interp_parse.c revision 64187
138716Smsmith/*
238716Smsmith * Redistribution and use in source and binary forms, with or without
338716Smsmith * modification, are permitted provided that the following conditions
438716Smsmith * are met:
538716Smsmith * 1. Redistributions of source code must retain the above copyright
638716Smsmith *    notice, this list of conditions and the following disclaimer.
738716Smsmith * 2. Redistributions in binary form must reproduce the above copyright
838716Smsmith *    notice, this list of conditions and the following disclaimer in the
938716Smsmith *    documentation and/or other materials provided with the distribution.
1038716Smsmith *
1138716Smsmith * Jordan K. Hubbard
1238716Smsmith * 29 August 1998
1338716Smsmith *
1450477Speter * $FreeBSD: head/sys/boot/common/interp_parse.c 64187 2000-08-03 09:14:02Z jhb $
1538716Smsmith *
1638716Smsmith * The meat of the simple parser.
1738716Smsmith */
1838716Smsmith
1938716Smsmith#include <stand.h>
2038716Smsmith#include <string.h>
2164187Sjhb#include "bootstrap.h"
2238716Smsmith
2364187Sjhbstatic void	 clean(void);
2464187Sjhbstatic int	 insert(int *argcp, char *buf);
2564187Sjhbstatic char	*variable_lookup(char *name);
2638716Smsmith
2738716Smsmith#define PARSE_BUFSIZE	1024	/* maximum size of one element */
2838716Smsmith#define MAXARGS		20	/* maximum number of elements */
2938716Smsmithstatic char		*args[MAXARGS];
3038716Smsmith
3138716Smsmith/*
3238716Smsmith * parse: accept a string of input and "parse" it for backslash
3338716Smsmith * substitutions and environment variable expansions (${var}),
3438716Smsmith * returning an argc/argv style vector of whitespace separated
3538716Smsmith * arguments.  Returns 0 on success, 1 on failure (ok, ok, so I
3638716Smsmith * wimped-out on the error codes! :).
3738716Smsmith *
3838716Smsmith * Note that the argv array returned must be freed by the caller, but
3938716Smsmith * we own the space allocated for arguments and will free that on next
4038716Smsmith * invocation.  This allows argv consumers to modify the array if
4138716Smsmith * required.
4238716Smsmith *
4338716Smsmith * NB: environment variables that expand to more than one whitespace
4438716Smsmith * separated token will be returned as a single argv[] element, not
4538716Smsmith * split in turn.  Expanded text is also immune to further backslash
4638716Smsmith * elimination or expansion since this is a one-pass, non-recursive
4738716Smsmith * parser.  You didn't specify more than this so if you want more, ask
4838716Smsmith * me. - jkh
4938716Smsmith */
5038716Smsmith
5138716Smsmith#define PARSE_FAIL(expr) \
5238716Smsmithif (expr) { \
5338716Smsmith    printf("fail at line %d\n", __LINE__); \
5438716Smsmith    clean(); \
5538716Smsmith    free(copy); \
5638716Smsmith    free(buf); \
5738716Smsmith    return 1; \
5838716Smsmith}
5938716Smsmith
6038716Smsmith/* Accept the usual delimiters for a variable, returning counterpart */
6138716Smsmithstatic char
6264187Sjhbisdelim(int ch)
6338716Smsmith{
6438716Smsmith    if (ch == '{')
6538716Smsmith	return '}';
6638716Smsmith    else if (ch == '(')
6738716Smsmith	return ')';
6838716Smsmith    return '\0';
6938716Smsmith}
7038716Smsmith
7138765Sjkhstatic int
7264187Sjhbisquote(int ch)
7338765Sjkh{
7438765Sjkh    return (ch == '\'' || ch == '"');
7538765Sjkh}
7638765Sjkh
7738716Smsmithint
7838716Smsmithparse(int *argc, char ***argv, char *str)
7938716Smsmith{
8038716Smsmith    int ac;
8138716Smsmith    char *val, *p, *q, *copy = NULL;
8264187Sjhb    size_t i = 0;
8338765Sjkh    char token, tmp, quote, *buf;
8438716Smsmith    enum { STR, VAR, WHITE } state;
8538716Smsmith
8638716Smsmith    ac = *argc = 0;
8738765Sjkh    quote = 0;
8838716Smsmith    if (!str || (p = copy = backslash(str)) == NULL)
