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 * 3. 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 const char *argstr(const char *, int, struct worddest *); 95 static const char *exptilde(const char *, int); 96 static const char *expari(const char *, int, struct worddest *); 97 static void expbackq(union node *, int, int, struct worddest *); 98 static void subevalvar_trim(const char *, int, int, int); 99 static int subevalvar_misc(const char *, const char *, int, int, int); 100 static const char *evalvar(const 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 ${var+word} etc. 260 * If EXP_GLOB or EXP_CASE are set, keep and/or generate CTLESC 261 * characters to allow for further processing. 262 * 263 * If EXP_SPLIT is set, dst receives any complete words produced. 264 */ 265 static const char * 266 argstr(const char *p, int flag, struct worddest *dst) 267 { 268 char c; 269 int quotes = flag & (EXP_GLOB | EXP_CASE); /* do CTLESC */ 270 int firsteq = 1; 271 int split_lit; 272 int lit_quoted; 273 274 split_lit = flag & EXP_SPLIT_LIT; 275 lit_quoted = flag & EXP_LIT_QUOTED; 276 flag &= ~(EXP_SPLIT_LIT | EXP_LIT_QUOTED); 277 if (*p == '~' && (flag & (EXP_TILDE | EXP_VARTILDE))) 278 p = exptilde(p, flag); 279 for (;;) { 280 CHECKSTRSPACE(2, expdest); 281 switch (c = *p++) { 282 case '\0': 283 return (p - 1); 284 case CTLENDVAR: 285 case CTLENDARI: 286 return (p); 287 case CTLQUOTEMARK: 288 lit_quoted = 1; 289 /* "$@" 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 const char * 356 exptilde(const char *p, int flag) 357 { 358 char c; 359 const char *startp = p; 360 const char *user; 361 struct passwd *pw; 362 char *home; 363 int len; 364 365 for (;;) { 366 c = *p; 367 switch(c) { 368 case CTLESC: /* This means CTL* are always considered quoted. */ 369 case CTLVAR: 370 case CTLBACKQ: 371 case CTLBACKQ | CTLQUOTE: 372 case CTLARI: 373 case CTLENDARI: 374 case CTLQUOTEMARK: 375 return (startp); 376 case ':': 377 if ((flag & EXP_VARTILDE) == 0) 378 break; 379 /* FALLTHROUGH */ 380 case '\0': 381 case '/': 382 case CTLENDVAR: 383 len = p - startp - 1; 384 STPUTBIN(startp + 1, len, expdest); 385 STACKSTRNUL(expdest); 386 user = expdest - len; 387 if (*user == '\0') { 388 home = lookupvar("HOME"); 389 } else { 390 pw = getpwnam(user); 391 home = pw != NULL ? pw->pw_dir : NULL; 392 } 393 STADJUST(-len, expdest); 394 if (home == NULL || *home == '\0') 395 return (startp); 396 strtodest(home, flag, VSNORMAL, 1, NULL); 397 return (p); 398 } 399 p++; 400 } 401 } 402 403 404 /* 405 * Expand arithmetic expression. 406 */ 407 static const char * 408 expari(const char *p, int flag, struct worddest *dst) 409 { 410 char *q, *start; 411 arith_t result; 412 int begoff; 413 int quoted; 414 int adj; 415 416 quoted = *p++ == '"'; 417 begoff = expdest - stackblock(); 418 p = argstr(p, 0, NULL); 419 STPUTC('\0', expdest); 420 start = stackblock() + begoff; 421 422 q = grabstackstr(expdest); 423 result = arith(start); 424 ungrabstackstr(q, expdest); 425 426 start = stackblock() + begoff; 427 adj = start - expdest; 428 STADJUST(adj, expdest); 429 430 CHECKSTRSPACE((int)(DIGITS(result) + 1), expdest); 431 fmtstr(expdest, DIGITS(result), ARITH_FORMAT_STR, result); 432 adj = strlen(expdest); 433 STADJUST(adj, expdest); 434 if (!quoted) 435 reprocess(expdest - adj - stackblock(), flag, VSNORMAL, 0, dst); 436 return p; 437 } 438 439 440 /* 441 * Perform command substitution. 