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 93 static const char *argstr(const char *, struct nodelist **restrict, int, 94 struct worddest *); 95 static const char *exptilde(const char *, int); 96 static const char *expari(const char *, struct nodelist **restrict, int, 97 struct worddest *); 98 static void expbackq(union node *, int, int, struct worddest *); 99 static const char *subevalvar_trim(const char *, struct nodelist **restrict, 100 int, int, int); 101 static const char *subevalvar_misc(const char *, struct nodelist **restrict, 102 const char *, int, int, int); 103 static const char *evalvar(const char *, struct nodelist **restrict, int, 104 struct worddest *); 105 static int varisset(const char *, int); 106 static void strtodest(const char *, int, int, int, struct worddest *); 107 static void reprocess(int, int, int, int, struct worddest *); 108 static void varvalue(const char *, int, int, int, struct worddest *); 109 static void expandmeta(char *, struct arglist *); 110 static void expmeta(char *, char *, struct arglist *); 111 static int expsortcmp(const void *, const void *); 112 static int patmatch(const char *, const char *); 113 static void cvtnum(int, char *); 114 static int collate_range_cmp(wchar_t, wchar_t); 115 116 void 117 emptyarglist(struct arglist *list) 118 { 119 120 list->args = list->smallarg; 121 list->count = 0; 122 list->capacity = sizeof(list->smallarg) / sizeof(list->smallarg[0]); 123 } 124 125 void 126 appendarglist(struct arglist *list, char *str) 127 { 128 char **newargs; 129 int newcapacity; 130 131 if (list->count >= list->capacity) { 132 newcapacity = list->capacity * 2; 133 if (newcapacity < 16) 134 newcapacity = 16; 135 if (newcapacity > INT_MAX / (int)sizeof(newargs[0])) 136 error("Too many entries in arglist"); 137 newargs = stalloc(newcapacity * sizeof(newargs[0])); 138 memcpy(newargs, list->args, list->count * sizeof(newargs[0])); 139 list->args = newargs; 140 list->capacity = newcapacity; 141 } 142 list->args[list->count++] = str; 143 } 144 145 static int 146 collate_range_cmp(wchar_t c1, wchar_t c2) 147 { 148 wchar_t s1[2], s2[2]; 149 150 s1[0] = c1; 151 s1[1] = L'\0'; 152 s2[0] = c2; 153 s2[1] = L'\0'; 154 return (wcscoll(s1, s2)); 155 } 156 157 static char * 158 stputs_quotes(const char *data, const char *syntax, char *p) 159 { 160 while (*data) { 161 CHECKSTRSPACE(2, p); 162 if (syntax[(int)*data] == CCTL) 163 USTPUTC(CTLESC, p); 164 USTPUTC(*data++, p); 165 } 166 return (p); 167 } 168 #define STPUTS_QUOTES(data, syntax, p) p = stputs_quotes((data), syntax, p) 169 170 static char * 171 nextword(char c, int flag, char *p, struct worddest *dst) 172 { 173 int is_ws; 174 175 is_ws = c == '\t' || c == '\n' || c == ' '; 176 if (p != stackblock() || (is_ws ? dst->state == WORD_QUOTEMARK : 177 dst->state != WORD_WS_DELIMITED) || c == '\0') { 178 STPUTC('\0', p); 179 if (flag & EXP_GLOB) 180 expandmeta(grabstackstr(p), dst->list); 181 else 182 appendarglist(dst->list, grabstackstr(p)); 183 dst->state = is_ws ? WORD_WS_DELIMITED : WORD_IDLE; 184 } else if (!is_ws && dst->state == WORD_WS_DELIMITED) 185 dst->state = WORD_IDLE; 186 /* Reserve space while the stack string is empty. */ 187 appendarglist(dst->list, NULL); 188 dst->list->count--; 189 STARTSTACKSTR(p); 190 return p; 191 } 192 #define NEXTWORD(c, flag, p, dstlist) p = nextword(c, flag, p, dstlist) 193 194 static char * 195 stputs_split(const char *data, const char *syntax, int flag, char *p, 196 struct worddest *dst) 197 { 198 const char *ifs; 199 char c; 200 201 ifs = ifsset() ? ifsval() : " \t\n"; 202 while (*data) { 203 CHECKSTRSPACE(2, p); 204 c = *data++; 205 if (strchr(ifs, c) != NULL) { 206 NEXTWORD(c, flag, p, dst); 207 continue; 208 } 209 if (flag & EXP_GLOB && syntax[(int)c] == CCTL) 210 USTPUTC(CTLESC, p); 211 USTPUTC(c, p); 212 } 213 return (p); 214 } 215 #define STPUTS_SPLIT(data, syntax, flag, p, dst) p = stputs_split((data), syntax, flag, p, dst) 216 217 /* 218 * Perform expansions on an argument, placing the resulting list of arguments 219 * in arglist. Parameter expansion, command substitution and arithmetic 220 * expansion are always performed; additional expansions can be requested 221 * via flag (EXP_*). 222 * The result is left in the stack string. 223 * When arglist is NULL, perform here document expansion. 224 * 225 * Caution: this function uses global state and is not reentrant. 226 * However, a new invocation after an interrupted invocation is safe 227 * and will reset the global state for the new call. 228 */ 229 void 230 expandarg(union node *arg, struct arglist *arglist, int flag) 231 { 232 struct worddest exparg; 233 struct nodelist *argbackq; 234 235 if (fflag) 236 flag &= ~EXP_GLOB; 237 argbackq = arg->narg.backquote; 238 exparg.list = arglist; 239 exparg.state = WORD_IDLE; 240 STARTSTACKSTR(expdest); 241 argstr(arg->narg.text, &argbackq, flag, &exparg); 242 if (arglist == NULL) { 243 STACKSTRNUL(expdest); 244 return; /* here document expanded */ 245 } 246 if ((flag & EXP_SPLIT) == 0 || expdest != stackblock() || 247 exparg.state == WORD_QUOTEMARK) { 248 STPUTC('\0', expdest); 249 if (flag & EXP_SPLIT) { 250 if (flag & EXP_GLOB) 251 expandmeta(grabstackstr(expdest), exparg.list); 252 else 253 appendarglist(exparg.list, grabstackstr(expdest)); 254 } 255 } 256 if ((flag & EXP_SPLIT) == 0) 257 appendarglist(arglist, grabstackstr(expdest)); 258 } 259 260 261 262 /* 263 * Perform parameter expansion, command substitution and arithmetic 264 * expansion, and tilde expansion if requested via EXP_TILDE/EXP_VARTILDE. 265 * Processing ends at a CTLENDVAR or CTLENDARI character as well as '\0'. 266 * This is used to expand word in ${var+word} etc. 267 * If EXP_GLOB or EXP_CASE are set, keep and/or generate CTLESC 268 * characters to allow for further processing. 269 * 270 * If EXP_SPLIT is set, dst receives any complete words produced. 