1 /* $NetBSD: expand.c,v 1.147 2025/05/07 14:01:01 kre Exp $ */ 2 3 /*- 4 * Copyright (c) 1991, 1993 5 * The Regents of the University of California. All rights reserved. 6 * 7 * This code is derived from software contributed to Berkeley by 8 * Kenneth Almquist. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 3. 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 #include <sys/cdefs.h> 36 #ifndef lint 37 #if 0 38 static char sccsid[] = "@(#)expand.c 8.5 (Berkeley) 5/15/95"; 39 #else 40 __RCSID("$NetBSD: expand.c,v 1.147 2025/05/07 14:01:01 kre Exp $"); 41 #endif 42 #endif /* not lint */ 43 44 #include <sys/types.h> 45 #include <sys/time.h> 46 #include <sys/stat.h> 47 #include <errno.h> 48 #include <dirent.h> 49 #include <unistd.h> 50 #include <pwd.h> 51 #include <limits.h> 52 #include <stdlib.h> 53 #include <stdio.h> 54 #include <wctype.h> 55 #include <wchar.h> 56 57 /* 58 * Routines to expand arguments to commands. We have to deal with 59 * backquotes, shell variables, and file metacharacters. 60 */ 61 62 #include "shell.h" 63 #include "main.h" 64 #include "nodes.h" 65 #include "eval.h" 66 #include "expand.h" 67 #include "syntax.h" 68 #include "arithmetic.h" 69 #include "parser.h" 70 #include "jobs.h" 71 #include "options.h" 72 #include "builtins.h" 73 #include "var.h" 74 #include "input.h" 75 #include "output.h" 76 #include "memalloc.h" 77 #include "error.h" 78 #include "mystring.h" 79 #include "show.h" 80 81 /* 82 * Structure specifying which parts of the string should be searched 83 * for IFS characters. 84 */ 85 86 struct ifsregion { 87 struct ifsregion *next; /* next region in list */ 88 int begoff; /* offset of start of region */ 89 int endoff; /* offset of end of region */ 90 int inquotes; /* search for nul bytes only */ 91 }; 92 93 94 char *expdest; /* output of current string */ 95 struct nodelist *argbackq; /* list of back quote expressions */ 96 struct ifsregion ifsfirst; /* first struct in list of ifs regions */ 97 struct ifsregion *ifslastp; /* last struct in list */ 98 struct arglist exparg; /* holds expanded arg list */ 99 100 static int empty_dollar_at; /* have expanded "$@" to nothing */ 101 102 STATIC const char *argstr(const char *, int); 103 STATIC const char *exptilde(const char *, int); 104 STATIC void expbackq(union node *, int, int); 105 STATIC const char *expari(const char *); 106 STATIC int subevalvar(const char *, const char *, int, int, int); 107 STATIC int subevalvar_trim(const char *, int, int, int, int, int); 108 STATIC const char *evalvar(const char *, int); 109 STATIC int varisset(const char *, int); 110 STATIC void varvalue(const char *, int, int, int); 111 STATIC void recordregion(int, int, int); 112 STATIC void removerecordregions(int); 113 STATIC void ifsbreakup(char *, struct arglist *); 114 STATIC void ifsfree(void); 115 STATIC void expandmeta(struct strlist *, int); 116 STATIC void expmeta(char *, char *); 117 STATIC void addfname(char *); 118 STATIC struct strlist *expsort(struct strlist *); 119 STATIC struct strlist *msort(struct strlist *, int); 120 STATIC int patmatch(const char *, const char *, int); 121 STATIC char *cvtnum(int, char *); 122 static int collate_range_cmp(wchar_t, wchar_t); 123 STATIC void add_args(struct strlist *); 124 STATIC void rmescapes_nl(char *); 125 126 #ifdef DEBUG 127 #define NULLTERM_4_TRACE(p) STACKSTRNUL(p) 128 #else 129 #define NULLTERM_4_TRACE(p) do { /* nothing */ } while (0) 130 #endif 131 132 #define IS_BORING(_ch) \ 133 ((_ch) == CTLQUOTEMARK || (_ch) == CTLQUOTEEND || (_ch) == CTLNONL) 134 #define SKIP_BORING(p) \ 135 do { \ 136 char _ch; \ 137 \ 138 while ((_ch = *(p)), IS_BORING(_ch)) \ 139 (p)++; \ 140 } while (0) 141 142 /* 143 * Expand shell variables and backquotes inside a here document. 144 */ 145 146 char * 147 expandhere(union node *arg) 148 { 149 int len; 150 151 VTRACE(DBG_EXPAND|DBG_REDIR, ("expandhere(%p)\n", arg)); 152 expandarg(arg, NULL, 0); 153 len = rmescapes(stackblock()); 154 VTRACE(DBG_EXPAND|DBG_REDIR, ("expandhere() -> %d\n", len)); 155 return stalloc(len + 1); /* include the \0 */ 156 } 157 158 159 static int 160 collate_range_cmp(wchar_t c1, wchar_t c2) 161 { 162 wchar_t s1[2], s2[2]; 163 164 s1[0] = c1; 165 s1[1] = L'\0'; 166 s2[0] = c2; 167 s2[1] = L'\0'; 168 return (wcscoll(s1, s2)); 169 } 170 171 /* 172 * Perform variable substitution and command substitution on an argument, 173 * placing the resulting list of arguments in arglist. If EXP_FULL is true, 174 * perform splitting and file name expansion. When arglist is NULL, perform 175 * here document expansion. 176 */ 177 178 void 179 expandarg(union node *arg, struct arglist *arglist, int flag) 180 { 181 struct strlist *sp; 182 char *p; 183 184 CTRACE(DBG_EXPAND, ("expandarg(fl=%#x)\n", flag)); 185 if (fflag) /* no filename expandsion */ 186 flag &= ~EXP_GLOB; 187 188 empty_dollar_at = 0; 189 argbackq = arg->narg.backquote; 190 STARTSTACKSTR(expdest); 191 ifsfirst.next = NULL; 192 ifslastp = NULL; 193 line_number = arg->narg.lineno; 194 argstr(arg->narg.text, flag); 195 if (arglist == NULL) { 196 STACKSTRNUL(expdest); 197 CTRACE(DBG_EXPAND, 198 ("expandarg: no arglist, done[%d] (len %d) \"%s\"\n", 199 back_exitstatus, expdest - stackblock(), stackblock())); 200 return; /* here document expanded */ 201 } 202 STPUTC('\0', expdest); 203 CTRACE(DBG_EXPAND, ("expandarg: arglist got (%d) \"%s\"\n", 204 expdest - stackblock() - 1, stackblock())); 205 p = grabstackstr(expdest); 206 exparg.lastp = &exparg.list; 207 /* 208 * TODO - EXP_REDIR 209 */ 210 if (flag & EXP_SPLIT) { 211 ifsbreakup(p, &exparg); 212 *exparg.lastp = NULL; 213 exparg.lastp = &exparg.list; 214 if (flag & EXP_GLOB) 215 expandmeta(exparg.list, flag); 216 else 217 add_args(exparg.list); 218 #if 0 219 } else if (flag & EXP_REDIR) { 220 /* if EXP_REDIR ever happens, it happens here */ 221 /* for now just (below) remove escapes, and leave it alone */ 222 #endif 223 } else { 224 rmescapes(p); /* we might have escaped CTL bytes to remove */ 225 sp = stalloc(sizeof(*sp)); 226 sp->text = p; 227 *exparg.lastp = sp; 228 exparg.lastp = &sp->next; 229 } 230 ifsfree(); 231 *exparg.lastp = NULL; 232 if (exparg.list) { 233 *arglist->lastp = exparg.list; 234 arglist->lastp = exparg.lastp; 235 } 236 } 237 238 239 240 /* 241 * Perform variable and command substitution. 242 * If EXP_GLOB is set, output CTLESC characters to allow for further processing. 243 * If EXP_SPLIT is set, remember location of result for later, 244 * Otherwise treat $@ like $* since no splitting will be performed. 