442 */ 443 static void 444 expbackq(union node *cmd, int quoted, int flag, struct worddest *dst) 445 { 446 struct backcmd in; 447 int i; 448 char buf[128]; 449 char *p; 450 char *dest = expdest; 451 struct nodelist *saveargbackq; 452 char lastc; 453 char const *syntax = quoted? DQSYNTAX : BASESYNTAX; 454 int quotes = flag & (EXP_GLOB | EXP_CASE); 455 size_t nnl; 456 const char *ifs; 457 458 INTOFF; 459 saveargbackq = argbackq; 460 p = grabstackstr(dest); 461 evalbackcmd(cmd, &in); 462 ungrabstackstr(p, dest); 463 464 p = in.buf; 465 nnl = 0; 466 if (!quoted && flag & EXP_SPLIT) 467 ifs = ifsset() ? ifsval() : " \t\n"; 468 else 469 ifs = ""; 470 /* Don't copy trailing newlines */ 471 for (;;) { 472 if (--in.nleft < 0) { 473 if (in.fd < 0) 474 break; 475 while ((i = read(in.fd, buf, sizeof buf)) < 0 && errno == EINTR) 476 ; 477 TRACE(("expbackq: read returns %d\n", i)); 478 if (i <= 0) 479 break; 480 p = buf; 481 in.nleft = i - 1; 482 } 483 lastc = *p++; 484 if (lastc == '\0') 485 continue; 486 if (lastc == '\n') { 487 nnl++; 488 } else { 489 if (nnl > 0) { 490 if (strchr(ifs, '\n') != NULL) { 491 NEXTWORD('\n', flag, dest, dst); 492 nnl = 0; 493 } else { 494 CHECKSTRSPACE(nnl + 2, dest); 495 while (nnl > 0) { 496 nnl--; 497 USTPUTC('\n', dest); 498 } 499 } 500 } 501 if (strchr(ifs, lastc) != NULL) 502 NEXTWORD(lastc, flag, dest, dst); 503 else { 504 CHECKSTRSPACE(2, dest); 505 if (quotes && syntax[(int)lastc] == CCTL) 506 USTPUTC(CTLESC, dest); 507 USTPUTC(lastc, dest); 508 } 509 } 510 } 511 512 if (in.fd >= 0) 513 close(in.fd); 514 if (in.buf) 515 ckfree(in.buf); 516 if (in.jp) { 517 p = grabstackstr(dest); 518 exitstatus = waitforjob(in.jp, (int *)NULL); 519 ungrabstackstr(p, dest); 520 } 521 TRACE(("expbackq: size=%td: \"%.*s\"\n", 522 ((dest - stackblock()) - startloc), 523 (int)((dest - stackblock()) - startloc), 524 stackblock() + startloc)); 525 argbackq = saveargbackq; 526 expdest = dest; 527 INTON; 528 } 529 530 531 532 static void 533 recordleft(const char *str, const char *loc, char *startp) 534 { 535 int amount; 536 537 amount = ((str - 1) - (loc - startp)) - expdest; 538 STADJUST(amount, expdest); 539 while (loc != str - 1) 540 *startp++ = *loc++; 541 } 542 543 static void 544 subevalvar_trim(const char *p, int strloc, int subtype, int startloc) 545 { 546 char *startp; 547 char *loc = NULL; 548 char *str; 549 int c = 0; 550 struct nodelist *saveargbackq = argbackq; 551 int amount; 552 553 argstr(p, EXP_CASE | EXP_TILDE, NULL); 554 STACKSTRNUL(expdest); 555 argbackq = saveargbackq; 556 startp = stackblock() + startloc; 557 str = stackblock() + strloc; 558 559 switch (subtype) { 560 case VSTRIMLEFT: 561 for (loc = startp; loc < str; loc++) { 562 c = *loc; 563 *loc = '\0'; 564 if (patmatch(str, startp)) { 565 *loc = c; 566 recordleft(str, loc, startp); 567 return; 568 } 569 *loc = c; 570 } 571 break; 572 573 case VSTRIMLEFTMAX: 574 for (loc = str - 1; loc >= startp;) { 575 c = *loc; 576 *loc = '\0'; 577 if (patmatch(str, startp)) { 578 *loc = c; 579 recordleft(str, loc, startp); 580 return; 581 } 582 *loc = c; 583 loc--; 584 } 585 break; 586 587 case VSTRIMRIGHT: 588 for (loc = str - 1; loc >= startp;) { 589 if (patmatch(str, loc)) { 590 amount = loc - expdest; 591 STADJUST(amount, expdest); 592 return; 593 } 594 loc--; 595 } 596 break; 597 598 case VSTRIMRIGHTMAX: 599 for (loc = startp; loc < str - 1; loc++) { 600 if (patmatch(str, loc)) { 601 amount = loc - expdest; 602 STADJUST(amount, expdest); 603 return; 