Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
#include <assert.h>
|
|
|
|
#include <string.h>
|
2014-02-14 13:49:10 +01:00
|
|
|
#include <sys/types.h>
|
|
|
|
#include <dirent.h>
|
2014-02-24 20:47:20 +01:00
|
|
|
#include <math.h>
|
2014-02-14 13:49:10 +01:00
|
|
|
|
|
|
|
#include "osdep/io.h"
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
|
|
|
|
#include <lua.h>
|
|
|
|
#include <lualib.h>
|
|
|
|
#include <lauxlib.h>
|
|
|
|
|
|
|
|
#include "talloc.h"
|
|
|
|
|
2013-12-17 02:39:45 +01:00
|
|
|
#include "common/common.h"
|
2013-12-17 02:02:25 +01:00
|
|
|
#include "options/m_property.h"
|
2013-12-17 02:39:45 +01:00
|
|
|
#include "common/msg.h"
|
lua: allow scripts to snoop messages
Adds the following Lua function to enable message events:
mp.enable_messages(size, level)
size is the maximum number of messages the ringbuffer consists of. level
is the minimum log level for a message to be added to the ringbuffer,
and uses the same values as the mp.log() function. (Actually not yet,
but this will be fixed in the following commit.)
The messages will be delivered via the mp_event() in the user script,
using "message" as event name. The event argument is a table with the
following fields:
level: log level of the message (string as in mp.log())
prefix: string identifying the module of origin
text: contents of the message
As of currently, the message text will contain newline characters. A
message can consist of several lines. It is also possible that a
message doesn't end with a newline, and a caller can use multiple
messages to "build" a line. Most messages will contain exactly 1 line
ending with a single newline character, though.
If the message buffer overflows (messages are not read quickly enough),
new messages are lost until the queued up messages are read. At the
point of the overflow, a special overflow message is inserted. It will
have prefix set to "overflow", and the message text is set to "".
Care should be taken not to print any messages from the message event
handler. This would lead to an infinite loop (the event handler would be
called again after returning, because a new message is available). This
includes mp.log() and all mp.msg.* functions. Keep in mind that the Lua
print() function is mapped to mp.msg.info().
2014-01-16 21:34:58 +01:00
|
|
|
#include "common/msg_control.h"
|
2013-12-17 02:02:25 +01:00
|
|
|
#include "options/m_option.h"
|
2013-12-17 01:23:09 +01:00
|
|
|
#include "input/input.h"
|
2013-12-17 02:02:25 +01:00
|
|
|
#include "options/path.h"
|
2013-12-17 02:39:45 +01:00
|
|
|
#include "bstr/bstr.h"
|
2013-10-29 21:35:29 +01:00
|
|
|
#include "osdep/timer.h"
|
2014-02-10 21:03:59 +01:00
|
|
|
#include "osdep/threads.h"
|
2013-11-24 12:58:06 +01:00
|
|
|
#include "sub/osd.h"
|
2013-12-17 01:08:53 +01:00
|
|
|
#include "core.h"
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
#include "command.h"
|
2014-02-10 21:03:59 +01:00
|
|
|
#include "client.h"
|
|
|
|
#include "libmpv/client.h"
|
2013-12-17 01:15:48 +01:00
|
|
|
#include "lua.h"
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
|
|
|
|
// List of builtin modules and their contents as strings.
|
2013-12-17 00:53:22 +01:00
|
|
|
// All these are generated from player/lua/*.lua
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
static const char *builtin_lua_scripts[][2] = {
|
|
|
|
{"mp.defaults",
|
2013-12-17 00:53:22 +01:00
|
|
|
# include "player/lua/defaults.inc"
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
},
|
|
|
|
{"mp.assdraw",
|
2013-12-17 00:53:22 +01:00
|
|
|
# include "player/lua/assdraw.inc"
|
2013-09-26 00:56:41 +02:00
|
|
|
},
|
|
|
|
{"@osc",
|
2013-12-17 00:53:22 +01:00
|
|
|
# include "player/lua/osc.inc"
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
},
|
|
|
|
{0}
|
|
|
|
};
|
|
|
|
|
|
|
|
// Represents a loaded script. Each has its own Lua state.
|
|
|
|
struct script_ctx {
|
|
|
|
const char *name;
|
|
|
|
lua_State *state;
|
|
|
|
struct mp_log *log;
|
2014-02-10 21:03:59 +01:00
|
|
|
struct mpv_handle *client;
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
struct MPContext *mpctx;
|
2014-02-10 21:03:59 +01:00
|
|
|
int suspended;
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
};
|
|
|
|
|
|
|
|
static struct script_ctx *get_ctx(lua_State *L)
|
|
|
|
{
|
|
|
|
lua_getfield(L, LUA_REGISTRYINDEX, "ctx");
|
|
|
|
struct script_ctx *ctx = lua_touserdata(L, -1);
|
|
|
|
lua_pop(L, 1);
|
|
|
|
assert(ctx);
|
|
|
|
return ctx;
|
|
|
|
}
|
|
|
|
|
|
|
|
static struct MPContext *get_mpctx(lua_State *L)
|
|
|
|
{
|
|
|
|
return get_ctx(L)->mpctx;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int wrap_cpcall(lua_State *L)
|
|
|
|
{
|
|
|
|
lua_CFunction fn = lua_touserdata(L, -1);
|
|
|
|
lua_pop(L, 1);
|
|
|
|
return fn(L);
|
|
|
|
}
|
|
|
|
|
|
|
|
// Call the given function fn under a Lua error handler (similar to lua_cpcall).
|
|
|
|
// Pass the given number of args from the Lua stack to fn.
|
|
|
|
// Returns 0 (and empty stack) on success.
|
|
|
|
// Returns LUA_ERR[RUN|MEM|ERR] otherwise, with the error value on the stack.
|
|
|
|
static int mp_cpcall(lua_State *L, lua_CFunction fn, int args)
|
|
|
|
{
|
|
|
|
// Don't use lua_pushcfunction() - it allocates memory on Lua 5.1.
|
|
|
|
// Instead, emulate C closures by making wrap_cpcall call fn.
|
|
|
|
lua_pushlightuserdata(L, fn); // args... fn
|
|
|
|
// Will always succeed if mp_lua_init() set it up correctly.
|
|
|
|
lua_getfield(L, LUA_REGISTRYINDEX, "wrap_cpcall"); // args... fn wrap_cpcall
|
|
|
|
lua_insert(L, -(args + 2)); // wrap_cpcall args... fn
|
|
|
|
return lua_pcall(L, args + 1, 0, 0);
|
|
|
|
}
|
|
|
|
|
|
|
|
static void report_error(lua_State *L)
|
|
|
|
{
|
2014-02-14 13:49:10 +01:00
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
const char *err = lua_tostring(L, -1);
|
2014-02-14 13:49:10 +01:00
|
|
|
MP_WARN(ctx, "Error: %s\n", err ? err : "[unknown]");
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
lua_pop(L, 1);
|
|
|
|
}
|
|
|
|
|
2014-02-10 21:03:59 +01:00
|
|
|
// Check client API error code:
|
2014-02-11 00:10:25 +01:00
|
|
|
// if err >= 0, push "true" to the stack, and return 1
|
|
|
|
// if err < 0, push nil and then the error string to the stack, and return 2
|
2014-02-10 21:03:59 +01:00
|
|
|
static int check_error(lua_State *L, int err)
|
|
|
|
{
|
2014-02-11 00:10:25 +01:00
|
|
|
if (err >= 0) {
|
|
|
|
lua_pushboolean(L, 1);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
lua_pushnil(L);
|
|
|
|
lua_pushstring(L, mpv_error_string(err));
|
|
|
|
return 2;
|
2014-02-10 21:03:59 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
static int run_event_loop(lua_State *L)
|
|
|
|
{
|
|
|
|
lua_getglobal(L, "mp_event_loop");
|
|
|
|
if (lua_isnil(L, -1))
|
|
|
|
luaL_error(L, "no event loop function\n");
|
|
|
|
lua_call(L, 0, 0);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
static void add_functions(struct script_ctx *ctx);
|
|
|
|
|
2014-02-10 21:03:59 +01:00
|
|
|
static char *script_name_from_filename(void *talloc_ctx, const char *fname)
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
{
|
|
|
|
fname = mp_basename(fname);
|
|
|
|
if (fname[0] == '@')
|
|
|
|
fname += 1;
|
|
|
|
char *name = talloc_strdup(talloc_ctx, fname);
|
|
|
|
// Drop .lua extension
|
|
|
|
char *dot = strrchr(name, '.');
|
|
|
|
if (dot)
|
|
|
|
*dot = '\0';
|
|
|
|
// Turn it into a safe identifier - this is used with e.g. dispatching
|
|
|
|
// input via: "send scriptname ..."
