expand.c revision 223024
1/*- 2 * Copyright (c) 1991, 1993 3 * The Regents of the University of California. All rights reserved. 4 * Copyright (c) 1997-2005 5 * Herbert Xu <herbert@gondor.apana.org.au>. All rights reserved. 6 * 7 * This code is derived from software contributed to Berkeley by 8 * Kenneth Almquist. 9 * 10 * Redistribution and use in source and binary forms, with or without 11 * modification, are permitted provided that the following conditions 12 * are met: 13 * 1. Redistributions of source code must retain the above copyright 14 * notice, this list of conditions and the following disclaimer. 15 * 2. Redistributions in binary form must reproduce the above copyright 16 * notice, this list of conditions and the following disclaimer in the 17 * documentation and/or other materials provided with the distribution. 18 * 4. Neither the name of the University nor the names of its contributors 19 * may be used to endorse or promote products derived from this software 20 * without specific prior written permission. 21 * 22 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 25 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 32 * SUCH DAMAGE. 33 */ 34 35#ifndef lint 36#if 0 37static char sccsid[] = "@(#)expand.c 8.5 (Berkeley) 5/15/95"; 38#endif 39#endif /* not lint */ 40#include <sys/cdefs.h> 41__FBSDID("$FreeBSD: head/bin/sh/expand.c 223024 2011-06-12 23:06:04Z jilles $"); 42 43#include <sys/types.h> 44#include <sys/time.h> 45#include <sys/stat.h> 46#include <dirent.h> 47#include <errno.h> 48#include <inttypes.h> 49#include <limits.h> 50#include <pwd.h> 51#include <stdio.h> 52#include <stdlib.h> 53#include <string.h> 54#include <unistd.h> 55#include <wchar.h> 56 57/* 58 * Routines to expand arguments to commands. We have to deal with 59 * backquotes, shell variables, and file metacharacters. 60 */ 61 62#include "shell.h" 63#include "main.h" 64#include "nodes.h" 65#include "eval.h" 66#include "expand.h" 67#include "syntax.h" 68#include "parser.h" 69#include "jobs.h" 70#include "options.h" 71#include "var.h" 72#include "input.h" 73#include "output.h" 74#include "memalloc.h" 75#include "error.h" 76#include "mystring.h" 77#include "arith.h" 78#include "show.h" 79 80/* 81 * Structure specifying which parts of the string should be searched 82 * for IFS characters. 83 */ 84 85struct ifsregion { 86 struct ifsregion *next; /* next region in list */ 87 int begoff; /* offset of start of region */ 88 int endoff; /* offset of end of region */ 89 int inquotes; /* search for nul bytes only */ 90}; 91 92 93static char *expdest; /* output of current string */ 94static struct nodelist *argbackq; /* list of back quote expressions */ 95static struct ifsregion ifsfirst; /* first struct in list of ifs regions */ 96static struct ifsregion *ifslastp; /* last struct in list */ 97static struct arglist exparg; /* holds expanded arg list */ 98 99static void argstr(char *, int); 100static char *exptilde(char *, int); 101static void expbackq(union node *, int, int); 102static int subevalvar(char *, char *, int, int, int, int, int); 103static char *evalvar(char *, int); 104static int varisset(char *, int); 105static void varvalue(char *, int, int, int); 106static void recordregion(int, int, int); 107static void removerecordregions(int); 108static void ifsbreakup(char *, struct arglist *); 109static void expandmeta(struct strlist *, int); 110static void expmeta(char *, char *); 111static void addfname(char *); 112static struct strlist *expsort(struct strlist *); 113static struct strlist *msort(struct strlist *, int); 114static char *cvtnum(int, char *); 115static int collate_range_cmp(wchar_t, wchar_t); 116 117static int 118collate_range_cmp(wchar_t c1, wchar_t c2) 119{ 120 static wchar_t s1[2], s2[2]; 121 122 s1[0] = c1; 123 s2[0] = c2; 124 return (wcscoll(s1, s2)); 125} 126 127/* 128 * Expand shell variables and backquotes inside a here document. 129 * union node *arg the document 130 * int fd; where to write the expanded version 131 */ 132 133void 134expandhere(union node *arg, int fd) 135{ 136 expandarg(arg, (struct arglist *)NULL, 0); 137 xwrite(fd, stackblock(), expdest - stackblock()); 138} 139 140static char * 141stputs_quotes(const char *data, const char *syntax, char *p) 142{ 143 while (*data) { 144 CHECKSTRSPACE(2, p); 145 if (syntax[(int)*data] == CCTL) 146 USTPUTC(CTLESC, p); 147 USTPUTC(*data++, p); 148 } 149 return (p); 150} 151#define STPUTS_QUOTES(data, syntax, p) p = stputs_quotes((data), syntax, p) 152 153/* 154 * Perform expansions on an argument, placing the resulting list of arguments 155 * in arglist. Parameter expansion, command substitution and arithmetic 156 * expansion are always performed; additional expansions can be requested 157 * via flag (EXP_*). 