1 /* $NetBSD: expand.c,v 1.141 2021/11/22 05:17:43 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 1.141 2021/11/22 05:17:43 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; 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 */ 461 if (home == NULL /* || *home == '\0' */) { 462 CTRACE(DBG_EXPAND, (": returning unused \"%s\"\n", startp)); 463 return startp; 464 } while ((c = *home++) != '\0') { 465 if ((quotes && NEEDESC(c)) || ISCTL(c)) 466 STPUTC(CTLESC, expdest); 467 STPUTC(c, expdest); 468 } 469 CTRACE(DBG_EXPAND, (": added %d \"%.*s\" returning \"%s\"\n", 470 expdest - stackblock() - offs, expdest - stackblock() - offs, 471 stackblock() + offs, p)); 472 473 return (p); 474 } 475 476 477 STATIC void 478 removerecordregions(int endoff) 479 { 480 481 VTRACE(DBG_EXPAND, ("removerecordregions(%d):", endoff)); 482 if (ifslastp == NULL) { 483 VTRACE(DBG_EXPAND, (" none\n", endoff)); 484 return; 485 } 486 487 if (ifsfirst.endoff > endoff) { 488 VTRACE(DBG_EXPAND, (" first(%d)", ifsfirst.endoff)); 489 while (ifsfirst.next != NULL) { 490 struct ifsregion *ifsp; 491 INTOFF; 492 ifsp = ifsfirst.next->next; 493 ckfree(ifsfirst.next); 494 ifsfirst.next = ifsp; 495 INTON; 496 } 497 if (ifsfirst.begoff > endoff) 498 ifslastp = NULL; 499 else { 500 VTRACE(DBG_EXPAND,("->(%d,%d)",ifsfirst.begoff,endoff)); 501 ifslastp = &ifsfirst; 502 ifsfirst.endoff = endoff; 503 } 504 VTRACE(DBG_EXPAND, ("\n")); 505 return; 506 } 507 508 ifslastp = &ifsfirst; 509 while (ifslastp->next && ifslastp->next->begoff < endoff) 510 ifslastp=ifslastp->next; 511 VTRACE(DBG_EXPAND, (" found(%d,%d)", ifslastp->begoff,ifslastp->endoff)); 512 while (ifslastp->next != NULL) { 513 struct ifsregion *ifsp; 514 INTOFF; 515 ifsp = ifslastp->next->next; 516 ckfree(ifslastp->next); 517 ifslastp->next = ifsp; 518 INTON; 519 } 520 if (ifslastp->endoff > endoff) 521 ifslastp->endoff = endoff; 522 VTRACE(DBG_EXPAND, ("->(%d,%d)", ifslastp->begoff,ifslastp->endoff)); 523 } 524 525 526 /* 527 * Expand arithmetic expression. 528 * 529 * In this incarnation, we start at the beginning (yes, "Let's start at the 530 * very beginning. A very good place to start.") and collect the expression 531 * until the end - which means expanding anything contained within. 532 * 533 * Fortunately, argstr() just happens to do that for us... 534 */ 535 STATIC const char * 536 expari(const char *p) 537 { 538 char *q, *start; 539 intmax_t result; 540 int adjustment; 541 int begoff; 542 int quoted; 543 struct stackmark smark; 544 545 /* ifsfree(); */ 546 547 /* 548 * SPACE_NEEDED is enough for all possible digits (rounded up) 549 * plus possible "-", and the terminating '\0', hence, plus 2 550 * 551 * The calculation produces the number of bytes needed to 552 * represent the biggest possible value, in octal. We only 553 * generate decimal, which takes (often) less digits (never more) 554 * so this is safe, if occasionally slightly wasteful. 555 */ 556 #define SPACE_NEEDED ((int)((sizeof(intmax_t) * CHAR_BIT + 2) / 3 + 2)) 557 558 quoted = *p++ == '"'; 559 begoff = expdest - stackblock(); 560 VTRACE(DBG_EXPAND, ("expari%s: \"%s\" begoff %d\n", 561 quoted ? "(quoted)" : "", p, begoff)); 562 563 p = argstr(p, EXP_NL); /* expand $(( )) string */ 564 STPUTC('\0', expdest); 565 start = stackblock() + begoff; 566 567 removerecordregions(begoff); /* nothing there is kept */ 568 rmescapes_nl(start); /* convert CRTNONL back into \n's */ 569 570 setstackmark(&smark); 571 q = grabstackstr(expdest); /* keep the expression while eval'ing */ 572 result = arith(start, line_number); 573 popstackmark(&smark); /* return the stack to before grab */ 574 575 start = stackblock() + begoff; /* block may have moved */ 576 adjustment = expdest - start; 577 STADJUST(-adjustment, expdest); /* remove the argstr() result */ 578 579 CHECKSTRSPACE(SPACE_NEEDED, expdest); /* nb: stack block might move */ 580 fmtstr(expdest, SPACE_NEEDED, "%"PRIdMAX, result); 581 582 for (q = expdest; *q++ != '\0'; ) /* find end of what we added */ 583 ; 584 585 if (quoted == 0) /* allow weird splitting */ 586 recordregion(begoff, begoff + q - 1 - expdest, 0); 587 adjustment = q - expdest - 1; 588 STADJUST(adjustment, expdest); /* move expdest to end */ 589 VTRACE(DBG_EXPAND, ("expari: adding %d \"%s\", returning \"%.5s...\"\n", 590 adjustment, stackblock() + begoff, p)); 591 592 return p; 593 } 594 595 596 /* 597 * Expand stuff in backwards quotes (these days, any command substitution). 