xref: /freebsd/stand/common/interp_parse.c (revision 16bb6523bc4a99566a377f8a70b5d0139074f612)
1ca987d46SWarner Losh /*-
2ca987d46SWarner Losh  * Redistribution and use in source and binary forms, with or without
3ca987d46SWarner Losh  * modification, are permitted provided that the following conditions
4ca987d46SWarner Losh  * are met:
5ca987d46SWarner Losh  * 1. Redistributions of source code must retain the above copyright
6ca987d46SWarner Losh  *    notice, this list of conditions and the following disclaimer.
7ca987d46SWarner Losh  * 2. Redistributions in binary form must reproduce the above copyright
8ca987d46SWarner Losh  *    notice, this list of conditions and the following disclaimer in the
9ca987d46SWarner Losh  *    documentation and/or other materials provided with the distribution.
10ca987d46SWarner Losh  *
11ca987d46SWarner Losh  * Jordan K. Hubbard
12ca987d46SWarner Losh  * 29 August 1998
13ca987d46SWarner Losh  *
14ca987d46SWarner Losh  * The meat of the simple parser.
15ca987d46SWarner Losh  */
16ca987d46SWarner Losh 
17ca987d46SWarner Losh #include <sys/cdefs.h>
18ca987d46SWarner Losh __FBSDID("$FreeBSD$");
19ca987d46SWarner Losh 
20ca987d46SWarner Losh #include <stand.h>
21ca987d46SWarner Losh #include <string.h>
22ca987d46SWarner Losh #include "bootstrap.h"
23ca987d46SWarner Losh 
24ca987d46SWarner Losh static void	 clean(void);
25ca987d46SWarner Losh static int	 insert(int *argcp, char *buf);
26ca987d46SWarner Losh static char	*variable_lookup(char *name);
27ca987d46SWarner Losh 
28ca987d46SWarner Losh #define PARSE_BUFSIZE	1024	/* maximum size of one element */
29ca987d46SWarner Losh #define MAXARGS		20	/* maximum number of elements */
30ca987d46SWarner Losh static char		*args[MAXARGS];
31ca987d46SWarner Losh 
32ca987d46SWarner Losh /*
33ca987d46SWarner Losh  * parse: accept a string of input and "parse" it for backslash
34ca987d46SWarner Losh  * substitutions and environment variable expansions (${var}),
35ca987d46SWarner Losh  * returning an argc/argv style vector of whitespace separated
36ca987d46SWarner Losh  * arguments.  Returns 0 on success, 1 on failure (ok, ok, so I
37ca987d46SWarner Losh  * wimped-out on the error codes! :).
38ca987d46SWarner Losh  *
39ca987d46SWarner Losh  * Note that the argv array returned must be freed by the caller, but
40ca987d46SWarner Losh  * we own the space allocated for arguments and will free that on next
41ca987d46SWarner Losh  * invocation.  This allows argv consumers to modify the array if
42ca987d46SWarner Losh  * required.
43ca987d46SWarner Losh  *
44ca987d46SWarner Losh  * NB: environment variables that expand to more than one whitespace
45ca987d46SWarner Losh  * separated token will be returned as a single argv[] element, not
46ca987d46SWarner Losh  * split in turn.  Expanded text is also immune to further backslash
47ca987d46SWarner Losh  * elimination or expansion since this is a one-pass, non-recursive
48ca987d46SWarner Losh  * parser.  You didn't specify more than this so if you want more, ask
49ca987d46SWarner Losh  * me. - jkh
50ca987d46SWarner Losh  */
51ca987d46SWarner Losh 
52ca987d46SWarner Losh #define PARSE_FAIL(expr) \
53ca987d46SWarner Losh 	if (expr) { \
54ca987d46SWarner Losh 		printf("fail at line %d\n", __LINE__); \
55ca987d46SWarner Losh 		clean(); \
56ca987d46SWarner Losh 		free(copy); \
57ca987d46SWarner Losh 		free(buf); \
58ca987d46SWarner Losh 		return 1; \
59ca987d46SWarner Losh 	}
60ca987d46SWarner Losh 