158 * The result is left in the stack string. 159 * When arglist is NULL, perform here document expansion. 160 * 161 * Caution: this function uses global state and is not reentrant. 162 * However, a new invocation after an interrupted invocation is safe 163 * and will reset the global state for the new call. 164 */ 165void 166expandarg(union node *arg, struct arglist *arglist, int flag) 167{ 168 struct strlist *sp; 169 char *p; 170 171 argbackq = arg->narg.backquote; 172 STARTSTACKSTR(expdest); 173 ifsfirst.next = NULL; 174 ifslastp = NULL; 175 argstr(arg->narg.text, flag); 176 if (arglist == NULL) { 177 STACKSTRNUL(expdest); 178 return; /* here document expanded */ 179 } 180 STPUTC('\0', expdest); 181 p = grabstackstr(expdest); 182 exparg.lastp = &exparg.list; 183 /* 184 * TODO - EXP_REDIR 185 */ 186 if (flag & EXP_FULL) { 187 ifsbreakup(p, &exparg); 188 *exparg.lastp = NULL; 189 exparg.lastp = &exparg.list; 190 expandmeta(exparg.list, flag); 191 } else { 192 if (flag & EXP_REDIR) /*XXX - for now, just remove escapes */ 193 rmescapes(p); 194 sp = (struct strlist *)stalloc(sizeof (struct strlist)); 195 sp->text = p; 196 *exparg.lastp = sp; 197 exparg.lastp = &sp->next; 198 } 199 while (ifsfirst.next != NULL) { 200 struct ifsregion *ifsp; 201 INTOFF; 202 ifsp = ifsfirst.next->next; 203 ckfree(ifsfirst.next); 204 ifsfirst.next = ifsp; 205 INTON; 206 } 207 *exparg.lastp = NULL; 208 if (exparg.list) { 209 *arglist->lastp = exparg.list; 210 arglist->lastp = exparg.lastp; 211 } 212} 213 214 215 216/* 217 * Perform parameter expansion, command substitution and arithmetic 218 * expansion, and tilde expansion if requested via EXP_TILDE/EXP_VARTILDE. 219 * Processing ends at a CTLENDVAR character as well as '\0'. 220 * This is used to expand word in ${var+word} etc. 221 * If EXP_FULL, EXP_CASE or EXP_REDIR are set, keep and/or generate CTLESC 222 * characters to allow for further processing. 223 * If EXP_FULL is set, also preserve CTLQUOTEMARK characters. 224 */ 225static void 226argstr(char *p, int flag) 227{ 228 char c; 229 int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR); /* do CTLESC */ 230 int firsteq = 1; 231 int split_lit; 232 int lit_quoted; 233 234 split_lit = flag & EXP_SPLIT_LIT; 235 lit_quoted = flag & EXP_LIT_QUOTED; 236 flag &= ~(EXP_SPLIT_LIT | EXP_LIT_QUOTED); 237 if (*p == '~' && (flag & (EXP_TILDE | EXP_VARTILDE))) 238 p = exptilde(p, flag); 239 for (;;) { 240 CHECKSTRSPACE(2, expdest); 241 switch (c = *p++) { 242 case '\0': 243 case CTLENDVAR: 244 goto breakloop; 245 case CTLQUOTEMARK: 246 lit_quoted = 1; 247 /* "$@" syntax adherence hack */ 248 if (p[0] == CTLVAR && p[2] == '@' && p[3] == '=') 249 break; 250 if ((flag & EXP_FULL) != 0) 251 USTPUTC(c, expdest); 252 break; 253 case CTLQUOTEEND: 254 lit_quoted = 0; 255 break; 256 case CTLESC: 257 if (quotes) 258 USTPUTC(c, expdest); 259 c = *p++; 260 USTPUTC(c, expdest); 261 if (split_lit && !lit_quoted) 262 recordregion(expdest - stackblock() - 263 (quotes ? 2 : 1), 264 expdest - stackblock(), 0); 265 break; 266 case CTLVAR: 267 p = evalvar(p, flag); 268 break; 269 case CTLBACKQ: 270 case CTLBACKQ|CTLQUOTE: 271 expbackq(argbackq->n, c & CTLQUOTE, flag); 272 argbackq = argbackq->next; 273 break; 274 case CTLENDARI: 275 expari(flag); 276 break; 277 case ':': 278 case '=': 279 /* 280 * sort of a hack - expand tildes in variable 281 * assignments (after the first '=' and after ':'s). 282 */ 283 USTPUTC(c, expdest); 284 if (split_lit && !lit_quoted) 285 recordregion(expdest - stackblock() - 1, 286 expdest - stackblock(), 0); 287 if (flag & EXP_VARTILDE && *p == '~' && 288 (c != '=' || firsteq)) { 289 if (c == '=') 290 firsteq = 0; 291 p = exptilde(p, flag); 292 } 293 break; 294 default: 295 USTPUTC(c, expdest); 296 if (split_lit && !lit_quoted) 297 recordregion(expdest - stackblock() - 1, 298 expdest - stackblock(), 0); 299 } 300 } 301breakloop:; 302} 303 304/* 305 * Perform tilde expansion, placing the result in the stack string and 306 * returning the next position in the input string to process. 