1 /*- 2 * Copyright (c) 1991, 1993 3 * The Regents of the University of California. All rights reserved. 4 * Copyright (c) 1997-2005 5 * Herbert Xu <herbert@gondor.apana.org.au>. All rights reserved. 6 * 7 * This code is derived from software contributed to Berkeley by 8 * Kenneth Almquist. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 4. Neither the name of the University nor the names of its contributors 19 * may be used to endorse or promote products derived from this software 20 * without specific prior written permission. 21 * 22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 25 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 32 * SUCH DAMAGE. 33 */ 34 35 #ifndef lint 36 #if 0 37 static char sccsid[] = "@(#)expand.c 8.5 (Berkeley) 5/15/95"; 38 #endif 39 #endif /* not lint */ 40 #include <sys/cdefs.h> 41 __FBSDID("$FreeBSD$"); 42 43 #include <sys/types.h> 44 #include <sys/time.h> 45 #include <sys/stat.h> 46 #include <dirent.h> 47 #include <errno.h> 48 #include <inttypes.h> 49 #include <limits.h> 50 #include <pwd.h> 51 #include <stdio.h> 52 #include <stdlib.h> 53 #include <string.h> 54 #include <unistd.h> 55 #include <wchar.h> 56 #include <wctype.h> 57 58 /* 59 * Routines to expand arguments to commands. We have to deal with 60 * backquotes, shell variables, and file metacharacters. 61 */ 62 63 #include "shell.h" 64 #include "main.h" 65 #include "nodes.h" 66 #include "eval.h" 67 #include "expand.h" 68 #include "syntax.h" 69 #include "parser.h" 70 #include "jobs.h" 71 #include "options.h" 72 #include "var.h" 73 #include "input.h" 74 #include "output.h" 75 #include "memalloc.h" 76 #include "error.h" 77 #include "mystring.h" 78 #include "arith.h" 79 #include "show.h" 80 #include "builtins.h" 81 82 /* 83 * Structure specifying which parts of the string should be searched 84 * for IFS characters. 85 */ 86 87 struct ifsregion { 88 struct ifsregion *next; /* next region in list */ 89 int begoff; /* offset of start of region */ 90 int endoff; /* offset of end of region */ 91 int inquotes; /* search for nul bytes only */ 92 }; 93 94 95 static char *expdest; /* output of current string */ 96 static struct nodelist *argbackq; /* list of back quote expressions */ 97 static struct ifsregion ifsfirst; /* first struct in list of ifs regions */ 98 static struct ifsregion *ifslastp; /* last struct in list */ 99 100 static char *argstr(char *, int); 101 static char *exptilde(char *, int); 102 static char *expari(char *); 103 static void expbackq(union node *, int, int); 104 static int subevalvar(char *, char *, int, int, int, int, int); 105 static char *evalvar(char *, int); 106 static int varisset(const char *, int); 107 static void strtodest(const char *, int, int, int); 108 static void varvalue(const char *, int, int, int); 109 static void recordregion(int, int, int); 110 static void removerecordregions(int); 111 static void ifsbreakup(char *, struct arglist *); 112 static void expandmeta(struct arglist *, struct arglist *); 113 static void expmeta(char *, char *, struct arglist *); 114 static int expsortcmp(const void *, const void *); 115 static int patmatch(const char *, const char *, int); 116 static char *cvtnum(int, char *); 117 static int collate_range_cmp(wchar_t, wchar_t); 118 119 void 120 emptyarglist(struct arglist *list) 121 { 122 123 list->args = list->smallarg; 124 list->count = 0; 125 list->capacity = sizeof(list->smallarg) / sizeof(list->smallarg[0]); 126 } 127 128 void 129 appendarglist(struct arglist *list, char *str) 130 { 131 char **newargs; 132 int newcapacity; 133 134 if (list->count >= list->capacity) { 135 newcapacity = list->capacity * 2; 136 if (newcapacity < 16) 137 newcapacity = 16; 138 if (newcapacity > INT_MAX / (int)sizeof(newargs[0])) 139 error("Too many entries in arglist"); 140 newargs = stalloc(newcapacity * sizeof(newargs[0])); 141 memcpy(newargs, list->args, list->count * sizeof(newargs[0])); 142 list->args = newargs; 143 list->capacity = newcapacity; 144 } 145 list->args[list->count++] = str; 146 } 147 148 static int 149 collate_range_cmp(wchar_t c1, wchar_t c2) 150 { 151 static wchar_t s1[2], s2[2]; 152 153 s1[0] = c1; 154 s2[0] = c2; 155 return (wcscoll(s1, s2)); 156 } 157 158 static char * 159 stputs_quotes(const char *data, const char *syntax, char *p) 160 { 161 while (*data) { 162 CHECKSTRSPACE(2, p); 163 if (syntax[(int)*data] == CCTL) 164 USTPUTC(CTLESC, p); 165 USTPUTC(*data++, p); 166 } 167 return (p); 168 } 169 #define STPUTS_QUOTES(data, syntax, p) p = stputs_quotes((data), syntax, p) 170 171 /* 172 * Perform expansions on an argument, placing the resulting list of arguments 173 * in arglist. Parameter expansion, command substitution and arithmetic 174 * expansion are always performed; additional expansions can be requested 175 * via flag (EXP_*). 