expand.c revision 287750
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 37static char sccsid[] = "@(#)expand.c 8.5 (Berkeley) 5/15/95"; 38#endif 39#endif /* not lint */ 40#include <sys/cdefs.h> 41__FBSDID("$FreeBSD: stable/10/bin/sh/expand.c 287750 2015-09-13 13:43:08Z jilles $"); 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 87struct 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 95static char *expdest; /* output of current string */ 96static struct nodelist *argbackq; /* list of back quote expressions */ 97static struct ifsregion ifsfirst; /* first struct in list of ifs regions */ 98static struct ifsregion *ifslastp; /* last struct in list */ 99static struct arglist exparg; /* holds expanded arg list */ 100 101static char *argstr(char *, int); 102static char *exptilde(char *, int); 103static char *expari(char *); 104static void expbackq(union node *, int, int); 105static int subevalvar(char *, char *, int, int, int, int, int); 106static char *evalvar(char *, int); 107static int varisset(const char *, int); 108static void varvalue(const char *, int, int, int); 109static void recordregion(int, int, int); 110static void removerecordregions(int); 111static void ifsbreakup(char *, struct arglist *); 112static void expandmeta(struct strlist *, int); 113static void expmeta(char *, char *); 114static void addfname(char *); 115static struct strlist *expsort(struct strlist *); 116static struct strlist *msort(struct strlist *, int); 117static int patmatch(const char *, const char *, int); 118static char *cvtnum(int, char *); 119static int collate_range_cmp(wchar_t, wchar_t); 120 121static int 122collate_range_cmp(wchar_t c1, wchar_t c2) 123{ 124 static wchar_t s1[2], s2[2]; 125 126 s1[0] = c1; 127 s2[0] = c2; 128 return (wcscoll(s1, s2)); 129} 130 131static char * 132stputs_quotes(const char *data, const char *syntax, char *p) 133{ 134 while (*data) { 135 CHECKSTRSPACE(2, p); 136 if (syntax[(int)*data] == CCTL) 137 USTPUTC(CTLESC, p); 138 USTPUTC(*data++, p); 139 } 140 return (p); 141} 142#define STPUTS_QUOTES(data, syntax, p) p = stputs_quotes((data), syntax, p) 143 144/* 145 * Perform expansions on an argument, placing the resulting list of arguments 146 * in arglist. Parameter expansion, command substitution and arithmetic 147 * expansion are always performed; additional expansions can be requested 148 * via flag (EXP_*). 149 * The result is left in the stack string. 150 * When arglist is NULL, perform here document expansion. 151 * 152 * Caution: this function uses global state and is not reentrant. 153 * However, a new invocation after an interrupted invocation is safe 154 * and will reset the global state for the new call. 155 */ 156void 157expandarg(union node *arg, struct arglist *arglist, int flag) 158{ 159 struct strlist *sp; 160 char *p; 161 162 argbackq = arg->narg.backquote; 163 STARTSTACKSTR(expdest); 164 ifsfirst.next = NULL; 165 ifslastp = NULL; 166 argstr(arg->narg.text, flag); 167 if (arglist == NULL) { 168 STACKSTRNUL(expdest); 169 return; /* here document expanded */ 170 } 171 STPUTC('\0', expdest); 172 p = grabstackstr(expdest); 173 exparg.lastp = &exparg.list; 174 if (flag & EXP_FULL) { 175 ifsbreakup(p, &exparg); 176 *exparg.lastp = NULL; 177 exparg.lastp = &exparg.list; 178 expandmeta(exparg.list, flag); 179 } else { 180 sp = (struct strlist *)stalloc(sizeof (struct strlist)); 181 sp->text = p; 182 *exparg.lastp = sp; 183 exparg.lastp = &sp->next; 184 } 185 while (ifsfirst.next != NULL) { 186 struct ifsregion *ifsp; 187 INTOFF; 188 ifsp = ifsfirst.next->next; 189 ckfree(ifsfirst.next); 190 ifsfirst.next = ifsp; 191 INTON; 192 } 193 *exparg.lastp = NULL; 194 if (exparg.list) { 195 *arglist->lastp = exparg.list; 196 arglist->lastp = exparg.lastp; 197 } 198} 199 200 201 202/* 203 * Perform parameter expansion, command substitution and arithmetic 204 * expansion, and tilde expansion if requested via EXP_TILDE/EXP_VARTILDE. 205 * Processing ends at a CTLENDVAR or CTLENDARI character as well as '\0'. 206 * This is used to expand word in ${var+word} etc. 207 * If EXP_FULL or EXP_CASE are set, keep and/or generate CTLESC 208 * characters to allow for further processing. 209 * If EXP_FULL is set, also preserve CTLQUOTEMARK characters. 