307 */ 308static char * 309exptilde(char *p, int flag) 310{ 311 char c, *startp = p; 312 struct passwd *pw; 313 char *home; 314 int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR); 315 316 while ((c = *p) != '\0') { 317 switch(c) { 318 case CTLESC: /* This means CTL* are always considered quoted. */ 319 case CTLVAR: 320 case CTLBACKQ: 321 case CTLBACKQ | CTLQUOTE: 322 case CTLARI: 323 case CTLENDARI: 324 case CTLQUOTEMARK: 325 return (startp); 326 case ':': 327 if (flag & EXP_VARTILDE) 328 goto done; 329 break; 330 case '/': 331 case CTLENDVAR: 332 goto done; 333 } 334 p++; 335 } 336done: 337 *p = '\0'; 338 if (*(startp+1) == '\0') { 339 if ((home = lookupvar("HOME")) == NULL) 340 goto lose; 341 } else { 342 if ((pw = getpwnam(startp+1)) == NULL) 343 goto lose; 344 home = pw->pw_dir; 345 } 346 if (*home == '\0') 347 goto lose; 348 *p = c; 349 if (quotes) 350 STPUTS_QUOTES(home, SQSYNTAX, expdest); 351 else 352 STPUTS(home, expdest); 353 return (p); 354lose: 355 *p = c; 356 return (startp); 357} 358 359 360static void 361removerecordregions(int endoff) 362{ 363 if (ifslastp == NULL) 364 return; 365 366 if (ifsfirst.endoff > endoff) { 367 while (ifsfirst.next != NULL) { 368 struct ifsregion *ifsp; 369 INTOFF; 370 ifsp = ifsfirst.next->next; 371 ckfree(ifsfirst.next); 372 ifsfirst.next = ifsp; 373 INTON; 374 } 375 if (ifsfirst.begoff > endoff) 376 ifslastp = NULL; 377 else { 378 ifslastp = &ifsfirst; 379 ifsfirst.endoff = endoff; 380 } 381 return; 382 } 383 384 ifslastp = &ifsfirst; 385 while (ifslastp->next && ifslastp->next->begoff < endoff) 386 ifslastp=ifslastp->next; 387 while (ifslastp->next != NULL) { 388 struct ifsregion *ifsp; 389 INTOFF; 390 ifsp = ifslastp->next->next; 391 ckfree(ifslastp->next); 392 ifslastp->next = ifsp; 393 INTON; 394 } 395 if (ifslastp->endoff > endoff) 396 ifslastp->endoff = endoff; 397} 398 399/* 400 * Expand arithmetic expression. Backup to start of expression, 401 * evaluate, place result in (backed up) result, adjust string position. 402 */ 403void 404expari(int flag) 405{ 406 char *p, *q, *start; 407 arith_t result; 408 int begoff; 409 int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR); 410 int quoted; 411 412 /* 413 * This routine is slightly over-complicated for 414 * efficiency. First we make sure there is 415 * enough space for the result, which may be bigger 416 * than the expression. Next we 417 * scan backwards looking for the start of arithmetic. If the 418 * next previous character is a CTLESC character, then we 419 * have to rescan starting from the beginning since CTLESC 420 * characters have to be processed left to right. 421 */ 422 CHECKSTRSPACE(DIGITS(result) - 2, expdest); 423 USTPUTC('\0', expdest); 424 start = stackblock(); 425 p = expdest - 2; 426 while (p >= start && *p != CTLARI) 427 --p; 428 if (p < start || *p != CTLARI) 429 error("missing CTLARI (shouldn't happen)"); 430 if (p > start && *(p - 1) == CTLESC) 431 for (p = start; *p != CTLARI; p++) 432 if (*p == CTLESC) 433 p++; 434 435 if (p[1] == '"') 436 quoted=1; 437 else 438 quoted=0; 439 begoff = p - start; 440 removerecordregions(begoff); 441 if (quotes) 442 rmescapes(p+2); 443 q = grabstackstr(expdest); 444 result = arith(p+2); 445 ungrabstackstr(q, expdest); 446 fmtstr(p, DIGITS(result), ARITH_FORMAT_STR, result); 447 while (*p++) 448 ; 449 if (quoted == 0) 450 recordregion(begoff, p - 1 - start, 0); 451 result = expdest - p + 1; 452 STADJUST(-result, expdest); 453} 454 455 456/* 457 * Perform command substitution. 458 */ 459static void 460expbackq(union node *cmd, int quoted, int flag) 461{ 462 struct backcmd in; 463 int i; 464 char buf[128]; 465 char *p; 466 char *dest = expdest; 467 struct ifsregion saveifs, *savelastp; 468 struct nodelist *saveargbackq; 469 char lastc; 470 int startloc = dest - stackblock(); 471 char const *syntax = quoted? DQSYNTAX : BASESYNTAX; 472 int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR); 473 int nnl; 474 475 INTOFF; 476 saveifs = ifsfirst; 477 savelastp = ifslastp; 478 saveargbackq = argbackq; 479 p = grabstackstr(dest); 480 evalbackcmd(cmd, &in); 481 ungrabstackstr(p, dest); 482 ifsfirst = saveifs; 483 ifslastp = savelastp; 484 argbackq = saveargbackq; 485 486 p = in.buf; 487 lastc = '\0'; 488 nnl = 0; 489 /* Don't copy trailing newlines */ 490 for (;;) { 491 if (--in.nleft < 0) { 492 if (in.fd < 0) 493 break; 494 while ((i = read(in.fd, buf, sizeof buf)) < 0 && errno == EINTR); 495 TRACE(("expbackq: read returns %d\n", i)); 496 if (i <= 0) 497 break; 498 p = buf; 499 in.nleft = i - 1; 500 } 501 lastc = *p++; 502 if (lastc != '\0') { 503 if (lastc == '\n') { 504 nnl++; 505 } else { 506 CHECKSTRSPACE(nnl + 2, dest); 507 while (nnl > 0) { 508 nnl--; 509 USTPUTC('\n', dest); 510 } 511 if (quotes && syntax[(int)lastc] == CCTL) 512 USTPUTC(CTLESC, dest); 513 USTPUTC(lastc, dest); 514 } 515 } 516 } 517 518 if (in.fd >= 0) 519 close(in.fd); 520 if (in.buf) 521 ckfree(in.buf); 522 if (in.jp) 523 exitstatus = waitforjob(in.jp, (int *)NULL); 524 if (quoted == 0) 525 recordregion(startloc, dest - stackblock(), 0); 526 TRACE(("expbackq: size=%td: \"%.