src/eval.c

   1 /*      $OpenBSD: eval.c,v 1.40 2013/09/14 20:09:30 millert Exp $       */
   2
   3 /*-
   4  * Copyright (c) 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010,
   5  *               2011, 2012, 2013, 2014
   6  *      Thorsten Glaser <tg@mirbsd.org>
   7  *
   8  * Provided that these terms and disclaimer and all copyright notices
   9  * are retained or reproduced in an accompanying document, permission
  10  * is granted to deal in this work without restriction, including un-
  11  * limited rights to use, publicly perform, distribute, sell, modify,
  12  * merge, give away, or sublicence.
  13  *
  14  * This work is provided "AS IS" and WITHOUT WARRANTY of any kind, to
  15  * the utmost extent permitted by applicable law, neither express nor
  16  * implied; without malicious intent or gross negligence. In no event
  17  * may a licensor, author or contributor be held liable for indirect,
  18  * direct, other damage, loss, or other issues arising in any way out
  19  * of dealing in the work, even if advised of the possibility of such
  20  * damage or existence of a defect, except proven that it results out
  21  * of said person's immediate fault when using the work as intended.
  22  */
  23
  24 #include "sh.h"
  25
  26 __RCSID("$MirOS: src/bin/mksh/eval.c,v 1.150 2014/06/09 11:16:07 tg Exp $");
  27
  28 /*
  29  * string expansion
  30  *
  31  * first pass: quoting, IFS separation, ~, ${}, $() and $(()) substitution.
  32  * second pass: alternation ({,}), filename expansion (*?[]).
  33  */
  34
  35 /* expansion generator state */
  36 typedef struct {
  37         /* not including an "int type;" member, see expand() */
  38         /* string */
  39         const char *str;
  40         /* source */
  41         union {
  42                 /* string[] */
  43                 const char **strv;
  44                 /* file */
  45                 struct shf *shf;
  46         } u;
  47         /* variable in ${var...} */
  48         struct tbl *var;
  49         /* split "$@" / call waitlast in $() */
  50         bool split;
  51 } Expand;
  52
  53 #define XBASE           0       /* scanning original */
  54 #define XSUB            1       /* expanding ${} string */
  55 #define XARGSEP         2       /* ifs0 between "$*" */
  56 #define XARG            3       /* expanding $*, $@ */
  57 #define XCOM            4       /* expanding $() */
  58 #define XNULLSUB        5       /* "$@" when $# is 0 (don't generate word) */
  59 #define XSUBMID         6       /* middle of expanding ${} */
  60
  61 /* States used for field splitting */
  62 #define IFS_WORD        0       /* word has chars (or quotes) */
  63 #define IFS_WS          1       /* have seen IFS white-space */
  64 #define IFS_NWS         2       /* have seen IFS non-white-space */
  65
  66 static int varsub(Expand *, const char *, const char *, int *, int *);
  67 static int comsub(Expand *, const char *, int);
  68 static char *valsub(struct op *, Area *);
  69 static char *trimsub(char *, char *, int);
  70 static void glob(char *, XPtrV *, bool);
  71 static void globit(XString *, char **, char *, XPtrV *, int);
  72 static const char *maybe_expand_tilde(const char *, XString *, char **, int);
  73 #ifndef MKSH_NOPWNAM
  74 static char *homedir(char *);
  75 #endif
  76 static void alt_expand(XPtrV *, char *, char *, char *, int);
  77 static int utflen(const char *) MKSH_A_PURE;
  78 static void utfincptr(const char *, mksh_ari_t *);
  79
  80 /* UTFMODE functions */
  81 static int
  82 utflen(const char *s)
  83 {
  84         size_t n;
  85
  86         if (UTFMODE) {
  87                 n = 0;
  88                 while (*s) {
  89                         s += utf_ptradj(s);
  90                         ++n;
  91                 }
  92         } else
  93                 n = strlen(s);
  94
  95         if (n > 2147483647)
  96                 n = 2147483647;
  97         return ((int)n);
  98 }
  99
 100 static void
 101 utfincptr(const char *s, mksh_ari_t *lp)
 102 {
 103         const char *cp = s;
 104
 105         while ((*lp)--)
 106                 cp += utf_ptradj(cp);
 107         *lp = cp - s;
 108 }
 109
 110 /* compile and expand word */
 111 char *
 112 substitute(const char *cp, int f)
 113 {
 114         struct source *s, *sold;
 115
 116         sold = source;
 117         s = pushs(SWSTR, ATEMP);
 118         s->start = s->str = cp;
 119         source = s;
 120         if (yylex(ONEWORD) != LWORD)
 121                 internal_errorf("bad substitution");
 122         source = sold;
 123         afree(s, ATEMP);
 124         return (evalstr(yylval.cp, f));
 125 }
 126
 127 /*
 128  * expand arg-list
 129  */
 130 char **
 131 eval(const char **ap, int f)
 132 {
 133         XPtrV w;
 134
 135         if (*ap == NULL) {
 136                 union mksh_ccphack vap;
 137
 138                 vap.ro = ap;
 139                 return (vap.rw);
 140         }
 141         XPinit(w, 32);
 142         /* space for shell name */
 143         XPput(w, NULL);
 144         while (*ap != NULL)
 145                 expand(*ap++, &w, f);
 146         XPput(w, NULL);
 147         return ((char **)XPclose(w) + 1);
 148 }
 149
 150 /*
 151  * expand string
 152  */
 153 char *
 154 evalstr(const char *cp, int f)
 155 {
 156         XPtrV w;
 157         char *dp = null;
 158
 159         XPinit(w, 1);
 160         expand(cp, &w, f);
 161         if (XPsize(w))
 162                 dp = *XPptrv(w);
 163         XPfree(w);
 164         return (dp);
 165 }
 166
 167 /*
 168  * expand string - return only one component
 169  * used from iosetup to expand redirection files
 170  */
 171 char *
 172 evalonestr(const char *cp, int f)
 173 {
 174         XPtrV w;
 175         char *rv;
 176
 177         XPinit(w, 1);
 178         expand(cp, &w, f);
 179         switch (XPsize(w)) {
 180         case 0:
 181                 rv = null;
 182                 break;
 183         case 1:
 184                 rv = (char *) *XPptrv(w);
 185                 break;
 186         default:
 187                 rv = evalstr(cp, f&~DOGLOB);
 188                 break;
 189         }
 190         XPfree(w);
 191         return (rv);
 192 }
 193
 194 /* for nested substitution: ${var:=$var2} */
 195 typedef struct SubType {
 196         struct tbl *var;        /* variable for ${var..} */
 197         struct SubType *prev;   /* old type */
 198         struct SubType *next;   /* poped type (to avoid re-allocating) */
 199         size_t  base;           /* begin position of expanded word */
 200         short   stype;          /* [=+-?%#] action after expanded word */
 201         short   f;              /* saved value of f (DOPAT, etc) */
 202         uint8_t quotep;         /* saved value of quote (for ${..[%#]..}) */
 203         uint8_t quotew;         /* saved value of quote (for ${..[+-=]..}) */
 204 } SubType;
 205
 206 void
 207 expand(
 208     /* input word */
 209     const char *ccp,
 210     /* output words */
 211     XPtrV *wp,
 212     /* DO* flags */
 213     int f)
 214 {
 215         int c = 0;
 216         /* expansion type */
 217         int type;
 218         /* quoted */
 219         int quote = 0;
 220         /* destination string and live pointer */
 221         XString ds;
 222         char *dp;
 223         /* source */
 224         const char *sp;
 225         /* second pass flags */
 226         int fdo;
 227         /* have word */
 228         int word;
 229         /* field splitting of parameter/command substitution */
 230         int doblank;
 231         /* expansion variables */
 232         Expand x = {
 233                 NULL, { NULL }, NULL, 0
 234         };
 235         SubType st_head, *st;
 236         /* record number of trailing newlines in COMSUB */
 237         int newlines = 0;
 238         bool saw_eq, make_magic;
 239         int tilde_ok;
 240         size_t len;
 241         char *cp;
 242
 243         if (ccp == NULL)
 244                 internal_errorf("expand(NULL)");
 245         /* for alias, readonly, set, typeset commands */
 246         if ((f & DOVACHECK) && is_wdvarassign(ccp)) {
 247                 f &= ~(DOVACHECK|DOBLANK|DOGLOB|DOTILDE);
 248                 f |= DOASNTILDE;
 249         }
 250         if (Flag(FNOGLOB))
 251                 f &= ~DOGLOB;
 252         if (Flag(FMARKDIRS))
 253                 f |= DOMARKDIRS;
 254         if (Flag(FBRACEEXPAND) && (f & DOGLOB))
 255                 f |= DOBRACE;
 256
 257         /* init destination string */
 258         Xinit(ds, dp, 128, ATEMP);
 259         type = XBASE;
 260         sp = ccp;
 261         fdo = 0;
 262         saw_eq = false;
 263         /* must be 1/0 */
 264         tilde_ok = (f & (DOTILDE|DOASNTILDE)) ? 1 : 0;
 265         doblank = 0;
 266         make_magic = false;
 267         word = (f&DOBLANK) ? IFS_WS : IFS_WORD;
 268         /* clang doesn't know OSUBST comes before CSUBST */
 269         memset(&st_head, 0, sizeof(st_head));
 270         st = &st_head;
 271
 272         while (/* CONSTCOND */ 1) {
 273                 Xcheck(ds, dp);
 274
 275                 switch (type) {
 276                 case XBASE:
 277                         /* original prefixed string */
 278                         c = *sp++;
 279                         switch (c) {
 280                         case EOS:
 281                                 c = 0;
 282                                 break;
 283                         case CHAR:
 284                                 c = *sp++;
 285                                 break;
 286                         case QCHAR:
 287                                 /* temporary quote */
 288                                 quote |= 2;
 289                                 c = *sp++;
 290                                 break;
 291                         case OQUOTE:
 292                                 word = IFS_WORD;
 293                                 tilde_ok = 0;
 294                                 quote = 1;
 295                                 continue;
 296                         case CQUOTE:
 297                                 quote = st->quotew;
 298                                 continue;
 299                         case COMSUB:
 300                         case FUNSUB:
 301                         case VALSUB:
 302                                 tilde_ok = 0;
 303                                 if (f & DONTRUNCOMMAND) {
 304                                         word = IFS_WORD;
 305                                         *dp++ = '$';
 306                                         *dp++ = c == COMSUB ? '(' : '{';
 307                                         if (c != COMSUB)
 308                                                 *dp++ = c == FUNSUB ? ' ' : '|';
 309                                         while (*sp != '\0') {
 310                                                 Xcheck(ds, dp);
 311                                                 *dp++ = *sp++;
 312                                         }
 313                                         if (c != COMSUB) {
 314                                                 *dp++ = ';';
 315                                                 *dp++ = '}';
 316                                         } else
 317                                                 *dp++ = ')';
 318                                 } else {
 319                                         type = comsub(&x, sp, c);
 320                                         if (type != XBASE && (f & DOBLANK))
 321                                                 doblank++;