|
|
|
|
for (int n = 0; name[n]; n++) {
|
|
|
|
char c = name[n];
|
|
|
|
if (!(c >= 'A' && c <= 'Z') && !(c >= 'a' && c <= 'z') &&
|
|
|
|
!(c >= '0' && c <= '9'))
|
|
|
|
name[n] = '_';
|
|
|
|
}
|
2014-02-10 21:03:59 +01:00
|
|
|
return talloc_asprintf(talloc_ctx, "lua/%s", name);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
static int load_file(struct script_ctx *ctx, const char *fname)
|
|
|
|
{
|
|
|
|
int r = 0;
|
|
|
|
lua_State *L = ctx->state;
|
2014-01-21 21:23:42 +01:00
|
|
|
char *res_name = mp_get_user_path(NULL, ctx->mpctx->global, fname);
|
2014-02-14 13:49:10 +01:00
|
|
|
MP_VERBOSE(ctx, "loading file %s\n", res_name);
|
2014-01-21 21:23:42 +01:00
|
|
|
if (luaL_loadfile(L, res_name) || lua_pcall(L, 0, 0, 0)) {
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
report_error(L);
|
|
|
|
r = -1;
|
|
|
|
}
|
|
|
|
assert(lua_gettop(L) == 0);
|
2014-01-21 21:23:42 +01:00
|
|
|
talloc_free(res_name);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
return r;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int load_builtin(lua_State *L)
|
|
|
|
{
|
|
|
|
const char *name = luaL_checkstring(L, 1);
|
|
|
|
for (int n = 0; builtin_lua_scripts[n][0]; n++) {
|
|
|
|
if (strcmp(name, builtin_lua_scripts[n][0]) == 0) {
|
|
|
|
if (luaL_loadstring(L, builtin_lua_scripts[n][1]))
|
|
|
|
lua_error(L);
|
|
|
|
lua_call(L, 0, 1);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
// Execute "require " .. name
|
|
|
|
static bool require(lua_State *L, const char *name)
|
|
|
|
{
|
2014-02-14 13:49:10 +01:00
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
|
|
|
MP_VERBOSE(ctx, "loading %s\n", name);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
// Lazy, but better than calling the "require" function manually
|
2014-02-14 13:49:10 +01:00
|
|
|
char buf[80];
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
snprintf(buf, sizeof(buf), "require '%s'", name);
|
|
|
|
if (luaL_loadstring(L, buf) || lua_pcall(L, 0, 0, 0)) {
|
|
|
|
report_error(L);
|
|
|
|
return false;
|
|
|
|
}
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
2014-02-10 21:03:59 +01:00
|
|
|
struct thread_arg {
|
|
|
|
struct MPContext *mpctx;
|
|
|
|
mpv_handle *client;
|
|
|
|
const char *fname;
|
|
|
|
};
|
|
|
|
|
|
|
|
static void *lua_thread(void *p)
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
{
|
2014-02-10 21:03:59 +01:00
|
|
|
pthread_detach(pthread_self());
|
|
|
|
|
|
|
|
struct thread_arg *arg = p;
|
|
|
|
struct MPContext *mpctx = arg->mpctx;
|
|
|
|
const char *fname = arg->fname;
|
|
|
|
mpv_handle *client = arg->client;
|
|
|
|
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
struct script_ctx *ctx = talloc_ptrtype(NULL, ctx);
|
|
|
|
*ctx = (struct script_ctx) {
|
|
|
|
.mpctx = mpctx,
|
2014-02-10 21:03:59 +01:00
|
|
|
.client = client,
|
|
|
|
.name = mpv_client_name(client),
|
|
|
|
.log = mp_client_get_log(client),
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
};
|
|
|
|
|
|
|
|
lua_State *L = ctx->state = luaL_newstate();
|
|
|
|
if (!L)
|
|
|
|
goto error_out;
|
|
|
|
|
|
|
|
// used by get_ctx()
|
|
|
|
lua_pushlightuserdata(L, ctx); // ctx
|
|
|
|
lua_setfield(L, LUA_REGISTRYINDEX, "ctx"); // -
|
|
|
|
|
|
|
|
lua_pushcfunction(L, wrap_cpcall); // closure
|
|
|
|
lua_setfield(L, LUA_REGISTRYINDEX, "wrap_cpcall"); // -
|
|
|
|
|
|
|
|
luaL_openlibs(L);
|
|
|
|
|
|
|
|
lua_newtable(L); // mp
|
|
|
|
lua_pushvalue(L, -1); // mp mp
|
|
|
|
lua_setglobal(L, "mp"); // mp
|
|
|
|
|
|
|
|
add_functions(ctx); // mp
|
|
|
|
|
|
|
|
lua_pushstring(L, ctx->name); // mp name
|
|
|
|
lua_setfield(L, -2, "script_name"); // mp
|
|
|
|
|
2014-02-24 20:47:20 +01:00
|
|
|
// used by pushnode()
|
|
|
|
lua_newtable(L); // mp table
|
|
|
|
lua_pushvalue(L, -1); // mp table table
|
|
|
|
lua_setfield(L, LUA_REGISTRYINDEX, "UNKNOWN_TYPE"); // mp table
|
|
|
|
lua_setfield(L, -2, "UNKNOWN_TYPE"); // mp
|
|
|
|
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
lua_pop(L, 1); // -
|
|
|
|
|
2014-02-24 20:47:20 +01:00
|
|
|
assert(lua_gettop(L) == 0);
|
|
|
|
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
// Add a preloader for each builtin Lua module
|
|
|
|
lua_getglobal(L, "package"); // package
|
|
|
|
assert(lua_type(L, -1) == LUA_TTABLE);
|
|
|
|
lua_getfield(L, -1, "preload"); // package preload
|
|
|
|
assert(lua_type(L, -1) == LUA_TTABLE);
|
|
|
|
for (int n = 0; builtin_lua_scripts[n][0]; n++) {
|
|
|
|
lua_pushcfunction(L, load_builtin); // package preload load_builtin
|
|
|
|
lua_setfield(L, -2, builtin_lua_scripts[n][0]);
|
|
|
|
}
|
|
|
|
lua_pop(L, 2); // -
|
|
|
|
|
|
|
|
assert(lua_gettop(L) == 0);
|
|
|
|
|
|
|
|
if (!require(L, "mp.defaults")) {
|
|
|
|
report_error(L);
|
|
|
|
goto error_out;
|
|
|
|
}
|
|
|
|
|
|
|
|
assert(lua_gettop(L) == 0);
|
|
|
|
|
|
|
|
if (fname[0] == '@') {
|
|
|
|
if (!require(L, fname))
|
|
|
|
goto error_out;
|
|
|
|
} else {
|
|
|
|
if (load_file(ctx, fname) < 0)
|
|
|
|
goto error_out;
|
|
|
|
}
|
|
|
|
|
2014-02-10 21:03:59 +01:00
|
|
|
// Call the script's event loop runs until the script terminates and unloads.
|
|
|
|
if (mp_cpcall(L, run_event_loop, 0) != 0)
|
|
|
|
report_error(L);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
|
|
|
|
error_out:
|
2014-02-14 13:49:10 +01:00
|
|
|
MP_VERBOSE(ctx, "exiting.\n");
|
2014-02-10 21:03:59 +01:00
|
|
|
if (ctx->suspended)
|
|
|
|
mpv_resume(ctx->client);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
if (ctx->state)
|
|
|
|
lua_close(ctx->state);
|
2014-02-10 21:03:59 +01:00
|
|
|
mpv_destroy(ctx->client);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
talloc_free(ctx);
|
2014-02-10 21:03:59 +01:00
|
|
|
talloc_free(arg);
|
|
|
|
return NULL;
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
}
|
|
|
|
|
2014-02-10 21:03:59 +01:00
|
|
|
static void mp_lua_load_script(struct MPContext *mpctx, const char *fname)
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
{
|
2014-02-10 21:03:59 +01:00
|
|
|
struct thread_arg *arg = talloc_ptrtype(NULL, arg);
|
|
|
|
char *name = script_name_from_filename(arg, fname);
|
|
|
|
*arg = (struct thread_arg){
|
|
|
|
.mpctx = mpctx,
|
|
|
|
.fname = talloc_strdup(arg, fname),
|
|
|
|
// Create the client before creating the thread; otherwise a race
|
|
|
|
// condition could happen, where MPContext is destroyed while the
|
|
|
|
// thread tries to create the client.
|
|
|
|
.client = mp_new_client(mpctx->clients, name),
|
|
|
|
};
|
|
|
|
if (!arg->client) {
|
|
|
|
talloc_free(arg);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
return;
|
|
|
|
}
|
2014-02-10 21:03:59 +01:00
|
|
|
|
|
|
|
pthread_t thread;
|
|
|
|
if (pthread_create(&thread, NULL, lua_thread, arg))
|
|
|
|
talloc_free(arg);
|
|
|
|
|
|
|
|
return;
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
}
|
|
|
|
|
lua: allow scripts to snoop messages
Adds the following Lua function to enable message events:
mp.enable_messages(size, level)
size is the maximum number of messages the ringbuffer consists of. level
is the minimum log level for a message to be added to the ringbuffer,
and uses the same values as the mp.log() function. (Actually not yet,
but this will be fixed in the following commit.)