*s\"\n", 527 ((dest - stackblock()) - startloc), 528 (int)((dest - stackblock()) - startloc), 529 stackblock() + startloc)); 530 expdest = dest; 531 INTON; 532} 533 534 535 536static int 537subevalvar(char *p, char *str, int strloc, int subtype, int startloc, 538 int varflags, int quotes) 539{ 540 char *startp; 541 char *loc = NULL; 542 char *q; 543 int c = 0; 544 struct nodelist *saveargbackq = argbackq; 545 int amount; 546 547 argstr(p, (subtype == VSTRIMLEFT || subtype == VSTRIMLEFTMAX || 548 subtype == VSTRIMRIGHT || subtype == VSTRIMRIGHTMAX ? 549 EXP_CASE : 0) | EXP_TILDE); 550 STACKSTRNUL(expdest); 551 argbackq = saveargbackq; 552 startp = stackblock() + startloc; 553 if (str == NULL) 554 str = stackblock() + strloc; 555 556 switch (subtype) { 557 case VSASSIGN: 558 setvar(str, startp, 0); 559 amount = startp - expdest; 560 STADJUST(amount, expdest); 561 varflags &= ~VSNUL; 562 return 1; 563 564 case VSQUESTION: 565 if (*p != CTLENDVAR) { 566 outfmt(out2, "%s\n", startp); 567 error((char *)NULL); 568 } 569 error("%.*s: parameter %snot set", (int)(p - str - 1), 570 str, (varflags & VSNUL) ? "null or " 571 : nullstr); 572 return 0; 573 574 case VSTRIMLEFT: 575 for (loc = startp; loc < str; loc++) { 576 c = *loc; 577 *loc = '\0'; 578 if (patmatch(str, startp, quotes)) { 579 *loc = c; 580 goto recordleft; 581 } 582 *loc = c; 583 if (quotes && *loc == CTLESC) 584 loc++; 585 } 586 return 0; 587 588 case VSTRIMLEFTMAX: 589 for (loc = str - 1; loc >= startp;) { 590 c = *loc; 591 *loc = '\0'; 592 if (patmatch(str, startp, quotes)) { 593 *loc = c; 594 goto recordleft; 595 } 596 *loc = c; 597 loc--; 598 if (quotes && loc > startp && *(loc - 1) == CTLESC) { 599 for (q = startp; q < loc; q++) 600 if (*q == CTLESC) 601 q++; 602 if (q > loc) 603 loc--; 604 } 605 } 606 return 0; 607 608 case VSTRIMRIGHT: 609 for (loc = str - 1; loc >= startp;) { 610 if (patmatch(str, loc, quotes)) { 611 amount = loc - expdest; 612 STADJUST(amount, expdest); 613 return 1; 614 } 615 loc--; 616 if (quotes && loc > startp && *(loc - 1) == CTLESC) { 617 for (q = startp; q < loc; q++) 618 if (*q == CTLESC) 619 q++; 620 if (q > loc) 621 loc--; 622 } 623 } 624 return 0; 625 626 case VSTRIMRIGHTMAX: 627 for (loc = startp; loc < str - 1; loc++) { 628 if (patmatch(str, loc, quotes)) { 629 amount = loc - expdest; 630 STADJUST(amount, expdest); 631 return 1; 632 } 633 if (quotes && *loc == CTLESC) 634 loc++; 635 } 636 return 0; 637 638 639 default: 640 abort(); 641 } 642 643recordleft: 644 amount = ((str - 1) - (loc - startp)) - expdest; 645 STADJUST(amount, expdest); 646 while (loc != str - 1) 647 *startp++ = *loc++; 648 return 1; 649} 650 651 652/* 653 * Expand a variable, and return a pointer to the next character in the 654 * input string. 655 */ 656 657static char * 658evalvar(char *p, int flag) 659{ 660 int subtype; 661 int varflags; 662 char *var; 663 char *val; 664 int patloc; 665 int c; 666 int set; 667 int special; 668 int startloc; 669 int varlen; 670 int varlenb; 671 int easy; 672 int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR); 673 674 varflags = (unsigned char)*p++; 675 subtype = varflags & VSTYPE; 676 var = p; 677 special = 0; 678 if (! is_name(*p)) 679 special = 1; 680 p = strchr(p, '=') + 1; 681again: /* jump here after setting a variable with ${var=text} */ 682 if (varflags & VSLINENO) { 683 set = 1; 684 special = 0; 685 val = var; 686 p[-1] = '\0'; /* temporarily overwrite '=' to have \0 687 terminated string */ 688 } else if (special) { 689 set = varisset(var, varflags & VSNUL); 690 val = NULL; 691 } else { 692 val = bltinlookup(var, 1); 693 if (val == NULL || ((varflags & VSNUL) && val[0] == '\0')) { 694 val = NULL; 695 set = 0; 696 } else 697 set = 1; 698 } 699 varlen = 0; 700 startloc = expdest - stackblock(); 701 if (!set && uflag && *var != '@' && *var != '*') { 702 switch (subtype) { 703 case VSNORMAL: 704 case VSTRIMLEFT: 705 case VSTRIMLEFTMAX: 706 case VSTRIMRIGHT: 707 case VSTRIMRIGHTMAX: 708 case VSLENGTH: 709 error("%.*s: parameter not set", (int)(p - var - 1), 710 var); 711 } 712 } 713 if (set && subtype != VSPLUS) { 714 /* insert the value of the variable */ 715 if (special) { 716 varvalue(var, varflags & VSQUOTE, subtype, flag); 717 if (subtype == VSLENGTH) { 718 varlenb = expdest - stackblock() - startloc; 719 varlen = varlenb; 720 if (localeisutf8) { 721 val = stackblock() + startloc; 722 for (;val != expdest; val++) 723 if ((*val & 0xC0) == 0x80) 724 varlen--; 725 } 726 STADJUST(-varlenb, expdest); 727 } 728 } else { 729 char const *syntax = (varflags & VSQUOTE) ? DQSYNTAX 730 : BASESYNTAX; 731 732 if (subtype == VSLENGTH) { 733 for (;*val; val++) 734 if (!localeisutf8 || 735 (*val & 0xC0) != 0x80) 736 varlen++; 737 } 738 else { 739 if (quotes) 740 STPUTS_QUOTES(val, syntax, expdest); 741 else 742 STPUTS(val, expdest); 743 744 } 745 } 746 } 747 748 if (subtype == VSPLUS) 749 set = ! set; 750 751 