176 * The result is left in the stack string. 177 * When arglist is NULL, perform here document expansion. 178 * 179 * Caution: this function uses global state and is not reentrant. 180 * However, a new invocation after an interrupted invocation is safe 181 * and will reset the global state for the new call. 182 */ 183 void 184 expandarg(union node *arg, struct arglist *arglist, int flag) 185 { 186 struct arglist exparg; 187 char *p; 188 189 argbackq = arg->narg.backquote; 190 STARTSTACKSTR(expdest); 191 ifsfirst.next = NULL; 192 ifslastp = NULL; 193 argstr(arg->narg.text, flag); 194 if (arglist == NULL) { 195 STACKSTRNUL(expdest); 196 return; /* here document expanded */ 197 } 198 STPUTC('\0', expdest); 199 p = grabstackstr(expdest); 200 emptyarglist(&exparg); 201 if (flag & EXP_FULL) { 202 ifsbreakup(p, &exparg); 203 expandmeta(&exparg, arglist); 204 } else 205 appendarglist(arglist, p); 206 while (ifsfirst.next != NULL) { 207 struct ifsregion *ifsp; 208 INTOFF; 209 ifsp = ifsfirst.next->next; 210 ckfree(ifsfirst.next); 211 ifsfirst.next = ifsp; 212 INTON; 213 } 214 } 215 216 217 218 /* 219 * Perform parameter expansion, command substitution and arithmetic 220 * expansion, and tilde expansion if requested via EXP_TILDE/EXP_VARTILDE. 221 * Processing ends at a CTLENDVAR or CTLENDARI character as well as '\0'. 222 * This is used to expand word in ${var+word} etc. 223 * If EXP_FULL or EXP_CASE are set, keep and/or generate CTLESC 224 * characters to allow for further processing. 225 * If EXP_FULL is set, also preserve CTLQUOTEMARK characters. 226 */ 227 static char * 228 argstr(char *p, int flag) 229 { 230 char c; 231 int quotes = flag & (EXP_FULL | EXP_CASE); /* do CTLESC */ 232 int firsteq = 1; 233 int split_lit; 234 int lit_quoted; 235 236 split_lit = flag & EXP_SPLIT_LIT; 237 lit_quoted = flag & EXP_LIT_QUOTED; 238 flag &= ~(EXP_SPLIT_LIT | EXP_LIT_QUOTED); 239 if (*p == '~' && (flag & (EXP_TILDE | EXP_VARTILDE))) 240 p = exptilde(p, flag); 241 for (;;) { 242 CHECKSTRSPACE(2, expdest); 243 switch (c = *p++) { 244 case '\0': 245 return (p - 1); 246 case CTLENDVAR: 247 case CTLENDARI: 248 return (p); 249 case CTLQUOTEMARK: 250 lit_quoted = 1; 251 /* "$@" syntax adherence hack */ 252 if (p[0] == CTLVAR && p[2] == '@' && p[3] == '=') 253 break; 254 if ((flag & EXP_FULL) != 0) 255 USTPUTC(c, expdest); 256 break; 257 case CTLQUOTEEND: 258 lit_quoted = 0; 259 break; 260 case CTLESC: 261 if (quotes) 262 USTPUTC(c, expdest); 263 c = *p++; 264 USTPUTC(c, expdest); 265 if (split_lit && !lit_quoted) 266 recordregion(expdest - stackblock() - 267 (quotes ? 2 : 1), 268 expdest - stackblock(), 0); 269 break; 270 case CTLVAR: 271 p = evalvar(p, flag); 272 break; 273 case CTLBACKQ: 274 case CTLBACKQ|CTLQUOTE: 275 expbackq(argbackq->n, c & CTLQUOTE, flag); 276 argbackq = argbackq->next; 277 break; 278 case CTLARI: 279 p = expari(p); 280 break; 281 case ':': 282 case '=': 283 /* 284 * sort of a hack - expand tildes in variable 285 * assignments (after the first '=' and after ':'s). 286 */ 287 USTPUTC(c, expdest); 288 if (split_lit && !lit_quoted) 289 recordregion(expdest - stackblock() - 1, 290 expdest - stackblock(), 0); 291 if (flag & EXP_VARTILDE && *p == '~' && 292 (c != '=' || firsteq)) { 293 if (c == '=') 294 firsteq = 0; 295 p = exptilde(p, flag); 296 } 297 break; 298 default: 299 USTPUTC(c, expdest); 300 if (split_lit && !lit_quoted) 301 recordregion(expdest - stackblock() - 1, 302 expdest - stackblock(), 0); 303 } 304 } 305 } 306 307 /* 308 * Perform tilde expansion, placing the result in the stack string and 309 * returning the next position in the input string to process. 310 */ 311 static char * 312 exptilde(char *p, int flag) 313 { 314 char c, *startp = p; 315 struct passwd *pw; 316 char *home; 317 318 for (;;) { 319 c = *p; 320 switch(c) { 321 case CTLESC: /* This means CTL* are always considered quoted. */ 322 case CTLVAR: 323 case CTLBACKQ: 324 case CTLBACKQ | CTLQUOTE: 325 case CTLARI: 326 case CTLENDARI: 327 case CTLQUOTEMARK: 328 return (startp); 329 case ':': 330 if ((flag & EXP_VARTILDE) == 0) 331 break; 332 /* FALLTHROUGH */ 333 case '\0': 334 case '/': 335 case CTLENDVAR: 336 *p = '\0'; 337 if (*(startp+1) == '\0') { 338 home = lookupvar("HOME"); 339 } else { 340 pw = getpwnam(startp+1); 341 home = pw != NULL ? pw->pw_dir : NULL; 342 } 343 *p = c; 344 if (home == NULL || *home == '\0') 345 return (startp); 346 strtodest(home, flag, VSNORMAL, 1); 347 return (p); 348 } 349 p++; 350 } 351 } 352 353 354 static void 355 removerecordregions(int endoff) 356 { 357 if (ifslastp == NULL) 358 return; 359 360 if (ifsfirst.endoff > endoff) { 361 while (ifsfirst.next != NULL) { 362 struct ifsregion *ifsp; 363 INTOFF; 364 ifsp = ifsfirst.next->next; 365 ckfree(ifsfirst.next); 366 ifsfirst.next = ifsp; 367 INTON; 368 } 369 if (ifsfirst.begoff > endoff) 370 ifslastp = NULL; 371 else { 372 ifslastp = &ifsfirst; 373 ifsfirst.endoff = endoff; 374 } 375 return; 376 } 377 378 ifslastp = &ifsfirst; 379 while (ifslastp->next && ifslastp->next->begoff < endoff) 380 ifslastp=ifslastp->next; 381 while (ifslastp->next != NULL) { 382 struct ifsregion *ifsp; 383 INTOFF; 384 ifsp = ifslastp->next->next; 385 ckfree(ifslastp->next); 386 ifslastp->next = ifsp; 387 INTON; 388 } 389 if (ifslastp->endoff > endoff) 390 ifslastp->endoff = endoff; 391 } 392 393 /* 394 * Expand arithmetic expression. 