bin/sh/expand.c

   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  * 3. All advertising materials mentioning features or use of this software
  19  *    must display the following acknowledgement:
  20  *      This product includes software developed by the University of
  21  *      California, Berkeley and its contributors.
  22  * 4. Neither the name of the University nor the names of its contributors
  23  *    may be used to endorse or promote products derived from this software
  24  *    without specific prior written permission.
  25  *
  26  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
  27  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
  28  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
  29  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
  30  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
  31  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
  32  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
  33  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
  34  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
  35  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
  36  * SUCH DAMAGE.
  37  *
  38  * @(#)expand.c 8.5 (Berkeley) 5/15/95
  39  * $FreeBSD: head/bin/sh/expand.c 246288 2013-02-03 15:54:57Z jilles $
  40  */
  41
  42 #include <sys/types.h>
  43 #include <sys/time.h>
  44 #include <sys/stat.h>
  45 #include <dirent.h>
  46 #include <errno.h>
  47 #include <inttypes.h>
  48 #include <limits.h>
  49 #include <pwd.h>
  50 #include <stdio.h>
  51 #include <stdlib.h>
  52 #include <string.h>
  53 #include <unistd.h>
  54 #include <wchar.h>
  55 #include <wctype.h>
  56
  57 /*
  58  * Routines to expand arguments to commands.  We have to deal with
  59  * backquotes, shell variables, and file metacharacters.
  60  */
  61
  62 #include "shell.h"
  63 #include "main.h"
  64 #include "nodes.h"
  65 #include "eval.h"
  66 #include "expand.h"
  67 #include "syntax.h"
  68 #include "parser.h"
  69 #include "jobs.h"
  70 #include "options.h"
  71 #include "var.h"
  72 #include "input.h"
  73 #include "output.h"
  74 #include "memalloc.h"
  75 #include "error.h"
  76 #include "mystring.h"
  77 #include "arith.h"
  78 #include "show.h"
  79 #include "builtins.h"
  80
  81 /*
  82  * Structure specifying which parts of the string should be searched
  83  * for IFS characters.
  84  */
  85
  86 struct ifsregion {
  87         struct ifsregion *next; /* next region in list */
  88         int begoff;             /* offset of start of region */
  89         int endoff;             /* offset of end of region */
  90         int inquotes;           /* search for nul bytes only */
  91 };
  92
  93
  94 static char *expdest;                   /* output of current string */
  95 static struct nodelist *argbackq;       /* list of back quote expressions */
  96 static struct ifsregion ifsfirst;       /* first struct in list of ifs regions */
  97 static struct ifsregion *ifslastp;      /* last struct in list */
  98 static struct arglist exparg;           /* holds expanded arg list */
  99
 100 static void argstr(char *, int);
 101 static char *exptilde(char *, int);
 102 static void expbackq(union node *, int, int);
 103 static int subevalvar(char *, char *, int, int, int, int, int);
 104 static char *evalvar(char *, int);
 105 static int varisset(char *, int);
 106 static void varvalue(char *, int, int, int);
 107 static void recordregion(int, int, int);
 108 static void removerecordregions(int);
 109 static void ifsbreakup(char *, struct arglist *);
 110 static void expandmeta(struct strlist *, int);
 111 static void expmeta(char *, char *);
 112 static void addfname(char *);
 113 static struct strlist *expsort(struct strlist *);
 114 static struct strlist *msort(struct strlist *, int);
 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 static int
 120 collate_range_cmp(wchar_t c1, wchar_t c2)
 121 {
 122         static wchar_t s1[2], s2[2];
 123
 124         s1[0] = c1;
 125         s2[0] = c2;
 126         return (wcscoll(s1, s2));
 127 }
 128
 129 static char *
 130 stputs_quotes(const char *data, const char *syntax, char *p)
 131 {
 132         while (*data) {
 133                 CHECKSTRSPACE(2, p);
 134                 if (syntax[(int)*data] == CCTL)
 135                         USTPUTC(CTLESC, p);
 136                 USTPUTC(*data++, p);
 137         }
 138         return (p);
 139 }
 140 #define STPUTS_QUOTES(data, syntax, p) p = stputs_quotes((data), syntax, p)
 141
 142 /*
 143  * Perform expansions on an argument, placing the resulting list of arguments
 144  * in arglist.  Parameter expansion, command substitution and arithmetic
 145  * expansion are always performed; additional expansions can be requested
 146  * via flag (EXP_*).
 147  * The result is left in the stack string.
 148  * When arglist is NULL, perform here document expansion.
 149  *
 150  * Caution: this function uses global state and is not reentrant.
 151  * However, a new invocation after an interrupted invocation is safe
 152  * and will reset the global state for the new call.
 153  */
 154 void
 155 expandarg(union node *arg, struct arglist *arglist, int flag)
 156 {
 157         struct strlist *sp;
 158         char *p;
 159
 160         argbackq = arg->narg.backquote;
 161         STARTSTACKSTR(expdest);
 162         ifsfirst.next = NULL;
 163         ifslastp = NULL;
 164         argstr(arg->narg.text, flag);
 165         if (arglist == NULL) {
 166                 STACKSTRNUL(expdest);
 167                 return;                 /* here document expanded */
 168         }
 169         STPUTC('\0', expdest);
 170         p = grabstackstr(expdest);
 171         exparg.lastp = &exparg.list;
 172         /*
 173          * TODO - EXP_REDIR
 174          */
 175         if (flag & EXP_FULL) {
 176                 ifsbreakup(p, &exparg);
 177                 *exparg.lastp = NULL;
 178                 exparg.lastp = &exparg.list;
 179                 expandmeta(exparg.list, flag);
 180         } else {
 181                 if (flag & EXP_REDIR) /*XXX - for now, just remove escapes */
 182                         rmescapes(p);
 183                 sp = (struct strlist *)stalloc(sizeof (struct strlist));
 184                 sp->text = p;
 185                 *exparg.lastp = sp;
 186                 exparg.lastp = &sp->next;
 187         }
 188         while (ifsfirst.next != NULL) {
 189                 struct ifsregion *ifsp;
 190                 INTOFF;
 191                 ifsp = ifsfirst.next->next;
 192                 ckfree(ifsfirst.next);
 193                 ifsfirst.next = ifsp;
 194                 INTON;
 195         }
 196         *exparg.lastp = NULL;
 197         if (exparg.list) {
 198                 *arglist->lastp = exparg.list;
 199                 arglist->lastp = exparg.lastp;
 200         }
 201 }
 202
 203
 204
 205 /*
 206  * Perform parameter expansion, command substitution and arithmetic
 207  * expansion, and tilde expansion if requested via EXP_TILDE/EXP_VARTILDE.
 208  * Processing ends at a CTLENDVAR character as well as '\0'.
 209  * This is used to expand word in ${var+word} etc.
