Commit | Line | Data |
---|---|---|
205a2d85 | 1 | #ifndef lint |
efef1041 | 2 | static char sccsid[] = "@(#)old.bin.grep.c 4.8 (Berkeley) %G%"; |
205a2d85 | 3 | #endif |
49d08d24 KM |
4 | |
5 | /* | |
6 | * grep -- print lines matching (or not matching) a pattern | |
7 | * | |
8 | * status returns: | |
9 | * 0 - ok, and some matches | |
10 | * 1 - ok, but no matches | |
11 | * 2 - some error | |
12 | */ | |
13 | ||
14 | #include <stdio.h> | |
15 | #include <ctype.h> | |
16 | ||
17 | #define CBRA 1 | |
18 | #define CCHR 2 | |
19 | #define CDOT 4 | |
20 | #define CCL 6 | |
21 | #define NCCL 8 | |
22 | #define CDOL 10 | |
23 | #define CEOF 11 | |
24 | #define CKET 12 | |
6a6b31b3 SL |
25 | #define CBRC 14 |
26 | #define CLET 15 | |
49d08d24 KM |
27 | #define CBACK 18 |
28 | ||
29 | #define STAR 01 | |
30 | ||
31 | #define LBSIZE BUFSIZ | |
32 | #define ESIZE 256 | |
33 | #define NBRA 9 | |
34 | ||
35 | char expbuf[ESIZE]; | |
36 | long lnum; | |
37 | char linebuf[LBSIZE+1]; | |
38 | char ybuf[ESIZE]; | |
39 | int bflag; | |
40 | int lflag; | |
41 | int nflag; | |
42 | int cflag; | |
43 | int vflag; | |
44 | int nfile; | |
45 | int hflag = 1; | |
7df1ddda | 46 | int oflag; |
49d08d24 KM |
47 | int sflag; |
48 | int yflag; | |
6a6b31b3 | 49 | int wflag; |
b6920e2a | 50 | int retcode = 0; |
49d08d24 KM |
51 | int circf; |
52 | int blkno; | |
53 | long tln; | |
54 | int nsucc; | |
55 | char *braslist[NBRA]; | |
56 | char *braelist[NBRA]; | |
57 | char bittab[] = { | |
58 | 1, | |
59 | 2, | |
60 | 4, | |
61 | 8, | |
62 | 16, | |
63 | 32, | |
64 | 64, | |
65 | 128 | |
66 | }; | |
67 | ||
68 | main(argc, argv) | |
09d77a2a KB |
69 | int argc; |
70 | char **argv; | |
49d08d24 | 71 | { |
09d77a2a KB |
72 | extern char *optarg; |
73 | extern int optind; | |
74 | int ch; | |
49d08d24 | 75 | |
09d77a2a KB |
76 | while ((ch = getopt(argc, argv, "iywohsvblcne")) != EOF) |
77 | switch((char)ch) { | |
6a6b31b3 | 78 | case 'i': |
49d08d24 KM |
79 | case 'y': |
80 | yflag++; | |
09d77a2a | 81 | break; |
6a6b31b3 SL |
82 | case 'w': |
83 | wflag++; | |
09d77a2a | 84 | break; |
7df1ddda KB |
85 | case 'o': |
86 | oflag++; | |
09d77a2a | 87 | break; |
49d08d24 KM |
88 | case 'h': |
89 | hflag = 0; | |
09d77a2a | 90 | break; |
49d08d24 KM |
91 | case 's': |
92 | sflag++; | |
09d77a2a | 93 | break; |
49d08d24 KM |
94 | case 'v': |
95 | vflag++; | |
09d77a2a | 96 | break; |
49d08d24 KM |
97 | case 'b': |
98 | bflag++; | |
09d77a2a | 99 | break; |
49d08d24 KM |
100 | case 'l': |
101 | lflag++; | |
09d77a2a | 102 | break; |
49d08d24 KM |
103 | case 'c': |
104 | cflag++; | |
09d77a2a | 105 | break; |
49d08d24 KM |
106 | case 'n': |
107 | nflag++; | |
09d77a2a | 108 | break; |
49d08d24 | 109 | case 'e': |
09d77a2a KB |
110 | argv += optind; |
111 | argc -= optind; | |
112 | *argv = optarg; | |
49d08d24 | 113 | goto out; |
09d77a2a | 114 | case '?': |
49d08d24 KM |
115 | default: |
116 | errexit("grep: unknown flag\n", (char *)NULL); | |
49d08d24 | 117 | } |
09d77a2a KB |
118 | argv += optind; |
119 | argc -= optind; | |
120 | ||
49d08d24 KM |
121 | out: |
122 | if (argc<=0) | |
123 | exit(2); | |
