1 /* $OpenBSD: xargs.c,v 1.31 2015/12/09 19:29:49 mmcc Exp $ */ 2 /* $FreeBSD: xargs.c,v 1.51 2003/05/03 19:09:11 obrien Exp $ */ 3 4 /*- 5 * Copyright (c) 1990, 1993 6 * The Regents of the University of California. All rights reserved. 7 * 8 * This code is derived from software contributed to Berkeley by 9 * John B. Roll Jr. 10 * 11 * Redistribution and use in source and binary forms, with or without 12 * modification, are permitted provided that the following conditions 13 * are met: 14 * 1. Redistributions of source code must retain the above copyright 15 * notice, this list of conditions and the following disclaimer. 16 * 2. Redistributions in binary form must reproduce the above copyright 17 * notice, this list of conditions and the following disclaimer in the 18 * documentation and/or other materials provided with the distribution. 19 * 3. Neither the name of the University nor the names of its contributors 20 * may be used to endorse or promote products derived from this software 21 * without specific prior written permission. 22 * 23 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 24 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 25 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 26 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 27 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 28 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 29 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 30 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 31 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 32 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 33 * SUCH DAMAGE. 34 * 35 * $xMach: xargs.c,v 1.6 2002/02/23 05:27:47 tim Exp $ 36 */ 37 38 #include <sys/wait.h> 39 40 #include <ctype.h> 41 #include <err.h> 42 #include <errno.h> 43 #include <fcntl.h> 44 #include <langinfo.h> 45 #include <locale.h> 46 #include <paths.h> 47 #include <regex.h> 48 #include <signal.h> 49 #include <stdio.h> 50 #include <stdlib.h> 51 #include <string.h> 52 #include <unistd.h> 53 #include <limits.h> 54 55 #include "pathnames.h" 56 57 static void parse_input(int, char *[]); 58 static void prerun(int, char *[]); 59 static int prompt(void); 60 static void run(char **); 61 static void usage(void); 62 void strnsubst(char **, const char *, const char *, size_t); 63 static void waitchildren(const char *, int); 64 65 static char **av, **bxp, **ep, **endxp, **xp; 66 static char *argp, *bbp, *ebp, *inpline, *p, *replstr; 67 static const char *eofstr; 68 static int count, insingle, indouble, oflag, pflag, tflag, Rflag, rval, zflag; 69 static int cnt, Iflag, jfound, Lflag, wasquoted, xflag, runeof = 1; 70 static int curprocs, maxprocs; 71 static size_t inpsize; 72 73 extern char **environ; 74 75 int 76 main(int argc, char *argv[]) 77 { 78 long arg_max; 79 int ch, Jflag, nargs, nflag, nline; 80 size_t linelen; 81 char *endptr; 82 const char *errstr; 83 84 inpline = replstr = NULL; 85 ep = environ; 86 eofstr = ""; 87 Jflag = nflag = 0; 88 89 (void)setlocale(LC_MESSAGES, ""); 90 91 /* 92 * POSIX.2 limits the exec line length to ARG_MAX - 2K. Running that 93 * caused some E2BIG errors, so it was changed to ARG_MAX - 4K. Given 94 * that the smallest argument is 2 bytes in length, this means that 95 * the number of arguments is limited to: 96 * 97 * (ARG_MAX - 4K - LENGTH(utility + arguments)) / 2. 