395 * Note that flag is not required as digits never require CTLESC characters. 396 */ 397 static char * 398 expari(char *p) 399 { 400 char *q, *start; 401 arith_t result; 402 int begoff; 403 int quoted; 404 int adj; 405 406 quoted = *p++ == '"'; 407 begoff = expdest - stackblock(); 408 p = argstr(p, 0); 409 removerecordregions(begoff); 410 STPUTC('\0', expdest); 411 start = stackblock() + begoff; 412 413 q = grabstackstr(expdest); 414 result = arith(start); 415 ungrabstackstr(q, expdest); 416 417 start = stackblock() + begoff; 418 adj = start - expdest; 419 STADJUST(adj, expdest); 420 421 CHECKSTRSPACE((int)(DIGITS(result) + 1), expdest); 422 fmtstr(expdest, DIGITS(result), ARITH_FORMAT_STR, result); 423 adj = strlen(expdest); 424 STADJUST(adj, expdest); 425 if (!quoted) 426 recordregion(begoff, expdest - stackblock(), 0); 427 return p; 428 } 429 430 431 /* 432 * Perform command substitution. 433 */ 434 static void 435 expbackq(union node *cmd, int quoted, int flag) 436 { 437 struct backcmd in; 438 int i; 439 char buf[128]; 440 char *p; 441 char *dest = expdest; 442 struct ifsregion saveifs, *savelastp; 443 struct nodelist *saveargbackq; 444 char lastc; 445 int startloc = dest - stackblock(); 446 char const *syntax = quoted? DQSYNTAX : BASESYNTAX; 447 int quotes = flag & (EXP_FULL | EXP_CASE); 448 size_t nnl; 449 450 INTOFF; 451 saveifs = ifsfirst; 452 savelastp = ifslastp; 453 saveargbackq = argbackq; 454 p = grabstackstr(dest); 455 evalbackcmd(cmd, &in); 456 ungrabstackstr(p, dest); 457 ifsfirst = saveifs; 458 ifslastp = savelastp; 459 argbackq = saveargbackq; 460 461 p = in.buf; 462 lastc = '\0'; 463 nnl = 0; 464 /* Don't copy trailing newlines */ 465 for (;;) { 466 if (--in.nleft < 0) { 467 if (in.fd < 0) 468 break; 469 while ((i = read(in.fd, buf, sizeof buf)) < 0 && errno == EINTR); 470 TRACE(("expbackq: read returns %d\n", i)); 471 if (i <= 0) 472 break; 473 p = buf; 474 in.nleft = i - 1; 475 } 476 lastc = *p++; 477 if (lastc != '\0') { 478 if (lastc == '\n') { 479 nnl++; 480 } else { 481 CHECKSTRSPACE(nnl + 2, dest); 482 while (nnl > 0) { 483 nnl--; 484 USTPUTC('\n', dest); 485 } 486 if (quotes && syntax[(int)lastc] == CCTL) 487 USTPUTC(CTLESC, dest); 488 USTPUTC(lastc, dest); 489 } 490 } 491 } 492 493 if (in.fd >= 0) 494 close(in.fd); 495 if (in.buf) 496 ckfree(in.buf); 497 if (in.jp) 498 exitstatus = waitforjob(in.jp, (int *)NULL); 499 if (quoted == 0) 500 recordregion(startloc, dest - stackblock(), 0); 501 TRACE(("expbackq: size=%td: \"%.*s\"\n", 502 ((dest - stackblock()) - startloc), 503 (int)((dest - stackblock()) - startloc), 504 stackblock() + startloc)); 505 expdest = dest; 506 INTON; 507 } 508 509 510 511 static void 512 recordleft(const char *str, const char *loc, char *startp) 513 { 514 int amount; 515 516 amount = ((str - 1) - (loc - startp)) - expdest; 517 STADJUST(amount, expdest); 518 while (loc != str - 1) 519 *startp++ = *loc++; 520 } 521 522 static int 523 subevalvar(char *p, char *str, int strloc, int subtype, int startloc, 524 int varflags, int quotes) 525 { 526 char *startp; 527 char *loc = NULL; 528 char *q; 529 int c = 0; 530 struct nodelist *saveargbackq = argbackq; 531 int amount; 532 533 argstr(p, (subtype == VSTRIMLEFT || subtype == VSTRIMLEFTMAX || 534 subtype == VSTRIMRIGHT || subtype == VSTRIMRIGHTMAX ? 535 EXP_CASE : 0) | EXP_TILDE); 536 STACKSTRNUL(expdest); 537 argbackq = saveargbackq; 538 startp = stackblock() + startloc; 539 if (str == NULL) 540 str = stackblock() + strloc; 541 542 switch (subtype) { 543 case VSASSIGN: 544 setvar(str, startp, 0); 545 amount = startp - expdest; 546 STADJUST(amount, expdest); 547 varflags &= ~VSNUL; 548 return 1; 549 550 case VSQUESTION: 551 if (*p != CTLENDVAR) { 552 outfmt(out2, "%s\n", startp); 553 error((char *)NULL); 554 } 555 error("%.