604 } 605 } 606 break; 607 608 609 default: 610 abort(); 611 } 612 amount = (expdest - stackblock() - strloc) + 1; 613 STADJUST(-amount, expdest); 614 } 615 616 617 static int 618 subevalvar_misc(const char *p, const char *var, int subtype, int startloc, 619 int varflags) 620 { 621 char *startp; 622 struct nodelist *saveargbackq = argbackq; 623 int amount; 624 625 argstr(p, EXP_TILDE, NULL); 626 STACKSTRNUL(expdest); 627 argbackq = saveargbackq; 628 startp = stackblock() + startloc; 629 630 switch (subtype) { 631 case VSASSIGN: 632 setvar(var, startp, 0); 633 amount = startp - expdest; 634 STADJUST(amount, expdest); 635 return 1; 636 637 case VSQUESTION: 638 if (*p != CTLENDVAR) { 639 outfmt(out2, "%s\n", startp); 640 error((char *)NULL); 641 } 642 error("%.*s: parameter %snot set", (int)(p - var - 1), 643 var, (varflags & VSNUL) ? "null or " : ""); 644 return 0; 645 646 default: 647 abort(); 648 } 649 } 650 651 652 /* 653 * Expand a variable, and return a pointer to the next character in the 654 * input string. 655 */ 656 657 static const char * 658 evalvar(const char *p, int flag, struct worddest *dst) 659 { 660 int subtype; 661 int varflags; 662 const char *var; 663 const char *val; 664 int patloc; 665 int c; 666 int set; 667 int special; 668 int startloc; 669 int varlen; 670 int varlenb; 671 char buf[21]; 672 673 varflags = (unsigned char)*p++; 674 subtype = varflags & VSTYPE; 675 var = p; 676 special = 0; 677 if (! is_name(*p)) 678 special = 1; 679 p = strchr(p, '=') + 1; 680 again: /* jump here after setting a variable with ${var=text} */ 681 if (varflags & VSLINENO) { 682 set = 1; 683 special = 1; 684 val = NULL; 685 } else if (special) { 686 set = varisset(var, varflags & VSNUL); 687 val = NULL; 688 } else { 689 val = bltinlookup(var, 1); 690 if (val == NULL || ((varflags & VSNUL) && val[0] == '\0')) { 691 val = NULL; 692 set = 0; 693 } else 694 set = 1; 695 } 696 varlen = 0; 697 startloc = expdest - stackblock(); 698 if (!set && uflag && *var != '@' && *var != '*') { 699 switch (subtype) { 700 case VSNORMAL: 701 case VSTRIMLEFT: 702 case VSTRIMLEFTMAX: 703 case VSTRIMRIGHT: 704 case VSTRIMRIGHTMAX: 705 case VSLENGTH: 706 error("%.*s: parameter not set", (int)(p - var - 1), 707 var); 708 } 709 } 710 if (set && subtype != VSPLUS) { 711 /* insert the value of the variable */ 712 if (special) { 713 if (varflags & VSLINENO) { 714 if (p - var > (ptrdiff_t)sizeof(buf)) 715 abort(); 716 memcpy(buf, var, p - var - 1); 717 buf[p - var - 1] = '\0'; 718 strtodest(buf, flag, subtype, 719 varflags & VSQUOTE, dst); 720 } else 721 varvalue(var, varflags & VSQUOTE, subtype, flag, 722 dst); 723 if (subtype == VSLENGTH) { 724 varlenb = expdest - stackblock() - startloc; 725 varlen = varlenb; 726 if (localeisutf8) { 727 val = stackblock() + startloc; 728 for (;val != expdest; val++) 729 if ((*val & 0xC0) == 0x80) 730 varlen--; 731 } 732 STADJUST(-varlenb, expdest); 733 } 734 } else { 735 if (subtype == VSLENGTH) { 736 for (;*val; val++) 737 if (!localeisutf8 || 738 (*val & 0xC0) != 0x80) 739 varlen++; 740 } 741 else 742 strtodest(val, flag, subtype, 743 varflags & VSQUOTE, dst); 744 } 745 } 746 747 if (subtype == VSPLUS) 748 set = ! set; 749 750 switch (subtype) { 751 case VSLENGTH: 752 cvtnum(varlen, buf); 753 strtodest(buf, flag, VSNORMAL, varflags & VSQUOTE, dst); 754 break; 755 756 case VSNORMAL: 757 break; 758 759 case VSPLUS: 760 case VSMINUS: 761 if (!set) { 762 argstr(p, flag | (flag & EXP_SPLIT ? EXP_SPLIT_LIT : 0) | 763 (varflags & VSQUOTE ? EXP_LIT_QUOTED : 0), dst); 764 break; 765 } 766 break; 767 768 case VSTRIMLEFT: 769 case VSTRIMLEFTMAX: 770 case VSTRIMRIGHT: 771 case VSTRIMRIGHTMAX: 772 if (!set) { 773 set = 1; 774 break; 775 } 776 /* 777 * Terminate the string and start recording the pattern 778 * right after it 779 */ 780 STPUTC('\0', expdest); 781 patloc = expdest - stackblock(); 782 subevalvar_trim(p, patloc, subtype, startloc); 783 reprocess(startloc, flag, VSNORMAL, varflags & VSQUOTE, dst); 784 if (flag & EXP_SPLIT && *var == '@' && varflags & VSQUOTE) 785 dst->state = WORD_QUOTEMARK; 786 break; 787 788 case VSASSIGN: 789 case VSQUESTION: 790 if (!set) { 791 if (subevalvar_misc(p, var, subtype, startloc, 792 varflags)) { 793 varflags &= ~VSNUL; 794 goto again; 795 } 796 break; 797 } 798 break; 799 800 case VSERROR: 801 c = p - var - 1; 802 error("${%.*s%s}: Bad substitution", c, var, 803 (c > 0 && *p != CTLENDVAR) ? "..." : ""); 804 805 default: 806 abort(); 807 } 808 809 if (subtype != VSNORMAL) { /* skip to end of alternative */ 810 int nesting = 1; 811 for (;;) { 812 if ((c = *p++) == CTLESC) 813 p++; 814 else if (c == CTLBACKQ || c == (CTLBACKQ|CTLQUOTE)) { 815 if (set) 816 argbackq = argbackq->next; 817 } else if (c == CTLVAR) { 818 if ((*p++ & VSTYPE) != VSNORMAL) 819 nesting++; 820 } else if (c == CTLENDVAR) { 821 if (--nesting == 0) 822 break; 823 } 824 } 825 } 826 return p; 827 } 828 829 830 831 /* 832 * Test whether a specialized variable is set. 833 */ 834 835 static int 836 varisset(const char *name, int nulok) 837 { 838 839 if (*name == '!') 840 return backgndpidset(); 841 else if (*name == '@' || *name == '*') { 842 if (*shellparam.p == NULL) 843 return 0; 844 845 if (nulok) { 846 char **av; 847 848 for (av = shellparam.p; *av; av++) 849 if (**av != '\0') 850 return 1; 851 return 0; 852 } 853 } else if (is_digit(*name)) { 854 char *ap; 855 long num; 856 857 errno = 0; 858 num = strtol(name, NULL, 10); 859 if (errno != 0 || num > shellparam.nparam) 860 return 0; 861 862 if (num == 0) 863 ap = arg0; 864 else 865 ap = shellparam.p[num - 1]; 866 867 if (nulok && (ap == NULL || *ap == '\0')) 868 return 0; 869 } 870 return 1; 871 } 872 873 static void 874 strtodest(const char *p, int flag, int subtype, int quoted, 875 struct worddest *dst) 876 { 877 if (subtype == VSLENGTH || subtype == VSTRIMLEFT || 878 subtype == VSTRIMLEFTMAX || subtype == VSTRIMRIGHT || 879 subtype == VSTRIMRIGHTMAX) 880 STPUTS(p, expdest); 881 else if (flag & EXP_SPLIT && !quoted && dst != NULL) 882 STPUTS_SPLIT(p, BASESYNTAX, flag, expdest, dst); 883 else if (flag & (EXP_GLOB | EXP_CASE)) 884 STPUTS_QUOTES(p, quoted ? DQSYNTAX : BASESYNTAX, expdest); 885 else 886 STPUTS(p, expdest); 887 } 888 889 static void 890 reprocess(int startloc, int flag, int subtype, int quoted, 891 struct worddest *dst) 892 { 893 static char *buf = NULL; 894 static size_t buflen = 0; 895 char *startp; 896 size_t len, zpos, zlen; 897 898 startp = stackblock() + startloc; 899 len = expdest - startp; 900 if (len >= SIZE_MAX / 2) 901 abort(); 902 INTOFF; 903 if (len >= buflen) { 904 ckfree(buf); 905 buf = NULL; 906 } 907 if (buflen < 128) 908 buflen = 128; 909 while (len >= buflen) 910 buflen <<= 1; 911 if (buf == NULL) 912 buf = ckmalloc(buflen); 913 INTON; 914 memcpy(buf, startp, len); 915 buf[len] = '\0'; 916 STADJUST(-len, expdest); 917 for (zpos = 0;;) { 918 zlen = strlen(buf + zpos); 919 strtodest(buf + zpos, flag, subtype, quoted, dst); 920 zpos += zlen + 1; 921 if (zpos == len + 1) 922 break; 923 if (flag & EXP_SPLIT && (quoted || (zlen > 0 && zpos < len))) 924 NEXTWORD('\0', flag, expdest, dst); 925 } 926 } 927 928 /* 929 * Add the value of a specialized variable to the stack string. 