The messages will be delivered via the mp_event() in the user script,
using "message" as event name. The event argument is a table with the
following fields:
level: log level of the message (string as in mp.log())
prefix: string identifying the module of origin
text: contents of the message
As of currently, the message text will contain newline characters. A
message can consist of several lines. It is also possible that a
message doesn't end with a newline, and a caller can use multiple
messages to "build" a line. Most messages will contain exactly 1 line
ending with a single newline character, though.
If the message buffer overflows (messages are not read quickly enough),
new messages are lost until the queued up messages are read. At the
point of the overflow, a special overflow message is inserted. It will
have prefix set to "overflow", and the message text is set to "".
Care should be taken not to print any messages from the message event
handler. This would lead to an infinite loop (the event handler would be
called again after returning, because a new message is available). This
includes mp.log() and all mp.msg.* functions. Keep in mind that the Lua
print() function is mapped to mp.msg.info().
2014-01-16 21:34:58 +01:00
|
|
|
static int check_loglevel(lua_State *L, int arg)
|
|
|
|
{
|
|
|
|
const char *level = luaL_checkstring(L, arg);
|
2014-01-16 21:37:29 +01:00
|
|
|
for (int n = 0; n < MSGL_MAX; n++) {
|
|
|
|
if (mp_log_levels[n] && strcasecmp(mp_log_levels[n], level) == 0)
|
lua: allow scripts to snoop messages
Adds the following Lua function to enable message events:
mp.enable_messages(size, level)
size is the maximum number of messages the ringbuffer consists of. level
is the minimum log level for a message to be added to the ringbuffer,
and uses the same values as the mp.log() function. (Actually not yet,
but this will be fixed in the following commit.)
The messages will be delivered via the mp_event() in the user script,
using "message" as event name. The event argument is a table with the
following fields:
level: log level of the message (string as in mp.log())
prefix: string identifying the module of origin
text: contents of the message
As of currently, the message text will contain newline characters. A
message can consist of several lines. It is also possible that a
message doesn't end with a newline, and a caller can use multiple
messages to "build" a line. Most messages will contain exactly 1 line
ending with a single newline character, though.
If the message buffer overflows (messages are not read quickly enough),
new messages are lost until the queued up messages are read. At the
point of the overflow, a special overflow message is inserted. It will
have prefix set to "overflow", and the message text is set to "".
Care should be taken not to print any messages from the message event
handler. This would lead to an infinite loop (the event handler would be
called again after returning, because a new message is available). This
includes mp.log() and all mp.msg.* functions. Keep in mind that the Lua
print() function is mapped to mp.msg.info().
2014-01-16 21:34:58 +01:00
|
|
|
return n;
|
|
|
|
}
|
|
|
|
luaL_error(L, "Invalid log level '%s'", level);
|
|
|
|
abort();
|
|
|
|
}
|
|
|
|
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
static int script_log(lua_State *L)
|
|
|
|
{
|
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
|
|
|
|
lua: allow scripts to snoop messages
Adds the following Lua function to enable message events:
mp.enable_messages(size, level)
size is the maximum number of messages the ringbuffer consists of. level
is the minimum log level for a message to be added to the ringbuffer,
and uses the same values as the mp.log() function. (Actually not yet,
but this will be fixed in the following commit.)
The messages will be delivered via the mp_event() in the user script,
using "message" as event name. The event argument is a table with the
following fields:
level: log level of the message (string as in mp.log())
prefix: string identifying the module of origin
text: contents of the message
As of currently, the message text will contain newline characters. A
message can consist of several lines. It is also possible that a
message doesn't end with a newline, and a caller can use multiple
messages to "build" a line. Most messages will contain exactly 1 line
ending with a single newline character, though.
If the message buffer overflows (messages are not read quickly enough),
new messages are lost until the queued up messages are read. At the
point of the overflow, a special overflow message is inserted. It will
have prefix set to "overflow", and the message text is set to "".
Care should be taken not to print any messages from the message event
handler. This would lead to an infinite loop (the event handler would be
called again after returning, because a new message is available). This
includes mp.log() and all mp.msg.* functions. Keep in mind that the Lua
print() function is mapped to mp.msg.info().
2014-01-16 21:34:58 +01:00
|
|
|
int msgl = check_loglevel(L, 1);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
|
|
|
|
int last = lua_gettop(L);
|
|
|
|
lua_getglobal(L, "tostring"); // args... tostring
|
|
|
|
for (int i = 2; i <= last; i++) {
|
|
|
|
lua_pushvalue(L, -1); // args... tostring tostring
|
|
|
|
lua_pushvalue(L, i); // args... tostring tostring args[i]
|
|
|
|
lua_call(L, 1, 1); // args... tostring str
|
|
|
|
const char *s = lua_tostring(L, -1);
|
|
|
|
if (s == NULL)
|
|
|
|
return luaL_error(L, "Invalid argument");
|
2013-12-21 21:49:13 +01:00
|
|
|
mp_msg(ctx->log, msgl, "%s%s", s, i > 0 ? " " : "");
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
lua_pop(L, 1); // args... tostring
|
|
|
|
}
|
2013-12-21 21:49:13 +01:00
|
|
|
mp_msg(ctx->log, msgl, "\n");
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int script_find_config_file(lua_State *L)
|
|
|
|
{
|
2013-12-21 20:45:19 +01:00
|
|
|
struct MPContext *mpctx = get_mpctx(L);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
const char *s = luaL_checkstring(L, 1);
|
2013-12-21 20:45:19 +01:00
|
|
|
char *path = mp_find_user_config_file(NULL, mpctx->global, s);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
if (path) {
|
|
|
|
lua_pushstring(L, path);
|
|
|
|
} else {
|
|
|
|
lua_pushnil(L);
|
|
|
|
}
|
|
|
|
talloc_free(path);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2014-02-10 21:03:59 +01:00
|
|
|
static int script_suspend(lua_State *L)
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
{
|
2014-02-10 21:03:59 +01:00
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
|
|
|
if (!ctx->suspended)
|
|
|
|
mpv_suspend(ctx->client);
|
|
|
|
ctx->suspended++;
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
2014-02-10 21:03:59 +01:00
|
|
|
static int script_resume(lua_State *L)
|
lua: allow scripts to snoop messages
Adds the following Lua function to enable message events:
mp.enable_messages(size, level)
size is the maximum number of messages the ringbuffer consists of. level
is the minimum log level for a message to be added to the ringbuffer,
and uses the same values as the mp.log() function. (Actually not yet,
but this will be fixed in the following commit.)
The messages will be delivered via the mp_event() in the user script,
using "message" as event name. The event argument is a table with the
following fields:
level: log level of the message (string as in mp.log())
prefix: string identifying the module of origin
text: contents of the message
As of currently, the message text will contain newline characters. A
message can consist of several lines. It is also possible that a
message doesn't end with a newline, and a caller can use multiple
messages to "build" a line. Most messages will contain exactly 1 line
ending with a single newline character, though.
If the message buffer overflows (messages are not read quickly enough),
new messages are lost until the queued up messages are read. At the
point of the overflow, a special overflow message is inserted. It will
have prefix set to "overflow", and the message text is set to "".
