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 <stand.h>
18ca987d46SWarner Losh #include <string.h>
19ca987d46SWarner Losh #include "bootstrap.h"
20ca987d46SWarner Losh
21ca987d46SWarner Losh static void clean(void);
22ca987d46SWarner Losh static int insert(int *argcp, char *buf);
23ca987d46SWarner Losh static char *variable_lookup(char *name);
24ca987d46SWarner Losh
25ca987d46SWarner Losh #define PARSE_BUFSIZE 1024 /* maximum size of one element */
26ca987d46SWarner Losh #define MAXARGS 20 /* maximum number of elements */
27ca987d46SWarner Losh static char *args[MAXARGS];
28ca987d46SWarner Losh
29ca987d46SWarner Losh /*
30ca987d46SWarner Losh * parse: accept a string of input and "parse" it for backslash
31ca987d46SWarner Losh * substitutions and environment variable expansions (${var}),
32ca987d46SWarner Losh * returning an argc/argv style vector of whitespace separated
33ca987d46SWarner Losh * arguments. Returns 0 on success, 1 on failure (ok, ok, so I
34ca987d46SWarner Losh * wimped-out on the error codes! :).
35ca987d46SWarner Losh *
36ca987d46SWarner Losh * Note that the argv array returned must be freed by the caller, but
37ca987d46SWarner Losh * we own the space allocated for arguments and will free that on next
38ca987d46SWarner Losh * invocation. This allows argv consumers to modify the array if
39ca987d46SWarner Losh * required.
40ca987d46SWarner Losh *
41ca987d46SWarner Losh * NB: environment variables that expand to more than one whitespace
42ca987d46SWarner Losh * separated token will be returned as a single argv[] element, not
43ca987d46SWarner Losh * split in turn. Expanded text is also immune to further backslash
44ca987d46SWarner Losh * elimination or expansion since this is a one-pass, non-recursive
45ca987d46SWarner Losh * parser. You didn't specify more than this so if you want more, ask
46ca987d46SWarner Losh * me. - jkh
47ca987d46SWarner Losh */
48ca987d46SWarner Losh
49ca987d46SWarner Losh #define PARSE_FAIL(expr) \
50ca987d46SWarner Losh if (expr) { \
51ca987d46SWarner Losh printf("fail at line %d\n", __LINE__); \
52ca987d46SWarner Losh clean(); \
53ca987d46SWarner Losh free(copy); \
54ca987d46SWarner Losh free(buf); \
55ca987d46SWarner Losh return 1; \
56ca987d46SWarner Losh }
57ca987d46SWarner Losh
58ca987d46SWarner Losh /* Accept the usual delimiters for a variable, returning counterpart */
59ca987d46SWarner Losh static char
isdelim(int ch)60ca987d46SWarner Losh isdelim(int ch)
61ca987d46SWarner Losh {
62*16bb6523SWarner Losh
63ca987d46SWarner Losh if (ch == '{')
64ca987d46SWarner Losh return '}';
65ca987d46SWarner Losh else if (ch == '(')
66ca987d46SWarner Losh return ')';
67ca987d46SWarner Losh return '\0';
68ca987d46SWarner Losh }
69ca987d46SWarner Losh
70ca987d46SWarner Losh static int
isquote(int ch)71ca987d46SWarner Losh isquote(int ch)
72ca987d46SWarner Losh {
73*16bb6523SWarner Losh
74ca987d46SWarner Losh return (ch == '\'');
75ca987d46SWarner Losh }
76ca987d46SWarner Losh
77ca987d46SWarner Losh static int