271 */ 272 static const char * 273 argstr(const char *p, struct nodelist **restrict argbackq, int flag, 274 struct worddest *dst) 275 { 276 char c; 277 int quotes = flag & (EXP_GLOB | EXP_CASE); /* do CTLESC */ 278 int firsteq = 1; 279 int split_lit; 280 int lit_quoted; 281 282 split_lit = flag & EXP_SPLIT_LIT; 283 lit_quoted = flag & EXP_LIT_QUOTED; 284 flag &= ~(EXP_SPLIT_LIT | EXP_LIT_QUOTED); 285 if (*p == '~' && (flag & (EXP_TILDE | EXP_VARTILDE))) 286 p = exptilde(p, flag); 287 for (;;) { 288 CHECKSTRSPACE(2, expdest); 289 switch (c = *p++) { 290 case '\0': 291 return (p - 1); 292 case CTLENDVAR: 293 case CTLENDARI: 294 return (p); 295 case CTLQUOTEMARK: 296 lit_quoted = 1; 297 /* "$@" syntax adherence hack */ 298 if (p[0] == CTLVAR && (p[1] & VSQUOTE) != 0 && 299 p[2] == '@' && p[3] == '=') 300 break; 301 if ((flag & EXP_SPLIT) != 0 && expdest == stackblock()) 302 dst->state = WORD_QUOTEMARK; 303 break; 304 case CTLQUOTEEND: 305 lit_quoted = 0; 306 break; 307 case CTLESC: 308 c = *p++; 309 if (split_lit && !lit_quoted && 310 strchr(ifsset() ? ifsval() : " \t\n", c) != NULL) { 311 NEXTWORD(c, flag, expdest, dst); 312 break; 313 } 314 if (quotes) 315 USTPUTC(CTLESC, expdest); 316 USTPUTC(c, expdest); 317 break; 318 case CTLVAR: 319 p = evalvar(p, argbackq, flag, dst); 320 break; 321 case CTLBACKQ: 322 case CTLBACKQ|CTLQUOTE: 323 expbackq((*argbackq)->n, c & CTLQUOTE, flag, dst); 324 *argbackq = (*argbackq)->next; 325 break; 326 case CTLARI: 327 p = expari(p, argbackq, flag, dst); 328 break; 329 case ':': 330 case '=': 331 /* 332 * sort of a hack - expand tildes in variable 333 * assignments (after the first '=' and after ':'s). 334 */ 335 if (split_lit && !lit_quoted && 336 strchr(ifsset() ? ifsval() : " \t\n", c) != NULL) { 337 NEXTWORD(c, flag, expdest, dst); 338 break; 339 } 340 USTPUTC(c, expdest); 341 if (flag & EXP_VARTILDE && *p == '~' && 342 (c != '=' || firsteq)) { 343 if (c == '=') 344 firsteq = 0; 345 p = exptilde(p, flag); 346 } 347 break; 348 default: 349 if (split_lit && !lit_quoted && 350 strchr(ifsset() ? ifsval() : " \t\n", c) != NULL) { 351 NEXTWORD(c, flag, expdest, dst); 352 break; 353 } 354 USTPUTC(c, expdest); 355 } 356 } 357 } 358 359 /* 360 * Perform tilde expansion, placing the result in the stack string and 361 * returning the next position in the input string to process. 362 */ 363 static const char * 364 exptilde(const char *p, int flag) 365 { 366 char c; 367 const char *startp = p; 368 const char *user; 369 struct passwd *pw; 370 char *home; 371 int len; 372 373 for (;;) { 374 c = *p; 375 switch(c) { 376 case CTLESC: /* This means CTL* are always considered quoted. */ 377 case CTLVAR: 378 case CTLBACKQ: 379 case CTLBACKQ | CTLQUOTE: 380 case CTLARI: 381 case CTLENDARI: 382 case CTLQUOTEMARK: 383 return (startp); 384 case ':': 385 if ((flag & EXP_VARTILDE) == 0) 386 break; 387 /* FALLTHROUGH */ 388 case '\0': 389 case '/': 390 case CTLENDVAR: 391 len = p - startp - 1; 392 STPUTBIN(startp + 1, len, expdest); 393 STACKSTRNUL(expdest); 394 user = expdest - len; 395 if (*user == '\0') { 396 home = lookupvar("HOME"); 397 } else { 398 pw = getpwnam(user); 399 home = pw != NULL ? pw->pw_dir : NULL; 400 } 401 STADJUST(-len, expdest); 402 if (home == NULL || *home == '\0') 403 return (startp); 404 strtodest(home, flag, VSNORMAL, 1, NULL); 405 return (p); 406 } 407 p++; 408 } 409 } 410 411 412 /* 413 * Expand arithmetic expression. 