easy = ((varflags & VSQUOTE) == 0 || 752 (*var == '@' && shellparam.nparam != 1)); 753 754 755 switch (subtype) { 756 case VSLENGTH: 757 expdest = cvtnum(varlen, expdest); 758 goto record; 759 760 case VSNORMAL: 761 if (!easy) 762 break; 763record: 764 recordregion(startloc, expdest - stackblock(), 765 varflags & VSQUOTE || (ifsset() && ifsval()[0] == '\0' && 766 (*var == '@' || *var == '*'))); 767 break; 768 769 case VSPLUS: 770 case VSMINUS: 771 if (!set) { 772 argstr(p, flag | (flag & EXP_FULL ? EXP_SPLIT_LIT : 0) | 773 (varflags & VSQUOTE ? EXP_LIT_QUOTED : 0)); 774 break; 775 } 776 if (easy) 777 goto record; 778 break; 779 780 case VSTRIMLEFT: 781 case VSTRIMLEFTMAX: 782 case VSTRIMRIGHT: 783 case VSTRIMRIGHTMAX: 784 if (!set) 785 break; 786 /* 787 * Terminate the string and start recording the pattern 788 * right after it 789 */ 790 STPUTC('\0', expdest); 791 patloc = expdest - stackblock(); 792 if (subevalvar(p, NULL, patloc, subtype, 793 startloc, varflags, quotes) == 0) { 794 int amount = (expdest - stackblock() - patloc) + 1; 795 STADJUST(-amount, expdest); 796 } 797 /* Remove any recorded regions beyond start of variable */ 798 removerecordregions(startloc); 799 goto record; 800 801 case VSASSIGN: 802 case VSQUESTION: 803 if (!set) { 804 if (subevalvar(p, var, 0, subtype, startloc, varflags, 805 quotes)) { 806 varflags &= ~VSNUL; 807 /* 808 * Remove any recorded regions beyond 809 * start of variable 810 */ 811 removerecordregions(startloc); 812 goto again; 813 } 814 break; 815 } 816 if (easy) 817 goto record; 818 break; 819 820 case VSERROR: 821 c = p - var - 1; 822 error("${%.*s%s}: Bad substitution", c, var, 823 (c > 0 && *p != CTLENDVAR) ? "..." : ""); 824 825 default: 826 abort(); 827 } 828 p[-1] = '='; /* recover overwritten '=' */ 829 830 if (subtype != VSNORMAL) { /* skip to end of alternative */ 831 int nesting = 1; 832 for (;;) { 833 if ((c = *p++) == CTLESC) 834 p++; 835 else if (c == CTLBACKQ || c == (CTLBACKQ|CTLQUOTE)) { 836 if (set) 837 argbackq = argbackq->next; 838 } else if (c == CTLVAR) { 839 if ((*p++ & VSTYPE) != VSNORMAL) 840 nesting++; 841 } else if (c == CTLENDVAR) { 842 if (--nesting == 0) 843 break; 844 } 845 } 846 } 847 return p; 848} 849 850 851 852/* 853 * Test whether a specialized variable is set. 854 */ 855 856static int 857varisset(char *name, int nulok) 858{ 859 860 if (*name == '!') 861 return backgndpidset(); 862 else if (*name == '@' || *name == '*') { 863 if (*shellparam.p == NULL) 864 return 0; 865 866 if (nulok) { 867 char **av; 868 869 for (av = shellparam.p; *av; av++) 870 if (**av != '\0') 871 return 1; 872 return 0; 873 } 874 } else if (is_digit(*name)) { 875 char *ap; 876 int num = atoi(name); 877 878 if (num > shellparam.nparam) 879 return 0; 880 881 if (num == 0) 882 ap = arg0; 883 else 884 ap = shellparam.p[num - 1]; 885 886 if (nulok && (ap == NULL || *ap == '\0')) 887 return 0; 888 } 889 return 1; 890} 891 892static void 893strtodest(const char *p, int flag, int subtype, int quoted) 894{ 895 if (flag & (EXP_FULL | EXP_CASE) && subtype != VSLENGTH) 896 STPUTS_QUOTES(p, quoted ? DQSYNTAX : BASESYNTAX, expdest); 897 else 898 STPUTS(p, expdest); 899} 900 901/* 902 * Add the value of a specialized variable to the stack string. 903 */ 904 905static void 906varvalue(char *name, int quoted, int subtype, int flag) 907{ 908 int num; 909 char *p; 910 int i; 911 char sep; 912 char **ap; 913 914 switch (*name) { 915 case '$': 916 num = rootpid; 917 goto numvar; 918 case '?': 919 num = oexitstatus; 920 goto numvar; 921 case '#': 922 num = shellparam.nparam; 923 goto numvar; 924 case '!': 925 num = backgndpidval(); 926numvar: 927 expdest = cvtnum(num, expdest); 928 break; 929 case '-': 930 for (i = 0 ; i < NOPTS ; i++) { 931 if (optlist[i].val) 932 STPUTC(optlist[i].letter, expdest); 933 } 934 break; 935 case '@': 936 if (flag & EXP_FULL && quoted) { 937 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 938 strtodest(p, flag, subtype, quoted); 939 if (*ap) 940 STPUTC('\0', expdest); 941 } 942 break; 943 } 944 /* FALLTHROUGH */ 945 case '*': 946 if (ifsset()) 947 sep = ifsval()[0]; 948 else 949 sep = ' '; 950 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 951 strtodest(p, flag, subtype, quoted); 952 if (!*ap) 953 break; 954 if (sep || (flag & EXP_FULL && !quoted && **ap != '\0')) 955 STPUTC(sep, expdest); 956 } 957 break; 958 case '0': 959 p = arg0; 960 strtodest(p, flag, subtype, quoted); 961 break; 962 default: 963 if (is_digit(*name)) { 964 num = atoi(name); 965 if (num > 0 && num <= shellparam.nparam) { 966 p = shellparam.p[num - 1]; 967 strtodest(p, flag, subtype, quoted); 968 } 969 } 970 break; 971 } 972} 973 974 975 976/* 977 * Record the fact that we have to scan this region of the 978 * string for IFS characters. 