 210  * If EXP_FULL, EXP_CASE or EXP_REDIR are set, keep and/or generate CTLESC
 211  * characters to allow for further processing.
 212  * If EXP_FULL is set, also preserve CTLQUOTEMARK characters.
 213  */
 214 static void
 215 argstr(char *p, int flag)
 216 {
 217         char c;
 218         int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR);  /* do CTLESC */
 219         int firsteq = 1;
 220         int split_lit;
 221         int lit_quoted;
 222
 223         split_lit = flag & EXP_SPLIT_LIT;
 224         lit_quoted = flag & EXP_LIT_QUOTED;
 225         flag &= ~(EXP_SPLIT_LIT | EXP_LIT_QUOTED);
 226         if (*p == '~' && (flag & (EXP_TILDE | EXP_VARTILDE)))
 227                 p = exptilde(p, flag);
 228         for (;;) {
 229                 CHECKSTRSPACE(2, expdest);
 230                 switch (c = *p++) {
 231                 case '\0':
 232                 case CTLENDVAR:
 233                         goto breakloop;
 234                 case CTLQUOTEMARK:
 235                         lit_quoted = 1;
 236                         /* "$@" syntax adherence hack */
 237                         if (p[0] == CTLVAR && p[2] == '@' && p[3] == '=')
 238                                 break;
 239                         if ((flag & EXP_FULL) != 0)
 240                                 USTPUTC(c, expdest);
 241                         break;
 242                 case CTLQUOTEEND:
 243                         lit_quoted = 0;
 244                         break;
 245                 case CTLESC:
 246                         if (quotes)
 247                                 USTPUTC(c, expdest);
 248                         c = *p++;
 249                         USTPUTC(c, expdest);
 250                         if (split_lit && !lit_quoted)
 251                                 recordregion(expdest - stackblock() -
 252                                     (quotes ? 2 : 1),
 253                                     expdest - stackblock(), 0);
 254                         break;
 255                 case CTLVAR:
 256                         p = evalvar(p, flag);
 257                         break;
 258                 case CTLBACKQ:
 259                 case CTLBACKQ|CTLQUOTE:
 260                         expbackq(argbackq->n, c & CTLQUOTE, flag);
 261                         argbackq = argbackq->next;
 262                         break;
 263                 case CTLENDARI:
 264                         expari(flag);
 265                         break;
 266                 case ':':
 267                 case '=':
 268                         /*
 269                          * sort of a hack - expand tildes in variable
 270                          * assignments (after the first '=' and after ':'s).
 271                          */
 272                         USTPUTC(c, expdest);
 273                         if (split_lit && !lit_quoted)
 274                                 recordregion(expdest - stackblock() - 1,
 275                                     expdest - stackblock(), 0);
 276                         if (flag & EXP_VARTILDE && *p == '~' &&
 277                             (c != '=' || firsteq)) {
 278                                 if (c == '=')
 279                                         firsteq = 0;
 280                                 p = exptilde(p, flag);
 281                         }
 282                         break;
 283                 default:
 284                         USTPUTC(c, expdest);
 285                         if (split_lit && !lit_quoted)
 286                                 recordregion(expdest - stackblock() - 1,
 287                                     expdest - stackblock(), 0);
 288                 }
 289         }
 290 breakloop:;
 291 }
 292
 293 /*
 294  * Perform tilde expansion, placing the result in the stack string and
 295  * returning the next position in the input string to process.
 296  */
 297 static char *
 298 exptilde(char *p, int flag)
 299 {
 300         char c, *startp = p;
 301         struct passwd *pw;
 302         char *home;
 303         int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR);
 304
 305         while ((c = *p) != '\0') {
 306                 switch(c) {
 307                 case CTLESC: /* This means CTL* are always considered quoted. */
 308                 case CTLVAR:
 309                 case CTLBACKQ:
 310                 case CTLBACKQ | CTLQUOTE:
 311                 case CTLARI:
 312                 case CTLENDARI:
 313                 case CTLQUOTEMARK:
 314                         return (startp);
 315                 case ':':
 316                         if (flag & EXP_VARTILDE)
 317                                 goto done;
 318                         break;
 319                 case '/':
 320                 case CTLENDVAR:
 321                         goto done;
 322                 }
 323                 p++;
 324         }
 325 done:
 326         *p = '\0';
 327         if (*(startp+1) == '\0') {
 328                 if ((home = lookupvar("HOME")) == NULL)
 329                         goto lose;
 330         } else {
 331                 if ((pw = getpwnam(startp+1)) == NULL)
 332                         goto lose;
 333                 home = pw->pw_dir;
 334         }
 335         if (*home == '\0')
 336                 goto lose;
 337         *p = c;
 338         if (quotes)
 339                 STPUTS_QUOTES(home, SQSYNTAX, expdest);
 340         else
 341                 STPUTS(home, expdest);
 342         return (p);
 343 lose:
 344         *p = c;
 345         return (startp);
 346 }
 347
 348
 349 static void
 350 removerecordregions(int endoff)
 351 {
 352         if (ifslastp == NULL)
 353                 return;
 354
 355         if (ifsfirst.endoff > endoff) {
 356                 while (ifsfirst.next != NULL) {
 357                         struct ifsregion *ifsp;
 358                         INTOFF;
 359                         ifsp = ifsfirst.next->next;
 360                         ckfree(ifsfirst.next);
 361                         ifsfirst.next = ifsp;
 362                         INTON;
 363                 }
 364                 if (ifsfirst.begoff > endoff)
 365                         ifslastp = NULL;
 366                 else {
 367                         ifslastp = &ifsfirst;
 368                         ifsfirst.endoff = endoff;
 369                 }
 370                 return;
 371         }
 372
 373         ifslastp = &ifsfirst;
 374         while (ifslastp->next && ifslastp->next->begoff < endoff)
 375                 ifslastp=ifslastp->next;
 376         while (ifslastp->next != NULL) {
 377                 struct ifsregion *ifsp;
 378                 INTOFF;
 379                 ifsp = ifslastp->next->next;
 380                 ckfree(ifslastp->next);
 381                 ifslastp->next = ifsp;
 382                 INTON;
 383         }
 384         if (ifslastp->endoff > endoff)
 385                 ifslastp->endoff = endoff;
 386 }
 387
 388 /*
 389  * Expand arithmetic expression.  Backup to start of expression,
 390  * evaluate, place result in (backed up) result, adjust string position.
 391  */
 392 void
 393 expari(int flag)
 394 {
 395         char *p, *q, *start;
 396         arith_t result;
 397         int begoff;
 398         int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR);
 399         int quoted;
 400
 401         /*
 402          * This routine is slightly over-complicated for
 403          * efficiency.  First we make sure there is
 404          * enough space for the result, which may be bigger
 405          * than the expression.  Next we
 406          * scan backwards looking for the start of arithmetic.  If the
 407          * next previous character is a CTLESC character, then we
 408          * have to rescan starting from the beginning since CTLESC
 409          * characters have to be processed left to right.