124 | if (yflag) { | |
125 | register char *p, *s; | |
126 | for (s = ybuf, p = *argv; *p; ) { | |
127 | if (*p == '\\') { | |
128 | *s++ = *p++; | |
129 | if (*p) | |
130 | *s++ = *p++; | |
131 | } else if (*p == '[') { | |
132 | while (*p != '\0' && *p != ']') | |
133 | *s++ = *p++; | |
134 | } else if (islower(*p)) { | |
135 | *s++ = '['; | |
136 | *s++ = toupper(*p); | |
137 | *s++ = *p++; | |
138 | *s++ = ']'; | |
139 | } else | |
140 | *s++ = *p++; | |
141 | if (s >= ybuf+ESIZE-5) | |
142 | errexit("grep: argument too long\n", (char *)NULL); | |
143 | } | |
144 | *s = '\0'; | |
145 | *argv = ybuf; | |
146 | } | |
147 | compile(*argv); | |
148 | nfile = --argc; | |
149 | if (argc<=0) { | |
150 | if (lflag) | |
151 | exit(1); | |
152 | execute((char *)NULL); | |
153 | } else while (--argc >= 0) { | |
154 | argv++; | |
155 | execute(*argv); | |
156 | } | |
b6920e2a | 157 | exit(retcode != 0 ? retcode : nsucc == 0); |
49d08d24 KM |
158 | } |
159 | ||
160 | compile(astr) | |
161 | char *astr; | |
162 | { | |
163 | register c; | |
164 | register char *ep, *sp; | |
165 | char *cstart; | |
166 | char *lastep; | |
167 | int cclcnt; | |
168 | char bracket[NBRA], *bracketp; | |
169 | int closed; | |
170 | char numbra; | |
171 | char neg; | |
172 | ||
173 | ep = expbuf; | |
174 | sp = astr; | |
175 | lastep = 0; | |
176 | bracketp = bracket; | |
177 | closed = numbra = 0; | |
178 | if (*sp == '^') { | |
179 | circf++; | |
180 | sp++; | |
181 | } | |
6a6b31b3 SL |
182 | if (wflag) |
183 | *ep++ = CBRC; | |
49d08d24 KM |
184 | for (;;) { |
185 | if (ep >= &expbuf[ESIZE]) | |
186 | goto cerror; | |
187 | if ((c = *sp++) != '*') | |
188 | lastep = ep; | |
189 | switch (c) { | |
190 | ||
191 | case '\0': | |
6a6b31b3 SL |
192 | if (wflag) |
193 | *ep++ = CLET; | |
49d08d24 KM |
194 | *ep++ = CEOF; |
195 | return; | |
196 | ||
197 | case '.': | |
198 | *ep++ = CDOT; | |
199 | continue; | |
200 | ||
201 | case '*': | |
6a6b31b3 SL |
202 | if (lastep==0 || *lastep==CBRA || *lastep==CKET || |
203 | *lastep == CBRC || *lastep == CLET) | |
49d08d24 KM |
204 | goto defchar; |
205 | *lastep |= STAR; | |
206 | continue; | |
207 | ||
208 | case '$': | |
209 | if (*sp != '\0') | |
210 | goto defchar; | |
211 | *ep++ = CDOL; | |
212 | continue; | |
213 | ||
214 | case '[': | |
215 | if(&ep[17] >= &expbuf[ESIZE]) | |
216 | goto cerror; | |
217 | *ep++ = CCL; | |
218 | neg = 0; | |
219 | if((c = *sp++) == '^') { | |
220 | neg = 1; | |
221 | c = *sp++; | |
222 | } | |
223 | cstart = sp; | |
224 | do { | |
225 | if (c=='\0') | |
226 | goto cerror; | |
227 | if (c=='-' && sp>cstart && *sp!=']') { | |
228 | for (c = sp[-2]; c<*sp; c++) | |
229 | ep[c>>3] |= bittab[c&07]; | |
230 | sp++; | |
231 | } | |
232 | ep[c>>3] |= bittab[c&07]; | |
233 | } while((c = *sp++) != ']'); | |
234 | if(neg) { | |
235 | for(cclcnt = 0; cclcnt < 16; cclcnt++) | |
236 | ep[cclcnt] ^= -1; | |
237 | ep[0] &= 0376; | |
238 | } | |
239 | ||
240 | ep += 16; | |
241 | ||
242 | continue; | |
243 | ||
244 | case '\\': | |
6a6b31b3 SL |
245 | if((c = *sp++) == 0) |
246 | goto cerror; | |
247 | if(c == '<') { | |
248 | *ep++ = CBRC; | |