Care should be taken not to print any messages from the message event
handler. This would lead to an infinite loop (the event handler would be
called again after returning, because a new message is available). This
includes mp.log() and all mp.msg.* functions. Keep in mind that the Lua
print() function is mapped to mp.msg.info().
2014-01-16 21:34:58 +01:00
|
|
|
{
|
2014-02-10 21:03:59 +01:00
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
2014-02-11 00:57:40 +01:00
|
|
|
if (ctx->suspended < 1)
|
|
|
|
luaL_error(L, "trying to resume, but core is not suspended");
|
|
|
|
ctx->suspended--;
|
|
|
|
if (!ctx->suspended)
|
|
|
|
mpv_resume(ctx->client);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int script_resume_all(lua_State *L)
|
|
|
|
{
|
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
|
|
|
if (ctx->suspended)
|
|
|
|
mpv_resume(ctx->client);
|
|
|
|
ctx->suspended = 0;
|
2014-02-10 21:03:59 +01:00
|
|
|
return 0;
|
|
|
|
}
|
lua: allow scripts to snoop messages
Adds the following Lua function to enable message events:
mp.enable_messages(size, level)
size is the maximum number of messages the ringbuffer consists of. level
is the minimum log level for a message to be added to the ringbuffer,
and uses the same values as the mp.log() function. (Actually not yet,
but this will be fixed in the following commit.)
The messages will be delivered via the mp_event() in the user script,
using "message" as event name. The event argument is a table with the
following fields:
level: log level of the message (string as in mp.log())
prefix: string identifying the module of origin
text: contents of the message
As of currently, the message text will contain newline characters. A
message can consist of several lines. It is also possible that a
message doesn't end with a newline, and a caller can use multiple
messages to "build" a line. Most messages will contain exactly 1 line
ending with a single newline character, though.
If the message buffer overflows (messages are not read quickly enough),
new messages are lost until the queued up messages are read. At the
point of the overflow, a special overflow message is inserted. It will
have prefix set to "overflow", and the message text is set to "".
Care should be taken not to print any messages from the message event
handler. This would lead to an infinite loop (the event handler would be
called again after returning, because a new message is available). This
includes mp.log() and all mp.msg.* functions. Keep in mind that the Lua
print() function is mapped to mp.msg.info().
2014-01-16 21:34:58 +01:00
|
|
|
|
2014-02-10 21:03:59 +01:00
|
|
|
static int script_wait_event(lua_State *L)
|
|
|
|
{
|
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
lua: allow scripts to snoop messages
Adds the following Lua function to enable message events:
mp.enable_messages(size, level)
size is the maximum number of messages the ringbuffer consists of. level
is the minimum log level for a message to be added to the ringbuffer,
and uses the same values as the mp.log() function. (Actually not yet,
but this will be fixed in the following commit.)
The messages will be delivered via the mp_event() in the user script,
using "message" as event name. The event argument is a table with the
following fields:
level: log level of the message (string as in mp.log())
prefix: string identifying the module of origin
text: contents of the message
As of currently, the message text will contain newline characters. A
message can consist of several lines. It is also possible that a
message doesn't end with a newline, and a caller can use multiple
messages to "build" a line. Most messages will contain exactly 1 line
ending with a single newline character, though.
If the message buffer overflows (messages are not read quickly enough),
new messages are lost until the queued up messages are read. At the
point of the overflow, a special overflow message is inserted. It will
have prefix set to "overflow", and the message text is set to "".
Care should be taken not to print any messages from the message event
handler. This would lead to an infinite loop (the event handler would be
called again after returning, because a new message is available). This
includes mp.log() and all mp.msg.* functions. Keep in mind that the Lua
print() function is mapped to mp.msg.info().
2014-01-16 21:34:58 +01:00
|
|
|
|
2014-02-10 21:03:59 +01:00
|
|
|
double timeout = luaL_optnumber(L, 1, 1e20);
|
|
|
|
|
|
|
|
// This will almost surely lead to a deadlock. (Polling is still ok.)
|
|
|
|
if (ctx->suspended && timeout > 0)
|
|
|
|
luaL_error(L, "attempting to wait while core is suspended");
|
|
|
|
|
|
|
|
mpv_event *event = mpv_wait_event(ctx->client, timeout);
|
lua: allow scripts to snoop messages
Adds the following Lua function to enable message events:
mp.enable_messages(size, level)
size is the maximum number of messages the ringbuffer consists of. level
is the minimum log level for a message to be added to the ringbuffer,
and uses the same values as the mp.log() function. (Actually not yet,
but this will be fixed in the following commit.)
The messages will be delivered via the mp_event() in the user script,
using "message" as event name. The event argument is a table with the
following fields:
level: log level of the message (string as in mp.log())
prefix: string identifying the module of origin
text: contents of the message
As of currently, the message text will contain newline characters. A
message can consist of several lines. It is also possible that a
message doesn't end with a newline, and a caller can use multiple
messages to "build" a line. Most messages will contain exactly 1 line
ending with a single newline character, though.
If the message buffer overflows (messages are not read quickly enough),
new messages are lost until the queued up messages are read. At the
point of the overflow, a special overflow message is inserted. It will
have prefix set to "overflow", and the message text is set to "".
Care should be taken not to print any messages from the message event
handler. This would lead to an infinite loop (the event handler would be
called again after returning, because a new message is available). This
includes mp.log() and all mp.msg.* functions. Keep in mind that the Lua
print() function is mapped to mp.msg.info().
2014-01-16 21:34:58 +01:00
|
|
|
|
2014-02-10 21:03:59 +01:00
|
|
|
lua_newtable(L); // event
|
|
|
|
lua_pushstring(L, mpv_event_name(event->event_id)); // event name
|
|
|
|
lua_setfield(L, -2, "event"); // event
|
lua: allow scripts to snoop messages
Adds the following Lua function to enable message events:
mp.enable_messages(size, level)
size is the maximum number of messages the ringbuffer consists of. level
is the minimum log level for a message to be added to the ringbuffer,
and uses the same values as the mp.log() function. (Actually not yet,
but this will be fixed in the following commit.)
The messages will be delivered via the mp_event() in the user script,
using "message" as event name. The event argument is a table with the
following fields:
level: log level of the message (string as in mp.log())
prefix: string identifying the module of origin
text: contents of the message
As of currently, the message text will contain newline characters. A
message can consist of several lines. It is also possible that a
message doesn't end with a newline, and a caller can use multiple
messages to "build" a line. Most messages will contain exactly 1 line
ending with a single newline character, though.
If the message buffer overflows (messages are not read quickly enough),
new messages are lost until the queued up messages are read. At the
point of the overflow, a special overflow message is inserted. It will
have prefix set to "overflow", and the message text is set to "".
Care should be taken not to print any messages from the message event
handler. This would lead to an infinite loop (the event handler would be
called again after returning, because a new message is available). This
includes mp.log() and all mp.msg.* functions. Keep in mind that the Lua
print() function is mapped to mp.msg.info().
2014-01-16 21:34:58 +01:00
|
|
|
|
2014-02-10 21:03:59 +01:00
|
|
|
if (event->error < 0) {
|
|
|
|
lua_pushstring(L, mpv_error_string(event->error)); // event err
|
|
|
|
lua_setfield(L, -2, "error"); // event
|
|
|
|
}
|
lua: allow scripts to snoop messages
Adds the following Lua function to enable message events:
mp.enable_messages(size, level)
size is the maximum number of messages the ringbuffer consists of. level
is the minimum log level for a message to be added to the ringbuffer,
and uses the same values as the mp.log() function. (Actually not yet,
but this will be fixed in the following commit.)
The messages will be delivered via the mp_event() in the user script,
using "message" as event name. The event argument is a table with the
following fields:
level: log level of the message (string as in mp.log())
prefix: string identifying the module of origin
text: contents of the message
As of currently, the message text will contain newline characters. A
message can consist of several lines. It is also possible that a
message doesn't end with a newline, and a caller can use multiple
messages to "build" a line. Most messages will contain exactly 1 line
ending with a single newline character, though.
If the message buffer overflows (messages are not read quickly enough),
new messages are lost until the queued up messages are read. At the
point of the overflow, a special overflow message is inserted. It will
have prefix set to "overflow", and the message text is set to "".
Care should be taken not to print any messages from the message event
handler. This would lead to an infinite loop (the event handler would be
called again after returning, because a new message is available). This
includes mp.log() and all mp.msg.* functions. Keep in mind that the Lua
print() function is mapped to mp.msg.info().
