1 |
#ifndef lint |
2 |
static const char RCSid[] = "$Id: rexpr.c,v 2.8 2003/06/30 14:59:11 schorsch Exp $"; |
3 |
#endif |
4 |
/* |
5 |
* Regular expression parsing routines. |
6 |
* |
7 |
* External symbols declared in standard.h |
8 |
*/ |
9 |
|
10 |
#include "copyright.h" |
11 |
|
12 |
#include <stdio.h> |
13 |
#include <stdlib.h> |
14 |
#include <ctype.h> |
15 |
#include <string.h> |
16 |
|
17 |
#include "rtio.h" |
18 |
|
19 |
/* |
20 |
* rexpr.c - regular expression parser (ala grep) |
21 |
*/ |
22 |
|
23 |
#define CCHR 2 |
24 |
#define CDOT 4 |
25 |
#define CCL 6 |
26 |
#define NCCL 8 |
27 |
#define CDOL 10 |
28 |
#define CEOF 11 |
29 |
#define CBRC 14 |
30 |
#define CLET 15 |
31 |
#define STAR 01 |
32 |
|
33 |
#define ESIZE 255 |
34 |
|
35 |
#define same(a,b) (a==b || (iflag && (a^b)==' ' && isalpha(a))) |
36 |
|
37 |
|
38 |
static int advance(char *, char *); |
39 |
static int cclass(char *, int c, int af); |
40 |
|
41 |
static char expbuf[ESIZE]; |
42 |
static int iflag; |
43 |
static int circf; |
44 |
|
45 |
int explen; /* length of the last expression found */ |
46 |
|
47 |
int |
48 |
ecompile(sp, iflg, wflag) /* compile the expression */ |
49 |
register char *sp; |
50 |
int iflg, wflag; |
51 |
{ |
52 |
register int c; |
53 |
register char *ep; |
54 |
char *lastep = NULL; |
55 |
int cclcnt; |
56 |
|
57 |
iflag = iflg; |
58 |
ep = expbuf; |
59 |
explen = 0; |
60 |
if (*sp == '^') { |
61 |
circf = 1; |
62 |
sp++; |
63 |
} else |
64 |
circf = 0; |
65 |
if (wflag) |
66 |
*ep++ = CBRC; |
67 |
for (;;) { |
68 |
if (ep >= &expbuf[ESIZE]) |
69 |
return(-1); |
70 |
if ((c = *sp++) != '*') |
71 |
lastep = ep; |
72 |
switch (c) { |
73 |
|
74 |
case '\0': |
75 |
if (wflag) |
76 |
*ep++ = CLET; |
77 |
*ep++ = CEOF; |
78 |
explen = ep - expbuf; |
79 |
return(0); |
80 |
|
81 |
case '.': |
82 |
*ep++ = CDOT; |
83 |
continue; |
84 |
|
85 |
case '*': |
86 |
if (lastep==0) |
87 |
goto defchar; |
88 |
*lastep |= STAR; |
89 |
continue; |
90 |
|
91 |
case '$': |
92 |
if (*sp != '\0') |
93 |
goto defchar; |
94 |
*ep++ = CDOL; |
95 |
continue; |
96 |
|
97 |
case '[': |
98 |
*ep++ = CCL; |
99 |
*ep++ = 0; |
100 |
cclcnt = 1; |
101 |
if ((c = *sp++) == '^') { |
102 |
c = *sp++; |
103 |
ep[-2] = NCCL; |
104 |
} |
105 |
do { |
106 |
*ep++ = c; |
107 |
cclcnt++; |
108 |
if (c=='\0' || ep >= &expbuf[ESIZE]) |
109 |
return(-1); |
110 |
} while ((c = *sp++) != ']'); |
111 |
lastep[1] = cclcnt; |
112 |
continue; |
113 |
|
114 |
case '\\': |
115 |
if ((c = *sp++) == '\0') |
116 |
return(-1); |
117 |
if (c == '<') { |
118 |
if (ep == expbuf || ep[-1] != CBRC) |
119 |
*ep++ = CBRC; |
120 |
continue; |
121 |
} |
122 |
if (c == '>') { |
123 |
*ep++ = CLET; |
124 |
continue; |
125 |
} |
126 |
defchar: |
127 |
default: |
128 |
*ep++ = CCHR; |
129 |
*ep++ = c; |
130 |
} |
131 |
} |
132 |
} |
133 |
|
134 |
char * |
135 |
expsave() /* save compiled string */ |
136 |
{ |
137 |
register char *ep; |
138 |
|
139 |
if (explen == 0) |
140 |
return(NULL); |
141 |
if ((ep = (char *)malloc(explen+3)) == NULL) |
142 |
return(NULL); |
143 |
ep[0] = iflag; |
144 |
ep[1] = circf; |
145 |
ep[2] = explen; |
146 |
