docs/cut

annotate code/cut.c__freebsd.2012-11-24 @ 38:ec76f8926598

find changesets by author, revision, files, or words in the commit message
clarify a statement Thanks to Francesc for the suggestion.
author markus schnalke <meillo@marmaro.de>
date Tue, 06 Oct 2015 10:43:26 +0200
parents
children
rev   line source
meillo@14 1 /*
meillo@14 2 * Copyright (c) 1989, 1993
meillo@14 3 * The Regents of the University of California. All rights reserved.
meillo@14 4 *
meillo@14 5 * This code is derived from software contributed to Berkeley by
meillo@14 6 * Adam S. Moskowitz of Menlo Consulting and Marciano Pitargue.
meillo@14 7 *
meillo@14 8 * Redistribution and use in source and binary forms, with or without
meillo@14 9 * modification, are permitted provided that the following conditions
meillo@14 10 * are met:
meillo@14 11 * 1. Redistributions of source code must retain the above copyright
meillo@14 12 * notice, this list of conditions and the following disclaimer.
meillo@14 13 * 2. Redistributions in binary form must reproduce the above copyright
meillo@14 14 * notice, this list of conditions and the following disclaimer in the
meillo@14 15 * documentation and/or other materials provided with the distribution.
meillo@14 16 * 4. Neither the name of the University nor the names of its contributors
meillo@14 17 * may be used to endorse or promote products derived from this software
meillo@14 18 * without specific prior written permission.
meillo@14 19 *
meillo@14 20 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
meillo@14 21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
meillo@14 22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
meillo@14 23 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
meillo@14 24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
meillo@14 25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
meillo@14 26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
meillo@14 27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
meillo@14 28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
meillo@14 29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
meillo@14 30 * SUCH DAMAGE.
meillo@14 31 */
meillo@14 32
meillo@14 33 #ifndef lint
meillo@14 34 static const char copyright[] =
meillo@14 35 "@(#) Copyright (c) 1989, 1993\n\
meillo@14 36 The Regents of the University of California. All rights reserved.\n";
meillo@14 37 static const char sccsid[] = "@(#)cut.c 8.3 (Berkeley) 5/4/95";
meillo@14 38 #endif /* not lint */
meillo@14 39 #include <sys/cdefs.h>
meillo@14 40 __FBSDID("$FreeBSD$");
meillo@14 41
meillo@14 42 #include <ctype.h>
meillo@14 43 #include <err.h>
meillo@14 44 #include <errno.h>
meillo@14 45 #include <limits.h>
meillo@14 46 #include <locale.h>
meillo@14 47 #include <stdio.h>
meillo@14 48 #include <stdlib.h>
meillo@14 49 #include <string.h>
meillo@14 50 #include <unistd.h>
meillo@14 51 #include <wchar.h>
meillo@14 52
meillo@14 53 static int bflag;
meillo@14 54 static int cflag;
meillo@14 55 static wchar_t dchar;
meillo@14 56 static char dcharmb[MB_LEN_MAX + 1];
meillo@14 57 static int dflag;
meillo@14 58 static int fflag;
meillo@14 59 static int nflag;
meillo@14 60 static int sflag;
meillo@14 61 static int wflag;
meillo@14 62
meillo@14 63 static size_t autostart, autostop, maxval;
meillo@14 64 static char * positions;
meillo@14 65
