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 * Copyright (c) 2010-2015 7 * Jilles Tjoelker <jilles@stack.nl>. All rights reserved. 8 * 9 * This code is derived from software contributed to Berkeley by 10 * Kenneth Almquist. 11 * 12 * Redistribution and use in source and binary forms, with or without 13 * modification, are permitted provided that the following conditions 14 * are met: 15 * 1. Redistributions of source code must retain the above copyright 16 * notice, this list of conditions and the following disclaimer. 17 * 2. Redistributions in binary form must reproduce the above copyright 18 * notice, this list of conditions and the following disclaimer in the 19 * documentation and/or other materials provided with the distribution. 20 * 4. Neither the name of the University nor the names of its contributors 21 * may be used to endorse or promote products derived from this software 22 * without specific prior written permission. 23 * 24 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 25 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 26 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 27 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 28 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 29 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 30 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 31 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 32 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 33 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 34 * SUCH DAMAGE. 35 */ 36 37 #ifndef lint 38 #if 0 39 static char sccsid[] = "@(#)expand.c 8.5 (Berkeley) 5/15/95"; 40 #endif 41 #endif /* not lint */ 42 #include <sys/cdefs.h> 43 __FBSDID("$FreeBSD$"); 44 45 #include <sys/types.h> 46 #include <sys/time.h> 47 #include <sys/stat.h> 48 #include <dirent.h> 49 #include <errno.h> 50 #include <inttypes.h> 51 #include <limits.h> 52 #include <pwd.h> 53 #include <stdio.h> 54 #include <stdlib.h> 55 #include <string.h> 56 #include <unistd.h> 57 #include <wchar.h> 58 #include <wctype.h> 59 60 /* 61 * Routines to expand arguments to commands. We have to deal with 62 * backquotes, shell variables, and file metacharacters. 63 */ 64 65 #include "shell.h" 66 #include "main.h" 67 #include "nodes.h" 68 #include "eval.h" 69 #include "expand.h" 70 #include "syntax.h" 71 #include "parser.h" 72 #include "jobs.h" 73 #include "options.h" 74 #include "var.h" 75 #include "input.h" 76 #include "output.h" 77 #include "memalloc.h" 78 #include "error.h" 79 #include "mystring.h" 80 #include "arith.h" 81 #include "show.h" 82 #include "builtins.h" 83 84 enum wordstate { WORD_IDLE, WORD_WS_DELIMITED, WORD_QUOTEMARK }; 85 86 struct worddest { 87 struct arglist *list; 88 enum wordstate state; 89 }; 90 91 static char *expdest; /* output of current string */ 92 static struct nodelist *argbackq; /* list of back quote expressions */ 93 94 static char *argstr(char *, int, struct worddest *); 95 static char *exptilde(char *, int); 96 static char *expari(char *, int, struct worddest *); 97 static void expbackq(union node *, int, int, struct worddest *); 98 static void subevalvar_trim(char *, int, int, int); 99 static int subevalvar_misc(char *, const char *, int, int, int); 100 static char *evalvar(char *, int, struct worddest *); 101 static int varisset(const char *, int); 102 static void strtodest(const char *, int, int, int, struct worddest *); 103 static void reprocess(int, int, int, int, struct worddest *); 104 static void varvalue(const char *, int, int, int, struct worddest *); 105 static void expandmeta(char *, struct arglist *); 106 static void expmeta(char *, char *, struct arglist *); 107 static int expsortcmp(const void *, const void *); 108 static int patmatch(const char *, const char *); 109 static void cvtnum(int, char *); 110 static int collate_range_cmp(wchar_t, wchar_t); 111 112 void 113 emptyarglist(struct arglist *list) 114 { 115 116 list->args = list->smallarg; 117 list->count = 0; 118 list->capacity = sizeof(list->smallarg) / sizeof(list->smallarg[0]); 119 } 120 121 void 122 appendarglist(struct arglist *list, char *str) 123 { 124 char **newargs; 125 int newcapacity; 126 127 if (list->count >= list->capacity) { 128 newcapacity = list->capacity * 2; 129 if (newcapacity < 16) 130 newcapacity = 16; 131 if (newcapacity > INT_MAX / (int)sizeof(newargs[0])) 132 error("Too many entries in arglist"); 133 newargs = stalloc(newcapacity * sizeof(newargs[0])); 134 memcpy(newargs, list->args, list->count * sizeof(newargs[0])); 135 list->args = newargs; 136 