598 */ 599 600 STATIC void 601 expbackq(union node *cmd, int quoted, int flag) 602 { 603 struct backcmd in; 604 int i; 605 char buf[128]; 606 char *p; 607 char *dest = expdest; /* expdest may be reused by eval, use an alt */ 608 struct ifsregion saveifs, *savelastp; 609 struct nodelist *saveargbackq; 610 char lastc; 611 int startloc = dest - stackblock(); 612 int saveherefd; 613 const int quotes = flag & EXP_QNEEDED; 614 int nnl; 615 struct stackmark smark; 616 617 VTRACE(DBG_EXPAND, ("expbackq( ..., q=%d flag=%#x) have %d\n", 618 quoted, flag, startloc)); 619 INTOFF; 620 saveifs = ifsfirst; 621 savelastp = ifslastp; 622 saveargbackq = argbackq; 623 saveherefd = herefd; 624 herefd = -1; 625 626 setstackmark(&smark); /* preserve the stack */ 627 p = grabstackstr(dest); /* save what we have there currently */ 628 evalbackcmd(cmd, &in); /* evaluate the $( ) tree (using stack) */ 629 popstackmark(&smark); /* and return stack to when we entered */ 630 631 ifsfirst = saveifs; 632 ifslastp = savelastp; 633 argbackq = saveargbackq; 634 herefd = saveherefd; 635 636 p = in.buf; /* now extract the results */ 637 nnl = 0; /* dropping trailing \n's */ 638 for (;;) { 639 if (--in.nleft < 0) { 640 if (in.fd < 0) 641 break; 642 INTON; 643 while ((i = read(in.fd, buf, sizeof buf)) < 0 && errno == EINTR) 644 continue; 645 INTOFF; 646 VTRACE(DBG_EXPAND, ("expbackq: read returns %d\n", i)); 647 if (i <= 0) 648 break; 649 p = buf; 650 in.nleft = i - 1; 651 } 652 lastc = *p++; 653 if (lastc != '\0') { 654 if (lastc == '\n') /* don't save \n yet */ 655 nnl++; /* it might be trailing */ 656 else { 657 /* 658 * We have something other than \n 659 * 660 * Before saving it, we need to insert 661 * any \n's that we have just skipped. 662 */ 663 664 /* XXX 665 * this hack is just because our 666 * CHECKSTRSPACE() is lazy, and only 667 * ever grows the stack once, even 668 * if that does not allocate the space 669 * we requested. ie: safe for small 670 * requests, but not large ones. 671 * FIXME someday... 672 */ 673 if (nnl < 20) { 674 CHECKSTRSPACE(nnl + 2, dest); 675 while (nnl > 0) { 676 nnl--; 677 USTPUTC('\n', dest); 678 } 679 } else { 680 /* The slower, safer, way */ 681 while (nnl > 0) { 682 nnl--; 683 STPUTC('\n', dest); 684 } 685 CHECKSTRSPACE(2, dest); 686 } 687 if ((quotes && quoted && NEEDESC(lastc)) || 688 ISCTL(lastc)) 689 USTPUTC(CTLESC, dest); 690 USTPUTC(lastc, dest); 691 } 692 } 693 } 694 695 if (in.fd >= 0) 696 close(in.fd); 697 if (in.buf) 698 ckfree(in.buf); 699 if (in.jp) 700 back_exitstatus = waitforjob(in.jp); 701 if (quoted == 0) 702 recordregion(startloc, dest - stackblock(), 0); 703 CTRACE(DBG_EXPAND, ("evalbackq: [%d] size=%d: \"%.*s\"\n", 704 back_exitstatus, 705 (int)((dest - stackblock()) - startloc), 706 (int)((dest - stackblock()) - startloc), 707 stackblock() + startloc)); 708 709 expdest = dest; /* all done, expdest is all ours again */ 710 INTON; 711 } 712 713 714 STATIC int 715 subevalvar(const char *p, const char *str, int subtype, int startloc, 716 int varflags) 717 { 718 char *startp; 719 int saveherefd = herefd; 720 struct nodelist *saveargbackq = argbackq; 721 int amount; 722 723 herefd = -1; 724 VTRACE(DBG_EXPAND, ("subevalvar(%d) \"%.20s\" ${%.*s} sloc=%d vf=%x\n", 725 subtype, p, p-str, str, startloc, varflags)); 726 argstr(p, subtype == VSASSIGN ? EXP_VARTILDE : EXP_TILDE); 727 STACKSTRNUL(expdest); 728 herefd = saveherefd; 729 argbackq = saveargbackq; 730 startp = stackblock() + startloc; 731 732 switch (subtype) { 733 case VSASSIGN: 734 setvar(str, startp, 0); 735 amount = startp - expdest; /* remove what argstr added */ 736 STADJUST(amount, expdest); 737 varflags &= ~VSNUL; /*XXX Huh? What's that achieve? */ 738 return 1; /* go back and eval var again */ 739 740 case VSQUESTION: 741 if (*p != CTLENDVAR) { 742 outfmt(&errout, "%s\n", startp); 743 error(NULL); 744 } 745 error("%.*s: parameter %snot set", 746 (int)(p - str - 1), 747 str, (varflags & VSNUL) ? "null or " 748 : nullstr); 749 /* NOTREACHED */ 750 751 default: 752 abort(); 753 } 754 } 755 756 STATIC int 757 subevalvar_trim(const char *p, int strloc, int subtype, int startloc, 758 int varflags, int quotes) 759 { 760 char *startp; 761 char *str; 762 char *loc = NULL; 763 char *q; 764 int c = 0; 765 int saveherefd = herefd; 766 struct nodelist *saveargbackq = argbackq; 767 int amount; 768 769 herefd = -1; 770 switch (subtype) { 771 case VSTRIMLEFT: 772 case VSTRIMLEFTMAX: 773 case VSTRIMRIGHT: 774 case VSTRIMRIGHTMAX: 775 break; 776 default: 777 abort(); 778 break; 779 } 780 781 VTRACE(DBG_EXPAND, 782 ("subevalvar_trim(\"%.9s\", STR@%d, SUBT=%d, start@%d, vf=%x, q=%x)\n", 783 p, strloc, subtype, startloc, varflags, quotes)); 784 785 argstr(p, (varflags & (VSQUOTE|VSPATQ)) == VSQUOTE ? 0 : EXP_CASE); 786 STACKSTRNUL(expdest); 787 herefd = saveherefd; 788 argbackq = saveargbackq; 789 startp = stackblock() + startloc; 790 str = stackblock() + strloc; 791 792 switch (subtype) { 793 794 case VSTRIMLEFT: 795 for (loc = startp; loc < str; loc++) { 796 c = *loc; 797 *loc = '\0'; 798 if (patmatch(str, startp, quotes)) 799 goto recordleft; 800 *loc = c; 801 if (quotes && *loc == CTLESC) 802 loc++; 803 } 804 return 0; 805 806 case VSTRIMLEFTMAX: 807 for (loc = str - 1; loc >= startp;) { 808 c = *loc; 809 *loc = '\0'; 810 if (patmatch(str, startp, quotes)) 811 goto recordleft; 812 *loc = c; 813 loc--; 814 if (quotes && loc > startp && 815 *(loc - 1) == CTLESC) { 816 for (q = startp; q < loc; q++) 817 if (*q == CTLESC) 818 q++; 819 if (q > loc) 820 loc--; 821 } 822 } 823 return 0; 824 825 case VSTRIMRIGHT: 826 for (loc = str - 1; loc >= startp;) { 827 if (patmatch(str, loc, quotes)) 828 goto recordright; 829 loc--; 830 if (quotes && loc > startp && 831 *(loc - 1) == CTLESC) { 832 for (q = startp; q < loc; q++) 833 if (*q == CTLESC) 834 q++; 835 if (q > loc) 836 loc--; 837 } 838 } 839 return 0; 840 841 case VSTRIMRIGHTMAX: 842 for (loc = startp; loc < str - 1; loc++) { 843 if (patmatch(str, loc, quotes)) 844 goto recordright; 845 if (quotes && *loc == CTLESC) 846 loc++; 847 } 848 return 0; 849 850 default: 851 abort(); 852 } 853 854 recordleft: 855 *loc = c; 856 amount = ((str - 1) - (loc - startp)) - expdest; 857 STADJUST(amount, expdest); 858 while (loc != str - 1) 859 *startp++ = *loc++; 860 return 1; 861 862 recordright: 863 amount = loc - expdest; 864 STADJUST(amount, expdest); 865 STPUTC('\0', expdest); 866 STADJUST(-1, expdest); 867 return 1; 868 } 869 870 871 /* 872 * Expand a variable, and return a pointer to the next character in the 873 * input string. 