Commit | Line | Data |
---|---|---|
ca7072d6 KM |
1 | /* old.ucb.grep.c 4.1 82/05/07 */ |
2 | ||
3 | #include <stdio.h> | |
4 | /* | |
5 | * grep -- print lines matching (or not matching) a pattern | |
6 | */ | |
7 | ||
8 | #define CCHR 2 | |
9 | #define CDOT 4 | |
10 | #define CCL 6 | |
11 | #define NCCL 8 | |
12 | #define CDOL 10 | |
13 | #define CEOF 11 | |
14 | ||
15 | #define CBRC 14 | |
16 | #define CLET 15 | |
17 | #define STAR 01 | |
18 | ||
19 | #define LBSIZE BUFSIZ | |
20 | #define ESIZE 256 | |
21 | ||
22 | char expbuf[ESIZE]; | |
23 | long lnum; | |
24 | char linebuf[LBSIZE+1]; | |
25 | int bflag; | |
26 | int nflag; | |
27 | int cflag; | |
28 | int vflag; | |
29 | int nfile; | |
30 | int iflag; | |
31 | int lflag; | |
32 | int wflag; | |
33 | int sflag; | |
34 | int nsucc; | |
35 | int circf; | |
36 | int blkno; | |
37 | char ibuf[BUFSIZ]; | |
38 | long tln; | |
39 | ||
40 | main(argc, argv) | |
41 | char **argv; | |
42 | { | |
43 | char obuf[BUFSIZ]; | |
44 | ||
45 | setbuf(stdout, obuf); | |
46 | while (--argc > 0 && (++argv)[0][0]=='-') { | |
47 | char *cp = argv[0] + 1; | |
48 | while (*cp) switch (*cp++) { | |
49 | ||
50 | case 'v': | |
51 | vflag++; | |
52 | continue; | |
53 | ||
54 | case 'b': | |
55 | bflag++; | |
56 | continue; | |
57 | ||
58 | case 'i': | |
59 | case 'y': /* -y for compatibility with btl grep */ | |
60 | iflag++; | |
61 | continue; | |
62 | ||
63 | case 'l': | |
64 | lflag++; | |
65 | case 'c': | |
66 | cflag++; | |
67 | continue; | |
68 | ||
69 | case 'w': | |
70 | wflag++; | |
71 | continue; | |
72 | ||
73 | case 's': | |
74 | sflag++; | |
75 | continue; | |
76 | ||
77 | case 'n': | |
78 | nflag++; | |
79 | continue; | |
80 | ||
81 | case 'e': | |
82 | --argc; | |
83 | ++argv; | |
84 | goto out; | |
85 | ||
86 | default: | |
87 | fprintf(stderr, "Unknown flag\n"); | |
88 | continue; | |
89 | } | |
90 | } | |
91 | out: | |
92 | if (argc<=0) | |
93 | exit(2); | |
94 | compile(*argv); | |
95 | nfile = --argc; | |
96 | if (argc<=0) { | |
97 | if (lflag) | |
98 | exit(1); | |
99 | execute(0); | |
100 | } | |
101 | else while (--argc >= 0) { | |
102 | argv++; | |
103 | execute(*argv); | |
104 | } | |
105 | exit(nsucc == 0); | |
106 | } | |
107 | ||
108 | compile(astr) | |
109 | char *astr; | |
110 | { | |
111 | register c; | |
112 | register char *ep, *sp; | |
113 | char *lastep; | |
114 | int cclcnt; | |
115 | ||
116 | ep = expbuf; | |
117 | sp = astr; | |
118 | if (*sp == '^') { | |
119 | circf++; | |
120 | sp++; | |
121 | } | |
122 | if (wflag) | |
123 | *ep++ = CBRC; | |
124 | for (;;) { | |
125 | if (ep >= &expbuf[ESIZE]) | |
126 | goto cerror; | |
127 | if ((c = *sp++) != '*') | |
128 | lastep = ep; | |
129 | switch (c) { | |
130 | ||
131 | case '\0': | |
132 | if (wflag) | |
133 | *ep++ = CLET; | |
134 | *ep++ = CEOF; | |
135 | return; | |
136 | ||
137 | case '.': | |
138 | *ep++ = CDOT; | |
139 | continue; | |
140 | ||
141 | case '*': | |
142 | if (lastep==0) | |
143 | goto defchar; | |
144 | *lastep |= STAR; | |
145 | continue; | |
146 | ||
147 | case '$': | |
148 | if (*sp != '\0') | |
149 | goto defchar; | |