(void)memcpy(ep+3, expbuf, explen); |
147 |
return(ep); |
148 |
} |
149 |
|
150 |
void |
151 |
expset(ep) /* install saved string */ |
152 |
register char *ep; |
153 |
{ |
154 |
iflag = ep[0]; |
155 |
circf = ep[1]; |
156 |
(void)memcpy(expbuf, ep+3, ep[2]&0xff); |
157 |
} |
158 |
|
159 |
char * |
160 |
eindex(sp) /* find the expression in string sp */ |
161 |
register char *sp; |
162 |
{ |
163 |
/* check for match at beginning of line, watch CBRC */ |
164 |
if (advance(sp, expbuf[0]==CBRC ? expbuf+1 : expbuf)) |
165 |
return(sp); |
166 |
if (circf) |
167 |
return(NULL); |
168 |
/* fast check for first character */ |
169 |
if (expbuf[0]==CCHR) { |
170 |
register int c = expbuf[1]; |
171 |
while (*++sp) |
172 |
if (same(*sp, c) && advance(sp, expbuf)) |
173 |
return(sp); |
174 |
return(NULL); |
175 |
} |
176 |
/* regular algorithm */ |
177 |
while (*++sp) |
178 |
if (advance(sp, expbuf)) |
179 |
return(sp); |
180 |
return(NULL); |
181 |
} |
182 |
|
183 |
static int |
184 |
advance(alp, ep) |
185 |
char *alp; |
186 |
register char *ep; |
187 |
{ |
188 |
register char *lp; |
189 |
char *curlp; |
190 |
|
191 |
lp = alp; |
192 |
for (;;) switch (*ep++) { |
193 |
|
194 |
case CCHR: |
195 |
if (!same(*ep, *lp)) |
196 |
return (0); |
197 |
ep++, lp++; |
198 |
continue; |
199 |
|
200 |
case CDOT: |
201 |
if (*lp++) |
202 |
continue; |
203 |
return(0); |
204 |
|
205 |
case CDOL: |
206 |
if (*lp==0) |
207 |
continue; |
208 |
return(0); |
209 |
|
210 |
case CEOF: |
211 |
explen = lp - alp; |
212 |
return(1); |
213 |
|
214 |
case CCL: |
215 |
if (cclass(ep, *lp++, 1)) { |
216 |
ep += *ep; |
217 |
continue; |
218 |
} |
219 |
return(0); |
220 |
|
221 |
case NCCL: |
222 |
if (cclass(ep, *lp++, 0)) { |
223 |
ep += *ep; |
224 |
continue; |
225 |
} |
226 |
return(0); |
227 |
|
228 |
case CDOT|STAR: |
229 |
curlp = lp; |
230 |
while (*lp++); |
231 |
goto star; |
232 |
|
233 |
case CCHR|STAR: |
234 |
curlp = lp; |
235 |
while (same(*lp, *ep)) |
236 |
lp++; |
237 |
lp++; |
238 |
ep++; |
239 |
goto star; |
240 |
|
241 |
case CCL|STAR: |
242 |
case NCCL|STAR: |
243 |
curlp = lp; |
244 |
while (cclass(ep, *lp++, ep[-1]==(CCL|STAR))); |
245 |
ep += *ep; |
246 |
star: |
247 |
do { |
248 |
lp--; |
249 |
if (advance(lp, ep)) { |
250 |
explen += lp - alp; |
251 |
return(1); |
252 |
} |
253 |
} while (lp > curlp); |
254 |
return(0); |
255 |
|
256 |
case CBRC: |
257 |
if ((isalnum(*lp) || *lp == '_') && !(isalnum(lp[-1]) || lp[-1] == '_')) |
258 |
continue; |
259 |
return (0); |
260 |
|
261 |
case CLET: |
262 |
if (!isalnum(*lp) && *lp != '_') |
263 |
continue; |
264 |
return (0); |
265 |
|
266 |
default: |
267 |
fprintf(stderr, "RE botch\n"); |
268 |
} |
269 |
} |
270 |
|
271 |
static int |
272 |
cclass(set, c, af) |
273 |
register char *set; |
274 |
register int c; |
275 |
int af; |
276 |
{ |
277 |
register int n; |
278 |
|
279 |
if (c == 0) |
280 |
return(0); |
281 |
n = *set++; |
282 |
while (--n) |
283 |
if (n > 2 && set[1] == '-') { |
284 |
if (c >= set[0] && c <= set[2]) |
285 |
return (af); |
286 |
set += 3; |
287 |
n -= 2; |
288 |
} else |
289 |
if (*set++ == c) |
290 |
return(af); |
291 |
return(!af); |
292 |
} |