isdquote(int ch)78ca987d46SWarner Losh isdquote(int ch)
79ca987d46SWarner Losh {
80*16bb6523SWarner Losh
81ca987d46SWarner Losh return (ch == '"');
82ca987d46SWarner Losh }
83ca987d46SWarner Losh
84ca987d46SWarner Losh int
parse(int * argc,char *** argv,const char * str)85bd04a914SWarner Losh parse(int *argc, char ***argv, const char *str)
86ca987d46SWarner Losh {
87ca987d46SWarner Losh int ac;
88ca987d46SWarner Losh char *val, *p, *q, *copy = NULL;
89ca987d46SWarner Losh size_t i = 0;
90ca987d46SWarner Losh char token, tmp, quote, dquote, *buf;
91ca987d46SWarner Losh enum { STR, VAR, WHITE } state;
92ca987d46SWarner Losh
93ca987d46SWarner Losh ac = *argc = 0;
94ca987d46SWarner Losh dquote = quote = 0;
95ca987d46SWarner Losh if (!str || (p = copy = backslash(str)) == NULL)
96ca987d46SWarner Losh return 1;
97ca987d46SWarner Losh
98ca987d46SWarner Losh /* Initialize vector and state */
99ca987d46SWarner Losh clean();
100ca987d46SWarner Losh state = STR;
101ca987d46SWarner Losh buf = (char *)malloc(PARSE_BUFSIZE);
102ca987d46SWarner Losh token = 0;
103ca987d46SWarner Losh
104ca987d46SWarner Losh /* And awaaaaaaaaay we go! */
105ca987d46SWarner Losh while (*p) {
106ca987d46SWarner Losh switch (state) {
107ca987d46SWarner Losh case STR:
108ca987d46SWarner Losh if ((*p == '\\') && p[1]) {
109ca987d46SWarner Losh p++;
110ca987d46SWarner Losh PARSE_FAIL(i == (PARSE_BUFSIZE - 1));
111ca987d46SWarner Losh buf[i++] = *p++;
112ca987d46SWarner Losh } else if (isquote(*p)) {
113ca987d46SWarner Losh quote = quote ? 0 : *p;
114ca987d46SWarner Losh if (dquote) { /* keep quote */
115ca987d46SWarner Losh PARSE_FAIL(i == (PARSE_BUFSIZE - 1));
116ca987d46SWarner Losh buf[i++] = *p++;
117ca987d46SWarner Losh } else
118ca987d46SWarner Losh ++p;
119ca987d46SWarner Losh } else if (isdquote(*p)) {
120ca987d46SWarner Losh dquote = dquote ? 0 : *p;
121ca987d46SWarner Losh if (quote) { /* keep dquote */
122ca987d46SWarner Losh PARSE_FAIL(i == (PARSE_BUFSIZE - 1));
123ca987d46SWarner Losh buf[i++] = *p++;
124ca987d46SWarner Losh } else
125ca987d46SWarner Losh ++p;
126ca987d46SWarner Losh } else if (isspace(*p) && !quote && !dquote) {
127ca987d46SWarner Losh state = WHITE;
128ca987d46SWarner Losh if (i) {
129ca987d46SWarner Losh buf[i] = '\0';
130ca987d46SWarner Losh PARSE_FAIL(insert(&ac, buf));
131ca987d46SWarner Losh i = 0;
132ca987d46SWarner Losh }
133ca987d46SWarner Losh ++p;
134ca987d46SWarner Losh } else if (*p == '$' && !quote) {
135ca987d46SWarner Losh token = isdelim(*(p + 1));
136ca987d46SWarner Losh if (token)
137ca987d46SWarner Losh p += 2;
138ca987d46SWarner Losh else
139ca987d46SWarner Losh ++p;
140ca987d46SWarner Losh state = VAR;
141ca987d46SWarner Losh } else {
142ca987d46SWarner Losh PARSE_FAIL(i == (PARSE_BUFSIZE - 1));
143ca987d46SWarner Losh buf[i++] = *p++;
144ca987d46SWarner Losh }
145ca987d46SWarner Losh break;
146ca987d46SWarner Losh
147ca987d46SWarner Losh case WHITE:
148ca987d46SWarner Losh if (isspace(*p))
149ca987d46SWarner Losh ++p;