*s: parameter %snot set", (int)(p - str - 1), 556 str, (varflags & VSNUL) ? "null or " : ""); 557 return 0; 558 559 case VSTRIMLEFT: 560 for (loc = startp; loc < str; loc++) { 561 c = *loc; 562 *loc = '\0'; 563 if (patmatch(str, startp, quotes)) { 564 *loc = c; 565 recordleft(str, loc, startp); 566 return 1; 567 } 568 *loc = c; 569 if (quotes && *loc == CTLESC) 570 loc++; 571 } 572 return 0; 573 574 case VSTRIMLEFTMAX: 575 for (loc = str - 1; loc >= startp;) { 576 c = *loc; 577 *loc = '\0'; 578 if (patmatch(str, startp, quotes)) { 579 *loc = c; 580 recordleft(str, loc, startp); 581 return 1; 582 } 583 *loc = c; 584 loc--; 585 if (quotes && loc > startp && *(loc - 1) == CTLESC) { 586 for (q = startp; q < loc; q++) 587 if (*q == CTLESC) 588 q++; 589 if (q > loc) 590 loc--; 591 } 592 } 593 return 0; 594 595 case VSTRIMRIGHT: 596 for (loc = str - 1; loc >= startp;) { 597 if (patmatch(str, loc, quotes)) { 598 amount = loc - expdest; 599 STADJUST(amount, expdest); 600 return 1; 601 } 602 loc--; 603 if (quotes && loc > startp && *(loc - 1) == CTLESC) { 604 for (q = startp; q < loc; q++) 605 if (*q == CTLESC) 606 q++; 607 if (q > loc) 608 loc--; 609 } 610 } 611 return 0; 612 613 case VSTRIMRIGHTMAX: 614 for (loc = startp; loc < str - 1; loc++) { 615 if (patmatch(str, loc, quotes)) { 616 amount = loc - expdest; 617 STADJUST(amount, expdest); 618 return 1; 619 } 620 if (quotes && *loc == CTLESC) 621 loc++; 622 } 623 return 0; 624 625 626 default: 627 abort(); 628 } 629 } 630 631 632 /* 633 * Expand a variable, and return a pointer to the next character in the 634 * input string. 635 */ 636 637 static char * 638 evalvar(char *p, int flag) 639 { 640 int subtype; 641 int varflags; 642 char *var; 643 const char *val; 644 int patloc; 645 int c; 646 int set; 647 int special; 648 int startloc; 649 int varlen; 650 int varlenb; 651 int easy; 652 int quotes = flag & (EXP_FULL | EXP_CASE); 653 int record = 0; 654 655 varflags = (unsigned char)*p++; 656 subtype = varflags & VSTYPE; 657 var = p; 658 special = 0; 659 if (! is_name(*p)) 660 special = 1; 661 p = strchr(p, '=') + 1; 662 again: /* jump here after setting a variable with ${var=text} */ 663 if (varflags & VSLINENO) { 664 set = 1; 665 special = 1; 666 val = NULL; 667 } else if (special) { 668 set = varisset(var, varflags & VSNUL); 669 val = NULL; 670 } else { 671 val = bltinlookup(var, 1); 672 if (val == NULL || ((varflags & VSNUL) && val[0] == '\0')) { 673 val = NULL; 674 set = 0; 675 } else 676 set = 1; 677 } 678 varlen = 0; 679 startloc = expdest - stackblock(); 680 if (!set && uflag && *var != '@' && *var != '*') { 681 switch (subtype) { 682 case VSNORMAL: 683 case VSTRIMLEFT: 684 case VSTRIMLEFTMAX: 685 case VSTRIMRIGHT: 686 case VSTRIMRIGHTMAX: 687 case VSLENGTH: 688 error("%.*s: parameter not set", (int)(p - var - 1), 689 var); 690 } 691 } 692 if (set && subtype != VSPLUS) { 693 /* insert the value of the variable */ 694 if (special) { 695 if (varflags & VSLINENO) 696 STPUTBIN(var, p - var - 1, expdest); 697 else 698 varvalue(var, varflags & VSQUOTE, subtype, flag); 699 if (subtype == VSLENGTH) { 700 varlenb = expdest - stackblock() - startloc; 701 varlen = varlenb; 702 if (localeisutf8) { 703 val = stackblock() + startloc; 704 for (;val != expdest; val++) 705 if ((*val & 0xC0) == 0x80) 706 varlen--; 707 } 708 STADJUST(-varlenb, expdest); 709 } 710 } else { 711 if (subtype == VSLENGTH) { 712 for (;*val; val++) 713 if (!localeisutf8 || 714 (*val & 0xC0) != 0x80) 715 varlen++; 716 } 717 else 718 strtodest(val, flag, subtype, 719 varflags & VSQUOTE); 720 } 721 } 722 723 if (subtype == VSPLUS) 724 set = ! set; 725 726 easy = ((varflags & VSQUOTE) == 0 || 727 (*var == '@' && shellparam.nparam != 1)); 728 729 730 switch (subtype) { 731 case VSLENGTH: 732 expdest = cvtnum(varlen, expdest); 733 record = 1; 734 break; 735 736 case VSNORMAL: 737 record = easy; 738 break; 739 740 case VSPLUS: 741 case VSMINUS: 742 if (!set) { 743 argstr(p, flag | (flag & EXP_FULL ? EXP_SPLIT_LIT : 0) | 744 (varflags & VSQUOTE ? EXP_LIT_QUOTED : 0)); 745 break; 746 } 747 record = easy; 748 break; 749 750 case VSTRIMLEFT: 751 case VSTRIMLEFTMAX: 752 case VSTRIMRIGHT: 753 case VSTRIMRIGHTMAX: 754 if (!set) 755 break; 756 /* 757 * Terminate the string and start recording the pattern 758 * right after it 759 */ 760 STPUTC('\0', expdest); 761 patloc = expdest - stackblock(); 762 if (subevalvar(p, NULL, patloc, subtype, 763 startloc, varflags, quotes) == 0) { 764 int amount = (expdest - stackblock() - patloc) + 1; 765 STADJUST(-amount, expdest); 766 } 767 /* Remove any recorded regions beyond start of variable */ 768 removerecordregions(startloc); 769 record = 1; 770 break; 771 772 case VSASSIGN: 773 case VSQUESTION: 774 if (!set) { 775 if (subevalvar(p, var, 0, subtype, startloc, varflags, 776 quotes)) { 777 varflags &= ~VSNUL; 778 /* 779 * Remove any recorded regions beyond 780 * start of variable 781 */ 782 removerecordregions(startloc); 783 goto again; 784 } 785 break; 786 } 787 record = easy; 788 break; 789 790 case VSERROR: 791 c = p - var - 1; 792 error("${%.