1*22028508SToomas Soome /*
2*22028508SToomas Soome * Redistribution and use in source and binary forms, with or without
3*22028508SToomas Soome * modification, are permitted provided that the following conditions
4*22028508SToomas Soome * are met:
5*22028508SToomas Soome * 1. Redistributions of source code must retain the above copyright
6*22028508SToomas Soome * notice, this list of conditions and the following disclaimer.
7*22028508SToomas Soome * 2. Redistributions in binary form must reproduce the above copyright
8*22028508SToomas Soome * notice, this list of conditions and the following disclaimer in the
9*22028508SToomas Soome * documentation and/or other materials provided with the distribution.
10*22028508SToomas Soome *
11*22028508SToomas Soome * Jordan K. Hubbard
12*22028508SToomas Soome * 29 August 1998
13*22028508SToomas Soome *
14*22028508SToomas Soome * The meat of the simple parser.
15*22028508SToomas Soome */
16*22028508SToomas Soome
17*22028508SToomas Soome #include <sys/cdefs.h>
18*22028508SToomas Soome
19*22028508SToomas Soome #include <stand.h>
20*22028508SToomas Soome #include <string.h>
21*22028508SToomas Soome #include "bootstrap.h"
22*22028508SToomas Soome
23*22028508SToomas Soome static void clean(void);
24*22028508SToomas Soome static int insert(int *argcp, char *buf);
25*22028508SToomas Soome static char *variable_lookup(char *name);
26*22028508SToomas Soome
27*22028508SToomas Soome #define PARSE_BUFSIZE 1024 /* maximum size of one element */
28*22028508SToomas Soome #define MAXARGS 20 /* maximum number of elements */
29*22028508SToomas Soome static char *args[MAXARGS];
30*22028508SToomas Soome
31*22028508SToomas Soome /*
32*22028508SToomas Soome * parse: accept a string of input and "parse" it for backslash
33*22028508SToomas Soome * substitutions and environment variable expansions (${var}),
34*22028508SToomas Soome * returning an argc/argv style vector of whitespace separated
35*22028508SToomas Soome * arguments. Returns 0 on success, 1 on failure (ok, ok, so I
36*22028508SToomas Soome * wimped-out on the error codes! :).
37*22028508SToomas Soome *
38*22028508SToomas Soome * Note that the argv array returned must be freed by the caller, but
39*22028508SToomas Soome * we own the space allocated for arguments and will free that on next
40*22028508SToomas Soome * invocation. This allows argv consumers to modify the array if
41*22028508SToomas Soome * required.
42*22028508SToomas Soome *
43*22028508SToomas Soome * NB: environment variables that expand to more than one whitespace
44*22028508SToomas Soome * separated token will be returned as a single argv[] element, not
45*22028508SToomas Soome * split in turn. Expanded text is also immune to further backslash
46*22028508SToomas Soome * elimination or expansion since this is a one-pass, non-recursive
47*22028508SToomas Soome * parser. You didn't specify more than this so if you want more, ask
48*22028508SToomas Soome * me. - jkh
49*22028508SToomas Soome */
50*22028508SToomas Soome
51*22028508SToomas Soome #define PARSE_FAIL(expr) \
52*22028508SToomas Soome if (expr) { \
53*22028508SToomas Soome printf("fail at line %d\n", __LINE__); \
54*22028508SToomas Soome clean(); \
55*22028508SToomas Soome free(copy); \
56*22028508SToomas Soome free(buf); \
57*22028508SToomas Soome return (1); \
58*22028508SToomas Soome }
59*22028508SToomas Soome
60*22028508SToomas Soome /* Accept the usual delimiters for a variable, returning counterpart */