 410          */
 411         CHECKSTRSPACE(DIGITS(result) - 2, expdest);
 412         USTPUTC('\0', expdest);
 413         start = stackblock();
 414         p = expdest - 2;
 415         while (p >= start && *p != CTLARI)
 416                 --p;
 417         if (p < start || *p != CTLARI)
 418                 error("missing CTLARI (shouldn't happen)");
 419         if (p > start && *(p - 1) == CTLESC)
 420                 for (p = start; *p != CTLARI; p++)
 421                         if (*p == CTLESC)
 422                                 p++;
 423
 424         if (p[1] == '"')
 425                 quoted=1;
 426         else
 427                 quoted=0;
 428         begoff = p - start;
 429         removerecordregions(begoff);
 430         if (quotes)
 431                 rmescapes(p+2);
 432         q = grabstackstr(expdest);
 433         result = arith(p+2);
 434         ungrabstackstr(q, expdest);
 435         fmtstr(p, DIGITS(result), ARITH_FORMAT_STR, result);
 436         while (*p++)
 437                 ;
 438         if (quoted == 0)
 439                 recordregion(begoff, p - 1 - start, 0);
 440         result = expdest - p + 1;
 441         STADJUST(-result, expdest);
 442 }
 443
 444
 445 /*
 446  * Perform command substitution.
 447  */
 448 static void
 449 expbackq(union node *cmd, int quoted, int flag)
 450 {
 451         struct backcmd in;
 452         int i;
 453         char buf[128];
 454         char *p;
 455         char *dest = expdest;
 456         struct ifsregion saveifs, *savelastp;
 457         struct nodelist *saveargbackq;
 458         char lastc;
 459         int startloc = dest - stackblock();
 460         char const *syntax = quoted? DQSYNTAX : BASESYNTAX;
 461         int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR);
 462         int nnl;
 463
 464         INTOFF;
 465         saveifs = ifsfirst;
 466         savelastp = ifslastp;
 467         saveargbackq = argbackq;
 468         p = grabstackstr(dest);
 469         evalbackcmd(cmd, &in);
 470         ungrabstackstr(p, dest);
 471         ifsfirst = saveifs;
 472         ifslastp = savelastp;
 473         argbackq = saveargbackq;
 474
 475         p = in.buf;
 476         lastc = '\0';
 477         nnl = 0;
 478         /* Don't copy trailing newlines */
 479         for (;;) {
 480                 if (--in.nleft < 0) {
 481                         if (in.fd < 0)
 482                                 break;
 483                         while ((i = read(in.fd, buf, sizeof buf)) < 0 && errno == EINTR);
 484                         TRACE(("expbackq: read returns %d\n", i));
 485                         if (i <= 0)
 486                                 break;
 487                         p = buf;
 488                         in.nleft = i - 1;
 489                 }
 490                 lastc = *p++;
 491                 if (lastc != '\0') {
 492                         if (lastc == '\n') {
 493                                 nnl++;
 494                         } else {
 495                                 CHECKSTRSPACE(nnl + 2, dest);
 496                                 while (nnl > 0) {
 497                                         nnl--;
 498                                         USTPUTC('\n', dest);
 499                                 }
 500                                 if (quotes && syntax[(int)lastc] == CCTL)
 501                                         USTPUTC(CTLESC, dest);
 502                                 USTPUTC(lastc, dest);
 503                         }
 504                 }
 505         }
 506
 507         if (in.fd >= 0)
 508                 close(in.fd);
 509         if (in.buf)
 510                 ckfree(in.buf);
 511         if (in.jp)
 512                 exitstatus = waitforjob(in.jp, NULL);
 513         if (quoted == 0)
 514                 recordregion(startloc, dest - stackblock(), 0);
 515         TRACE(("expbackq: size=%td: \"%.*s\"\n",
 516                 ((dest - stackblock()) - startloc),
 517                 (int)((dest - stackblock()) - startloc),
 518                 stackblock() + startloc));
 519         expdest = dest;
 520         INTON;
 521 }
 522
 523
 524
 525 static int
 526 subevalvar(char *p, char *str, int strloc, int subtype, int startloc,
 527   int varflags, int quotes)
 528 {
 529         char *startp;
 530         char *loc = NULL;
 531         char *q;
 532         int c = 0;
 533         struct nodelist *saveargbackq = argbackq;
 534         int amount;
 535
 536         argstr(p, (subtype == VSTRIMLEFT || subtype == VSTRIMLEFTMAX ||
 537             subtype == VSTRIMRIGHT || subtype == VSTRIMRIGHTMAX ?
 538             EXP_CASE : 0) | EXP_TILDE);
 539         STACKSTRNUL(expdest);
 540         argbackq = saveargbackq;
 541         startp = stackblock() + startloc;
 542         if (str == NULL)
 543             str = stackblock() + strloc;
 544
 545         switch (subtype) {
 546         case VSASSIGN:
 547                 setvar(str, startp, 0);
 548                 amount = startp - expdest;
 549                 STADJUST(amount, expdest);
 550                 varflags &= ~VSNUL;
 551                 return 1;
 552
 553         case VSQUESTION:
 554                 if (*p != CTLENDVAR) {
 555                         outfmt(out2, "%s\n", startp);
 556                         error(NULL);
 557                 }
 558                 error("%.*s: parameter %snot set", (int)(p - str - 1),
 559                       str, (varflags & VSNUL) ? "null or "
 560                                               : nullstr);
 561                 return 0;
 562
 563         case VSTRIMLEFT:
 564                 for (loc = startp; loc < str; loc++) {
 565                         c = *loc;
 566                         *loc = '\0';
 567                         if (patmatch(str, startp, quotes)) {
 568                                 *loc = c;
 569                                 goto recordleft;
 570                         }
 571                         *loc = c;
 572                         if (quotes && *loc == CTLESC)
 573                                 loc++;
 574                 }
 575                 return 0;
 576
 577         case VSTRIMLEFTMAX:
 578                 for (loc = str - 1; loc >= startp;) {
 579                         c = *loc;
 580                         *loc = '\0';
 581                         if (patmatch(str, startp, quotes)) {
 582                                 *loc = c;
 583                                 goto recordleft;
 584                         }
 585                         *loc = c;
 586                         loc--;
 587                         if (quotes && loc > startp && *(loc - 1) == CTLESC) {
 588                                 for (q = startp; q < loc; q++)
 589                                         if (*q == CTLESC)
 590                                                 q++;
 591                                 if (q > loc)
 592                                         loc--;
 593                         }
 594                 }
 595                 return 0;
 596
 597         case VSTRIMRIGHT:
 598                 for (loc = str - 1; loc >= startp;) {
 599                         if (patmatch(str, loc, quotes)) {
 600                                 amount = loc - expdest;
 601                                 STADJUST(amount, expdest);
 602                                 return 1;
 603                         }
 604                         loc--;
 605                         if (quotes && loc > startp && *(loc - 1) == CTLESC) {
 606                                 for (q = startp; q < loc; q++)
 607                                         if (*q == CTLESC)
 608                                                 q++;
 609                                 if (q > loc)
 610                                         loc--;
 611                         }
 612                 }
 613                 return 0;
 614
 615         case VSTRIMRIGHTMAX:
 616                 for (loc = startp; loc < str - 1; loc++) {
 617                         if (patmatch(str, loc, quotes)) {
 618                                 amount = loc - expdest;
 619                                 STADJUST(amount, expdest);
 620                                 return 1;
 621                         }
 622                         if (quotes && *loc == CTLESC)
 623                                 loc++;
 624                 }
 625                 return 0;
 626
 627
 628         default:
 629                 abort();
 630         }
 631
 632 recordleft:
 633         amount = ((str - 1) - (loc - startp)) - expdest;
 634         STADJUST(amount, expdest);
 635         while (loc != str - 1)
 636                 *startp++ = *loc++;
 637         return 1;
 638 }
 639
 640
 641 /*
 642  * Expand a variable, and return a pointer to the next character in the
 643  * input string.
