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