list->capacity = newcapacity; 137 } 138 list->args[list->count++] = str; 139 } 140 141 static int 142 collate_range_cmp(wchar_t c1, wchar_t c2) 143 { 144 static wchar_t s1[2], s2[2]; 145 146 s1[0] = c1; 147 s2[0] = c2; 148 return (wcscoll(s1, s2)); 149 } 150 151 static char * 152 stputs_quotes(const char *data, const char *syntax, char *p) 153 { 154 while (*data) { 155 CHECKSTRSPACE(2, p); 156 if (syntax[(int)*data] == CCTL) 157 USTPUTC(CTLESC, p); 158 USTPUTC(*data++, p); 159 } 160 return (p); 161 } 162 #define STPUTS_QUOTES(data, syntax, p) p = stputs_quotes((data), syntax, p) 163 164 static char * 165 nextword(char c, int flag, char *p, struct worddest *dst) 166 { 167 int is_ws; 168 169 is_ws = c == '\t' || c == '\n' || c == ' '; 170 if (p != stackblock() || (is_ws ? dst->state == WORD_QUOTEMARK : 171 dst->state != WORD_WS_DELIMITED) || c == '\0') { 172 STPUTC('\0', p); 173 if (flag & EXP_GLOB) 174 expandmeta(grabstackstr(p), dst->list); 175 else 176 appendarglist(dst->list, grabstackstr(p)); 177 dst->state = is_ws ? WORD_WS_DELIMITED : WORD_IDLE; 178 } else if (!is_ws && dst->state == WORD_WS_DELIMITED) 179 dst->state = WORD_IDLE; 180 /* Reserve space while the stack string is empty. */ 181 appendarglist(dst->list, NULL); 182 dst->list->count--; 183 STARTSTACKSTR(p); 184 return p; 185 } 186 #define NEXTWORD(c, flag, p, dstlist) p = nextword(c, flag, p, dstlist) 187 188 static char * 189 stputs_split(const char *data, const char *syntax, int flag, char *p, 190 struct worddest *dst) 191 { 192 const char *ifs; 193 char c; 194 195 ifs = ifsset() ? ifsval() : " \t\n"; 196 while (*data) { 197 CHECKSTRSPACE(2, p); 198 c = *data++; 199 if (strchr(ifs, c) != NULL) { 200 NEXTWORD(c, flag, p, dst); 201 continue; 202 } 203 if (flag & EXP_GLOB && syntax[(int)c] == CCTL) 204 USTPUTC(CTLESC, p); 205 USTPUTC(c, p); 206 } 207 return (p); 208 } 209 #define STPUTS_SPLIT(data, syntax, flag, p, dst) p = stputs_split((data), syntax, flag, p, dst) 210 211 /* 212 * Perform expansions on an argument, placing the resulting list of arguments 213 * in arglist. Parameter expansion, command substitution and arithmetic 214 * expansion are always performed; additional expansions can be requested 215 * via flag (EXP_*). 216 * The result is left in the stack string. 217 * When arglist is NULL, perform here document expansion. 218 * 219 * Caution: this function uses global state and is not reentrant. 220 * However, a new invocation after an interrupted invocation is safe 221 * and will reset the global state for the new call. 222 */ 223 void 224 expandarg(union node *arg, struct arglist *arglist, int flag) 225 { 226 struct worddest exparg; 227 228 if (fflag) 229 flag &= ~EXP_GLOB; 230 argbackq = arg->narg.backquote; 231 exparg.list = arglist; 232 exparg.state = WORD_IDLE; 233 STARTSTACKSTR(expdest); 234 argstr(arg->narg.text, flag, &exparg); 235 if (arglist == NULL) { 236 STACKSTRNUL(expdest); 237 return; /* here document expanded */ 238 } 239 if ((flag & EXP_SPLIT) == 0 || expdest != stackblock() || 240 exparg.state == WORD_QUOTEMARK) { 241 STPUTC('\0', expdest); 242 if (flag & EXP_SPLIT) { 243 if (flag & EXP_GLOB) 244 expandmeta(grabstackstr(expdest), exparg.list); 245 else 246 appendarglist(exparg.list, grabstackstr(expdest)); 247 } 248 } 249 if ((flag & EXP_SPLIT) == 0) 250 appendarglist(arglist, grabstackstr(expdest)); 251 } 252 253 254 255 /* 256 * Perform parameter expansion, command substitution and arithmetic 257 * expansion, and tilde expansion if requested via EXP_TILDE/EXP_VARTILDE. 258 * Processing ends at a CTLENDVAR or CTLENDARI character as well as '\0'. 259 * This is used to expand word in ${var+word} etc. 260 * If EXP_GLOB or EXP_CASE are set, keep and/or generate CTLESC 261 * characters to allow for further processing. 262 * 263 * If EXP_SPLIT is set, dst receives any complete words produced. 264 */ 265 static char * 266 argstr(char *p, int flag, struct worddest *dst) 267 { 268 char c; 269 int quotes = flag & (EXP_GLOB | EXP_CASE); /* do CTLESC */ 270 int firsteq = 1; 271 int split_lit; 272 int lit_quoted; 273 274 split_lit = flag & EXP_SPLIT_LIT; 275 lit_quoted = flag & EXP_LIT_QUOTED; 276 flag &= ~(EXP_SPLIT_LIT | EXP_LIT_QUOTED); 277 if (*p == '~' && (flag & (EXP_TILDE | EXP_VARTILDE))) 278 p = exptilde(p, flag); 279 for (;;) { 280 CHECKSTRSPACE(2, expdest); 281 switch (c = *p++) { 282 case '\0': 283 return (p - 1); 284 case CTLENDVAR: 285 case CTLENDARI: 286 return (p); 287 case CTLQUOTEMARK: 288 lit_quoted = 1; 289 /* "$@" syntax adherence hack */ 290 if (p[0] == CTLVAR && (p[1] & VSQUOTE) != 