meillo@14 66 static int b_cut(FILE *, const char *);
meillo@14 67 static int b_n_cut(FILE *, const char *);
meillo@14 68 static int c_cut(FILE *, const char *);
meillo@14 69 static int f_cut(FILE *, const char *);
meillo@14 70 static void get_list(char *);
meillo@14 71 static int is_delim(wchar_t);
meillo@14 72 static void needpos(size_t);
meillo@14 73 static void usage(void);
meillo@14 74
meillo@14 75 int
meillo@14 76 main(int argc, char *argv[])
meillo@14 77 {
meillo@14 78 FILE *fp;
meillo@14 79 int (*fcn)(FILE *, const char *);
meillo@14 80 int ch, rval;
meillo@14 81 size_t n;
meillo@14 82
meillo@14 83 setlocale(LC_ALL, "");
meillo@14 84
meillo@14 85 fcn = NULL;
meillo@14 86 dchar = '\t'; /* default delimiter is \t */
meillo@14 87 strcpy(dcharmb, "\t");
meillo@14 88
meillo@14 89 while ((ch = getopt(argc, argv, "b:c:d:f:snw")) != -1)
meillo@14 90 switch(ch) {
meillo@14 91 case 'b':
meillo@14 92 get_list(optarg);
meillo@14 93 bflag = 1;
meillo@14 94 break;
meillo@14 95 case 'c':
meillo@14 96 get_list(optarg);
meillo@14 97 cflag = 1;
meillo@14 98 break;
meillo@14 99 case 'd':
meillo@14 100 n = mbrtowc(&dchar, optarg, MB_LEN_MAX, NULL);
meillo@14 101 if (dchar == '\0' || n != strlen(optarg))
meillo@14 102 errx(1, "bad delimiter");
meillo@14 103 strcpy(dcharmb, optarg);
meillo@14 104 dflag = 1;
meillo@14 105 break;
meillo@14 106 case 'f':
meillo@14 107 get_list(optarg);
meillo@14 108 fflag = 1;
meillo@14 109 break;
meillo@14 110 case 's':
meillo@14 111 sflag = 1;
meillo@14 112 break;
meillo@14 113 case 'n':
meillo@14 114 nflag = 1;
meillo@14 115 break;
meillo@14 116 case 'w':
meillo@14 117 wflag = 1;
meillo@14 118 break;
meillo@14 119 case '?':
meillo@14 120 default:
meillo@14 121 usage();
meillo@14 122 }
meillo@14 123 argc -= optind;
meillo@14 124 argv += optind;
meillo@14 125
meillo@14 126 if (fflag) {
meillo@14 127 if (bflag || cflag || nflag || (wflag && dflag))
meillo@14 128 usage();
meillo@14 129 } else if (!(bflag || cflag) || dflag || sflag || wflag)
meillo@14 130 usage();
meillo@14 131 else if (!bflag && nflag)
meillo@14 132 usage();
meillo@14 133
meillo@14 134 if (fflag)
meillo@14 135 fcn = f_cut;
meillo@14 136 else if (cflag)
meillo@14 137 fcn = MB_CUR_MAX > 1 ? c_cut : b_cut;
meillo@14 138 else if (bflag)
meillo@14 139 fcn = nflag && MB_CUR_MAX > 1 ? b_n_cut : b_cut;
meillo@14 140
meillo@14 141 rval = 0;
meillo@14 142 if (*argv)
meillo@14 143 for (; *argv; ++argv) {
meillo@14 144 if (strcmp(*argv, "-") == 0)
meillo@14 145 rval |= fcn(stdin, "stdin");
meillo@14 146 else {
meillo@14 147 if (!(fp = fopen(*argv, "r"))) {
meillo@14 148 warn("%s", *argv);
meillo@14 149 rval = 1;
meillo@14 150 continue;
meillo@14 151 }
meillo@14 152 fcn(fp, *argv);
meillo@14 153 (void)fclose(fp);
meillo@14 154 }
meillo@14 155 }
meillo@14 156 else
meillo@14 157 rval = fcn(stdin, "stdin");
meillo@14 158 exit(rval);
meillo@14 159 }
meillo@14 160
meillo@14 161 static void
meillo@14 162 get_list(char *list)
meillo@14 163 {
meillo@14 164 size_t setautostart, start, stop;
meillo@14 165 char *pos;
meillo@14 166 char *p;
meillo@14 167
meillo@14 168 /*
meillo@14 169 * set a byte in the positions array to indicate if a field or
meillo@14 170 * column is to be selected; use +1, it's 1-based, not 0-based.
meillo@14 171 * Numbers and number ranges may be overlapping, repeated, and in
meillo@14 172 * any order. We handle "-3-5" although there's no real reason to.