874 */ 875 876 STATIC const char * 877 evalvar(const char *p, int flag) 878 { 879 int subtype; 880 int varflags; 881 const char *var; 882 char *val; 883 int patloc; 884 int c; 885 int set; 886 int special; 887 int startloc; 888 int varlen; 889 int apply_ifs; 890 const int quotes = flag & EXP_QNEEDED; 891 892 varflags = (unsigned char)*p++; 893 subtype = varflags & VSTYPE; 894 var = p; 895 special = !is_name(*p); 896 p = strchr(p, '=') + 1; 897 898 CTRACE(DBG_EXPAND, 899 ("evalvar \"%.*s\", flag=%#X quotes=%#X vf=%#X subtype=%X\n", 900 p - var - 1, var, flag, quotes, varflags, subtype)); 901 902 again: /* jump here after setting a variable with ${var=text} */ 903 if (varflags & VSLINENO) { 904 if (line_num.flags & VUNSET) { 905 set = 0; 906 val = NULL; 907 } else { 908 set = 1; 909 special = p - var; 910 val = NULL; 911 } 912 } else if (special) { 913 set = varisset(var, varflags & VSNUL); 914 val = NULL; 915 if (!set && *var == '@') 916 empty_dollar_at = 1; 917 } else { 918 val = lookupvar(var); 919 if (val == NULL || ((varflags & VSNUL) && val[0] == '\0')) { 920 val = NULL; 921 set = 0; 922 } else 923 set = 1; 924 } 925 926 varlen = 0; 927 startloc = expdest - stackblock(); 928 929 if (!set && uflag && *var != '@' && *var != '*') { 930 switch (subtype) { 931 case VSNORMAL: 932 case VSTRIMLEFT: 933 case VSTRIMLEFTMAX: 934 case VSTRIMRIGHT: 935 case VSTRIMRIGHTMAX: 936 case VSLENGTH: 937 error("%.*s: parameter not set", 938 (int)(p - var - 1), var); 939 /* NOTREACHED */ 940 } 941 } 942 943 #if 0 /* no longer need this $@ evil ... */ 944 if (!set && subtype != VSPLUS && special && *var == '@') 945 if (startloc > 0 && expdest[-1] == CTLQUOTEMARK) 946 expdest--, startloc--; 947 #endif 948 949 if (set && subtype != VSPLUS) { 950 /* insert the value of the variable */ 951 if (special) { 952 if (varflags & VSLINENO) { 953 /* 954 * The LINENO hack (expansion part) 955 */ 956 while (--special > 0) { 957 /* not needed, it is a number... 958 if (quotes && NEEDESC(*var)) 959 STPUTC(CTLESC, expdest); 960 */ 961 STPUTC(*var++, expdest); 962 } 963 } else 964 varvalue(var, varflags&VSQUOTE, subtype, flag); 965 if (subtype == VSLENGTH) { 966 varlen = expdest - stackblock() - startloc; 967 STADJUST(-varlen, expdest); 968 } 969 } else { 970 971 if (subtype == VSLENGTH) { 972 for (; *val; val++) 973 varlen++; 974 } else if (quotes && varflags & VSQUOTE) { 975 /* 976 * If we are going to look for magic in the 977 * value (quotes is set) and the expansion 978 * occurs inside "" (VSQUOTE) then any char 979 * that has any potential special meaning 980 * needs to have that meaning suppressed, 981 * so supply a CTLESC prefix for it. 982 */ 983 for (; (c = *val) != '\0'; val++) { 984 if (NEEDESC(c)) 985 STPUTC(CTLESC, expdest); 986 STPUTC(c, expdest); 987 } 988 } else { 989 /* 990 * We are going to rmescapes() later, 991 * so make sure that any data char that 992 * might be mistaken for one of our CTLxxx 993 * magic chars is protected ... always. 994 */ 995 for (; (c = *val) != '\0'; val++) { 996 if (ISCTL(c)) 997 STPUTC(CTLESC, expdest); 998 STPUTC(c, expdest); 999 } 1000 } 1001 } 1002 } 1003 1004 1005 if (varflags & VSQUOTE) { 1006 if (*var == '@' && shellparam.nparam != 1) 1007 apply_ifs = 1; 1008 else { 1009 /* 1010 * Mark so that we don't apply IFS if we recurse through 1011 * here expanding $bar from "${foo-$bar}". 1012 */ 1013 flag |= EXP_IN_QUOTES; 1014 apply_ifs = 0; 1015 } 1016 } else if (flag & EXP_IN_QUOTES) { 1017 apply_ifs = 0; 1018 } else 1019 apply_ifs = 1; 1020 1021 switch (subtype) { 1022 case VSLENGTH: 1023 expdest = cvtnum(varlen, expdest); 1024 break; 1025 1026 case VSNORMAL: 1027 break; 1028 1029 case VSPLUS: 1030 set = !set; 1031 /* FALLTHROUGH */ 1032 case VSMINUS: 1033 if (!set) { 1034 argstr(p, flag | (apply_ifs ? EXP_IFS_SPLIT : 0)); 1035 /* 1036 * ${x-a b c} doesn't get split, but removing the 1037 * 'apply_ifs = 0' apparently breaks ${1+"$@"}.. 