210 */ 211static char * 212argstr(char *p, int flag) 213{ 214 char c; 215 int quotes = flag & (EXP_FULL | EXP_CASE); /* do CTLESC */ 216 int firsteq = 1; 217 int split_lit; 218 int lit_quoted; 219 220 split_lit = flag & EXP_SPLIT_LIT; 221 lit_quoted = flag & EXP_LIT_QUOTED; 222 flag &= ~(EXP_SPLIT_LIT | EXP_LIT_QUOTED); 223 if (*p == '~' && (flag & (EXP_TILDE | EXP_VARTILDE))) 224 p = exptilde(p, flag); 225 for (;;) { 226 CHECKSTRSPACE(2, expdest); 227 switch (c = *p++) { 228 case '\0': 229 return (p - 1); 230 case CTLENDVAR: 231 case CTLENDARI: 232 return (p); 233 case CTLQUOTEMARK: 234 lit_quoted = 1; 235 /* "$@" syntax adherence hack */ 236 if (p[0] == CTLVAR && p[2] == '@' && p[3] == '=') 237 break; 238 if ((flag & EXP_FULL) != 0) 239 USTPUTC(c, expdest); 240 break; 241 case CTLQUOTEEND: 242 lit_quoted = 0; 243 break; 244 case CTLESC: 245 if (quotes) 246 USTPUTC(c, expdest); 247 c = *p++; 248 USTPUTC(c, expdest); 249 if (split_lit && !lit_quoted) 250 recordregion(expdest - stackblock() - 251 (quotes ? 2 : 1), 252 expdest - stackblock(), 0); 253 break; 254 case CTLVAR: 255 p = evalvar(p, flag); 256 break; 257 case CTLBACKQ: 258 case CTLBACKQ|CTLQUOTE: 259 expbackq(argbackq->n, c & CTLQUOTE, flag); 260 argbackq = argbackq->next; 261 break; 262 case CTLARI: 263 p = expari(p); 264 break; 265 case ':': 266 case '=': 267 /* 268 * sort of a hack - expand tildes in variable 269 * assignments (after the first '=' and after ':'s). 270 */ 271 USTPUTC(c, expdest); 272 if (split_lit && !lit_quoted) 273 recordregion(expdest - stackblock() - 1, 274 expdest - stackblock(), 0); 275 if (flag & EXP_VARTILDE && *p == '~' && 276 (c != '=' || firsteq)) { 277 if (c == '=') 278 firsteq = 0; 279 p = exptilde(p, flag); 280 } 281 break; 282 default: 283 USTPUTC(c, expdest); 284 if (split_lit && !lit_quoted) 285 recordregion(expdest - stackblock() - 1, 286 expdest - stackblock(), 0); 287 } 288 } 289} 290 291/* 292 * Perform tilde expansion, placing the result in the stack string and 293 * returning the next position in the input string to process. 294 */ 295static char * 296exptilde(char *p, int flag) 297{ 298 char c, *startp = p; 299 struct passwd *pw; 300 char *home; 301 int quotes = flag & (EXP_FULL | EXP_CASE); 302 303 while ((c = *p) != '\0') { 304 switch(c) { 305 case CTLESC: /* This means CTL* are always considered quoted. */ 306 case CTLVAR: 307 case CTLBACKQ: 308 case CTLBACKQ | CTLQUOTE: 309 case CTLARI: 310 case CTLENDARI: 311 case CTLQUOTEMARK: 312 return (startp); 313 case ':': 314 if (flag & EXP_VARTILDE) 315 goto done; 316 break; 317 case '/': 318 case CTLENDVAR: 319 goto done; 320 } 321 p++; 322 } 323done: 324 *p = '\0'; 325 if (*(startp+1) == '\0') { 326 home = lookupvar("HOME"); 327 } else { 328 pw = getpwnam(startp+1); 329 home = pw != NULL ? pw->pw_dir : NULL; 330 } 331 *p = c; 332 if (home == NULL || *home == '\0') 333 return (startp); 334 if (quotes) 335 STPUTS_QUOTES(home, DQSYNTAX, expdest); 336 else 337 STPUTS(home, expdest); 338 return (p); 339} 340 341 342static void 343removerecordregions(int endoff) 344{ 345 if (ifslastp == NULL) 346 return; 347 348 if (ifsfirst.endoff > endoff) { 349 while (ifsfirst.next != NULL) { 350 struct ifsregion *ifsp; 351 INTOFF; 352 ifsp = ifsfirst.next->next; 353 ckfree(ifsfirst.next); 354 ifsfirst.next = ifsp; 355 INTON; 356 } 357 if (ifsfirst.begoff > endoff) 358 ifslastp = NULL; 359 else { 360 ifslastp = &ifsfirst; 361 ifsfirst.endoff = endoff; 362 } 363 return; 364 } 365 366 ifslastp = &ifsfirst; 367 while (ifslastp->next && ifslastp->next->begoff < endoff) 368 ifslastp=ifslastp->next; 369 while (ifslastp->next != NULL) { 370 struct ifsregion *ifsp; 371 INTOFF; 372 ifsp = ifslastp->next->next; 373 ckfree(ifslastp->next); 374 ifslastp->next = ifsp; 375 INTON; 376 } 377 if (ifslastp->endoff > endoff) 378 ifslastp->endoff = endoff; 379} 380 381/* 382 * Expand arithmetic expression. 383 * Note that flag is not required as digits never require CTLESC characters. 384 */ 385static char * 386expari(char *p) 387{ 388 char *q, *start; 389 arith_t result; 390 int begoff; 391 int quoted; 392 int adj; 393 394 quoted = *p++ == '"'; 395 begoff = expdest - stackblock(); 396 p = argstr(p, 0); 397 removerecordregions(begoff); 398 STPUTC('\0', expdest); 399 start = stackblock() + begoff; 400 401 q = grabstackstr(expdest); 402 result = arith(start); 403 ungrabstackstr(q, expdest); 404 405 start = stackblock() + begoff; 406 adj = start - expdest; 407 STADJUST(adj, expdest); 408 409 CHECKSTRSPACE((int)(DIGITS(result) + 1), expdest); 410 fmtstr(expdest, DIGITS(result), ARITH_FORMAT_STR, result); 411 adj = strlen(expdest); 412 STADJUST(adj, expdest); 413 if (!quoted) 414 recordregion(begoff, expdest - stackblock(), 0); 415 return p; 416} 417 418 419/* 420 * Perform command substitution. 