0 && 291 p[2] == '@' && p[3] == '=') 292 break; 293 if ((flag & EXP_SPLIT) != 0 && expdest == stackblock()) 294 dst->state = WORD_QUOTEMARK; 295 break; 296 case CTLQUOTEEND: 297 lit_quoted = 0; 298 break; 299 case CTLESC: 300 c = *p++; 301 if (split_lit && !lit_quoted && 302 strchr(ifsset() ? ifsval() : " \t\n", c) != NULL) { 303 NEXTWORD(c, flag, expdest, dst); 304 break; 305 } 306 if (quotes) 307 USTPUTC(CTLESC, expdest); 308 USTPUTC(c, expdest); 309 break; 310 case CTLVAR: 311 p = evalvar(p, flag, dst); 312 break; 313 case CTLBACKQ: 314 case CTLBACKQ|CTLQUOTE: 315 expbackq(argbackq->n, c & CTLQUOTE, flag, dst); 316 argbackq = argbackq->next; 317 break; 318 case CTLARI: 319 p = expari(p, flag, dst); 320 break; 321 case ':': 322 case '=': 323 /* 324 * sort of a hack - expand tildes in variable 325 * assignments (after the first '=' and after ':'s). 326 */ 327 if (split_lit && !lit_quoted && 328 strchr(ifsset() ? ifsval() : " \t\n", c) != NULL) { 329 NEXTWORD(c, flag, expdest, dst); 330 break; 331 } 332 USTPUTC(c, expdest); 333 if (flag & EXP_VARTILDE && *p == '~' && 334 (c != '=' || firsteq)) { 335 if (c == '=') 336 firsteq = 0; 337 p = exptilde(p, flag); 338 } 339 break; 340 default: 341 if (split_lit && !lit_quoted && 342 strchr(ifsset() ? ifsval() : " \t\n", c) != NULL) { 343 NEXTWORD(c, flag, expdest, dst); 344 break; 345 } 346 USTPUTC(c, expdest); 347 } 348 } 349 } 350 351 /* 352 * Perform tilde expansion, placing the result in the stack string and 353 * returning the next position in the input string to process. 354 */ 355 static char * 356 exptilde(char *p, int flag) 357 { 358 char c, *startp = p; 359 struct passwd *pw; 360 char *home; 361 362 for (;;) { 363 c = *p; 364 switch(c) { 365 case CTLESC: /* This means CTL* are always considered quoted. */ 366 case CTLVAR: 367 case CTLBACKQ: 368 case CTLBACKQ | CTLQUOTE: 369 case CTLARI: 370 case CTLENDARI: 371 case CTLQUOTEMARK: 372 return (startp); 373 case ':': 374 if ((flag & EXP_VARTILDE) == 0) 375 break; 376 /* FALLTHROUGH */ 377 case '\0': 378 case '/': 379 case CTLENDVAR: 380 *p = '\0'; 381 if (*(startp+1) == '\0') { 382 home = lookupvar("HOME"); 383 } else { 384 pw = getpwnam(startp+1); 385 home = pw != NULL ? pw->pw_dir : NULL; 386 } 387 *p = c; 388 if (home == NULL || *home == '\0') 389 return (startp); 390 strtodest(home, flag, VSNORMAL, 1, NULL); 391 return (p); 392 } 393 p++; 394 } 395 } 396 397 398 /* 399 * Expand arithmetic expression. 400 */ 401 static char * 402 expari(char *p, int flag, struct worddest *dst) 403 { 404 char *q, *start; 405 arith_t result; 406 int begoff; 407 int quoted; 408 int adj; 409 410 quoted = *p++ == '"'; 411 begoff = expdest - stackblock(); 412 p = argstr(p, 0, NULL); 413 STPUTC('\0', expdest); 414 start = stackblock() + begoff; 415 416 q = grabstackstr(expdest); 417 result = arith(start); 418 ungrabstackstr(q, expdest); 419 420 start = stackblock() + begoff; 421 adj = start - expdest; 422 STADJUST(adj, expdest); 423 424 CHECKSTRSPACE((int)(DIGITS(result) + 1), expdest); 425 fmtstr(expdest, DIGITS(result), ARITH_FORMAT_STR, result); 426 adj = strlen(expdest); 427 STADJUST(adj, expdest); 428 if (!quoted) 429 reprocess(expdest - adj - stackblock(), flag, VSNORMAL, 0, dst); 430 return p; 431 } 432 433 434 /* 435 * Perform command substitution. 436 */ 437 static void 438 expbackq(union node *cmd, int quoted, int flag, struct worddest *dst) 439 { 440 struct backcmd in; 441 int i; 442 char buf[128]; 443 char *p; 444 char *dest = expdest; 445 struct nodelist *saveargbackq; 446 char lastc; 447 char const *syntax = quoted? DQSYNTAX : BASESYNTAX; 448 int quotes = flag & (EXP_GLOB | EXP_CASE); 449 size_t nnl; 450 const char *ifs; 451 452 INTOFF; 453 saveargbackq = argbackq; 454 p = grabstackstr(dest); 455 evalbackcmd(cmd, &in); 456 ungrabstackstr(p, dest); 457 argbackq = saveargbackq; 458 459 p = in.buf; 460 lastc = '\0'; 461 nnl = 0; 462 if (!quoted && flag & EXP_SPLIT) 463 ifs = ifsset() ? ifsval() : " \t\n"; 464 else 465 ifs = ""; 466 /* Don't copy trailing newlines */ 467 for (;;) { 468 if (--in.nleft < 0) { 469 if (in.fd < 0) 470 break; 471 while ((i = read(in.fd, buf, sizeof buf)) < 0 && errno == EINTR); 472 TRACE(("expbackq: read returns %d\n", i)); 473 if (i <= 0) 474 break; 475 p = buf; 476 in.nleft = i - 1; 477 } 478 lastc = *p++; 479 if (lastc == '\0') 480 continue; 481 if (lastc == '\n') { 482 nnl++; 483 } else { 484 if (nnl > 0) { 485 if (strchr(ifs, '\n') != NULL) { 486 NEXTWORD('\n', flag, dest, dst); 487 nnl = 0; 488 } else { 489 CHECKSTRSPACE(nnl + 2, dest); 490 while (nnl > 0) { 491 nnl--; 492 USTPUTC('\n', dest); 493 } 494 } 495 } 496 if (strchr(ifs, lastc) != NULL) 497 NEXTWORD(lastc, flag, dest, dst); 498 else { 499 CHECKSTRSPACE(2, dest); 500 if (quotes && syntax[(int)lastc] == CCTL) 501 USTPUTC(CTLESC, dest); 502 USTPUTC(lastc, dest); 503 } 504 } 505 } 506 507 if (in.fd >= 0) 508 close(in.fd); 509 if (in.buf) 510 ckfree(in.buf); 511 if (in.jp) 512 exitstatus = waitforjob(in.jp, (int *)NULL); 513 TRACE(("expbackq: size=%td: \"%.