2014-01-16 21:34:58 +01:00
|
|
|
|
2014-02-10 21:03:59 +01:00
|
|
|
switch (event->event_id) {
|
|
|
|
case MPV_EVENT_LOG_MESSAGE: {
|
|
|
|
mpv_event_log_message *msg = event->data;
|
|
|
|
|
|
|
|
lua_pushstring(L, msg->prefix); // event s
|
|
|
|
lua_setfield(L, -2, "prefix"); // event
|
|
|
|
lua_pushstring(L, msg->level); // event s
|
|
|
|
lua_setfield(L, -2, "level"); // event
|
|
|
|
lua_pushstring(L, msg->text); // event s
|
|
|
|
lua_setfield(L, -2, "text"); // event
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
case MPV_EVENT_SCRIPT_INPUT_DISPATCH: {
|
|
|
|
mpv_event_script_input_dispatch *msg = event->data;
|
|
|
|
|
|
|
|
lua_pushinteger(L, msg->arg0); // event i
|
|
|
|
lua_setfield(L, -2, "arg0"); // event
|
|
|
|
lua_pushstring(L, msg->type); // event s
|
|
|
|
lua_setfield(L, -2, "type"); // event
|
|
|
|
break;
|
|
|
|
}
|
2014-02-17 02:33:47 +01:00
|
|
|
case MPV_EVENT_CLIENT_MESSAGE: {
|
|
|
|
mpv_event_client_message *msg = event->data;
|
|
|
|
|
|
|
|
lua_newtable(L); // event args
|
|
|
|
for (int n = 0; n < msg->num_args; n++) {
|
|
|
|
lua_pushinteger(L, n + 1); // event args N
|
|
|
|
lua_pushstring(L, msg->args[n]); // event args N val
|
|
|
|
lua_settable(L, -3); // event args
|
|
|
|
}
|
|
|
|
lua_setfield(L, -2, "args"); // event
|
|
|
|
break;
|
|
|
|
}
|
2014-02-10 21:03:59 +01:00
|
|
|
default: ;
|
lua: allow scripts to snoop messages
Adds the following Lua function to enable message events:
mp.enable_messages(size, level)
size is the maximum number of messages the ringbuffer consists of. level
is the minimum log level for a message to be added to the ringbuffer,
and uses the same values as the mp.log() function. (Actually not yet,
but this will be fixed in the following commit.)
The messages will be delivered via the mp_event() in the user script,
using "message" as event name. The event argument is a table with the
following fields:
level: log level of the message (string as in mp.log())
prefix: string identifying the module of origin
text: contents of the message
As of currently, the message text will contain newline characters. A
message can consist of several lines. It is also possible that a
message doesn't end with a newline, and a caller can use multiple
messages to "build" a line. Most messages will contain exactly 1 line
ending with a single newline character, though.
If the message buffer overflows (messages are not read quickly enough),
new messages are lost until the queued up messages are read. At the
point of the overflow, a special overflow message is inserted. It will
have prefix set to "overflow", and the message text is set to "".
Care should be taken not to print any messages from the message event
handler. This would lead to an infinite loop (the event handler would be
called again after returning, because a new message is available). This
includes mp.log() and all mp.msg.* functions. Keep in mind that the Lua
print() function is mapped to mp.msg.info().
2014-01-16 21:34:58 +01:00
|
|
|
}
|
2014-02-10 21:03:59 +01:00
|
|
|
|
2014-02-17 02:33:47 +01:00
|
|
|
// return event
|
2014-02-10 21:03:59 +01:00
|
|
|
return 1;
|
lua: allow scripts to snoop messages
Adds the following Lua function to enable message events:
mp.enable_messages(size, level)
size is the maximum number of messages the ringbuffer consists of. level
is the minimum log level for a message to be added to the ringbuffer,
and uses the same values as the mp.log() function. (Actually not yet,
but this will be fixed in the following commit.)
The messages will be delivered via the mp_event() in the user script,
using "message" as event name. The event argument is a table with the
following fields:
level: log level of the message (string as in mp.log())
prefix: string identifying the module of origin
text: contents of the message
As of currently, the message text will contain newline characters. A
message can consist of several lines. It is also possible that a
message doesn't end with a newline, and a caller can use multiple
messages to "build" a line. Most messages will contain exactly 1 line
ending with a single newline character, though.
If the message buffer overflows (messages are not read quickly enough),
new messages are lost until the queued up messages are read. At the
point of the overflow, a special overflow message is inserted. It will
have prefix set to "overflow", and the message text is set to "".
Care should be taken not to print any messages from the message event
handler. This would lead to an infinite loop (the event handler would be
called again after returning, because a new message is available). This
includes mp.log() and all mp.msg.* functions. Keep in mind that the Lua
print() function is mapped to mp.msg.info().
2014-01-16 21:34:58 +01:00
|
|
|
}
|
|
|
|
|
2014-02-10 21:03:59 +01:00
|
|
|
static int script_request_event(lua_State *L)
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
{
|
2014-02-10 21:03:59 +01:00
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
|
|
|
const char *event = luaL_checkstring(L, 1);
|
|
|
|
bool enable = lua_toboolean(L, 2);
|
|
|
|
// brute force event name -> id; stops working for events > assumed max
|
|
|
|
int event_id = -1;
|
|
|
|
for (int n = 0; n < 256; n++) {
|
|
|
|
const char *name = mpv_event_name(n);
|
|
|
|
if (name && strcmp(name, event) == 0) {
|
|
|
|
event_id = n;
|
|
|
|
break;
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
}
|
|
|
|
}
|
2014-02-10 21:03:59 +01:00
|
|
|
lua_pushboolean(L, mpv_request_event(ctx->client, event_id, enable) >= 0);
|
|
|
|
return 1;
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
}
|
|
|
|
|
lua: allow scripts to snoop messages
Adds the following Lua function to enable message events:
mp.enable_messages(size, level)
size is the maximum number of messages the ringbuffer consists of. level
is the minimum log level for a message to be added to the ringbuffer,
and uses the same values as the mp.log() function. (Actually not yet,
but this will be fixed in the following commit.)
The messages will be delivered via the mp_event() in the user script,
using "message" as event name. The event argument is a table with the
following fields:
level: log level of the message (string as in mp.log())
prefix: string identifying the module of origin
text: contents of the message
As of currently, the message text will contain newline characters. A
message can consist of several lines. It is also possible that a
message doesn't end with a newline, and a caller can use multiple
messages to "build" a line. Most messages will contain exactly 1 line
ending with a single newline character, though.
If the message buffer overflows (messages are not read quickly enough),
new messages are lost until the queued up messages are read. At the
point of the overflow, a special overflow message is inserted. It will
have prefix set to "overflow", and the message text is set to "".
Care should be taken not to print any messages from the message event
handler. This would lead to an infinite loop (the event handler would be
called again after returning, because a new message is available). This
includes mp.log() and all mp.msg.* functions. Keep in mind that the Lua
print() function is mapped to mp.msg.info().