414 */ 415 static const char * 416 expari(const char *p, struct nodelist **restrict argbackq, int flag, 417 struct worddest *dst) 418 { 419 char *q, *start; 420 arith_t result; 421 int begoff; 422 int quoted; 423 int adj; 424 425 quoted = *p++ == '"'; 426 begoff = expdest - stackblock(); 427 p = argstr(p, argbackq, 0, NULL); 428 STPUTC('\0', expdest); 429 start = stackblock() + begoff; 430 431 q = grabstackstr(expdest); 432 result = arith(start); 433 ungrabstackstr(q, expdest); 434 435 start = stackblock() + begoff; 436 adj = start - expdest; 437 STADJUST(adj, expdest); 438 439 CHECKSTRSPACE((int)(DIGITS(result) + 1), expdest); 440 fmtstr(expdest, DIGITS(result), ARITH_FORMAT_STR, result); 441 adj = strlen(expdest); 442 STADJUST(adj, expdest); 443 if (!quoted) 444 reprocess(expdest - adj - stackblock(), flag, VSNORMAL, 0, dst); 445 return p; 446 } 447 448 449 /* 450 * Perform command substitution. 451 */ 452 static void 453 expbackq(union node *cmd, int quoted, int flag, struct worddest *dst) 454 { 455 struct backcmd in; 456 int i; 457 char buf[128]; 458 char *p; 459 char *dest = expdest; 460 char lastc; 461 char const *syntax = quoted? DQSYNTAX : BASESYNTAX; 462 int quotes = flag & (EXP_GLOB | EXP_CASE); 463 size_t nnl; 464 const char *ifs; 465 466 INTOFF; 467 p = grabstackstr(dest); 468 evalbackcmd(cmd, &in); 469 ungrabstackstr(p, dest); 470 471 p = in.buf; 472 nnl = 0; 473 if (!quoted && flag & EXP_SPLIT) 474 ifs = ifsset() ? ifsval() : " \t\n"; 475 else 476 ifs = ""; 477 /* Don't copy trailing newlines */ 478 for (;;) { 479 if (--in.nleft < 0) { 480 if (in.fd < 0) 481 break; 482 while ((i = read(in.fd, buf, sizeof buf)) < 0 && errno == EINTR) 483 ; 484 TRACE(("expbackq: read returns %d\n", i)); 485 if (i <= 0) 486 break; 487 p = buf; 488 in.nleft = i - 1; 489 } 490 lastc = *p++; 491 if (lastc == '\0') 492 continue; 493 if (lastc == '\n') { 494 nnl++; 495 } else { 496 if (nnl > 0) { 497 if (strchr(ifs, '\n') != NULL) { 498 NEXTWORD('\n', flag, dest, dst); 499 nnl = 0; 500 } else { 501 CHECKSTRSPACE(nnl + 2, dest); 502 while (nnl > 0) { 503 nnl--; 504 USTPUTC('\n', dest); 505 } 506 } 507 } 508 if (strchr(ifs, lastc) != NULL) 509 NEXTWORD(lastc, flag, dest, dst); 510 else { 511 CHECKSTRSPACE(2, dest); 512 if (quotes && syntax[(int)lastc] == CCTL) 513 USTPUTC(CTLESC, dest); 514 USTPUTC(lastc, dest); 515 } 516 } 517 } 518 519 if (in.fd >= 0) 520 close(in.fd); 521 if (in.buf) 522 ckfree(in.buf); 523 if (in.jp) { 524 p = grabstackstr(dest); 525 exitstatus = waitforjob(in.jp, (int *)NULL); 526 ungrabstackstr(p, dest); 527 } 528 TRACE(("expbackq: done\n")); 529 expdest = dest; 530 INTON; 531 } 532 533 534 535 static void 536 recordleft(const char *str, const char *loc, char *startp) 537 { 538 int amount; 539 540 amount = ((str - 1) - (loc - startp)) - expdest; 541 STADJUST(amount, expdest); 542 while (loc != str - 1) 543 *startp++ = *loc++; 544 } 545 546 static const char * 547 subevalvar_trim(const char *p, struct nodelist **restrict argbackq, int strloc, 548 int subtype, int startloc) 549 { 550 char *startp; 551 char *loc = NULL; 552 char *str; 553 int c = 0; 554 int amount; 555 556 p = argstr(p, argbackq, EXP_CASE | EXP_TILDE, NULL); 