meillo@14 173 */
meillo@14 174 for (; (p = strsep(&list, ", \t")) != NULL;) {
meillo@14 175 setautostart = start = stop = 0;
meillo@14 176 if (*p == '-') {
meillo@14 177 ++p;
meillo@14 178 setautostart = 1;
meillo@14 179 }
meillo@14 180 if (isdigit((unsigned char)*p)) {
meillo@14 181 start = stop = strtol(p, &p, 10);
meillo@14 182 if (setautostart && start > autostart)
meillo@14 183 autostart = start;
meillo@14 184 }
meillo@14 185 if (*p == '-') {
meillo@14 186 if (isdigit((unsigned char)p[1]))
meillo@14 187 stop = strtol(p + 1, &p, 10);
meillo@14 188 if (*p == '-') {
meillo@14 189 ++p;
meillo@14 190 if (!autostop || autostop > stop)
meillo@14 191 autostop = stop;
meillo@14 192 }
meillo@14 193 }
meillo@14 194 if (*p)
meillo@14 195 errx(1, "[-bcf] list: illegal list value");
meillo@14 196 if (!stop || !start)
meillo@14 197 errx(1, "[-bcf] list: values may not include zero");
meillo@14 198 if (maxval < stop) {
meillo@14 199 maxval = stop;
meillo@14 200 needpos(maxval + 1);
meillo@14 201 }
meillo@14 202 for (pos = positions + start; start++ <= stop; *pos++ = 1);
meillo@14 203 }
meillo@14 204
meillo@14 205 /* overlapping ranges */
meillo@14 206 if (autostop && maxval > autostop) {
meillo@14 207 maxval = autostop;
meillo@14 208 needpos(maxval + 1);
meillo@14 209 }
meillo@14 210
meillo@14 211 /* set autostart */
meillo@14 212 if (autostart)
meillo@14 213 memset(positions + 1, '1', autostart);
meillo@14 214 }
meillo@14 215
meillo@14 216 static void
meillo@14 217 needpos(size_t n)
meillo@14 218 {
meillo@14 219 static size_t npos;
meillo@14 220 size_t oldnpos;
meillo@14 221
meillo@14 222 /* Grow the positions array to at least the specified size. */
meillo@14 223 if (n > npos) {
meillo@14 224 oldnpos = npos;
meillo@14 225 if (npos == 0)
meillo@14 226 npos = n;
meillo@14 227 while (n > npos)
meillo@14 228 npos *= 2;
meillo@14 229 if ((positions = realloc(positions, npos)) == NULL)
meillo@14 230 err(1, "realloc");
meillo@14 231 memset((char *)positions + oldnpos, 0, npos - oldnpos);
meillo@14 232 }
meillo@14 233 }
meillo@14 234
meillo@14 235 static int
meillo@14 236 b_cut(FILE *fp, const char *fname __unused)
meillo@14 237 {
meillo@14 238 int ch, col;
meillo@14 239 char *pos;
meillo@14 240
meillo@14 241 ch = 0;
meillo@14 242 for (;;) {
meillo@14 243 pos = positions + 1;
meillo@14 244 for (col = maxval; col; --col) {
meillo@14 245 if ((ch = getc(fp)) == EOF)
meillo@14 246 return (0);
meillo@14 247 if (ch == '\n')
meillo@14 248 break;
meillo@14 249 if (*pos++)
meillo@14 250 (void)putchar(ch);
meillo@14 251 }
meillo@14 252 if (ch != '\n') {
meillo@14 253 if (autostop)
meillo@14 254 while ((ch = getc(fp)) != EOF && ch != '\n')
meillo@14 255 (void)putchar(ch);
meillo@14 256 else
meillo@14 257 while ((ch = getc(fp)) != EOF && ch != '\n');
meillo@14 258 }
meillo@14 259 (void)putchar('\n');
meillo@14 260 }
meillo@14 261 return (0);
meillo@14 262 }
meillo@14 263
meillo@14 264 /*
meillo@14 265 * Cut based on byte positions, taking care not to split multibyte characters.
meillo@14 266 * Although this function also handles the case where -n is not specified,
meillo@14 267 * b_cut() ought to be much faster.
meillo@14 268 */
meillo@14 269 static int
meillo@14 270 b_n_cut(FILE *fp, const char *fname)
meillo@14 271 {
meillo@14 272 size_t col, i, lbuflen;
meillo@14 273 char *lbuf;
meillo@14 274 int canwrite, clen, warned;
meillo@14 275 mbstate_t mbs;
meillo@14 276
meillo@14 277 memset(&mbs, 0, sizeof(mbs));
meillo@14 278 warned = 0;
meillo@14 279 while ((lbuf = fgetln(fp, &lbuflen)) != NULL) {
meillo@14 280 for (col = 0; lbuflen > 0; col += clen) {
meillo@14 281 if ((clen = mbrlen(lbuf, lbuflen, &mbs)) < 0) {
meillo@14 282 if (!warned) {
meillo@14 283 warn("%s", fname);
meillo@14 284 warned = 1;
meillo@14 285 }
meillo@14 286 memset(&mbs, 0, sizeof(mbs));
meillo@14 287 clen = 1;
meillo@14 288 }
meillo@14 289 if (clen == 0 || *lbuf == '\n')
meillo@14 290 break;
meillo@14 291 if (col < maxval && !positions[1 + col]) {
meillo@14 292 /*
meillo@14 293 * Print the character if (1) after an initial
meillo@14 294 * segment of un-selected bytes, the rest of
meillo@14 295 * it is selected, and (2) the last byte is
meillo@14 296 * selected.