 644  */
 645
 646 static char *
 647 evalvar(char *p, int flag)
 648 {
 649         int subtype;
 650         int varflags;
 651         char *var;
 652         char *val;
 653         int patloc;
 654         int c;
 655         int set;
 656         int special;
 657         int startloc;
 658         int varlen;
 659         int varlenb;
 660         int easy;
 661         int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR);
 662
 663         varflags = (unsigned char)*p++;
 664         subtype = varflags & VSTYPE;
 665         var = p;
 666         special = 0;
 667         if (! is_name(*p))
 668                 special = 1;
 669         p = strchr(p, '=') + 1;
 670 again: /* jump here after setting a variable with ${var=text} */
 671         if (varflags & VSLINENO) {
 672                 set = 1;
 673                 special = 0;
 674                 val = var;
 675                 p[-1] = '\0';   /* temporarily overwrite '=' to have \0
 676                                    terminated string */
 677         } else if (special) {
 678                 set = varisset(var, varflags & VSNUL);
 679                 val = NULL;
 680         } else {
 681                 val = bltinlookup(var, 1);
 682                 if (val == NULL || ((varflags & VSNUL) && val[0] == '\0')) {
 683                         val = NULL;
 684                         set = 0;
 685                 } else
 686                         set = 1;
 687         }
 688         varlen = 0;
 689         startloc = expdest - stackblock();
 690         if (!set && uflag && *var != '@' && *var != '*') {
 691                 switch (subtype) {
 692                 case VSNORMAL:
 693                 case VSTRIMLEFT:
 694                 case VSTRIMLEFTMAX:
 695                 case VSTRIMRIGHT:
 696                 case VSTRIMRIGHTMAX:
 697                 case VSLENGTH:
 698                         error("%.*s: parameter not set", (int)(p - var - 1),
 699                             var);
 700                 }
 701         }
 702         if (set && subtype != VSPLUS) {
 703                 /* insert the value of the variable */
 704                 if (special) {
 705                         varvalue(var, varflags & VSQUOTE, subtype, flag);
 706                         if (subtype == VSLENGTH) {
 707                                 varlenb = expdest - stackblock() - startloc;
 708                                 varlen = varlenb;
 709                                 if (localeisutf8) {
 710                                         val = stackblock() + startloc;
 711                                         for (;val != expdest; val++)
 712                                                 if ((*val & 0xC0) == 0x80)
 713                                                         varlen--;
 714                                 }
 715                                 STADJUST(-varlenb, expdest);
 716                         }
 717                 } else {
 718                         char const *syntax = (varflags & VSQUOTE) ? DQSYNTAX
 719                                                                   : BASESYNTAX;
 720
 721                         if (subtype == VSLENGTH) {
 722                                 for (;*val; val++)
 723                                         if (!localeisutf8 ||
 724                                             (*val & 0xC0) != 0x80)
 725                                                 varlen++;
 726                         }
 727                         else {
 728                                 if (quotes)
 729                                         STPUTS_QUOTES(val, syntax, expdest);
 730                                 else
 731                                         STPUTS(val, expdest);
 732
 733                         }
 734                 }
 735         }
 736
 737         if (subtype == VSPLUS)
 738                 set = ! set;
 739
 740         easy = ((varflags & VSQUOTE) == 0 ||
 741                 (*var == '@' && shellparam.nparam != 1));
 742
 743
 744         switch (subtype) {
 745         case VSLENGTH:
 746                 expdest = cvtnum(varlen, expdest);
 747                 goto record;
 748
 749         case VSNORMAL:
 750                 if (!easy)
 751                         break;
 752 record:
 753                 recordregion(startloc, expdest - stackblock(),
 754                     varflags & VSQUOTE || (ifsset() && ifsval()[0] == '\0' &&
 755                     (*var == '@' || *var == '*')));
 756                 break;
 757
 758         case VSPLUS:
 759         case VSMINUS:
 760                 if (!set) {
 761                         argstr(p, flag | (flag & EXP_FULL ? EXP_SPLIT_LIT : 0) |
 762                             (varflags & VSQUOTE ? EXP_LIT_QUOTED : 0));
 763                         break;
 764                 }
 765                 if (easy)
 766                         goto record;
 767                 break;
 768
 769         case VSTRIMLEFT:
 770         case VSTRIMLEFTMAX:
 771         case VSTRIMRIGHT:
 772         case VSTRIMRIGHTMAX:
 773                 if (!set)
 774                         break;
 775                 /*
 776                  * Terminate the string and start recording the pattern
 777                  * right after it
 778                  */
 779                 STPUTC('\0', expdest);
 780                 patloc = expdest - stackblock();
 781                 if (subevalvar(p, NULL, patloc, subtype,
 782                     startloc, varflags, quotes) == 0) {
 783                         int amount = (expdest - stackblock() - patloc) + 1;
 784                         STADJUST(-amount, expdest);
 785                 }
 786                 /* Remove any recorded regions beyond start of variable */
 787                 removerecordregions(startloc);
 788                 goto record;
 789
 790         case VSASSIGN:
 791         case VSQUESTION:
 792                 if (!set) {
 793                         if (subevalvar(p, var, 0, subtype, startloc, varflags,
 794                             quotes)) {
 795                                 varflags &= ~VSNUL;
 796                                 /*
 797                                  * Remove any recorded regions beyond
 798                                  * start of variable
 799                                  */
 800                                 removerecordregions(startloc);
 801                                 goto again;
 802                         }
 803                         break;
 804                 }
 805                 if (easy)
 806                         goto record;
 807                 break;
 808
 809         case VSERROR:
 810                 c = p - var - 1;
 811                 error("${%.*s%s}: Bad substitution", c, var,
 812                     (c > 0 && *p != CTLENDVAR) ? "..." : "");
 813
 814         default:
 815                 abort();
 816         }
 817         p[-1] = '=';    /* recover overwritten '=' */
 818
 819         if (subtype != VSNORMAL) {      /* skip to end of alternative */
 820                 int nesting = 1;
 821                 for (;;) {
 822                         if ((c = *p++) == CTLESC)
 823                                 p++;
 824                         else if (c == CTLBACKQ || c == (CTLBACKQ|CTLQUOTE)) {
 825                                 if (set)
 826                                         argbackq = argbackq->next;
 827                         } else if (c == CTLVAR) {
 828                                 if ((*p++ & VSTYPE) != VSNORMAL)
 829                                         nesting++;
 830                         } else if (c == CTLENDVAR) {
 831                                 if (--nesting == 0)
 832                                         break;
 833                         }
 834                 }
 835         }
 836         return p;
 837 }
 838
 839
 840
 841 /*
 842  * Test whether a specialized variable is set.