979 */ 980 981static void 982recordregion(int start, int end, int inquotes) 983{ 984 struct ifsregion *ifsp; 985 986 if (ifslastp == NULL) { 987 ifsp = &ifsfirst; 988 } else { 989 if (ifslastp->endoff == start 990 && ifslastp->inquotes == inquotes) { 991 /* extend previous area */ 992 ifslastp->endoff = end; 993 return; 994 } 995 ifsp = (struct ifsregion *)ckmalloc(sizeof (struct ifsregion)); 996 ifslastp->next = ifsp; 997 } 998 ifslastp = ifsp; 999 ifslastp->next = NULL; 1000 ifslastp->begoff = start; 1001 ifslastp->endoff = end; 1002 ifslastp->inquotes = inquotes; 1003} 1004 1005 1006 1007/* 1008 * Break the argument string into pieces based upon IFS and add the 1009 * strings to the argument list. The regions of the string to be 1010 * searched for IFS characters have been stored by recordregion. 1011 * CTLESC characters are preserved but have little effect in this pass 1012 * other than escaping CTL* characters. In particular, they do not escape 1013 * IFS characters: that should be done with the ifsregion mechanism. 1014 * CTLQUOTEMARK characters are used to preserve empty quoted strings. 1015 * This pass treats them as a regular character, making the string non-empty. 1016 * Later, they are removed along with the other CTL* characters. 1017 */ 1018static void 1019ifsbreakup(char *string, struct arglist *arglist) 1020{ 1021 struct ifsregion *ifsp; 1022 struct strlist *sp; 1023 char *start; 1024 char *p; 1025 char *q; 1026 const char *ifs; 1027 const char *ifsspc; 1028 int had_param_ch = 0; 1029 1030 start = string; 1031 1032 if (ifslastp == NULL) { 1033 /* Return entire argument, IFS doesn't apply to any of it */ 1034 sp = (struct strlist *)stalloc(sizeof *sp); 1035 sp->text = start; 1036 *arglist->lastp = sp; 1037 arglist->lastp = &sp->next; 1038 return; 1039 } 1040 1041 ifs = ifsset() ? ifsval() : " \t\n"; 1042 1043 for (ifsp = &ifsfirst; ifsp != NULL; ifsp = ifsp->next) { 1044 p = string + ifsp->begoff; 1045 while (p < string + ifsp->endoff) { 1046 q = p; 1047 if (*p == CTLESC) 1048 p++; 1049 if (ifsp->inquotes) { 1050 /* Only NULs (should be from "$@") end args */ 1051 had_param_ch = 1; 1052 if (*p != 0) { 1053 p++; 1054 continue; 1055 } 1056 ifsspc = NULL; 1057 } else { 1058 if (!strchr(ifs, *p)) { 1059 had_param_ch = 1; 1060 p++; 1061 continue; 1062 } 1063 ifsspc = strchr(" \t\n", *p); 1064 1065 /* Ignore IFS whitespace at start */ 1066 if (q == start && ifsspc != NULL) { 1067 p++; 1068 start = p; 1069 continue; 1070 } 1071 had_param_ch = 0; 1072 } 1073 1074 /* Save this argument... */ 1075 *q = '\0'; 1076 sp = (struct strlist *)stalloc(sizeof *sp); 1077 sp->text = start; 1078 *arglist->lastp = sp; 1079 arglist->lastp = &sp->next; 1080 p++; 1081 1082 if (ifsspc != NULL) { 1083 /* Ignore further trailing IFS whitespace */ 1084 for (; p < string + ifsp->endoff; p++) { 1085 q = p; 1086 if (*p == CTLESC) 1087 p++; 1088 if (strchr(ifs, *p) == NULL) { 1089 p = q; 1090 break; 1091 } 1092 if (strchr(" \t\n", *p) == NULL) { 1093 p++; 1094 break; 1095 } 1096 } 1097 } 1098 start = p; 1099 } 1100 } 1101 1102 /* 1103 * Save anything left as an argument. 1104 * Traditionally we have treated 'IFS=':'; set -- x$IFS' as 1105 * generating 2 arguments, the second of which is empty. 1106 * Some recent clarification of the Posix spec say that it 1107 * should only generate one.... 1108 */ 1109 if (had_param_ch || *start != 0) { 1110 sp = (struct strlist *)stalloc(sizeof *sp); 1111 sp->text = start; 1112 *arglist->lastp = sp; 1113 arglist->lastp = &sp->next; 1114 } 1115} 1116 1117 1118static char expdir[PATH_MAX]; 1119#define expdir_end (expdir + sizeof(expdir)) 1120 1121/* 1122 * Perform pathname generation and remove control characters. 1123 * At this point, the only control characters should be CTLESC and CTLQUOTEMARK. 1124 * The results are stored in the list exparg. 1125 */ 1126static void 1127expandmeta(struct strlist *str, int flag __unused) 1128{ 1129 char *p; 1130 struct strlist **savelastp; 1131 struct strlist *sp; 1132 char c; 1133 /* TODO - EXP_REDIR */ 1134 1135 while (str) { 1136 if (fflag) 1137 goto nometa; 1138 p = str->text; 1139 for (;;) { /* fast check for meta chars */ 1140 if ((c = *p++) == '\0') 1141 goto nometa; 1142 if (c == '*' || c == '?' || c == '[') 1143 break; 1144 } 1145 savelastp = exparg.lastp; 1146 INTOFF; 1147 expmeta(expdir, str->text); 1148 INTON; 1149 if (exparg.lastp == savelastp) { 1150 /* 1151 * no matches 1152 */ 1153nometa: 1154 *exparg.lastp = str; 1155 rmescapes(str->text); 1156 exparg.lastp = &str->next; 1157 } else { 1158 *exparg.lastp = NULL; 1159 *savelastp = sp = expsort(*savelastp); 1160 while (sp->next != NULL) 1161 sp = sp->next; 1162 exparg.lastp = &sp->next; 1163 } 1164 str = str->next; 1165 } 1166} 1167 1168 1169/* 1170 * Do metacharacter (i.e. *, ?, [...]) expansion. 