2014-01-16 21:34:58 +01:00
|
|
|
static int script_enable_messages(lua_State *L)
|
|
|
|
{
|
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
2014-02-10 21:03:59 +01:00
|
|
|
check_loglevel(L, 1);
|
|
|
|
const char *level = luaL_checkstring(L, 1);
|
|
|
|
return check_error(L, mpv_request_log_messages(ctx->client, level));
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
}
|
|
|
|
|
2014-02-10 22:42:20 +01:00
|
|
|
static int script_command(lua_State *L)
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
{
|
2014-02-10 21:03:59 +01:00
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
const char *s = luaL_checkstring(L, 1);
|
|
|
|
|
2014-02-10 21:03:59 +01:00
|
|
|
return check_error(L, mpv_command_string(ctx->client, s));
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
}
|
|
|
|
|
2014-02-10 22:42:20 +01:00
|
|
|
static int script_commandv(lua_State *L)
|
2013-12-20 18:01:04 +01:00
|
|
|
{
|
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
|
|
|
int num = lua_gettop(L);
|
2014-02-10 21:03:59 +01:00
|
|
|
const char *args[50];
|
|
|
|
if (num + 1 > MP_ARRAY_SIZE(args))
|
2013-12-20 18:01:04 +01:00
|
|
|
luaL_error(L, "too many arguments");
|
|
|
|
for (int n = 1; n <= num; n++) {
|
2014-02-10 21:03:59 +01:00
|
|
|
const char *s = lua_tostring(L, n);
|
2013-12-20 18:01:04 +01:00
|
|
|
if (!s)
|
|
|
|
luaL_error(L, "argument %d is not a string", n);
|
2014-02-10 21:03:59 +01:00
|
|
|
args[n - 1] = s;
|
2013-12-20 18:01:04 +01:00
|
|
|
}
|
2014-02-10 21:03:59 +01:00
|
|
|
args[num] = NULL;
|
|
|
|
return check_error(L, mpv_command(ctx->client, args));
|
2013-12-20 18:01:04 +01:00
|
|
|
}
|
|
|
|
|
2014-02-11 00:23:10 +01:00
|
|
|
static int script_set_property(lua_State *L)
|
|
|
|
{
|
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
|
|
|
const char *p = luaL_checkstring(L, 1);
|
|
|
|
const char *v = luaL_checkstring(L, 2);
|
|
|
|
|
|
|
|
return check_error(L, mpv_set_property_string(ctx->client, p, v));
|
|
|
|
}
|
|
|
|
|
2014-02-24 20:47:20 +01:00
|
|
|
static int script_set_property_bool(lua_State *L)
|
|
|
|
{
|
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
|
|
|
const char *p = luaL_checkstring(L, 1);
|
|
|
|
int v = lua_toboolean(L, 2);
|
|
|
|
|
|
|
|
return check_error(L, mpv_set_property(ctx->client, p, MPV_FORMAT_FLAG, &v));
|
|
|
|
}
|
|
|
|
|
|
|
|
static int script_set_property_number(lua_State *L)
|
|
|
|
{
|
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
|
|
|
const char *p = luaL_checkstring(L, 1);
|
|
|
|
double d = luaL_checknumber(L, 2);
|
|
|
|
// If the number might be an integer, then set it as integer. The mpv core
|
|
|
|
// will (probably) convert INT64 to DOUBLE when setting, but not the other
|
|
|
|
// way around.
|
|
|
|
int64_t v = d;
|
|
|
|
int res;
|
|
|
|
if (d == (double)v) {
|
|
|
|
res = mpv_set_property(ctx->client, p, MPV_FORMAT_INT64, &v);
|
|
|
|
} else {
|
|
|
|
res = mpv_set_property(ctx->client, p, MPV_FORMAT_DOUBLE, &d);
|
|
|
|
}
|
|
|
|
return check_error(L, res);
|
|
|
|
}
|
|
|
|
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
static int script_property_list(lua_State *L)
|
|
|
|
{
|
|
|
|
const struct m_option *props = mp_get_property_list();
|
|
|
|
lua_newtable(L);
|
|
|
|
for (int i = 0; props[i].name; i++) {
|
|
|
|
lua_pushinteger(L, i + 1);
|
|
|
|
lua_pushstring(L, props[i].name);
|
|
|
|
lua_settable(L, -3);
|
|
|
|
}
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2014-02-10 22:42:20 +01:00
|
|
|
static int script_get_property(lua_State *L)
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
{
|
2014-02-10 21:03:59 +01:00
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
const char *name = luaL_checkstring(L, 1);
|
|
|
|
int type = lua_tointeger(L, lua_upvalueindex(1))
|
2014-02-10 21:03:59 +01:00
|
|
|
? MPV_FORMAT_OSD_STRING : MPV_FORMAT_STRING;
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
|
|
|
|
char *result = NULL;
|
2014-02-10 21:03:59 +01:00
|
|
|
int err = mpv_get_property(ctx->client, name, type, &result);
|
|
|
|
if (err >= 0) {
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
lua_pushstring(L, result);
|
|
|
|
talloc_free(result);
|
|
|
|
return 1;
|
2014-02-24 20:47:20 +01:00
|
|
|
} else {
|
|
|
|
if (lua_isnoneornil(L, 2) && type == MPV_FORMAT_OSD_STRING) {
|
|
|
|
lua_pushstring(L, "");
|
|
|
|
} else {
|
|
|
|
lua_pushvalue(L, 2);
|
|
|
|
}
|
|
|
|
lua_pushstring(L, mpv_error_string(err));
|
|
|
|
return 2;
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
}
|
2014-02-24 20:47:20 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
static int script_get_property_bool(lua_State *L)
|
|
|
|
{
|
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
|
|
|
const char *name = luaL_checkstring(L, 1);
|
|
|
|
|
|
|
|
int result = 0;
|
|
|
|
int err = mpv_get_property(ctx->client, name, MPV_FORMAT_FLAG, &result);
|
|
|
|
if (err >= 0) {
|
|
|
|
lua_pushboolean(L, !!result);
|
|
|
|
return 1;
|
|
|
|
} else {
|
|
|
|
lua_pushvalue(L, 2);
|
|
|
|
lua_pushstring(L, mpv_error_string(err));
|
|
|
|
return 2;
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static int script_get_property_number(lua_State *L)
|
|
|
|
{
|
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
|
|
|
const char *name = luaL_checkstring(L, 1);
|
|
|
|
|
|
|
|
// Note: the mpv core will (hopefully) convert INT64 to DOUBLE
|
|
|
|
double result = 0;
|
|
|
|
int err = mpv_get_property(ctx->client, name, MPV_FORMAT_DOUBLE, &result);
|
|
|
|
if (err >= 0) {
|
|
|
|
lua_pushnumber(L, result);
|
|
|
|
return 1;
|
|
|
|
} else {
|
|
|
|
lua_pushvalue(L, 2);
|
2014-02-11 00:10:25 +01:00
|
|
|
lua_pushstring(L, mpv_error_string(err));
|
2014-02-17 00:19:30 +01:00
|
|
|
return 2;
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
}
|
2014-02-24 20:47:20 +01:00
|
|
|
}
|
|
|
|
|
|
|
|
static bool pushnode(lua_State *L, mpv_node *node, int depth)
|
|
|
|
{
|
|
|
|
depth--;
|
|
|
|
if (depth < 0)
|
|
|
|
return false;
|
|
|
|
luaL_checkstack(L, 6, "stack overflow");
|
|
|
|
|
|
|
|
switch (node->format) {
|
|
|
|
case MPV_FORMAT_STRING:
|
|
|
|
lua_pushstring(L, node->u.string);
|
|
|
|
break;
|
|
|
|
case MPV_FORMAT_INT64:
|
|
|
|
lua_pushnumber(L, node->u.int64);
|
|
|
|
break;
|
|
|
|
case MPV_FORMAT_DOUBLE:
|
|
|
|
lua_pushnumber(L, node->u.double_);
|
|
|
|
break;
|
|
|
|
case MPV_FORMAT_NONE:
|
|
|
|
lua_pushnil(L);
|
|
|
|
break;
|
|
|
|
case MPV_FORMAT_FLAG:
|
|
|
|
lua_pushboolean(L, node->u.flag);
|
|
|
|
break;
|
|
|
|
case MPV_FORMAT_NODE_ARRAY:
|
|
|
|
lua_newtable(L); // table
|
|
|
|
for (int n = 0; n < node->u.list->num; n++) {
|
|
|
|
if (!pushnode(L, &node->u.list->values[n], depth)) // table value
|
|
|
|
return false;
|
|
|
|
lua_rawseti(L, -2, n + 1); // table
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
case MPV_FORMAT_NODE_MAP:
|
|
|
|
lua_newtable(L); // table
|
|
|
|
for (int n = 0; n < node->u.list->num; n++) {
|
|
|
|
lua_pushstring(L, node->u.list->keys[n]); // table key
|
|
|
|
if (!pushnode(L, &node->u.list->values[n], depth)) // table key value
|
|
|
|
return false;
|
|
|
|
lua_rawset(L, -3);
|
|
|
|
}
|
|
|
|
break;
|
|
|
|
default:
|
|
|
|
// unknown value - what do we do?