1038 * ${x-'a b' c} should generate 2 args. 1039 */ 1040 if (*p != CTLENDVAR) 1041 /* We should have marked stuff already */ 1042 apply_ifs = 0; 1043 } 1044 break; 1045 1046 case VSTRIMLEFT: 1047 case VSTRIMLEFTMAX: 1048 case VSTRIMRIGHT: 1049 case VSTRIMRIGHTMAX: 1050 if (!set) { 1051 set = 1; /* allow argbackq to be advanced if needed */ 1052 break; 1053 } 1054 /* 1055 * Terminate the string and start recording the pattern 1056 * right after it 1057 */ 1058 STPUTC('\0', expdest); 1059 patloc = expdest - stackblock(); 1060 if (subevalvar_trim(p, patloc, subtype, startloc, varflags, 1061 quotes) == 0) { 1062 int amount = (expdest - stackblock() - patloc) + 1; 1063 STADJUST(-amount, expdest); 1064 } 1065 /* Remove any recorded regions beyond start of variable */ 1066 removerecordregions(startloc); 1067 apply_ifs = 1; 1068 break; 1069 1070 case VSASSIGN: 1071 case VSQUESTION: 1072 if (set) 1073 break; 1074 if (subevalvar(p, var, subtype, startloc, varflags)) { 1075 /* if subevalvar() returns, it always returns 1 */ 1076 1077 varflags &= ~VSNUL; 1078 /* 1079 * Remove any recorded regions beyond 1080 * start of variable 1081 */ 1082 removerecordregions(startloc); 1083 goto again; 1084 } 1085 apply_ifs = 0; /* never executed */ 1086 break; 1087 1088 default: 1089 abort(); 1090 } 1091 1092 if (apply_ifs) 1093 recordregion(startloc, expdest - stackblock(), 1094 varflags & VSQUOTE); 1095 1096 if (subtype != VSNORMAL) { /* skip to end of alternative */ 1097 int nesting = 1; 1098 for (;;) { 1099 if ((c = *p++) == CTLESC) 1100 p++; 1101 else if (c == CTLNONL) 1102 ; 1103 else if (c == CTLBACKQ || c == (CTLBACKQ|CTLQUOTE)) { 1104 if (set) 1105 argbackq = argbackq->next; 1106 } else if (c == CTLVAR) { 1107 if ((*p++ & VSTYPE) != VSNORMAL) 1108 nesting++; 1109 } else if (c == CTLENDVAR) { 1110 if (--nesting == 0) 1111 break; 1112 } 1113 } 1114 } 1115 return p; 1116 } 1117 1118 1119 1120 /* 1121 * Test whether a special parameter is set. 1122 */ 1123 1124 STATIC int 1125 varisset(const char *name, int nulok) 1126 { 1127 if (*name == '!') 1128 return backgndpid != -1; 1129 else if (*name == '@' || *name == '*') { 1130 if (*shellparam.p == NULL) 1131 return 0; 1132 1133 if (nulok) { 1134 char **av; 1135 1136 for (av = shellparam.p; *av; av++) 1137 if (**av != '\0') 1138 return 1; 1139 return 0; 1140 } 1141 } else if (is_digit(*name)) { 1142 char *ap; 1143 long num; 1144 1145 /* 1146 * handle overflow sensibly (the *ap tests should never fail) 1147 */ 1148 errno = 0; 1149 num = strtol(name, &ap, 10); 1150 if (errno != 0 || (*ap != '\0' && *ap != '=')) 1151 return 0; 1152 1153 if (num == 0) 1154 ap = arg0; 1155 else if (num > shellparam.nparam) 1156 return 0; 1157 else 1158 ap = shellparam.p[num - 1]; 1159 1160 if (nulok && (ap == NULL || *ap == '\0')) 1161 return 0; 1162 } 1163 return 1; 1164 } 1165 1166 1167 1168 /* 1169 * Add the value of a specialized variable to the stack string. 1170 */ 1171 1172 STATIC void 1173 varvalue(const char *name, int quoted, int subtype, int flag) 1174 { 1175 int num; 1176 char *p; 1177 int i; 1178 int sep; 1179 char **ap; 1180 #ifdef DEBUG 1181 char *start = expdest; 1182 #endif 1183 1184 VTRACE(DBG_EXPAND, ("varvalue(%c%s, sub=%d, fl=%#x)", *name, 1185 quoted ? ", quoted" : "", subtype, flag)); 1186 1187 if (subtype == VSLENGTH) /* no magic required ... */ 1188 flag &= ~EXP_FULL; 1189 1190 #define STRTODEST(p) \ 1191 do {\ 1192 if ((flag & EXP_QNEEDED) && quoted) { \ 1193 while (*p) { \ 1194 if (NEEDESC(*p)) \ 1195 STPUTC(CTLESC, expdest); \ 1196 STPUTC(*p++, expdest); \ 1197 } \ 1198 } else \ 1199 while (*p) { \ 1200 if (ISCTL(*p)) \ 1201 STPUTC(CTLESC, expdest); \ 1202 STPUTC(*p++, expdest); \ 1203 } \ 1204 } while (0) 1205 1206 1207 switch (*name) { 1208 case '$': 1209 num = rootpid; 1210 break; 1211 case '?': 1212 num = exitstatus; 1213 break; 1214 case '#': 1215 num = shellparam.nparam; 1216 break; 1217 case '!': 1218 num = backgndpid; 1219 break; 1220 case '-': 1221 for (i = 0; i < option_flags; i++) { 1222 if (optlist[optorder[i]].val) 1223 STPUTC(optlist[optorder[i]].letter, expdest); 1224 } 1225 VTRACE(DBG_EXPAND, (": %.*s\n", expdest-start, start)); 1226 return; 1227 case '@': 1228 if (flag & EXP_SPLIT && quoted) { 1229 VTRACE(DBG_EXPAND, (": $@ split (%d)\n", 1230 shellparam.nparam)); 1231 #if 0 1232 /* GROSS HACK */ 1233 if (shellparam.nparam == 0 && 1234 expdest[-1] == CTLQUOTEMARK) 1235 expdest--; 1236 /* KCAH SSORG */ 1237 #endif 1238 if (shellparam.nparam == 0) 1239 empty_dollar_at = 1; 1240 1241 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 1242 if (*p == '\0') { 1243 /* retain an explicit null string */ 1244 STPUTC(CTLQUOTEMARK, expdest); 1245 STPUTC(CTLQUOTEEND, expdest); 1246 } else 1247 STRTODEST(p); 1248 if (*ap) 1249 /* A NUL separates args inside "" */ 1250 STPUTC('\0', expdest); 1251 } 1252 return; 1253 } 1254 /* fall through */ 1255 case '*': 1256 sep = ifsval()[0]; 1257 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 1258 STRTODEST(p); 1259 if (!*ap) 1260 break; 1261 if (sep) { 1262 if (quoted && (flag & EXP_QNEEDED) && 1263 NEEDESC(sep)) 1264 STPUTC(CTLESC, expdest); 1265 STPUTC(sep, expdest); 1266 } else 1267 if ((flag & (EXP_SPLIT|EXP_IN_QUOTES)) == EXP_SPLIT 1268 && !quoted && **ap != '\0') 1269 STPUTC('\0', expdest); 1270 } 1271 VTRACE(DBG_EXPAND, (": %.*s\n", expdest-start, start)); 1272 return; 1273 default: 1274 if (is_digit(*name)) { 1275 long lnum; 1276 1277 errno = 0; 1278 lnum = strtol(name, &p, 10); 1279 if (errno != 0 || (*p != '\0' && *p != '=')) 1280 return; 1281 1282 if (lnum == 0) 1283 p = arg0; 1284 else if (lnum > 0 && lnum <= shellparam.nparam) 1285 p = shellparam.p[lnum - 1]; 1286 else 1287 return; 1288 STRTODEST(p); 1289 } 1290 VTRACE(DBG_EXPAND, (": %.