245 */ 246 247 STATIC const char * 248 argstr(const char *p, int flag) 249 { 250 char c; 251 const int quotes = flag & EXP_QNEEDED; /* do CTLESC */ 252 int firsteq = 1; 253 int had_dol_at = 0; 254 int startoff; 255 const char *ifs = NULL; 256 int ifs_split = EXP_IFS_SPLIT; 257 258 if (flag & EXP_IFS_SPLIT) 259 ifs = ifsval(); 260 261 CTRACE(DBG_EXPAND, ("argstr(\"%s\", %#x) quotes=%#x\n", p,flag,quotes)); 262 263 startoff = expdest - stackblock(); 264 if (*p == '~' && (flag & (EXP_TILDE | EXP_VARTILDE))) 265 p = exptilde(p, flag); 266 for (;;) { 267 switch (c = *p++) { 268 case '\0': 269 NULLTERM_4_TRACE(expdest); 270 VTRACE(DBG_EXPAND, ("argstr returning at \"\" " 271 "added \"%s\" to expdest\n", stackblock())); 272 return p - 1; 273 case CTLENDVAR: /* end of expanding yyy in ${xxx-yyy} */ 274 case CTLENDARI: /* end of a $(( )) string */ 275 if (had_dol_at && *p == CTLQUOTEEND) 276 p++; 277 NULLTERM_4_TRACE(expdest); 278 VTRACE(DBG_EXPAND, ("argstr returning at \"%.6s\"..." 279 " after %2.2X; added \"%s\" to expdest\n", 280 p, (c & 0xff), stackblock())); 281 return p; 282 case CTLQUOTEMARK: 283 /* "$@" syntax adherence hack */ 284 if (p[0] == CTLVAR && p[1] & VSQUOTE && 285 p[2] == '@' && p[3] == '=') { 286 had_dol_at = 1; 287 break; 288 } 289 had_dol_at = 0; 290 empty_dollar_at = 0; 291 if ((flag & EXP_SPLIT) != 0) 292 STPUTC(c, expdest); 293 ifs_split = 0; 294 break; 295 case CTLNONL: 296 if (flag & EXP_NL) 297 STPUTC(c, expdest); 298 line_number++; 299 break; 300 case CTLCNL: 301 STPUTC('\n', expdest); /* no line_number++ */ 302 break; 303 case CTLQUOTEEND: 304 if (empty_dollar_at && 305 expdest - stackblock() > startoff && 306 expdest[-1] == CTLQUOTEMARK) 307 expdest--; 308 else if (!had_dol_at && (flag & EXP_SPLIT) != 0) 309 STPUTC(c, expdest); 310 ifs_split = EXP_IFS_SPLIT; 311 had_dol_at = 0; 312 break; 313 case CTLESC: 314 if (quotes || ISCTL(*p)) 315 STPUTC(c, expdest); 316 c = *p++; 317 STPUTC(c, expdest); 318 if (c == '\n') /* should not happen, but ... */ 319 line_number++; 320 break; 321 case CTLVAR: { 322 #ifdef DEBUG 323 unsigned int pos = expdest - stackblock(); 324 NULLTERM_4_TRACE(expdest); 325 #endif 326 p = evalvar(p, (flag & ~EXP_IFS_SPLIT) | (flag & ifs_split)); 327 NULLTERM_4_TRACE(expdest); 328 VTRACE(DBG_EXPAND, ("argstr evalvar " 329 "added %zd \"%s\" to expdest\n", 330 (size_t)(expdest - (stackblock() + pos)), 331 stackblock() + pos)); 332 break; 333 } 334 case CTLBACKQ: 335 case CTLBACKQ|CTLQUOTE: { 336 #ifdef DEBUG 337 unsigned int pos = expdest - stackblock(); 338 #endif 339 expbackq(argbackq->n, c & CTLQUOTE, flag); 340 argbackq = argbackq->next; 341 NULLTERM_4_TRACE(expdest); 342 VTRACE(DBG_EXPAND, ("argstr expbackq added \"%s\" " 343 "to expdest\n", stackblock() + pos)); 344 break; 345 } 346 case CTLARI: { 347 #ifdef DEBUG 348 unsigned int pos = expdest - stackblock(); 349 #endif 350 p = expari(p); 351 NULLTERM_4_TRACE(expdest); 352 VTRACE(DBG_EXPAND, ("argstr expari " 353 "+ \"%s\" to expdest p=\"%.5s...\"\n", 354 stackblock() + pos, p)); 355 break; 356 } 357 case ':': 358 case '=': 359 /* 360 * sort of a hack - expand tildes in variable 361 * assignments (after the first '=' and after ':'s). 362 */ 363 STPUTC(c, expdest); 364 if (flag & EXP_VARTILDE && *p == '~') { 365 if (c == '=') { 366 if (firsteq) 367 firsteq = 0; 368 else 369 break; 370 } 371 p = exptilde(p, flag); 372 } 373 break; 374 default: 375 if (c == '\n') 376 line_number++; 377 STPUTC(c, expdest); 378 if (flag & ifs_split && strchr(ifs, c) != NULL) { 379 /* We need to get the output split here... */ 380 recordregion(expdest - stackblock() - 1, 381 expdest - stackblock(), 0); 382 } 383 break; 384 } 385 } 386 } 387 388 STATIC const char * 389 exptilde(const char *p, int flag) 390 { 391 char c, last; 392 const char *startp = p; 393 struct passwd *pw; 394 const char *home; 395 const int quotes = flag & EXP_QNEEDED; 396 char *user; 397 struct stackmark smark; 398 #ifdef DEBUG 399 unsigned int offs = expdest - stackblock(); 400 #endif 401 402 setstackmark(&smark); 403 (void) grabstackstr(expdest); 404 user = stackblock(); /* we will just borrow top of stack */ 405 406 while ((c = *++p) != '\0') { 407 switch(c) { 408 case CTLESC: /* any of these occurring */ 409 case CTLVAR: /* means ~ expansion */ 410 case CTLBACKQ: /* does not happen at all */ 411 case CTLBACKQ | CTLQUOTE: 412 case CTLARI: /* just leave original unchanged */ 413 case CTLENDARI: 414 case CTLQUOTEMARK: 415 case '\n': 416 popstackmark(&smark); 417 return (startp); 418 case CTLNONL: 419 continue; 420 case ':': 421 if (!posix || flag & EXP_VARTILDE) 422 goto done; 423 break; 424 case CTLENDVAR: 425 case '/': 426 goto done; 427 } 428 STPUTC(c, user); 429 } 430 done: 431 STACKSTRNUL(user); 432 user = stackblock(); /* to start of collected username */ 433 434 CTRACE(DBG_EXPAND, ("exptilde, found \"~%s\"", user)); 435 if (*user == '\0') { 436 home = lookupvar("HOME"); 437 /* 438 * if HOME is unset, results are unspecified... 439 * we used to just leave the ~ unchanged, but 440 * (some) other shells do ... and this seems more useful. 441 */ 442 if (home == NULL && (pw = getpwuid(getuid())) != NULL) 443 home = pw->pw_dir; 444 } else if ((pw = getpwnam(user)) == NULL) { 445 /* 446 * If user does not exist, results are undefined. 447 * so we can abort() here if we want, but let's not! 448 */ 449 home = NULL; 450 } else 451 home = pw->pw_dir; 452 453 VTRACE(DBG_EXPAND, (" ->\"%s\"", home ? home : "<<NULL>>")); 454 popstackmark(&smark); /* now expdest is valid again */ 455 456 /* 457 * Posix XCU 2.6.1: The value of $HOME (for ~) or the initial 458 * working directory from getpwnam() for ~user 459 * Nothing there about "except if a null string". So do what it wants. 460 * In later drafts (to become Issue 8), it is even required that in 461 * this case, (where HOME='') a bare ~ expands to "" (which must not 462 * be reduced to nothing). 463 */ 464 last = '\0'; /* just in case *home == '\0' (already) */ 465 if (home == NULL) { 466 CTRACE(DBG_EXPAND, (": returning unused \"%s\"\n", startp)); 467 return startp; 468 } 469 while ((c = *home++) != '\0') { 470 if ((quotes && NEEDESC(c)) || ISCTL(c)) 471 STPUTC(CTLESC, expdest); 472 STPUTC(c, expdest); 473 last = c; 474 } 475 476 /* 477 * If HOME (or whatver) ended in a '/' (last == '/'), and 478 * the ~prefix was terminated by a '/', then only keep one 479 * of them - since we already took the one from HOME, just 480 * skip over the one that ended the tilde prefix. 481 * 482 * Current (Issue 8) drafts say this is permitted, and recommend 483 * it - a later version of the standard will probably require it. 484 * This is to prevent ~/foo generating //foo when HOME=/ (and 485 * other cases like it, but that's the important one). 486 */ 487 if (last == '/' && *p == '/') 488 p++; 489 490 CTRACE(DBG_EXPAND, (": added %d \"%.