/* MI Command Set - MI parser.
Copyright (C) 2000-2022 Free Software Foundation, Inc.
Contributed by Cygnus Solutions (a Red Hat company).
This file is part of GDB.
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General Public License as published by
the Free Software Foundation; either version 3 of the License, or
(at your option) any later version.
This program is distributed in the hope that it will be useful,
but WITHOUT ANY WARRANTY; without even the implied warranty of
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
GNU General Public License for more details.
You should have received a copy of the GNU General Public License
along with this program. If not, see . */
#include "defs.h"
#include "mi-cmds.h"
#include "mi-parse.h"
#include "charset.h"
#include
#include "cli/cli-utils.h"
#include "language.h"
static const char mi_no_values[] = "--no-values";
static const char mi_simple_values[] = "--simple-values";
static const char mi_all_values[] = "--all-values";
/* Like parse_escape, but leave the results as a host char, not a
target char. */
static int
mi_parse_escape (const char **string_ptr)
{
int c = *(*string_ptr)++;
switch (c)
{
case '\n':
return -2;
case 0:
(*string_ptr)--;
return 0;
case '0':
case '1':
case '2':
case '3':
case '4':
case '5':
case '6':
case '7':
{
int i = fromhex (c);
int count = 0;
while (++count < 3)
{
c = (**string_ptr);
if (isdigit (c) && c != '8' && c != '9')
{
(*string_ptr)++;
i *= 8;
i += fromhex (c);
}
else
{
break;
}
}
return i;
}
case 'a':
c = '\a';
break;
case 'b':
c = '\b';
break;
case 'f':
c = '\f';
break;
case 'n':
c = '\n';
break;
case 'r':
c = '\r';
break;
case 't':
c = '\t';
break;
case 'v':
c = '\v';
break;
default:
break;
}
return c;
}
void
mi_parse_argv (const char *args, struct mi_parse *parse)
{
const char *chp = args;
int argc = 0;
char **argv = XNEWVEC (char *, argc + 1);
argv[argc] = NULL;
while (1)
{
char *arg;
/* Skip leading white space. */
chp = skip_spaces (chp);
/* Three possibilities: EOF, quoted string, or other text. */
switch (*chp)
{
case '\0':
parse->argv = argv;
parse->argc = argc;
return;
case '"':
{
/* A quoted string. */
int len;
const char *start = chp + 1;
/* Determine the buffer size. */
chp = start;
len = 0;
while (*chp != '\0' && *chp != '"')
{
if (*chp == '\\')
{
chp++;
if (mi_parse_escape (&chp) <= 0)
{
/* Do not allow split lines or "\000". */
freeargv (argv);
return;
}
}
else
chp++;
len++;
}
/* Insist on a closing quote. */
if (*chp != '"')
{
freeargv (argv);
return;
}
/* Insist on trailing white space. */
if (chp[1] != '\0' && !isspace (chp[1]))
{
freeargv (argv);
return;
}
/* Create the buffer and copy characters in. */
arg = XNEWVEC (char, len + 1);
chp = start;
len = 0;
while (*chp != '\0' && *chp != '"')
{
if (*chp == '\\')
{
chp++;
arg[len] = mi_parse_escape (&chp);
}
else
arg[len] = *chp++;
len++;
}
arg[len] = '\0';
chp++; /* That closing quote. */
break;
}
default:
{
/* An unquoted string. Accumulate all non-blank
characters into a buffer. */
int len;
const char *start = chp;
while (*chp != '\0' && !isspace (*chp))
{
chp++;
}
len = chp - start;
arg = XNEWVEC (char, len + 1);
strncpy (arg, start, len);
arg[len] = '\0';
break;
}
}
/* Append arg to argv. */
argv = XRESIZEVEC (char *, argv, argc + 2);
argv[argc++] = arg;
argv[argc] = NULL;
}
}
mi_parse::mi_parse ()
: op (MI_COMMAND),
command (NULL),
token (NULL),
cmd (NULL),
cmd_start (NULL),
args (NULL),
argv (NULL),
argc (0),
all (0),
thread_group (-1),
thread (-1),
frame (-1),
language (language_unknown)
{
}
mi_parse::~mi_parse ()
{
xfree (command);
xfree (token);
xfree (args);
freeargv (argv);
}
std::unique_ptr