930 */ 931 932 static void 933 varvalue(const char *name, int quoted, int subtype, int flag, 934 struct worddest *dst) 935 { 936 int num; 937 char *p; 938 int i; 939 int splitlater; 940 char sep[2]; 941 char **ap; 942 char buf[(NSHORTOPTS > 10 ? NSHORTOPTS : 10) + 1]; 943 944 if (subtype == VSLENGTH) 945 flag &= ~EXP_FULL; 946 splitlater = subtype == VSTRIMLEFT || subtype == VSTRIMLEFTMAX || 947 subtype == VSTRIMRIGHT || subtype == VSTRIMRIGHTMAX; 948 949 switch (*name) { 950 case '$': 951 num = rootpid; 952 break; 953 case '?': 954 num = oexitstatus; 955 break; 956 case '#': 957 num = shellparam.nparam; 958 break; 959 case '!': 960 num = backgndpidval(); 961 break; 962 case '-': 963 p = buf; 964 for (i = 0 ; i < NSHORTOPTS ; i++) { 965 if (optval[i]) 966 *p++ = optletter[i]; 967 } 968 *p = '\0'; 969 strtodest(buf, flag, subtype, quoted, dst); 970 return; 971 case '@': 972 if (flag & EXP_SPLIT && quoted) { 973 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 974 strtodest(p, flag, subtype, quoted, dst); 975 if (*ap) { 976 if (splitlater) 977 STPUTC('\0', expdest); 978 else 979 NEXTWORD('\0', flag, expdest, 980 dst); 981 } 982 } 983 if (shellparam.nparam > 0) 984 dst->state = WORD_QUOTEMARK; 985 return; 986 } 987 /* FALLTHROUGH */ 988 case '*': 989 if (ifsset()) 990 sep[0] = ifsval()[0]; 991 else 992 sep[0] = ' '; 993 sep[1] = '\0'; 994 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 995 strtodest(p, flag, subtype, quoted, dst); 996 if (!*ap) 997 break; 998 if (sep[0]) 999 strtodest(sep, flag, subtype, quoted, dst); 1000 else if (flag & EXP_SPLIT && !quoted && **ap != '\0') { 1001 if (splitlater) 1002 STPUTC('\0', expdest); 1003 else 1004 NEXTWORD('\0', flag, expdest, dst); 1005 } 1006 } 1007 return; 1008 default: 1009 if (is_digit(*name)) { 1010 num = atoi(name); 1011 if (num == 0) 1012 p = arg0; 1013 else if (num > 0 && num <= shellparam.nparam) 1014 p = shellparam.p[num - 1]; 1015 else 1016 return; 1017 strtodest(p, flag, subtype, quoted, dst); 1018 } 1019 return; 1020 } 1021 cvtnum(num, buf); 1022 strtodest(buf, flag, subtype, quoted, dst); 1023 } 1024 1025 1026 1027 static char expdir[PATH_MAX]; 1028 #define expdir_end (expdir + sizeof(expdir)) 1029 1030 /* 1031 * Perform pathname generation and remove control characters. 1032 * At this point, the only control characters should be CTLESC. 1033 * The results are stored in the list dstlist. 1034 */ 1035 static void 1036 expandmeta(char *pattern, struct arglist *dstlist) 1037 { 1038 char *p; 1039 int firstmatch; 1040 char c; 1041 1042 firstmatch = dstlist->count; 1043 p = pattern; 1044 for (; (c = *p) != '\0'; p++) { 1045 /* fast check for meta chars */ 1046 if (c == '*' || c == '?' || c == '[') { 1047 INTOFF; 1048 expmeta(expdir, pattern, dstlist); 1049 INTON; 1050 break; 1051 } 1052 } 1053 if (dstlist->count == firstmatch) { 1054 /* 1055 * no matches 1056 */ 1057 rmescapes(pattern); 1058 appendarglist(dstlist, pattern); 1059 } else { 1060 qsort(&dstlist->args[firstmatch], 1061 dstlist->count - firstmatch, 1062 sizeof(dstlist->args[0]), expsortcmp); 1063 } 1064 } 1065 1066 1067 /* 1068 * Do metacharacter (i.e. *, ?, [...]) expansion. 