*s\" returning \"%s\"\n", 491 expdest - stackblock() - offs, expdest - stackblock() - offs, 492 stackblock() + offs, p)); 493 494 return (p); 495 } 496 497 498 STATIC void 499 removerecordregions(int endoff) 500 { 501 502 VTRACE(DBG_EXPAND, ("removerecordregions(%d):", endoff)); 503 if (ifslastp == NULL) { 504 VTRACE(DBG_EXPAND, (" none\n", endoff)); 505 return; 506 } 507 508 if (ifsfirst.endoff > endoff) { 509 VTRACE(DBG_EXPAND, (" first(%d)", ifsfirst.endoff)); 510 while (ifsfirst.next != NULL) { 511 struct ifsregion *ifsp; 512 INTOFF; 513 ifsp = ifsfirst.next->next; 514 ckfree(ifsfirst.next); 515 ifsfirst.next = ifsp; 516 INTON; 517 } 518 if (ifsfirst.begoff > endoff) 519 ifslastp = NULL; 520 else { 521 VTRACE(DBG_EXPAND,("->(%d,%d)",ifsfirst.begoff,endoff)); 522 ifslastp = &ifsfirst; 523 ifsfirst.endoff = endoff; 524 } 525 VTRACE(DBG_EXPAND, ("\n")); 526 return; 527 } 528 529 ifslastp = &ifsfirst; 530 while (ifslastp->next && ifslastp->next->begoff < endoff) 531 ifslastp=ifslastp->next; 532 VTRACE(DBG_EXPAND, (" found(%d,%d)", ifslastp->begoff,ifslastp->endoff)); 533 while (ifslastp->next != NULL) { 534 struct ifsregion *ifsp; 535 INTOFF; 536 ifsp = ifslastp->next->next; 537 ckfree(ifslastp->next); 538 ifslastp->next = ifsp; 539 INTON; 540 } 541 if (ifslastp->endoff > endoff) 542 ifslastp->endoff = endoff; 543 VTRACE(DBG_EXPAND, ("->(%d,%d)", ifslastp->begoff,ifslastp->endoff)); 544 } 545 546 547 /* 548 * Expand arithmetic expression. 549 * 550 * In this incarnation, we start at the beginning (yes, "Let's start at the 551 * very beginning. A very good place to start.") and collect the expression 552 * until the end - which means expanding anything contained within. 553 * 554 * Fortunately, argstr() just happens to do that for us... 555 */ 556 STATIC const char * 557 expari(const char *p) 558 { 559 char *q, *start; 560 intmax_t result; 561 int adjustment; 562 int begoff; 563 int quoted; 564 struct stackmark smark; 565 566 /* ifsfree(); */ 567 568 /* 569 * SPACE_NEEDED is enough for all possible digits (rounded up) 570 * plus possible "-", and the terminating '\0', hence, plus 2 571 * 572 * The calculation produces the number of bytes needed to 573 * represent the biggest possible value, in octal. We only 574 * generate decimal, which takes (often) less digits (never more) 575 * so this is safe, if occasionally slightly wasteful. 576 */ 577 #define SPACE_NEEDED ((int)((sizeof(intmax_t) * CHAR_BIT + 2) / 3 + 2)) 578 579 quoted = *p++ == '"'; 580 begoff = expdest - stackblock(); 581 VTRACE(DBG_EXPAND, ("expari%s: \"%s\" begoff %d\n", 582 quoted ? "(quoted)" : "", p, begoff)); 583 584 p = argstr(p, EXP_NL); /* expand $(( )) string */ 585 STPUTC('\0', expdest); 586 start = stackblock() + begoff; 587 588 removerecordregions(begoff); /* nothing there is kept */ 589 rmescapes_nl(start); /* convert CRTNONL back into \n's */ 590 591 setstackmark(&smark); 592 q = grabstackstr(expdest); /* keep the expression while eval'ing */ 593 result = arith(start, line_number); 594 popstackmark(&smark); /* return the stack to before grab */ 595 596 start = stackblock() + begoff; /* block may have moved */ 597 adjustment = expdest - start; 598 STADJUST(-adjustment, expdest); /* remove the argstr() result */ 599 600 CHECKSTRSPACE(SPACE_NEEDED, expdest); /* nb: stack block might move */ 601 fmtstr(expdest, SPACE_NEEDED, "%"PRIdMAX, result); 602 603 for (q = expdest; *q++ != '\0'; ) /* find end of what we added */ 604 ; 605 606 if (quoted == 0) /* allow weird splitting */ 607 recordregion(begoff, begoff + q - 1 - expdest, 0); 608 adjustment = q - expdest - 1; 609 STADJUST(adjustment, expdest); /* move expdest to end */ 610 VTRACE(DBG_EXPAND, ("expari: adding %d \"%s\", returning \"%.5s...\"\n", 611 adjustment, stackblock() + begoff, p)); 612 613 return p; 614 } 615 616 617 /* 618 * Expand stuff in backwards quotes (these days, any command substitution). 619 */ 620 621 STATIC void 622 expbackq(union node *cmd, int quoted, int flag) 623 { 624 struct backcmd in; 625 int i; 626 char buf[128]; 627 char *p; 628 char *dest = expdest; /* expdest may be reused by eval, use an alt */ 629 struct ifsregion saveifs, *savelastp; 630 struct nodelist *saveargbackq; 631 char lastc; 632 int startloc = dest - stackblock(); 633 const int quotes = flag & EXP_QNEEDED; 634 int nnl; 635 struct stackmark smark; 636 637 VTRACE(DBG_EXPAND, ("expbackq( ..., q=%d flag=%#x) have %d\n", 638 quoted, flag, startloc)); 639 INTOFF; 640 saveifs = ifsfirst; 641 savelastp = ifslastp; 642 saveargbackq = argbackq; 643 644 setstackmark(&smark); /* preserve the stack */ 645 p = grabstackstr(dest); /* save what we have there currently */ 646 evalbackcmd(cmd, &in); /* evaluate the $( ) tree (using stack) */ 647 popstackmark(&smark); /* and return stack to when we entered */ 648 649 ifsfirst = saveifs; 650 ifslastp = savelastp; 651 argbackq = saveargbackq; 652 653 p = in.buf; /* now extract the results */ 654 nnl = 0; /* dropping trailing \n's */ 655 for (;;) { 656 if (--in.nleft < 0) { 657 if (in.fd < 0) 658 break; 659 INTON; 660 while ((i = read(in.fd, buf, sizeof buf)) < 0 && errno == EINTR) 661 continue; 662 INTOFF; 663 VTRACE(DBG_EXPAND, ("expbackq: read returns %d\n", i)); 664 if (i <= 0) 665 break; 666 p = buf; 667 in.nleft = i - 1; 668 } 669 lastc = *p++; 670 if (lastc != '\0') { 671 if (lastc == '\n') /* don't save \n yet */ 672 nnl++; /* it might be trailing */ 673 else { 674 /* 675 * We have something other than \n 676 * 677 * Before saving it, we need to insert 678 * any \n's that we have just skipped. 679 */ 680 681 /* XXX 682 * this hack is just because our 683 * CHECKSTRSPACE() is lazy, and only 684 * ever grows the stack once, even 685 * if that does not allocate the space 686 * we requested. ie: safe for small 687 * requests, but not large ones. 688 * FIXME someday... 689 */ 690 if (nnl < 20) { 691 CHECKSTRSPACE(nnl + 2, dest); 692 while (nnl > 0) { 693 nnl--; 694 USTPUTC('\n', dest); 695 } 696 } else { 697 /* The slower, safer, way */ 698 while (nnl > 0) { 699 nnl--; 700 STPUTC('\n', dest); 701 } 702 CHECKSTRSPACE(2, dest); 703 } 704 if ((quotes && quoted && NEEDESC(lastc)) || 705 ISCTL(lastc)) 706 USTPUTC(CTLESC, dest); 707 USTPUTC(lastc, dest); 708 } 709 } 710 } 711 712 if (in.fd >= 0) 713 close(in.fd); 714 if (in.buf) 715 ckfree(in.buf); 716 if (in.jp) 717 back_exitstatus = waitforjob(in.jp); 718 if (quoted == 0) 719 recordregion(startloc, dest - stackblock(), 0); 720 CTRACE(DBG_EXPAND, ("evalbackq: [%d] size=%d: \"%.*s\"\n", 721 back_exitstatus, 722 (int)((dest - stackblock()) - startloc), 723 (int)((dest - stackblock()) - startloc), 724 stackblock() + startloc)); 725 726 expdest = dest; /* all done, expdest is all ours again */ 727 INTON; 728 } 729 730 731 STATIC int 732 subevalvar(const char *p, const char *str, int subtype, int startloc, 733 int varflags) 734 { 735 char *startp; 736 struct nodelist *saveargbackq = argbackq; 737 int amount; 738 739 VTRACE(DBG_EXPAND, ("subevalvar(%d) \"%.20s\" ${%.