 322                                         sp = strnul(sp) + 1;
 323                                         newlines = 0;
 324                                 }
 325                                 continue;
 326                         case EXPRSUB:
 327                                 tilde_ok = 0;
 328                                 if (f & DONTRUNCOMMAND) {
 329                                         word = IFS_WORD;
 330                                         *dp++ = '$'; *dp++ = '('; *dp++ = '(';
 331                                         while (*sp != '\0') {
 332                                                 Xcheck(ds, dp);
 333                                                 *dp++ = *sp++;
 334                                         }
 335                                         *dp++ = ')'; *dp++ = ')';
 336                                 } else {
 337                                         struct tbl v;
 338
 339                                         v.flag = DEFINED|ISSET|INTEGER;
 340                                         /* not default */
 341                                         v.type = 10;
 342                                         v.name[0] = '\0';
 343                                         v_evaluate(&v, substitute(sp, 0),
 344                                             KSH_UNWIND_ERROR, true);
 345                                         sp = strnul(sp) + 1;
 346                                         x.str = str_val(&v);
 347                                         type = XSUB;
 348                                         if (f & DOBLANK)
 349                                                 doblank++;
 350                                 }
 351                                 continue;
 352                         case OSUBST: {
 353                                 /* ${{#}var{:}[=+-?#%]word} */
 354                         /*-
 355                          * format is:
 356                          *      OSUBST [{x] plain-variable-part \0
 357                          *          compiled-word-part CSUBST [}x]
 358                          * This is where all syntax checking gets done...
 359                          */
 360                                 /* skip the { or x (}) */
 361                                 const char *varname = ++sp;
 362                                 int stype;
 363                                 int slen = 0;
 364
 365                                 /* skip variable */
 366                                 sp = cstrchr(sp, '\0') + 1;
 367                                 type = varsub(&x, varname, sp, &stype, &slen);
 368                                 if (type < 0) {
 369                                         char *beg, *end, *str;
 370  unwind_substsyn:
 371                                         /* restore sp */
 372                                         sp = varname - 2;
 373                                         end = (beg = wdcopy(sp, ATEMP)) +
 374                                             (wdscan(sp, CSUBST) - sp);
 375                                         /* ({) the } or x is already skipped */
 376                                         if (end < wdscan(beg, EOS))
 377                                                 *end = EOS;
 378                                         str = snptreef(NULL, 64, "%S", beg);
 379                                         afree(beg, ATEMP);
 380                                         errorf("%s: %s", str, "bad substitution");
 381                                 }
 382                                 if (f & DOBLANK)
 383                                         doblank++;
 384                                 tilde_ok = 0;
 385                                 if (type == XBASE) {
 386                                         /* expand? */
 387                                         if (!st->next) {
 388                                                 SubType *newst;
 389
 390                                                 newst = alloc(sizeof(SubType), ATEMP);
 391                                                 newst->next = NULL;
 392                                                 newst->prev = st;
 393                                                 st->next = newst;
 394                                         }
 395                                         st = st->next;
 396                                         st->stype = stype;
 397                                         st->base = Xsavepos(ds, dp);
 398                                         st->f = f;
 399                                         if (x.var == &vtemp) {
 400                                                 st->var = tempvar();
 401                                                 st->var->flag &= ~INTEGER;
 402                                                 /* can't fail here */
 403                                                 setstr(st->var,
 404                                                     str_val(x.var),
 405                                                     KSH_RETURN_ERROR | 0x4);
 406                                         } else
 407                                                 st->var = x.var;
 408
 409                                         st->quotew = st->quotep = quote;
 410                                         /* skip qualifier(s) */
 411                                         if (stype)
 412                                                 sp += slen;
 413                                         switch (stype & 0x17F) {
 414                                         case 0x100 | '#':
 415                                                 x.str = shf_smprintf("%08X",
 416                                                     (unsigned int)hash(str_val(st->var)));
 417                                                 break;
 418                                         case 0x100 | 'Q': {
 419                                                 struct shf shf;
 420
 421                                                 shf_sopen(NULL, 0, SHF_WR|SHF_DYNAMIC, &shf);
 422                                                 print_value_quoted(&shf, str_val(st->var));
 423                                                 x.str = shf_sclose(&shf);
 424                                                 break;
 425                                         }
 426                                         case '0': {
 427                                                 char *beg, *mid, *end, *stg;
 428                                                 mksh_ari_t from = 0, num = -1, flen, finc = 0;
 429
 430                                                 beg = wdcopy(sp, ATEMP);
 431                                                 mid = beg + (wdscan(sp, ADELIM) - sp);
 432                                                 stg = beg + (wdscan(sp, CSUBST) - sp);
 433                                                 if (mid >= stg)
 434                                                         goto unwind_substsyn;
 435                                                 mid[-2] = EOS;
 436                                                 if (mid[-1] == /*{*/'}') {
 437                                                         sp += mid - beg - 1;
 438                                                         end = NULL;
 439                                                 } else {
 440                                                         end = mid +
 441                                                             (wdscan(mid, ADELIM) - mid);
 442                                                         if (end >= stg ||
 443                                                             /* more than max delimiters */
 444                                                             end[-1] != /*{*/ '}')
 445                                                                 goto unwind_substsyn;
 446                                                         end[-2] = EOS;
 447                                                         sp += end - beg - 1;
 448                                                 }
 449                                                 evaluate(substitute(stg = wdstrip(beg, 0), 0),
 450                                                     &from, KSH_UNWIND_ERROR, true);
 451                                                 afree(stg, ATEMP);
 452                                                 if (end) {
 453                                                         evaluate(substitute(stg = wdstrip(mid, 0), 0),
 454                                                             &num, KSH_UNWIND_ERROR, true);
 455                                                         afree(stg, ATEMP);
 456                                                 }
 457                                                 afree(beg, ATEMP);
 458                                                 beg = str_val(st->var);
 459                                                 flen = utflen(beg);
 460                                                 if (from < 0) {
 461                                                         if (-from < flen)
 462                                                                 finc = flen + from;
 463                                                 } else
 464                                                         finc = from < flen ? from : flen;
 465                                                 if (UTFMODE)
 466                                                         utfincptr(beg, &finc);
 467                                                 beg += finc;
 468                                                 flen = utflen(beg);
 469                                                 if (num < 0 || num > flen)
 470                                                         num = flen;
 471                                                 if (UTFMODE)
 472                                                         utfincptr(beg, &num);
 473                                                 strndupx(x.str, beg, num, ATEMP);
 474                                                 goto do_CSUBST;
 475                                         }
 476                                         case '/': {
 477                                                 char *s, *p, *d, *sbeg, *end;
 478                                                 char *pat, *rrep;
 479                                                 char *tpat0, *tpat1, *tpat2;
 480
 481                                                 s = wdcopy(sp, ATEMP);
 482                                                 p = s + (wdscan(sp, ADELIM) - sp);
 483                                                 d = s + (wdscan(sp, CSUBST) - sp);
 484                                                 if (p >= d)
 485                                                         goto unwind_substsyn;
 486                                                 p[-2] = EOS;
 487                                                 if (p[-1] == /*{*/'}')
 488                                                         d = NULL;
 489                                                 else
 490                                                         d[-2] = EOS;
 491                                                 sp += (d ? d : p) - s - 1;
 492                                                 tpat0 = wdstrip(s,
 493                                                     WDS_KEEPQ | WDS_MAGIC);
 494                                                 pat = substitute(tpat0, 0);
 495                                                 if (d) {