98 * 99 * We arbitrarily limit the number of arguments to 5000. This is 100 * allowed by POSIX.2 as long as the resulting minimum exec line is 101 * at least LINE_MAX. Realloc'ing as necessary is possible, but 102 * probably not worthwhile. 103 */ 104 nargs = 5000; 105 if ((arg_max = sysconf(_SC_ARG_MAX)) == -1) 106 errx(1, "sysconf(_SC_ARG_MAX) failed"); 107 108 if (pledge("stdio rpath proc exec", NULL) == -1) 109 err(1, "pledge"); 110 111 nline = arg_max - 4 * 1024; 112 while (*ep != NULL) { 113 /* 1 byte for each '\0' */ 114 nline -= strlen(*ep++) + 1 + sizeof(*ep); 115 } 116 maxprocs = 1; 117 while ((ch = getopt(argc, argv, "0E:I:J:L:n:oP:pR:rs:tx")) != -1) 118 switch (ch) { 119 case 'E': 120 eofstr = optarg; 121 break; 122 case 'I': 123 Jflag = 0; 124 Iflag = 1; 125 Lflag = 1; 126 replstr = optarg; 127 break; 128 case 'J': 129 Iflag = 0; 130 Jflag = 1; 131 replstr = optarg; 132 break; 133 case 'L': 134 Lflag = strtonum(optarg, 0, INT_MAX, &errstr); 135 if (errstr) 136 errx(1, "-L %s: %s", optarg, errstr); 137 break; 138 case 'n': 139 nflag = 1; 140 nargs = strtonum(optarg, 1, INT_MAX, &errstr); 141 if (errstr) 142 errx(1, "-n %s: %s", optarg, errstr); 143 break; 144 case 'o': 145 oflag = 1; 146 break; 147 case 'P': 148 maxprocs = strtonum(optarg, 1, INT_MAX, &errstr); 149 if (errstr) 150 errx(1, "-P %s: %s", optarg, errstr); 151 break; 152 case 'p': 153 pflag = 1; 154 break; 155 case 'r': 156 runeof = 0; 157 break; 158 case 'R': 159 Rflag = strtol(optarg, &endptr, 10); 160 if (*endptr != '\0') 161 errx(1, "replacements must be a number"); 162 break; 163 case 's': 164 nline = strtonum(optarg, 0, INT_MAX, &errstr); 165 if (errstr) 166 errx(1, "-s %s: %s", optarg, errstr); 167 break; 168 case 't': 169 tflag = 1; 170 break; 171 case 'x': 172 xflag = 1; 173 break; 174 case '0': 175 zflag = 1; 176 break; 177 case '?': 178 default: 179 usage(); 180 } 181 argc -= optind; 182 argv += optind; 183 184 if (!Iflag && Rflag) 185 usage(); 186 if (Iflag && !Rflag) 187 Rflag = 5; 188 if (xflag && !nflag) 189 usage(); 190 if (Iflag || Lflag) 191 xflag = 1; 192 if (replstr != NULL && *replstr == '\0') 193 errx(1, "replstr may not be empty"); 194 195 /* 196 * Allocate pointers for the utility name, the utility arguments, 197 * the maximum arguments to be read from stdin and the trailing 198 * NULL. 199 */ 200 linelen = 1 + argc + nargs + 1; 201 if ((av = bxp = calloc(linelen, sizeof(char **))) == NULL) 202 err(1, NULL); 203 204 /* 205 * Use the user's name for the utility as argv[0], just like the 206 * shell. Echo is the default. Set up pointers for the user's 207 * arguments. 