*s} sloc=%d vf=%x\n", 740 subtype, p, p-str, str, startloc, varflags)); 741 argstr(p, subtype == VSASSIGN ? EXP_VARTILDE : EXP_TILDE); 742 STACKSTRNUL(expdest); 743 argbackq = saveargbackq; 744 startp = stackblock() + startloc; 745 746 switch (subtype) { 747 case VSASSIGN: 748 setvar(str, startp, 0); 749 amount = startp - expdest; /* remove what argstr added */ 750 STADJUST(amount, expdest); 751 varflags &= ~VSNUL; /*XXX Huh? What's that achieve? */ 752 return 1; /* go back and eval var again */ 753 754 case VSQUESTION: 755 if (*p != CTLENDVAR) { 756 outfmt(&errout, "%s\n", startp); 757 error(NULL); 758 } 759 error("%.*s: parameter %snot set", 760 (int)(p - str - 1), 761 str, (varflags & VSNUL) ? "null or " 762 : nullstr); 763 /* NOTREACHED */ 764 765 default: 766 abort(); 767 } 768 } 769 770 STATIC int 771 subevalvar_trim(const char *p, int strloc, int subtype, int startloc, 772 int varflags, int quotes) 773 { 774 char *startp; 775 char *str; 776 char *loc = NULL; 777 char *q; 778 int c = 0; 779 struct nodelist *saveargbackq = argbackq; 780 int amount; 781 782 switch (subtype) { 783 case VSTRIMLEFT: 784 case VSTRIMLEFTMAX: 785 case VSTRIMRIGHT: 786 case VSTRIMRIGHTMAX: 787 break; 788 default: 789 abort(); 790 break; 791 } 792 793 VTRACE(DBG_EXPAND, 794 ("subevalvar_trim(\"%.9s\", STR@%d, SUBT=%d, start@%d, vf=%x, q=%x)\n", 795 p, strloc, subtype, startloc, varflags, quotes)); 796 797 argstr(p, (varflags & (VSQUOTE|VSPATQ)) == VSQUOTE ? 0 : EXP_CASE); 798 STACKSTRNUL(expdest); 799 argbackq = saveargbackq; 800 startp = stackblock() + startloc; 801 str = stackblock() + strloc; 802 803 switch (subtype) { 804 805 case VSTRIMLEFT: 806 for (loc = startp; loc < str; loc++) { 807 c = *loc; 808 *loc = '\0'; 809 if (patmatch(str, startp, quotes)) 810 goto recordleft; 811 *loc = c; 812 if (quotes && *loc == CTLESC) 813 loc++; 814 } 815 return 0; 816 817 case VSTRIMLEFTMAX: 818 for (loc = str - 1; loc >= startp;) { 819 c = *loc; 820 *loc = '\0'; 821 if (patmatch(str, startp, quotes)) 822 goto recordleft; 823 *loc = c; 824 loc--; 825 if (quotes && loc > startp && 826 *(loc - 1) == CTLESC) { 827 for (q = startp; q < loc; q++) 828 if (*q == CTLESC) 829 q++; 830 if (q > loc) 831 loc--; 832 } 833 } 834 return 0; 835 836 case VSTRIMRIGHT: 837 for (loc = str - 1; loc >= startp;) { 838 if (patmatch(str, loc, quotes)) 839 goto recordright; 840 loc--; 841 if (quotes && loc > startp && 842 *(loc - 1) == CTLESC) { 843 for (q = startp; q < loc; q++) 844 if (*q == CTLESC) 845 q++; 846 if (q > loc) 847 loc--; 848 } 849 } 850 return 0; 851 852 case VSTRIMRIGHTMAX: 853 for (loc = startp; loc < str - 1; loc++) { 854 if (patmatch(str, loc, quotes)) 855 goto recordright; 856 if (quotes && *loc == CTLESC) 857 loc++; 858 } 859 return 0; 860 861 default: 862 abort(); 863 } 864 865 recordleft: 866 *loc = c; 867 amount = ((str - 1) - (loc - startp)) - expdest; 868 STADJUST(amount, expdest); 869 while (loc != str - 1) 870 *startp++ = *loc++; 871 return 1; 872 873 recordright: 874 amount = loc - expdest; 875 STADJUST(amount, expdest); 876 STPUTC('\0', expdest); 877 STADJUST(-1, expdest); 878 return 1; 879 } 880 881 882 /* 883 * Expand a variable, and return a pointer to the next character in the 884 * input string. 885 */ 886 887 STATIC const char * 888 evalvar(const char *p, int flag) 889 { 890 int subtype; 891 int varflags; 892 const char *var; 893 char *val; 894 int patloc; 895 int c; 896 int set; 897 int special; 898 int startloc; 899 int varlen; 900 int apply_ifs; 901 const int quotes = flag & EXP_QNEEDED; 902 903 varflags = (unsigned char)*p++; 904 subtype = varflags & VSTYPE; 905 var = p; 906 special = subtype != VSUNKNOWN && !is_name(*p); 907 p = strchr(p, '=') + 1; 908 909 CTRACE(DBG_EXPAND, 910 ("evalvar \"%.*s\", flag=%#X quotes=%#X vf=%#X subtype=%X\n", 911 (int)(p - var - 1), var, flag, quotes, varflags, subtype)); 912 913 again: /* jump here after setting a variable with ${var=text} */ 914 if (varflags & VSLINENO) { 915 if (line_num.flags & VUNSET) { 916 set = 0; 917 val = NULL; 918 } else { 919 set = 1; 920 special = p - var; 921 val = NULL; 922 } 923 } else if (special) { 924 set = varisset(var, varflags & VSNUL); 925 val = NULL; 926 if (!set && *var == '@') 927 empty_dollar_at = 1; 928 } else { 929 val = lookupvar(var); 930 if (val == NULL || ((varflags & VSNUL) && val[0] == '\0')) { 931 val = NULL; 932 set = 0; 933 } else 934 set = 1; 935 } 936 937 varlen = 0; 938 startloc = expdest - stackblock(); 939 940 if (!set && uflag && *var != '@' && *var != '*') { 941 switch (subtype) { 942 case VSNORMAL: 943 case VSTRIMLEFT: 944 case VSTRIMLEFTMAX: 945 case VSTRIMRIGHT: 946 case VSTRIMRIGHTMAX: 947 case VSLENGTH: 948 error("%.*s: parameter not set", 949 (int)(p - var - 1), var); 950 /* NOTREACHED */ 951 } 952 } 953 954 #if 0 /* no longer need this $@ evil ... */ 955 if (!set && subtype != VSPLUS && special && *var == '@') 956 if (startloc > 0 && expdest[-1] == CTLQUOTEMARK) 957 expdest--, startloc--; 958 #endif 959 960 if (set && subtype != VSPLUS) { 961 /* insert the value of the variable */ 962 if (special) { 963 if (varflags & VSLINENO) { 964 /* 965 * The LINENO hack (expansion part) 966 */ 967 while (--special > 0) { 968 /* not needed, it is a number... 969 if (quotes && NEEDESC(*var)) 970 STPUTC(CTLESC, expdest); 971 */ 972 STPUTC(*var++, expdest); 973 } 974 } else 975 varvalue(var, varflags&VSQUOTE, subtype, flag); 976 if (subtype == VSLENGTH) { 977 varlen = expdest - stackblock() - startloc; 978 STADJUST(-varlen, expdest); 979 } 980 } else { 981 982 if (subtype == VSLENGTH) { 983 for (; *val; val++) 984 varlen++; 985 } else if (quotes && varflags & VSQUOTE) { 986 /* 987 * If we are going to look for magic in the 988 * value (quotes is set) and the expansion 989 * occurs inside "" (VSQUOTE) then any char 990 * that has any potential special meaning 991 * needs to have that meaning suppressed, 992 * so supply a CTLESC prefix for it. 993 */ 994 for (; (c = *val) != '\0'; val++) { 995 if (NEEDESC(c)) 996 STPUTC(CTLESC, expdest); 997 STPUTC(c, expdest); 998 } 999 } else { 1000 /* 1001 * We are going to rmescapes() later, 1002 * so make sure that any data char that 1003 * might be mistaken for one of our CTLxxx 1004 * magic chars is protected ... always. 1005 */ 1006 for (; (c = *val) != '\0'; val++) { 1007 if (ISCTL(c)) 1008 STPUTC(CTLESC, expdest); 1009 STPUTC(c, expdest); 1010 } 1011 } 1012 } 1013 } 1014 1015 1016 if (varflags & VSQUOTE) { 1017 if (*var == '@' && shellparam.nparam != 1) 1018 apply_ifs = 1; 1019 else { 1020 /* 1021 * Mark so that we don't apply IFS if we recurse through 1022 * here expanding $bar from "${foo-$bar}". 