421 */ 422static void 423expbackq(union node *cmd, int quoted, int flag) 424{ 425 struct backcmd in; 426 int i; 427 char buf[128]; 428 char *p; 429 char *dest = expdest; 430 struct ifsregion saveifs, *savelastp; 431 struct nodelist *saveargbackq; 432 char lastc; 433 int startloc = dest - stackblock(); 434 char const *syntax = quoted? DQSYNTAX : BASESYNTAX; 435 int quotes = flag & (EXP_FULL | EXP_CASE); 436 size_t nnl; 437 438 INTOFF; 439 saveifs = ifsfirst; 440 savelastp = ifslastp; 441 saveargbackq = argbackq; 442 p = grabstackstr(dest); 443 evalbackcmd(cmd, &in); 444 ungrabstackstr(p, dest); 445 ifsfirst = saveifs; 446 ifslastp = savelastp; 447 argbackq = saveargbackq; 448 449 p = in.buf; 450 lastc = '\0'; 451 nnl = 0; 452 /* Don't copy trailing newlines */ 453 for (;;) { 454 if (--in.nleft < 0) { 455 if (in.fd < 0) 456 break; 457 while ((i = read(in.fd, buf, sizeof buf)) < 0 && errno == EINTR); 458 TRACE(("expbackq: read returns %d\n", i)); 459 if (i <= 0) 460 break; 461 p = buf; 462 in.nleft = i - 1; 463 } 464 lastc = *p++; 465 if (lastc != '\0') { 466 if (lastc == '\n') { 467 nnl++; 468 } else { 469 CHECKSTRSPACE(nnl + 2, dest); 470 while (nnl > 0) { 471 nnl--; 472 USTPUTC('\n', dest); 473 } 474 if (quotes && syntax[(int)lastc] == CCTL) 475 USTPUTC(CTLESC, dest); 476 USTPUTC(lastc, dest); 477 } 478 } 479 } 480 481 if (in.fd >= 0) 482 close(in.fd); 483 if (in.buf) 484 ckfree(in.buf); 485 if (in.jp) 486 exitstatus = waitforjob(in.jp, (int *)NULL); 487 if (quoted == 0) 488 recordregion(startloc, dest - stackblock(), 0); 489 TRACE(("expbackq: size=%td: \"%.*s\"\n", 490 ((dest - stackblock()) - startloc), 491 (int)((dest - stackblock()) - startloc), 492 stackblock() + startloc)); 493 expdest = dest; 494 INTON; 495} 496 497 498 499static int 500subevalvar(char *p, char *str, int strloc, int subtype, int startloc, 501 int varflags, int quotes) 502{ 503 char *startp; 504 char *loc = NULL; 505 char *q; 506 int c = 0; 507 struct nodelist *saveargbackq = argbackq; 508 int amount; 509 510 argstr(p, (subtype == VSTRIMLEFT || subtype == VSTRIMLEFTMAX || 511 subtype == VSTRIMRIGHT || subtype == VSTRIMRIGHTMAX ? 512 EXP_CASE : 0) | EXP_TILDE); 513 STACKSTRNUL(expdest); 514 argbackq = saveargbackq; 515 startp = stackblock() + startloc; 516 if (str == NULL) 517 str = stackblock() + strloc; 518 519 switch (subtype) { 520 case VSASSIGN: 521 setvar(str, startp, 0); 522 amount = startp - expdest; 523 STADJUST(amount, expdest); 524 varflags &= ~VSNUL; 525 return 1; 526 527 case VSQUESTION: 528 if (*p != CTLENDVAR) { 529 outfmt(out2, "%s\n", startp); 530 error((char *)NULL); 531 } 532 error("%.*s: parameter %snot set", (int)(p - str - 1), 533 str, (varflags & VSNUL) ? "null or " 534 : nullstr); 535 return 0; 536 537 case VSTRIMLEFT: 538 for (loc = startp; loc < str; loc++) { 539 c = *loc; 540 *loc = '\0'; 541 if (patmatch(str, startp, quotes)) { 542 *loc = c; 543 goto recordleft; 544 } 545 *loc = c; 546 if (quotes && *loc == CTLESC) 547 loc++; 548 } 549 return 0; 550 551 case VSTRIMLEFTMAX: 552 for (loc = str - 1; loc >= startp;) { 553 c = *loc; 554 *loc = '\0'; 555 if (patmatch(str, startp, quotes)) { 556 *loc = c; 557 goto recordleft; 558 } 559 *loc = c; 560 loc--; 561 if (quotes && loc > startp && *(loc - 1) == CTLESC) { 562 for (q = startp; q < loc; q++) 563 if (*q == CTLESC) 564 q++; 565 if (q > loc) 566 loc--; 567 } 568 } 569 return 0; 570 571 case VSTRIMRIGHT: 572 for (loc = str - 1; loc >= startp;) { 573 if (patmatch(str, loc, quotes)) { 574 amount = loc - expdest; 575 STADJUST(amount, expdest); 576 return 1; 577 } 578 loc--; 579 if (quotes && loc > startp && *(loc - 1) == CTLESC) { 580 for (q = startp; q < loc; q++) 581 if (*q == CTLESC) 582 q++; 583 if (q > loc) 584 loc--; 585 } 586 } 587 return 0; 588 589 case VSTRIMRIGHTMAX: 590 for (loc = startp; loc < str - 1; loc++) { 591 if (patmatch(str, loc, quotes)) { 592 amount = loc - expdest; 593 STADJUST(amount, expdest); 594 return 1; 595 } 596 if (quotes && *loc == CTLESC) 597 loc++; 598 } 599 return 0; 600 601 602 default: 603 abort(); 604 } 605 606recordleft: 607 amount = ((str - 1) - (loc - startp)) - expdest; 608 STADJUST(amount, expdest); 609 while (loc != str - 1) 610 *startp++ = *loc++; 611 return 1; 612} 613 614 615/* 616 * Expand a variable, and return a pointer to the next character in the 617 * input string. 