 496                                                         d = wdstrip(p, WDS_KEEPQ);
 497                                                         rrep = substitute(d, 0);
 498                                                         afree(d, ATEMP);
 499                                                 } else
 500                                                         rrep = null;
 501                                                 afree(s, ATEMP);
 502                                                 s = d = pat;
 503                                                 while (*s)
 504                                                         if (*s != '\\' ||
 505                                                             s[1] == '%' ||
 506                                                             s[1] == '#' ||
 507                                                             s[1] == '\0' ||
 508                                 /* XXX really? */           s[1] == '\\' ||
 509                                                             s[1] == '/')
 510                                                                 *d++ = *s++;
 511                                                         else
 512                                                                 s++;
 513                                                 *d = '\0';
 514                                                 afree(tpat0, ATEMP);
 515
 516                                                 /* check for special cases */
 517                                                 d = str_val(st->var);
 518                                                 mkssert(d != NULL);
 519                                                 switch (*pat) {
 520                                                 case '#':
 521                                                         /* anchor at begin */
 522                                                         tpat0 = pat + 1;
 523                                                         tpat1 = rrep;
 524                                                         tpat2 = d;
 525                                                         break;
 526                                                 case '%':
 527                                                         /* anchor at end */
 528                                                         tpat0 = pat + 1;
 529                                                         tpat1 = d;
 530                                                         tpat2 = rrep;
 531                                                         break;
 532                                                 case '\0':
 533                                                         /* empty pattern */
 534                                                         goto no_repl;
 535                                                 default:
 536                                                         tpat0 = pat;
 537                                                         /* silence gcc */
 538                                                         tpat1 = tpat2 = NULL;
 539                                                 }
 540                                                 if (gmatchx(null, tpat0, false)) {
 541                                                         /*
 542                                                          * pattern matches
 543                                                          * the empty string
 544                                                          */
 545                                                         if (tpat0 == pat)
 546                                                                 goto no_repl;
 547                                                         /* but is anchored */
 548                                                         s = shf_smprintf("%s%s",
 549                                                             tpat1, tpat2);
 550                                                         goto do_repl;
 551                                                 }
 552
 553                                                 /* prepare string on which to work */
 554                                                 strdupx(s, d, ATEMP);
 555                                                 sbeg = s;
 556
 557                                                 /* first see if we have any match at all */
 558                                                 tpat0 = pat;
 559                                                 if (*pat == '#') {
 560                                                         /* anchor at the beginning */
 561                                                         tpat1 = shf_smprintf("%s%c*", ++tpat0, MAGIC);
 562                                                         tpat2 = tpat1;
 563                                                 } else if (*pat == '%') {
 564                                                         /* anchor at the end */
 565                                                         tpat1 = shf_smprintf("%c*%s", MAGIC, ++tpat0);
 566                                                         tpat2 = tpat0;
 567                                                 } else {
 568                                                         /* float */
 569                                                         tpat1 = shf_smprintf("%c*%s%c*", MAGIC, pat, MAGIC);
 570                                                         tpat2 = tpat1 + 2;
 571                                                 }
 572  again_repl:
 573                                                 /*
 574                                                  * this would not be necessary if gmatchx would return
 575                                                  * the start and end values of a match found, like re*
 576                                                  */
 577                                                 if (!gmatchx(sbeg, tpat1, false))
 578                                                         goto end_repl;
 579                                                 end = strnul(s);
 580                                                 /* now anchor the beginning of the match */
 581                                                 if (*pat != '#')
 582                                                         while (sbeg <= end) {
 583                                                                 if (gmatchx(sbeg, tpat2, false))
 584                                                                         break;
 585                                                                 else
 586                                                                         sbeg++;
 587                                                         }
 588                                                 /* now anchor the end of the match */
 589                                                 p = end;
 590                                                 if (*pat != '%')
 591                                                         while (p >= sbeg) {
 592                                                                 bool gotmatch;
 593
 594                                                                 c = *p;
 595                                                                 *p = '\0';
 596                                                                 gotmatch = tobool(gmatchx(sbeg, tpat0, false));
 597                                                                 *p = c;
 598                                                                 if (gotmatch)
 599                                                                         break;
 600                                                                 p--;
 601                                                         }
 602                                                 strndupx(end, s, sbeg - s, ATEMP);
 603                                                 d = shf_smprintf("%s%s%s", end, rrep, p);
 604                                                 afree(end, ATEMP);
 605                                                 sbeg = d + (sbeg - s) + strlen(rrep);
 606                                                 afree(s, ATEMP);
 607                                                 s = d;
 608                                                 if (stype & 0x80)
 609                                                         goto again_repl;
 610  end_repl:
 611                                                 afree(tpat1, ATEMP);
 612  do_repl:
 613                                                 x.str = s;
 614  no_repl:
 615                                                 afree(pat, ATEMP);
 616                                                 if (rrep != null)
 617                                                         afree(rrep, ATEMP);
 618                                                 goto do_CSUBST;
 619                                         }
 620                                         case '#':
 621                                         case '%':
 622                                                 /* ! DOBLANK,DOBRACE,DOTILDE */
 623                                                 f = (f & DONTRUNCOMMAND) |
 624                                                     DOPAT | DOTEMP;
 625                                                 st->quotew = quote = 0;
 626                                                 /*
 627                                                  * Prepend open pattern (so |
 628                                                  * in a trim will work as
 629                                                  * expected)
 630                                                  */
 631                                                 if (!Flag(FSH)) {
 632                                                         *dp++ = MAGIC;
 633                                                         *dp++ = '@' | 0x80;
 634                                                 }
 635                                                 break;
 636                                         case '=':
 637                                                 /*
 638                                                  * Enabling tilde expansion
 639                                                  * after :s here is
 640                                                  * non-standard ksh, but is
 641                                                  * consistent with rules for
 642                                                  * other assignments. Not
 643                                                  * sure what POSIX thinks of
 644                                                  * this.
 645                                                  * Not doing tilde expansion
 646                                                  * for integer variables is a
 647                                                  * non-POSIX thing - makes
 648                                                  * sense though, since ~ is
 649                                                  * a arithmetic operator.
 650                                                  */
 651                                                 if (!(x.var->flag & INTEGER))
 652                                                         f |= DOASNTILDE|DOTILDE;
 653                                                 f |= DOTEMP;
 654                                                 /*
 655                                                  * These will be done after the
 656                                                  * value has been assigned.