150 | *ep++ = CDOL; | |
151 | continue; | |
152 | ||
153 | case '[': | |
154 | *ep++ = CCL; | |
155 | *ep++ = 0; | |
156 | cclcnt = 1; | |
157 | if ((c = *sp++) == '^') { | |
158 | c = *sp++; | |
159 | ep[-2] = NCCL; | |
160 | } | |
161 | do { | |
162 | *ep++ = c; | |
163 | cclcnt++; | |
164 | if (c=='\0' || ep >= &expbuf[ESIZE]) | |
165 | goto cerror; | |
166 | } while ((c = *sp++) != ']'); | |
167 | lastep[1] = cclcnt; | |
168 | continue; | |
169 | ||
170 | case '\\': | |
171 | if ((c = *sp++) == '\0') | |
172 | goto cerror; | |
173 | if (c == '<') { | |
174 | *ep++ = CBRC; | |
175 | continue; | |
176 | } | |
177 | if (c == '>') { | |
178 | *ep++ = CLET; | |
179 | continue; | |
180 | } | |
181 | defchar: | |
182 | default: | |
183 | *ep++ = CCHR; | |
184 | *ep++ = c; | |
185 | } | |
186 | } | |
187 | cerror: | |
188 | fprintf(stderr, "RE error\n"); | |
189 | } | |
190 | ||
191 | same(a, b) | |
192 | register int a, b; | |
193 | { | |
194 | ||
195 | return (a == b || iflag && (a ^ b) == ' ' && letter(a) == letter(b)); | |
196 | } | |
197 | ||
198 | letter(c) | |
199 | register int c; | |
200 | { | |
201 | ||
202 | if (c >= 'a' && c <= 'z') | |
203 | return (c); | |
204 | if (c >= 'A' && c <= 'Z') | |
205 | return (c + 'a' - 'A'); | |
206 | return (0); | |
207 | } | |
208 | ||
209 | execute(file) | |
210 | { | |
211 | register char *p1, *p2; | |
212 | register c; | |
213 | int f; | |
214 | char *ebp, *cbp; | |
215 | ||
216 | if (file) { | |
217 | if ((f = open(file, 0)) < 0) { | |
218 | fprintf(stderr, "Can't open %s\n", file); | |
219 | } | |
220 | } else | |
221 | f = 0; | |
222 | ebp = ibuf; | |
223 | cbp = ibuf; | |
224 | lnum = 0; | |
225 | tln = 0; | |
226 | blkno = -1; | |
227 | for (;;) { | |
228 | lnum++; | |
229 | if((lnum&0377) == 0) | |
230 | fflush(stdout); | |
231 | p1 = linebuf; | |
232 | p2 = cbp; | |
233 | for (;;) { | |
234 | if (p2 >= ebp) { | |
235 | if ((c = read(f, ibuf, BUFSIZ)) <= 0) { | |
236 | close(f); | |
237 | if (cflag) { | |
238 | if (lflag) { | |
239 | if (tln) | |
240 | printf("%s\n", file); | |
241 | } else { | |
242 | if (nfile > 1) | |
243 | printf("%s:", file); | |
244 | printf("%ld\n", tln); | |
245 | } | |
246 | } | |
247 | return; | |
248 | } | |
249 | blkno++; | |
250 | p2 = ibuf; | |
251 | ebp = ibuf+c; | |
252 | } | |
253 | if ((c = *p2++) == '\n') | |
254 | break; | |
255 | if(c) | |
256 | if (p1 < &linebuf[LBSIZE-1]) | |
257 | *p1++ = c; | |
258 | } | |
259 | *p1++ = 0; | |
260 | cbp = p2; | |
261 | p1 = linebuf; | |
262 | p2 = expbuf; | |
263 | if (circf) { | |
264 | if (advance(p1, p2)) | |
265 | goto found; | |
266 | goto nfound; | |
267 | } | |
268 | /* fast check for first character */ | |
269 | if (*p2==CCHR) { | |
270 | c = p2[1]; | |
271 | do { | |
272 | if (*p1!=c && (!iflag || (c ^ *p1) != ' ' | |
273 | || letter(c) != letter(*p1))) | |
274 | continue; | |
275 | if (advance(p1, p2)) | |
276 | goto found; | |
277 | } while (*p1++); | |
278 | goto nfound; | |
279 | } | |
280 | /* regular algorithm */ | |
281 | do { | |
282 | if (advance(p1, p2)) | |
283 | goto found; | |
284 | } while (*p1++); | |
285 | nfound: | |
