hush/coreutils/expr.c

507 lines
9.4 KiB
C
Raw Normal View History

/* vi: set sw=4 ts=4: */
/*
* Mini expr implementation for busybox
*
* based on GNU expr Mike Parker.
* Copyright (C) 86, 1991-1997, 1999 Free Software Foundation, Inc.
*
* Busybox modifications
* Copyright (c) 2000 Edward Betts <edward@debian.org>.
2006-01-25 13:56:03 +00:00
* Copyright (C) 2003-2005 Vladimir Oleynik <dzo@simtreas.ru>
* - reduced 464 bytes.
* - 64 math support
*
* Licensed under GPLv2 or later, see file LICENSE in this tarball for details.
*/
/* This program evaluates expressions. Each token (operator, operand,
* parenthesis) of the expression must be a separate argument. The
* parser used is a reasonably general one, though any incarnation of
* it is language-specific. It is especially nice for expressions.
*
* No parse tree is needed; a new node is evaluated immediately.
* One function can handle multiple operators all of equal precedence,
* provided they all associate ((x op x) op x). */
/* no getopt needed */
#include "busybox.h"
#include "xregex.h"
/* The kinds of value we can have. */
enum valtype {
integer,
string
};
typedef enum valtype TYPE;
2006-01-25 13:56:03 +00:00
#if ENABLE_EXPR_MATH_SUPPORT_64
typedef int64_t arith_t;
2006-01-25 13:56:03 +00:00
#define PF_REZ "ll"
#define PF_REZ_TYPE (long long)
2006-01-25 13:56:03 +00:00
#define STRTOL(s, e, b) strtoll(s, e, b)
#else
typedef long arith_t;
2006-01-25 13:56:03 +00:00
#define PF_REZ "l"
#define PF_REZ_TYPE (long)
2006-01-25 13:56:03 +00:00
#define STRTOL(s, e, b) strtol(s, e, b)
#endif
/* TODO: use bb_strtol[l]? It's easier to check for errors... */
/* A value is.... */
struct valinfo {
TYPE type; /* Which kind. */
union { /* The value itself. */
2006-01-25 13:56:03 +00:00
arith_t i;
char *s;
} u;
};
typedef struct valinfo VALUE;
/* The arguments given to the program, minus the program name. */
static char **args;
static VALUE *docolon(VALUE * sv, VALUE * pv);
static VALUE *eval(void);
static VALUE *int_value(arith_t i);
static VALUE *str_value(const char *s);
static int nextarg(const char *str);
static int null(VALUE * v);
static int toarith(VALUE * v);
static void freev(VALUE * v);
static void tostring(VALUE * v);
int expr_main(int argc, char **argv);
int expr_main(int argc, char **argv)
{
VALUE *v;
if (argc == 1) {
2003-03-19 09:13:01 +00:00
bb_error_msg_and_die("too few arguments");
}
args = argv + 1;
v = eval();
if (*args)
bb_error_msg_and_die("syntax error");
if (v->type == integer)
2006-10-26 23:21:47 +00:00
printf("%" PF_REZ "d\n", PF_REZ_TYPE v->u.i);
else
puts(v->u.s);
2006-10-26 23:21:47 +00:00
fflush_stdout_and_exit(null(v));
}
/* Return a VALUE for I. */
static VALUE *int_value(arith_t i)
{
VALUE *v;
v = xmalloc(sizeof(VALUE));
v->type = integer;
v->u.i = i;
return v;
}
/* Return a VALUE for S. */
static VALUE *str_value(const char *s)
{
VALUE *v;
v = xmalloc(sizeof(VALUE));
v->type = string;
v->u.s = xstrdup(s);
return v;
}
/* Free VALUE V, including structure components. */
static void freev(VALUE * v)
{
if (v->type == string)
free(v->u.s);
free(v);
}
/* Return nonzero if V is a null-string or zero-number. */
static int null(VALUE * v)
{
if (v->type == integer)
return v->u.i == 0;
else /* string: */
return v->u.s[0] == '\0' || LONE_CHAR(v->u.s, '0');
}
/* Coerce V to a string value (can't fail). */
static void tostring(VALUE * v)
{
if (v->type == integer) {
v->u.s = xasprintf("%" PF_REZ "d", PF_REZ_TYPE v->u.i);
v->type = string;
}
}
/* Coerce V to an integer value. Return 1 on success, 0 on failure. */
static int toarith(VALUE * v)
{
if (v->type == string) {
2006-01-25 13:56:03 +00:00
arith_t i;
char *e;
/* Don't interpret the empty string as an integer. */
/* Currently does not worry about overflow or int/long differences. */
2006-01-25 13:56:03 +00:00
i = STRTOL(v->u.s, &e, 10);
if ((v->u.s == e) || *e)
return 0;
free(v->u.s);
v->u.i = i;
v->type = integer;
}
2003-08-29 07:29:30 +00:00
return 1;
}
/* Return nonzero if the next token matches STR exactly.