618 */ 619 620static char * 621evalvar(char *p, int flag) 622{ 623 int subtype; 624 int varflags; 625 char *var; 626 const char *val; 627 int patloc; 628 int c; 629 int set; 630 int special; 631 int startloc; 632 int varlen; 633 int varlenb; 634 int easy; 635 int quotes = flag & (EXP_FULL | EXP_CASE); 636 637 varflags = (unsigned char)*p++; 638 subtype = varflags & VSTYPE; 639 var = p; 640 special = 0; 641 if (! is_name(*p)) 642 special = 1; 643 p = strchr(p, '=') + 1; 644again: /* jump here after setting a variable with ${var=text} */ 645 if (varflags & VSLINENO) { 646 set = 1; 647 special = 1; 648 val = NULL; 649 } else if (special) { 650 set = varisset(var, varflags & VSNUL); 651 val = NULL; 652 } else { 653 val = bltinlookup(var, 1); 654 if (val == NULL || ((varflags & VSNUL) && val[0] == '\0')) { 655 val = NULL; 656 set = 0; 657 } else 658 set = 1; 659 } 660 varlen = 0; 661 startloc = expdest - stackblock(); 662 if (!set && uflag && *var != '@' && *var != '*') { 663 switch (subtype) { 664 case VSNORMAL: 665 case VSTRIMLEFT: 666 case VSTRIMLEFTMAX: 667 case VSTRIMRIGHT: 668 case VSTRIMRIGHTMAX: 669 case VSLENGTH: 670 error("%.*s: parameter not set", (int)(p - var - 1), 671 var); 672 } 673 } 674 if (set && subtype != VSPLUS) { 675 /* insert the value of the variable */ 676 if (special) { 677 if (varflags & VSLINENO) 678 STPUTBIN(var, p - var - 1, expdest); 679 else 680 varvalue(var, varflags & VSQUOTE, subtype, flag); 681 if (subtype == VSLENGTH) { 682 varlenb = expdest - stackblock() - startloc; 683 varlen = varlenb; 684 if (localeisutf8) { 685 val = stackblock() + startloc; 686 for (;val != expdest; val++) 687 if ((*val & 0xC0) == 0x80) 688 varlen--; 689 } 690 STADJUST(-varlenb, expdest); 691 } 692 } else { 693 char const *syntax = (varflags & VSQUOTE) ? DQSYNTAX 694 : BASESYNTAX; 695 696 if (subtype == VSLENGTH) { 697 for (;*val; val++) 698 if (!localeisutf8 || 699 (*val & 0xC0) != 0x80) 700 varlen++; 701 } 702 else { 703 if (quotes) 704 STPUTS_QUOTES(val, syntax, expdest); 705 else 706 STPUTS(val, expdest); 707 708 } 709 } 710 } 711 712 if (subtype == VSPLUS) 713 set = ! set; 714 715 easy = ((varflags & VSQUOTE) == 0 || 716 (*var == '@' && shellparam.nparam != 1)); 717 718 719 switch (subtype) { 720 case VSLENGTH: 721 expdest = cvtnum(varlen, expdest); 722 goto record; 723 724 case VSNORMAL: 725 if (!easy) 726 break; 727record: 728 recordregion(startloc, expdest - stackblock(), 729 varflags & VSQUOTE || (ifsset() && ifsval()[0] == '\0' && 730 (*var == '@' || *var == '*'))); 731 break; 732 733 case VSPLUS: 734 case VSMINUS: 735 if (!set) { 736 argstr(p, flag | (flag & EXP_FULL ? EXP_SPLIT_LIT : 0) | 737 (varflags & VSQUOTE ? EXP_LIT_QUOTED : 0)); 738 break; 739 } 740 if (easy) 741 goto record; 742 break; 743 744 case VSTRIMLEFT: 745 case VSTRIMLEFTMAX: 746 case VSTRIMRIGHT: 747 case VSTRIMRIGHTMAX: 748 if (!set) 749 break; 750 /* 751 * Terminate the string and start recording the pattern 752 * right after it 753 */ 754 STPUTC('\0', expdest); 755 patloc = expdest - stackblock(); 756 if (subevalvar(p, NULL, patloc, subtype, 757 startloc, varflags, quotes) == 0) { 758 int amount = (expdest - stackblock() - patloc) + 1; 759 STADJUST(-amount, expdest); 760 } 761 /* Remove any recorded regions beyond start of variable */ 762 removerecordregions(startloc); 763 goto record; 764 765 case VSASSIGN: 766 case VSQUESTION: 767 if (!set) { 768 if (subevalvar(p, var, 0, subtype, startloc, varflags, 769 quotes)) { 770 varflags &= ~VSNUL; 771 /* 772 * Remove any recorded regions beyond 773 * start of variable 774 */ 775 removerecordregions(startloc); 776 goto again; 777 } 778 break; 779 } 780 if (easy) 781 goto record; 782 break; 783 784 case VSERROR: 785 c = p - var - 1; 786 error("${%.*s%s}: Bad substitution", c, var, 787 (c > 0 && *p != CTLENDVAR) ? "..." : ""); 788 789 default: 790 abort(); 791 } 792 793 if (subtype != VSNORMAL) { /* skip to end of alternative */ 794 int nesting = 1; 795 for (;;) { 796 if ((c = *p++) == CTLESC) 797 p++; 798 else if (c == CTLBACKQ || c == (CTLBACKQ|CTLQUOTE)) { 799 if (set) 800 argbackq = argbackq->next; 801 } else if (c == CTLVAR) { 802 if ((*p++ & VSTYPE) != VSNORMAL) 803 nesting++; 804 } else if (c == CTLENDVAR) { 805 if (--nesting == 0) 806 break; 807 } 808 } 809 } 810 return p; 811} 812 813 814 815/* 816 * Test whether a specialized variable is set. 817 */ 818 819static int 820varisset(const char *name, int nulok) 821{ 822 823 if (*name == '!') 