249 | continue; | |
250 | } | |
251 | if(c == '>') { | |
252 | *ep++ = CLET; | |
253 | continue; | |
254 | } | |
255 | if(c == '(') { | |
49d08d24 KM |
256 | if(numbra >= NBRA) { |
257 | goto cerror; | |
258 | } | |
259 | *bracketp++ = numbra; | |
260 | *ep++ = CBRA; | |
261 | *ep++ = numbra++; | |
262 | continue; | |
263 | } | |
264 | if(c == ')') { | |
265 | if(bracketp <= bracket) { | |
266 | goto cerror; | |
267 | } | |
268 | *ep++ = CKET; | |
269 | *ep++ = *--bracketp; | |
270 | closed++; | |
271 | continue; | |
272 | } | |
273 | ||
274 | if(c >= '1' && c <= '9') { | |
275 | if((c -= '1') >= closed) | |
276 | goto cerror; | |
277 | *ep++ = CBACK; | |
278 | *ep++ = c; | |
279 | continue; | |
280 | } | |
281 | ||
282 | defchar: | |
283 | default: | |
284 | *ep++ = CCHR; | |
285 | *ep++ = c; | |
286 | } | |
287 | } | |
288 | cerror: | |
289 | errexit("grep: RE error\n", (char *)NULL); | |
290 | } | |
291 | ||
292 | execute(file) | |
293 | char *file; | |
294 | { | |
295 | register char *p1, *p2; | |
296 | register c; | |
297 | ||
298 | if (file) { | |
b6920e2a | 299 | if (freopen(file, "r", stdin) == NULL) { |
78d90aa3 | 300 | perror(file); |
b6920e2a EW |
301 | retcode = 2; |
302 | } | |
49d08d24 KM |
303 | } |
304 | lnum = 0; | |
305 | tln = 0; | |
306 | for (;;) { | |
307 | lnum++; | |
308 | p1 = linebuf; | |
309 | while ((c = getchar()) != '\n') { | |
310 | if (c == EOF) { | |
311 | if (cflag) { | |
312 | if (nfile>1) | |
313 | printf("%s:", file); | |
314 | printf("%D\n", tln); | |
315 | fflush(stdout); | |
316 | } | |
317 | return; | |
318 | } | |
319 | *p1++ = c; | |
320 | if (p1 >= &linebuf[LBSIZE-1]) | |
321 | break; | |
322 | } | |
323 | *p1++ = '\0'; | |
324 | p1 = linebuf; | |
325 | p2 = expbuf; | |
326 | if (circf) { | |
327 | if (advance(p1, p2)) | |
328 | goto found; | |
329 | goto nfound; | |
330 | } | |
331 | /* fast check for first character */ | |
332 | if (*p2==CCHR) { | |
333 | c = p2[1]; | |
334 | do { | |
335 | if (*p1!=c) | |
336 | continue; | |
337 | if (advance(p1, p2)) | |
338 | goto found; | |
339 | } while (*p1++); | |
340 | goto nfound; | |
341 | } | |
342 | /* regular algorithm */ | |
343 | do { | |
344 | if (advance(p1, p2)) | |
345 | goto found; | |
346 | } while (*p1++); | |
347 | nfound: | |
348 | if (vflag) | |
349 | succeed(file); | |
350 | continue; | |
351 | found: | |
352 | if (vflag==0) | |
353 | succeed(file); | |
354 | } | |
355 | } | |
356 | ||
357 | advance(lp, ep) | |
358 | register char *lp, *ep; | |
359 | { | |
360 | register char *curlp; | |
361 | char c; | |
362 | char *bbeg; | |
363 | int ct; | |
364 | ||
365 | for (;;) switch (*ep++) { | |
366 | ||
367 | case CCHR: | |
368 | if (*ep++ == *lp++) | |
369 | continue; | |
370 | return(0); | |
371 | ||
372 | case CDOT: | |
373 | if (*lp++) | |
374 | continue; | |
375 | return(0); | |
376 | ||
377 | case CDOL: | |
378 | if (*lp==0) | |
379 | continue; | |
380 | return(0); | |
381 | ||
382 | case CEOF: | |
383 | return(1); | |
384 | ||
385 | case CCL: | |
386 | c = *lp++ & 0177; | |
387 | if(ep[c>>3] & bittab[c & 07]) { | |
388 | ep += 16; | |
389 | continue; | |
