1 /* $OpenBSD: xargs.c,v 1.35 2020/07/19 13:19:25 schwarze 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 <paths.h> 45 #include <signal.h> 46 #include <stdio.h> 47 #include <stdlib.h> 48 #include <string.h> 49 #include <unistd.h> 50 #include <limits.h> 51 52 #include "pathnames.h" 53 54 static void parse_input(int, char *[]); 55 static void prerun(int, char *[]); 56 static int prompt(void); 57 static void run(char **); 58 static void usage(void); 59 void strnsubst(char **, const char *, const char *, size_t); 60 static void waitchildren(const char *, int); 61 62 static char **av, **bxp, **ep, **endxp, **xp; 63 static char *argp, *bbp, *ebp, *inpline, *p, *replstr; 64 static const char *eofstr; 65 static int count, insingle, indouble, oflag, pflag, tflag, Rflag, rval, zflag; 66 static int cnt, Iflag, jfound, Lflag, wasquoted, xflag, runeof = 1; 67 static int curprocs, maxprocs; 68 static size_t inpsize; 69 70 extern char **environ; 71 72 int 73 main(int argc, char *argv[]) 74 { 75 long arg_max; 76 int ch, Jflag, nargs, nflag, nline; 77 size_t linelen; 78 char *endptr; 79 const char *errstr; 80 81 inpline = replstr = NULL; 82 ep = environ; 83 eofstr = ""; 84 Jflag = nflag = 0; 85 86 /* 87 * POSIX.2 limits the exec line length to ARG_MAX - 2K. Running that 88 * caused some E2BIG errors, so it was changed to ARG_MAX - 4K. Given 89 * that the smallest argument is 2 bytes in length, this means that 90 * the number of arguments is limited to: 91 * 92 * (ARG_MAX - 4K - LENGTH(utility + arguments)) / 2. 93 * 94 * We arbitrarily limit the number of arguments to 5000. This is 95 * allowed by POSIX.2 as long as the resulting minimum exec line is 96 * at least LINE_MAX. Realloc'ing as necessary is possible, but 97 * probably not worthwhile. 98 */ 99 nargs = 5000; 100 if ((arg_max = sysconf(_SC_ARG_MAX)) == -1) 101 errx(1, "sysconf(_SC_ARG_MAX) failed"); 102 103 if (pledge("stdio rpath proc exec", NULL) == -1) 104 err(1, "pledge"); 105 106 nline = arg_max - 4 * 1024; 107 while (*ep != NULL) { 108 /* 1 byte for each '\0' */ 109 nline -= strlen(*ep++) + 1 + sizeof(*ep); 110 } 111 maxprocs = 1; 112 while ((ch = getopt(argc, argv, "0E:I:J:L:n:oP:pR:rs:tx")) != -1) 113 switch (ch) { 114 case 'E': 115 eofstr = optarg; 116 break; 117 case 'I': 118 Jflag = 0; 119 Iflag = 1; 120 Lflag = 1; 121 replstr = optarg; 122 break; 123 case 'J': 124 Iflag = 0; 125 Jflag = 1; 126 replstr = optarg; 127 break; 128 case 'L': 129 Lflag = strtonum(optarg, 0, INT_MAX, &errstr); 130 if (errstr) 131 errx(1, "-L %s: %s", optarg, errstr); 132 break; 133 case 'n': 134 nflag = 1; 135 nargs = strtonum(optarg, 1, INT_MAX, &errstr); 136 if (errstr) 137 errx(1, "-n %s: %s", optarg, errstr); 138 break; 139 case 'o': 140 oflag = 1; 141 break; 142 case 'P': 143 maxprocs = strtonum(optarg, 1, INT_MAX, &errstr); 144 if (errstr) 145 errx(1, "-P %s: %s", optarg, errstr); 146 break; 147 case 'p': 148 pflag = 1; 149 break; 150 case 'r': 151 runeof = 0; 152 break; 153 case 'R': 154 Rflag = strtol(optarg, &endptr, 10); 155 if (*endptr != '\0') 156 errx(1, "replacements must be a number"); 157 break; 158 case 's': 159 nline = strtonum(optarg, 0, INT_MAX, &errstr); 160 if (errstr) 161 errx(1, "-s %s: %s", optarg, errstr); 162 break; 163 case 't': 164 tflag = 1; 165 break; 166 case 'x': 167 xflag = 1; 168 break; 169 case '0': 170 zflag = 1; 171 break; 172 case '?': 173 default: 174 usage(); 175 } 176 argc -= optind; 177 argv += optind; 178 179 if (!Iflag && Rflag) 180 usage(); 181 if (Iflag && !Rflag) 182 Rflag = 5; 183 