61*22028508SToomas Soome static char
isdelim(int ch)62*22028508SToomas Soome isdelim(int ch)
63*22028508SToomas Soome {
64*22028508SToomas Soome if (ch == '{')
65*22028508SToomas Soome return ('}');
66*22028508SToomas Soome else if (ch == '(')
67*22028508SToomas Soome return (')');
68*22028508SToomas Soome return ('\0');
69*22028508SToomas Soome }
70*22028508SToomas Soome
71*22028508SToomas Soome static int
isquote(int ch)72*22028508SToomas Soome isquote(int ch)
73*22028508SToomas Soome {
74*22028508SToomas Soome return (ch == '\'');
75*22028508SToomas Soome }
76*22028508SToomas Soome
77*22028508SToomas Soome static int
isdquote(int ch)78*22028508SToomas Soome isdquote(int ch)
79*22028508SToomas Soome {
80*22028508SToomas Soome return (ch == '"');
81*22028508SToomas Soome }
82*22028508SToomas Soome
83*22028508SToomas Soome int
parse(int * argc,char *** argv,char * str)84*22028508SToomas Soome parse(int *argc, char ***argv, char *str)
85*22028508SToomas Soome {
86*22028508SToomas Soome int ac;
87*22028508SToomas Soome char *val, *p, *q, *copy = NULL;
88*22028508SToomas Soome size_t i = 0;
89*22028508SToomas Soome char token, tmp, quote, dquote, *buf;
90*22028508SToomas Soome enum { STR, VAR, WHITE } state;
91*22028508SToomas Soome
92*22028508SToomas Soome ac = *argc = 0;
93*22028508SToomas Soome dquote = quote = 0;
94*22028508SToomas Soome if (!str || (p = copy = backslash(str)) == NULL)
95*22028508SToomas Soome return (1);
96*22028508SToomas Soome
97*22028508SToomas Soome /* Initialize vector and state */
98*22028508SToomas Soome clean();
99*22028508SToomas Soome state = STR;
100*22028508SToomas Soome buf = malloc(PARSE_BUFSIZE);
101*22028508SToomas Soome token = 0;
102*22028508SToomas Soome
103*22028508SToomas Soome /* And awaaaaaaaaay we go! */
104*22028508SToomas Soome while (*p) {
105*22028508SToomas Soome switch (state) {
106*22028508SToomas Soome case STR:
107*22028508SToomas Soome if ((*p == '\\') && p[1]) {
108*22028508SToomas Soome p++;
109*22028508SToomas Soome PARSE_FAIL(i == (PARSE_BUFSIZE - 1));
110*22028508SToomas Soome buf[i++] = *p++;
111*22028508SToomas Soome } else if (isquote(*p)) {
112*22028508SToomas Soome quote = quote ? 0 : *p;
113*22028508SToomas Soome if (dquote) { /* keep quote */
114*22028508SToomas Soome PARSE_FAIL(i == (PARSE_BUFSIZE - 1));
115*22028508SToomas Soome buf[i++] = *p++;
116*22028508SToomas Soome } else {
117*22028508SToomas Soome ++p;
118*22028508SToomas Soome }
119*22028508SToomas Soome } else if (isdquote(*p)) {
120*22028508SToomas Soome dquote = dquote ? 0 : *p;
121*22028508SToomas Soome if (quote) { /* keep dquote */
122*22028508SToomas Soome PARSE_FAIL(i == (PARSE_BUFSIZE - 1));
123*22028508SToomas Soome buf[i++] = *p++;
124*22028508SToomas Soome } else {
125*22028508SToomas Soome ++p;
126*22028508SToomas Soome }
127*22028508SToomas Soome } else if (isspace(*p) && !quote && !dquote) {
128*22028508SToomas Soome state = WHITE;
129*22028508SToomas Soome if (i) {
130*22028508SToomas Soome buf[i] = '\0';
131*22028508SToomas Soome PARSE_FAIL(insert(&ac, buf));
132*22028508SToomas Soome i = 0;
133*22028508SToomas Soome }
134*22028508SToomas Soome ++p;
135*22028508SToomas Soome } else if (*p == '$' && !quote) {
136*22028508SToomas Soome token = isdelim(*(p + 1));
137*22028508SToomas Soome if (token)
138*22028508SToomas Soome p += 2;
139*22028508SToomas Soome else
140*22028508SToomas Soome ++p;