*s\n", expdest-start, start)); 1291 return; 1292 } 1293 /* 1294 * only the specials with an int value arrive here 1295 */ 1296 VTRACE(DBG_EXPAND, ("(%d)", num)); 1297 expdest = cvtnum(num, expdest); 1298 VTRACE(DBG_EXPAND, (": %.*s\n", expdest-start, start)); 1299 } 1300 1301 1302 1303 /* 1304 * Record the fact that we have to scan this region of the 1305 * string for IFS characters. 1306 */ 1307 1308 STATIC void 1309 recordregion(int start, int end, int inquotes) 1310 { 1311 struct ifsregion *ifsp; 1312 1313 VTRACE(DBG_EXPAND, ("recordregion(%d,%d,%d)\n", start, end, inquotes)); 1314 if (ifslastp == NULL) { 1315 ifsp = &ifsfirst; 1316 } else { 1317 if (ifslastp->endoff == start 1318 && ifslastp->inquotes == inquotes) { 1319 /* extend previous area */ 1320 ifslastp->endoff = end; 1321 return; 1322 } 1323 ifsp = (struct ifsregion *)ckmalloc(sizeof (struct ifsregion)); 1324 ifslastp->next = ifsp; 1325 } 1326 ifslastp = ifsp; 1327 ifslastp->next = NULL; 1328 ifslastp->begoff = start; 1329 ifslastp->endoff = end; 1330 ifslastp->inquotes = inquotes; 1331 } 1332 1333 1334 1335 /* 1336 * Break the argument string into pieces based upon IFS and add the 1337 * strings to the argument list. The regions of the string to be 1338 * searched for IFS characters have been stored by recordregion. 1339 */ 1340 STATIC void 1341 ifsbreakup(char *string, struct arglist *arglist) 1342 { 1343 struct ifsregion *ifsp; 1344 struct strlist *sp; 1345 char *start; 1346 char *p; 1347 char *q; 1348 const char *ifs; 1349 const char *ifsspc; 1350 int had_param_ch = 0; 1351 1352 start = string; 1353 1354 VTRACE(DBG_EXPAND, ("ifsbreakup(\"%s\")", string)); /* misses \0's */ 1355 if (ifslastp == NULL) { 1356 /* Return entire argument, IFS doesn't apply to any of it */ 1357 VTRACE(DBG_EXPAND, ("no regions\n", string)); 1358 sp = stalloc(sizeof(*sp)); 1359 sp->text = start; 1360 *arglist->lastp = sp; 1361 arglist->lastp = &sp->next; 1362 return; 1363 } 1364 1365 ifs = ifsval(); 1366 1367 for (ifsp = &ifsfirst; ifsp != NULL; ifsp = ifsp->next) { 1368 p = string + ifsp->begoff; 1369 VTRACE(DBG_EXPAND, (" !%.*s!(%d)", ifsp->endoff-ifsp->begoff, 1370 p, ifsp->endoff-ifsp->begoff)); 1371 while (p < string + ifsp->endoff) { 1372 had_param_ch = 1; 1373 q = p; 1374 if (IS_BORING(*p)) { 1375 p++; 1376 continue; 1377 } 1378 if (*p == CTLESC) 1379 p++; 1380 if (ifsp->inquotes) { 1381 /* Only NULs (should be from "$@") end args */ 1382 if (*p != 0) { 1383 p++; 1384 continue; 1385 } 1386 ifsspc = NULL; 1387 VTRACE(DBG_EXPAND, (" \\0 nxt:\"%s\" ", p)); 1388 } else { 1389 if (!strchr(ifs, *p)) { 1390 p++; 1391 continue; 1392 } 1393 had_param_ch = 0; 1394 ifsspc = strchr(" \t\n", *p); 1395 1396 /* Ignore IFS whitespace at start */ 1397 if (q == start && ifsspc != NULL) { 1398 p++; 1399 start = p; 1400 continue; 1401 } 1402 } 1403 1404 /* Save this argument... */ 1405 *q = '\0'; 1406 VTRACE(DBG_EXPAND, ("<%s>", start)); 1407 sp = stalloc(sizeof(*sp)); 1408 sp->text = start; 1409 *arglist->lastp = sp; 1410 arglist->lastp = &sp->next; 1411 p++; 1412 1413 if (ifsspc != NULL) { 1414 /* Ignore further trailing IFS whitespace */ 1415 for (; p < string + ifsp->endoff; p++) { 1416 q = p; 1417 if (*p == CTLNONL) 1418 continue; 1419 if (*p == CTLESC) 1420 p++; 1421 if (strchr(ifs, *p) == NULL) { 1422 p = q; 1423 break; 1424 } 1425 if (strchr(" \t\n", *p) == NULL) { 1426 p++; 1427 break; 1428 } 1429 } 1430 } 1431 start = p; 1432 } 1433 } 1434 1435 /* 1436 while (*start == CTLQUOTEEND) 1437 start++; 1438 */ 1439 1440 /* 1441 * Save anything left as an argument. 1442 * Traditionally we have treated 'IFS=':'; set -- x$IFS' as 1443 * generating 2 arguments, the second of which is empty. 1444 * Some recent clarification of the Posix spec say that it 1445 * should only generate one.... 1446 */ 1447 if (had_param_ch || *start != 0) { 1448 VTRACE(DBG_EXPAND, (" T<%s>", start)); 1449 sp = stalloc(sizeof(*sp)); 1450 sp->text = start; 1451 *arglist->lastp = sp; 1452 arglist->lastp = &sp->next; 1453 } 1454 VTRACE(DBG_EXPAND, ("\n")); 1455 } 1456 1457 STATIC void 1458 ifsfree(void) 1459 { 1460 while (ifsfirst.next != NULL) { 1461 struct ifsregion *ifsp; 1462 INTOFF; 1463 ifsp = ifsfirst.next->next; 1464 ckfree(ifsfirst.next); 1465 ifsfirst.next = ifsp; 1466 INTON; 1467 } 1468 ifslastp = NULL; 1469 ifsfirst.next = NULL; 1470 } 1471 1472 1473 1474 /* 1475 * Expand shell metacharacters. At this point, the only control characters 1476 * should be escapes. The results are stored in the list exparg. 1477 */ 1478 1479 char *expdir; 1480 1481 1482 STATIC void 1483 expandmeta(struct strlist *str, int flag) 1484 { 1485 char *p; 1486 struct strlist **savelastp; 1487 struct strlist *sp; 1488 char c; 1489 /* TODO - EXP_REDIR */ 1490 1491 while (str) { 1492 p = str->text; 1493 for (;;) { /* fast check for meta chars */ 1494 if ((c = *p++) == '\0') 1495 goto nometa; 1496 if (c == '*' || c == '?' || c == '[' /* || c == '!' */) 1497 break; 1498 } 1499 savelastp = exparg.lastp; 1500 INTOFF; 1501 if (expdir == NULL) { 1502 int i = strlen(str->text); 1503 expdir = ckmalloc(i < 2048 ? 