 843  */
 844
 845 static int
 846 varisset(char *name, int nulok)
 847 {
 848
 849         if (*name == '!')
 850                 return backgndpidset();
 851         else if (*name == '@' || *name == '*') {
 852                 if (*shellparam.p == NULL)
 853                         return 0;
 854
 855                 if (nulok) {
 856                         char **av;
 857
 858                         for (av = shellparam.p; *av; av++)
 859                                 if (**av != '\0')
 860                                         return 1;
 861                         return 0;
 862                 }
 863         } else if (is_digit(*name)) {
 864                 char *ap;
 865                 int num = atoi(name);
 866
 867                 if (num > shellparam.nparam)
 868                         return 0;
 869
 870                 if (num == 0)
 871                         ap = arg0;
 872                 else
 873                         ap = shellparam.p[num - 1];
 874
 875                 if (nulok && (ap == NULL || *ap == '\0'))
 876                         return 0;
 877         }
 878         return 1;
 879 }
 880
 881 static void
 882 strtodest(const char *p, int flag, int subtype, int quoted)
 883 {
 884         if (flag & (EXP_FULL | EXP_CASE) && subtype != VSLENGTH)
 885                 STPUTS_QUOTES(p, quoted ? DQSYNTAX : BASESYNTAX, expdest);
 886         else
 887                 STPUTS(p, expdest);
 888 }
 889
 890 /*
 891  * Add the value of a specialized variable to the stack string.
 892  */
 893
 894 static void
 895 varvalue(char *name, int quoted, int subtype, int flag)
 896 {
 897         int num;
 898         char *p;
 899         int i;
 900         char sep;
 901         char **ap;
 902
 903         switch (*name) {
 904         case '$':
 905                 num = rootpid;
 906                 goto numvar;
 907         case '?':
 908                 num = oexitstatus;
 909                 goto numvar;
 910         case '#':
 911                 num = shellparam.nparam;
 912                 goto numvar;
 913         case '!':
 914                 num = backgndpidval();
 915 numvar:
 916                 expdest = cvtnum(num, expdest);
 917                 break;
 918         case '-':
 919                 for (i = 0 ; i < NOPTS ; i++) {
 920                         if (optlist[i].val)
 921                                 STPUTC(optlist[i].letter, expdest);
 922                 }
 923                 break;
 924         case '@':
 925                 if (flag & EXP_FULL && quoted) {
 926                         for (ap = shellparam.p ; (p = *ap++) != NULL ; ) {
 927                                 strtodest(p, flag, subtype, quoted);
 928                                 if (*ap)
 929                                         STPUTC('\0', expdest);
 930                         }
 931                         break;
 932                 }
 933                 /* FALLTHROUGH */
 934         case '*':
 935                 if (ifsset())
 936                         sep = ifsval()[0];
 937                 else
 938                         sep = ' ';
 939                 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) {
 940                         strtodest(p, flag, subtype, quoted);
 941                         if (!*ap)
 942                                 break;
 943                         if (sep || (flag & EXP_FULL && !quoted && **ap != '\0'))
 944                                 STPUTC(sep, expdest);
 945                 }
 946                 break;
 947         case '0':
 948                 p = arg0;
 949                 strtodest(p, flag, subtype, quoted);
 950                 break;
 951         default:
 952                 if (is_digit(*name)) {
 953                         num = atoi(name);
 954                         if (num > 0 && num <= shellparam.nparam) {
 955                                 p = shellparam.p[num - 1];
 956                                 strtodest(p, flag, subtype, quoted);
 957                         }
 958                 }
 959                 break;
 960         }
 961 }
 962
 963
 964
 965 /*
 966  * Record the fact that we have to scan this region of the
 967  * string for IFS characters.
 968  */
 969
 970 static void
 971 recordregion(int start, int end, int inquotes)
 972 {
 973         struct ifsregion *ifsp;
 974
 975         if (ifslastp == NULL) {
 976                 ifsp = &ifsfirst;
 977         } else {
 978                 if (ifslastp->endoff == start
 979                     && ifslastp->inquotes == inquotes) {
 980                         /* extend previous area */
 981                         ifslastp->endoff = end;
 982                         return;
 983                 }
 984                 ifsp = (struct ifsregion *)ckmalloc(sizeof (struct ifsregion));
 985                 ifslastp->next = ifsp;
 986         }
 987         ifslastp = ifsp;
 988         ifslastp->next = NULL;
 989         ifslastp->begoff = start;
 990         ifslastp->endoff = end;
 991         ifslastp->inquotes = inquotes;
 992 }
 993
 994
 995
 996 /*
 997  * Break the argument string into pieces based upon IFS and add the
 998  * strings to the argument list.  The regions of the string to be
 999  * searched for IFS characters have been stored by recordregion.
1000  * CTLESC characters are preserved but have little effect in this pass
1001  * other than escaping CTL* characters.  In particular, they do not escape
1002  * IFS characters: that should be done with the ifsregion mechanism.
1003  * CTLQUOTEMARK characters are used to preserve empty quoted strings.
1004  * This pass treats them as a regular character, making the string non-empty.
1005  * Later, they are removed along with the other CTL* characters.