1023 */ 1024 flag |= EXP_IN_QUOTES; 1025 apply_ifs = 0; 1026 } 1027 } else if (flag & EXP_IN_QUOTES) { 1028 apply_ifs = 0; 1029 } else 1030 apply_ifs = 1; 1031 1032 switch (subtype) { 1033 case VSLENGTH: 1034 expdest = cvtnum(varlen, expdest); 1035 break; 1036 1037 case VSNORMAL: 1038 break; 1039 1040 case VSPLUS: 1041 set = !set; 1042 /* FALLTHROUGH */ 1043 case VSMINUS: 1044 if (!set) { 1045 argstr(p, flag | (apply_ifs ? EXP_IFS_SPLIT : 0)); 1046 /* 1047 * ${x-a b c} doesn't get split, but removing the 1048 * 'apply_ifs = 0' apparently breaks ${1+"$@"}.. 1049 * ${x-'a b' c} should generate 2 args. 1050 */ 1051 if (*p != CTLENDVAR) 1052 /* We should have marked stuff already */ 1053 apply_ifs = 0; 1054 } 1055 break; 1056 1057 case VSTRIMLEFT: 1058 case VSTRIMLEFTMAX: 1059 case VSTRIMRIGHT: 1060 case VSTRIMRIGHTMAX: 1061 if (!set) { 1062 set = 1; /* allow argbackq to be advanced if needed */ 1063 break; 1064 } 1065 /* 1066 * Terminate the string and start recording the pattern 1067 * right after it 1068 */ 1069 STPUTC('\0', expdest); 1070 patloc = expdest - stackblock(); 1071 if (subevalvar_trim(p, patloc, subtype, startloc, varflags, 1072 quotes) == 0) { 1073 int amount = (expdest - stackblock() - patloc) + 1; 1074 STADJUST(-amount, expdest); 1075 } 1076 /* Remove any recorded regions beyond start of variable */ 1077 removerecordregions(startloc); 1078 apply_ifs = 1; 1079 break; 1080 1081 case VSASSIGN: 1082 case VSQUESTION: 1083 if (set) 1084 break; 1085 if (subevalvar(p, var, subtype, startloc, varflags)) { 1086 /* if subevalvar() returns, it always returns 1 */ 1087 1088 varflags &= ~VSNUL; 1089 /* 1090 * Remove any recorded regions beyond 1091 * start of variable 1092 */ 1093 removerecordregions(startloc); 1094 goto again; 1095 } 1096 apply_ifs = 0; /* never executed */ 1097 break; 1098 1099 case VSUNKNOWN: 1100 VTRACE(DBG_EXPAND, 1101 ("evalvar \"%.*s\", unknown [%p %p] \"%.3s\" (%#2x %#2x)\n", 1102 (int)(p - var - 1), var, var, p, p, p[0] & 0xFF, p[1] & 0xFF)); 1103 1104 if ((p - var) <= 1) 1105 error("%d: unknown expansion type", line_number); 1106 else { 1107 if (*p == '#') /* only VSUNKNOWN as a ${#var:...} */ 1108 error("%d: ${#%.*s%c..}: unknown modifier", 1109 line_number, (int)(p - var - 1), 1110 var, p[1]&0xFF); 1111 1112 if (*p == CTLESC) 1113 p++; 1114 error("%d: ${%.*s%c..}: unknown modifier", 1115 line_number, (int)(p - var - 1), var, (*p & 0xFF)); 1116 } 1117 /* NOTREACHED */ 1118 1119 default: 1120 abort(); 1121 } 1122 1123 if (apply_ifs) 1124 recordregion(startloc, expdest - stackblock(), 1125 varflags & VSQUOTE); 1126 1127 if (subtype != VSNORMAL) { /* skip to end of alternative */ 1128 int nesting = 1; 1129 for (;;) { 1130 if ((c = *p++) == CTLESC) 1131 p++; 1132 else if (c == CTLNONL) 1133 ; 1134 else if (c == CTLBACKQ || c == (CTLBACKQ|CTLQUOTE)) { 1135 if (set) 1136 argbackq = argbackq->next; 1137 } else if (c == CTLVAR) { 1138 if ((*p++ & VSTYPE) != VSNORMAL) 1139 nesting++; 1140 } else if (c == CTLENDVAR) { 1141 if (--nesting == 0) 1142 break; 1143 } 1144 } 1145 } 1146 return p; 1147 } 1148 1149 1150 1151 /* 1152 * Test whether a special parameter is set. 1153 */ 1154 1155 STATIC int 1156 varisset(const char *name, int nulok) 1157 { 1158 if (*name == '!') 1159 return backgndpid != -1; 1160 else if (*name == '@' || *name == '*') { 1161 if (*shellparam.p == NULL) 1162 return 0; 1163 1164 if (nulok) { 1165 char **av; 1166 1167 for (av = shellparam.p; *av; av++) 1168 if (**av != '\0') 1169 return 1; 1170 return 0; 1171 } 1172 } else if (is_digit(*name)) { 1173 char *ap; 1174 long num; 1175 1176 /* 1177 * handle overflow sensibly (the *ap tests should never fail) 1178 */ 1179 errno = 0; 1180 num = strtol(name, &ap, 10); 1181 if (errno != 0 || (*ap != '\0' && *ap != '=')) 1182 return 0; 1183 1184 if (num == 0) 1185 ap = arg0; 1186 else if (num > shellparam.nparam) 1187 return 0; 1188 else 1189 ap = shellparam.p[num - 1]; 1190 1191 if (nulok && (ap == NULL || *ap == '\0')) 1192 return 0; 1193 } 1194 return 1; 1195 } 1196 1197 1198 1199 /* 1200 * Add the value of a specialized variable to the stack string. 1201 */ 1202 1203 STATIC void 1204 varvalue(const char *name, int quoted, int subtype, int flag) 1205 { 1206 int num; 1207 char *p; 1208 int i; 1209 int sep; 1210 char **ap; 1211 #ifdef DEBUG 1212 char *start = expdest; 1213 #endif 1214 1215 VTRACE(DBG_EXPAND, ("varvalue(%c%s, sub=%d, fl=%#x)", *name, 1216 quoted ? ", quoted" : "", subtype, flag)); 1217 1218 if (subtype == VSLENGTH) /* no magic required ... */ 1219 flag &= ~(EXP_FULL | EXP_QNEEDED); 1220 1221 #define STRTODEST(p) \ 1222 do {\ 1223 if ((flag & EXP_QNEEDED) && quoted) { \ 1224 while (*p) { \ 1225 if (NEEDESC(*p)) \ 1226 STPUTC(CTLESC, expdest); \ 1227 STPUTC(*p++, expdest); \ 1228 } \ 1229 } else \ 1230 while (*p) { \ 1231 if ((flag & EXP_QNEEDED) && ISCTL(*p)) \ 1232 STPUTC(CTLESC, expdest); \ 1233 STPUTC(*p++, expdest); \ 1234 } \ 1235 } while (0) 1236 1237 1238 switch (*name) { 1239 case '$': 1240 num = rootpid; 1241 break; 1242 case '?': 1243 num = exitstatus; 1244 break; 1245 case '#': 1246 num = shellparam.nparam; 1247 break; 1248 case '!': 1249 num = backgndpid; 1250 break; 1251 case '-': 1252 for (i = 0; i < option_flags; i++) { 1253 if (optlist[optorder[i]].val) 1254 STPUTC(optlist[optorder[i]].letter, expdest); 1255 } 1256 VTRACE(DBG_EXPAND, (": %.*s\n", expdest-start, start)); 1257 return; 1258 case '@': 1259 if (flag & EXP_SPLIT && quoted) { 1260 VTRACE(DBG_EXPAND, (": $@ split (%d)\n", 1261 shellparam.nparam)); 1262 #if 0 1263 /* GROSS HACK */ 1264 if (shellparam.nparam == 0 && 1265 expdest[-1] == CTLQUOTEMARK) 1266 expdest--; 1267 /* KCAH SSORG */ 1268 #endif 1269 if (shellparam.nparam == 0) 1270 empty_dollar_at = 1; 1271 1272 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 1273 if (*p == '\0') { 1274 /* retain an explicit null string */ 1275 STPUTC(CTLQUOTEMARK, expdest); 1276 STPUTC(CTLQUOTEEND, expdest); 1277 } else 1278 STRTODEST(p); 1279 if (*ap) 1280 /* A NUL separates args inside "" */ 1281 STPUTC('\0', expdest); 1282 } 1283 return; 1284 } 1285 /* fall through */ 1286 case '*': 1287 sep = ifsval()[0]; 1288 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 1289 STRTODEST(p); 1290 if (!*ap) 1291 break; 1292 if (sep) { 1293 if (quoted && (flag & EXP_QNEEDED) && 1294 NEEDESC(sep)) 1295 STPUTC(CTLESC, expdest); 1296 STPUTC(sep, expdest); 1297 } else 1298 if ((flag & (EXP_SPLIT|EXP_IN_QUOTES)) == EXP_SPLIT 1299 && !quoted && **ap != '\0') 1300 STPUTC('\0', expdest); 1301 } 1302 VTRACE(DBG_EXPAND, (": %.*s\n", expdest-start, start)); 1303 return; 1304 default: 1305 if (is_digit(*name)) { 1306 long lnum; 1307 1308 errno = 0; 1309 lnum = strtol(name, &p, 10); 1310 if (errno != 0 || (*p != '\0' && *p != '=')) 1311 return; 1312 1313 if (lnum == 0) 1314 p = arg0; 1315 else if (lnum > 0 && lnum <= shellparam.nparam) 1316 p = shellparam.p[lnum - 1]; 1317 else 1318 return; 1319 STRTODEST(p); 1320 } 1321 VTRACE(DBG_EXPAND, (": %.