1171 */ 1172 1173static void 1174expmeta(char *enddir, char *name) 1175{ 1176 char *p; 1177 char *q; 1178 char *start; 1179 char *endname; 1180 int metaflag; 1181 struct stat statb; 1182 DIR *dirp; 1183 struct dirent *dp; 1184 int atend; 1185 int matchdot; 1186 int esc; 1187 1188 metaflag = 0; 1189 start = name; 1190 for (p = name; esc = 0, *p; p += esc + 1) { 1191 if (*p == '*' || *p == '?') 1192 metaflag = 1; 1193 else if (*p == '[') { 1194 q = p + 1; 1195 if (*q == '!' || *q == '^') 1196 q++; 1197 for (;;) { 1198 while (*q == CTLQUOTEMARK) 1199 q++; 1200 if (*q == CTLESC) 1201 q++; 1202 if (*q == '/' || *q == '\0') 1203 break; 1204 if (*++q == ']') { 1205 metaflag = 1; 1206 break; 1207 } 1208 } 1209 } else if (*p == '\0') 1210 break; 1211 else if (*p == CTLQUOTEMARK) 1212 continue; 1213 else { 1214 if (*p == CTLESC) 1215 esc++; 1216 if (p[esc] == '/') { 1217 if (metaflag) 1218 break; 1219 start = p + esc + 1; 1220 } 1221 } 1222 } 1223 if (metaflag == 0) { /* we've reached the end of the file name */ 1224 if (enddir != expdir) 1225 metaflag++; 1226 for (p = name ; ; p++) { 1227 if (*p == CTLQUOTEMARK) 1228 continue; 1229 if (*p == CTLESC) 1230 p++; 1231 *enddir++ = *p; 1232 if (*p == '\0') 1233 break; 1234 if (enddir == expdir_end) 1235 return; 1236 } 1237 if (metaflag == 0 || lstat(expdir, &statb) >= 0) 1238 addfname(expdir); 1239 return; 1240 } 1241 endname = p; 1242 if (start != name) { 1243 p = name; 1244 while (p < start) { 1245 while (*p == CTLQUOTEMARK) 1246 p++; 1247 if (*p == CTLESC) 1248 p++; 1249 *enddir++ = *p++; 1250 if (enddir == expdir_end) 1251 return; 1252 } 1253 } 1254 if (enddir == expdir) { 1255 p = "."; 1256 } else if (enddir == expdir + 1 && *expdir == '/') { 1257 p = "/"; 1258 } else { 1259 p = expdir; 1260 enddir[-1] = '\0'; 1261 } 1262 if ((dirp = opendir(p)) == NULL) 1263 return; 1264 if (enddir != expdir) 1265 enddir[-1] = '/'; 1266 if (*endname == 0) { 1267 atend = 1; 1268 } else { 1269 atend = 0; 1270 *endname = '\0'; 1271 endname += esc + 1; 1272 } 1273 matchdot = 0; 1274 p = start; 1275 while (*p == CTLQUOTEMARK) 1276 p++; 1277 if (*p == CTLESC) 1278 p++; 1279 if (*p == '.') 1280 matchdot++; 1281 while (! int_pending() && (dp = readdir(dirp)) != NULL) { 1282 if (dp->d_name[0] == '.' && ! matchdot) 1283 continue; 1284 if (patmatch(start, dp->d_name, 0)) { 1285 if (enddir + dp->d_namlen + 1 > expdir_end) 1286 continue; 1287 memcpy(enddir, dp->d_name, dp->d_namlen + 1); 1288 if (atend) 1289 addfname(expdir); 1290 else { 1291 if (enddir + dp->d_namlen + 2 > expdir_end) 1292 continue; 1293 enddir[dp->d_namlen] = '/'; 1294 enddir[dp->d_namlen + 1] = '\0'; 1295 expmeta(enddir + dp->d_namlen + 1, endname); 1296 } 1297 } 1298 } 1299 closedir(dirp); 1300 if (! atend) 1301 endname[-esc - 1] = esc ? CTLESC : '/'; 1302} 1303 1304 1305/* 1306 * Add a file name to the list. 1307 */ 1308 1309static void 1310addfname(char *name) 1311{ 1312 char *p; 1313 struct strlist *sp; 1314 1315 p = stalloc(strlen(name) + 1); 1316 scopy(name, p); 1317 sp = (struct strlist *)stalloc(sizeof *sp); 1318 sp->text = p; 1319 *exparg.lastp = sp; 1320 exparg.lastp = &sp->next; 1321} 1322 1323 1324/* 1325 * Sort the results of file name expansion. It calculates the number of 1326 * strings to sort and then calls msort (short for merge sort) to do the 1327 * work. 1328 */ 1329 1330static struct strlist * 1331expsort(struct strlist *str) 1332{ 1333 int len; 1334 struct strlist *sp; 1335 1336 len = 0; 1337 for (sp = str ; sp ; sp = sp->next) 1338 len++; 1339 return msort(str, len); 1340} 1341 1342 1343static struct strlist * 1344msort(struct strlist *list, int len) 1345{ 1346 struct strlist *p, *q = NULL; 1347 struct strlist **lpp; 1348 int half; 1349 int n; 1350 1351 if (len <= 1) 1352 return list; 1353 half = len >> 1; 1354 p = list; 1355 for (n = half ; --n >= 0 ; ) { 1356 q = p; 1357 p = p->next; 1358 } 1359 q->next = NULL; /* terminate first half of list */ 1360 q = msort(list, half); /* sort first half of list */ 1361 p = msort(p, len - half); /* sort second half */ 1362 lpp = &list; 1363 for (;;) { 1364 if (strcmp(p->text, q->text) < 0) { 1365 *lpp = p; 1366 lpp = &p->next; 1367 if ((p = *lpp) == NULL) { 1368 *lpp = q; 1369 break; 1370 } 1371 } else { 1372 *lpp = q; 1373 lpp = &q->next; 1374 if ((q = *lpp) == NULL) { 1375 *lpp = p; 1376 break; 1377 } 1378 } 1379 } 1380 return list; 1381} 1382 1383 1384 1385static wchar_t 1386get_wc(const char **p) 1387{ 1388 wchar_t c; 1389 int chrlen; 1390 1391 chrlen = mbtowc(&c, *p, 4); 1392 if (chrlen == 0) 1393 return 0; 1394 else if (chrlen == -1) 1395 c = 0; 1396 else 1397 *p += chrlen; 1398 return c; 1399} 1400 1401 1402/* 1403 * Returns true if the pattern matches the string. 