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