 657                                                  */
 658                                                 f &= ~(DOBLANK|DOGLOB|DOBRACE);
 659                                                 tilde_ok = 1;
 660                                                 break;
 661                                         case '?':
 662                                                 f &= ~DOBLANK;
 663                                                 f |= DOTEMP;
 664                                                 /* FALLTHROUGH */
 665                                         default:
 666                                                 /* Enable tilde expansion */
 667                                                 tilde_ok = 1;
 668                                                 f |= DOTILDE;
 669                                         }
 670                                 } else
 671                                         /* skip word */
 672                                         sp += wdscan(sp, CSUBST) - sp;
 673                                 continue;
 674                         }
 675                         case CSUBST:
 676                                 /* only get here if expanding word */
 677  do_CSUBST:
 678                                 /* ({) skip the } or x */
 679                                 sp++;
 680                                 /* in case of ${unset:-} */
 681                                 tilde_ok = 0;
 682                                 *dp = '\0';
 683                                 quote = st->quotep;
 684                                 f = st->f;
 685                                 if (f & DOBLANK)
 686                                         doblank--;
 687                                 switch (st->stype & 0x17F) {
 688                                 case '#':
 689                                 case '%':
 690                                         if (!Flag(FSH)) {
 691                                                 /* Append end-pattern */
 692                                                 *dp++ = MAGIC;
 693                                                 *dp++ = ')';
 694                                         }
 695                                         *dp = '\0';
 696                                         dp = Xrestpos(ds, dp, st->base);
 697                                         /*
 698                                          * Must use st->var since calling
 699                                          * global would break things
 700                                          * like x[i+=1].
 701                                          */
 702                                         x.str = trimsub(str_val(st->var),
 703                                                 dp, st->stype);
 704                                         if (x.str[0] != '\0') {
 705                                                 word = IFS_WS;
 706                                                 type = XSUB;
 707                                         } else
 708                                                 type = quote ? XSUB : XNULLSUB;
 709                                         if (f & DOBLANK)
 710                                                 doblank++;
 711                                         st = st->prev;
 712                                         continue;
 713                                 case '=':
 714                                         /*
 715                                          * Restore our position and substitute
 716                                          * the value of st->var (may not be
 717                                          * the assigned value in the presence
 718                                          * of integer/right-adj/etc attributes).
 719                                          */
 720                                         dp = Xrestpos(ds, dp, st->base);
 721                                         /*
 722                                          * Must use st->var since calling
 723                                          * global would cause with things
 724                                          * like x[i+=1] to be evaluated twice.
 725                                          */
 726                                         /*
 727                                          * Note: not exported by FEXPORT
 728                                          * in AT&T ksh.
 729                                          */
 730                                         /*
 731                                          * XXX POSIX says readonly is only
 732                                          * fatal for special builtins (setstr
 733                                          * does readonly check).
 734                                          */
 735                                         len = strlen(dp) + 1;
 736                                         setstr(st->var,
 737                                             debunk(alloc(len, ATEMP),
 738                                             dp, len), KSH_UNWIND_ERROR);
 739                                         x.str = str_val(st->var);
 740                                         type = XSUB;
 741                                         if (f & DOBLANK)
 742                                                 doblank++;
 743                                         st = st->prev;
 744                                         continue;
 745                                 case '?': {
 746                                         char *s = Xrestpos(ds, dp, st->base);
 747
 748                                         errorf("%s: %s", st->var->name,
 749                                             dp == s ?
 750                                             "parameter null or not set" :
 751                                             (debunk(s, s, strlen(s) + 1), s));
 752                                 }
 753                                 case '0':
 754                                 case '/':
 755                                 case 0x100 | '#':
 756                                 case 0x100 | 'Q':
 757                                         dp = Xrestpos(ds, dp, st->base);
 758                                         type = XSUB;
 759                                         if (f & DOBLANK)
 760                                                 doblank++;
 761                                         st = st->prev;
 762                                         continue;
 763                                 }
 764                                 st = st->prev;
 765                                 type = XBASE;
 766                                 continue;
 767
 768                         case OPAT:
 769                                 /* open pattern: *(foo|bar) */
 770                                 /* Next char is the type of pattern */
 771                                 make_magic = true;
 772                                 c = *sp++ | 0x80;
 773                                 break;
 774
 775                         case SPAT:
 776                                 /* pattern separator (|) */
 777                                 make_magic = true;
 778                                 c = '|';
 779                                 break;
 780
 781                         case CPAT:
 782                                 /* close pattern */
 783                                 make_magic = true;
 784                                 c = /*(*/ ')';
 785                                 break;
 786                         }
 787                         break;
 788
 789                 case XNULLSUB:
 790                         /*
 791                          * Special case for "$@" (and "${foo[@]}") - no
 792                          * word is generated if $# is 0 (unless there is
 793                          * other stuff inside the quotes).
 794                          */
 795                         type = XBASE;
 796                         if (f & DOBLANK) {
 797                                 doblank--;
 798                                 /*
 799                                  * XXX not really correct:
 800                                  *      x=; "$x$@"
 801                                  * should generate a null argument and
 802                                  *      set A; "${@:+}"
 803                                  * shouldn't.
 804                                  */
 805                                 if (dp == Xstring(ds, dp))
 806                                         word = IFS_WS;
 807                         }
 808                         continue;
 809
 810                 case XSUB:
 811                 case XSUBMID:
 812                         if ((c = *x.str++) == 0) {
 813                                 type = XBASE;
 814                                 if (f & DOBLANK)
 815                                         doblank--;
 816                                 continue;
 817                         }
 818                         break;
 819
 820                 case XARGSEP:
 821                         type = XARG;
 822                         quote = 1;
 823                         /* FALLTHROUGH */
 824                 case XARG:
 825                         if ((c = *x.str++) == '\0') {
 826                                 /*
 827                                  * force null words to be created so
 828                                  * set -- '' 2 ''; foo "$@" will do
 829                                  * the right thing
 830                                  */
 831                                 if (quote && x.split)
 832                                         word = IFS_WORD;
 833                                 if ((x.str = *x.u.strv++) == NULL) {
 834                                         type = XBASE;
 835                                         if (f & DOBLANK)
 836                                                 doblank--;
 837                                         continue;
 838                                 }
 839                                 c = ifs0;
 840                                 if (c == 0) {
 841                                         if (quote && !x.split)
 842                                                 continue;
 843                                         /* this is so we don't terminate */
 844                                         c = ' ';
 845                                         /* now force-emit a word */
 846                                         goto emit_word;
 847                                 }
 848                                 if (quote && x.split) {
 849                                         /* terminate word for "$@" */
 850                                         type = XARGSEP;
 851                                         quote = 0;
 852                                 }
 853                         }
 854                         break;
 855
 856                 case XCOM:
 857                         if (x.u.shf == NULL) {
 858                                 /* $(<...) failed */
 859                                 subst_exstat = 1;
 860                                 /* fake EOF */
 861                                 c = EOF;
 862                         } else if (newlines) {
 863                                 /* spit out saved NLs */
 864                                 c = '\n';
 865                                 --newlines;
 866                         } else {
 867                                 while ((c = shf_getc(x.u.shf)) == 0 || c == '\n')
 868                                         if (c == '\n')
 869                                                 /* save newlines */
 870                                                 newlines++;
 871                                 if (newlines && c != EOF) {
 872                                         shf_ungetc(c, x.u.shf);
 873                                         c = '\n';
 874                                         --newlines;
 875                                 }
 876                         }
 877                         if (c == EOF) {
 878                                 newlines = 0;
 879                                 if (x.u.shf)
 880                                         shf_close(x.u.shf);
 881                                 if (x.split)
 882                                         subst_exstat = waitlast();
 883                                 type = XBASE;
 884                                 if (f & DOBLANK)
 885                                         doblank--;
 886                                 continue;
 887                         }
 888                         break;
 889                 }
 890
 891                 /* check for end of word or IFS separation */
 892                 if (c == 0 || (!quote && (f & DOBLANK) && doblank &&
 893                     !make_magic && ctype(c, C_IFS))) {
 894                         /*-
 895                          * How words are broken up:
 896                          *                      |       value of c
 897                          *      word            |       ws      nws     0
 898                          *      -----------------------------------
 899                          *      IFS_WORD                w/WS    w/NWS   w
 900                          *      IFS_WS                  -/WS    w/NWS   -
 901                          *      IFS_NWS                 -/NWS   w/NWS   w
 902                          * (w means generate a word)
 903                          * Note that IFS_NWS/0 generates a word (AT&T ksh
 904                          * doesn't do this, but POSIX does).