1404 */ 1405 1406int 1407patmatch(const char *pattern, const char *string, int squoted) 1408{ 1409 const char *p, *q; 1410 char c; 1411 wchar_t wc, wc2; 1412 1413 p = pattern; 1414 q = string; 1415 for (;;) { 1416 switch (c = *p++) { 1417 case '\0': 1418 goto breakloop; 1419 case CTLESC: 1420 if (squoted && *q == CTLESC) 1421 q++; 1422 if (*q++ != *p++) 1423 return 0; 1424 break; 1425 case CTLQUOTEMARK: 1426 continue; 1427 case '?': 1428 if (squoted && *q == CTLESC) 1429 q++; 1430 if (localeisutf8) 1431 wc = get_wc(&q); 1432 else 1433 wc = (unsigned char)*q++; 1434 if (wc == '\0') 1435 return 0; 1436 break; 1437 case '*': 1438 c = *p; 1439 while (c == CTLQUOTEMARK || c == '*') 1440 c = *++p; 1441 if (c != CTLESC && c != CTLQUOTEMARK && 1442 c != '?' && c != '*' && c != '[') { 1443 while (*q != c) { 1444 if (squoted && *q == CTLESC && 1445 q[1] == c) 1446 break; 1447 if (*q == '\0') 1448 return 0; 1449 if (squoted && *q == CTLESC) 1450 q++; 1451 q++; 1452 } 1453 } 1454 do { 1455 if (patmatch(p, q, squoted)) 1456 return 1; 1457 if (squoted && *q == CTLESC) 1458 q++; 1459 } while (*q++ != '\0'); 1460 return 0; 1461 case '[': { 1462 const char *endp; 1463 int invert, found; 1464 wchar_t chr; 1465 1466 endp = p; 1467 if (*endp == '!' || *endp == '^') 1468 endp++; 1469 for (;;) { 1470 while (*endp == CTLQUOTEMARK) 1471 endp++; 1472 if (*endp == '\0') 1473 goto dft; /* no matching ] */ 1474 if (*endp == CTLESC) 1475 endp++; 1476 if (*++endp == ']') 1477 break; 1478 } 1479 invert = 0; 1480 if (*p == '!' || *p == '^') { 1481 invert++; 1482 p++; 1483 } 1484 found = 0; 1485 if (squoted && *q == CTLESC) 1486 q++; 1487 if (localeisutf8) 1488 chr = get_wc(&q); 1489 else 1490 chr = (unsigned char)*q++; 1491 if (chr == '\0') 1492 return 0; 1493 c = *p++; 1494 do { 1495 if (c == CTLQUOTEMARK) 1496 continue; 1497 if (c == CTLESC) 1498 c = *p++; 1499 if (localeisutf8 && c & 0x80) { 1500 p--; 1501 wc = get_wc(&p); 1502 if (wc == 0) /* bad utf-8 */ 1503 return 0; 1504 } else 1505 wc = (unsigned char)c; 1506 if (*p == '-' && p[1] != ']') { 1507 p++; 1508 while (*p == CTLQUOTEMARK) 1509 p++; 1510 if (*p == CTLESC) 1511 p++; 1512 if (localeisutf8) { 1513 wc2 = get_wc(&p); 1514 if (wc2 == 0) /* bad utf-8 */ 1515 return 0; 1516 } else 1517 wc2 = (unsigned char)*p++; 1518 if ( collate_range_cmp(chr, wc) >= 0 1519 && collate_range_cmp(chr, wc2) <= 0 1520 ) 1521 found = 1; 1522 } else { 1523 if (chr == wc) 1524 found = 1; 1525 } 1526 } while ((c = *p++) != ']'); 1527 if (found == invert) 1528 return 0; 1529 break; 1530 } 1531dft: default: 1532 if (squoted && *q == CTLESC) 1533 q++; 1534 if (*q++ != c) 1535 return 0; 1536 break; 1537 } 1538 } 1539breakloop: 1540 if (*q != '\0') 1541 return 0; 1542 return 1; 1543} 1544 1545 1546 1547/* 1548 * Remove any CTLESC and CTLQUOTEMARK characters from a string. 1549 */ 1550 1551void 1552rmescapes(char *str) 1553{ 1554 char *p, *q; 1555 1556 p = str; 1557 while (*p != CTLESC && *p != CTLQUOTEMARK && *p != CTLQUOTEEND) { 1558 if (*p++ == '\0') 1559 return; 1560 } 1561 q = p; 1562 while (*p) { 1563 if (*p == CTLQUOTEMARK || *p == CTLQUOTEEND) { 1564 p++; 1565 continue; 1566 } 1567 if (*p == CTLESC) 1568 p++; 1569 *q++ = *p++; 1570 } 1571 *q = '\0'; 1572} 1573 1574 1575 1576/* 1577 * See if a pattern matches in a case statement. 1578 */ 1579 1580int 1581casematch(union node *pattern, const char *val) 1582{ 1583 struct stackmark smark; 1584 int result; 1585 char *p; 1586 1587 setstackmark(&smark); 1588 argbackq = pattern->narg.backquote; 1589 STARTSTACKSTR(expdest); 1590 ifslastp = NULL; 1591 argstr(pattern->narg.text, EXP_TILDE | EXP_CASE); 1592 STPUTC('\0', expdest); 1593 p = grabstackstr(expdest); 1594 result = patmatch(p, val, 0); 1595 popstackmark(&smark); 1596 return result; 1597} 1598 1599/* 1600 * Our own itoa(). 1601 */ 1602 1603static char * 1604cvtnum(int num, char *buf) 1605{ 1606 char temp[32]; 1607 int neg = num < 0; 1608 char *p = temp + 31; 1609 1610 temp[31] = '\0'; 1611 1612 do { 1613 *--p = num % 10 + '0'; 1614 } while ((num /= 10) != 0); 1615 1616 if (neg) 1617 *--p = '-'; 1618 1619 STPUTS(p, buf); 1620 return buf; 1621} 1622 1623/* 1624 * Do most of the work for wordexp(3). 1625 */ 1626 1627int 1628wordexpcmd(int argc, char **argv) 1629{ 1630 size_t len; 1631 int i; 1632 1633 out1fmt("%08x", argc - 1); 1634 for (i = 1, len = 0; i < argc; i++) 1635 len += strlen(argv[i]); 1636 out1fmt("%08x", (int)len); 1637 for (i = 1; i < argc; i++) 1638 outbin(argv[i], strlen(argv[i]) + 1, out1); 1639 return (0); 1640} 1641