1 |
#ifndef lint |
2 |
static const char RCSid[] = "$Id: rexpr.c,v 2.7 2003/06/07 12:50:20 schorsch Exp $"; |
3 |
#endif |
4 |
/* |
5 |
* Regular expression parsing routines. |
6 |
* |
7 |
* External symbols declared in standard.h |
8 |
*/ |
9 |
|
10 |
#include "copyright.h" |
11 |
|
12 |
#include <stdio.h> |
13 |
#include <stdlib.h> |
14 |
#include <ctype.h> |
15 |
#include <string.h> |
16 |
|
17 |
/* |
18 |
* rexpr.c - regular expression parser (ala grep) |
19 |
*/ |
20 |
|
21 |
#define CCHR 2 |
22 |
#define CDOT 4 |
23 |
#define CCL 6 |
24 |
#define NCCL 8 |
25 |
#define CDOL 10 |
26 |
#define CEOF 11 |
27 |
#define CBRC 14 |
28 |
#define CLET 15 |
29 |
#define STAR 01 |
30 |
|
31 |
#define ESIZE 255 |
32 |
|
33 |
#define same(a,b) (a==b || (iflag && (a^b)==' ' && isalpha(a))) |
34 |
|
35 |
|
36 |
static int advance(), cclass(); |
37 |
|
38 |
static char expbuf[ESIZE]; |
39 |
static int iflag; |
40 |
static int circf; |
41 |
|
42 |
int explen; /* length of the last expression found */ |
43 |
|
44 |
int |
45 |
ecompile(sp, iflg, wflag) /* compile the expression */ |
46 |
register char *sp; |
47 |
int iflg, wflag; |
48 |
{ |
49 |
register c; |
50 |
register char *ep; |
51 |
char *lastep; |
52 |
int cclcnt; |
53 |
|
54 |
iflag = iflg; |
55 |
ep = expbuf; |
56 |
explen = 0; |
57 |
if (*sp == '^') { |
58 |
circf = 1; |
59 |
sp++; |
60 |
} else |
61 |
circf = 0; |
62 |
if (wflag) |
63 |
*ep++ = CBRC; |
64 |
for (;;) { |
65 |
if (ep >= &expbuf[ESIZE]) |
66 |
return(-1); |
67 |
if ((c = *sp++) != '*') |
68 |
lastep = ep; |
69 |
switch (c) { |
70 |
|
71 |
case '\0': |
72 |
if (wflag) |
73 |
*ep++ = CLET; |
74 |
*ep++ = CEOF; |
75 |
explen = ep - expbuf; |
76 |
return(0); |
77 |
|
78 |
case '.': |
79 |
*ep++ = CDOT; |
80 |
continue; |
81 |
|
82 |
case '*': |
83 |
if (lastep==0) |
84 |
goto defchar; |
85 |
*lastep |= STAR; |
86 |
continue; |
87 |
|
88 |
case '$': |
89 |
if (*sp != '\0') |
90 |
goto defchar; |
91 |
*ep++ = CDOL; |
92 |
continue; |
93 |
|
94 |
case '[': |
95 |
*ep++ = CCL; |
96 |
*ep++ = 0; |
97 |
cclcnt = 1; |
98 |
if ((c = *sp++) == '^') { |
99 |
c = *sp++; |
100 |
ep[-2] = NCCL; |
101 |
} |
102 |
do { |
103 |
*ep++ = c; |
104 |
cclcnt++; |
105 |
if (c=='\0' || ep >= &expbuf[ESIZE]) |
106 |
return(-1); |
107 |
} while ((c = *sp++) != ']'); |
108 |
lastep[1] = cclcnt; |
109 |
continue; |
110 |
|
111 |
case '\\': |
112 |
if ((c = *sp++) == '\0') |
113 |
return(-1); |
114 |
if (c == '<') { |
115 |
if (ep == expbuf || ep[-1] != CBRC) |
116 |
*ep++ = CBRC; |
117 |
continue; |
118 |
} |
119 |
if (c == '>') { |
120 |
*ep++ = CLET; |
121 |
continue; |
122 |
} |
123 |
defchar: |
124 |
default: |
125 |
*ep++ = CCHR; |
126 |
*ep++ = c; |
127 |
} |
128 |
} |
129 |
} |
130 |
|
131 |
char * |
132 |
expsave() /* save compiled string */ |
133 |
{ |
134 |
register char *ep; |
135 |
|
136 |
if (explen == 0) |
137 |
return(NULL); |
138 |
if ((ep = (char *)malloc(explen+3)) == NULL) |
139 |
return(NULL); |
140 |
ep[0] = iflag; |
141 |
ep[1] = circf; |
142 |
ep[2] = explen; |
143 |
(void)memcpy(ep+3, expbuf, explen); |
144 |
return(ep); |