*s%s}: Bad substitution", c, var, 793 (c > 0 && *p != CTLENDVAR) ? "..." : ""); 794 795 default: 796 abort(); 797 } 798 799 if (record) 800 recordregion(startloc, expdest - stackblock(), 801 varflags & VSQUOTE || (ifsset() && ifsval()[0] == '\0' && 802 (*var == '@' || *var == '*'))); 803 804 if (subtype != VSNORMAL) { /* skip to end of alternative */ 805 int nesting = 1; 806 for (;;) { 807 if ((c = *p++) == CTLESC) 808 p++; 809 else if (c == CTLBACKQ || c == (CTLBACKQ|CTLQUOTE)) { 810 if (set) 811 argbackq = argbackq->next; 812 } else if (c == CTLVAR) { 813 if ((*p++ & VSTYPE) != VSNORMAL) 814 nesting++; 815 } else if (c == CTLENDVAR) { 816 if (--nesting == 0) 817 break; 818 } 819 } 820 } 821 return p; 822 } 823 824 825 826 /* 827 * Test whether a specialized variable is set. 828 */ 829 830 static int 831 varisset(const char *name, int nulok) 832 { 833 834 if (*name == '!') 835 return backgndpidset(); 836 else if (*name == '@' || *name == '*') { 837 if (*shellparam.p == NULL) 838 return 0; 839 840 if (nulok) { 841 char **av; 842 843 for (av = shellparam.p; *av; av++) 844 if (**av != '\0') 845 return 1; 846 return 0; 847 } 848 } else if (is_digit(*name)) { 849 char *ap; 850 long num; 851 852 errno = 0; 853 num = strtol(name, NULL, 10); 854 if (errno != 0 || num > shellparam.nparam) 855 return 0; 856 857 if (num == 0) 858 ap = arg0; 859 else 860 ap = shellparam.p[num - 1]; 861 862 if (nulok && (ap == NULL || *ap == '\0')) 863 return 0; 864 } 865 return 1; 866 } 867 868 static void 869 strtodest(const char *p, int flag, int subtype, int quoted) 870 { 871 if (flag & (EXP_FULL | EXP_CASE) && subtype != VSLENGTH) 872 STPUTS_QUOTES(p, quoted ? DQSYNTAX : BASESYNTAX, expdest); 873 else 874 STPUTS(p, expdest); 875 } 876 877 /* 878 * Add the value of a specialized variable to the stack string. 879 */ 880 881 static void 882 varvalue(const char *name, int quoted, int subtype, int flag) 883 { 884 int num; 885 char *p; 886 int i; 887 char sep[2]; 888 char **ap; 889 890 switch (*name) { 891 case '$': 892 num = rootpid; 893 break; 894 case '?': 895 num = oexitstatus; 896 break; 897 case '#': 898 num = shellparam.nparam; 899 break; 900 case '!': 901 num = backgndpidval(); 902 break; 903 case '-': 904 for (i = 0 ; i < NSHORTOPTS ; i++) { 905 if (optlist[i].val) 906 STPUTC(optlist[i].letter, expdest); 907 } 908 return; 909 case '@': 910 if (flag & EXP_FULL && quoted) { 911 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 912 strtodest(p, flag, subtype, quoted); 913 if (*ap) 914 STPUTC('\0', expdest); 915 } 916 return; 917 } 918 /* FALLTHROUGH */ 919 case '*': 920 if (ifsset()) 921 sep[0] = ifsval()[0]; 922 else 923 sep[0] = ' '; 924 sep[1] = '\0'; 925 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 926 strtodest(p, flag, subtype, quoted); 927 if (!*ap) 928 break; 929 if (sep[0]) 930 strtodest(sep, flag, subtype, quoted); 931 else if (flag & EXP_FULL && !quoted && **ap != '\0') 932 STPUTC('\0', expdest); 933 } 934 return; 935 default: 936 if (is_digit(*name)) { 937 num = atoi(name); 938 if (num == 0) 939 p = arg0; 940 else if (num > 0 && num <= shellparam.nparam) 941 p = shellparam.p[num - 1]; 942 else 943 return; 944 strtodest(p, flag, subtype, quoted); 945 } 946 return; 947 } 948 expdest = cvtnum(num, expdest); 949 } 950 951 952 953 /* 954 * Record the fact that we have to scan this region of the 955 * string for IFS characters. 956 */ 957 958 static void 959 recordregion(int start, int end, int inquotes) 960 { 961 struct ifsregion *ifsp; 962 963 INTOFF; 964 if (ifslastp == NULL) { 965 ifsp = &ifsfirst; 966 } else { 967 if (ifslastp->endoff == start 968 && ifslastp->inquotes == inquotes) { 969 /* extend previous area */ 970 ifslastp->endoff = end; 971 INTON; 972 return; 973 } 974 ifsp = (struct ifsregion *)ckmalloc(sizeof (struct ifsregion)); 975 ifslastp->next = ifsp; 976 } 977 ifslastp = ifsp; 978 ifslastp->next = NULL; 979 ifslastp->begoff = start; 980 ifslastp->endoff = end; 981 ifslastp->inquotes = inquotes; 982 INTON; 983 } 984 985 986 987 /* 988 * Break the argument string into pieces based upon IFS and add the 989 * strings to the argument list. The regions of the string to be 990 * searched for IFS characters have been stored by recordregion. 991 * CTLESC characters are preserved but have little effect in this pass 992 * other than escaping CTL* characters. In particular, they do not escape 993 * IFS characters: that should be done with the ifsregion mechanism. 994 * CTLQUOTEMARK characters are used to preserve empty quoted strings. 995 * This pass treats them as a regular character, making the string non-empty. 