1069 */ 1070 1071 static void 1072 expmeta(char *enddir, char *name, struct arglist *arglist) 1073 { 1074 const char *p; 1075 const char *q; 1076 const char *start; 1077 char *endname; 1078 int metaflag; 1079 struct stat statb; 1080 DIR *dirp; 1081 struct dirent *dp; 1082 int atend; 1083 int matchdot; 1084 int esc; 1085 int namlen; 1086 1087 metaflag = 0; 1088 start = name; 1089 for (p = name; esc = 0, *p; p += esc + 1) { 1090 if (*p == '*' || *p == '?') 1091 metaflag = 1; 1092 else if (*p == '[') { 1093 q = p + 1; 1094 if (*q == '!' || *q == '^') 1095 q++; 1096 for (;;) { 1097 if (*q == CTLESC) 1098 q++; 1099 if (*q == '/' || *q == '\0') 1100 break; 1101 if (*++q == ']') { 1102 metaflag = 1; 1103 break; 1104 } 1105 } 1106 } else if (*p == '\0') 1107 break; 1108 else { 1109 if (*p == CTLESC) 1110 esc++; 1111 if (p[esc] == '/') { 1112 if (metaflag) 1113 break; 1114 start = p + esc + 1; 1115 } 1116 } 1117 } 1118 if (metaflag == 0) { /* we've reached the end of the file name */ 1119 if (enddir != expdir) 1120 metaflag++; 1121 for (p = name ; ; p++) { 1122 if (*p == CTLESC) 1123 p++; 1124 *enddir++ = *p; 1125 if (*p == '\0') 1126 break; 1127 if (enddir == expdir_end) 1128 return; 1129 } 1130 if (metaflag == 0 || lstat(expdir, &statb) >= 0) 1131 appendarglist(arglist, stsavestr(expdir)); 1132 return; 1133 } 1134 endname = name + (p - name); 1135 if (start != name) { 1136 p = name; 1137 while (p < start) { 1138 if (*p == CTLESC) 1139 p++; 1140 *enddir++ = *p++; 1141 if (enddir == expdir_end) 1142 return; 1143 } 1144 } 1145 if (enddir == expdir) { 1146 p = "."; 1147 } else if (enddir == expdir + 1 && *expdir == '/') { 1148 p = "/"; 1149 } else { 1150 p = expdir; 1151 enddir[-1] = '\0'; 1152 } 1153 if ((dirp = opendir(p)) == NULL) 1154 return; 1155 if (enddir != expdir) 1156 enddir[-1] = '/'; 1157 if (*endname == 0) { 1158 atend = 1; 1159 } else { 1160 atend = 0; 1161 *endname = '\0'; 1162 endname += esc + 1; 1163 } 1164 matchdot = 0; 1165 p = start; 1166 if (*p == CTLESC) 1167 p++; 1168 if (*p == '.') 1169 matchdot++; 1170 while (! int_pending() && (dp = readdir(dirp)) != NULL) { 1171 if (dp->d_name[0] == '.' && ! matchdot) 1172 continue; 1173 if (patmatch(start, dp->d_name)) { 1174 namlen = dp->d_namlen; 1175 if (enddir + namlen + 1 > expdir_end) 1176 continue; 1177 memcpy(enddir, dp->d_name, namlen + 1); 1178 if (atend) 1179 appendarglist(arglist, stsavestr(expdir)); 1180 else { 1181 if (dp->d_type != DT_UNKNOWN && 1182 dp->d_type != DT_DIR && 1183 dp->d_type != DT_LNK) 1184 continue; 1185 if (enddir + namlen + 2 > expdir_end) 1186 continue; 1187 enddir[namlen] = '/'; 1188 enddir[namlen + 1] = '\0'; 1189 expmeta(enddir + namlen + 1, endname, arglist); 1190 } 1191 } 1192 } 1193 closedir(dirp); 1194 if (! atend) 1195 endname[-esc - 1] = esc ? CTLESC : '/'; 1196 } 1197 1198 1199 static int 1200 expsortcmp(const void *p1, const void *p2) 1201 { 1202 const char *s1 = *(const char * const *)p1; 1203 const char *s2 = *(const char * const *)p2; 1204 1205 return (strcoll(s1, s2)); 1206 } 1207 1208 1209 1210 static wchar_t 1211 get_wc(const char **p) 1212 { 1213 wchar_t c; 1214 int chrlen; 1215 1216 chrlen = mbtowc(&c, *p, 4); 1217 if (chrlen == 0) 1218 return 0; 1219 else if (chrlen == -1) 1220 c = 0; 1221 else 1222 *p += chrlen; 1223 return c; 1224 } 1225 1226 1227 /* 1228 * See if a character matches a character class, starting at the first colon 1229 * of "[:class:]". 