824 return backgndpidset(); 825 else if (*name == '@' || *name == '*') { 826 if (*shellparam.p == NULL) 827 return 0; 828 829 if (nulok) { 830 char **av; 831 832 for (av = shellparam.p; *av; av++) 833 if (**av != '\0') 834 return 1; 835 return 0; 836 } 837 } else if (is_digit(*name)) { 838 char *ap; 839 long num; 840 841 errno = 0; 842 num = strtol(name, NULL, 10); 843 if (errno != 0 || num > shellparam.nparam) 844 return 0; 845 846 if (num == 0) 847 ap = arg0; 848 else 849 ap = shellparam.p[num - 1]; 850 851 if (nulok && (ap == NULL || *ap == '\0')) 852 return 0; 853 } 854 return 1; 855} 856 857static void 858strtodest(const char *p, int flag, int subtype, int quoted) 859{ 860 if (flag & (EXP_FULL | EXP_CASE) && subtype != VSLENGTH) 861 STPUTS_QUOTES(p, quoted ? DQSYNTAX : BASESYNTAX, expdest); 862 else 863 STPUTS(p, expdest); 864} 865 866/* 867 * Add the value of a specialized variable to the stack string. 868 */ 869 870static void 871varvalue(const char *name, int quoted, int subtype, int flag) 872{ 873 int num; 874 char *p; 875 int i; 876 char sep; 877 char **ap; 878 879 switch (*name) { 880 case '$': 881 num = rootpid; 882 goto numvar; 883 case '?': 884 num = oexitstatus; 885 goto numvar; 886 case '#': 887 num = shellparam.nparam; 888 goto numvar; 889 case '!': 890 num = backgndpidval(); 891numvar: 892 expdest = cvtnum(num, expdest); 893 break; 894 case '-': 895 for (i = 0 ; i < NOPTS ; i++) { 896 if (optlist[i].val) 897 STPUTC(optlist[i].letter, expdest); 898 } 899 break; 900 case '@': 901 if (flag & EXP_FULL && quoted) { 902 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 903 strtodest(p, flag, subtype, quoted); 904 if (*ap) 905 STPUTC('\0', expdest); 906 } 907 break; 908 } 909 /* FALLTHROUGH */ 910 case '*': 911 if (ifsset()) 912 sep = ifsval()[0]; 913 else 914 sep = ' '; 915 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 916 strtodest(p, flag, subtype, quoted); 917 if (!*ap) 918 break; 919 if (sep || (flag & EXP_FULL && !quoted && **ap != '\0')) 920 STPUTC(sep, expdest); 921 } 922 break; 923 case '0': 924 p = arg0; 925 strtodest(p, flag, subtype, quoted); 926 break; 927 default: 928 if (is_digit(*name)) { 929 num = atoi(name); 930 if (num > 0 && num <= shellparam.nparam) { 931 p = shellparam.p[num - 1]; 932 strtodest(p, flag, subtype, quoted); 933 } 934 } 935 break; 936 } 937} 938 939 940 941/* 942 * Record the fact that we have to scan this region of the 943 * string for IFS characters. 944 */ 945 946static void 947recordregion(int start, int end, int inquotes) 948{ 949 struct ifsregion *ifsp; 950 951 INTOFF; 952 if (ifslastp == NULL) { 953 ifsp = &ifsfirst; 954 } else { 955 if (ifslastp->endoff == start 956 && ifslastp->inquotes == inquotes) { 957 /* extend previous area */ 958 ifslastp->endoff = end; 959 INTON; 960 return; 961 } 962 ifsp = (struct ifsregion *)ckmalloc(sizeof (struct ifsregion)); 963 ifslastp->next = ifsp; 964 } 965 ifslastp = ifsp; 966 ifslastp->next = NULL; 967 ifslastp->begoff = start; 968 ifslastp->endoff = end; 969 ifslastp->inquotes = inquotes; 970 INTON; 971} 972 973 974 975/* 976 * Break the argument string into pieces based upon IFS and add the 977 * strings to the argument list. The regions of the string to be 978 * searched for IFS characters have been stored by recordregion. 979 * CTLESC characters are preserved but have little effect in this pass 980 * other than escaping CTL* characters. In particular, they do not escape 981 * IFS characters: that should be done with the ifsregion mechanism. 982 * CTLQUOTEMARK characters are used to preserve empty quoted strings. 983 * This pass treats them as a regular character, making the string non-empty. 984 * Later, they are removed along with the other CTL* characters. 985 */ 986static void 987ifsbreakup(char *string, struct arglist *arglist) 988{ 989 struct ifsregion *ifsp; 990 struct strlist *sp; 991 char *start; 992 char *p; 993 char *q; 994 const char *ifs; 995 const char *ifsspc; 996 int had_param_ch = 0; 997 998 start = string; 999 1000 if (ifslastp == NULL) { 1001 /* Return entire argument, IFS doesn't apply to any of it */ 1002 sp = (struct strlist *)stalloc(sizeof *sp); 1003 sp->text = start; 1004 *arglist->lastp = sp; 1005 arglist->lastp = &sp->next; 1006 return; 1007 } 1008 1009 ifs = ifsset() ? ifsval() : " \t\n"; 1010 1011 for (ifsp = &ifsfirst; ifsp != NULL; ifsp = ifsp->next) { 1012 p = string + ifsp->begoff; 1013 while (p < string + ifsp->endoff) { 1014 q = p; 1015 if (*p == CTLESC) 1016 p++; 1017 if (ifsp->inquotes) { 1018 /* Only NULs (should be from "$@") end args */ 1019 had_param_ch = 1; 1020 if (*p != 0) { 1021 p++; 1022 continue; 1023 } 1024 ifsspc = NULL; 1025 } else { 1026 if (!strchr(ifs, *p)) { 1027 had_param_ch = 1; 1028 p++; 1029 continue; 1030 } 1031 ifsspc = strchr(" \t\n", *p); 1032 1033 /* Ignore IFS whitespace at start */ 1034 if (q == start && ifsspc != NULL) { 1035 p++; 1036 start = p; 1037 continue; 1038 } 1039 had_param_ch = 0; 1040 } 1041 1042 /* Save this argument... */ 1043 *q = '\0'; 1044 sp = (struct strlist *)stalloc(sizeof *sp); 1045 sp->text = start; 1046 *arglist->lastp = sp; 1047 arglist->lastp = &sp->next; 1048 p++; 1049 1050 if (ifsspc != NULL) { 1051 /* Ignore further trailing IFS whitespace */ 1052 for (; p < string + ifsp->endoff; p++) { 1053 q = p; 1054 if (*p == CTLESC) 1055 p++; 1056 if (strchr(ifs, *p) == NULL) { 1057 p = q; 1058 break; 1059 } 1060 if (strchr(" \t\n", *p) == NULL) { 1061 p++; 1062 break; 1063 } 1064 } 1065 } 1066 start = p; 1067 } 1068 } 1069 1070 /* 1071 * Save anything left as an argument. 1072 * Traditionally we have treated 'IFS=':'; set -- x$IFS' as 1073 * generating 2 arguments, the second of which is empty. 1074 * Some recent clarification of the Posix spec say that it 1075 * should only generate one.... 1076 */ 1077 if (had_param_ch || *start != 0) { 1078 sp = (struct strlist *)stalloc(sizeof *sp); 1079 sp->text = start; 1080 *arglist->lastp = sp; 1081 arglist->lastp = &sp->next; 1082 } 1083} 1084 1085 1086static char expdir[PATH_MAX]; 1087#define expdir_end (expdir + sizeof(expdir)) 1088 1089/* 1090 * Perform pathname generation and remove control characters. 1091 * At this point, the only control characters should be CTLESC and CTLQUOTEMARK. 1092 * The results are stored in the list exparg. 1093 */ 1094static void 1095expandmeta(struct strlist *str, int flag __unused) 1096{ 1097 char *p; 1098 struct strlist **savelastp; 1099 struct strlist *sp; 1100 char c; 1101 1102 while (str) { 1103 if (fflag) 1104 goto nometa; 1105 p = str->text; 1106 for (;;) { /* fast check for meta chars */ 1107 if ((c = *p++) == '\0') 1108 goto nometa; 1109 if (c == '*' || c == '?' || c == '[') 1110 break; 1111 } 1112 savelastp = exparg.lastp; 1113 INTOFF; 1114 expmeta(expdir, str->text); 1115 INTON; 1116 if (exparg.lastp == savelastp) { 1117 /* 1118 * no matches 1119 */ 1120nometa: 1121 *exparg.lastp = str; 1122 rmescapes(str->text); 1123 exparg.lastp = &str->next; 1124 } else { 1125 *exparg.lastp = NULL; 1126 *savelastp = sp = expsort(*savelastp); 1127 while (sp->next != NULL) 1128 sp = sp->next; 1129 exparg.lastp = &sp->next; 1130 } 1131 str = str->next; 1132 } 1133} 1134 1135 1136/* 1137 * Do metacharacter (i.e. *, ?, [...]) expansion. 1138 */ 1139 1140static void 1141expmeta(char *enddir, char *name) 1142{ 1143 const char *p; 1144 const char *q; 1145 const char *start; 1146 char *endname; 1147 int metaflag; 1148 struct stat statb; 1149 DIR *dirp; 1150 struct dirent *dp; 1151 int atend; 1152 int matchdot; 1153 int esc; 1154 int namlen; 1155 1156 metaflag = 0; 1157 start = name; 1158 for (p = name; esc = 0, *p; p += esc + 1) { 1159 if (*p == '*' || *p == '?') 1160 metaflag = 1; 1161 else if (*p == '[') { 1162 q = p + 1; 1163 if (*q == '!' || *q == '^') 1164 q++; 1165 for (;;) { 1166 while (*q == CTLQUOTEMARK) 1167 q++; 1168 if (*q == CTLESC) 1169 q++; 1170 if (*q == '/' || *q == '\0') 1171 break; 1172 if (*++q == ']') { 1173 metaflag = 1; 1174 break; 1175 } 1176 } 1177 } else if (*p == '\0') 1178 break; 1179 else if (*p == CTLQUOTEMARK) 1180 continue; 1181 else { 1182 if (*p == CTLESC) 1183 esc++; 1184 if (p[esc] == '/') { 1185 if (metaflag) 1186 break; 1187 start = p + esc + 1; 1188 } 1189 } 1190 } 1191 if (metaflag == 0) { /* we've reached the end of the file name */ 1192 if (enddir != expdir) 1193 metaflag++; 1194 for (p = name ; ; p++) { 1195 if (*p == CTLQUOTEMARK) 1196 continue; 1197 if (*p == CTLESC) 1198 p++; 1199 *enddir++ = *p; 1200 if (*p == '\0') 1201 break; 1202 if (enddir == expdir_end) 1203 return; 1204 } 1205 if (metaflag == 0 || lstat(expdir, &statb) >= 0) 1206 addfname(expdir); 1207 return; 1208 } 1209 endname = name + (p - name); 1210 if (start != name) { 1211 p = name; 1212 while (p < start) { 1213 while (*p == CTLQUOTEMARK) 1214 p++; 1215 if (*p == CTLESC) 1216 p++; 1217 *enddir++ = *p++; 1218 if (enddir == expdir_end) 1219 return; 1220 } 1221 } 1222 if (enddir == expdir) { 1223 p = "."; 1224 } else if (enddir == expdir + 1 && *expdir == '/') { 1225 p = "/"; 1226 } else { 1227 p = expdir; 1228 enddir[-1] = '\0'; 1229 } 1230 if ((dirp = opendir(p)) == NULL) 1231 return; 1232 if (enddir != expdir) 1233 enddir[-1] = '/'; 1234 if (*endname == 0) { 1235 atend = 1; 1236 } else { 1237 atend = 0; 1238 *endname = '\0'; 1239 endname += esc + 1; 1240 } 1241 matchdot = 0; 1242 p = start; 1243 while (*p == CTLQUOTEMARK) 1244 p++; 1245 if (*p == CTLESC) 1246 p++; 1247 if (*p == '.') 