if (xflag && !nflag) 184 usage(); 185 if (Iflag || Lflag) 186 xflag = 1; 187 if (replstr != NULL && *replstr == '\0') 188 errx(1, "replstr may not be empty"); 189 190 /* 191 * Allocate pointers for the utility name, the utility arguments, 192 * the maximum arguments to be read from stdin and the trailing 193 * NULL. 194 */ 195 linelen = 1 + argc + nargs + 1; 196 if ((av = bxp = calloc(linelen, sizeof(char *))) == NULL) 197 err(1, NULL); 198 199 /* 200 * Use the user's name for the utility as argv[0], just like the 201 * shell. Echo is the default. Set up pointers for the user's 202 * arguments. 203 */ 204 if (*argv == NULL) 205 cnt = strlen(*bxp++ = _PATH_ECHO); 206 else { 207 do { 208 if (Jflag && strcmp(*argv, replstr) == 0) { 209 char **avj; 210 jfound = 1; 211 argv++; 212 for (avj = argv; *avj; avj++) 213 cnt += strlen(*avj) + 1; 214 break; 215 } 216 cnt += strlen(*bxp++ = *argv) + 1; 217 } while (*++argv != NULL); 218 } 219 220 /* 221 * Set up begin/end/traversing pointers into the array. The -n 222 * count doesn't include the trailing NULL pointer, so the malloc 223 * added in an extra slot. 224 */ 225 endxp = (xp = bxp) + nargs; 226 227 /* 228 * Allocate buffer space for the arguments read from stdin and the 229 * trailing NULL. Buffer space is defined as the default or specified 230 * space, minus the length of the utility name and arguments. Set up 231 * begin/end/traversing pointers into the array. The -s count does 232 * include the trailing NULL, so the malloc didn't add in an extra 233 * slot. 234 */ 235 nline -= cnt; 236 if (nline <= 0) 237 errx(1, "insufficient space for command"); 238 239 if ((bbp = malloc((size_t)(nline + 1))) == NULL) 240 err(1, NULL); 241 ebp = (argp = p = bbp) + nline - 1; 242 for (;;) 243 parse_input(argc, argv); 244 } 245 246 static void 247 parse_input(int argc, char *argv[]) 248 { 249 int hasblank = 0; 250 static int hadblank = 0; 251 int ch, foundeof = 0; 252 char **avj; 253 254 ch = getchar(); 255 if (isblank(ch)) { 256 /* Quotes escape tabs and spaces. */ 257 if (insingle || indouble) 258 goto addch; 259 hasblank = 1; 260 if (zflag) 261 goto addch; 262 goto arg2; 263 } 264 265 switch (ch) { 266 case EOF: 267 /* No arguments since last exec. */ 268 if (p == bbp) { 269 if (runeof) 270 prerun(0, av); 271 waitchildren(*argv, 1); 272 exit(rval); 273 } 274 goto arg1; 275 case '\0': 276 if (zflag) { 277 /* 278 * Increment 'count', so that nulls will be treated 279 * as end-of-line, as well as end-of-argument. This 280 * is needed so -0 works properly with -I and -L. 281 */ 282 count++; 283 goto arg2; 284 } 285 goto addch; 286 case '\n': 287 if (zflag) 288 goto addch; 289 hasblank = 1; 290 if (hadblank == 0) 291 count++; 292 293 /* Quotes do not escape newlines. */ 294 arg1: if (insingle || indouble) 295 errx(1, "unterminated quote"); 296 arg2: 297 foundeof = *eofstr != '\0' && 298 strcmp(argp, eofstr) == 0; 299 300 /* Do not make empty args unless they are quoted */ 301 if ((argp != p || wasquoted) && !foundeof) { 302 *p++ = '\0'; 303 *xp++ = argp; 304 if (Iflag) { 305 size_t curlen; 306 307 if (inpline == NULL) 308 curlen = 0; 309 else { 310 /* 311 * If this string is not zero 312 * length, append a space for 313 * separation before the next 314 * argument. 