286 | if (vflag) | |
287 | succeed(file); | |
288 | continue; | |
289 | found: | |
290 | if (vflag==0) | |
291 | succeed(file); | |
292 | } | |
293 | } | |
294 | ||
295 | advance(alp, aep) | |
296 | char *alp, *aep; | |
297 | { | |
298 | register char *lp, *ep, *curlp; | |
299 | char *nextep; | |
300 | ||
301 | lp = alp; | |
302 | ep = aep; | |
303 | for (;;) switch (*ep++) { | |
304 | ||
305 | case CCHR: | |
306 | if (!same(*ep, *lp)) | |
307 | return (0); | |
308 | ep++, lp++; | |
309 | continue; | |
310 | ||
311 | case CDOT: | |
312 | if (*lp++) | |
313 | continue; | |
314 | return(0); | |
315 | ||
316 | case CDOL: | |
317 | if (*lp==0) | |
318 | continue; | |
319 | return(0); | |
320 | ||
321 | case CEOF: | |
322 | return(1); | |
323 | ||
324 | case CCL: | |
325 | if (cclass(ep, *lp++, 1)) { | |
326 | ep += *ep; | |
327 | continue; | |
328 | } | |
329 | return(0); | |
330 | ||
331 | case NCCL: | |
332 | if (cclass(ep, *lp++, 0)) { | |
333 | ep += *ep; | |
334 | continue; | |
335 | } | |
336 | return(0); | |
337 | ||
338 | case CDOT|STAR: | |
339 | curlp = lp; | |
340 | while (*lp++); | |
341 | goto star; | |
342 | ||
343 | case CCHR|STAR: | |
344 | curlp = lp; | |
345 | while (same(*lp, *ep)) | |
346 | lp++; | |
347 | lp++; | |
348 | ep++; | |
349 | goto star; | |
350 | ||
351 | case CCL|STAR: | |
352 | case NCCL|STAR: | |
353 | curlp = lp; | |
354 | while (cclass(ep, *lp++, ep[-1]==(CCL|STAR))); | |
355 | ep += *ep; | |
356 | goto star; | |
357 | ||
358 | star: | |
359 | do { | |
360 | lp--; | |
361 | if (advance(lp, ep)) | |
362 | return(1); | |
363 | } while (lp > curlp); | |
364 | return(0); | |
365 | ||
366 | case CBRC: | |
367 | if (lp == expbuf) | |
368 | continue; | |
369 | #define uletter(c) (letter(c) || c == '_') | |
370 | if ( ( uletter(*lp) || digit ( * lp ) ) && !uletter(lp[-1]) && !digit(lp[-1])) | |
371 | continue; | |
372 | return (0); | |
373 | ||
374 | case CLET: | |
375 | if (!uletter(*lp) && !digit(*lp)) | |
376 | continue; | |
377 | return (0); | |
378 | ||
379 | default: | |
380 | fprintf(stderr, "RE botch\n"); | |
381 | } | |
382 | } | |
383 | ||
384 | cclass(aset, ac, af) | |
385 | char *aset; | |
386 | { | |
387 | register char *set, c; | |
388 | register n; | |
389 | ||
390 | set = aset; | |
391 | if ((c = ac) == 0) | |
392 | return(0); | |
393 | n = *set++; | |
394 | while (--n) | |
395 | if (n > 2 && set[1] == '-') { | |
396 | if (c >= (set[0] & 0177) && c <= (set[2] & 0177)) | |
397 | return (af); | |
398 | set += 3; | |
399 | n -= 2; | |
400 | } else | |
401 | if ((*set++ & 0177) == c) | |
402 | return(af); | |
403 | return(!af); | |
404 | } | |
405 | ||
406 | succeed(f) | |
407 | { | |
408 | nsucc = 1; | |
409 | if (sflag) | |
410 | return; | |
411 | if (cflag) { | |
412 | tln++; | |
413 | return; | |
414 | } | |
415 | if (nfile > 1) | |
416 | printf("%s:", f); | |
417 | if (bflag) | |
418 | printf("%d:", blkno); | |
419 | if (nflag) | |
420 | printf("%ld:", lnum); | |
421 | printf("%s\n", linebuf); | |
422 | } | |
423 | ||
424 | digit(c) | |
425 | char c; | |
426 | { | |
427 | return (c>='0' && c<='9'); | |
428 | } |