141*22028508SToomas Soome state = VAR;
142*22028508SToomas Soome } else {
143*22028508SToomas Soome PARSE_FAIL(i == (PARSE_BUFSIZE - 1));
144*22028508SToomas Soome buf[i++] = *p++;
145*22028508SToomas Soome }
146*22028508SToomas Soome break;
147*22028508SToomas Soome
148*22028508SToomas Soome case WHITE:
149*22028508SToomas Soome if (isspace(*p))
150*22028508SToomas Soome ++p;
151*22028508SToomas Soome else
152*22028508SToomas Soome state = STR;
153*22028508SToomas Soome break;
154*22028508SToomas Soome
155*22028508SToomas Soome case VAR:
156*22028508SToomas Soome if (token) {
157*22028508SToomas Soome PARSE_FAIL((q = strchr(p, token)) == NULL);
158*22028508SToomas Soome } else {
159*22028508SToomas Soome q = p;
160*22028508SToomas Soome while (*q && !isspace(*q))
161*22028508SToomas Soome ++q;
162*22028508SToomas Soome }
163*22028508SToomas Soome tmp = *q;
164*22028508SToomas Soome *q = '\0';
165*22028508SToomas Soome if ((val = variable_lookup(p)) != NULL) {
166*22028508SToomas Soome size_t len = strlen(val);
167*22028508SToomas Soome
168*22028508SToomas Soome strncpy(buf + i, val, PARSE_BUFSIZE - (i + 1));
169*22028508SToomas Soome i += min(len, PARSE_BUFSIZE - 1);
170*22028508SToomas Soome }
171*22028508SToomas Soome *q = tmp; /* restore value */
172*22028508SToomas Soome p = q + (token ? 1 : 0);
173*22028508SToomas Soome state = STR;
174*22028508SToomas Soome break;
175*22028508SToomas Soome }
176*22028508SToomas Soome }
177*22028508SToomas Soome /* missing terminating ' or " */
178*22028508SToomas Soome PARSE_FAIL(quote || dquote);
179*22028508SToomas Soome /* If at end of token, add it */
180*22028508SToomas Soome if (i && state == STR) {
181*22028508SToomas Soome buf[i] = '\0';
182*22028508SToomas Soome PARSE_FAIL(insert(&ac, buf));
183*22028508SToomas Soome }
184*22028508SToomas Soome args[ac] = NULL;
185*22028508SToomas Soome *argc = ac;
186*22028508SToomas Soome *argv = malloc((sizeof (char *) * ac + 1));
187*22028508SToomas Soome bcopy(args, *argv, sizeof (char *) * ac + 1);
188*22028508SToomas Soome free(buf);
189*22028508SToomas Soome free(copy);
190*22028508SToomas Soome return (0);
191*22028508SToomas Soome }
192*22028508SToomas Soome
193*22028508SToomas Soome #define MAXARGS 20
194*22028508SToomas Soome
195*22028508SToomas Soome /* Clean vector space */
196*22028508SToomas Soome static void
clean(void)197*22028508SToomas Soome clean(void)
198*22028508SToomas Soome {
199*22028508SToomas Soome int i;
200*22028508SToomas Soome
201*22028508SToomas Soome for (i = 0; i < MAXARGS; i++) {
202*22028508SToomas Soome free(args[i]);
203*22028508SToomas Soome args[i] = NULL;
204*22028508SToomas Soome }
205*22028508SToomas Soome }
206*22028508SToomas Soome
207*22028508SToomas Soome static int
insert(int * argcp,char * buf)208*22028508SToomas Soome insert(int *argcp, char *buf)
209*22028508SToomas Soome {
210*22028508SToomas Soome if (*argcp >= MAXARGS)
211*22028508SToomas Soome return (1);
212*22028508SToomas Soome args[(*argcp)++] = strdup(buf);
213*22028508SToomas Soome return (0);
214*22028508SToomas Soome }
215*22028508SToomas Soome
216*22028508SToomas Soome static char *
variable_lookup(char * name)217*22028508SToomas Soome variable_lookup(char *name)
218*22028508SToomas Soome {
219*22028508SToomas Soome
220*22028508SToomas Soome /* XXX search "special variable" space first? */
221*22028508SToomas Soome return (getenv(name));
222*22028508SToomas Soome }
223