1006  */
1007 static void
1008 ifsbreakup(char *string, struct arglist *arglist)
1009 {
1010         struct ifsregion *ifsp;
1011         struct strlist *sp;
1012         char *start;
1013         char *p;
1014         char *q;
1015         const char *ifs;
1016         const char *ifsspc;
1017         int had_param_ch = 0;
1018
1019         start = string;
1020
1021         if (ifslastp == NULL) {
1022                 /* Return entire argument, IFS doesn't apply to any of it */
1023                 sp = (struct strlist *)stalloc(sizeof *sp);
1024                 sp->text = start;
1025                 *arglist->lastp = sp;
1026                 arglist->lastp = &sp->next;
1027                 return;
1028         }
1029
1030         ifs = ifsset() ? ifsval() : " \t\n";
1031
1032         for (ifsp = &ifsfirst; ifsp != NULL; ifsp = ifsp->next) {
1033                 p = string + ifsp->begoff;
1034                 while (p < string + ifsp->endoff) {
1035                         q = p;
1036                         if (*p == CTLESC)
1037                                 p++;
1038                         if (ifsp->inquotes) {
1039                                 /* Only NULs (should be from "$@") end args */
1040                                 had_param_ch = 1;
1041                                 if (*p != 0) {
1042                                         p++;
1043                                         continue;
1044                                 }
1045                                 ifsspc = NULL;
1046                         } else {
1047                                 if (!strchr(ifs, *p)) {
1048                                         had_param_ch = 1;
1049                                         p++;
1050                                         continue;
1051                                 }
1052                                 ifsspc = strchr(" \t\n", *p);
1053
1054                                 /* Ignore IFS whitespace at start */
1055                                 if (q == start && ifsspc != NULL) {
1056                                         p++;
1057                                         start = p;
1058                                         continue;
1059                                 }
1060                                 had_param_ch = 0;
1061                         }
1062
1063                         /* Save this argument... */
1064                         *q = '\0';
1065                         sp = (struct strlist *)stalloc(sizeof *sp);
1066                         sp->text = start;
1067                         *arglist->lastp = sp;
1068                         arglist->lastp = &sp->next;
1069                         p++;
1070
1071                         if (ifsspc != NULL) {
1072                                 /* Ignore further trailing IFS whitespace */
1073                                 for (; p < string + ifsp->endoff; p++) {
1074                                         q = p;
1075                                         if (*p == CTLESC)
1076                                                 p++;
1077                                         if (strchr(ifs, *p) == NULL) {
1078                                                 p = q;
1079                                                 break;
1080                                         }
1081                                         if (strchr(" \t\n", *p) == NULL) {
1082                                                 p++;
1083                                                 break;
1084                                         }
1085                                 }
1086                         }
1087                         start = p;
1088                 }
1089         }
1090
1091         /*
1092          * Save anything left as an argument.
1093          * Traditionally we have treated 'IFS=':'; set -- x$IFS' as
1094          * generating 2 arguments, the second of which is empty.
1095          * Some recent clarification of the Posix spec say that it
1096          * should only generate one....
1097          */
1098         if (had_param_ch || *start != 0) {
1099                 sp = (struct strlist *)stalloc(sizeof *sp);
1100                 sp->text = start;
1101                 *arglist->lastp = sp;
1102                 arglist->lastp = &sp->next;
1103         }
1104 }
1105
1106
1107 static char expdir[PATH_MAX];
1108 #define expdir_end (expdir + sizeof(expdir))
1109
1110 /*
1111  * Perform pathname generation and remove control characters.
1112  * At this point, the only control characters should be CTLESC and CTLQUOTEMARK.
1113  * The results are stored in the list exparg.
1114  */
1115 static void
1116 expandmeta(struct strlist *str, int flag __unused)
1117 {
1118         char *p;
1119         struct strlist **savelastp;
1120         struct strlist *sp;
1121         char c;
1122         /* TODO - EXP_REDIR */
1123
1124         while (str) {
1125                 if (fflag)
1126                         goto nometa;
1127                 p = str->text;
1128                 for (;;) {                      /* fast check for meta chars */
1129                         if ((c = *p++) == '\0')
1130                                 goto nometa;
1131                         if (c == '*' || c == '?' || c == '[')
1132                                 break;
1133                 }
1134                 savelastp = exparg.lastp;
1135                 INTOFF;
1136                 expmeta(expdir, str->text);
1137                 INTON;
1138                 if (exparg.lastp == savelastp) {
1139                         /*
1140                          * no matches
1141                          */
1142 nometa:
1143                         *exparg.lastp = str;
1144                         rmescapes(str->text);
1145                         exparg.lastp = &str->next;
1146                 } else {
1147                         *exparg.lastp = NULL;
1148                         *savelastp = sp = expsort(*savelastp);
1149                         while (sp->next != NULL)
1150                                 sp = sp->next;
1151                         exparg.lastp = &sp->next;
1152                 }
1153                 str = str->next;
1154         }
1155 }
1156
1157
1158 /*
1159  * Do metacharacter (i.e. *, ?, [...]) expansion.
1160  */
1161
1162 static void
1163 expmeta(char *enddir, char *name)
1164 {
1165         char *p;
1166         char *q;
1167         char *start;
1168         char *endname;
1169         int metaflag;
1170         struct stat statb;
1171         DIR *dirp;
1172         struct dirent *dp;
1173         int atend;
1174         int matchdot;
1175         int esc;
1176         int namlen;
1177
1178         metaflag = 0;
1179         start = name;
1180         for (p = name; esc = 0, *p; p += esc + 1) {
1181                 if (*p == '*' || *p == '?')
1182                         metaflag = 1;
1183                 else if (*p == '[') {
1184                         q = p + 1;
1185                         if (*q == '!' || *q == '^')
1186                                 q++;
1187                         for (;;) {
1188                                 while (*q == CTLQUOTEMARK)
1189                                         q++;
1190                                 if (*q == CTLESC)
1191                                         q++;
1192                                 if (*q == '/' || *q == '\0')
1193                                         break;
1194                                 if (*++q == ']') {
1195                                         metaflag = 1;
1196                                         break;
1197                                 }
1198                         }
1199                 } else if (*p == '\0')
1200                         break;
1201                 else if (*p == CTLQUOTEMARK)
1202                         continue;
1203                 else {
1204                         if (*p == CTLESC)
1205                                 esc++;
1206                         if (p[esc] == '/') {
1207                                 if (metaflag)
1208                                         break;
1209                                 start = p + esc + 1;
1210                         }
1211                 }
1212         }
1213         if (metaflag == 0) {    /* we've reached the end of the file name */
1214                 if (enddir != expdir)
1215                         metaflag++;
1216                 for (p = name ; ; p++) {
1217                         if (*p == CTLQUOTEMARK)
1218                                 continue;
1219                         if (*p == CTLESC)
1220                                 p++;
1221                         *enddir++ = *p;
1222                         if (*p == '\0')
1223                                 break;
1224                         if (enddir == expdir_end)
1225                                 return;
1226                 }
1227                 if (metaflag == 0 || lstat(expdir, &statb) >= 0)
1228                         addfname(expdir);
1229                 return;
1230         }
1231         endname = p;
1232         if (start != name) {
1233                 p = name;
1234                 while (p < start) {
1235                         while (*p == CTLQUOTEMARK)
1236                                 p++;
1237                         if (*p == CTLESC)
1238                                 p++;
1239                         *enddir++ = *p++;
1240                         if (enddir == expdir_end)
1241                                 return;
1242                 }
1243         }
1244         if (enddir == expdir) {
1245                 p = __DECONST(char *, ".");
1246         } else if (enddir == expdir + 1 && *expdir == '/') {
1247                 p = __DECONST(char *, "/");
1248         } else {
1249                 p = expdir;
1250                 enddir[-1] = '\0';
1251         }
1252         if ((dirp = opendir(p)) == NULL)
1253                 return;
1254         if (enddir != expdir)
1255                 enddir[-1] = '/';
1256         if (*endname == 0) {
1257                 atend = 1;
1258         } else {
1259                 atend = 0;
1260                 *endname = '\0';
1261                 endname += esc + 1;
1262         }
1263         matchdot = 0;
1264         p = start;
1265         while (*p == CTLQUOTEMARK)
1266                 p++;
1267         if (*p == CTLESC)
1268                 p++;
1269         if (*p == '.')
