| 1 | /* Copyright (c) 1979 Regents of the University of California */ |
| 2 | #include <stdio.h> |
| 3 | /* |
| 4 | * grep -- print lines matching (or not matching) a pattern |
| 5 | */ |
| 6 | |
| 7 | #define CCHR 2 |
| 8 | #define CDOT 4 |
| 9 | #define CCL 6 |
| 10 | #define NCCL 8 |
| 11 | #define CDOL 10 |
| 12 | #define CEOF 11 |
| 13 | |
| 14 | #define CBRC 14 |
| 15 | #define CLET 15 |
| 16 | #define STAR 01 |
| 17 | |
| 18 | #define LBSIZE 256 |
| 19 | #define ESIZE 256 |
| 20 | |
| 21 | char expbuf[ESIZE]; |
| 22 | long lnum; |
| 23 | char linebuf[LBSIZE+1]; |
| 24 | int bflag; |
| 25 | int nflag; |
| 26 | int cflag; |
| 27 | int vflag; |
| 28 | int nfile; |
| 29 | int iflag; |
| 30 | int lflag; |
| 31 | int wflag; |
| 32 | int circf; |
| 33 | int blkno; |
| 34 | char ibuf[512]; |
| 35 | long tln; |
| 36 | |
| 37 | main(argc, argv) |
| 38 | char **argv; |
| 39 | { |
| 40 | char obuf[BUFSIZ]; |
| 41 | |
| 42 | setbuf(stdout, obuf); |
| 43 | while (--argc > 0 && (++argv)[0][0]=='-') { |
| 44 | char *cp = argv[0] + 1; |
| 45 | while (*cp) switch (*cp++) { |
| 46 | |
| 47 | case 'v': |
| 48 | vflag++; |
| 49 | continue; |
| 50 | |
| 51 | case 'b': |
| 52 | bflag++; |
| 53 | continue; |
| 54 | |
| 55 | case 'i': |
| 56 | iflag++; |
| 57 | continue; |
| 58 | |
| 59 | case 'l': |
| 60 | lflag++; |
| 61 | case 'c': |
| 62 | cflag++; |
| 63 | continue; |
| 64 | |
| 65 | case 'w': |
| 66 | wflag++; |
| 67 | continue; |
| 68 | |
| 69 | case 'n': |
| 70 | nflag++; |
| 71 | continue; |
| 72 | |
| 73 | default: |
| 74 | fprintf(stderr, "Unknown flag\n"); |
| 75 | continue; |
| 76 | } |
| 77 | } |
| 78 | if (argc<=0) |
| 79 | exit(2); |
| 80 | compile(*argv); |
| 81 | nfile = --argc; |
| 82 | if (argc<=0) |
| 83 | execute(0); |
| 84 | else while (--argc >= 0) { |
| 85 | argv++; |
| 86 | execute(*argv); |
| 87 | } |
| 88 | exit(0); |
| 89 | } |
| 90 | |
| 91 | compile(astr) |
| 92 | char *astr; |
| 93 | { |
| 94 | register c; |
| 95 | register char *ep, *sp; |
| 96 | char *lastep; |
| 97 | int cclcnt; |
| 98 | |
| 99 | ep = expbuf; |
| 100 | sp = astr; |
| 101 | if (*sp == '^') { |
| 102 | circf++; |
| 103 | sp++; |
| 104 | } |
| 105 | if (wflag) |
| 106 | *ep++ = CBRC; |
| 107 | for (;;) { |
| 108 | if (ep >= &expbuf[ESIZE]) |
| 109 | goto cerror; |
| 110 | if ((c = *sp++) != '*') |
| 111 | lastep = ep; |
| 112 | switch (c) { |
| 113 | |
| 114 | case '\0': |
| 115 | if (wflag) |
| 116 | *ep++ = CLET; |
| 117 | *ep++ = CEOF; |
| 118 | return; |
| 119 | |
| 120 | case '.': |
| 121 | *ep++ = CDOT; |
| 122 | continue; |
| 123 | |
| 124 | case '*': |
| 125 | if (lastep==0) |
| 126 | goto defchar; |
| 127 | *lastep =| STAR; |
| 128 | continue; |
| 129 | |
| 130 | case '$': |
| 131 | if (*sp != '\0') |
| 132 | goto defchar; |
| 133 | *ep++ = CDOL; |
| 134 | continue; |
| 135 | |
| 136 | case '[': |
| 137 | *ep++ = CCL; |
| 138 | *ep++ = 0; |
| 139 | cclcnt = 1; |
| 140 | if ((c = *sp++) == '^') { |
| 141 | c = *sp++; |