1248 matchdot++; 1249 while (! int_pending() && (dp = readdir(dirp)) != NULL) { 1250 if (dp->d_name[0] == '.' && ! matchdot) 1251 continue; 1252 if (patmatch(start, dp->d_name, 0)) { 1253 namlen = dp->d_namlen; 1254 if (enddir + namlen + 1 > expdir_end) 1255 continue; 1256 memcpy(enddir, dp->d_name, namlen + 1); 1257 if (atend) 1258 addfname(expdir); 1259 else { 1260 if (dp->d_type != DT_UNKNOWN && 1261 dp->d_type != DT_DIR && 1262 dp->d_type != DT_LNK) 1263 continue; 1264 if (enddir + namlen + 2 > expdir_end) 1265 continue; 1266 enddir[namlen] = '/'; 1267 enddir[namlen + 1] = '\0'; 1268 expmeta(enddir + namlen + 1, endname); 1269 } 1270 } 1271 } 1272 closedir(dirp); 1273 if (! atend) 1274 endname[-esc - 1] = esc ? CTLESC : '/'; 1275} 1276 1277 1278/* 1279 * Add a file name to the list. 1280 */ 1281 1282static void 1283addfname(char *name) 1284{ 1285 char *p; 1286 struct strlist *sp; 1287 size_t len; 1288 1289 len = strlen(name); 1290 p = stalloc(len + 1); 1291 memcpy(p, name, len + 1); 1292 sp = (struct strlist *)stalloc(sizeof *sp); 1293 sp->text = p; 1294 *exparg.lastp = sp; 1295 exparg.lastp = &sp->next; 1296} 1297 1298 1299/* 1300 * Sort the results of file name expansion. It calculates the number of 1301 * strings to sort and then calls msort (short for merge sort) to do the 1302 * work. 1303 */ 1304 1305static struct strlist * 1306expsort(struct strlist *str) 1307{ 1308 int len; 1309 struct strlist *sp; 1310 1311 len = 0; 1312 for (sp = str ; sp ; sp = sp->next) 1313 len++; 1314 return msort(str, len); 1315} 1316 1317 1318static struct strlist * 1319msort(struct strlist *list, int len) 1320{ 1321 struct strlist *p, *q = NULL; 1322 struct strlist **lpp; 1323 int half; 1324 int n; 1325 1326 if (len <= 1) 1327 return list; 1328 half = len >> 1; 1329 p = list; 1330 for (n = half ; --n >= 0 ; ) { 1331 q = p; 1332 p = p->next; 1333 } 1334 q->next = NULL; /* terminate first half of list */ 1335 q = msort(list, half); /* sort first half of list */ 1336 p = msort(p, len - half); /* sort second half */ 1337 lpp = &list; 1338 for (;;) { 1339 if (strcmp(p->text, q->text) < 0) { 1340 *lpp = p; 1341 lpp = &p->next; 1342 if ((p = *lpp) == NULL) { 1343 *lpp = q; 1344 break; 1345 } 1346 } else { 1347 *lpp = q; 1348 lpp = &q->next; 1349 if ((q = *lpp) == NULL) { 1350 *lpp = p; 1351 break; 1352 } 1353 } 1354 } 1355 return list; 1356} 1357 1358 1359 1360static wchar_t 1361get_wc(const char **p) 1362{ 1363 wchar_t c; 1364 int chrlen; 1365 1366 chrlen = mbtowc(&c, *p, 4); 1367 if (chrlen == 0) 1368 return 0; 1369 else if (chrlen == -1) 1370 c = 0; 1371 else 1372 *p += chrlen; 1373 return c; 1374} 1375 1376 1377/* 1378 * See if a character matches a character class, starting at the first colon 1379 * of "[:class:]". 1380 * If a valid character class is recognized, a pointer to the next character 1381 * after the final closing bracket is stored into *end, otherwise a null 1382 * pointer is stored into *end. 1383 */ 1384static int 1385match_charclass(const char *p, wchar_t chr, const char **end) 1386{ 1387 char name[20]; 1388 const char *nameend; 1389 wctype_t cclass; 1390 1391 *end = NULL; 1392 p++; 1393 nameend = strstr(p, ":]"); 1394 if (nameend == NULL || (size_t)(nameend - p) >= sizeof(name) || 1395 nameend == p) 1396 return 0; 1397 memcpy(name, p, nameend - p); 1398 name[nameend - p] = '\0'; 1399 *end = nameend + 2; 1400 cclass = wctype(name); 1401 /* An unknown class matches nothing but is valid nevertheless. */ 1402 if (cclass == 0) 1403 return 0; 1404 return iswctype(chr, cclass); 1405} 1406 1407 1408/* 1409 * Returns true if the pattern matches the string. 1410 */ 1411 1412static int 1413patmatch(const char *pattern, const char *string, int squoted) 1414{ 1415 const char *p, *q, *end; 1416 const char *bt_p, *bt_q; 1417 char c; 1418 wchar_t wc, wc2; 1419 1420 p = pattern; 1421 q = string; 1422 bt_p = NULL; 1423 bt_q = NULL; 1424 for (;;) { 1425 switch (c = *p++) { 1426 case '\0': 1427 if (*q != '\0') 1428 goto backtrack; 1429 return 1; 1430 case CTLESC: 1431 if (squoted && *q == CTLESC) 1432 q++; 1433 if (*q++ != *p++) 1434 goto backtrack; 1435 break; 1436 case CTLQUOTEMARK: 1437 continue; 1438 case '?': 1439 if (squoted && *q == CTLESC) 1440 q++; 1441 if (*q == '\0') 1442 return 0; 1443 if (localeisutf8) { 1444 wc = get_wc(&q); 1445 /* 1446 * A '?' does not match invalid UTF-8 but a 1447 * '*' does, so backtrack. 