208 */ 209 if (*argv == NULL) 210 cnt = strlen(*bxp++ = _PATH_ECHO); 211 else { 212 do { 213 if (Jflag && strcmp(*argv, replstr) == 0) { 214 char **avj; 215 jfound = 1; 216 argv++; 217 for (avj = argv; *avj; avj++) 218 cnt += strlen(*avj) + 1; 219 break; 220 } 221 cnt += strlen(*bxp++ = *argv) + 1; 222 } while (*++argv != NULL); 223 } 224 225 /* 226 * Set up begin/end/traversing pointers into the array. The -n 227 * count doesn't include the trailing NULL pointer, so the malloc 228 * added in an extra slot. 229 */ 230 endxp = (xp = bxp) + nargs; 231 232 /* 233 * Allocate buffer space for the arguments read from stdin and the 234 * trailing NULL. Buffer space is defined as the default or specified 235 * space, minus the length of the utility name and arguments. Set up 236 * begin/end/traversing pointers into the array. The -s count does 237 * include the trailing NULL, so the malloc didn't add in an extra 238 * slot. 239 */ 240 nline -= cnt; 241 if (nline <= 0) 242 errx(1, "insufficient space for command"); 243 244 if ((bbp = malloc((size_t)(nline + 1))) == NULL) 245 err(1, NULL); 246 ebp = (argp = p = bbp) + nline - 1; 247 for (;;) 248 parse_input(argc, argv); 249 } 250 251 static void 252 parse_input(int argc, char *argv[]) 253 { 254 int hasblank = 0; 255 static int hadblank = 0; 256 int ch, foundeof = 0; 257 char **avj; 258 259 ch = getchar(); 260 if (isblank(ch)) { 261 /* Quotes escape tabs and spaces. */ 262 if (insingle || indouble) 263 goto addch; 264 hasblank = 1; 265 if (zflag) 266 goto addch; 267 goto arg2; 268 } 269 270 switch (ch) { 271 case EOF: 272 /* No arguments since last exec. */ 273 if (p == bbp) { 274 if (runeof) 275 prerun(0, av); 276 waitchildren(*argv, 1); 277 exit(rval); 278 } 279 goto arg1; 280 case '\0': 281 if (zflag) 282 goto arg2; 283 goto addch; 284 case '\n': 285 hasblank = 1; 286 if (hadblank == 0) 287 count++; 288 if (zflag) 289 goto addch; 290 291 /* Quotes do not escape newlines. */ 292 arg1: if (insingle || indouble) 293 errx(1, "unterminated quote"); 294 arg2: 295 foundeof = *eofstr != '\0' && 296 strcmp(argp, eofstr) == 0; 297 298 /* Do not make empty args unless they are quoted */ 299 if ((argp != p || wasquoted) && !foundeof) { 300 *p++ = '\0'; 301 *xp++ = argp; 302 if (Iflag) { 303 size_t curlen; 304 305 if (inpline == NULL) 306 curlen = 0; 307 else { 308 /* 309 * If this string is not zero 310 * length, append a space for 311 * separation before the next 312 * argument. 313 */ 314 if ((curlen = strlen(inpline))) 315 strlcat(inpline, " ", inpsize); 316 } 317 curlen++; 318 /* 319 * Allocate enough to hold what we will 320 * be holding in a second, and to append 321 * a space next time through, if we have 322 * to. 323 */ 324 inpsize = curlen + 2 + strlen(argp); 325 inpline = realloc(inpline, inpsize); 326 if (inpline == NULL) 327 errx(1, "realloc failed"); 328 if (curlen == 1) 329 strlcpy(inpline, argp, inpsize); 330 else 331 strlcat(inpline, argp, inpsize); 332 } 333 } 334 335 /* 336 * If max'd out on args or buffer, or reached EOF, 337 * run the command. If xflag and max'd out on buffer 338 * but not on args, object. Having reached the limit 339 * of input lines, as specified by -L is the same as 340 * maxing out on arguments. 