145 |
} |
146 |
|
147 |
void |
148 |
expset(ep) /* install saved string */ |
149 |
register char *ep; |
150 |
{ |
151 |
iflag = ep[0]; |
152 |
circf = ep[1]; |
153 |
(void)memcpy(expbuf, ep+3, ep[2]&0xff); |
154 |
} |
155 |
|
156 |
char * |
157 |
eindex(sp) /* find the expression in string sp */ |
158 |
register char *sp; |
159 |
{ |
160 |
/* check for match at beginning of line, watch CBRC */ |
161 |
if (advance(sp, expbuf[0]==CBRC ? expbuf+1 : expbuf)) |
162 |
return(sp); |
163 |
if (circf) |
164 |
return(NULL); |
165 |
/* fast check for first character */ |
166 |
if (expbuf[0]==CCHR) { |
167 |
register c = expbuf[1]; |
168 |
while (*++sp) |
169 |
if (same(*sp, c) && advance(sp, expbuf)) |
170 |
return(sp); |
171 |
return(NULL); |
172 |
} |
173 |
/* regular algorithm */ |
174 |
while (*++sp) |
175 |
if (advance(sp, expbuf)) |
176 |
return(sp); |
177 |
return(NULL); |
178 |
} |
179 |
|
180 |
static int |
181 |
advance(alp, ep) |
182 |
char *alp; |
183 |
register char *ep; |
184 |
{ |
185 |
register char *lp; |
186 |
char *curlp; |
187 |
|
188 |
lp = alp; |
189 |
for (;;) switch (*ep++) { |
190 |
|
191 |
case CCHR: |
192 |
if (!same(*ep, *lp)) |
193 |
return (0); |
194 |
ep++, lp++; |
195 |
continue; |
196 |
|
197 |
case CDOT: |
198 |
if (*lp++) |
199 |
continue; |
200 |
return(0); |
201 |
|
202 |
case CDOL: |
203 |
if (*lp==0) |
204 |
continue; |
205 |
return(0); |
206 |
|
207 |
case CEOF: |
208 |
explen = lp - alp; |
209 |
return(1); |
210 |
|
211 |
case CCL: |
212 |
if (cclass(ep, *lp++, 1)) { |
213 |
ep += *ep; |
214 |
continue; |
215 |
} |
216 |
return(0); |
217 |
|
218 |
case NCCL: |
219 |
if (cclass(ep, *lp++, 0)) { |
220 |
ep += *ep; |
221 |
continue; |
222 |
} |
223 |
return(0); |
224 |
|
225 |
case CDOT|STAR: |
226 |
curlp = lp; |
227 |
while (*lp++); |
228 |
goto star; |
229 |
|
230 |
case CCHR|STAR: |
231 |
curlp = lp; |
232 |
while (same(*lp, *ep)) |
233 |
lp++; |
234 |
lp++; |
235 |
ep++; |
236 |
goto star; |
237 |
|
238 |
case CCL|STAR: |
239 |
case NCCL|STAR: |
240 |
curlp = lp; |
241 |
while (cclass(ep, *lp++, ep[-1]==(CCL|STAR))); |
242 |
ep += *ep; |
243 |
star: |
244 |
do { |
245 |
lp--; |
246 |
if (advance(lp, ep)) { |
247 |
explen += lp - alp; |
248 |
return(1); |
249 |
} |
250 |
} while (lp > curlp); |
251 |
return(0); |
252 |
|
253 |
case CBRC: |
254 |
if ((isalnum(*lp) || *lp == '_') && !(isalnum(lp[-1]) || lp[-1] == '_')) |
255 |
continue; |
256 |
return (0); |
257 |
|
258 |
case CLET: |
259 |
if (!isalnum(*lp) && *lp != '_') |
260 |
continue; |
261 |
return (0); |
262 |
|
263 |
default: |
264 |
fprintf(stderr, "RE botch\n"); |
265 |
} |
266 |
} |
267 |
|
268 |
static int |
269 |
cclass(set, c, af) |
270 |
register char *set; |
271 |
register c; |
272 |
int af; |
273 |
{ |
274 |
register n; |
275 |
|
276 |
if (c == 0) |
277 |
return(0); |
278 |
n = *set++; |
279 |
while (--n) |
280 |
if (n > 2 && set[1] == '-') { |
281 |
if (c >= set[0] && c <= set[2]) |
282 |
return (af); |
283 |
set += 3; |
284 |
n -= 2; |
285 |
} else |
286 |
if (*set++ == c) |
287 |
return(af); |
288 |
return(!af); |
289 |
} |