mi_parse (const char *cmd, char **token)
{
const char *chp;
std::unique_ptr parse (new struct mi_parse);
/* Before starting, skip leading white space. */
cmd = skip_spaces (cmd);
/* Find/skip any token and then extract it. */
for (chp = cmd; *chp >= '0' && *chp <= '9'; chp++)
;
*token = (char *) xmalloc (chp - cmd + 1);
memcpy (*token, cmd, (chp - cmd));
(*token)[chp - cmd] = '\0';
/* This wasn't a real MI command. Return it as a CLI_COMMAND. */
if (*chp != '-')
{
chp = skip_spaces (chp);
parse->command = xstrdup (chp);
parse->op = CLI_COMMAND;
return parse;
}
/* Extract the command. */
{
const char *tmp = chp + 1; /* discard ``-'' */
for (; *chp && !isspace (*chp); chp++)
;
parse->command = (char *) xmalloc (chp - tmp + 1);
memcpy (parse->command, tmp, chp - tmp);
parse->command[chp - tmp] = '\0';
}
/* Find the command in the MI table. */
parse->cmd = mi_cmd_lookup (parse->command);
if (parse->cmd == NULL)
throw_error (UNDEFINED_COMMAND_ERROR,
_("Undefined MI command: %s"), parse->command);
/* Skip white space following the command. */
chp = skip_spaces (chp);
/* Parse the --thread and --frame options, if present. At present,
some important commands, like '-break-*' are implemented by
forwarding to the CLI layer directly. We want to parse --thread
and --frame here, so as not to leave those option in the string
that will be passed to CLI.
Same for the --language option. */
for (;;)
{
const char *option;
size_t as = sizeof ("--all ") - 1;
size_t tgs = sizeof ("--thread-group ") - 1;
size_t ts = sizeof ("--thread ") - 1;
size_t fs = sizeof ("--frame ") - 1;
size_t ls = sizeof ("--language ") - 1;
if (strncmp (chp, "--all ", as) == 0)
{
parse->all = 1;
chp += as;
}
/* See if --all is the last token in the input. */
if (strcmp (chp, "--all") == 0)
{
parse->all = 1;
chp += strlen (chp);
}
if (strncmp (chp, "--thread-group ", tgs) == 0)
{
char *endp;
option = "--thread-group";
if (parse->thread_group != -1)
error (_("Duplicate '--thread-group' option"));
chp += tgs;
if (*chp != 'i')
error (_("Invalid thread group id"));
chp += 1;
parse->thread_group = strtol (chp, &endp, 10);
chp = endp;
}
else if (strncmp (chp, "--thread ", ts) == 0)
{
char *endp;
option = "--thread";
if (parse->thread != -1)
error (_("Duplicate '--thread' option"));
chp += ts;
parse->thread = strtol (chp, &endp, 10);
chp = endp;
}
else if (strncmp (chp, "--frame ", fs) == 0)
{
char *endp;
option = "--frame";
if (parse->frame != -1)
error (_("Duplicate '--frame' option"));
chp += fs;
parse->frame = strtol (chp, &endp, 10);
chp = endp;
}
else if (strncmp (chp, "--language ", ls) == 0)
{
option = "--language";
chp += ls;
std::string lang_name = extract_arg (&chp);
parse->language = language_enum (lang_name.c_str ());
if (parse->language == language_unknown
|| parse->language == language_auto)
error (_("Invalid --language argument: %s"), lang_name.c_str ());
}
else
break;
if (*chp != '\0' && !isspace (*chp))
error (_("Invalid value for the '%s' option"), option);
chp = skip_spaces (chp);
}
/* Save the rest of the arguments for the command. */
parse->args = xstrdup (chp);
/* Fully parsed, flag as an MI command. */
parse->op = MI_COMMAND;
return parse;
}
enum print_values
mi_parse_print_values (const char *name)
{
if (strcmp (name, "0") == 0
|| strcmp (name, mi_no_values) == 0)
return PRINT_NO_VALUES;
else if (strcmp (name, "1") == 0
|| strcmp (name, mi_all_values) == 0)
return PRINT_ALL_VALUES;
else if (strcmp (name, "2") == 0
|| strcmp (name, mi_simple_values) == 0)
return PRINT_SIMPLE_VALUES;
else
error (_("Unknown value for PRINT_VALUES: must be: \
0 or \"%s\", 1 or \"%s\", 2 or \"%s\""),
mi_no_values, mi_all_values, mi_simple_values);
}