341 */ 342 if (xp == endxp || p > ebp || ch == EOF || 343 (Lflag <= count && xflag) || foundeof) { 344 if (xflag && xp != endxp && p > ebp) 345 errx(1, "insufficient space for arguments"); 346 if (jfound) { 347 for (avj = argv; *avj; avj++) 348 *xp++ = *avj; 349 } 350 prerun(argc, av); 351 if (ch == EOF || foundeof) { 352 waitchildren(*argv, 1); 353 exit(rval); 354 } 355 p = bbp; 356 xp = bxp; 357 count = 0; 358 } 359 argp = p; 360 wasquoted = 0; 361 break; 362 case '\'': 363 if (indouble || zflag) 364 goto addch; 365 insingle = !insingle; 366 wasquoted = 1; 367 break; 368 case '"': 369 if (insingle || zflag) 370 goto addch; 371 indouble = !indouble; 372 wasquoted = 1; 373 break; 374 case '\\': 375 if (zflag) 376 goto addch; 377 /* Backslash escapes anything, is escaped by quotes. */ 378 if (!insingle && !indouble && (ch = getchar()) == EOF) 379 errx(1, "backslash at EOF"); 380 /* FALLTHROUGH */ 381 default: 382 addch: if (p < ebp) { 383 *p++ = ch; 384 break; 385 } 386 387 /* If only one argument, not enough buffer space. */ 388 if (bxp == xp) 389 errx(1, "insufficient space for argument"); 390 /* Didn't hit argument limit, so if xflag object. */ 391 if (xflag) 392 errx(1, "insufficient space for arguments"); 393 394 if (jfound) { 395 for (avj = argv; *avj; avj++) 396 *xp++ = *avj; 397 } 398 prerun(argc, av); 399 xp = bxp; 400 cnt = ebp - argp; 401 memcpy(bbp, argp, (size_t)cnt); 402 p = (argp = bbp) + cnt; 403 *p++ = ch; 404 break; 405 } 406 hadblank = hasblank; 407 } 408 409 /* 410 * Do things necessary before run()'ing, such as -I substitution, 411 * and then call run(). 412 */ 413 static void 414 prerun(int argc, char *argv[]) 415 { 416 char **tmp, **tmp2, **avj; 417 int repls; 418 419 repls = Rflag; 420 runeof = 0; 421 422 if (argc == 0 || repls == 0) { 423 *xp = NULL; 424 run(argv); 425 return; 426 } 427 428 avj = argv; 429 430 /* 431 * Allocate memory to hold the argument list, and 432 * a NULL at the tail. 433 */ 434 tmp = calloc(argc + 1, sizeof(char**)); 435 if (tmp == NULL) 436 err(1, NULL); 437 tmp2 = tmp; 438 439 /* 440 * Save the first argument and iterate over it, we 441 * cannot do strnsubst() to it. 442 */ 443 if ((*tmp++ = strdup(*avj++)) == NULL) 444 err(1, NULL); 445 446 /* 447 * For each argument to utility, if we have not used up 448 * the number of replacements we are allowed to do, and 449 * if the argument contains at least one occurrence of 450 * replstr, call strnsubst(), else just save the string. 451 * Iterations over elements of avj and tmp are done 452 * where appropriate. 453 */ 454 while (--argc) { 455 *tmp = *avj++; 456 if (repls && strstr(*tmp, replstr) != NULL) { 457 strnsubst(tmp++, replstr, inpline, (size_t)255); 458 if (repls > 0) 459 repls--; 460 } else { 461 if ((*tmp = strdup(*tmp)) == NULL) 462 err(1, NULL); 463 tmp++; 464 } 465 } 466 467 /* 468 * Run it. 469 */ 470 *tmp = NULL; 471 run(tmp2); 472 473 /* 474 * Walk from the tail to the head, free along the way. 475 */ 476 for (; tmp2 != tmp; tmp--) 477 free(*tmp); 478 /* 479 * Now free the list itself. 480 */ 481 free(tmp2); 482 483 /* 484 * Free the input line buffer, if we have one. 485 */ 486 free(inpline); 487 inpline = NULL; 488 } 489 490 static void 491 run(char **argv) 492 { 493 pid_t pid; 494 int fd; 495 char **avec; 496 497 /* 498 * If the user wants to be notified of each command before it is 499 * executed, notify them. If they want the notification to be 500 * followed by a prompt, then prompt them. 501 */ 502 if (tflag || pflag) { 503 (void)fprintf(stderr, "%s", *argv); 504 for (avec = argv + 1; *avec != NULL; ++avec) 505 (void)fprintf(stderr, " %s", *avec); 506 /* 507 * If the user has asked to be prompted, do so. 508 */ 509 if (pflag) 510 /* 511 * If they asked not to exec, return without execution 512 * but if they asked to, go to the execution. If we 513 * could not open their tty, break the switch and drop 514 * back to -t behaviour. 