1230 * If a valid character class is recognized, a pointer to the next character 1231 * after the final closing bracket is stored into *end, otherwise a null 1232 * pointer is stored into *end. 1233 */ 1234 static int 1235 match_charclass(const char *p, wchar_t chr, const char **end) 1236 { 1237 char name[20]; 1238 const char *nameend; 1239 wctype_t cclass; 1240 1241 *end = NULL; 1242 p++; 1243 nameend = strstr(p, ":]"); 1244 if (nameend == NULL || (size_t)(nameend - p) >= sizeof(name) || 1245 nameend == p) 1246 return 0; 1247 memcpy(name, p, nameend - p); 1248 name[nameend - p] = '\0'; 1249 *end = nameend + 2; 1250 cclass = wctype(name); 1251 /* An unknown class matches nothing but is valid nevertheless. */ 1252 if (cclass == 0) 1253 return 0; 1254 return iswctype(chr, cclass); 1255 } 1256 1257 1258 /* 1259 * Returns true if the pattern matches the string. 1260 */ 1261 1262 static int 1263 patmatch(const char *pattern, const char *string) 1264 { 1265 const char *p, *q, *end; 1266 const char *bt_p, *bt_q; 1267 char c; 1268 wchar_t wc, wc2; 1269 1270 p = pattern; 1271 q = string; 1272 bt_p = NULL; 1273 bt_q = NULL; 1274 for (;;) { 1275 switch (c = *p++) { 1276 case '\0': 1277 if (*q != '\0') 1278 goto backtrack; 1279 return 1; 1280 case CTLESC: 1281 if (*q++ != *p++) 1282 goto backtrack; 1283 break; 1284 case '?': 1285 if (*q == '\0') 1286 return 0; 1287 if (localeisutf8) { 1288 wc = get_wc(&q); 1289 /* 1290 * A '?' does not match invalid UTF-8 but a 1291 * '*' does, so backtrack. 1292 */ 1293 if (wc == 0) 1294 goto backtrack; 1295 } else 1296 q++; 1297 break; 1298 case '*': 1299 c = *p; 1300 while (c == '*') 1301 c = *++p; 1302 /* 1303 * If the pattern ends here, we know the string 1304 * matches without needing to look at the rest of it. 1305 */ 1306 if (c == '\0') 1307 return 1; 1308 /* 1309 * First try the shortest match for the '*' that 1310 * could work. We can forget any earlier '*' since 1311 * there is no way having it match more characters 1312 * can help us, given that we are already here. 1313 */ 1314 bt_p = p; 1315 bt_q = q; 1316 break; 1317 case '[': { 1318 const char *savep, *saveq; 1319 int invert, found; 1320 wchar_t chr; 1321 1322 savep = p, saveq = q; 1323 invert = 0; 1324 if (*p == '!' || *p == '^') { 1325 invert++; 1326 p++; 1327 } 1328 found = 0; 1329 if (*q == '\0') 1330 return 0; 1331 if (localeisutf8) { 1332 chr = get_wc(&q); 1333 if (chr == 0) 1334 goto backtrack; 1335 } else 1336 chr = (unsigned char)*q++; 1337 c = *p++; 1338 do { 1339 if (c == '\0') { 1340 p = savep, q = saveq; 1341 c = '['; 1342 goto dft; 1343 } 1344 if (c == '[' && *p == ':') { 1345 found |= match_charclass(p, chr, &end); 1346 if (end != NULL) 1347 p = end; 1348 } 1349 if (c == CTLESC) 1350 c = *p++; 1351 if (localeisutf8 && c & 0x80) { 1352 p--; 1353 wc = get_wc(&p); 1354 if (wc == 0) /* bad utf-8 */ 1355 return 0; 1356 } else 1357 wc = (unsigned char)c; 1358 if (*p == '-' && p[1] != ']') { 1359 p++; 1360 if (*p == CTLESC) 1361 p++; 1362 if (localeisutf8) { 1363 wc2 = get_wc(&p); 1364 if (wc2 == 0) /* bad utf-8 */ 1365 return 0; 1366 } else 1367 wc2 = (unsigned char)*p++; 1368 if ( collate_range_cmp(chr, wc) >= 0 1369 && collate_range_cmp(chr, wc2) <= 0 1370 ) 1371 found = 1; 1372 } else { 1373 if (chr == wc) 1374 found = 1; 1375 } 1376 } while ((c = *p++) != ']'); 1377 if (found == invert) 1378 goto backtrack; 1379 break; 1380 } 1381 dft: default: 1382 if (*q == '\0') 1383 return 0; 1384 if (*q++ == c) 1385 break; 1386 backtrack: 1387 /* 1388 * If we have a mismatch (other than hitting the end 1389 * of the string), go back to the last '*' seen and 1390 * have it match one additional character. 