1270                 matchdot++;
1271         while (! int_pending() && (dp = readdir(dirp)) != NULL) {
1272                 if (dp->d_name[0] == '.' && ! matchdot)
1273                         continue;
1274                 if (patmatch(start, dp->d_name, 0)) {
1275                         namlen = dp->d_namlen;
1276                         if (enddir + namlen + 1 > expdir_end)
1277                                 continue;
1278                         memcpy(enddir, dp->d_name, namlen + 1);
1279                         if (atend)
1280                                 addfname(expdir);
1281                         else {
1282                                 if (dp->d_type != DT_UNKNOWN &&
1283                                     dp->d_type != DT_DIR &&
1284                                     dp->d_type != DT_LNK)
1285                                         continue;
1286                                 if (enddir + namlen + 2 > expdir_end)
1287                                         continue;
1288                                 enddir[namlen] = '/';
1289                                 enddir[namlen + 1] = '\0';
1290                                 expmeta(enddir + namlen + 1, endname);
1291                         }
1292                 }
1293         }
1294         closedir(dirp);
1295         if (! atend)
1296                 endname[-esc - 1] = esc ? CTLESC : '/';
1297 }
1298
1299
1300 /*
1301  * Add a file name to the list.
1302  */
1303
1304 static void
1305 addfname(char *name)
1306 {
1307         char *p;
1308         struct strlist *sp;
1309
1310         p = stalloc(strlen(name) + 1);
1311         scopy(name, p);
1312         sp = (struct strlist *)stalloc(sizeof *sp);
1313         sp->text = p;
1314         *exparg.lastp = sp;
1315         exparg.lastp = &sp->next;
1316 }
1317
1318
1319 /*
1320  * Sort the results of file name expansion.  It calculates the number of
1321  * strings to sort and then calls msort (short for merge sort) to do the
1322  * work.
1323  */
1324
1325 static struct strlist *
1326 expsort(struct strlist *str)
1327 {
1328         int len;
1329         struct strlist *sp;
1330
1331         len = 0;
1332         for (sp = str ; sp ; sp = sp->next)
1333                 len++;
1334         return msort(str, len);
1335 }
1336
1337
1338 static struct strlist *
1339 msort(struct strlist *list, int len)
1340 {
1341         struct strlist *p, *q = NULL;
1342         struct strlist **lpp;
1343         int half;
1344         int n;
1345
1346         if (len <= 1)
1347                 return list;
1348         half = len >> 1;
1349         p = list;
1350         for (n = half ; --n >= 0 ; ) {
1351                 q = p;
1352                 p = p->next;
1353         }
1354         q->next = NULL;                 /* terminate first half of list */
1355         q = msort(list, half);          /* sort first half of list */
1356         p = msort(p, len - half);               /* sort second half */
1357         lpp = &list;
1358         for (;;) {
1359                 if (strcmp(p->text, q->text) < 0) {
1360                         *lpp = p;
1361                         lpp = &p->next;
1362                         if ((p = *lpp) == NULL) {
1363                                 *lpp = q;
1364                                 break;
1365                         }
1366                 } else {
1367                         *lpp = q;
1368                         lpp = &q->next;
1369                         if ((q = *lpp) == NULL) {
1370                                 *lpp = p;
1371                                 break;
1372                         }
1373                 }
1374         }
1375         return list;
1376 }
1377
1378
1379
1380 static wchar_t
1381 get_wc(const char **p)
1382 {
1383         wchar_t c;
1384         int chrlen;
1385
1386         chrlen = mbtowc(&c, *p, 4);
1387         if (chrlen == 0)
1388                 return 0;
1389         else if (chrlen == -1)
1390                 c = 0;
1391         else
1392                 *p += chrlen;
1393         return c;
1394 }
1395
1396
1397 /*
1398  * See if a character matches a character class, starting at the first colon
1399  * of "[:class:]".
1400  * If a valid character class is recognized, a pointer to the next character
1401  * after the final closing bracket is stored into *end, otherwise a null
1402  * pointer is stored into *end.
1403  */
1404 static int
1405 match_charclass(const char *p, wchar_t chr, const char **end)
1406 {
1407         char name[20];
1408         const char *nameend;
1409         wctype_t cclass;
1410
1411         *end = NULL;
1412         p++;
1413         nameend = strstr(p, ":]");
1414         if (nameend == NULL || nameend - p >= (int)sizeof(name) || nameend == p)
1415                 return 0;
1416         memcpy(name, p, nameend - p);
1417         name[nameend - p] = '\0';
1418         *end = nameend + 2;
1419         cclass = wctype(name);
1420         /* An unknown class matches nothing but is valid nevertheless. */
1421         if (cclass == 0)
1422                 return 0;
1423         return iswctype(chr, cclass);
1424 }
1425
1426
1427 /*
1428  * Returns true if the pattern matches the string.
1429  */
1430
1431 static int
1432 patmatch(const char *pattern, const char *string, int squoted)
1433 {
1434         const char *p, *q, *end;
1435         const char *bt_p, *bt_q;
1436         char c;
1437         wchar_t wc, wc2;
1438
1439         p = pattern;
1440         q = string;
1441         bt_p = NULL;
1442         bt_q = NULL;
1443         for (;;) {
1444                 switch (c = *p++) {
1445                 case '\0':
1446                         if (*q != '\0')
1447                                 goto backtrack;
1448                         return 1;
1449                 case CTLESC:
1450                         if (squoted && *q == CTLESC)
1451                                 q++;
1452                         if (*q++ != *p++)
1453                                 goto backtrack;
1454                         break;
1455                 case CTLQUOTEMARK:
1456                         continue;
1457                 case '?':
1458                         if (squoted && *q == CTLESC)
1459                                 q++;
1460                         if (*q == '\0')
1461                                 return 0;
1462                         if (localeisutf8) {
1463                                 wc = get_wc(&q);
1464                                 /*
1465                                  * A '?' does not match invalid UTF-8 but a
1466                                  * '*' does, so backtrack.
