mirror of https://github.com/flysand7/ciabatta.git
117 lines
3.8 KiB
C
117 lines
3.8 KiB
C
|
|
#define CMDLINE_CMD_MAX 32767
|
|
#define CMDLINE_ARGV_MAX (16384+(98298+(int)sizeof(char*))/(int)sizeof(char*))
|
|
|
|
// https://github.com/skeeto/scratch/blob/master/misc/cmdline.c#L27
|
|
static int cmdline_to_argv8(const wchar_t *cmd, char **argv) {
|
|
int argc = 1; // worst case: argv[0] is an empty string
|
|
int state = 6; // special argv[0] state
|
|
int slash = 0;
|
|
char *buf = (char *)(argv + 16384); // second half: byte buffer
|
|
|
|
argv[0] = buf;
|
|
while (*cmd) {
|
|
int c = *cmd++;
|
|
if (c>>10 == 0x36 && *cmd>>10 == 0x37) { // surrogates?
|
|
c = 0x10000 + ((c - 0xd800)<<10) + (*cmd++ - 0xdc00);
|
|
}
|
|
|
|
switch (state) {
|
|
case 0: switch (c) { // outside token
|
|
case 0x09:
|
|
case 0x20: continue;
|
|
case 0x22: argv[argc++] = buf;
|
|
state = 2;
|
|
continue;
|
|
case 0x5c: argv[argc++] = buf;
|
|
slash = 1;
|
|
state = 3;
|
|
break;
|
|
default : argv[argc++] = buf;
|
|
state = 1;
|
|
} break;
|
|
case 1: switch (c) { // inside unquoted token
|
|
case 0x09:
|
|
case 0x20: *buf++ = 0;
|
|
state = 0;
|
|
continue;
|
|
case 0x22: state = 2;
|
|
continue;
|
|
case 0x5c: slash = 1;
|
|
state = 3;
|
|
break;
|
|
} break;
|
|
case 2: switch (c) { // inside quoted token
|
|
case 0x22: state = 5;
|
|
continue;
|
|
case 0x5c: slash = 1;
|
|
state = 4;
|
|
break;
|
|
} break;
|
|
case 3:
|
|
case 4: switch (c) { // backslash sequence
|
|
case 0x22: buf -= (1 + slash) >> 1;
|
|
if (slash & 1) {
|
|
state -= 2;
|
|
break;
|
|
} // fallthrough
|
|
default : cmd--;
|
|
state -= 2;
|
|
continue;
|
|
case 0x5c: slash++;
|
|
} break;
|
|
case 5: switch (c) { // quoted token exit
|
|
default : cmd--;
|
|
state = 1;
|
|
continue;
|
|
case 0x22: state = 1;
|
|
} break;
|
|
case 6: switch (c) { // begin argv[0]
|
|
case 0x09:
|
|
case 0x20: *buf++ = 0;
|
|
state = 0;
|
|
continue;
|
|
case 0x22: state = 8;
|
|
continue;
|
|
default : state = 7;
|
|
} break;
|
|
case 7: switch (c) { // unquoted argv[0]
|
|
case 0x09:
|
|
case 0x20: *buf++ = 0;
|
|
state = 0;
|
|
continue;
|
|
} break;
|
|
case 8: switch (c) { // quoted argv[0]
|
|
case 0x22: *buf++ = 0;
|
|
state = 0;
|
|
continue;
|
|
} break;
|
|
}
|
|
|
|
switch (c & 0x1f0880) { // WTF-8/UTF-8 encoding
|
|
case 0x00000: *buf++ = 0x00 | ((c >> 0) ); break;
|
|
case 0x00080: *buf++ = 0xc0 | ((c >> 6) );
|
|
*buf++ = 0x80 | ((c >> 0) & 63); break;
|
|
case 0x00800:
|
|
case 0x00880: *buf++ = 0xe0 | ((c >> 12) );
|
|
*buf++ = 0x80 | ((c >> 6) & 63);
|
|
*buf++ = 0x80 | ((c >> 0) & 63); break;
|
|
default : *buf++ = 0xf0 | ((c >> 18) );
|
|
*buf++ = 0x80 | ((c >> 12) & 63);
|
|
*buf++ = 0x80 | ((c >> 6) & 63);
|
|
*buf++ = 0x80 | ((c >> 0) & 63);
|
|
}
|
|
}
|
|
|
|
*buf = 0;
|
|
argv[argc] = 0;
|
|
return argc;
|
|
}
|
|
|
|
static char **get_command_args(int *argc_ptr) {
|
|
static char *argv_buffer[CMDLINE_ARGV_MAX];
|
|
wchar_t *cmdline = GetCommandLineW();
|
|
*argc_ptr = cmdline_to_argv8(cmdline, argv_buffer);
|
|
return argv_buffer;
|
|
}
|