150ca987d46SWarner Losh else
151ca987d46SWarner Losh state = STR;
152ca987d46SWarner Losh break;
153ca987d46SWarner Losh
154ca987d46SWarner Losh case VAR:
155ca987d46SWarner Losh if (token) {
156ca987d46SWarner Losh PARSE_FAIL((q = strchr(p, token)) == NULL);
157ca987d46SWarner Losh } else {
158ca987d46SWarner Losh q = p;
159ca987d46SWarner Losh while (*q && !isspace(*q))
160ca987d46SWarner Losh ++q;
161ca987d46SWarner Losh }
162ca987d46SWarner Losh tmp = *q;
163ca987d46SWarner Losh *q = '\0';
164ca987d46SWarner Losh if ((val = variable_lookup(p)) != NULL) {
165ca987d46SWarner Losh size_t len = strlen(val);
166ca987d46SWarner Losh
167ca987d46SWarner Losh strncpy(buf + i, val, PARSE_BUFSIZE - (i + 1));
168ca987d46SWarner Losh i += min(len, PARSE_BUFSIZE - 1);
169ca987d46SWarner Losh }
170ca987d46SWarner Losh *q = tmp; /* restore value */
171ca987d46SWarner Losh p = q + (token ? 1 : 0);
172ca987d46SWarner Losh state = STR;
173ca987d46SWarner Losh break;
174ca987d46SWarner Losh }
175ca987d46SWarner Losh }
176ca987d46SWarner Losh /* missing terminating ' or " */
177ca987d46SWarner Losh PARSE_FAIL(quote || dquote);
178ca987d46SWarner Losh /* If at end of token, add it */
179ca987d46SWarner Losh if (i && state == STR) {
180ca987d46SWarner Losh buf[i] = '\0';
181ca987d46SWarner Losh PARSE_FAIL(insert(&ac, buf));
182ca987d46SWarner Losh }
183ca987d46SWarner Losh args[ac] = NULL;
184ca987d46SWarner Losh *argc = ac;
185ca987d46SWarner Losh *argv = (char **)malloc((sizeof(char *) * ac + 1));
186ca987d46SWarner Losh bcopy(args, *argv, sizeof(char *) * ac + 1);
187ca987d46SWarner Losh free(buf);
188ca987d46SWarner Losh free(copy);
189ca987d46SWarner Losh return 0;
190ca987d46SWarner Losh }
191ca987d46SWarner Losh
192ca987d46SWarner Losh #define MAXARGS 20
193ca987d46SWarner Losh
194ca987d46SWarner Losh /* Clean vector space */
195ca987d46SWarner Losh static void
clean(void)196ca987d46SWarner Losh clean(void)
197ca987d46SWarner Losh {
198ca987d46SWarner Losh int i;
199ca987d46SWarner Losh
200ca987d46SWarner Losh for (i = 0; i < MAXARGS; i++) {
201ca987d46SWarner Losh if (args[i] != NULL) {
202ca987d46SWarner Losh free(args[i]);
203ca987d46SWarner Losh args[i] = NULL;
204ca987d46SWarner Losh }
205ca987d46SWarner Losh }
206ca987d46SWarner Losh }
207ca987d46SWarner Losh
208ca987d46SWarner Losh static int
insert(int * argcp,char * buf)209ca987d46SWarner Losh insert(int *argcp, char *buf)
210ca987d46SWarner Losh {
211*16bb6523SWarner Losh
212ca987d46SWarner Losh if (*argcp >= MAXARGS)
213ca987d46SWarner Losh return 1;
214ca987d46SWarner Losh args[(*argcp)++] = strdup(buf);
215ca987d46SWarner Losh return 0;
216ca987d46SWarner Losh }
217ca987d46SWarner Losh
218ca987d46SWarner Losh static char *
variable_lookup(char * name)219ca987d46SWarner Losh variable_lookup(char *name)
220ca987d46SWarner Losh {
221*16bb6523SWarner Losh
222ca987d46SWarner Losh /* XXX search "special variable" space first? */
223ca987d46SWarner Losh return (char *)getenv(name);
224ca987d46SWarner Losh }
225