61ca987d46SWarner Losh /* Accept the usual delimiters for a variable, returning counterpart */
62ca987d46SWarner Losh static char
63ca987d46SWarner Losh isdelim(int ch)
64ca987d46SWarner Losh {
65*16bb6523SWarner Losh 
66ca987d46SWarner Losh 	if (ch == '{')
67ca987d46SWarner Losh 		return '}';
68ca987d46SWarner Losh 	else if (ch == '(')
69ca987d46SWarner Losh 		return ')';
70ca987d46SWarner Losh 	return '\0';
71ca987d46SWarner Losh }
72ca987d46SWarner Losh 
73ca987d46SWarner Losh static int
74ca987d46SWarner Losh isquote(int ch)
75ca987d46SWarner Losh {
76*16bb6523SWarner Losh 
77ca987d46SWarner Losh 	return (ch == '\'');
78ca987d46SWarner Losh }
79ca987d46SWarner Losh 
80ca987d46SWarner Losh static int
81ca987d46SWarner Losh isdquote(int ch)
82ca987d46SWarner Losh {
83*16bb6523SWarner Losh 
84ca987d46SWarner Losh 	return (ch == '"');
85ca987d46SWarner Losh }
86ca987d46SWarner Losh 
87ca987d46SWarner Losh int
88bd04a914SWarner Losh parse(int *argc, char ***argv, const char *str)
89ca987d46SWarner Losh {
90ca987d46SWarner Losh 	int ac;
91ca987d46SWarner Losh 	char *val, *p, *q, *copy = NULL;
92ca987d46SWarner Losh 	size_t i = 0;
93ca987d46SWarner Losh 	char token, tmp, quote, dquote, *buf;
94ca987d46SWarner Losh 	enum { STR, VAR, WHITE } state;
95ca987d46SWarner Losh 
96ca987d46SWarner Losh 	ac = *argc = 0;
97ca987d46SWarner Losh 	dquote = quote = 0;
98ca987d46SWarner Losh 	if (!str || (p = copy = backslash(str)) == NULL)
99ca987d46SWarner Losh 		return 1;
100ca987d46SWarner Losh 
101ca987d46SWarner Losh 	/* Initialize vector and state */
102ca987d46SWarner Losh 	clean();
103ca987d46SWarner Losh 	state = STR;
104ca987d46SWarner Losh 	buf = (char *)malloc(PARSE_BUFSIZE);
105ca987d46SWarner Losh 	token = 0;
106ca987d46SWarner Losh 
107ca987d46SWarner Losh 	/* And awaaaaaaaaay we go! */
108ca987d46SWarner Losh 	while (*p) {
109ca987d46SWarner Losh 		switch (state) {
110ca987d46SWarner Losh 		case STR:
111ca987d46SWarner Losh 			if ((*p == '\\') && p[1]) {
112ca987d46SWarner Losh 				p++;
113ca987d46SWarner Losh 				PARSE_FAIL(i == (PARSE_BUFSIZE - 1));
114ca987d46SWarner Losh 				buf[i++] = *p++;
115ca987d46SWarner Losh 			} else if (isquote(*p)) {
116ca987d46SWarner Losh 				quote = quote ? 0 : *p;
117ca987d46SWarner Losh 				if (dquote) { /* keep quote */
118ca987d46SWarner Losh 					PARSE_FAIL(i == (PARSE_BUFSIZE - 1));
119ca987d46SWarner Losh 					buf[i++] = *p++;
120ca987d46SWarner Losh 				} else
121ca987d46SWarner Losh 					++p;
122ca987d46SWarner Losh 			} else if (isdquote(*p)) {
123ca987d46SWarner Losh 				dquote = dquote ? 0 : *p;
124ca987d46SWarner Losh 				if (quote) { /* keep dquote */
125ca987d46SWarner Losh 					PARSE_FAIL(i == (PARSE_BUFSIZE - 1));
126ca987d46SWarner Losh 					buf[i++] = *p++;
127ca987d46SWarner Losh 				} else
128ca987d46SWarner Losh 					++p;
129ca987d46SWarner Losh 			} else if (isspace(*p) && !quote && !dquote) {
130ca987d46SWarner Losh 				state = WHITE;
131ca987d46SWarner Losh 				if (i) {
132ca987d46SWarner Losh 					buf[i] = '\0';
133ca987d46SWarner Losh 					PARSE_FAIL(insert(&ac, buf));
134ca987d46SWarner Losh 					i = 0;
135ca987d46SWarner Losh 				}
136ca987d46SWarner Losh 				++p;
137ca987d46SWarner Losh 			} else if (*p == '$' && !quote) {
138ca987d46SWarner Losh 				token = isdelim(*(p + 1));
139ca987d46SWarner Losh 				if (token)
140ca987d46SWarner Losh 					p += 2;
141ca987d46SWarner Losh 				else
142ca987d46SWarner Losh 					++p;