*s\n", expdest-start, start)); 1322 return; 1323 } 1324 /* 1325 * only the specials with an int value arrive here 1326 */ 1327 VTRACE(DBG_EXPAND, ("(%d)", num)); 1328 expdest = cvtnum(num, expdest); 1329 VTRACE(DBG_EXPAND, (": %.*s\n", expdest-start, start)); 1330 } 1331 1332 1333 1334 /* 1335 * Record the fact that we have to scan this region of the 1336 * string for IFS characters. 1337 */ 1338 1339 STATIC void 1340 recordregion(int start, int end, int inquotes) 1341 { 1342 struct ifsregion *ifsp; 1343 1344 VTRACE(DBG_EXPAND, ("recordregion(%d,%d,%d)\n", start, end, inquotes)); 1345 if (ifslastp == NULL) { 1346 ifsp = &ifsfirst; 1347 } else { 1348 if (ifslastp->endoff == start 1349 && ifslastp->inquotes == inquotes) { 1350 /* extend previous area */ 1351 ifslastp->endoff = end; 1352 return; 1353 } 1354 ifsp = (struct ifsregion *)ckmalloc(sizeof (struct ifsregion)); 1355 ifslastp->next = ifsp; 1356 } 1357 ifslastp = ifsp; 1358 ifslastp->next = NULL; 1359 ifslastp->begoff = start; 1360 ifslastp->endoff = end; 1361 ifslastp->inquotes = inquotes; 1362 } 1363 1364 1365 1366 /* 1367 * Break the argument string into pieces based upon IFS and add the 1368 * strings to the argument list. The regions of the string to be 1369 * searched for IFS characters have been stored by recordregion. 1370 */ 1371 STATIC void 1372 ifsbreakup(char *string, struct arglist *arglist) 1373 { 1374 struct ifsregion *ifsp; 1375 struct strlist *sp; 1376 char *start; 1377 char *p; 1378 char *q; 1379 const char *ifs; 1380 const char *ifsspc; 1381 int had_param_ch = 0; 1382 1383 start = string; 1384 1385 VTRACE(DBG_EXPAND, ("ifsbreakup(\"%s\")", string)); /* misses \0's */ 1386 if (ifslastp == NULL) { 1387 /* Return entire argument, IFS doesn't apply to any of it */ 1388 VTRACE(DBG_EXPAND, ("no regions\n", string)); 1389 sp = stalloc(sizeof(*sp)); 1390 sp->text = start; 1391 *arglist->lastp = sp; 1392 arglist->lastp = &sp->next; 1393 return; 1394 } 1395 1396 ifs = ifsval(); 1397 1398 for (ifsp = &ifsfirst; ifsp != NULL; ifsp = ifsp->next) { 1399 p = string + ifsp->begoff; 1400 VTRACE(DBG_EXPAND, (" !%.*s!(%d)", ifsp->endoff-ifsp->begoff, 1401 p, ifsp->endoff-ifsp->begoff)); 1402 while (p < string + ifsp->endoff) { 1403 had_param_ch = 1; 1404 q = p; 1405 if (IS_BORING(*p)) { 1406 p++; 1407 continue; 1408 } 1409 if (*p == CTLESC) 1410 p++; 1411 if (ifsp->inquotes) { 1412 /* Only NULs (should be from "$@") end args */ 1413 if (*p != 0) { 1414 p++; 1415 continue; 1416 } 1417 ifsspc = NULL; 1418 VTRACE(DBG_EXPAND, (" \\0 nxt:\"%s\" ", p)); 1419 } else { 1420 if (!strchr(ifs, *p)) { 1421 p++; 1422 continue; 1423 } 1424 had_param_ch = 0; 1425 ifsspc = strchr(" \t\n", *p); 1426 1427 /* Ignore IFS whitespace at start */ 1428 if (q == start && ifsspc != NULL) { 1429 p++; 1430 start = p; 1431 continue; 1432 } 1433 } 1434 1435 /* Save this argument... */ 1436 *q = '\0'; 1437 VTRACE(DBG_EXPAND, ("<%s>", start)); 1438 sp = stalloc(sizeof(*sp)); 1439 sp->text = start; 1440 *arglist->lastp = sp; 1441 arglist->lastp = &sp->next; 1442 p++; 1443 1444 if (ifsspc != NULL) { 1445 /* Ignore further trailing IFS whitespace */ 1446 for (; p < string + ifsp->endoff; p++) { 1447 q = p; 1448 if (*p == CTLNONL) 1449 continue; 1450 if (*p == CTLESC) 1451 p++; 1452 if (strchr(ifs, *p) == NULL) { 1453 p = q; 1454 break; 1455 } 1456 if (strchr(" \t\n", *p) == NULL) { 1457 p++; 1458 break; 1459 } 1460 } 1461 } 1462 start = p; 1463 } 1464 } 1465 1466 /* 1467 while (*start == CTLQUOTEEND) 1468 start++; 1469 */ 1470 1471 /* 1472 * Save anything left as an argument. 1473 * Traditionally we have treated 'IFS=':'; set -- x$IFS' as 1474 * generating 2 arguments, the second of which is empty. 1475 * Some recent clarification of the Posix spec say that it 1476 * should only generate one.... 1477 */ 1478 if (had_param_ch || *start != 0) { 1479 VTRACE(DBG_EXPAND, (" T<%s>", start)); 1480 sp = stalloc(sizeof(*sp)); 1481 sp->text = start; 1482 *arglist->lastp = sp; 1483 arglist->lastp = &sp->next; 1484 } 1485 VTRACE(DBG_EXPAND, ("\n")); 1486 } 1487 1488 STATIC void 1489 ifsfree(void) 1490 { 1491 while (ifsfirst.next != NULL) { 1492 struct ifsregion *ifsp; 1493 INTOFF; 1494 ifsp = ifsfirst.next->next; 1495 ckfree(ifsfirst.next); 1496 ifsfirst.next = ifsp; 1497 INTON; 1498 } 1499 ifslastp = NULL; 1500 ifsfirst.next = NULL; 1501 } 1502 1503 1504 1505 /* 1506 * Expand shell metacharacters. At this point, the only control characters 1507 * should be escapes. The results are stored in the list exparg. 1508 */ 1509 1510 char *expdir; 1511 1512 1513 STATIC void 1514 expandmeta(struct strlist *str, int flag) 1515 { 1516 char *p; 1517 struct strlist **savelastp; 1518 struct strlist *sp; 1519 char c; 1520 /* TODO - EXP_REDIR */ 1521 1522 while (str) { 1523 p = str->text; 1524 for (;;) { /* fast check for meta chars */ 1525 if ((c = *p++) == '\0') 1526 goto nometa; 1527 if (c == '*' || c == '?' || c == '[' /* || c == '!' */) 1528 break; 1529 } 1530 savelastp = exparg.lastp; 1531 INTOFF; 1532 if (expdir == NULL) { 1533 int i = strlen(str->text); 1534 expdir = ckmalloc(i < 2048 ? 2048 : i); /* XXX */ 1535 } 1536 1537 expmeta(expdir, str->text); 1538 ckfree(expdir); 1539 expdir = NULL; 1540 INTON; 1541 if (exparg.lastp == savelastp) { 1542 /* 1543 * no matches 1544 */ 1545 nometa: 1546 *exparg.lastp = str; 1547 rmescapes(str->text); 1548 exparg.lastp = &str->next; 1549 } else { 1550 *exparg.lastp = NULL; 1551 *savelastp = sp = expsort(*savelastp); 1552 while (sp->next != NULL) 1553 sp = sp->next; 1554 exparg.lastp = &sp->next; 1555 } 1556 str = str->next; 1557 } 1558 } 1559 1560 STATIC void 1561 add_args(struct strlist *str) 1562 { 1563 while (str) { 1564 *exparg.lastp = str; 1565 rmescapes(str->text); 1566 exparg.lastp = &str->next; 1567 str = str->next; 1568 } 1569 } 1570 1571 1572 /* 1573 * Do metacharacter (i.e. *, ?, [...]) expansion. 1574 */ 1575 1576 STATIC void 1577 expmeta(char *enddir, char *name) 1578 { 1579 char *p; 1580 const char *cp; 1581 char *q; 1582 char *start; 1583 char *endname; 1584 int metaflag; 1585 struct stat statb; 1586 DIR *dirp; 1587 struct dirent *dp; 1588 int atend; 1589 int matchdot; 1590 1591 CTRACE(DBG_EXPAND|DBG_MATCH, ("expmeta(\"%s\")\n", name)); 1592 metaflag = 0; 1593 start = name; 1594 for (p = name ; ; p++) { 1595 if (*p == '*' || *p == '?') 1596 metaflag = 1; 1597 else if (*p == '[') { 1598 q = p + 1; 1599 if (*q == '!' || *q == '^') 1600 q++; 1601 for (;;) { 1602 while (IS_BORING(*q)) 1603 q++; 1604 if (*q == ']') { 1605 q++; 1606 metaflag = 1; 1607 break; 1608 } 1609 if (*q == '[' && q[1] == ':') { 1610 /* 1611 * character class, look for :] ending 1612 * also stop on ']' (end bracket expr) 1613 * or '\0' or '/' (end pattern) 1614 */ 1615 while (*++q != '\0' && *q != ']' && 1616 *q != '/') { 1617 if (*q == CTLESC) { 1618 if (*++q == '\0') 1619 break; 1620 if (*q == '/') 1621 break; 1622 } else if (*q == ':' && 1623 q[1] == ']') 1624 break; 1625 } 1626 if (*q == ':') { 1627 /* 1628 * stopped at ':]' 1629 * still in [...] 