|
|
|
|
// for now, set a unique dummy value
|
|
|
|
lua_getfield(L, LUA_REGISTRYINDEX, "UNKNOWN_TYPE");
|
|
|
|
break;
|
|
|
|
}
|
|
|
|
return true;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int script_get_property_native(lua_State *L)
|
|
|
|
{
|
|
|
|
struct script_ctx *ctx = get_ctx(L);
|
|
|
|
const char *name = luaL_checkstring(L, 1);
|
|
|
|
|
|
|
|
mpv_node node;
|
|
|
|
int err = mpv_get_property(ctx->client, name, MPV_FORMAT_NODE, &node);
|
|
|
|
const char *errstr = mpv_error_string(err);
|
|
|
|
if (err >= 0) {
|
|
|
|
bool ok = pushnode(L, &node, 50);
|
|
|
|
mpv_free_node_contents(&node);
|
|
|
|
if (ok)
|
|
|
|
return 1;
|
|
|
|
errstr = "value too large";
|
|
|
|
}
|
|
|
|
lua_pushvalue(L, 2);
|
|
|
|
lua_pushstring(L, errstr);
|
|
|
|
return 2;
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
}
|
|
|
|
|
|
|
|
static int script_set_osd_ass(lua_State *L)
|
|
|
|
{
|
|
|
|
struct MPContext *mpctx = get_mpctx(L);
|
|
|
|
int res_x = luaL_checkinteger(L, 1);
|
|
|
|
int res_y = luaL_checkinteger(L, 2);
|
|
|
|
const char *text = luaL_checkstring(L, 3);
|
2014-01-18 01:19:20 +01:00
|
|
|
osd_set_external(mpctx->osd, res_x, res_y, (char *)text);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int script_get_osd_resolution(lua_State *L)
|
|
|
|
{
|
|
|
|
struct MPContext *mpctx = get_mpctx(L);
|
|
|
|
int w, h;
|
2014-01-18 01:19:20 +01:00
|
|
|
osd_object_get_resolution(mpctx->osd, OSDTYPE_EXTERNAL, &w, &h);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
lua_pushnumber(L, w);
|
|
|
|
lua_pushnumber(L, h);
|
|
|
|
return 2;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int script_get_screen_size(lua_State *L)
|
|
|
|
{
|
|
|
|
struct MPContext *mpctx = get_mpctx(L);
|
2014-01-18 01:19:20 +01:00
|
|
|
struct mp_osd_res vo_res = osd_get_vo_res(mpctx->osd, OSDTYPE_EXTERNAL);
|
|
|
|
double aspect = 1.0 * vo_res.w / MPMAX(vo_res.h, 1) /
|
|
|
|
vo_res.display_par;
|
|
|
|
lua_pushnumber(L, vo_res.w);
|
|
|
|
lua_pushnumber(L, vo_res.h);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
lua_pushnumber(L, aspect);
|
|
|
|
return 3;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int script_get_mouse_pos(lua_State *L)
|
|
|
|
{
|
|
|
|
struct MPContext *mpctx = get_mpctx(L);
|
2013-09-27 15:39:28 +02:00
|
|
|
int px, py;
|
|
|
|
mp_input_get_mouse_pos(mpctx->input, &px, &py);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
double sw, sh;
|
2014-01-18 01:19:20 +01:00
|
|
|
osd_object_get_scale_factor(mpctx->osd, OSDTYPE_EXTERNAL, &sw, &sh);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
lua_pushnumber(L, px * sw);
|
|
|
|
lua_pushnumber(L, py * sh);
|
|
|
|
return 2;
|
|
|
|
}
|
|
|
|
|
2014-02-10 22:42:20 +01:00
|
|
|
static int script_get_time(lua_State *L)
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
{
|
|
|
|
lua_pushnumber(L, mp_time_sec());
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int script_input_define_section(lua_State *L)
|
|
|
|
{
|
|
|
|
struct MPContext *mpctx = get_mpctx(L);
|
|
|
|
char *section = (char *)luaL_checkstring(L, 1);
|
|
|
|
char *contents = (char *)luaL_checkstring(L, 2);
|
2014-02-17 02:38:07 +01:00
|
|
|
char *flags = (char *)luaL_optstring(L, 3, "");
|
|
|
|
bool builtin = true;
|
|
|
|
if (strcmp(flags, "builtin") == 0) {
|
|
|
|
builtin = true;
|
|
|
|
} else if (strcmp(flags, "default") == 0) {
|
|
|
|
builtin = false;
|
|
|
|
} else if (strcmp(flags, "") == 0) {
|
|
|
|
//pass
|
|
|
|
} else {
|
|
|
|
luaL_error(L, "invalid flags: '%*'", flags);
|
|
|
|
}
|
|
|
|
mp_input_define_section(mpctx->input, section, "<script>", contents, builtin);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int script_input_enable_section(lua_State *L)
|
|
|
|
{
|
|
|
|
struct MPContext *mpctx = get_mpctx(L);
|
|
|
|
char *section = (char *)luaL_checkstring(L, 1);
|
|
|
|
char *sflags = (char *)luaL_optstring(L, 2, "");
|
|
|
|
bstr bflags = bstr0(sflags);
|
|
|
|
int flags = 0;
|
|
|
|
while (bflags.len) {
|
|
|
|
bstr val;
|
|
|
|
bstr_split_tok(bflags, "|", &val, &bflags);
|
|
|
|
if (bstr_equals0(val, "allow-hide-cursor")) {
|
|
|
|
flags |= MP_INPUT_ALLOW_HIDE_CURSOR;
|
|
|
|
} else if (bstr_equals0(val, "allow-vo-dragging")) {
|
|
|
|
flags |= MP_INPUT_ALLOW_VO_DRAGGING;
|
|
|
|
} else if (bstr_equals0(val, "exclusive")) {
|
|
|
|
flags |= MP_INPUT_EXCLUSIVE;
|
|
|
|
} else {
|
|
|
|
luaL_error(L, "invalid flag: '%.*s'", BSTR_P(val));
|
|
|
|
}
|
|
|
|
}
|
|
|
|
mp_input_enable_section(mpctx->input, section, flags);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int script_input_disable_section(lua_State *L)
|
|
|
|
{
|
|
|
|
struct MPContext *mpctx = get_mpctx(L);
|
|
|
|
char *section = (char *)luaL_checkstring(L, 1);
|
|
|
|
mp_input_disable_section(mpctx->input, section);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int script_input_set_section_mouse_area(lua_State *L)
|
|
|
|
{
|
|
|
|
struct MPContext *mpctx = get_mpctx(L);
|
|
|
|
|
|
|
|
double sw, sh;
|
2014-01-18 01:19:20 +01:00
|
|
|
osd_object_get_scale_factor(mpctx->osd, OSDTYPE_EXTERNAL, &sw, &sh);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
|
|
|
|
char *section = (char *)luaL_checkstring(L, 1);
|
|
|
|
int x0 = luaL_checkinteger(L, 2) / sw;
|
|
|
|
int y0 = luaL_checkinteger(L, 3) / sh;
|
|
|
|
int x1 = luaL_checkinteger(L, 4) / sw;
|
|
|
|
int y1 = luaL_checkinteger(L, 5) / sh;
|
|
|
|
mp_input_set_section_mouse_area(mpctx->input, section, x0, y0, x1, y1);
|
|
|
|
return 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
static int script_format_time(lua_State *L)
|
|
|
|
{
|
|
|
|
double t = luaL_checknumber(L, 1);
|
|
|
|
const char *fmt = luaL_optstring(L, 2, "%H:%M:%S");
|
|
|
|
char *r = mp_format_time_fmt(fmt, t);
|
|
|
|
if (!r)
|
|
|
|
luaL_error(L, "Invalid time format string '%s'", fmt);
|
|
|
|
lua_pushstring(L, r);
|
|
|
|
talloc_free(r);
|
|
|
|
return 1;
|
|
|
|
}
|
|
|
|
|
2014-02-10 22:42:20 +01:00
|
|
|
static int script_get_opt(lua_State *L)
|
2014-01-16 23:06:06 +01:00
|
|
|
{
|
|
|
|
struct MPContext *mpctx = get_mpctx(L);
|
2014-02-10 21:03:59 +01:00
|
|
|
|
|
|
|
mp_dispatch_lock(mpctx->dispatch);
|
|
|
|
|
2014-01-16 23:06:06 +01:00
|
|
|
char **opts = mpctx->opts->lua_opts;
|
|
|
|
const char *name = luaL_checkstring(L, 1);
|
2014-02-10 21:03:59 +01:00
|
|
|
int r = 0;
|
2014-01-16 23:06:06 +01:00
|
|
|
|
|
|
|
for (int n = 0; opts && opts[n] && opts[n + 1]; n++) {
|
|
|
|
if (strcmp(opts[n], name) == 0) {
|
|
|
|
lua_pushstring(L, opts[n + 1]);
|
2014-02-10 21:03:59 +01:00
|
|
|
r = 1;
|
|
|
|
break;
|
2014-01-16 23:06:06 +01:00
|
|
|
}
|
|
|
|
}
|
2014-02-10 21:03:59 +01:00
|
|
|
|
|
|
|
mp_dispatch_unlock(mpctx->dispatch);
|
|
|
|
|
|
|
|
return r;
|
2014-01-16 23:06:06 +01:00
|
|
|
}
|
|
|
|
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
struct fn_entry {
|