996 * Later, they are removed along with the other CTL* characters. 997 */ 998 static void 999 ifsbreakup(char *string, struct arglist *arglist) 1000 { 1001 struct ifsregion *ifsp; 1002 char *start; 1003 char *p; 1004 char *q; 1005 const char *ifs; 1006 const char *ifsspc; 1007 int had_param_ch = 0; 1008 1009 start = string; 1010 1011 if (ifslastp == NULL) { 1012 /* Return entire argument, IFS doesn't apply to any of it */ 1013 appendarglist(arglist, start); 1014 return; 1015 } 1016 1017 ifs = ifsset() ? ifsval() : " \t\n"; 1018 1019 for (ifsp = &ifsfirst; ifsp != NULL; ifsp = ifsp->next) { 1020 p = string + ifsp->begoff; 1021 while (p < string + ifsp->endoff) { 1022 q = p; 1023 if (*p == CTLESC) 1024 p++; 1025 if (ifsp->inquotes) { 1026 /* Only NULs (should be from "$@") end args */ 1027 had_param_ch = 1; 1028 if (*p != 0) { 1029 p++; 1030 continue; 1031 } 1032 ifsspc = NULL; 1033 } else { 1034 if (!strchr(ifs, *p)) { 1035 had_param_ch = 1; 1036 p++; 1037 continue; 1038 } 1039 ifsspc = strchr(" \t\n", *p); 1040 1041 /* Ignore IFS whitespace at start */ 1042 if (q == start && ifsspc != NULL) { 1043 p++; 1044 start = p; 1045 continue; 1046 } 1047 had_param_ch = 0; 1048 } 1049 1050 /* Save this argument... */ 1051 *q = '\0'; 1052 appendarglist(arglist, start); 1053 p++; 1054 1055 if (ifsspc != NULL) { 1056 /* Ignore further trailing IFS whitespace */ 1057 for (; p < string + ifsp->endoff; p++) { 1058 q = p; 1059 if (*p == CTLESC) 1060 p++; 1061 if (strchr(ifs, *p) == NULL) { 1062 p = q; 1063 break; 1064 } 1065 if (strchr(" \t\n", *p) == NULL) { 1066 p++; 1067 break; 1068 } 1069 } 1070 } 1071 start = p; 1072 } 1073 } 1074 1075 /* 1076 * Save anything left as an argument. 1077 * Traditionally we have treated 'IFS=':'; set -- x$IFS' as 1078 * generating 2 arguments, the second of which is empty. 1079 * Some recent clarification of the Posix spec say that it 1080 * should only generate one.... 1081 */ 1082 if (had_param_ch || *start != 0) 1083 appendarglist(arglist, start); 1084 } 1085 1086 1087 static char expdir[PATH_MAX]; 1088 #define expdir_end (expdir + sizeof(expdir)) 1089 1090 /* 1091 * Perform pathname generation and remove control characters. 1092 * At this point, the only control characters should be CTLESC and CTLQUOTEMARK. 1093 * The results are stored in the list dstlist. 1094 */ 1095 static void 1096 expandmeta(struct arglist *srclist, struct arglist *dstlist) 1097 { 1098 char *p; 1099 int firstmatch; 1100 int i; 1101 char c; 1102 1103 for (i = 0; i < srclist->count; i++) { 1104 firstmatch = dstlist->count; 1105 if (!fflag) { 1106 p = srclist->args[i]; 1107 for (; (c = *p) != '\0'; p++) { 1108 /* fast check for meta chars */ 1109 if (c == '*' || c == '?' || c == '[') { 1110 INTOFF; 1111 expmeta(expdir, srclist->args[i], 1112 dstlist); 1113 INTON; 1114 break; 1115 } 1116 } 1117 } 1118 if (dstlist->count == firstmatch) { 1119 /* 1120 * no matches 1121 */ 1122 rmescapes(srclist->args[i]); 1123 appendarglist(dstlist, srclist->args[i]); 1124 } else { 1125 qsort(&dstlist->args[firstmatch], 1126 dstlist->count - firstmatch, 1127 sizeof(dstlist->args[0]), expsortcmp); 1128 } 1129 } 1130 } 1131 1132 1133 /* 1134 * Do metacharacter (i.e. *, ?, [...]) expansion. 1135 */ 1136 1137 static void 1138 expmeta(char *enddir, char *name, struct arglist *arglist) 1139 { 1140 const char *p; 1141 const char *q; 1142 const char *start; 1143 char *endname; 1144 int metaflag; 1145 struct stat statb; 1146 DIR *dirp; 1147 struct dirent *dp; 1148 int atend; 1149 int matchdot; 1150 int esc; 1151 int namlen; 1152 1153 metaflag = 0; 1154 start = name; 1155 for (p = name; esc = 0, *p; p += esc + 1) { 1156 if (*p == '*' || *p == '?') 1157 metaflag = 1; 1158 else if (*p == '[') { 1159 q = p + 1; 1160 if (*q == '!' || *q == '^') 1161 q++; 1162 for (;;) { 1163 while (*q == CTLQUOTEMARK) 1164 q++; 1165 if (*q == CTLESC) 1166 q++; 1167 if (*q == '/' || *q == '\0') 1168 break; 1169 if (*++q == ']') { 1170 metaflag = 1; 1171 break; 1172 } 1173 } 1174 } else if (*p == '\0') 1175 break; 1176 else if (*p == CTLQUOTEMARK) 1177 continue; 1178 else { 1179 if (*p == CTLESC) 1180 esc++; 1181 if (p[esc] == '/') { 1182 if (metaflag) 1183 break; 1184 start = p + esc + 1; 1185 } 1186 } 1187 } 1188 if (metaflag == 0) { /* we've reached the end of the file name */ 1189 if (enddir != expdir) 1190 metaflag++; 1191 for (p = name ; ; p++) { 1192 if (*p == CTLQUOTEMARK) 1193 continue; 1194 if (*p == CTLESC) 1195 p++; 1196 *enddir++ = *p; 1197 if (*p == '\0') 1198 break; 1199 if (enddir == expdir_end) 1200 return; 1201 } 1202 if (metaflag == 0 || lstat(expdir, &statb) >= 0) 1203 appendarglist(arglist, stsavestr(expdir)); 1204 return; 1205 } 1206 endname = name + (p - name); 1207 if (start != name) { 1208 p = name; 1209 while (p < start) { 1210 while (*p == CTLQUOTEMARK) 1211 p++; 1212 if (*p == CTLESC) 1213 p++; 1214 *enddir++ = *p++; 1215 if (enddir == expdir_end) 1216 return; 1217 } 1218 } 1219 if (enddir == expdir) { 1220 p = "."; 1221 } else if (enddir == expdir + 1 && *expdir == '/') { 1222 p = "/"; 1223 } else { 1224 p = expdir; 1225 enddir[-1] = '\0'; 1226 } 1227 if ((dirp = opendir(p)) == NULL) 1228 return; 1229 if (enddir != expdir) 1230 enddir[-1] = '/'; 1231 if (*endname == 0) { 1232 atend = 1; 1233 } else { 1234 atend = 0; 1235 *endname = '\0'; 1236 endname += esc + 1; 1237 } 1238 matchdot = 0; 1239 p = start; 1240 while (*p == CTLQUOTEMARK) 1241 p++; 1242 if (*p == CTLESC) 1243 p++; 1244 if (*p == '.') 