*s\"\n", 514 ((dest - stackblock()) - startloc), 515 (int)((dest - stackblock()) - startloc), 516 stackblock() + startloc)); 517 expdest = dest; 518 INTON; 519 } 520 521 522 523 static void 524 recordleft(const char *str, const char *loc, char *startp) 525 { 526 int amount; 527 528 amount = ((str - 1) - (loc - startp)) - expdest; 529 STADJUST(amount, expdest); 530 while (loc != str - 1) 531 *startp++ = *loc++; 532 } 533 534 static void 535 subevalvar_trim(char *p, int strloc, int subtype, int startloc) 536 { 537 char *startp; 538 char *loc = NULL; 539 char *str; 540 int c = 0; 541 struct nodelist *saveargbackq = argbackq; 542 int amount; 543 544 argstr(p, EXP_CASE | EXP_TILDE, NULL); 545 STACKSTRNUL(expdest); 546 argbackq = saveargbackq; 547 startp = stackblock() + startloc; 548 str = stackblock() + strloc; 549 550 switch (subtype) { 551 case VSTRIMLEFT: 552 for (loc = startp; loc < str; loc++) { 553 c = *loc; 554 *loc = '\0'; 555 if (patmatch(str, startp)) { 556 *loc = c; 557 recordleft(str, loc, startp); 558 return; 559 } 560 *loc = c; 561 } 562 break; 563 564 case VSTRIMLEFTMAX: 565 for (loc = str - 1; loc >= startp;) { 566 c = *loc; 567 *loc = '\0'; 568 if (patmatch(str, startp)) { 569 *loc = c; 570 recordleft(str, loc, startp); 571 return; 572 } 573 *loc = c; 574 loc--; 575 } 576 break; 577 578 case VSTRIMRIGHT: 579 for (loc = str - 1; loc >= startp;) { 580 if (patmatch(str, loc)) { 581 amount = loc - expdest; 582 STADJUST(amount, expdest); 583 return; 584 } 585 loc--; 586 } 587 break; 588 589 case VSTRIMRIGHTMAX: 590 for (loc = startp; loc < str - 1; loc++) { 591 if (patmatch(str, loc)) { 592 amount = loc - expdest; 593 STADJUST(amount, expdest); 594 return; 595 } 596 } 597 break; 598 599 600 default: 601 abort(); 602 } 603 amount = (expdest - stackblock() - strloc) + 1; 604 STADJUST(-amount, expdest); 605 } 606 607 608 static int 609 subevalvar_misc(char *p, const char *var, int subtype, int startloc, 610 int varflags) 611 { 612 char *startp; 613 struct nodelist *saveargbackq = argbackq; 614 int amount; 615 616 argstr(p, EXP_TILDE, NULL); 617 STACKSTRNUL(expdest); 618 argbackq = saveargbackq; 619 startp = stackblock() + startloc; 620 621 switch (subtype) { 622 case VSASSIGN: 623 setvar(var, startp, 0); 624 amount = startp - expdest; 625 STADJUST(amount, expdest); 626 return 1; 627 628 case VSQUESTION: 629 if (*p != CTLENDVAR) { 630 outfmt(out2, "%s\n", startp); 631 error((char *)NULL); 632 } 633 error("%.*s: parameter %snot set", (int)(p - var - 1), 634 var, (varflags & VSNUL) ? "null or " : ""); 635 return 0; 636 637 default: 638 abort(); 639 } 640 } 641 642 643 /* 644 * Expand a variable, and return a pointer to the next character in the 645 * input string. 646 */ 647 648 static char * 649 evalvar(char *p, int flag, struct worddest *dst) 650 { 651 int subtype; 652 int varflags; 653 char *var; 654 const char *val; 655 int patloc; 656 int c; 657 int set; 658 int special; 659 int startloc; 660 int varlen; 661 int varlenb; 662 char buf[21]; 663 664 varflags = (unsigned char)*p++; 665 subtype = varflags & VSTYPE; 666 var = p; 667 special = 0; 668 if (! is_name(*p)) 669 special = 1; 670 p = strchr(p, '=') + 1; 671 again: /* jump here after setting a variable with ${var=text} */ 672 if (varflags & VSLINENO) { 673 set = 1; 674 special = 1; 675 val = NULL; 676 } else if (special) { 677 set = varisset(var, varflags & VSNUL); 678 val = NULL; 679 } else { 680 val = bltinlookup(var, 1); 681 if (val == NULL || ((varflags & VSNUL) && val[0] == '\0')) { 682 val = NULL; 683 set = 0; 684 } else 685 set = 1; 686 } 687 varlen = 0; 688 startloc = expdest - stackblock(); 689 if (!set && uflag && *var != '@' && *var != '*') { 690 switch (subtype) { 691 case VSNORMAL: 692 case VSTRIMLEFT: 693 case VSTRIMLEFTMAX: 694 case VSTRIMRIGHT: 695 case VSTRIMRIGHTMAX: 696 case VSLENGTH: 697 error("%.