515 */ 516 switch (prompt()) { 517 case 0: 518 return; 519 case 1: 520 goto exec; 521 case 2: 522 break; 523 } 524 (void)fprintf(stderr, "\n"); 525 (void)fflush(stderr); 526 } 527 exec: 528 switch (pid = vfork()) { 529 case -1: 530 err(1, "vfork"); 531 case 0: 532 if (oflag) { 533 if ((fd = open(_PATH_TTY, O_RDONLY)) == -1) { 534 warn("can't open /dev/tty"); 535 _exit(1); 536 } 537 } else { 538 fd = open(_PATH_DEVNULL, O_RDONLY); 539 } 540 if (fd > STDIN_FILENO) { 541 if (dup2(fd, STDIN_FILENO) != 0) { 542 warn("can't dup2 to stdin"); 543 _exit(1); 544 } 545 close(fd); 546 } 547 execvp(argv[0], argv); 548 warn("%s", argv[0]); 549 _exit(errno == ENOENT ? 127 : 126); 550 } 551 curprocs++; 552 waitchildren(*argv, 0); 553 } 554 555 static void 556 waitchildren(const char *name, int waitall) 557 { 558 pid_t pid; 559 int status; 560 561 while ((pid = waitpid(-1, &status, !waitall && curprocs < maxprocs ? 562 WNOHANG : 0)) > 0) { 563 curprocs--; 564 /* 565 * According to POSIX, we have to exit if the utility exits 566 * with a 255 status, or is interrupted by a signal. 567 * We are allowed to return any exit status between 1 and 568 * 125 in these cases, but we'll use 124 and 125, the same 569 * values used by GNU xargs. 570 */ 571 if (WIFEXITED(status)) { 572 if (WEXITSTATUS(status) == 255) { 573 warnx("%s exited with status 255", name); 574 exit(124); 575 } else if (WEXITSTATUS(status) == 127 || 576 WEXITSTATUS(status) == 126) { 577 exit(WEXITSTATUS(status)); 578 } else if (WEXITSTATUS(status) != 0) { 579 rval = 123; 580 } 581 } else if (WIFSIGNALED(status)) { 582 if (WTERMSIG(status) != SIGPIPE) { 583 if (WTERMSIG(status) < NSIG) 584 warnx("%s terminated by SIG%s", name, 585 sys_signame[WTERMSIG(status)]); 586 else 587 warnx("%s terminated by signal %d", 588 name, WTERMSIG(status)); 589 } 590 exit(125); 591 } 592 } 593 if (pid == -1 && errno != ECHILD) 594 err(1, "waitpid"); 595 } 596 597 /* 598 * Prompt the user about running a command. 599 */ 600 static int 601 prompt(void) 602 { 603 regex_t cre; 604 size_t rsize; 605 int match; 606 char *response; 607 FILE *ttyfp; 608 609 if ((ttyfp = fopen(_PATH_TTY, "r")) == NULL) 610 return (2); /* Indicate that the TTY failed to open. */ 611 (void)fprintf(stderr, "?..."); 612 (void)fflush(stderr); 613 if ((response = fgetln(ttyfp, &rsize)) == NULL || 614 regcomp(&cre, nl_langinfo(YESEXPR), REG_BASIC) != 0) { 615 (void)fclose(ttyfp); 616 return (0); 617 } 618 response[rsize - 1] = '\0'; 619 match = regexec(&cre, response, 0, NULL, 0); 620 (void)fclose(ttyfp); 621 regfree(&cre); 622 return (match == 0); 623 } 624 625 static void 626 usage(void) 627 { 628 fprintf(stderr, 629 "usage: xargs [-0oprt] [-E eofstr] [-I replstr [-R replacements]] [-J replstr]\n" 630 " [-L number] [-n number [-x]] [-P maxprocs] [-s size]\n" 631 " [utility [argument ...]]\n"); 632 exit(1); 633 } 634