1245 matchdot++; 1246 while (! int_pending() && (dp = readdir(dirp)) != NULL) { 1247 if (dp->d_name[0] == '.' && ! matchdot) 1248 continue; 1249 if (patmatch(start, dp->d_name, 0)) { 1250 namlen = dp->d_namlen; 1251 if (enddir + namlen + 1 > expdir_end) 1252 continue; 1253 memcpy(enddir, dp->d_name, namlen + 1); 1254 if (atend) 1255 appendarglist(arglist, stsavestr(expdir)); 1256 else { 1257 if (dp->d_type != DT_UNKNOWN && 1258 dp->d_type != DT_DIR && 1259 dp->d_type != DT_LNK) 1260 continue; 1261 if (enddir + namlen + 2 > expdir_end) 1262 continue; 1263 enddir[namlen] = '/'; 1264 enddir[namlen + 1] = '\0'; 1265 expmeta(enddir + namlen + 1, endname, arglist); 1266 } 1267 } 1268 } 1269 closedir(dirp); 1270 if (! atend) 1271 endname[-esc - 1] = esc ? CTLESC : '/'; 1272 } 1273 1274 1275 static int 1276 expsortcmp(const void *p1, const void *p2) 1277 { 1278 const char *s1 = *(const char * const *)p1; 1279 const char *s2 = *(const char * const *)p2; 1280 1281 return (strcmp(s1, s2)); 1282 } 1283 1284 1285 1286 static wchar_t 1287 get_wc(const char **p) 1288 { 1289 wchar_t c; 1290 int chrlen; 1291 1292 chrlen = mbtowc(&c, *p, 4); 1293 if (chrlen == 0) 1294 return 0; 1295 else if (chrlen == -1) 1296 c = 0; 1297 else 1298 *p += chrlen; 1299 return c; 1300 } 1301 1302 1303 /* 1304 * See if a character matches a character class, starting at the first colon 1305 * of "[:class:]". 1306 * If a valid character class is recognized, a pointer to the next character 1307 * after the final closing bracket is stored into *end, otherwise a null 1308 * pointer is stored into *end. 1309 */ 1310 static int 1311 match_charclass(const char *p, wchar_t chr, const char **end) 1312 { 1313 char name[20]; 1314 const char *nameend; 1315 wctype_t cclass; 1316 1317 *end = NULL; 1318 p++; 1319 nameend = strstr(p, ":]"); 1320 if (nameend == NULL || (size_t)(nameend - p) >= sizeof(name) || 1321 nameend == p) 1322 return 0; 1323 memcpy(name, p, nameend - p); 1324 name[nameend - p] = '\0'; 1325 *end = nameend + 2; 1326 cclass = wctype(name); 1327 /* An unknown class matches nothing but is valid nevertheless. */ 1328 if (cclass == 0) 1329 return 0; 1330 return iswctype(chr, cclass); 1331 } 1332 1333 1334 /* 1335 * Returns true if the pattern matches the string. 1336 */ 1337 1338 static int 1339 patmatch(const char *pattern, const char *string, int squoted) 1340 { 1341 const char *p, *q, *end; 1342 const char *bt_p, *bt_q; 1343 char c; 1344 wchar_t wc, wc2; 1345 1346 p = pattern; 1347 q = string; 1348 bt_p = NULL; 1349 bt_q = NULL; 1350 for (;;) { 1351 switch (c = *p++) { 1352 case '\0': 1353 if (*q != '\0') 1354 goto backtrack; 1355 return 1; 1356 case CTLESC: 1357 if (squoted && *q == CTLESC) 1358 q++; 1359 if (*q++ != *p++) 1360 goto backtrack; 1361 break; 1362 case CTLQUOTEMARK: 1363 continue; 1364 case '?': 1365 if (squoted && *q == CTLESC) 1366 q++; 1367 if (*q == '\0') 1368 return 0; 1369 if (localeisutf8) { 1370 wc = get_wc(&q); 1371 /* 1372 * A '?' does not match invalid UTF-8 but a 1373 * '*' does, so backtrack. 1374 */ 1375 if (wc == 0) 1376 goto backtrack; 1377 } else 1378 wc = (unsigned char)*q++; 1379 break; 1380 case '*': 1381 c = *p; 1382 while (c == CTLQUOTEMARK || c == '*') 1383 c = *++p; 1384 /* 1385 * If the pattern ends here, we know the string 1386 * matches without needing to look at the rest of it. 1387 */ 1388 if (c == '\0') 1389 return 1; 1390 /* 1391 * First try the shortest match for the '*' that 1392 * could work. We can forget any earlier '*' since 1393 * there is no way having it match more characters 1394 * can help us, given that we are already here. 1395 */ 1396 bt_p = p; 1397 bt_q = q; 1398 break; 1399 case '[': { 1400 const char *savep, *saveq; 1401 int invert, found; 1402 wchar_t chr; 1403 1404 savep = p, saveq = q; 1405 invert = 0; 1406 if (*p == '!' || *p == '^') { 1407 invert++; 1408 p++; 1409 } 1410 found = 0; 1411 if (squoted && *q == CTLESC) 1412 q++; 1413 if (*q == '\0') 1414 return 0; 1415 if (localeisutf8) { 1416 chr = get_wc(&q); 1417 if (chr == 