143ca987d46SWarner Losh 				state = VAR;
144ca987d46SWarner Losh 			} else {
145ca987d46SWarner Losh 				PARSE_FAIL(i == (PARSE_BUFSIZE - 1));
146ca987d46SWarner Losh 				buf[i++] = *p++;
147ca987d46SWarner Losh 			}
148ca987d46SWarner Losh 			break;
149ca987d46SWarner Losh 
150ca987d46SWarner Losh 		case WHITE:
151ca987d46SWarner Losh 			if (isspace(*p))
152ca987d46SWarner Losh 				++p;
153ca987d46SWarner Losh 			else
154ca987d46SWarner Losh 				state = STR;
155ca987d46SWarner Losh 			break;
156ca987d46SWarner Losh 
157ca987d46SWarner Losh 		case VAR:
158ca987d46SWarner Losh 			if (token) {
159ca987d46SWarner Losh 				PARSE_FAIL((q = strchr(p, token)) == NULL);
160ca987d46SWarner Losh 			} else {
161ca987d46SWarner Losh 				q = p;
162ca987d46SWarner Losh 				while (*q && !isspace(*q))
163ca987d46SWarner Losh 					++q;
164ca987d46SWarner Losh 			}
165ca987d46SWarner Losh 			tmp = *q;
166ca987d46SWarner Losh 			*q = '\0';
167ca987d46SWarner Losh 			if ((val = variable_lookup(p)) != NULL) {
168ca987d46SWarner Losh 				size_t len = strlen(val);
169ca987d46SWarner Losh 
170ca987d46SWarner Losh 				strncpy(buf + i, val, PARSE_BUFSIZE - (i + 1));
171ca987d46SWarner Losh 				i += min(len, PARSE_BUFSIZE - 1);
172ca987d46SWarner Losh 			}
173ca987d46SWarner Losh 			*q = tmp;	/* restore value */
174ca987d46SWarner Losh 			p = q + (token ? 1 : 0);
175ca987d46SWarner Losh 			state = STR;
176ca987d46SWarner Losh 			break;
177ca987d46SWarner Losh 		}
178ca987d46SWarner Losh 	}
179ca987d46SWarner Losh 	/* missing terminating ' or " */
180ca987d46SWarner Losh 	PARSE_FAIL(quote || dquote);
181ca987d46SWarner Losh 	/* If at end of token, add it */
182ca987d46SWarner Losh 	if (i && state == STR) {
183ca987d46SWarner Losh 		buf[i] = '\0';
184ca987d46SWarner Losh 		PARSE_FAIL(insert(&ac, buf));
185ca987d46SWarner Losh 	}
186ca987d46SWarner Losh 	args[ac] = NULL;
187ca987d46SWarner Losh 	*argc = ac;
188ca987d46SWarner Losh 	*argv = (char **)malloc((sizeof(char *) * ac + 1));
189ca987d46SWarner Losh 	bcopy(args, *argv, sizeof(char *) * ac + 1);
190ca987d46SWarner Losh 	free(buf);
191ca987d46SWarner Losh 	free(copy);
192ca987d46SWarner Losh 	return 0;
193ca987d46SWarner Losh }
194ca987d46SWarner Losh 
195ca987d46SWarner Losh #define MAXARGS	20
196ca987d46SWarner Losh 
197ca987d46SWarner Losh /* Clean vector space */
198ca987d46SWarner Losh static void
199ca987d46SWarner Losh clean(void)
200ca987d46SWarner Losh {
201ca987d46SWarner Losh 	int		i;
202ca987d46SWarner Losh 
203ca987d46SWarner Losh 	for (i = 0; i < MAXARGS; i++) {
204ca987d46SWarner Losh 		if (args[i] != NULL) {
205ca987d46SWarner Losh 			free(args[i]);
206ca987d46SWarner Losh 			args[i] = NULL;
207ca987d46SWarner Losh 		}
208ca987d46SWarner Losh 	}
209ca987d46SWarner Losh }
210ca987d46SWarner Losh 
211ca987d46SWarner Losh static int
212ca987d46SWarner Losh insert(int *argcp, char *buf)
213ca987d46SWarner Losh {
214*16bb6523SWarner Losh 
215ca987d46SWarner Losh 	if (*argcp >= MAXARGS)
216ca987d46SWarner Losh 		return 1;
217ca987d46SWarner Losh 	args[(*argcp)++] = strdup(buf);
218ca987d46SWarner Losh 	return 0;
219ca987d46SWarner Losh }
220ca987d46SWarner Losh 
221ca987d46SWarner Losh static char *
222ca987d46SWarner Losh variable_lookup(char *name)
223ca987d46SWarner Losh {
224*16bb6523SWarner Losh 
225ca987d46SWarner Losh 	/* XXX search "special variable" space first? */
226ca987d46SWarner Losh 	return (char *)getenv(name);
227ca987d46SWarner Losh }
228