*s: parameter not set", (int)(p - var - 1), 698 var); 699 } 700 } 701 if (set && subtype != VSPLUS) { 702 /* insert the value of the variable */ 703 if (special) { 704 if (varflags & VSLINENO) { 705 if (p - var > (ptrdiff_t)sizeof(buf)) 706 abort(); 707 memcpy(buf, var, p - var - 1); 708 buf[p - var - 1] = '\0'; 709 strtodest(buf, flag, subtype, 710 varflags & VSQUOTE, dst); 711 } else 712 varvalue(var, varflags & VSQUOTE, subtype, flag, 713 dst); 714 if (subtype == VSLENGTH) { 715 varlenb = expdest - stackblock() - startloc; 716 varlen = varlenb; 717 if (localeisutf8) { 718 val = stackblock() + startloc; 719 for (;val != expdest; val++) 720 if ((*val & 0xC0) == 0x80) 721 varlen--; 722 } 723 STADJUST(-varlenb, expdest); 724 } 725 } else { 726 if (subtype == VSLENGTH) { 727 for (;*val; val++) 728 if (!localeisutf8 || 729 (*val & 0xC0) != 0x80) 730 varlen++; 731 } 732 else 733 strtodest(val, flag, subtype, 734 varflags & VSQUOTE, dst); 735 } 736 } 737 738 if (subtype == VSPLUS) 739 set = ! set; 740 741 switch (subtype) { 742 case VSLENGTH: 743 cvtnum(varlen, buf); 744 strtodest(buf, flag, VSNORMAL, varflags & VSQUOTE, dst); 745 break; 746 747 case VSNORMAL: 748 break; 749 750 case VSPLUS: 751 case VSMINUS: 752 if (!set) { 753 argstr(p, flag | (flag & EXP_SPLIT ? EXP_SPLIT_LIT : 0) | 754 (varflags & VSQUOTE ? EXP_LIT_QUOTED : 0), dst); 755 break; 756 } 757 break; 758 759 case VSTRIMLEFT: 760 case VSTRIMLEFTMAX: 761 case VSTRIMRIGHT: 762 case VSTRIMRIGHTMAX: 763 if (!set) 764 break; 765 /* 766 * Terminate the string and start recording the pattern 767 * right after it 768 */ 769 STPUTC('\0', expdest); 770 patloc = expdest - stackblock(); 771 subevalvar_trim(p, patloc, subtype, startloc); 772 reprocess(startloc, flag, VSNORMAL, varflags & VSQUOTE, dst); 773 if (flag & EXP_SPLIT && *var == '@' && varflags & VSQUOTE) 774 dst->state = WORD_QUOTEMARK; 775 break; 776 777 case VSASSIGN: 778 case VSQUESTION: 779 if (!set) { 780 if (subevalvar_misc(p, var, subtype, startloc, 781 varflags)) { 782 varflags &= ~VSNUL; 783 goto again; 784 } 785 break; 786 } 787 break; 788 789 case VSERROR: 790 c = p - var - 1; 791 error("${%.*s%s}: Bad substitution", c, var, 792 (c > 0 && *p != CTLENDVAR) ? "..." : ""); 793 794 default: 795 abort(); 796 } 797 798 if (subtype != VSNORMAL) { /* skip to end of alternative */ 799 int nesting = 1; 800 for (;;) { 801 if ((c = *p++) == CTLESC) 802 p++; 803 else if (c == CTLBACKQ || c == (CTLBACKQ|CTLQUOTE)) { 804 if (set) 805 argbackq = argbackq->next; 806 } else if (c == CTLVAR) { 807 if ((*p++ & VSTYPE) != VSNORMAL) 808 nesting++; 809 } else if (c == CTLENDVAR) { 810 if (--nesting == 0) 811 break; 812 } 813 } 814 } 815 return p; 816 } 817 818 819 820 /* 821 * Test whether a specialized variable is set. 822 */ 823 824 static int 825 varisset(const char *name, int nulok) 826 { 827 828 if (*name == '!') 829 return backgndpidset(); 830 else if (*name == '@' || *name == '*') { 831 if (*shellparam.p == NULL) 832 return 0; 833 834 if (nulok) { 835 char **av; 836 837 for (av = shellparam.p; *av; av++) 838 if (**av != '\0') 839 return 1; 840 return 0; 841 } 842 } else if (is_digit(*name)) { 843 char *ap; 844 long num; 845 846 errno = 0; 847 num = strtol(name, NULL, 10); 848 if (errno != 0 || num > shellparam.nparam) 849 return 0; 850 851 if (num == 0) 852 ap = arg0; 853 else 854 ap = shellparam.p[num - 1]; 855 856 if (nulok && (ap == NULL || *ap == '\0')) 857 return 0; 858 } 859 return 1; 860 } 861 862 static void 863 strtodest(const char *p, int flag, int subtype, int quoted, 864 struct worddest *dst) 865 { 866 if (subtype == VSLENGTH || subtype == VSTRIMLEFT || 867 subtype == VSTRIMLEFTMAX || subtype == VSTRIMRIGHT || 868 subtype == VSTRIMRIGHTMAX) 869 STPUTS(p, expdest); 870 else if (flag & EXP_SPLIT && !quoted && dst != NULL) 871 STPUTS_SPLIT(p, BASESYNTAX, flag, expdest, dst); 872 else if (flag & (EXP_GLOB | EXP_CASE)) 873 STPUTS_QUOTES(p, quoted ? DQSYNTAX : BASESYNTAX, expdest); 874 else 875 STPUTS(p, expdest); 876 } 877 878 static void 879 reprocess(int startloc, int flag, int subtype, int quoted, 880 struct worddest *dst) 881 { 882 static char *buf = NULL; 883 static size_t buflen = 0; 884 char *startp; 885 size_t len, zpos, zlen; 886 887 startp = stackblock() + startloc; 888 len = expdest - startp; 889 if (len >= SIZE_MAX / 2) 890 abort(); 891 INTOFF; 892 if (len >= buflen) { 893 ckfree(buf); 894 buf = NULL; 895 } 896 if (buflen < 128) 897 buflen = 128; 898 while (len >= buflen) 899 buflen <<= 1; 900 if (buf == NULL) 901 buf = ckmalloc(buflen); 902 INTON; 903 memcpy(buf, startp, len); 904 buf[len] = '\0'; 905 STADJUST(-len, expdest); 906 for (zpos = 0;;) { 907 zlen = strlen(buf + zpos); 908 strtodest(buf + zpos, flag, subtype, quoted, dst); 909 zpos += zlen + 1; 910 if (zpos == len + 1) 911 break; 912 if (flag & EXP_SPLIT && (quoted || (zlen > 0 && zpos < len))) 913 NEXTWORD('\0', flag, expdest, dst); 914 } 915 } 916 917 /* 918 * Add the value of a specialized variable to the stack string. 