0) 1418 goto backtrack; 1419 } else 1420 chr = (unsigned char)*q++; 1421 c = *p++; 1422 do { 1423 if (c == '\0') { 1424 p = savep, q = saveq; 1425 c = '['; 1426 goto dft; 1427 } 1428 if (c == CTLQUOTEMARK) 1429 continue; 1430 if (c == '[' && *p == ':') { 1431 found |= match_charclass(p, chr, &end); 1432 if (end != NULL) 1433 p = end; 1434 } 1435 if (c == CTLESC) 1436 c = *p++; 1437 if (localeisutf8 && c & 0x80) { 1438 p--; 1439 wc = get_wc(&p); 1440 if (wc == 0) /* bad utf-8 */ 1441 return 0; 1442 } else 1443 wc = (unsigned char)c; 1444 if (*p == '-' && p[1] != ']') { 1445 p++; 1446 while (*p == CTLQUOTEMARK) 1447 p++; 1448 if (*p == CTLESC) 1449 p++; 1450 if (localeisutf8) { 1451 wc2 = get_wc(&p); 1452 if (wc2 == 0) /* bad utf-8 */ 1453 return 0; 1454 } else 1455 wc2 = (unsigned char)*p++; 1456 if ( collate_range_cmp(chr, wc) >= 0 1457 && collate_range_cmp(chr, wc2) <= 0 1458 ) 1459 found = 1; 1460 } else { 1461 if (chr == wc) 1462 found = 1; 1463 } 1464 } while ((c = *p++) != ']'); 1465 if (found == invert) 1466 goto backtrack; 1467 break; 1468 } 1469 dft: default: 1470 if (squoted && *q == CTLESC) 1471 q++; 1472 if (*q == '\0') 1473 return 0; 1474 if (*q++ == c) 1475 break; 1476 backtrack: 1477 /* 1478 * If we have a mismatch (other than hitting the end 1479 * of the string), go back to the last '*' seen and 1480 * have it match one additional character. 1481 */ 1482 if (bt_p == NULL) 1483 return 0; 1484 if (squoted && *bt_q == CTLESC) 1485 bt_q++; 1486 if (*bt_q == '\0') 1487 return 0; 1488 bt_q++; 1489 p = bt_p; 1490 q = bt_q; 1491 break; 1492 } 1493 } 1494 } 1495 1496 1497 1498 /* 1499 * Remove any CTLESC and CTLQUOTEMARK characters from a string. 1500 */ 1501 1502 void 1503 rmescapes(char *str) 1504 { 1505 char *p, *q; 1506 1507 p = str; 1508 while (*p != CTLESC && *p != CTLQUOTEMARK && *p != CTLQUOTEEND) { 1509 if (*p++ == '\0') 1510 return; 1511 } 1512 q = p; 1513 while (*p) { 1514 if (*p == CTLQUOTEMARK || *p == CTLQUOTEEND) { 1515 p++; 1516 continue; 1517 } 1518 if (*p == CTLESC) 1519 p++; 1520 *q++ = *p++; 1521 } 1522 *q = '\0'; 1523 } 1524 1525 1526 1527 /* 1528 * See if a pattern matches in a case statement. 1529 */ 1530 1531 int 1532 casematch(union node *pattern, const char *val) 1533 { 1534 struct stackmark smark; 1535 int result; 1536 char *p; 1537 1538 setstackmark(&smark); 1539 argbackq = pattern->narg.backquote; 1540 STARTSTACKSTR(expdest); 1541 ifslastp = NULL; 1542 argstr(pattern->narg.text, EXP_TILDE | EXP_CASE); 1543 STPUTC('\0', expdest); 1544 p = grabstackstr(expdest); 1545 result = patmatch(p, val, 0); 1546 popstackmark(&smark); 1547 return result; 1548 } 1549 1550 /* 1551 * Our own itoa(). 1552 */ 1553 1554 static char * 1555 cvtnum(int num, char *buf) 1556 { 1557 char temp[32]; 1558 int neg = num < 0; 1559 char *p = temp + 31; 1560 1561 temp[31] = '\0'; 1562 1563 do { 1564 *--p = num % 10 + '0'; 1565 } while ((num /= 10) != 0); 1566 1567 if (neg) 1568 *--p = '-'; 1569 1570 STPUTS(p, buf); 1571 return buf; 1572 } 1573 1574 /* 1575 * Do most of the work for wordexp(3). 1576 */ 1577 1578 int 1579 wordexpcmd(int argc, char **argv) 1580 { 1581 size_t len; 1582 int i; 1583 1584 out1fmt("%08x", argc - 1); 1585 for (i = 1, len = 0; i < argc; i++) 1586 len += strlen(argv[i]); 1587 out1fmt("%08x", (int)len); 1588 for (i = 1; i < argc; i++) 1589 outbin(argv[i], strlen(argv[i]) + 1, out1); 1590 return (0); 1591 } 1592 1593 /* 1594 * Do most of the work for wordexp(3), new version. 1595 */ 1596 1597 int 1598 freebsd_wordexpcmd(int argc __unused, char **argv __unused) 1599 { 1600 struct arglist arglist; 1601 union node *args, *n; 1602 size_t len; 1603 int ch; 1604 int protected = 0; 1605 int fd = -1; 1606 int i; 1607 1608 while ((ch = nextopt("f:p")) != '\0') { 1609 switch (ch) { 1610 case 'f': 1611 fd = number(shoptarg); 1612 break; 1613 case 'p': 1614 protected = 1; 1615 break; 1616 } 1617 } 1618 if (*argptr != NULL) 1619 error("wrong number of arguments"); 1620 if (fd < 0) 1621 error("missing fd"); 1622 INTOFF; 1623 setinputfd(fd, 1); 1624 INTON; 1625 args = parsewordexp(); 1626 popfile(); /* will also close fd */ 1627 if (protected) 1628 for (n = args; n != NULL; n = n->narg.next) { 1629 if (n->narg.backquote != NULL) { 1630 outcslow('C', out1); 1631 error("command substitution disabled"); 1632 } 1633 } 1634 outcslow(' ', out1); 1635 emptyarglist(&arglist); 1636 for (n = args; n != NULL; n = n->narg.next) 1637 expandarg(n, &arglist, EXP_FULL | EXP_TILDE); 1638 for (i = 0, len = 0; i < arglist.count; i++) 1639 len += strlen(arglist.args[i]); 1640 out1fmt("%016x %016zx", arglist.count, len); 1641 for (i = 0; i < arglist.count; i++) 1642 outbin(arglist.args[i], strlen(arglist.args[i]) + 1, out1); 1643 return (0); 1644 } 1645