 905                          */
 906                         if (word == IFS_WORD ||
 907                             (!ctype(c, C_IFSWS) && c && word == IFS_NWS)) {
 908  emit_word:
 909                                 *dp++ = '\0';
 910                                 cp = Xclose(ds, dp);
 911                                 if (fdo & DOBRACE)
 912                                         /* also does globbing */
 913                                         alt_expand(wp, cp, cp,
 914                                             cp + Xlength(ds, (dp - 1)),
 915                                             fdo | (f & DOMARKDIRS));
 916                                 else if (fdo & DOGLOB)
 917                                         glob(cp, wp, tobool(f & DOMARKDIRS));
 918                                 else if ((f & DOPAT) || !(fdo & DOMAGIC))
 919                                         XPput(*wp, cp);
 920                                 else
 921                                         XPput(*wp, debunk(cp, cp,
 922                                             strlen(cp) + 1));
 923                                 fdo = 0;
 924                                 saw_eq = false;
 925                                 tilde_ok = (f & (DOTILDE|DOASNTILDE)) ? 1 : 0;
 926                                 if (c == 0)
 927                                         return;
 928                                 Xinit(ds, dp, 128, ATEMP);
 929                         } else if (c == 0) {
 930                                 return;
 931                         } else if (type == XSUB && ctype(c, C_IFS) &&
 932                             !ctype(c, C_IFSWS) && Xlength(ds, dp) == 0) {
 933                                 *(cp = alloc(1, ATEMP)) = '\0';
 934                                 XPput(*wp, cp);
 935                                 type = XSUBMID;
 936                         }
 937                         if (word != IFS_NWS)
 938                                 word = ctype(c, C_IFSWS) ? IFS_WS : IFS_NWS;
 939                 } else {
 940                         if (type == XSUB) {
 941                                 if (word == IFS_NWS &&
 942                                     Xlength(ds, dp) == 0) {
 943                                         *(cp = alloc(1, ATEMP)) = '\0';
 944                                         XPput(*wp, cp);
 945                                 }
 946                                 type = XSUBMID;
 947                         }
 948
 949                         /* age tilde_ok info - ~ code tests second bit */
 950                         tilde_ok <<= 1;
 951                         /* mark any special second pass chars */
 952                         if (!quote)
 953                                 switch (c) {
 954                                 case '[':
 955                                 case '!':
 956                                 case '-':
 957                                 case ']':
 958                                         /*
 959                                          * For character classes - doesn't hurt
 960                                          * to have magic !,-,]s outside of
 961                                          * [...] expressions.
 962                                          */
 963                                         if (f & (DOPAT | DOGLOB)) {
 964                                                 fdo |= DOMAGIC;
 965                                                 if (c == '[')
 966                                                         fdo |= f & DOGLOB;
 967                                                 *dp++ = MAGIC;
 968                                         }
 969                                         break;
 970                                 case '*':
 971                                 case '?':
 972                                         if (f & (DOPAT | DOGLOB)) {
 973                                                 fdo |= DOMAGIC | (f & DOGLOB);
 974                                                 *dp++ = MAGIC;
 975                                         }
 976                                         break;
 977                                 case '{':
 978                                 case '}':
 979                                 case ',':
 980                                         if ((f & DOBRACE) && (c == '{' /*}*/ ||
 981                                             (fdo & DOBRACE))) {
 982                                                 fdo |= DOBRACE|DOMAGIC;
 983                                                 *dp++ = MAGIC;
 984                                         }
 985                                         break;
 986                                 case '=':
 987                                         /* Note first unquoted = for ~ */
 988                                         if (!(f & DOTEMP) && !saw_eq &&
 989                                             (Flag(FBRACEEXPAND) ||
 990                                             (f & DOASNTILDE))) {
 991                                                 saw_eq = true;
 992                                                 tilde_ok = 1;
 993                                         }
 994                                         break;
 995                                 case ':':
 996                                         /* : */
 997                                         /* Note unquoted : for ~ */
 998                                         if (!(f & DOTEMP) && (f & DOASNTILDE))
 999                                                 tilde_ok = 1;
1000                                         break;
1001                                 case '~':
1002                                         /*
1003                                          * tilde_ok is reset whenever
1004                                          * any of ' " $( $(( ${ } are seen.
1005                                          * Note that tilde_ok must be preserved
1006                                          * through the sequence ${A=a=}~
1007                                          */
1008                                         if (type == XBASE &&
1009                                             (f & (DOTILDE|DOASNTILDE)) &&
1010                                             (tilde_ok & 2)) {
1011                                                 const char *tcp;
1012                                                 char *tdp = dp;
1013
1014                                                 tcp = maybe_expand_tilde(sp,
1015                                                     &ds, &tdp,
1016                                                     f & DOASNTILDE);
1017                                                 if (tcp) {
1018                                                         if (dp != tdp)
1019                                                                 word = IFS_WORD;
1020                                                         dp = tdp;
1021                                                         sp = tcp;
1022                                                         continue;
1023                                                 }
1024                                         }
1025                                         break;
1026                                 }
1027                         else
1028                                 /* undo temporary */
1029                                 quote &= ~2;
1030
1031                         if (make_magic) {
1032                                 make_magic = false;
1033                                 fdo |= DOMAGIC | (f & DOGLOB);
1034                                 *dp++ = MAGIC;
1035                         } else if (ISMAGIC(c)) {
1036                                 fdo |= DOMAGIC;
1037                                 *dp++ = MAGIC;
1038                         }
1039                         /* save output char */
1040                         *dp++ = c;
1041                         word = IFS_WORD;
1042                 }
1043         }
1044 }
1045
1046 /*
1047  * Prepare to generate the string returned by ${} substitution.
1048  */
1049 static int
1050 varsub(Expand *xp, const char *sp, const char *word,
1051     int *stypep,        /* becomes qualifier type */
1052     int *slenp)         /* " " len (=, :=, etc.) valid iff *stypep != 0 */
1053 {
1054         int c;
1055         int state;      /* next state: XBASE, XARG, XSUB, XNULLSUB */
1056         int stype;      /* substitution type */
1057         int slen;
1058         const char *p;
1059         struct tbl *vp;
1060         bool zero_ok = false;
1061
1062         if ((stype = sp[0]) == '\0')
1063                 /* Bad variable name */
1064                 return (-1);
1065
1066         xp->var = NULL;
1067
1068         /*-
1069          * ${#var}, string length (-U: characters, +U: octets) or array size
1070          * ${%var}, string width (-U: screen columns, +U: octets)
1071          */
1072         c = sp[1];
1073         if (stype == '%' && c == '\0')
1074                 return (-1);
1075         if ((stype == '#' || stype == '%') && c != '\0') {
1076                 /* Can't have any modifiers for ${#...} or ${%...} */
1077                 if (*word != CSUBST)
1078                         return (-1);
1079                 sp++;
1080                 /* Check for size of array */
1081                 if ((p = cstrchr(sp, '[')) && (p[1] == '*' || p[1] == '@') &&
1082                     p[2] == ']') {
1083                         int n = 0;
1084
1085                         if (stype != '#')
1086                                 return (-1);
1087                         vp = global(arrayname(sp));
1088                         if (vp->flag & (ISSET|ARRAY))
1089                                 zero_ok = true;
1090                         for (; vp; vp = vp->u.array)
1091                                 if (vp->flag & ISSET)
1092                                         n++;
1093                         c = n;
1094                 } else if (c == '*' || c == '@') {
1095                         if (stype != '#')
1096                                 return (-1);
1097                         c = e->loc->argc;
1098                 } else {
1099                         p = str_val(global(sp));
1100                         zero_ok = p != null;
1101                         if (stype == '#')
1102                                 c = utflen(p);
1103                         else {
1104                                 /* partial utf_mbswidth reimplementation */
1105                                 const char *s = p;
1106                                 unsigned int wc;
1107                                 size_t len;
1108                                 int cw;
1109
1110                                 c = 0;
1111                                 while (*s) {
1112                                         if (!UTFMODE || (len = utf_mbtowc(&wc,
1113                                             s)) == (size_t)-1)
1114                                                 /* not UTFMODE or not UTF-8 */
1115                                                 wc = (unsigned char)(*s++);
1116                                         else
1117                                                 /* UTFMODE and UTF-8 */
1118                                                 s += len;
1119                                         /* wc == char or wchar at s++ */
1120                                         if ((cw = utf_wcwidth(wc)) == -1) {
1121                                                 /* 646, 8859-1, 10646 C0/C1 */
1122                                                 c = -1;
1123                                                 break;
1124                                         }
1125                                         c += cw;
1126                                 }
1127                         }
1128                 }
1129                 if (Flag(FNOUNSET) && c == 0 && !zero_ok)
1130                         errorf("%s: %s", sp, "parameter not set");
1131                 /* unqualified variable/string substitution */
1132                 *stypep = 0;
1133                 xp->str = shf_smprintf("%d", c);
1134                 return (XSUB);
1135         }
1136
1137         /* Check for qualifiers in word part */
1138         stype = 0;
1139         c = word[slen = 0] == CHAR ? word[1] : 0;
1140         if (c == ':') {
1141                 slen += 2;
1142                 stype = 0x80;
1143                 c = word[slen + 0] == CHAR ? word[slen + 1] : 0;