2048 : i); /* XXX */ 1504 } 1505 1506 expmeta(expdir, str->text); 1507 ckfree(expdir); 1508 expdir = NULL; 1509 INTON; 1510 if (exparg.lastp == savelastp) { 1511 /* 1512 * no matches 1513 */ 1514 nometa: 1515 *exparg.lastp = str; 1516 rmescapes(str->text); 1517 exparg.lastp = &str->next; 1518 } else { 1519 *exparg.lastp = NULL; 1520 *savelastp = sp = expsort(*savelastp); 1521 while (sp->next != NULL) 1522 sp = sp->next; 1523 exparg.lastp = &sp->next; 1524 } 1525 str = str->next; 1526 } 1527 } 1528 1529 STATIC void 1530 add_args(struct strlist *str) 1531 { 1532 while (str) { 1533 *exparg.lastp = str; 1534 rmescapes(str->text); 1535 exparg.lastp = &str->next; 1536 str = str->next; 1537 } 1538 } 1539 1540 1541 /* 1542 * Do metacharacter (i.e. *, ?, [...]) expansion. 1543 */ 1544 1545 STATIC void 1546 expmeta(char *enddir, char *name) 1547 { 1548 char *p; 1549 const char *cp; 1550 char *q; 1551 char *start; 1552 char *endname; 1553 int metaflag; 1554 struct stat statb; 1555 DIR *dirp; 1556 struct dirent *dp; 1557 int atend; 1558 int matchdot; 1559 1560 CTRACE(DBG_EXPAND|DBG_MATCH, ("expmeta(\"%s\")\n", name)); 1561 metaflag = 0; 1562 start = name; 1563 for (p = name ; ; p++) { 1564 if (*p == '*' || *p == '?') 1565 metaflag = 1; 1566 else if (*p == '[') { 1567 q = p + 1; 1568 if (*q == '!' || *q == '^') 1569 q++; 1570 for (;;) { 1571 while (IS_BORING(*q)) 1572 q++; 1573 if (*q == ']') { 1574 q++; 1575 metaflag = 1; 1576 break; 1577 } 1578 if (*q == '[' && q[1] == ':') { 1579 /* 1580 * character class, look for :] ending 1581 * also stop on ']' (end bracket expr) 1582 * or '\0' or '/' (end pattern) 1583 */ 1584 while (*++q != '\0' && *q != ']' && 1585 *q != '/') { 1586 if (*q == CTLESC) { 1587 if (*++q == '\0') 1588 break; 1589 if (*q == '/') 1590 break; 1591 } else if (*q == ':' && 1592 q[1] == ']') 1593 break; 1594 } 1595 if (*q == ':') { 1596 /* 1597 * stopped at ':]' 1598 * still in [...] 1599 * skip ":]" and continue; 1600 */ 1601 q += 2; 1602 continue; 1603 } 1604 1605 /* done at end of pattern, not [...] */ 1606 if (*q == '\0' || *q == '/') 1607 break; 1608 1609 /* found the ']', we have a [...] */ 1610 metaflag = 1; 1611 q++; /* skip ']' */ 1612 break; 1613 } 1614 if (*q == CTLESC) 1615 q++; 1616 /* end of pattern cannot be escaped */ 1617 if (*q == '/' || *q == '\0') 1618 break; 1619 q++; 1620 } 1621 } else if (*p == '\0') 1622 break; 1623 else if (IS_BORING(*p)) 1624 continue; 1625 else if (*p == CTLESC) 1626 p++; 1627 if (*p == '/') { 1628 if (metaflag) 1629 break; 1630 start = p + 1; 1631 } 1632 } 1633 if (metaflag == 0) { /* we've reached the end of the file name */ 1634 if (enddir != expdir) 1635 metaflag++; 1636 for (p = name ; ; p++) { 1637 if (IS_BORING(*p)) 1638 continue; 1639 if (*p == CTLESC) 1640 p++; 1641 *enddir++ = *p; 1642 if (*p == '\0') 1643 break; 1644 } 1645 if (metaflag == 0 || lstat(expdir, &statb) >= 0) 1646 addfname(expdir); 1647 return; 1648 } 1649 endname = p; 1650 if (start != name) { 1651 p = name; 1652 while (p < start) { 1653 while (IS_BORING(*p)) 1654 p++; 1655 if (*p == CTLESC) 1656 p++; 1657 *enddir++ = *p++; 1658 } 1659 } 1660 if (enddir == expdir) { 1661 cp = "."; 1662 } else if (enddir == expdir + 1 && *expdir == '/') { 1663 cp = "/"; 1664 } else { 1665 cp = expdir; 1666 enddir[-1] = '\0'; 1667 } 1668 if ((dirp = opendir(cp)) == NULL) 1669 return; 1670 if (enddir != expdir) 1671 enddir[-1] = '/'; 1672 if (*endname == 0) { 1673 atend = 1; 1674 } else { 1675 atend = 0; 1676 *endname++ = '\0'; 1677 } 1678 matchdot = 0; 1679 p = start; 1680 while (IS_BORING(*p)) 1681 p++; 1682 if (*p == CTLESC) 1683 p++; 1684 if (*p == '.') 1685 matchdot++; 1686 while (! int_pending() && (dp = readdir(dirp)) != NULL) { 1687 if (dp->d_name[0] == '.' && ! matchdot) 1688 continue; 1689 if (patmatch(start, dp->d_name, 0)) { 1690 if (atend) { 1691 scopy(dp->d_name, enddir); 1692 addfname(expdir); 1693 } else { 1694 for (p = enddir, cp = dp->d_name; 1695 (*p++ = *cp++) != '\0';) 1696 continue; 1697 p[-1] = '/'; 1698 expmeta(p, endname); 1699 } 1700 } 1701 } 1702 closedir(dirp); 1703 if (! atend) 1704 endname[-1] = '/'; 1705 } 1706 1707 1708 /* 1709 * Add a file name to the list. 1710 */ 1711 1712 STATIC void 1713 addfname(char *name) 1714 { 1715 char *p; 1716 struct strlist *sp; 1717 1718 p = stalloc(strlen(name) + 1); 1719 scopy(name, p); 1720 sp = stalloc(sizeof(*sp)); 1721 sp->text = p; 1722 *exparg.lastp = sp; 1723 exparg.lastp = &sp->next; 1724 } 1725 1726 1727 /* 1728 * Sort the results of file name expansion. It calculates the number of 1729 * strings to sort and then calls msort (short for merge sort) to do the 1730 * work. 1731 */ 1732 1733 STATIC struct strlist * 1734 expsort(struct strlist *str) 1735 { 1736 int len; 1737 struct strlist *sp; 1738 1739 len = 0; 1740 for (sp = str ; sp ; sp = sp->next) 1741 len++; 1742 return msort(str, len); 1743 } 1744 1745 1746 STATIC struct strlist * 1747 msort(struct strlist *list, int len) 1748 { 1749 struct strlist *p, *q = NULL; 1750 