1467                                  */
1468                                 if (wc == 0)
1469                                         goto backtrack;
1470                         } else
1471                                 wc = (unsigned char)*q++;
1472                         break;
1473                 case '*':
1474                         c = *p;
1475                         while (c == CTLQUOTEMARK || c == '*')
1476                                 c = *++p;
1477                         /*
1478                          * If the pattern ends here, we know the string
1479                          * matches without needing to look at the rest of it.
1480                          */
1481                         if (c == '\0')
1482                                 return 1;
1483                         /*
1484                          * First try the shortest match for the '*' that
1485                          * could work. We can forget any earlier '*' since
1486                          * there is no way having it match more characters
1487                          * can help us, given that we are already here.
1488                          */
1489                         bt_p = p;
1490                         bt_q = q;
1491                         break;
1492                 case '[': {
1493                         const char *endp;
1494                         int invert, found;
1495                         wchar_t chr;
1496
1497                         endp = p;
1498                         if (*endp == '!' || *endp == '^')
1499                                 endp++;
1500                         for (;;) {
1501                                 while (*endp == CTLQUOTEMARK)
1502                                         endp++;
1503                                 if (*endp == 0)
1504                                         goto dft;               /* no matching ] */
1505                                 if (*endp == CTLESC)
1506                                         endp++;
1507                                 if (*++endp == ']')
1508                                         break;
1509                         }
1510                         invert = 0;
1511                         if (*p == '!' || *p == '^') {
1512                                 invert++;
1513                                 p++;
1514                         }
1515                         found = 0;
1516                         if (squoted && *q == CTLESC)
1517                                 q++;
1518                         if (*q == '\0')
1519                                 return 0;
1520                         if (localeisutf8) {
1521                                 chr = get_wc(&q);
1522                                 if (chr == 0)
1523                                         goto backtrack;
1524                         } else
1525                                 chr = (unsigned char)*q++;
1526                         c = *p++;
1527                         do {
1528                                 if (c == CTLQUOTEMARK)
1529                                         continue;
1530                                 if (c == '[' && *p == ':') {
1531                                         found |= match_charclass(p, chr, &end);
1532                                         if (end != NULL)
1533                                                 p = end;
1534                                 }
1535                                 if (c == CTLESC)
1536                                         c = *p++;
1537                                 if (localeisutf8 && c & 0x80) {
1538                                         p--;
1539                                         wc = get_wc(&p);
1540                                         if (wc == 0) /* bad utf-8 */
1541                                                 return 0;
1542                                 } else
1543                                         wc = (unsigned char)c;
1544                                 if (*p == '-' && p[1] != ']') {
1545                                         p++;
1546                                         while (*p == CTLQUOTEMARK)
1547                                                 p++;
1548                                         if (*p == CTLESC)
1549                                                 p++;
1550                                         if (localeisutf8) {
1551                                                 wc2 = get_wc(&p);
1552                                                 if (wc2 == 0) /* bad utf-8 */
1553                                                         return 0;
1554                                         } else
1555                                                 wc2 = (unsigned char)*p++;
1556                                         if (   collate_range_cmp(chr, wc) >= 0
1557                                             && collate_range_cmp(chr, wc2) <= 0
1558                                            )
1559                                                 found = 1;
1560                                 } else {
1561                                         if (chr == wc)
1562                                                 found = 1;
1563                                 }
1564                         } while ((c = *p++) != ']');
1565                         if (found == invert)
1566                                 goto backtrack;
1567                         break;
1568                 }
1569 dft:            default:
1570                         if (squoted && *q == CTLESC)
1571                                 q++;
1572                         if (*q == '\0')
1573                                 return 0;
1574                         if (*q++ == c)
1575                                 break;
1576 backtrack:
1577                         /*
1578                          * If we have a mismatch (other than hitting the end
1579                          * of the string), go back to the last '*' seen and
1580                          * have it match one additional character.
1581                          */
1582                         if (bt_p == NULL)
1583                                 return 0;
1584                         if (squoted && *bt_q == CTLESC)
1585                                 bt_q++;
1586                         if (*bt_q == '\0')
1587                                 return 0;
1588                         bt_q++;
1589                         p = bt_p;
1590                         q = bt_q;
1591                         break;
1592                 }
1593         }
1594 }
1595
1596
1597
1598 /*
1599  * Remove any CTLESC and CTLQUOTEMARK characters from a string.
1600  */
1601
1602 void
1603 rmescapes(char *str)
1604 {
1605         char *p, *q;
1606
1607         p = str;
1608         while (*p != CTLESC && *p != CTLQUOTEMARK && *p != CTLQUOTEEND) {
1609                 if (*p++ == '\0')
1610                         return;
1611         }
1612         q = p;
1613         while (*p) {
1614                 if (*p == CTLQUOTEMARK || *p == CTLQUOTEEND) {
1615                         p++;
1616                         continue;
1617                 }
1618                 if (*p == CTLESC)
1619                         p++;
1620                 *q++ = *p++;
1621         }
1622         *q = '\0';
1623 }
1624
1625
1626
1627 /*
1628  * See if a pattern matches in a case statement.
1629  */
1630
1631 int
1632 casematch(union node *pattern, const char *val)
1633 {
1634         struct stackmark smark;
1635         int result;
1636         char *p;
1637
1638         setstackmark(&smark);
1639         argbackq = pattern->narg.backquote;
1640         STARTSTACKSTR(expdest);
1641         ifslastp = NULL;
1642         argstr(pattern->narg.text, EXP_TILDE | EXP_CASE);
1643         STPUTC('\0', expdest);
1644         p = grabstackstr(expdest);
1645         result = patmatch(p, val, 0);
1646         popstackmark(&smark);
1647         return result;
1648 }
1649
1650 /*
1651  * Our own itoa().
1652  */
1653
1654 static char *
1655 cvtnum(int num, char *buf)
1656 {
1657         char temp[32];
1658         int neg = num < 0;
1659         char *p = temp + 31;
1660
1661         temp[31] = '\0';
1662
1663         do {
1664                 *--p = num % 10 + '0';
1665         } while ((num /= 10) != 0);
1666
1667         if (neg)
1668                 *--p = '-';
1669
1670         STPUTS(p, buf);
1671         return buf;
1672 }
1673
1674 /*
1675  * Do most of the work for wordexp(3).
1676  */
1677
1678 int
1679 wordexpcmd(int argc, char **argv)
1680 {
1681         size_t len;
1682         int i;
1683
1684         out1fmt("%08x", argc - 1);
1685         for (i = 1, len = 0; i < argc; i++)
1686                 len += strlen(argv[i]);
1687         out1fmt("%08x", (int)len);
1688         for (i = 1; i < argc; i++)
1689                 outbin(argv[i], strlen(argv[i]) + 1, out1);
1690         return (0);
1691 }