1144         }
1145         if (!stype && c == '/') {
1146                 slen += 2;
1147                 stype = c;
1148                 if (word[slen] == ADELIM) {
1149                         slen += 2;
1150                         stype |= 0x80;
1151                 }
1152         } else if (stype == 0x80 && (c == ' ' || c == '0')) {
1153                 stype |= '0';
1154         } else if (ctype(c, C_SUBOP1)) {
1155                 slen += 2;
1156                 stype |= c;
1157         } else if (ctype(c, C_SUBOP2)) {
1158                 /* Note: ksh88 allows :%, :%%, etc */
1159                 slen += 2;
1160                 stype = c;
1161                 if (word[slen + 0] == CHAR && c == word[slen + 1]) {
1162                         stype |= 0x80;
1163                         slen += 2;
1164                 }
1165         } else if (c == '@') {
1166                 /* @x where x is command char */
1167                 slen += 2;
1168                 stype |= 0x100;
1169                 if (word[slen] == CHAR) {
1170                         stype |= word[slen + 1];
1171                         slen += 2;
1172                 }
1173         } else if (stype)
1174                 /* : is not ok */
1175                 return (-1);
1176         if (!stype && *word != CSUBST)
1177                 return (-1);
1178         *stypep = stype;
1179         *slenp = slen;
1180
1181         c = sp[0];
1182         if (c == '*' || c == '@') {
1183                 switch (stype & 0x17F) {
1184                 /* can't assign to a vector */
1185                 case '=':
1186                 /* can't trim a vector (yet) */
1187                 case '%':
1188                 case '#':
1189                 case '0':
1190                 case '/':
1191                 case 0x100 | '#':
1192                 case 0x100 | 'Q':
1193                         return (-1);
1194                 }
1195                 if (e->loc->argc == 0) {
1196                         xp->str = null;
1197                         xp->var = global(sp);
1198                         state = c == '@' ? XNULLSUB : XSUB;
1199                 } else {
1200                         xp->u.strv = (const char **)e->loc->argv + 1;
1201                         xp->str = *xp->u.strv++;
1202                         /* $@ */
1203                         xp->split = tobool(c == '@');
1204                         state = XARG;
1205                 }
1206                 /* POSIX 2009? */
1207                 zero_ok = true;
1208         } else {
1209                 if ((p = cstrchr(sp, '[')) && (p[1] == '*' || p[1] == '@') &&
1210                     p[2] == ']') {
1211                         XPtrV wv;
1212
1213                         switch (stype & 0x17F) {
1214                         /* can't assign to a vector */
1215                         case '=':
1216                         /* can't trim a vector (yet) */
1217                         case '%':
1218                         case '#':
1219                         case '?':
1220                         case '0':
1221                         case '/':
1222                         case 0x100 | '#':
1223                         case 0x100 | 'Q':
1224                                 return (-1);
1225                         }
1226                         XPinit(wv, 32);
1227                         if ((c = sp[0]) == '!')
1228                                 ++sp;
1229                         vp = global(arrayname(sp));
1230                         for (; vp; vp = vp->u.array) {
1231                                 if (!(vp->flag&ISSET))
1232                                         continue;
1233                                 XPput(wv, c == '!' ? shf_smprintf("%lu",
1234                                     arrayindex(vp)) :
1235                                     str_val(vp));
1236                         }
1237                         if (XPsize(wv) == 0) {
1238                                 xp->str = null;
1239                                 state = p[1] == '@' ? XNULLSUB : XSUB;
1240                                 XPfree(wv);
1241                         } else {
1242                                 XPput(wv, 0);
1243                                 xp->u.strv = (const char **)XPptrv(wv);
1244                                 xp->str = *xp->u.strv++;
1245                                 /* ${foo[@]} */
1246                                 xp->split = tobool(p[1] == '@');
1247                                 state = XARG;
1248                         }
1249                 } else {
1250                         /* Can't assign things like $! or $1 */
1251                         if ((stype & 0x17F) == '=' &&
1252                             ctype(*sp, C_VAR1 | C_DIGIT))
1253                                 return (-1);
1254                         if (*sp == '!' && sp[1]) {
1255                                 ++sp;
1256                                 xp->var = global(sp);
1257                                 if (vstrchr(sp, '['))
1258                                         xp->str = shf_smprintf("%s[%lu]",
1259                                             xp->var->name,
1260                                             arrayindex(xp->var));
1261                                 else
1262                                         xp->str = xp->var->name;
1263                         } else {
1264                                 xp->var = global(sp);
1265                                 xp->str = str_val(xp->var);
1266                         }
1267                         state = XSUB;
1268                 }
1269         }
1270
1271         c = stype & 0x7F;
1272         /* test the compiler's code generator */
1273         if (((stype < 0x100) && (ctype(c, C_SUBOP2) || c == '/' ||
1274             (((stype&0x80) ? *xp->str=='\0' : xp->str==null) ? /* undef? */
1275             c == '=' || c == '-' || c == '?' : c == '+'))) ||
1276             stype == (0x80 | '0') || stype == (0x100 | '#') ||
1277             stype == (0x100 | 'Q'))
1278                 /* expand word instead of variable value */
1279                 state = XBASE;
1280         if (Flag(FNOUNSET) && xp->str == null && !zero_ok &&
1281             (ctype(c, C_SUBOP2) || (state != XBASE && c != '+')))
1282                 errorf("%s: %s", sp, "parameter not set");
1283         return (state);
1284 }
1285
1286 /*
1287  * Run the command in $(...) and read its output.
1288  */
1289 static int
1290 comsub(Expand *xp, const char *cp, int fn MKSH_A_UNUSED)
1291 {
1292         Source *s, *sold;
1293         struct op *t;
1294         struct shf *shf;
1295         uint8_t old_utfmode = UTFMODE;
1296
1297         s = pushs(SSTRING, ATEMP);
1298         s->start = s->str = cp;
1299         sold = source;
1300         t = compile(s, true);
1301         afree(s, ATEMP);
1302         source = sold;
1303
1304         UTFMODE = old_utfmode;
1305
1306         if (t == NULL)
1307                 return (XBASE);
1308
1309         /* no waitlast() unless specifically enabled later */
1310         xp->split = false;
1311
1312         if (t->type == TCOM &&
1313             *t->args == NULL && *t->vars == NULL && t->ioact != NULL) {
1314                 /* $(<file) */
1315                 struct ioword *io = *t->ioact;
1316                 char *name;
1317
1318                 if ((io->flag & IOTYPE) != IOREAD)
1319                         errorf("%s: %s", "funny $() command",
1320                             snptreef(NULL, 32, "%R", io));
1321                 shf = shf_open(name = evalstr(io->name, DOTILDE), O_RDONLY, 0,
1322                         SHF_MAPHI|SHF_CLEXEC);
1323                 if (shf == NULL)
1324                         warningf(!Flag(FTALKING), "%s: %s %s: %s", name,
1325                             "can't open", "$(<...) input", cstrerror(errno));
1326         } else if (fn == FUNSUB) {
1327                 int ofd1;
1328                 struct temp *tf = NULL;
1329
1330                 /*
1331                  * create a temporary file, open for reading and writing,
1332                  * with an shf open for reading (buffered) but yet unused
1333                  */
1334                 maketemp(ATEMP, TT_FUNSUB, &tf);
1335                 if (!tf->shf) {
1336                         errorf("can't %s temporary file %s: %s",
1337                             "create", tf->tffn, cstrerror(errno));
1338                 }
1339                 /* extract shf from temporary file, unlink and free it */
1340                 shf = tf->shf;
1341                 unlink(tf->tffn);
1342                 afree(tf, ATEMP);
1343                 /* save stdout and let it point to the tempfile */
1344                 ofd1 = savefd(1);
1345                 ksh_dup2(shf_fileno(shf), 1, false);
1346                 /*
1347                  * run tree, with output thrown into the tempfile,
1348                  * in a new function block
1349                  */
1350                 valsub(t, NULL);
1351                 subst_exstat = exstat & 0xFF;
1352                 /* rewind the tempfile and restore regular stdout */
1353                 lseek(shf_fileno(shf), (off_t)0, SEEK_SET);
1354                 restfd(1, ofd1);
1355         } else if (fn == VALSUB) {
1356                 xp->str = valsub(t, ATEMP);
1357                 subst_exstat = exstat & 0xFF;
1358                 return (XSUB);
1359         } else {
1360                 int ofd1, pv[2];
1361
1362                 openpipe(pv);
1363                 shf = shf_fdopen(pv[0], SHF_RD, NULL);
1364                 ofd1 = savefd(1);
1365                 if (pv[1] != 1) {
1366                         ksh_dup2(pv[1], 1, false);
1367                         close(pv[1]);
1368                 }
1369                 execute(t, XXCOM | XPIPEO | XFORK, NULL);
1370                 restfd(1, ofd1);
1371                 startlast();
1372                 /* waitlast() */
1373                 xp->split = true;
1374         }
1375
1376         xp->u.shf = shf;
1377         return (XCOM);
1378 }
1379
1380 /*
1381  * perform #pattern and %pattern substitution in ${}
1382  */
1383 static char *
1384 trimsub(char *str, char *pat, int how)
1385 {
1386         char *end = strnul(str);
1387         char *p, c;
1388
1389         switch (how & 0xFF) {
1390         case '#':
1391                 /* shortest match at beginning */
1392                 for (p = str; p <= end; p += utf_ptradj(p)) {
1393                         c = *p; *p = '\0';
1394                         if (gmatchx(str, pat, false)) {
1395                                 *p = c;
1396                                 return (p);
1397                         }
1398                         *p = c;
1399                 }
1400                 break;
1401         case '#'|0x80:
1402                 /* longest match at beginning */
1403                 for (p = end; p >= str; p--) {
1404                         c = *p; *p = '\0';
1405                         if (gmatchx(str, pat, false)) {
1406                                 *p = c;
1407                                 return (p);
1408                         }
1409                         *p = c;
1410                 }
1411                 break;
1412         case '%':
1413                 /* shortest match at end */
1414                 p = end;
1415                 while (p >= str) {
1416                         if (gmatchx(p, pat, false))
1417                                 goto trimsub_match;
1418                         if (UTFMODE) {
1419                                 char *op = p;
1420                                 while ((p-- > str) && ((*p & 0xC0) == 0x80))
1421                                         ;
1422                                 if ((p < str) || (p + utf_ptradj(p) != op))
1423                                         p = op - 1;
1424                         } else
1425                                 --p;
1426                 }
1427                 break;
1428         case '%'|0x80:
1429                 /* longest match at end */
1430                 for (p = str; p <= end; p++)
1431                         if (gmatchx(p, pat, false)) {
1432  trimsub_match:
1433                                 strndupx(end, str, p - str, ATEMP);
1434                                 return (end);
1435                         }
1436                 break;
1437         }
1438
1439         /* no match, return string */
1440         return (str);
1441 }
1442
1443 /*
1444  * glob
1445  * Name derived from V6's /etc/glob, the program that expanded filenames.