1630 * skip ":]" and continue; 1631 */ 1632 q += 2; 1633 continue; 1634 } 1635 1636 /* done at end of pattern, not [...] */ 1637 if (*q == '\0' || *q == '/') 1638 break; 1639 1640 /* found the ']', we have a [...] */ 1641 metaflag = 1; 1642 q++; /* skip ']' */ 1643 break; 1644 } 1645 if (*q == CTLESC) 1646 q++; 1647 /* end of pattern cannot be escaped */ 1648 if (*q == '/' || *q == '\0') 1649 break; 1650 q++; 1651 } 1652 } else if (*p == '\0') 1653 break; 1654 else if (IS_BORING(*p)) 1655 continue; 1656 else if (*p == CTLESC) 1657 p++; 1658 if (*p == '/') { 1659 if (metaflag) 1660 break; 1661 start = p + 1; 1662 } 1663 } 1664 if (metaflag == 0) { /* we've reached the end of the file name */ 1665 if (enddir != expdir) 1666 metaflag++; 1667 for (p = name ; ; p++) { 1668 if (IS_BORING(*p)) 1669 continue; 1670 if (*p == CTLESC) 1671 p++; 1672 *enddir++ = *p; 1673 if (*p == '\0') 1674 break; 1675 } 1676 if (metaflag == 0 || lstat(expdir, &statb) >= 0) 1677 addfname(expdir); 1678 return; 1679 } 1680 endname = p; 1681 if (start != name) { 1682 p = name; 1683 while (p < start) { 1684 while (IS_BORING(*p)) 1685 p++; 1686 if (*p == CTLESC) 1687 p++; 1688 *enddir++ = *p++; 1689 } 1690 } 1691 if (enddir == expdir) { 1692 cp = "."; 1693 } else if (enddir == expdir + 1 && *expdir == '/') { 1694 cp = "/"; 1695 } else { 1696 cp = expdir; 1697 enddir[-1] = '\0'; 1698 } 1699 if ((dirp = opendir(cp)) == NULL) 1700 return; 1701 if (enddir != expdir) 1702 enddir[-1] = '/'; 1703 if (*endname == 0) { 1704 atend = 1; 1705 } else { 1706 atend = 0; 1707 *endname++ = '\0'; 1708 } 1709 matchdot = 0; 1710 p = start; 1711 while (IS_BORING(*p)) 1712 p++; 1713 if (*p == CTLESC) 1714 p++; 1715 if (*p == '.') 1716 matchdot++; 1717 while (! int_pending() && (dp = readdir(dirp)) != NULL) { 1718 if (dp->d_name[0] == '.' && ! matchdot) 1719 continue; 1720 if (patmatch(start, dp->d_name, 0)) { 1721 if (atend) { 1722 scopy(dp->d_name, enddir); 1723 addfname(expdir); 1724 } else { 1725 for (p = enddir, cp = dp->d_name; 1726 (*p++ = *cp++) != '\0';) 1727 continue; 1728 p[-1] = '/'; 1729 expmeta(p, endname); 1730 } 1731 } 1732 } 1733 closedir(dirp); 1734 if (! atend) 1735 endname[-1] = '/'; 1736 } 1737 1738 1739 /* 1740 * Add a file name to the list. 1741 */ 1742 1743 STATIC void 1744 addfname(char *name) 1745 { 1746 char *p; 1747 struct strlist *sp; 1748 1749 p = stalloc(strlen(name) + 1); 1750 scopy(name, p); 1751 sp = stalloc(sizeof(*sp)); 1752 sp->text = p; 1753 *exparg.lastp = sp; 1754 exparg.lastp = &sp->next; 1755 } 1756 1757 1758 /* 1759 * Sort the results of file name expansion. It calculates the number of 1760 * strings to sort and then calls msort (short for merge sort) to do the 1761 * work. 1762 */ 1763 1764 STATIC struct strlist * 1765 expsort(struct strlist *str) 1766 { 1767 int len; 1768 struct strlist *sp; 1769 1770 len = 0; 1771 for (sp = str ; sp ; sp = sp->next) 1772 len++; 1773 return msort(str, len); 1774 } 1775 1776 1777 STATIC struct strlist * 1778 msort(struct strlist *list, int len) 1779 { 1780 struct strlist *p, *q = NULL; 1781 struct strlist **lpp; 1782 int half; 1783 int n; 1784 1785 if (len <= 1) 1786 return list; 1787 half = len >> 1; 1788 p = list; 1789 for (n = half ; --n >= 0 ; ) { 1790 q = p; 1791 p = p->next; 1792 } 1793 q->next = NULL; /* terminate first half of list */ 1794 q = msort(list, half); /* sort first half of list */ 1795 p = msort(p, len - half); /* sort second half */ 1796 lpp = &list; 1797 for (;;) { 1798 if (strcmp(p->text, q->text) < 0) { 1799 *lpp = p; 1800 lpp = &p->next; 1801 if ((p = *lpp) == NULL) { 1802 *lpp = q; 1803 break; 1804 } 1805 } else { 1806 *lpp = q; 1807 lpp = &q->next; 1808 if ((q = *lpp) == NULL) { 1809 *lpp = p; 1810 break; 1811 } 1812 } 1813 } 1814 return list; 1815 } 1816 1817 1818 /* 1819 * See if a character matches a character class, starting at the first colon 1820 * of "[:class:]". 1821 * If a valid character class is recognized, a pointer to the next character 1822 * after the final closing bracket is stored into *end, otherwise a null 1823 * pointer is stored into *end. 1824 */ 1825 static int 1826 match_charclass(const char *p, wchar_t chr, const char **end) 1827 { 1828 char name[20]; 1829 const char *nameend; 1830 wctype_t cclass; 1831 char *q; 1832 1833 *end = NULL; 1834 p++; 1835 q = &name[0]; 1836 nameend = strstr(p, ":]"); 1837 if (nameend == NULL || nameend == p) /* not a valid class */ 1838 return 0; 1839 1840 if (*p == CTLESC) { 1841 if (*++p == CTLESC) 1842 return 0; 1843 if (p == nameend) 1844 return 0; 1845 } 1846 if (!is_alpha(*p)) 1847 return 0; 1848 while (p < nameend) { 1849 if (*p == CTLESC) { 1850 p++; 1851 if (p == nameend) 1852 return 0; 1853 } 1854 if (!is_in_name(*p)) /* '_' is a local extension */ 1855 return 0; 1856 if (q < &name[sizeof name]) 1857 *q++ = *p++; 1858 else 1859 p++; 1860 } 1861 1862 *end = nameend + 2; /* committed to it being a char class */ 1863 1864 if (q < &name[sizeof name]) /* a usable name found */ 1865 *q++ = '\0'; 1866 else /* too long, valid, but no match */ 1867 return 0; 1868 1869 cclass = wctype(name); 1870 /* An unknown class matches nothing but is valid nevertheless. */ 1871 if (cclass == 0) 1872 return 0; 1873 return iswctype(chr, cclass); 1874 } 1875 1876 1877 /* 1878 * Returns true if the pattern matches the string. 1879 */ 1880 1881 STATIC int 1882 patmatch(const char *pattern, const char *string, int squoted) 1883 { 1884 const char *p, *q, *end; 1885 const char *bt_p, *bt_q; 1886 char c; 1887 wchar_t wc, wc2; 1888 1889 VTRACE(DBG_MATCH, ("patmatch(P=\"%s\", W=\"%s\"%s): ", 1890 pattern, string, squoted ? ", SQ" : "")); 1891 p = pattern; 1892 q = string; 1893 bt_p = NULL; 1894 bt_q = NULL; 1895 for (;;) { 1896 switch (c = *p++) { 1897 case '\0': 1898 if (squoted && *q == CTLESC) { 1899 if (q[1] == '\0') 1900 q++; 1901 } 1902 if (*q != '\0') 1903 goto backtrack; 1904 VTRACE(DBG_MATCH, ("match\n")); 1905 return 1; 1906 case CTLESC: 1907 if (squoted && *q == CTLESC) 1908 q++; 1909 if (*p == '\0' && *q == '\0') { 1910 VTRACE(DBG_MATCH, ("match-\\\n")); 1911 return 1; 1912 } 1913 if (*q++ != *p++) 1914 goto backtrack; 1915 break; 1916 case '\\': 1917 if (squoted && *q == CTLESC) 1918 q++; 1919 if (*q++ != *p++) 1920 goto backtrack; 1921 break; 1922 case CTLQUOTEMARK: 1923 case CTLQUOTEEND: 1924 case CTLNONL: 1925 continue; 1926 case '?': 1927 if (squoted && *q == CTLESC) 1928 q++; 1929 if (*q++ == '\0') { 1930 VTRACE(DBG_MATCH, ("?fail\n")); 1931 return 0; 1932 } 1933 break; 1934 case '*': 1935 c = *p; 1936 while (c == CTLQUOTEMARK || c == '*') 1937 c = *++p; 1938 if (c != CTLESC && !IS_BORING(c) && 1939 c != '?' && c != '*' && c != '[') { 1940 while (*q != c) { 1941 if (squoted && *q == CTLESC && 1942 q[1] == c) 1943 break; 1944 if (*q == '\0') { 1945 VTRACE(DBG_MATCH, ("*fail\n")); 1946 return 0; 1947 } 1948 if (squoted && *q == CTLESC) 1949 q++; 1950 q++; 1951 } 1952 } 1953 if (c == CTLESC && p[1] == '\0') { 1954 VTRACE(DBG_MATCH, ("match+\\\n")); 1955 return 1; 1956 } 1957 /* 1958 * First try the shortest match for the '*' that 1959 * could work. We can forget any earlier '*' since 1960 * there is no way having it match more characters 1961 * can help us, given that we are already here. 1962 */ 1963 bt_p = p; 1964 bt_q = q; 1965 break; 1966 case '[': { 1967 const char *savep, *saveq, *endp; 1968 int invert, found; 1969 unsigned char chr; 1970 1971 /* 1972 * First quick check to see if there is a 1973 * possible matching ']' - if not, then this 1974 * is not a char class, and the '[' is just 1975 * a literal '['. 