8938716Smsmith	return 1;
9038716Smsmith
9138716Smsmith    /* Initialize vector and state */
9238716Smsmith    clean();
9338716Smsmith    state = STR;
9438765Sjkh    buf = (char *)malloc(PARSE_BUFSIZE);
9538716Smsmith    token = 0;
9638716Smsmith
9738716Smsmith    /* And awaaaaaaaaay we go! */
9838716Smsmith    while (*p) {
9938716Smsmith	switch (state) {
10038716Smsmith	case STR:
10142465Smsmith	    if ((*p == '\\') && p[1]) {
10242465Smsmith		p++;
10342465Smsmith		PARSE_FAIL(i == (PARSE_BUFSIZE - 1));
10442620Smsmith		buf[i++] = *p++;
10542465Smsmith	    } else if (isquote(*p)) {
10638765Sjkh		quote = quote ? 0 : *p;
10738765Sjkh		++p;
10838765Sjkh	    }
10938765Sjkh	    else if (isspace(*p) && !quote) {
11038716Smsmith		state = WHITE;
11138716Smsmith		if (i) {
11238716Smsmith		    buf[i] = '\0';
11338716Smsmith		    PARSE_FAIL(insert(&ac, buf));
11438716Smsmith		    i = 0;
11538716Smsmith		}
11638716Smsmith		++p;
11738716Smsmith	    } else if (*p == '$') {
11838716Smsmith		token = isdelim(*(p + 1));
11938716Smsmith		if (token)
12038716Smsmith		    p += 2;
12138716Smsmith		else
12238716Smsmith		    ++p;
12338716Smsmith		state = VAR;
12438716Smsmith	    } else {
12538716Smsmith		PARSE_FAIL(i == (PARSE_BUFSIZE - 1));
12638716Smsmith		buf[i++] = *p++;
12738716Smsmith	    }
12838716Smsmith	    break;
12938716Smsmith
13038716Smsmith	case WHITE:
13138716Smsmith	    if (isspace(*p))
13238716Smsmith		++p;
13338716Smsmith	    else
13438716Smsmith		state = STR;
13538716Smsmith	    break;
13638716Smsmith
13738716Smsmith	case VAR:
13838716Smsmith	    if (token) {
13938716Smsmith		PARSE_FAIL((q = index(p, token)) == NULL);
14038716Smsmith	    } else {
14138716Smsmith		q = p;
14238716Smsmith		while (*q && !isspace(*q))
14338716Smsmith		    ++q;
14438716Smsmith	    }
14538716Smsmith	    tmp = *q;
14638716Smsmith	    *q = '\0';
14738716Smsmith	    if ((val = variable_lookup(p)) != NULL) {
14864187Sjhb		size_t len = strlen(val);
14938716Smsmith
15038716Smsmith		strncpy(buf + i, val, PARSE_BUFSIZE - (i + 1));
15138789Smsmith		i += min(len, PARSE_BUFSIZE - 1);
15238716Smsmith	    }
15338716Smsmith	    *q = tmp;	/* restore value */
15438716Smsmith	    p = q + (token ? 1 : 0);
15538716Smsmith	    state = STR;
15638716Smsmith	    break;
15738716Smsmith	}
15838716Smsmith    }
15938716Smsmith    /* If at end of token, add it */
16038716Smsmith    if (i && state == STR) {
16138716Smsmith	buf[i] = '\0';
16238716Smsmith	PARSE_FAIL(insert(&ac, buf));
16338716Smsmith    }
16438716Smsmith    args[ac] = NULL;
16538716Smsmith    *argc = ac;
16638765Sjkh    *argv = (char **)malloc((sizeof(char *) * ac + 1));
16738716Smsmith    bcopy(args, *argv, sizeof(char *) * ac + 1);
16844570Sdcs    free(buf);
16938716Smsmith    free(copy);
17038716Smsmith    return 0;
17138716Smsmith}
17238716Smsmith
17338716Smsmith#define MAXARGS	20
17438716Smsmith
17538716Smsmith/* Clean vector space */
17638716Smsmithstatic void
17738716Smsmithclean(void)
17838716Smsmith{
17938716Smsmith    int		i;
18038716Smsmith
18138716Smsmith    for (i = 0; i < MAXARGS; i++) {
18238716Smsmith	if (args[i] != NULL) {
18338716Smsmith	    free(args[i]);
18438716Smsmith	    args[i] = NULL;
18538716Smsmith	}
18638716Smsmith    }
18738716Smsmith}
18838716Smsmith
18938716Smsmithstatic int
19038716Smsmithinsert(int *argcp, char *buf)
19138716Smsmith{
19238716Smsmith    if (*argcp >= MAXARGS)
19338716Smsmith	return 1;
19438716Smsmith    args[(*argcp)++] = strdup(buf);
19538716Smsmith    return 0;
19638716Smsmith}
19738716Smsmith
19838716Smsmithstatic char *
19938716Smsmithvariable_lookup(char *name)
20038716Smsmith{
20138716Smsmith    /* XXX search "special variable" space first? */
20238716Smsmith    return (char *)getenv(name);
20338716Smsmith}
204