| 142 | ep[-2] = NCCL; |
| 143 | } |
| 144 | do { |
| 145 | *ep++ = c; |
| 146 | cclcnt++; |
| 147 | if (c=='\0' || ep >= &expbuf[ESIZE]) |
| 148 | goto cerror; |
| 149 | } while ((c = *sp++) != ']'); |
| 150 | lastep[1] = cclcnt; |
| 151 | continue; |
| 152 | |
| 153 | case '\\': |
| 154 | if ((c = *sp++) == '\0') |
| 155 | goto cerror; |
| 156 | if (c == '<') { |
| 157 | *ep++ = CBRC; |
| 158 | continue; |
| 159 | } |
| 160 | if (c == '>') { |
| 161 | *ep++ = CLET; |
| 162 | continue; |
| 163 | } |
| 164 | defchar: |
| 165 | default: |
| 166 | *ep++ = CCHR; |
| 167 | *ep++ = c; |
| 168 | } |
| 169 | } |
| 170 | cerror: |
| 171 | fprintf(stderr, "RE error\n"); |
| 172 | } |
| 173 | |
| 174 | same(a, b) |
| 175 | register int a, b; |
| 176 | { |
| 177 | |
| 178 | return (a == b || iflag && (a ^ b) == ' ' && letter(a) == letter(b)); |
| 179 | } |
| 180 | |
| 181 | letter(c) |
| 182 | register int c; |
| 183 | { |
| 184 | |
| 185 | if (c >= 'a' && c <= 'z') |
| 186 | return (c); |
| 187 | if (c >= 'A' && c <= 'Z') |
| 188 | return (c + 'a' - 'A'); |
| 189 | return (0); |
| 190 | } |
| 191 | |
| 192 | execute(file) |
| 193 | { |
| 194 | register char *p1, *p2; |
| 195 | register c; |
| 196 | int f; |
| 197 | char *ebp, *cbp; |
| 198 | |
| 199 | if (file) { |
| 200 | if ((f = open(file, 0)) < 0) { |
| 201 | fprintf(stderr, "Can't open %s\n", file); |
| 202 | } |
| 203 | } else |
| 204 | f = 0; |
| 205 | ebp = ibuf; |
| 206 | cbp = ibuf; |
| 207 | lnum = 0; |
| 208 | tln = 0; |
| 209 | blkno = -1; |
| 210 | for (;;) { |
| 211 | lnum++; |
| 212 | if((lnum&0377) == 0) |
| 213 | fflush(stdout); |
| 214 | p1 = linebuf; |
| 215 | p2 = cbp; |
| 216 | for (;;) { |
| 217 | if (p2 >= ebp) { |
| 218 | if ((c = read(f, ibuf, 512)) <= 0) { |
| 219 | close(f); |
| 220 | if (cflag) { |
| 221 | if (lflag) { |
| 222 | if (tln) |
| 223 | printf("%s\n", file); |
| 224 | } else { |
| 225 | if (nfile > 1) |
| 226 | printf("%s:", file); |
| 227 | printf("%ld\n", tln); |
| 228 | } |
| 229 | } |
| 230 | return; |
| 231 | } |
| 232 | blkno++; |
| 233 | p2 = ibuf; |
| 234 | ebp = ibuf+c; |
| 235 | } |
| 236 | if ((c = *p2++) == '\n') |
| 237 | break; |
| 238 | if(c) |
| 239 | if (p1 < &linebuf[LBSIZE-1]) |
| 240 | *p1++ = c; |
| 241 | } |
| 242 | *p1++ = 0; |
| 243 | cbp = p2; |
| 244 | p1 = linebuf; |
| 245 | p2 = expbuf; |
| 246 | if (circf) { |
| 247 | if (advance(p1, p2)) |
| 248 | goto found; |
| 249 | goto nfound; |
| 250 | } |
| 251 | /* fast check for first character */ |
| 252 | if (*p2==CCHR) { |
| 253 | c = p2[1]; |
| 254 | do { |
| 255 | if (*p1!=c && (!iflag || (c ^ *p1) != ' ' |
| 256 | || letter(c) != letter(*p1))) |
| 257 | continue; |
| 258 | if (advance(p1, p2)) |
| 259 | goto found; |
| 260 | } while (*p1++); |
| 261 | goto nfound; |
| 262 | } |
| 263 | /* regular algorithm */ |
| 264 | do { |
| 265 | if (advance(p1, p2)) |
| 266 | goto found; |
| 267 | } while (*p1++); |
| 268 | nfound: |
| 269 | if (vflag) |
| 270 | succeed(file); |