315 */ 316 if ((curlen = strlen(inpline))) 317 strlcat(inpline, " ", inpsize); 318 } 319 curlen++; 320 /* 321 * Allocate enough to hold what we will 322 * be holding in a second, and to append 323 * a space next time through, if we have 324 * to. 325 */ 326 inpsize = curlen + 2 + strlen(argp); 327 inpline = realloc(inpline, inpsize); 328 if (inpline == NULL) 329 errx(1, "realloc failed"); 330 if (curlen == 1) 331 strlcpy(inpline, argp, inpsize); 332 else 333 strlcat(inpline, argp, inpsize); 334 } 335 } 336 337 /* 338 * If max'd out on args or buffer, or reached EOF, 339 * run the command. If xflag and max'd out on buffer 340 * but not on args, object. Having reached the limit 341 * of input lines, as specified by -L is the same as 342 * maxing out on arguments. 343 */ 344 if (xp == endxp || p > ebp || ch == EOF || 345 (Lflag <= count && xflag) || foundeof) { 346 if (xflag && xp != endxp && p > ebp) 347 errx(1, "insufficient space for arguments"); 348 if (jfound) { 349 for (avj = argv; *avj; avj++) 350 *xp++ = *avj; 351 } 352 prerun(argc, av); 353 if (ch == EOF || foundeof) { 354 waitchildren(*argv, 1); 355 exit(rval); 356 } 357 p = bbp; 358 xp = bxp; 359 count = 0; 360 } 361 argp = p; 362 wasquoted = 0; 363 break; 364 case '\'': 365 if (indouble || zflag) 366 goto addch; 367 insingle = !insingle; 368 wasquoted = 1; 369 break; 370 case '"': 371 if (insingle || zflag) 372 goto addch; 373 indouble = !indouble; 374 wasquoted = 1; 375 break; 376 case '\\': 377 if (zflag) 378 goto addch; 379 /* Backslash escapes anything, is escaped by quotes. */ 380 if (!insingle && !indouble && (ch = getchar()) == EOF) 381 errx(1, "backslash at EOF"); 382 /* FALLTHROUGH */ 383 default: 384 addch: if (p < ebp) { 385 *p++ = ch; 386 break; 387 } 388 389 /* If only one argument, not enough buffer space. */ 390 if (bxp == xp) 391 errx(1, "insufficient space for argument"); 392 /* Didn't hit argument limit, so if xflag object. */ 393 if (xflag) 394 errx(1, "insufficient space for arguments"); 395 396 if (jfound) { 397 for (avj = argv; *avj; avj++) 398 *xp++ = *avj; 399 } 400 prerun(argc, av); 401 xp = bxp; 402 cnt = ebp - argp; 403 memmove(bbp, argp, (size_t)cnt); 404 p = (argp = bbp) + cnt; 405 *p++ = ch; 406 break; 407 } 408 hadblank = hasblank; 409 } 410 411 /* 412 * Do things necessary before run()'ing, such as -I substitution, 413 * and then call run(). 414 */ 415 static void 416 prerun(int argc, char *argv[]) 417 { 418 char **tmp, **tmp2, **avj; 419 int repls; 420 421 repls = Rflag; 422 runeof = 0; 423 424 if (argc == 0 || repls == 0) { 425 *xp = NULL; 426 run(argv); 427 return; 428 } 429 430 avj = argv; 431 432 /* 433 * Allocate memory to hold the argument list, and 434 * a NULL at the tail. 435 */ 436 tmp = calloc(argc + 1, sizeof(char *)); 437 if (tmp == NULL) 438 err(1, NULL); 439 tmp2 = tmp; 440 441 /* 442 * Save the first argument and iterate over it, we 443 * cannot do strnsubst() to it. 444 */ 445 if ((*tmp++ = strdup(*avj++)) == NULL) 446 err(1, NULL); 447 448 /* 449 * For each argument to utility, if we have not used up 450 * the number of replacements we are allowed to do, and 451 * if the argument contains at least one occurrence of 452 * replstr, call strnsubst(), else just save the string. 453 * Iterations over elements of avj and tmp are done 454 * where appropriate. 455 */ 456 while (--argc) { 457 *tmp = *avj++; 458 if (repls && strstr(*tmp, replstr) != NULL) { 459 strnsubst(tmp++, replstr, inpline, (size_t)255); 460 if (repls > 0) 461 repls--; 462 } else { 463 if ((*tmp = strdup(*tmp)) == NULL) 464 err(1, NULL); 465 tmp++; 466 } 467 } 468 469 /* 470 * Run it. 471 */ 472 *tmp = NULL; 473 run(tmp2); 474 475 /* 476 * Walk from the tail to the head, free along the way. 477 */ 478 for (; tmp2 != tmp; tmp--) 479 free(*tmp); 480 /* 481 * Now free the list itself. 