meillo@14 297 */
meillo@14 298 i = col;
meillo@14 299 while (i < col + clen && i < maxval &&
meillo@14 300 !positions[1 + i])
meillo@14 301 i++;
meillo@14 302 canwrite = i < col + clen;
meillo@14 303 for (; i < col + clen && i < maxval; i++)
meillo@14 304 canwrite &= positions[1 + i];
meillo@14 305 if (canwrite)
meillo@14 306 fwrite(lbuf, 1, clen, stdout);
meillo@14 307 } else {
meillo@14 308 /*
meillo@14 309 * Print the character if all of it has
meillo@14 310 * been selected.
meillo@14 311 */
meillo@14 312 canwrite = 1;
meillo@14 313 for (i = col; i < col + clen; i++)
meillo@14 314 if ((i >= maxval && !autostop) ||
meillo@14 315 (i < maxval && !positions[1 + i])) {
meillo@14 316 canwrite = 0;
meillo@14 317 break;
meillo@14 318 }
meillo@14 319 if (canwrite)
meillo@14 320 fwrite(lbuf, 1, clen, stdout);
meillo@14 321 }
meillo@14 322 lbuf += clen;
meillo@14 323 lbuflen -= clen;
meillo@14 324 }
meillo@14 325 if (lbuflen > 0)
meillo@14 326 putchar('\n');
meillo@14 327 }
meillo@14 328 return (warned);
meillo@14 329 }
meillo@14 330
meillo@14 331 static int
meillo@14 332 c_cut(FILE *fp, const char *fname)
meillo@14 333 {
meillo@14 334 wint_t ch;
meillo@14 335 int col;
meillo@14 336 char *pos;
meillo@14 337
meillo@14 338 ch = 0;
meillo@14 339 for (;;) {
meillo@14 340 pos = positions + 1;
meillo@14 341 for (col = maxval; col; --col) {
meillo@14 342 if ((ch = getwc(fp)) == WEOF)
meillo@14 343 goto out;
meillo@14 344 if (ch == '\n')
meillo@14 345 break;
meillo@14 346 if (*pos++)
meillo@14 347 (void)putwchar(ch);
meillo@14 348 }
meillo@14 349 if (ch != '\n') {
meillo@14 350 if (autostop)
meillo@14 351 while ((ch = getwc(fp)) != WEOF && ch != '\n')
meillo@14 352 (void)putwchar(ch);
meillo@14 353 else
meillo@14 354 while ((ch = getwc(fp)) != WEOF && ch != '\n');
meillo@14 355 }
meillo@14 356 (void)putwchar('\n');
meillo@14 357 }
meillo@14 358 out:
meillo@14 359 if (ferror(fp)) {
meillo@14 360 warn("%s", fname);
meillo@14 361 return (1);
meillo@14 362 }
meillo@14 363 return (0);
meillo@14 364 }
meillo@14 365
meillo@14 366 static int
meillo@14 367 is_delim(wchar_t ch)
meillo@14 368 {
meillo@14 369 if (wflag) {
meillo@14 370 if (ch == ' ' || ch == '\t')
meillo@14 371 return 1;
meillo@14 372 } else {
meillo@14 373 if (ch == dchar)
meillo@14 374 return 1;
meillo@14 375 }
meillo@14 376 return 0;
meillo@14 377 }
meillo@14 378
meillo@14 379 static int
meillo@14 380 f_cut(FILE *fp, const char *fname)
meillo@14 381 {
meillo@14 382 wchar_t ch;
meillo@14 383 int field, i, isdelim;
meillo@14 384 char *pos, *p;
meillo@14 385 int output;
meillo@14 386 char *lbuf, *mlbuf;
meillo@14 387 size_t clen, lbuflen, reallen;
meillo@14 388
meillo@14 389 mlbuf = NULL;
meillo@14 390 while ((lbuf = fgetln(fp, &lbuflen)) != NULL) {
meillo@14 391 reallen = lbuflen;
meillo@14 392 /* Assert EOL has a newline. */
meillo@14 393 if (*(lbuf + lbuflen - 1) != '\n') {