1391 */ 1392 if (bt_p == NULL) 1393 return 0; 1394 if (*bt_q == '\0') 1395 return 0; 1396 bt_q++; 1397 p = bt_p; 1398 q = bt_q; 1399 break; 1400 } 1401 } 1402 } 1403 1404 1405 1406 /* 1407 * Remove any CTLESC and CTLQUOTEMARK characters from a string. 1408 */ 1409 1410 void 1411 rmescapes(char *str) 1412 { 1413 char *p, *q; 1414 1415 p = str; 1416 while (*p != CTLESC && *p != CTLQUOTEMARK && *p != CTLQUOTEEND) { 1417 if (*p++ == '\0') 1418 return; 1419 } 1420 q = p; 1421 while (*p) { 1422 if (*p == CTLQUOTEMARK || *p == CTLQUOTEEND) { 1423 p++; 1424 continue; 1425 } 1426 if (*p == CTLESC) 1427 p++; 1428 *q++ = *p++; 1429 } 1430 *q = '\0'; 1431 } 1432 1433 1434 1435 /* 1436 * See if a pattern matches in a case statement. 1437 */ 1438 1439 int 1440 casematch(union node *pattern, const char *val) 1441 { 1442 struct stackmark smark; 1443 int result; 1444 char *p; 1445 1446 setstackmark(&smark); 1447 argbackq = pattern->narg.backquote; 1448 STARTSTACKSTR(expdest); 1449 argstr(pattern->narg.text, EXP_TILDE | EXP_CASE, NULL); 1450 STPUTC('\0', expdest); 1451 p = grabstackstr(expdest); 1452 result = patmatch(p, val); 1453 popstackmark(&smark); 1454 return result; 1455 } 1456 1457 /* 1458 * Our own itoa(). 1459 */ 1460 1461 static void 1462 cvtnum(int num, char *buf) 1463 { 1464 char temp[32]; 1465 int neg = num < 0; 1466 char *p = temp + 31; 1467 1468 temp[31] = '\0'; 1469 1470 do { 1471 *--p = num % 10 + '0'; 1472 } while ((num /= 10) != 0); 1473 1474 if (neg) 1475 *--p = '-'; 1476 1477 memcpy(buf, p, temp + 32 - p); 1478 } 1479 1480 /* 1481 * Do most of the work for wordexp(3). 1482 */ 1483 1484 int 1485 wordexpcmd(int argc, char **argv) 1486 { 1487 size_t len; 1488 int i; 1489 1490 out1fmt("%08x", argc - 1); 1491 for (i = 1, len = 0; i < argc; i++) 1492 len += strlen(argv[i]); 1493 out1fmt("%08x", (int)len); 1494 for (i = 1; i < argc; i++) 1495 outbin(argv[i], strlen(argv[i]) + 1, out1); 1496 return (0); 1497 } 1498 1499 /* 1500 * Do most of the work for wordexp(3), new version. 1501 */ 1502 1503 int 1504 freebsd_wordexpcmd(int argc __unused, char **argv __unused) 1505 { 1506 struct arglist arglist; 1507 union node *args, *n; 1508 size_t len; 1509 int ch; 1510 int protected = 0; 1511 int fd = -1; 1512 int i; 1513 1514 while ((ch = nextopt("f:p")) != '\0') { 1515 switch (ch) { 1516 case 'f': 1517 fd = number(shoptarg); 1518 break; 1519 case 'p': 1520 protected = 1; 1521 break; 1522 } 1523 } 1524 if (*argptr != NULL) 1525 error("wrong number of arguments"); 1526 if (fd < 0) 1527 error("missing fd"); 1528 INTOFF; 1529 setinputfd(fd, 1); 1530 INTON; 1531 args = parsewordexp(); 1532 popfile(); /* will also close fd */ 1533 if (protected) 1534 for (n = args; n != NULL; n = n->narg.next) { 1535 if (n->narg.backquote != NULL) { 1536 outcslow('C', out1); 1537 error("command substitution disabled"); 1538 } 1539 } 1540 outcslow(' ', out1); 1541 emptyarglist(&arglist); 1542 for (n = args; n != NULL; n = n->narg.next) 1543 expandarg(n, &arglist, EXP_FULL | EXP_TILDE); 1544 for (i = 0, len = 0; i < arglist.count; i++) 1545 len += strlen(arglist.args[i]); 1546 out1fmt("%016x %016zx", arglist.count, len); 1547 for (i = 0; i < arglist.count; i++) 1548 outbin(arglist.args[i], strlen(arglist.args[i]) + 1, out1); 1549 return (0); 1550 } 1551