STR must not be NULL. */
static int nextarg(const char *str)
{
if (*args == NULL)
return 0;
return strcmp(*args, str) == 0;
}
/* The comparison operator handling functions. */
static int cmp_common(VALUE * l, VALUE * r, int op)
2003-08-29 07:29:30 +00:00
{
int cmpval;
2003-08-29 07:29:30 +00:00
if (l->type == string || r->type == string) {
tostring(l);
tostring(r);
cmpval = strcmp(l->u.s, r->u.s);
} else
2003-08-29 07:29:30 +00:00
cmpval = l->u.i - r->u.i;
if (op == '<')
return cmpval < 0;
else if (op == ('L' + 'E'))
return cmpval <= 0;
else if (op == '=')
return cmpval == 0;
else if (op == '!')
return cmpval != 0;
else if (op == '>')
return cmpval > 0;
else /* >= */
return cmpval >= 0;
2003-08-29 07:29:30 +00:00
}
/* The arithmetic operator handling functions. */
static arith_t arithmetic_common(VALUE * l, VALUE * r, int op)
2003-08-29 07:29:30 +00:00
{
arith_t li, ri;
if (!toarith(l) || !toarith(r))
bb_error_msg_and_die("non-numeric argument");
li = l->u.i;
ri = r->u.i;
if ((op == '/' || op == '%') && ri == 0)
bb_error_msg_and_die("division by zero");
if (op == '+')
2003-08-29 07:29:30 +00:00
return li + ri;
else if (op == '-')
2003-08-29 07:29:30 +00:00
return li - ri;
else if (op == '*')
2003-08-29 07:29:30 +00:00
return li * ri;
else if (op == '/')
2003-08-29 07:29:30 +00:00
return li / ri;
else
2003-08-29 07:29:30 +00:00
return li % ri;
}
/* Do the : operator.
SV is the VALUE for the lhs (the string),
PV is the VALUE for the rhs (the pattern). */
static VALUE *docolon(VALUE * sv, VALUE * pv)
{
VALUE *v;
regex_t re_buffer;
const int NMATCH = 2;
regmatch_t re_regs[NMATCH];
tostring(sv);
tostring(pv);
if (pv->u.s[0] == '^') {
fprintf(stderr, "\
warning: unportable BRE: `%s': using `^' as the first character\n\
of a basic regular expression is not portable; it is being ignored", pv->u.s);
}
memset(&re_buffer, 0, sizeof(re_buffer));
memset(re_regs, 0, sizeof(*re_regs));
if (regcomp(&re_buffer, pv->u.s, 0) != 0)
bb_error_msg_and_die("invalid regular expression");
/* expr uses an anchored pattern match, so check that there was a
* match and that the match starts at offset 0. */
if (regexec(&re_buffer, sv->u.s, NMATCH, re_regs, 0) != REG_NOMATCH &&
re_regs[0].rm_so == 0) {
/* Were \(...\) used? */
if (re_buffer.re_nsub > 0) {
sv->u.s[re_regs[1].rm_eo] = '\0';
v = str_value(sv->u.s + re_regs[1].rm_so);
} else
v = int_value(re_regs[0].rm_eo);
} else {
/* Match failed -- return the right kind of null. */
if (re_buffer.re_nsub > 0)
v = str_value("");
else
v = int_value(0);
}
return v;
}
/* Handle bare operands and ( expr ) syntax. */
static VALUE *eval7(void)
{
VALUE *v;
if (!*args)
bb_error_msg_and_die("syntax error");
if (nextarg("(")) {
args++;
v = eval();
if (!nextarg(")"))
bb_error_msg_and_die("syntax error");
args++;
return v;
}
if (nextarg(")"))
bb_error_msg_and_die("syntax error");
return str_value(*args++);
}
/* Handle match, substr, index, length, and quote keywords. */
static VALUE *eval6(void)
{
VALUE *l, *r, *v, *i1, *i2;