1448 */ 1449 if (wc == 0) 1450 goto backtrack; 1451 } else 1452 wc = (unsigned char)*q++; 1453 break; 1454 case '*': 1455 c = *p; 1456 while (c == CTLQUOTEMARK || c == '*') 1457 c = *++p; 1458 /* 1459 * If the pattern ends here, we know the string 1460 * matches without needing to look at the rest of it. 1461 */ 1462 if (c == '\0') 1463 return 1; 1464 /* 1465 * First try the shortest match for the '*' that 1466 * could work. We can forget any earlier '*' since 1467 * there is no way having it match more characters 1468 * can help us, given that we are already here. 1469 */ 1470 bt_p = p; 1471 bt_q = q; 1472 break; 1473 case '[': { 1474 const char *endp; 1475 int invert, found; 1476 wchar_t chr; 1477 1478 endp = p; 1479 if (*endp == '!' || *endp == '^') 1480 endp++; 1481 for (;;) { 1482 while (*endp == CTLQUOTEMARK) 1483 endp++; 1484 if (*endp == 0) 1485 goto dft; /* no matching ] */ 1486 if (*endp == CTLESC) 1487 endp++; 1488 if (*++endp == ']') 1489 break; 1490 } 1491 invert = 0; 1492 if (*p == '!' || *p == '^') { 1493 invert++; 1494 p++; 1495 } 1496 found = 0; 1497 if (squoted && *q == CTLESC) 1498 q++; 1499 if (*q == '\0') 1500 return 0; 1501 if (localeisutf8) { 1502 chr = get_wc(&q); 1503 if (chr == 0) 1504 goto backtrack; 1505 } else 1506 chr = (unsigned char)*q++; 1507 c = *p++; 1508 do { 1509 if (c == CTLQUOTEMARK) 1510 continue; 1511 if (c == '[' && *p == ':') { 1512 found |= match_charclass(p, chr, &end); 1513 if (end != NULL) 1514 p = end; 1515 } 1516 if (c == CTLESC) 1517 c = *p++; 1518 if (localeisutf8 && c & 0x80) { 1519 p--; 1520 wc = get_wc(&p); 1521 if (wc == 0) /* bad utf-8 */ 1522 return 0; 1523 } else 1524 wc = (unsigned char)c; 1525 if (*p == '-' && p[1] != ']') { 1526 p++; 1527 while (*p == CTLQUOTEMARK) 1528 p++; 1529 if (*p == CTLESC) 1530 p++; 1531 if (localeisutf8) { 1532 wc2 = get_wc(&p); 1533 if (wc2 == 0) /* bad utf-8 */ 1534 return 0; 1535 } else 1536 wc2 = (unsigned char)*p++; 1537 if ( collate_range_cmp(chr, wc) >= 0 1538 && collate_range_cmp(chr, wc2) <= 0 1539 ) 1540 found = 1; 1541 } else { 1542 if (chr == wc) 1543 found = 1; 1544 } 1545 } while ((c = *p++) != ']'); 1546 if (found == invert) 1547 goto backtrack; 1548 break; 1549 } 1550dft: default: 1551 if (squoted && *q == CTLESC) 1552 q++; 1553 if (*q == '\0') 1554 return 0; 1555 if (*q++ == c) 1556 break; 1557backtrack: 1558 /* 1559 * If we have a mismatch (other than hitting the end 1560 * of the string), go back to the last '*' seen and 1561 * have it match one additional character. 1562 */ 1563 if (bt_p == NULL) 1564 return 0; 1565 if (squoted && *bt_q == CTLESC) 1566 bt_q++; 1567 if (*bt_q == '\0') 1568 return 0; 1569 bt_q++; 1570 p = bt_p; 1571 q = bt_q; 1572 break; 1573 } 1574 } 1575} 1576 1577 1578 1579/* 1580 * Remove any CTLESC and CTLQUOTEMARK characters from a string. 1581 */ 1582 1583void 1584rmescapes(char *str) 1585{ 1586 char *p, *q; 1587 1588 p = str; 1589 while (*p != CTLESC && *p != CTLQUOTEMARK && *p != CTLQUOTEEND) { 1590 if (*p++ == '\0') 1591 return; 1592 } 1593 q = p; 1594 while (*p) { 1595 if (*p == CTLQUOTEMARK || *p == CTLQUOTEEND) { 1596 p++; 1597 continue; 1598 } 1599 if (*p == CTLESC) 1600 p++; 1601 *q++ = *p++; 1602 } 1603 *q = '\0'; 1604} 1605 1606 1607 1608/* 1609 * See if a pattern matches in a case statement. 1610 */ 1611 1612int 1613casematch(union node *pattern, const char *val) 1614{ 1615 struct stackmark smark; 1616 int result; 1617 char *p; 1618 1619 setstackmark(&smark); 1620 argbackq = pattern->narg.backquote; 1621 STARTSTACKSTR(expdest); 1622 ifslastp = NULL; 1623 argstr(pattern->narg.text, EXP_TILDE | EXP_CASE); 1624 STPUTC('\0', expdest); 1625 p = grabstackstr(expdest); 1626 result = patmatch(p, val, 0); 1627 popstackmark(&smark); 1628 return result; 1629} 1630 1631/* 1632 * Our own itoa(). 1633 */ 1634 1635static char * 1636cvtnum(int num, char *buf) 1637{ 1638 char temp[32]; 1639 int neg = num < 0; 1640 char *p = temp + 31; 1641 1642 temp[31] = '\0'; 1643 1644 do { 1645 *--p = num % 10 + '0'; 1646 } while ((num /= 10) != 0); 1647 1648 if (neg) 1649 *--p = '-'; 1650 1651 STPUTS(p, buf); 1652 return buf; 1653} 1654 1655/* 1656 * Do most of the work for wordexp(3). 1657 */ 1658 1659int 1660wordexpcmd(int argc, char **argv) 1661{ 1662 size_t len; 1663 int i; 1664 1665 out1fmt("%08x", argc - 1); 1666 for (i = 1, len = 0; i < argc; i++) 1667 len += strlen(argv[i]); 1668 out1fmt("%08x", (int)len); 1669 for (i = 1; i < argc; i++) 1670 outbin(argv[i], strlen(argv[i]) + 1, out1); 1671 return (0); 1672} 1673