|
|
|
const char *name;
|
|
|
|
int (*fn)(lua_State *L);
|
|
|
|
};
|
|
|
|
|
|
|
|
#define FN_ENTRY(name) {#name, script_ ## name}
|
|
|
|
|
|
|
|
static struct fn_entry fn_list[] = {
|
|
|
|
FN_ENTRY(log),
|
2014-02-10 21:03:59 +01:00
|
|
|
FN_ENTRY(suspend),
|
|
|
|
FN_ENTRY(resume),
|
2014-02-11 00:57:40 +01:00
|
|
|
FN_ENTRY(resume_all),
|
2014-02-10 21:03:59 +01:00
|
|
|
FN_ENTRY(wait_event),
|
|
|
|
FN_ENTRY(request_event),
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
FN_ENTRY(find_config_file),
|
2014-02-10 22:42:20 +01:00
|
|
|
FN_ENTRY(command),
|
|
|
|
FN_ENTRY(commandv),
|
2014-02-24 20:47:20 +01:00
|
|
|
FN_ENTRY(get_property_bool),
|
|
|
|
FN_ENTRY(get_property_number),
|
|
|
|
FN_ENTRY(get_property_native),
|
2014-02-11 00:23:10 +01:00
|
|
|
FN_ENTRY(set_property),
|
2014-02-24 20:47:20 +01:00
|
|
|
FN_ENTRY(set_property_bool),
|
|
|
|
FN_ENTRY(set_property_number),
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
FN_ENTRY(property_list),
|
|
|
|
FN_ENTRY(set_osd_ass),
|
|
|
|
FN_ENTRY(get_osd_resolution),
|
|
|
|
FN_ENTRY(get_screen_size),
|
|
|
|
FN_ENTRY(get_mouse_pos),
|
2014-02-10 22:42:20 +01:00
|
|
|
FN_ENTRY(get_time),
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
FN_ENTRY(input_define_section),
|
|
|
|
FN_ENTRY(input_enable_section),
|
|
|
|
FN_ENTRY(input_disable_section),
|
|
|
|
FN_ENTRY(input_set_section_mouse_area),
|
|
|
|
FN_ENTRY(format_time),
|
lua: allow scripts to snoop messages
Adds the following Lua function to enable message events:
mp.enable_messages(size, level)
size is the maximum number of messages the ringbuffer consists of. level
is the minimum log level for a message to be added to the ringbuffer,
and uses the same values as the mp.log() function. (Actually not yet,
but this will be fixed in the following commit.)
The messages will be delivered via the mp_event() in the user script,
using "message" as event name. The event argument is a table with the
following fields:
level: log level of the message (string as in mp.log())
prefix: string identifying the module of origin
text: contents of the message
As of currently, the message text will contain newline characters. A
message can consist of several lines. It is also possible that a
message doesn't end with a newline, and a caller can use multiple
messages to "build" a line. Most messages will contain exactly 1 line
ending with a single newline character, though.
If the message buffer overflows (messages are not read quickly enough),
new messages are lost until the queued up messages are read. At the
point of the overflow, a special overflow message is inserted. It will
have prefix set to "overflow", and the message text is set to "".
Care should be taken not to print any messages from the message event
handler. This would lead to an infinite loop (the event handler would be
called again after returning, because a new message is available). This
includes mp.log() and all mp.msg.* functions. Keep in mind that the Lua
print() function is mapped to mp.msg.info().
2014-01-16 21:34:58 +01:00
|
|
|
FN_ENTRY(enable_messages),
|
2014-02-10 22:42:20 +01:00
|
|
|
FN_ENTRY(get_opt),
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
};
|
|
|
|
|
|
|
|
// On stack: mp table
|
|
|
|
static void add_functions(struct script_ctx *ctx)
|
|
|
|
{
|
|
|
|
lua_State *L = ctx->state;
|
|
|
|
|
|
|
|
for (int n = 0; n < MP_ARRAY_SIZE(fn_list); n++) {
|
|
|
|
lua_pushcfunction(L, fn_list[n].fn);
|
|
|
|
lua_setfield(L, -2, fn_list[n].name);
|
|
|
|
}
|
|
|
|
|
|
|
|
lua_pushinteger(L, 0);
|
2014-02-10 22:42:20 +01:00
|
|
|
lua_pushcclosure(L, script_get_property, 1);
|
|
|
|
lua_setfield(L, -2, "get_property");
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
|
|
|
|
lua_pushinteger(L, 1);
|
2014-02-10 22:42:20 +01:00
|
|
|
lua_pushcclosure(L, script_get_property, 1);
|
|
|
|
lua_setfield(L, -2, "get_property_osd");
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
}
|
|
|
|
|
2014-02-14 13:49:10 +01:00
|
|
|
static int compare_filename(const void *pa, const void *pb)
|
|
|
|
{
|
|
|
|
char *a = (char *)pa;
|
|
|
|
char *b = (char *)pb;
|
|
|
|
return strcmp(a, b);
|
|
|
|
}
|
|
|
|
|
|
|
|
static char **list_lua_files(void *talloc_ctx, char *path)
|
|
|
|
{
|
|
|
|
char **files = NULL;
|
|
|
|
int count = 0;
|
|
|
|
DIR *dp = opendir(path);
|
|
|
|
if (!dp)
|
|
|
|
return NULL;
|
|
|
|
struct dirent *ep;
|
|
|
|
while ((ep = readdir(dp))) {
|
|
|
|
char *ext = mp_splitext(ep->d_name, NULL);
|
|
|
|
if (!ext || strcasecmp(ext, "lua") != 0)
|
|
|
|
continue;
|
|
|
|
char *fname = mp_path_join(talloc_ctx, bstr0(path), bstr0(ep->d_name));
|
|
|
|
MP_TARRAY_APPEND(talloc_ctx, files, count, fname);
|
|
|
|
}
|
|
|
|
closedir(dp);
|
|
|
|
qsort(files, count, sizeof(char *), compare_filename);
|
|
|
|
MP_TARRAY_APPEND(talloc_ctx, files, count, NULL);
|
|
|
|
return files;
|
|
|
|
}
|
|
|
|
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
void mp_lua_init(struct MPContext *mpctx)
|
|
|
|
{
|
|
|
|
// Load scripts from options
|
2013-09-26 00:56:41 +02:00
|
|
|
if (mpctx->opts->lua_load_osc)
|
|
|
|
mp_lua_load_script(mpctx, "@osc");
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
char **files = mpctx->opts->lua_files;
|
|
|
|
for (int n = 0; files && files[n]; n++) {
|
|
|
|
if (files[n][0])
|
|
|
|
mp_lua_load_script(mpctx, files[n]);
|
|
|
|
}
|
2014-02-14 13:49:10 +01:00
|
|
|
// Load ~/.mpv/lua/*
|
|
|
|
void *tmp = talloc_new(NULL);
|
|
|
|
char *lua_path = mp_find_user_config_file(tmp, mpctx->global, "lua");
|
|
|
|
if (lua_path) {
|
|
|
|
files = list_lua_files(tmp, lua_path);
|
|
|
|
for (int n = 0; files && files[n]; n++)
|
|
|
|
mp_lua_load_script(mpctx, files[n]);
|
|
|
|
}
|
|
|
|
talloc_free(tmp);
|
Add initial Lua scripting support
This is preliminary. There are still tons of issues, and any aspect
of scripting may change in the future. I decided to merge this
(preliminary) work now because it makes it easier to develop it, not
because it's done. lua.rst is clear enough about it (plus some
sarcasm).
This requires linking to Lua. Lua has no official pkg-config file, but
there are distribution specific .pc files, all with different names.
Adding a non-pkg-config based configure test was considered, but we'd
rather not.
One major complication is that libquvi links against Lua too, and if
the Lua version is different from mpv's, you will get a crash as soon
as libquvi uses Lua. (libquvi by design always runs when a file is
opened.) I would consider this the problem of distros and whoever
builds mpv, but to make things easier for users, we add a terrible
runtime test to the configure script, which probes whether libquvi
will crash. This is disabled when cross-compiling, but in that case
we hope the user knows what he is doing.
2013-09-26 00:41:14 +02:00
|
|
|
}
|