1446  */
1447
1448 /* XXX cp not const 'cause slashes are temporarily replaced with NULs... */
1449 static void
1450 glob(char *cp, XPtrV *wp, bool markdirs)
1451 {
1452         int oldsize = XPsize(*wp);
1453
1454         if (glob_str(cp, wp, markdirs) == 0)
1455                 XPput(*wp, debunk(cp, cp, strlen(cp) + 1));
1456         else
1457                 qsort(XPptrv(*wp) + oldsize, XPsize(*wp) - oldsize,
1458                     sizeof(void *), xstrcmp);
1459 }
1460
1461 #define GF_NONE         0
1462 #define GF_EXCHECK      BIT(0)          /* do existence check on file */
1463 #define GF_GLOBBED      BIT(1)          /* some globbing has been done */
1464 #define GF_MARKDIR      BIT(2)          /* add trailing / to directories */
1465
1466 /*
1467  * Apply file globbing to cp and store the matching files in wp. Returns
1468  * the number of matches found.
1469  */
1470 int
1471 glob_str(char *cp, XPtrV *wp, bool markdirs)
1472 {
1473         int oldsize = XPsize(*wp);
1474         XString xs;
1475         char *xp;
1476
1477         Xinit(xs, xp, 256, ATEMP);
1478         globit(&xs, &xp, cp, wp, markdirs ? GF_MARKDIR : GF_NONE);
1479         Xfree(xs, xp);
1480
1481         return (XPsize(*wp) - oldsize);
1482 }
1483
1484 static void
1485 globit(XString *xs,     /* dest string */
1486     char **xpp,         /* ptr to dest end */
1487     char *sp,           /* source path */
1488     XPtrV *wp,          /* output list */
1489     int check)          /* GF_* flags */
1490 {
1491         char *np;               /* next source component */
1492         char *xp = *xpp;
1493         char *se;
1494         char odirsep;
1495
1496         /* This to allow long expansions to be interrupted */
1497         intrcheck();
1498
1499         if (sp == NULL) {
1500                 /* end of source path */
1501                 /*
1502                  * We only need to check if the file exists if a pattern
1503                  * is followed by a non-pattern (eg, foo*x/bar; no check
1504                  * is needed for foo* since the match must exist) or if
1505                  * any patterns were expanded and the markdirs option is set.
1506                  * Symlinks make things a bit tricky...
1507                  */
1508                 if ((check & GF_EXCHECK) ||
1509                     ((check & GF_MARKDIR) && (check & GF_GLOBBED))) {
1510 #define stat_check()    (stat_done ? stat_done : (stat_done = \
1511                             stat(Xstring(*xs, xp), &statb) < 0 ? -1 : 1))
1512                         struct stat lstatb, statb;
1513                         /* -1: failed, 1 ok, 0 not yet done */
1514                         int stat_done = 0;
1515
1516                         if (mksh_lstat(Xstring(*xs, xp), &lstatb) < 0)
1517                                 return;
1518                         /*
1519                          * special case for systems which strip trailing
1520                          * slashes from regular files (eg, /etc/passwd/).
1521                          * SunOS 4.1.3 does this...
1522                          */
1523                         if ((check & GF_EXCHECK) && xp > Xstring(*xs, xp) &&
1524                             xp[-1] == '/' && !S_ISDIR(lstatb.st_mode) &&
1525                             (!S_ISLNK(lstatb.st_mode) ||
1526                             stat_check() < 0 || !S_ISDIR(statb.st_mode)))
1527                                 return;
1528                         /*
1529                          * Possibly tack on a trailing / if there isn't already
1530                          * one and if the file is a directory or a symlink to a
1531                          * directory
1532                          */
1533                         if (((check & GF_MARKDIR) && (check & GF_GLOBBED)) &&
1534                             xp > Xstring(*xs, xp) && xp[-1] != '/' &&
1535                             (S_ISDIR(lstatb.st_mode) ||
1536                             (S_ISLNK(lstatb.st_mode) && stat_check() > 0 &&
1537                             S_ISDIR(statb.st_mode)))) {
1538                                 *xp++ = '/';
1539                                 *xp = '\0';
1540                         }
1541                 }
1542                 strndupx(np, Xstring(*xs, xp), Xlength(*xs, xp), ATEMP);
1543                 XPput(*wp, np);
1544                 return;
1545         }
1546
1547         if (xp > Xstring(*xs, xp))
1548                 *xp++ = '/';
1549         while (*sp == '/') {
1550                 Xcheck(*xs, xp);
1551                 *xp++ = *sp++;
1552         }
1553         np = strchr(sp, '/');
1554         if (np != NULL) {
1555                 se = np;
1556                 /* don't assume '/', can be multiple kinds */
1557                 odirsep = *np;
1558                 *np++ = '\0';
1559         } else {
1560                 odirsep = '\0'; /* keep gcc quiet */
1561                 se = sp + strlen(sp);
1562         }
1563
1564
1565         /*
1566          * Check if sp needs globbing - done to avoid pattern checks for strings
1567          * containing MAGIC characters, open [s without the matching close ],
1568          * etc. (otherwise opendir() will be called which may fail because the
1569          * directory isn't readable - if no globbing is needed, only execute
1570          * permission should be required (as per POSIX)).