meillo@14 394 /* Can't have > 1 line with no trailing newline. */
meillo@14 395 mlbuf = malloc(lbuflen + 1);
meillo@14 396 if (mlbuf == NULL)
meillo@14 397 err(1, "malloc");
meillo@14 398 memcpy(mlbuf, lbuf, lbuflen);
meillo@14 399 *(mlbuf + lbuflen) = '\n';
meillo@14 400 lbuf = mlbuf;
meillo@14 401 reallen++;
meillo@14 402 }
meillo@14 403 output = 0;
meillo@14 404 for (isdelim = 0, p = lbuf;; p += clen) {
meillo@14 405 clen = mbrtowc(&ch, p, lbuf + reallen - p, NULL);
meillo@14 406 if (clen == (size_t)-1 || clen == (size_t)-2) {
meillo@14 407 warnc(EILSEQ, "%s", fname);
meillo@14 408 free(mlbuf);
meillo@14 409 return (1);
meillo@14 410 }
meillo@14 411 if (clen == 0)
meillo@14 412 clen = 1;
meillo@14 413 /* this should work if newline is delimiter */
meillo@14 414 if (is_delim(ch))
meillo@14 415 isdelim = 1;
meillo@14 416 if (ch == '\n') {
meillo@14 417 if (!isdelim && !sflag)
meillo@14 418 (void)fwrite(lbuf, lbuflen, 1, stdout);
meillo@14 419 break;
meillo@14 420 }
meillo@14 421 }
meillo@14 422 if (!isdelim)
meillo@14 423 continue;
meillo@14 424
meillo@14 425 pos = positions + 1;
meillo@14 426 for (field = maxval, p = lbuf; field; --field, ++pos) {
meillo@14 427 if (*pos && output++)
meillo@14 428 for (i = 0; dcharmb[i] != '\0'; i++)
meillo@14 429 putchar(dcharmb[i]);
meillo@14 430 for (;;) {
meillo@14 431 clen = mbrtowc(&ch, p, lbuf + reallen - p,
meillo@14 432 NULL);
meillo@14 433 if (clen == (size_t)-1 || clen == (size_t)-2) {
meillo@14 434 warnc(EILSEQ, "%s", fname);
meillo@14 435 free(mlbuf);
meillo@14 436 return (1);
meillo@14 437 }
meillo@14 438 if (clen == 0)
meillo@14 439 clen = 1;
meillo@14 440 p += clen;
meillo@14 441 if (ch == '\n' || is_delim(ch)) {
meillo@14 442 /* compress whitespace */
meillo@14 443 if (wflag && ch != '\n')
meillo@14 444 while (is_delim(*p))
meillo@14 445 p++;
meillo@14 446 break;
meillo@14 447 }
meillo@14 448 if (*pos)
meillo@14 449 for (i = 0; i < (int)clen; i++)
meillo@14 450 putchar(p[i - clen]);
meillo@14 451 }
meillo@14 452 if (ch == '\n')
meillo@14 453 break;
meillo@14 454 }
meillo@14 455 if (ch != '\n') {
meillo@14 456 if (autostop) {
meillo@14 457 if (output)
meillo@14 458 for (i = 0; dcharmb[i] != '\0'; i++)
meillo@14 459 putchar(dcharmb[i]);
meillo@14 460 for (; (ch = *p) != '\n'; ++p)
meillo@14 461 (void)putchar(ch);
meillo@14 462 } else
meillo@14 463 for (; (ch = *p) != '\n'; ++p);
meillo@14 464 }
meillo@14 465 (void)putchar('\n');
meillo@14 466 }
meillo@14 467 free(mlbuf);
meillo@14 468 return (0);
meillo@14 469 }
meillo@14 470
meillo@14 471 static void
meillo@14 472 usage(void)
meillo@14 473 {
meillo@14 474 (void)fprintf(stderr, "%s\n%s\n%s\n",
meillo@14 475 "usage: cut -b list [-n] [file ...]",
meillo@14 476 " cut -c list [file ...]",
meillo@14 477 " cut -f list [-s] [-w | -d delim] [file ...]");
meillo@14 478 exit(1);
meillo@14 479 }