struct strlist **lpp; 1751 int half; 1752 int n; 1753 1754 if (len <= 1) 1755 return list; 1756 half = len >> 1; 1757 p = list; 1758 for (n = half ; --n >= 0 ; ) { 1759 q = p; 1760 p = p->next; 1761 } 1762 q->next = NULL; /* terminate first half of list */ 1763 q = msort(list, half); /* sort first half of list */ 1764 p = msort(p, len - half); /* sort second half */ 1765 lpp = &list; 1766 for (;;) { 1767 if (strcmp(p->text, q->text) < 0) { 1768 *lpp = p; 1769 lpp = &p->next; 1770 if ((p = *lpp) == NULL) { 1771 *lpp = q; 1772 break; 1773 } 1774 } else { 1775 *lpp = q; 1776 lpp = &q->next; 1777 if ((q = *lpp) == NULL) { 1778 *lpp = p; 1779 break; 1780 } 1781 } 1782 } 1783 return list; 1784 } 1785 1786 1787 /* 1788 * See if a character matches a character class, starting at the first colon 1789 * of "[:class:]". 1790 * If a valid character class is recognized, a pointer to the next character 1791 * after the final closing bracket is stored into *end, otherwise a null 1792 * pointer is stored into *end. 1793 */ 1794 static int 1795 match_charclass(const char *p, wchar_t chr, const char **end) 1796 { 1797 char name[20]; 1798 char *nameend; 1799 wctype_t cclass; 1800 char *q; 1801 1802 *end = NULL; 1803 p++; 1804 q = &name[0]; 1805 nameend = strstr(p, ":]"); 1806 if (nameend == NULL || nameend == p) /* not a valid class */ 1807 return 0; 1808 1809 if (*p == CTLESC) { 1810 if (*++p == CTLESC) 1811 return 0; 1812 if (p == nameend) 1813 return 0; 1814 } 1815 if (!is_alpha(*p)) 1816 return 0; 1817 while (p < nameend) { 1818 if (*p == CTLESC) { 1819 p++; 1820 if (p == nameend) 1821 return 0; 1822 } 1823 if (!is_in_name(*p)) /* '_' is a local extension */ 1824 return 0; 1825 if (q < &name[sizeof name]) 1826 *q++ = *p++; 1827 else 1828 p++; 1829 } 1830 1831 *end = nameend + 2; /* committed to it being a char class */ 1832 1833 if (q < &name[sizeof name]) /* a usable name found */ 1834 *q++ = '\0'; 1835 else /* too long, valid, but no match */ 1836 return 0; 1837 1838 cclass = wctype(name); 1839 /* An unknown class matches nothing but is valid nevertheless. */ 1840 if (cclass == 0) 1841 return 0; 1842 return iswctype(chr, cclass); 1843 } 1844 1845 1846 /* 1847 * Returns true if the pattern matches the string. 1848 */ 1849 1850 STATIC int 1851 patmatch(const char *pattern, const char *string, int squoted) 1852 { 1853 const char *p, *q, *end; 1854 const char *bt_p, *bt_q; 1855 char c; 1856 wchar_t wc, wc2; 1857 1858 VTRACE(DBG_MATCH, ("patmatch(P=\"%s\", W=\"%s\"%s): ", 1859 pattern, string, squoted ? ", SQ" : "")); 1860 p = pattern; 1861 q = string; 1862 bt_p = NULL; 1863 bt_q = NULL; 1864 for (;;) { 1865 switch (c = *p++) { 1866 case '\0': 1867 if (squoted && *q == CTLESC) { 1868 if (q[1] == '\0') 1869 q++; 1870 } 1871 if (*q != '\0') 1872 goto backtrack; 1873 VTRACE(DBG_MATCH, ("match\n")); 1874 return 1; 1875 case CTLESC: 1876 if (squoted && *q == CTLESC) 1877 q++; 1878 if (*p == '\0' && *q == '\0') { 1879 VTRACE(DBG_MATCH, ("match-\\\n")); 1880 return 1; 1881 } 1882 if (*q++ != *p++) 1883 goto backtrack; 1884 break; 1885 case '\\': 1886 if (squoted && *q == CTLESC) 1887 q++; 1888 if (*q++ != *p++) 1889 goto backtrack; 1890 break; 1891 case CTLQUOTEMARK: 1892 case CTLQUOTEEND: 1893 case CTLNONL: 1894 continue; 1895 case '?': 1896 if (squoted && *q == CTLESC) 1897 q++; 1898 if (*q++ == '\0') { 1899 VTRACE(DBG_MATCH, ("?fail\n")); 1900 return 0; 1901 } 1902 break; 1903 case '*': 1904 c = *p; 1905 while (c == CTLQUOTEMARK || c == '*') 1906 c = *++p; 1907 if (c != CTLESC && !IS_BORING(c) && 1908 c != '?' && c != '*' && c != '[') { 1909 while (*q != c) { 1910 if (squoted && *q == CTLESC && 1911 q[1] == c) 1912 break; 1913 if (*q == '\0') { 1914 VTRACE(DBG_MATCH, ("*fail\n")); 1915 return 0; 1916 } 1917 if (squoted && *q == CTLESC) 1918 q++; 1919 q++; 1920 } 1921 } 1922 if (c == CTLESC && p[1] == '\0') { 1923 VTRACE(DBG_MATCH, ("match+\\\n")); 1924 return 1; 1925 } 1926 /* 1927 * First try the shortest match for the '*' that 1928 * could work. We can forget any earlier '*' since 1929 * there is no way having it match more characters 1930 * can help us, given that we are already here. 1931 */ 1932 bt_p = p; 1933 bt_q = q; 1934 break; 1935 case '[': { 1936 const char *savep, *saveq, *endp; 1937 int invert, found; 1938 unsigned char chr; 1939 1940 /* 1941 * First quick check to see if there is a 1942 * possible matching ']' - if not, then this 1943 * is not a char class, and the '[' is just 1944 * a literal '['. 1945 * 1946 * This check will not detect all non classes, but 1947 * that's OK - It just means that we execute the 1948 * harder code sometimes when it it cannot succeed. 