| 271 | continue; |
| 272 | found: |
| 273 | if (vflag==0) |
| 274 | succeed(file); |
| 275 | } |
| 276 | } |
| 277 | |
| 278 | advance(alp, aep) |
| 279 | { |
| 280 | register char *lp, *ep, *curlp; |
| 281 | char *nextep; |
| 282 | |
| 283 | lp = alp; |
| 284 | ep = aep; |
| 285 | for (;;) switch (*ep++) { |
| 286 | |
| 287 | case CCHR: |
| 288 | if (!same(*ep, *lp)) |
| 289 | return (0); |
| 290 | ep++, lp++; |
| 291 | continue; |
| 292 | |
| 293 | case CDOT: |
| 294 | if (*lp++) |
| 295 | continue; |
| 296 | return(0); |
| 297 | |
| 298 | case CDOL: |
| 299 | if (*lp==0) |
| 300 | continue; |
| 301 | return(0); |
| 302 | |
| 303 | case CEOF: |
| 304 | return(1); |
| 305 | |
| 306 | case CCL: |
| 307 | if (cclass(ep, *lp++, 1)) { |
| 308 | ep =+ *ep; |
| 309 | continue; |
| 310 | } |
| 311 | return(0); |
| 312 | |
| 313 | case NCCL: |
| 314 | if (cclass(ep, *lp++, 0)) { |
| 315 | ep =+ *ep; |
| 316 | continue; |
| 317 | } |
| 318 | return(0); |
| 319 | |
| 320 | case CDOT|STAR: |
| 321 | curlp = lp; |
| 322 | while (*lp++); |
| 323 | goto star; |
| 324 | |
| 325 | case CCHR|STAR: |
| 326 | curlp = lp; |
| 327 | while (same(*lp, *ep)) |
| 328 | lp++; |
| 329 | lp++; |
| 330 | ep++; |
| 331 | goto star; |
| 332 | |
| 333 | case CCL|STAR: |
| 334 | case NCCL|STAR: |
| 335 | curlp = lp; |
| 336 | while (cclass(ep, *lp++, ep[-1]==(CCL|STAR))); |
| 337 | ep =+ *ep; |
| 338 | goto star; |
| 339 | |
| 340 | star: |
| 341 | do { |
| 342 | lp--; |
| 343 | if (advance(lp, ep)) |
| 344 | return(1); |
| 345 | } while (lp > curlp); |
| 346 | return(0); |
| 347 | |
| 348 | case CBRC: |
| 349 | if (lp == expbuf) |
| 350 | continue; |
| 351 | #define uletter(c) (letter(c) || c == '_') |
| 352 | if (uletter(*lp) && !uletter(lp[-1]) && !digit(lp[-1])) |
| 353 | continue; |
| 354 | return (0); |
| 355 | |
| 356 | case CLET: |
| 357 | if (!uletter(*lp) && !digit(*lp)) |
| 358 | continue; |
| 359 | return (0); |
| 360 | |
| 361 | default: |
| 362 | fprintf(stderr, "RE botch\n"); |
| 363 | } |
| 364 | } |
| 365 | |
| 366 | cclass(aset, ac, af) |
| 367 | { |
| 368 | register char *set, c; |
| 369 | register n; |
| 370 | |
| 371 | set = aset; |
| 372 | if ((c = ac) == 0) |
| 373 | return(0); |
| 374 | n = *set++; |
| 375 | while (--n) |
| 376 | if (n > 2 && set[1] == '-') { |
| 377 | if (c >= (set[0] & 0177) && c <= (set[2] & 0177)) |
| 378 | return (af); |
| 379 | set =+ 3; |
| 380 | n -= 2; |
| 381 | } else |
| 382 | if ((*set++ & 0177) == c) |
| 383 | return(af); |
| 384 | return(!af); |
| 385 | } |
| 386 | |
| 387 | succeed(f) |
| 388 | { |
| 389 | if (cflag) { |
| 390 | tln++; |
| 391 | return; |
| 392 | } |
| 393 | if (nfile > 1) |
| 394 | printf("%s:", f); |
| 395 | if (bflag) |
| 396 | printf("%l:", blkno); |
| 397 | if (nflag) |
| 398 | printf("%ld:", lnum); |
| 399 | printf("%s\n", linebuf); |
| 400 | } |
| 401 | |
| 402 | digit(c) |
| 403 | char c; |
| 404 | { |
| 405 | return (c>='0' && c<='9'); |
| 406 | } |