919 */ 920 921 static void 922 varvalue(const char *name, int quoted, int subtype, int flag, 923 struct worddest *dst) 924 { 925 int num; 926 char *p; 927 int i; 928 int splitlater; 929 char sep[2]; 930 char **ap; 931 char buf[(NSHORTOPTS > 10 ? NSHORTOPTS : 10) + 1]; 932 933 if (subtype == VSLENGTH) 934 flag &= ~EXP_FULL; 935 splitlater = subtype == VSTRIMLEFT || subtype == VSTRIMLEFTMAX || 936 subtype == VSTRIMRIGHT || subtype == VSTRIMRIGHTMAX; 937 938 switch (*name) { 939 case '$': 940 num = rootpid; 941 break; 942 case '?': 943 num = oexitstatus; 944 break; 945 case '#': 946 num = shellparam.nparam; 947 break; 948 case '!': 949 num = backgndpidval(); 950 break; 951 case '-': 952 p = buf; 953 for (i = 0 ; i < NSHORTOPTS ; i++) { 954 if (optval[i]) 955 *p++ = optletter[i]; 956 } 957 *p = '\0'; 958 strtodest(buf, flag, subtype, quoted, dst); 959 return; 960 case '@': 961 if (flag & EXP_SPLIT && quoted) { 962 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 963 strtodest(p, flag, subtype, quoted, dst); 964 if (*ap) { 965 if (splitlater) 966 STPUTC('\0', expdest); 967 else 968 NEXTWORD('\0', flag, expdest, 969 dst); 970 } 971 } 972 if (shellparam.nparam > 0) 973 dst->state = WORD_QUOTEMARK; 974 return; 975 } 976 /* FALLTHROUGH */ 977 case '*': 978 if (ifsset()) 979 sep[0] = ifsval()[0]; 980 else 981 sep[0] = ' '; 982 sep[1] = '\0'; 983 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 984 strtodest(p, flag, subtype, quoted, dst); 985 if (!*ap) 986 break; 987 if (sep[0]) 988 strtodest(sep, flag, subtype, quoted, dst); 989 else if (flag & EXP_SPLIT && !quoted && **ap != '\0') { 990 if (splitlater) 991 STPUTC('\0', expdest); 992 else 993 NEXTWORD('\0', flag, expdest, dst); 994 } 995 } 996 return; 997 default: 998 if (is_digit(*name)) { 999 num = atoi(name); 1000 if (num == 0) 1001 p = arg0; 1002 else if (num > 0 && num <= shellparam.nparam) 1003 p = shellparam.p[num - 1]; 1004 else 1005 return; 1006 strtodest(p, flag, subtype, quoted, dst); 1007 } 1008 return; 1009 } 1010 cvtnum(num, buf); 1011 strtodest(buf, flag, subtype, quoted, dst); 1012 } 1013 1014 1015 1016 static char expdir[PATH_MAX]; 1017 #define expdir_end (expdir + sizeof(expdir)) 1018 1019 /* 1020 * Perform pathname generation and remove control characters. 1021 * At this point, the only control characters should be CTLESC. 1022 * The results are stored in the list dstlist. 1023 */ 1024 static void 1025 expandmeta(char *pattern, struct arglist *dstlist) 1026 { 1027 char *p; 1028 int firstmatch; 1029 char c; 1030 1031 firstmatch = dstlist->count; 1032 p = pattern; 1033 for (; (c = *p) != '\0'; p++) { 1034 /* fast check for meta chars */ 1035 if (c == '*' || c == '?' || c == '[') { 1036 INTOFF; 1037 expmeta(expdir, pattern, dstlist); 1038 INTON; 1039 break; 1040 } 1041 } 1042 if (dstlist->count == firstmatch) { 1043 /* 1044 * no matches 1045 */ 1046 rmescapes(pattern); 1047 appendarglist(dstlist, pattern); 1048 } else { 1049 qsort(&dstlist->args[firstmatch], 1050 dstlist->count - firstmatch, 1051 sizeof(dstlist->args[0]), expsortcmp); 1052 } 1053 } 1054 1055 1056 /* 1057 * Do metacharacter (i.e. *, ?, [...]) expansion. 1058 */ 1059 1060 static void 1061 expmeta(char *enddir, char *name, struct arglist *arglist) 1062 { 1063 const char *p; 1064 const char *q; 1065 const char *start; 1066 char *endname; 1067 int metaflag; 1068 struct stat statb; 1069 DIR *dirp; 1070 struct dirent *dp; 1071 int atend; 1072 int matchdot; 1073 int esc; 1074 int namlen; 1075 1076 metaflag = 0; 1077 start = name; 1078 for (p = name; esc = 0, *p; p += esc + 1) { 1079 if (*p == '*' || *p == '?') 1080 metaflag = 1; 1081 else if (*p == '[') { 1082 q = p + 1; 1083 if (*q == '!' || *q == '^') 1084 q++; 1085 for (;;) { 1086 if (*q == CTLESC) 1087 q++; 1088 if (*q == '/' || *q == '\0') 1089 break; 1090 if (*++q == ']') { 1091 metaflag = 1; 1092 break; 1093 } 1094 } 1095 } else if (*p == '\0') 1096 break; 1097 else { 1098 if (*p == CTLESC) 1099 esc++; 1100 if (p[esc] == '/') { 1101 if (metaflag) 1102 break; 1103 start = p + esc + 1; 1104 } 1105 } 1106 } 1107 if (metaflag == 0) { /* we've reached the end of the file name */ 1108 if (enddir != expdir) 1109 metaflag++; 1110 for (p = name ; ; p++) { 1111 if (*p == CTLESC) 1112 p++; 1113 *enddir++ = *p; 1114 if (*p == '\0') 1115 break; 1116 if (enddir == expdir_end) 1117 return; 1118 } 1119 if (metaflag == 0 || lstat(expdir, &statb) >= 