1949 */ 1950 endp = p; 1951 if (*endp == '!' || *endp == '^') 1952 endp++; 1953 for (;;) { 1954 while (IS_BORING(*endp)) 1955 endp++; 1956 if (*endp == '\0') 1957 goto dft; /* no matching ] */ 1958 if (*endp++ == ']') 1959 break; 1960 } 1961 /* end shortcut */ 1962 1963 savep = p, saveq = q; 1964 invert = 0; 1965 if (*p == '!' || *p == '^') { 1966 invert++; 1967 p++; 1968 } 1969 found = 0; 1970 if (*q == '\0') { 1971 VTRACE(DBG_MATCH, ("[]fail\n")); 1972 return 0; 1973 } 1974 if (squoted && *q == CTLESC) 1975 q++; 1976 chr = (unsigned char)*q++; 1977 c = *p++; 1978 do { 1979 if (IS_BORING(c)) 1980 continue; 1981 if (c == '\0') { 1982 p = savep, q = saveq; 1983 c = '['; 1984 goto dft; 1985 } 1986 if (c == '[' && *p == ':') { 1987 found |= match_charclass(p, chr, &end); 1988 if (end != NULL) { 1989 p = end; 1990 continue; 1991 } 1992 } 1993 if (c == CTLESC || c == '\\') 1994 c = *p++; 1995 wc = (unsigned char)c; 1996 if (*p == '-' && p[1] != ']') { 1997 p++; 1998 if (*p == CTLESC || *p == '\\') 1999 p++; 2000 wc2 = (unsigned char)*p++; 2001 if ( collate_range_cmp(chr, wc) >= 0 2002 && collate_range_cmp(chr, wc2) <= 0 2003 ) 2004 found = 1; 2005 } else { 2006 if (chr == wc) 2007 found = 1; 2008 } 2009 } while ((c = *p++) != ']'); 2010 if (found == invert) 2011 goto backtrack; 2012 break; 2013 } 2014 dft: default: 2015 if (squoted && *q == CTLESC) 2016 q++; 2017 if (*q++ == c) 2018 break; 2019 backtrack: 2020 /* 2021 * If we have a mismatch (other than hitting the end 2022 * of the string), go back to the last '*' seen and 2023 * have it match one additional character. 2024 */ 2025 if (bt_p == NULL) { 2026 VTRACE(DBG_MATCH, ("BTP fail\n")); 2027 return 0; 2028 } 2029 if (*bt_q == '\0') { 2030 VTRACE(DBG_MATCH, ("BTQ fail\n")); 2031 return 0; 2032 } 2033 bt_q++; 2034 p = bt_p; 2035 q = bt_q; 2036 break; 2037 } 2038 } 2039 } 2040 2041 2042 2043 /* 2044 * Remove any CTLESC or CTLNONL characters from a string. 2045 * 2046 * String is modified in place, and we return the length of the result 2047 */ 2048 2049 int 2050 rmescapes(char *str) 2051 { 2052 char *p, *q; 2053 2054 p = str; 2055 while (!ISCTL(*p)) { 2056 if (*p++ == '\0') 2057 return ((int)(p - str) - 1); 2058 } 2059 q = p; 2060 while (*p) { 2061 if (IS_BORING(*p)) { 2062 p++; 2063 continue; 2064 } 2065 if (*p == CTLCNL) { 2066 p++; 2067 *q++ = '\n'; 2068 continue; 2069 } 2070 if (*p == CTLESC) 2071 p++; 2072 #ifdef DEBUG 2073 else if (ISCTL(*p)) 2074 abort(); 2075 #endif 2076 *q++ = *p++; 2077 } 2078 *q = '\0'; 2079 2080 return ((int)(q - str)); 2081 } 2082 2083 /* 2084 * and a special version for dealing with expressions to be parsed 2085 * by the arithmetic evaluator. That needs to be able to count \n's 2086 * even ones that were \newline elided \n's, so we have to put the 2087 * latter back into the string - just being careful to put them only 2088 * at a place where white space can reasonably occur in the string 2089 * -- then the \n we insert will just be white space, and ignored 2090 * for all purposes except line counting. 2091 */ 2092 2093 void 2094 rmescapes_nl(char *str) 2095 { 2096 char *p, *q; 2097 int nls = 0, holdnl = 0, holdlast; 2098 2099 p = str; 2100 while (!ISCTL(*p)) { 2101 if (*p++ == '\0') 2102 return; 2103 } 2104 if (p > str) /* must reprocess char before stopper (if any) */ 2105 --p; /* so we do not place a \n badly */ 2106 q = p; 2107 while (*p) { 2108 if (*p == CTLQUOTEMARK || *p == CTLQUOTEEND) { 2109 p++; 2110 continue; 2111 } 2112 if (*p == CTLNONL) { 2113 p++; 2114 nls++; 2115 continue; 2116 } 2117 if (*p == CTLCNL) { 2118 p++; 2119 *q++ = '\n'; 2120 continue; 2121 } 2122 if (*p == CTLESC) 2123 p++; 2124 #ifdef DEBUG 2125 else if (ISCTL(*p)) 2126 abort(); 2127 #endif 2128 2129 holdlast = holdnl; 2130 holdnl = is_in_name(*p); /* letters, digits, _ */ 2131 if (q == str || is_space(q[-1]) || (*p != '=' && q[-1] != *p)) { 2132 if (nls > 0 && holdnl != holdlast) { 2133 while (nls > 0) 2134 *q++ = '\n', nls--; 2135 } 2136 } 2137 *q++ = *p++; 2138 } 2139 while (--nls >= 0) 2140 *q++ = '\n'; 2141 *q = '\0'; 2142 } 2143 2144 2145 2146 /* 2147 * See if a pattern matches in a case statement. 2148 */ 2149 2150 int 2151 casematch(union node *pattern, char *val) 2152 { 2153 struct stackmark smark; 2154 int result; 2155 char *p; 2156 2157 CTRACE(DBG_MATCH, ("casematch(P=\"%s\", W=\"%s\")\n", 2158 pattern->narg.text, val)); 2159 setstackmark(&smark); 2160 argbackq = pattern->narg.backquote; 2161 STARTSTACKSTR(expdest); 2162 ifslastp = NULL; 2163 argstr(pattern->narg.text, EXP_TILDE | EXP_CASE); 2164 STPUTC('\0', expdest); 2165 p = grabstackstr(expdest); 2166 result = patmatch(p, val, 0); 2167 popstackmark(&smark); 2168 return result; 2169 } 2170 2171 /* 2172 * Our own itoa(). Assumes result buffer is on the stack 2173 */ 2174 2175 STATIC char * 2176 cvtnum(int num, char *buf) 2177 { 2178 char temp[32]; 2179 int neg = num < 0; 2180 char *p = temp + sizeof temp - 1; 2181 2182 if (neg) 2183 num = -num; 2184 2185 *p = '\0'; 2186 do { 2187 *--p = num % 10 + '0'; 2188 } while ((num /= 10) != 0 && p > temp + 1); 2189 2190 if (neg) 2191 *--p = '-'; 2192 2193 while (*p) 2194 STPUTC(*p++, buf); 2195 return buf; 2196 } 2197 2198 /* 2199 * Do most of the work for wordexp(3). 2200 */ 2201 2202 int 2203 wordexpcmd(int argc, char **argv) 2204 { 2205 size_t len; 2206 int i; 2207 2208 out1fmt("%d", argc - 1); 2209 out1c('\0'); 2210 for (i = 1, len = 0; i < argc; i++) 2211 len += strlen(argv[i]); 2212 out1fmt("%zu", len); 2213 out1c('\0'); 2214 for (i = 1; i < argc; i++) { 2215 out1str(argv[i]); 2216 out1c('\0'); 2217 } 2218 return (0); 2219 } 2220