1976 * 1977 * This check will not detect all non classes, but 1978 * that's OK - It just means that we execute the 1979 * harder code sometimes when it it cannot succeed. 1980 */ 1981 endp = p; 1982 if (*endp == '!' || *endp == '^') 1983 endp++; 1984 for (;;) { 1985 while (IS_BORING(*endp)) 1986 endp++; 1987 if (*endp == '\0') 1988 goto dft; /* no matching ] */ 1989 if (*endp++ == ']') 1990 break; 1991 } 1992 /* end shortcut */ 1993 1994 savep = p, saveq = q; 1995 invert = 0; 1996 if (*p == '!' || *p == '^') { 1997 invert++; 1998 p++; 1999 } 2000 found = 0; 2001 if (*q == '\0') { 2002 VTRACE(DBG_MATCH, ("[]fail\n")); 2003 return 0; 2004 } 2005 if (squoted && *q == CTLESC) 2006 q++; 2007 chr = (unsigned char)*q++; 2008 c = *p++; 2009 do { 2010 if (IS_BORING(c)) 2011 continue; 2012 if (c == '\0') { 2013 p = savep, q = saveq; 2014 c = '['; 2015 goto dft; 2016 } 2017 if (c == '[' && *p == ':') { 2018 found |= match_charclass(p, chr, &end); 2019 if (end != NULL) { 2020 p = end; 2021 continue; 2022 } 2023 } 2024 if (c == CTLESC || c == '\\') 2025 c = *p++; 2026 wc = (unsigned char)c; 2027 if (*p == '-' && p[1] != ']') { 2028 p++; 2029 if (*p == CTLESC || *p == '\\') 2030 p++; 2031 wc2 = (unsigned char)*p++; 2032 if ( collate_range_cmp(chr, wc) >= 0 2033 && collate_range_cmp(chr, wc2) <= 0 2034 ) 2035 found = 1; 2036 } else { 2037 if (chr == wc) 2038 found = 1; 2039 } 2040 } while ((c = *p++) != ']'); 2041 if (found == invert) 2042 goto backtrack; 2043 break; 2044 } 2045 dft: default: 2046 if (squoted && *q == CTLESC) 2047 q++; 2048 if (*q++ == c) 2049 break; 2050 backtrack: 2051 /* 2052 * If we have a mismatch (other than hitting the end 2053 * of the string), go back to the last '*' seen and 2054 * have it match one additional character. 2055 */ 2056 if (bt_p == NULL) { 2057 VTRACE(DBG_MATCH, ("BTP fail\n")); 2058 return 0; 2059 } 2060 if (*bt_q == '\0') { 2061 VTRACE(DBG_MATCH, ("BTQ fail\n")); 2062 return 0; 2063 } 2064 bt_q++; 2065 p = bt_p; 2066 q = bt_q; 2067 break; 2068 } 2069 } 2070 } 2071 2072 2073 2074 /* 2075 * Remove any CTLESC or CTLNONL characters from a string. 2076 * 2077 * String is modified in place, and we return the length of the result 2078 */ 2079 2080 int 2081 rmescapes(char *str) 2082 { 2083 char *p, *q; 2084 2085 p = str; 2086 while (!ISCTL(*p)) { 2087 if (*p++ == '\0') 2088 return ((int)(p - str) - 1); 2089 } 2090 q = p; 2091 while (*p) { 2092 if (IS_BORING(*p)) { 2093 p++; 2094 continue; 2095 } 2096 if (*p == CTLCNL) { 2097 p++; 2098 *q++ = '\n'; 2099 continue; 2100 } 2101 if (*p == CTLESC) 2102 p++; 2103 #ifdef DEBUG 2104 else if (ISCTL(*p)) 2105 abort(); 2106 #endif 2107 *q++ = *p++; 2108 } 2109 *q = '\0'; 2110 2111 return ((int)(q - str)); 2112 } 2113 2114 /* 2115 * and a special version for dealing with expressions to be parsed 2116 * by the arithmetic evaluator. That needs to be able to count \n's 2117 * even ones that were \newline elided \n's, so we have to put the 2118 * latter back into the string - just being careful to put them only 2119 * at a place where white space can reasonably occur in the string 2120 * -- then the \n we insert will just be white space, and ignored 2121 * for all purposes except line counting. 2122 */ 2123 2124 void 2125 rmescapes_nl(char *str) 2126 { 2127 char *p, *q; 2128 int nls = 0, holdnl = 0, holdlast; 2129 2130 p = str; 2131 while (!ISCTL(*p)) { 2132 if (*p++ == '\0') 2133 return; 2134 } 2135 if (p > str) /* must reprocess char before stopper (if any) */ 2136 --p; /* so we do not place a \n badly */ 2137 q = p; 2138 while (*p) { 2139 if (*p == CTLQUOTEMARK || *p == CTLQUOTEEND) { 2140 p++; 2141 continue; 2142 } 2143 if (*p == CTLNONL) { 2144 p++; 2145 nls++; 2146 continue; 2147 } 2148 if (*p == CTLCNL) { 2149 p++; 2150 *q++ = '\n'; 2151 continue; 2152 } 2153 if (*p == CTLESC) 2154 p++; 2155 #ifdef DEBUG 2156 else if (ISCTL(*p)) 2157 abort(); 2158 #endif 2159 2160 holdlast = holdnl; 2161 holdnl = is_in_name(*p); /* letters, digits, _ */ 2162 if (q == str || is_space(q[-1]) || (*p != '=' && q[-1] != *p)) { 2163 if (nls > 0 && holdnl != holdlast) { 2164 while (nls > 0) 2165 *q++ = '\n', nls--; 2166 } 2167 } 2168 *q++ = *p++; 2169 } 2170 while (--nls >= 0) 2171 *q++ = '\n'; 2172 *q = '\0'; 2173 } 2174 2175 2176 2177 /* 2178 * See if a pattern matches in a case statement. 2179 */ 2180 2181 int 2182 casematch(union node *pattern, char *val) 2183 { 2184 struct stackmark smark; 2185 int result; 2186 char *p; 2187 2188 CTRACE(DBG_MATCH, ("casematch(P=\"%s\", W=\"%s\")\n", 2189 pattern->narg.text, val)); 2190 setstackmark(&smark); 2191 argbackq = pattern->narg.backquote; 2192 STARTSTACKSTR(expdest); 2193 ifslastp = NULL; 2194 argstr(pattern->narg.text, EXP_TILDE | EXP_CASE); 2195 STPUTC('\0', expdest); 2196 p = grabstackstr(expdest); 2197 result = patmatch(p, val, 0); 2198 popstackmark(&smark); 2199 return result; 2200 } 2201 2202 /* 2203 * Our own itoa(). Assumes result buffer is on the stack 2204 */ 2205 2206 STATIC char * 2207 cvtnum(int num, char *buf) 2208 { 2209 char temp[32]; 2210 int neg = num < 0; 2211 char *p = temp + sizeof temp - 1; 2212 2213 if (neg) 2214 num = -num; 2215 2216 *p = '\0'; 2217 do { 2218 *--p = num % 10 + '0'; 2219 } while ((num /= 10) != 0 && p > temp + 1); 2220 2221 if (neg) 2222 *--p = '-'; 2223 2224 while (*p) 2225 STPUTC(*p++, buf); 2226 return buf; 2227 } 2228 2229 /* 2230 * Do most of the work for wordexp(3). 2231 */ 2232 2233 int 2234 wordexpcmd(int argc, char **argv) 2235 { 2236 size_t len; 2237 int i; 2238 2239 out1fmt("%d", argc - 1); 2240 out1c('\0'); 2241 for (i = 1, len = 0; i < argc; i++) 2242 len += strlen(argv[i]); 2243 out1fmt("%zu", len); 2244 out1c('\0'); 2245 for (i = 1; i < argc; i++) { 2246 out1str(argv[i]); 2247 out1c('\0'); 2248 } 2249 return (0); 2250 } 2251
Indexes created Sun Sep 21 16:09:51 GMT 2025