if (nextarg("quote")) {
args++;
if (!*args)
bb_error_msg_and_die("syntax error");
return str_value(*args++);
} else if (nextarg("length")) {
args++;
r = eval6();
tostring(r);
v = int_value(strlen(r->u.s));
freev(r);
return v;
} else if (nextarg("match")) {
args++;
l = eval6();
r = eval6();
v = docolon(l, r);
freev(l);
freev(r);
return v;
} else if (nextarg("index")) {
args++;
l = eval6();
r = eval6();
tostring(l);
tostring(r);
v = int_value(strcspn(l->u.s, r->u.s) + 1);
if (v->u.i == (arith_t) strlen(l->u.s) + 1)
v->u.i = 0;
freev(l);
freev(r);
return v;
} else if (nextarg("substr")) {
args++;
l = eval6();
i1 = eval6();
i2 = eval6();
tostring(l);
if (!toarith(i1) || !toarith(i2)
|| i1->u.i > (arith_t) strlen(l->u.s)
|| i1->u.i <= 0 || i2->u.i <= 0)
v = str_value("");
else {
v = xmalloc(sizeof(VALUE));
v->type = string;
v->u.s = xstrndup(l->u.s + i1->u.i - 1, i2->u.i);
}
freev(l);
freev(i1);
freev(i2);
return v;
} else
return eval7();
}
/* Handle : operator (pattern matching).
Calls docolon to do the real work. */
static VALUE *eval5(void)
{
VALUE *l, *r, *v;
l = eval6();
while (nextarg(":")) {
args++;
r = eval6();
v = docolon(l, r);
freev(l);
freev(r);
l = v;
}
return l;
}
/* Handle *, /, % operators. */
static VALUE *eval4(void)
{
VALUE *l, *r;
2006-01-25 13:56:03 +00:00
int op;
arith_t val;
l = eval5();
while (1) {
if (nextarg("*"))
2003-08-29 07:29:30 +00:00
op = '*';
else if (nextarg("/"))
2003-08-29 07:29:30 +00:00
op = '/';
else if (nextarg("%"))
2003-08-29 07:29:30 +00:00
op = '%';
else
return l;
args++;
r = eval5();
val = arithmetic_common(l, r, op);
freev(l);
freev(r);
l = int_value(val);
}
}
/* Handle +, - operators. */
static VALUE *eval3(void)
{
VALUE *l, *r;
2006-01-25 13:56:03 +00:00
int op;
arith_t val;
l = eval4();
while (1) {
if (nextarg("+"))
2003-08-29 07:29:30 +00:00
op = '+';
else if (nextarg("-"))
op = '-';
else
return l;
args++;
r = eval4();
val = arithmetic_common(l, r, op);
freev(l);
freev(r);
l = int_value(val);
}
}
/* Handle comparisons. */
static VALUE *eval2(void)
{
VALUE *l, *r;
2006-01-25 13:56:03 +00:00
int op;
arith_t val;
l = eval3();
while (1) {
if (nextarg("<"))
2003-08-29 07:29:30 +00:00
op = '<';
else if (nextarg("<="))
op = 'L' + 'E';
else if (nextarg("=") || nextarg("=="))
2003-08-29 07:29:30 +00:00
op = '=';
else if (nextarg("!="))
2003-08-29 07:29:30 +00:00
op = '!';
else if (nextarg(">="))
op = 'G' + 'E';
else if (nextarg(">"))
2003-08-29 07:29:30 +00:00
op = '>';
else
return l;
args++;
r = eval3();
toarith(l);
toarith(r);
val = cmp_common(l, r, op);
freev(l);
freev(r);
l = int_value(val);
}
}
/* Handle &. */
static VALUE *eval1(void)
{
VALUE *l, *r;
l = eval2();
while (nextarg("&")) {
args++;
r = eval2();
if (null(l) || null(r)) {
freev(l);
freev(r);
l = int_value(0);
} else
freev(r);
}
return l;
}
/* Handle |. */
static VALUE *eval(void)
{
VALUE *l, *r;
l = eval1();
while (nextarg("|")) {
args++;
r = eval1();
if (null(l)) {
freev(l);
l = r;
} else
freev(r);
}
return l;
}