482 */ 483 free(tmp2); 484 485 /* 486 * Free the input line buffer, if we have one. 487 */ 488 free(inpline); 489 inpline = NULL; 490 } 491 492 static void 493 run(char **argv) 494 { 495 pid_t pid; 496 int fd; 497 char **avec; 498 499 /* 500 * If the user wants to be notified of each command before it is 501 * executed, notify them. If they want the notification to be 502 * followed by a prompt, then prompt them. 503 */ 504 if (tflag || pflag) { 505 fprintf(stderr, "%s", *argv); 506 for (avec = argv + 1; *avec != NULL; ++avec) 507 fprintf(stderr, " %s", *avec); 508 /* 509 * If the user has asked to be prompted, do so. 510 */ 511 if (pflag) 512 /* 513 * If they asked not to exec, return without execution 514 * but if they asked to, go to the execution. If we 515 * could not open their tty, break the switch and drop 516 * back to -t behaviour. 517 */ 518 switch (prompt()) { 519 case 0: 520 return; 521 case 1: 522 goto exec; 523 case 2: 524 break; 525 } 526 fprintf(stderr, "\n"); 527 fflush(stderr); 528 } 529 exec: 530 switch (pid = vfork()) { 531 case -1: 532 err(1, "vfork"); 533 case 0: 534 if (oflag) { 535 if ((fd = open(_PATH_TTY, O_RDONLY)) == -1) { 536 warn("can't open /dev/tty"); 537 _exit(1); 538 } 539 } else { 540 fd = open(_PATH_DEVNULL, O_RDONLY); 541 } 542 if (fd > STDIN_FILENO) { 543 if (dup2(fd, STDIN_FILENO) != 0) { 544 warn("can't dup2 to stdin"); 545 _exit(1); 546 } 547 close(fd); 548 } 549 execvp(argv[0], argv); 550 warn("%s", argv[0]); 551 _exit(errno == ENOENT ? 127 : 126); 552 } 553 curprocs++; 554 waitchildren(*argv, 0); 555 } 556 557 static void 558 waitchildren(const char *name, int waitall) 559 { 560 pid_t pid; 561 int status; 562 563 while ((pid = waitpid(-1, &status, !waitall && curprocs < maxprocs ? 564 WNOHANG : 0)) > 0) { 565 curprocs--; 566 /* 567 * According to POSIX, we have to exit if the utility exits 568 * with a 255 status, or is interrupted by a signal. 569 * We are allowed to return any exit status between 1 and 570 * 125 in these cases, but we'll use 124 and 125, the same 571 * values used by GNU xargs. 572 */ 573 if (WIFEXITED(status)) { 574 if (WEXITSTATUS(status) == 255) { 575 warnx("%s exited with status 255", name); 576 exit(124); 577 } else if (WEXITSTATUS(status) == 127 || 578 WEXITSTATUS(status) == 126) { 579 exit(WEXITSTATUS(status)); 580 } else if (WEXITSTATUS(status) != 0) { 581 rval = 123; 582 } 583 } else if (WIFSIGNALED(status)) { 584 if (WTERMSIG(status) != SIGPIPE) { 585 if (WTERMSIG(status) < NSIG) 586 warnx("%s terminated by SIG%s", name, 587 sys_signame[WTERMSIG(status)]); 588 else 589 warnx("%s terminated by signal %d", 590 name, WTERMSIG(status)); 591 } 592 exit(125); 593 } 594 } 595 if (pid == -1 && errno != ECHILD) 596 err(1, "waitpid"); 597 } 598 599 /* 600 * Prompt the user about running a command. 601 */ 602 static int 603 prompt(void) 604 { 605 size_t rsize; 606 char *response; 607 FILE *ttyfp; 608 int doit = 0; 609 610 if ((ttyfp = fopen(_PATH_TTY, "r")) == NULL) 611 return (2); /* Indicate that the TTY failed to open. */ 612 fprintf(stderr, "?..."); 613 fflush(stderr); 614 response = fgetln(ttyfp, &rsize); 615 doit = response != NULL && (*response == 'y' || *response == 'Y'); 616 fclose(ttyfp); 617 return (doit); 618 } 619 620 static void 621 usage(void) 622 { 623 fprintf(stderr, 624 "usage: xargs [-0oprt] [-E eofstr] [-I replstr [-R replacements]] [-J replstr]\n" 625 " [-L number] [-n number [-x]] [-P maxprocs] [-s size]\n" 626 " [utility [argument ...]]\n"); 627 exit(1); 628 } 629