0) 1120 appendarglist(arglist, stsavestr(expdir)); 1121 return; 1122 } 1123 endname = name + (p - name); 1124 if (start != name) { 1125 p = name; 1126 while (p < start) { 1127 if (*p == CTLESC) 1128 p++; 1129 *enddir++ = *p++; 1130 if (enddir == expdir_end) 1131 return; 1132 } 1133 } 1134 if (enddir == expdir) { 1135 p = "."; 1136 } else if (enddir == expdir + 1 && *expdir == '/') { 1137 p = "/"; 1138 } else { 1139 p = expdir; 1140 enddir[-1] = '\0'; 1141 } 1142 if ((dirp = opendir(p)) == NULL) 1143 return; 1144 if (enddir != expdir) 1145 enddir[-1] = '/'; 1146 if (*endname == 0) { 1147 atend = 1; 1148 } else { 1149 atend = 0; 1150 *endname = '\0'; 1151 endname += esc + 1; 1152 } 1153 matchdot = 0; 1154 p = start; 1155 if (*p == CTLESC) 1156 p++; 1157 if (*p == '.') 1158 matchdot++; 1159 while (! int_pending() && (dp = readdir(dirp)) != NULL) { 1160 if (dp->d_name[0] == '.' && ! matchdot) 1161 continue; 1162 if (patmatch(start, dp->d_name)) { 1163 namlen = dp->d_namlen; 1164 if (enddir + namlen + 1 > expdir_end) 1165 continue; 1166 memcpy(enddir, dp->d_name, namlen + 1); 1167 if (atend) 1168 appendarglist(arglist, stsavestr(expdir)); 1169 else { 1170 if (dp->d_type != DT_UNKNOWN && 1171 dp->d_type != DT_DIR && 1172 dp->d_type != DT_LNK) 1173 continue; 1174 if (enddir + namlen + 2 > expdir_end) 1175 continue; 1176 enddir[namlen] = '/'; 1177 enddir[namlen + 1] = '\0'; 1178 expmeta(enddir + namlen + 1, endname, arglist); 1179 } 1180 } 1181 } 1182 closedir(dirp); 1183 if (! atend) 1184 endname[-esc - 1] = esc ? CTLESC : '/'; 1185 } 1186 1187 1188 static int 1189 expsortcmp(const void *p1, const void *p2) 1190 { 1191 const char *s1 = *(const char * const *)p1; 1192 const char *s2 = *(const char * const *)p2; 1193 1194 return (strcmp(s1, s2)); 1195 } 1196 1197 1198 1199 static wchar_t 1200 get_wc(const char **p) 1201 { 1202 wchar_t c; 1203 int chrlen; 1204 1205 chrlen = mbtowc(&c, *p, 4); 1206 if (chrlen == 0) 1207 return 0; 1208 else if (chrlen == -1) 1209 c = 0; 1210 else 1211 *p += chrlen; 1212 return c; 1213 } 1214 1215 1216 /* 1217 * See if a character matches a character class, starting at the first colon 1218 * of "[:class:]". 1219 * If a valid character class is recognized, a pointer to the next character 1220 * after the final closing bracket is stored into *end, otherwise a null 1221 * pointer is stored into *end. 1222 */ 1223 static int 1224 match_charclass(const char *p, wchar_t chr, const char **end) 1225 { 1226 char name[20]; 1227 const char *nameend; 1228 wctype_t cclass; 1229 1230 *end = NULL; 1231 p++; 1232 nameend = strstr(p, ":]"); 1233 if (nameend == NULL || (size_t)(nameend - p) >= sizeof(name) || 1234 nameend == p) 1235 return 0; 1236 memcpy(name, p, nameend - p); 1237 name[nameend - p] = '\0'; 1238 *end = nameend + 2; 1239 cclass = wctype(name); 1240 /* An unknown class matches nothing but is valid nevertheless. */ 1241 if (cclass == 0) 1242 return 0; 1243 return iswctype(chr, cclass); 1244 } 1245 1246 1247 /* 1248 * Returns true if the pattern matches the string. 1249 */ 1250 1251 static int 1252 patmatch(const char *pattern, const char *string) 1253 { 1254 const char *p, *q, *end; 1255 const char *bt_p, *bt_q; 1256 char c; 1257 wchar_t wc, wc2; 1258 1259 p = pattern; 1260 q = string; 1261 bt_p = NULL; 1262 bt_q = NULL; 1263 for (;;) { 1264 switch (c = *p++) { 1265 case '\0': 1266 if (*q != '\0') 1267 goto backtrack; 1268 return 1; 1269 case CTLESC: 1270 if (*q++ != *p++) 1271 goto backtrack; 1272 break; 1273 case '?': 1274 if (*q == '\0') 1275 return 0; 1276 if (localeisutf8) { 1277 wc = get_wc(&q); 1278 /* 1279 * A '?' does not match invalid UTF-8 but a 1280 * '*' does, so backtrack. 1281 */ 1282 if (wc == 0) 1283 goto backtrack; 1284 } else 1285 wc = (unsigned char)*q++; 1286 break; 1287 case '*': 1288 c = *p; 1289 while (c == '*') 1290 c = *++p; 1291 /* 1292 * If the pattern ends here, we know the string 1293 * matches without needing to look at the rest of it. 1294 */ 1295 if (c == '\0') 1296 return 1; 1297 /* 1298 * First try the shortest match for the '*' that 1299 * could work. We can forget any earlier '*' since 1300 * there is no way having it match more characters 1301 * can help us, given that we are already here. 1302 */ 1303 bt_p = p; 1304 bt_q = q; 1305 break; 1306 case '[': { 1307 const char *savep, *saveq; 1308 int invert, found; 1309 wchar_t chr; 1310 1311 savep = p, saveq = q; 1312 invert = 0; 1313 if (*p == '!' || *p == '^') { 1314 invert++; 1315 p++; 1316 } 1317 found = 0; 1318 if (*q == '\0') 1319 return 0; 1320 if (localeisutf8) { 