enabling input sections, because the OSC makes
this very extremely annoying.
2014-08-25 00:48:55 +02:00
wm4
ef9b399020 input: fix event wakeup
When a new event was added, merely a flag was set, instead of actually
waking up the core (if needed). This was ok in ancient times when all
event sources were part of the select() loop. But now there are several
cases where other threads can add input, and then you actually need to
wakeup the core in order to make it read the events at all.
2014-08-11 13:50:56 +02:00
wm4
5906041343 input: fix off by one error in command parser
Should fix #989.
2014-08-07 18:21:36 +02:00
wm4
58255e0e2b input: be stricter about rejecting mouse input with --no-input-cursor
Apparently this switch means all mouse input should be strictly
rejected. Some VO backends (such as X11) explicitly disable all mouse
events if this option is set, but others don't. So check them in
input.c, which increases consistency.
2014-07-27 22:00:55 +02:00
wm4
4c533fbb16 vo: remove vo_mouse_movement() wrapper
So that VO backends don't have to access the VO just for that.
2014-07-27 21:53:29 +02:00
wm4
89391e7c94 vo: different hack for VOs which need to mangle mouse input
Follow up on commit 760548da. Mouse handling is a bit confusing, because
there are at least 3 coordinate systems associated with it, and it
should be cleaned up. But that is hard, so just apply a hack which gets
the currently-annoying issue (VO backends needing access to the VO) out
of the way.
2014-07-27 21:33:11 +02:00
wm4
18c432b83a osdep: don't assume errno is positive
Apparently this is not necessarily the case, so just drop the silly idea
that depended on this assumption.
2014-07-25 14:32:45 +02:00
wm4
843f5f4723 command: add append-play loadfile mode
"loadfile filename append-play" will now always append the file to the
playlist, and if nothing is playing yet, start playback. I don't want to
change the semantics of "append" mode, so a new mode is needed.

Probably fixes issue #950.
2014-07-23 00:20:53 +02:00
foo86
6a556f524e input: enable wakeup on LIRC socket
Commit dc00b14 removed playloop polling. Enable wakeup on LIRC socket,
otherwise remote control doesn't work when paused.
2014-07-20 13:52:06 +02:00
wm4
a09329bcf7 input: skip BOM in input.conf 2014-07-12 21:25:32 +02:00
wm4
f8c2dd1b78 build: include <strings.h> for strcasecmp()
It happens to work without strings.h on glibc or with _GNU_SOURCE, but
the POSIX standard requires including <strings.h>.

Hopefully fixes OSX build.
2014-07-10 08:29:32 +02:00