1571          */
1572         if (!has_globbing(sp, se)) {
1573                 XcheckN(*xs, xp, se - sp + 1);
1574                 debunk(xp, sp, Xnleft(*xs, xp));
1575                 xp += strlen(xp);
1576                 *xpp = xp;
1577                 globit(xs, xpp, np, wp, check);
1578         } else {
1579                 DIR *dirp;
1580                 struct dirent *d;
1581                 char *name;
1582                 size_t len, prefix_len;
1583
1584                 /* xp = *xpp;   copy_non_glob() may have re-alloc'd xs */
1585                 *xp = '\0';
1586                 prefix_len = Xlength(*xs, xp);
1587                 dirp = opendir(prefix_len ? Xstring(*xs, xp) : ".");
1588                 if (dirp == NULL)
1589                         goto Nodir;
1590                 while ((d = readdir(dirp)) != NULL) {
1591                         name = d->d_name;
1592                         if (name[0] == '.' &&
1593                             (name[1] == 0 || (name[1] == '.' && name[2] == 0)))
1594                                 /* always ignore . and .. */
1595                                 continue;
1596                         if ((*name == '.' && *sp != '.') ||
1597                             !gmatchx(name, sp, true))
1598                                 continue;
1599
1600                         len = strlen(d->d_name) + 1;
1601                         XcheckN(*xs, xp, len);
1602                         memcpy(xp, name, len);
1603                         *xpp = xp + len - 1;
1604                         globit(xs, xpp, np, wp,
1605                                 (check & GF_MARKDIR) | GF_GLOBBED
1606                                 | (np ? GF_EXCHECK : GF_NONE));
1607                         xp = Xstring(*xs, xp) + prefix_len;
1608                 }
1609                 closedir(dirp);
1610  Nodir:
1611                 ;
1612         }
1613
1614         if (np != NULL)
1615                 *--np = odirsep;
1616 }
1617
1618 /* remove MAGIC from string */
1619 char *
1620 debunk(char *dp, const char *sp, size_t dlen)
1621 {
1622         char *d;
1623         const char *s;
1624
1625         if ((s = cstrchr(sp, MAGIC))) {
1626                 if (s - sp >= (ssize_t)dlen)
1627                         return (dp);
1628                 memmove(dp, sp, s - sp);
1629                 for (d = dp + (s - sp); *s && (d - dp < (ssize_t)dlen); s++)
1630                         if (!ISMAGIC(*s) || !(*++s & 0x80) ||
1631                             !vstrchr("*+?@! ", *s & 0x7f))
1632                                 *d++ = *s;
1633                         else {
1634                                 /* extended pattern operators: *+?@! */
1635                                 if ((*s & 0x7f) != ' ')
1636                                         *d++ = *s & 0x7f;
1637                                 if (d - dp < (ssize_t)dlen)
1638                                         *d++ = '(';
1639                         }
1640                 *d = '\0';
1641         } else if (dp != sp)
1642                 strlcpy(dp, sp, dlen);
1643         return (dp);
1644 }
1645
1646 /*
1647  * Check if p is an unquoted name, possibly followed by a / or :. If so
1648  * puts the expanded version in *dcp,dp and returns a pointer in p just
1649  * past the name, otherwise returns 0.
1650  */
1651 static const char *
1652 maybe_expand_tilde(const char *p, XString *dsp, char **dpp, int isassign)
1653 {
1654         XString ts;
1655         char *dp = *dpp;
1656         char *tp;
1657         const char *r;
1658
1659         Xinit(ts, tp, 16, ATEMP);
1660         /* : only for DOASNTILDE form */
1661         while (p[0] == CHAR && p[1] != '/' && (!isassign || p[1] != ':'))
1662         {
1663                 Xcheck(ts, tp);
1664                 *tp++ = p[1];
1665                 p += 2;
1666         }
1667         *tp = '\0';
1668         r = (p[0] == EOS || p[0] == CHAR || p[0] == CSUBST) ?
1669             tilde(Xstring(ts, tp)) : NULL;
1670         Xfree(ts, tp);
1671         if (r) {
1672                 while (*r) {
1673                         Xcheck(*dsp, dp);
1674                         if (ISMAGIC(*r))
1675                                 *dp++ = MAGIC;
1676                         *dp++ = *r++;
1677                 }
1678                 *dpp = dp;
1679                 r = p;
1680         }
1681         return (r);
1682 }
1683
1684 /*
1685  * tilde expansion
1686  *
1687  * based on a version by Arnold Robbins
1688  */
1689
1690 char *
1691 tilde(char *cp)
1692 {
1693         char *dp = null;
1694
1695         if (cp[0] == '\0')
1696                 dp = str_val(global("HOME"));
1697         else if (cp[0] == '+' && cp[1] == '\0')
1698                 dp = str_val(global("PWD"));
1699         else if (cp[0] == '-' && cp[1] == '\0')
1700                 dp = str_val(global("OLDPWD"));
1701 #ifndef MKSH_NOPWNAM
1702         else
1703                 dp = homedir(cp);
1704 #endif
1705         /* If HOME, PWD or OLDPWD are not set, don't expand ~ */
1706         return (dp == null ? NULL : dp);
1707 }
1708
1709 #ifndef MKSH_NOPWNAM
1710 /*
1711  * map userid to user's home directory.
1712  * note that 4.3's getpw adds more than 6K to the shell,
1713  * and the YP version probably adds much more.
1714  * we might consider our own version of getpwnam() to keep the size down.
1715  */
1716 static char *
1717 homedir(char *name)
1718 {
1719         struct tbl *ap;
1720
1721         ap = ktenter(&homedirs, name, hash(name));
1722         if (!(ap->flag & ISSET)) {
1723                 struct passwd *pw;
1724
1725                 pw = getpwnam(name);
1726                 if (pw == NULL)
1727                         return (NULL);
1728                 strdupx(ap->val.s, pw->pw_dir, APERM);
1729                 ap->flag |= DEFINED|ISSET|ALLOC;
1730         }
1731         return (ap->val.s);
1732 }
1733 #endif
1734
1735 static void
1736 alt_expand(XPtrV *wp, char *start, char *exp_start, char *end, int fdo)
1737 {
1738         int count = 0;
1739         char *brace_start, *brace_end, *comma = NULL;
1740         char *field_start;
1741         char *p;
1742
1743         /* search for open brace */
1744         for (p = exp_start; (p = strchr(p, MAGIC)) && p[1] != '{' /*}*/; p += 2)
1745                 ;
1746         brace_start = p;
1747
1748         /* find matching close brace, if any */
1749         if (p) {
1750                 comma = NULL;
1751                 count = 1;
1752                 for (p += 2; *p && count; p++) {
1753                         if (ISMAGIC(*p)) {
1754                                 if (*++p == '{' /*}*/)
1755                                         count++;
1756                                 else if (*p == /*{*/ '}')
1757                                         --count;
1758                                 else if (*p == ',' && count == 1)
1759                                         comma = p;
1760                         }
1761                 }
1762         }
1763         /* no valid expansions... */
1764         if (!p || count != 0) {
1765                 /*
1766                  * Note that given a{{b,c} we do not expand anything (this is
1767                  * what AT&T ksh does. This may be changed to do the {b,c}
1768                  * expansion. }
1769                  */
1770                 if (fdo & DOGLOB)
1771                         glob(start, wp, tobool(fdo & DOMARKDIRS));
1772                 else
1773                         XPput(*wp, debunk(start, start, end - start));
1774                 return;
1775         }
1776         brace_end = p;
1777         if (!comma) {
1778                 alt_expand(wp, start, brace_end, end, fdo);
1779                 return;
1780         }
1781
1782         /* expand expression */
1783         field_start = brace_start + 2;
1784         count = 1;
1785         for (p = brace_start + 2; p != brace_end; p++) {
1786                 if (ISMAGIC(*p)) {
1787                         if (*++p == '{' /*}*/)
1788                                 count++;
1789                         else if ((*p == /*{*/ '}' && --count == 0) ||
1790                             (*p == ',' && count == 1)) {
1791                                 char *news;
1792                                 int l1, l2, l3;
1793
1794                                 /*
1795                                  * addition safe since these operate on
1796                                  * one string (separate substrings)
1797                                  */
1798                                 l1 = brace_start - start;
1799                                 l2 = (p - 1) - field_start;
1800                                 l3 = end - brace_end;
1801                                 news = alloc(l1 + l2 + l3 + 1, ATEMP);
1802                                 memcpy(news, start, l1);
1803                                 memcpy(news + l1, field_start, l2);
1804                                 memcpy(news + l1 + l2, brace_end, l3);
1805                                 news[l1 + l2 + l3] = '\0';
1806                                 alt_expand(wp, news, news + l1,
1807                                     news + l1 + l2 + l3, fdo);
1808                                 field_start = p + 1;
1809                         }
1810                 }
1811         }
1812         return;
1813 }
1814
1815 /* helper function due to setjmp/longjmp woes */
1816 static char *
1817 valsub(struct op *t, Area *ap)
1818 {
1819         char * volatile cp = NULL;
1820         struct tbl * volatile vp = NULL;
1821
1822         newenv(E_FUNC);
1823         newblock();
1824         if (ap)
1825                 vp = local("REPLY", false);
1826         if (!kshsetjmp(e->jbuf))
1827                 execute(t, XXCOM | XERROK, NULL);
1828         if (vp)
1829                 strdupx(cp, str_val(vp), ap);
1830         quitenv(NULL);
1831
1832         return (cp);
1833 }