1321 chr = get_wc(&q); 1322 if (chr == 0) 1323 goto backtrack; 1324 } else 1325 chr = (unsigned char)*q++; 1326 c = *p++; 1327 do { 1328 if (c == '\0') { 1329 p = savep, q = saveq; 1330 c = '['; 1331 goto dft; 1332 } 1333 if (c == '[' && *p == ':') { 1334 found |= match_charclass(p, chr, &end); 1335 if (end != NULL) 1336 p = end; 1337 } 1338 if (c == CTLESC) 1339 c = *p++; 1340 if (localeisutf8 && c & 0x80) { 1341 p--; 1342 wc = get_wc(&p); 1343 if (wc == 0) /* bad utf-8 */ 1344 return 0; 1345 } else 1346 wc = (unsigned char)c; 1347 if (*p == '-' && p[1] != ']') { 1348 p++; 1349 if (*p == CTLESC) 1350 p++; 1351 if (localeisutf8) { 1352 wc2 = get_wc(&p); 1353 if (wc2 == 0) /* bad utf-8 */ 1354 return 0; 1355 } else 1356 wc2 = (unsigned char)*p++; 1357 if ( collate_range_cmp(chr, wc) >= 0 1358 && collate_range_cmp(chr, wc2) <= 0 1359 ) 1360 found = 1; 1361 } else { 1362 if (chr == wc) 1363 found = 1; 1364 } 1365 } while ((c = *p++) != ']'); 1366 if (found == invert) 1367 goto backtrack; 1368 break; 1369 } 1370 dft: default: 1371 if (*q == '\0') 1372 return 0; 1373 if (*q++ == c) 1374 break; 1375 backtrack: 1376 /* 1377 * If we have a mismatch (other than hitting the end 1378 * of the string), go back to the last '*' seen and 1379 * have it match one additional character. 1380 */ 1381 if (bt_p == NULL) 1382 return 0; 1383 if (*bt_q == '\0') 1384 return 0; 1385 bt_q++; 1386 p = bt_p; 1387 q = bt_q; 1388 break; 1389 } 1390 } 1391 } 1392 1393 1394 1395 /* 1396 * Remove any CTLESC and CTLQUOTEMARK characters from a string. 1397 */ 1398 1399 void 1400 rmescapes(char *str) 1401 { 1402 char *p, *q; 1403 1404 p = str; 1405 while (*p != CTLESC && *p != CTLQUOTEMARK && *p != CTLQUOTEEND) { 1406 if (*p++ == '\0') 1407 return; 1408 } 1409 q = p; 1410 while (*p) { 1411 if (*p == CTLQUOTEMARK || *p == CTLQUOTEEND) { 1412 p++; 1413 continue; 1414 } 1415 if (*p == CTLESC) 1416 p++; 1417 *q++ = *p++; 1418 } 1419 *q = '\0'; 1420 } 1421 1422 1423 1424 /* 1425 * See if a pattern matches in a case statement. 1426 */ 1427 1428 int 1429 casematch(union node *pattern, const char *val) 1430 { 1431 struct stackmark smark; 1432 int result; 1433 char *p; 1434 1435 setstackmark(&smark); 1436 argbackq = pattern->narg.backquote; 1437 STARTSTACKSTR(expdest); 1438 argstr(pattern->narg.text, EXP_TILDE | EXP_CASE, NULL); 1439 STPUTC('\0', expdest); 1440 p = grabstackstr(expdest); 1441 result = patmatch(p, val); 1442 popstackmark(&smark); 1443 return result; 1444 } 1445 1446 /* 1447 * Our own itoa(). 1448 */ 1449 1450 static void 1451 cvtnum(int num, char *buf) 1452 { 1453 char temp[32]; 1454 int neg = num < 0; 1455 char *p = temp + 31; 1456 1457 temp[31] = '\0'; 1458 1459 do { 1460 *--p = num % 10 + '0'; 1461 } while ((num /= 10) != 0); 1462 1463 if (neg) 1464 *--p = '-'; 1465 1466 memcpy(buf, p, temp + 32 - p); 1467 } 1468 1469 /* 1470 * Do most of the work for wordexp(3). 1471 */ 1472 1473 int 1474 wordexpcmd(int argc, char **argv) 1475 { 1476 size_t len; 1477 int i; 1478 1479 out1fmt("%08x", argc - 1); 1480 for (i = 1, len = 0; i < argc; i++) 1481 len += strlen(argv[i]); 1482 out1fmt("%08x", (int)len); 1483 for (i = 1; i < argc; i++) 1484 outbin(argv[i], strlen(argv[i]) + 1, out1); 1485 return (0); 1486 } 1487 1488 /* 1489 * Do most of the work for wordexp(3), new version. 1490 */ 1491 1492 int 1493 freebsd_wordexpcmd(int argc __unused, char **argv __unused) 1494 { 1495 struct arglist arglist; 1496 union node *args, *n; 1497 size_t len; 1498 int ch; 1499 int protected = 0; 1500 int fd = -1; 1501 int i; 1502 1503 while ((ch = nextopt("f:p")) != '\0') { 1504 switch (ch) { 1505 case 'f': 1506 fd = number(shoptarg); 1507 break; 1508 case 'p': 1509 protected = 1; 1510 break; 1511 } 1512 } 1513 if (*argptr != NULL) 1514 error("wrong number of arguments"); 1515 if (fd < 0) 1516 error("missing fd"); 1517 INTOFF; 1518 setinputfd(fd, 1); 1519 INTON; 1520 args = parsewordexp(); 1521 popfile(); /* will also close fd */ 1522 if (protected) 1523 for (n = args; n != NULL; n = n->narg.next) { 1524 if (n->narg.backquote != NULL) { 1525 outcslow('C', out1); 1526 error("command substitution disabled"); 1527 } 1528 } 1529 outcslow(' ', out1); 1530 emptyarglist(&arglist); 1531 for (n = args; n != NULL; n = n->narg.next) 1532 expandarg(n, &arglist, EXP_FULL | EXP_TILDE); 1533 for (i = 0, len = 0; i < arglist.count; i++) 1534 len += strlen(arglist.args[i]); 1535 out1fmt("%016x %016zx", arglist.count, len); 1536 for (i = 0; i < arglist.count; i++) 1537 outbin(arglist.args[i], strlen(arglist.args[i]) + 1, out1); 1538 return (0); 1539 } 1540