390 | } | |
391 | return(0); | |
392 | case CBRA: | |
393 | braslist[*ep++] = lp; | |
394 | continue; | |
395 | ||
396 | case CKET: | |
397 | braelist[*ep++] = lp; | |
398 | continue; | |
399 | ||
400 | case CBACK: | |
401 | bbeg = braslist[*ep]; | |
402 | if (braelist[*ep]==0) | |
403 | return(0); | |
404 | ct = braelist[*ep++] - bbeg; | |
405 | if(ecmp(bbeg, lp, ct)) { | |
406 | lp += ct; | |
407 | continue; | |
408 | } | |
409 | return(0); | |
410 | ||
411 | case CBACK|STAR: | |
412 | bbeg = braslist[*ep]; | |
413 | if (braelist[*ep]==0) | |
414 | return(0); | |
415 | ct = braelist[*ep++] - bbeg; | |
416 | curlp = lp; | |
417 | while(ecmp(bbeg, lp, ct)) | |
418 | lp += ct; | |
419 | while(lp >= curlp) { | |
420 | if(advance(lp, ep)) return(1); | |
421 | lp -= ct; | |
422 | } | |
423 | return(0); | |
424 | ||
425 | ||
426 | case CDOT|STAR: | |
427 | curlp = lp; | |
428 | while (*lp++); | |
429 | goto star; | |
430 | ||
431 | case CCHR|STAR: | |
432 | curlp = lp; | |
433 | while (*lp++ == *ep); | |
434 | ep++; | |
435 | goto star; | |
436 | ||
437 | case CCL|STAR: | |
438 | curlp = lp; | |
439 | do { | |
440 | c = *lp++ & 0177; | |
441 | } while(ep[c>>3] & bittab[c & 07]); | |
442 | ep += 16; | |
443 | goto star; | |
444 | ||
445 | star: | |
446 | if(--lp == curlp) { | |
447 | continue; | |
448 | } | |
449 | ||
450 | if(*ep == CCHR) { | |
451 | c = ep[1]; | |
452 | do { | |
453 | if(*lp != c) | |
454 | continue; | |
455 | if(advance(lp, ep)) | |
456 | return(1); | |
457 | } while(lp-- > curlp); | |
458 | return(0); | |
459 | } | |
460 | ||
461 | do { | |
462 | if (advance(lp, ep)) | |
463 | return(1); | |
464 | } while (lp-- > curlp); | |
465 | return(0); | |
466 | ||
6a6b31b3 SL |
467 | case CBRC: |
468 | if (lp == expbuf) | |
469 | continue; | |
470 | #define uletter(c) (isalpha(c) || (c) == '_') | |
471 | if (uletter(*lp) || isdigit(*lp)) | |
472 | if (!uletter(lp[-1]) && !isdigit(lp[-1])) | |
473 | continue; | |
474 | return (0); | |
475 | ||
476 | case CLET: | |
477 | if (!uletter(*lp) && !isdigit(*lp)) | |
478 | continue; | |
479 | return (0); | |
480 | ||
49d08d24 KM |
481 | default: |
482 | errexit("grep RE botch\n", (char *)NULL); | |
483 | } | |
484 | } | |
485 | ||
486 | succeed(f) | |
487 | char *f; | |
488 | { | |
489 | nsucc = 1; | |
490 | if (sflag) | |
491 | return; | |
492 | if (cflag) { | |
493 | tln++; | |
494 | return; | |
495 | } | |
496 | if (lflag) { | |
497 | printf("%s\n", f); | |
498 | fflush(stdout); | |
499 | fseek(stdin, 0l, 2); | |
500 | return; | |
501 | } | |
7df1ddda | 502 | if (nfile > 1 && hflag || oflag) |
49d08d24 KM |
503 | printf("%s:", f); |
504 | if (bflag) | |
505 | printf("%u:", blkno); | |
506 | if (nflag) | |
507 | printf("%ld:", lnum); | |
508 | printf("%s\n", linebuf); | |
509 | fflush(stdout); | |
510 | } | |
511 | ||
512 | ecmp(a, b, count) | |
513 | char *a, *b; | |
514 | { | |
515 | register cc = count; | |
516 | while(cc--) | |
517 | if(*a++ != *b++) return(0); | |
518 | return(1); | |
519 | } | |
520 | ||
521 | errexit(s, f) | |
522 | char *s, *f; | |
523 | { | |
524 | fprintf(stderr, s, f); | |
525 | exit(2); | |
526 | } |