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