557 STACKSTRNUL(expdest); 558 startp = stackblock() + startloc; 559 str = stackblock() + strloc; 560 561 switch (subtype) { 562 case VSTRIMLEFT: 563 for (loc = startp; loc < str; loc++) { 564 c = *loc; 565 *loc = '\0'; 566 if (patmatch(str, startp)) { 567 *loc = c; 568 recordleft(str, loc, startp); 569 return p; 570 } 571 *loc = c; 572 } 573 break; 574 575 case VSTRIMLEFTMAX: 576 for (loc = str - 1; loc >= startp;) { 577 c = *loc; 578 *loc = '\0'; 579 if (patmatch(str, startp)) { 580 *loc = c; 581 recordleft(str, loc, startp); 582 return p; 583 } 584 *loc = c; 585 loc--; 586 } 587 break; 588 589 case VSTRIMRIGHT: 590 for (loc = str - 1; loc >= startp;) { 591 if (patmatch(str, loc)) { 592 amount = loc - expdest; 593 STADJUST(amount, expdest); 594 return p; 595 } 596 loc--; 597 } 598 break; 599 600 case VSTRIMRIGHTMAX: 601 for (loc = startp; loc < str - 1; loc++) { 602 if (patmatch(str, loc)) { 603 amount = loc - expdest; 604 STADJUST(amount, expdest); 605 return p; 606 } 607 } 608 break; 609 610 611 default: 612 abort(); 613 } 614 amount = (expdest - stackblock() - strloc) + 1; 615 STADJUST(-amount, expdest); 616 return p; 617 } 618 619 620 static const char * 621 subevalvar_misc(const char *p, struct nodelist **restrict argbackq, 622 const char *var, int subtype, int startloc, int varflags) 623 { 624 char *startp; 625 int amount; 626 627 p = argstr(p, argbackq, EXP_TILDE, NULL); 628 STACKSTRNUL(expdest); 629 startp = stackblock() + startloc; 630 631 switch (subtype) { 632 case VSASSIGN: 633 setvar(var, startp, 0); 634 amount = startp - expdest; 635 STADJUST(amount, expdest); 636 return p; 637 638 case VSQUESTION: 639 if (*p != CTLENDVAR) { 640 outfmt(out2, "%s\n", startp); 641 error((char *)NULL); 642 } 643 error("%.*s: parameter %snot set", (int)(p - var - 1), 644 var, (varflags & VSNUL) ? "null or " : ""); 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, struct nodelist **restrict argbackq, int flag, 659 struct worddest *dst) 660 { 661 int subtype; 662 int varflags; 663 const char *var; 664 const char *val; 665 int patloc; 666 int c; 667 int set; 668 int special; 669 int startloc; 670 int varlen; 671 int varlenb; 672 char buf[21]; 673 674 varflags = (unsigned char)*p++; 675 subtype = varflags & VSTYPE; 676 var = p; 677 special = 0; 678 if (! is_name(*p)) 679 special = 1; 680 p = strchr(p, '=') + 1; 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 return p; 758 759 case VSPLUS: 760 case VSMINUS: 761 if (!set) { 762 return argstr(p, argbackq, 763 flag | (flag & EXP_SPLIT ? EXP_SPLIT_LIT : 0) | 764 (varflags & VSQUOTE ? EXP_LIT_QUOTED : 0), dst); 765 } 766 break; 767 768 case VSTRIMLEFT: 769 case VSTRIMLEFTMAX: 770 case VSTRIMRIGHT: 771 case VSTRIMRIGHTMAX: 772 if (!set) 773 break; 774 /* 775 * Terminate the string and start recording the pattern 776 * right after it 777 */ 778 STPUTC('\0', expdest); 779 patloc = expdest - stackblock(); 780 p = subevalvar_trim(p, argbackq, patloc, subtype, startloc); 781 reprocess(startloc, flag, VSNORMAL, varflags & VSQUOTE, dst); 782 if (flag & EXP_SPLIT && *var == '@' && varflags & VSQUOTE) 783 dst->state = WORD_QUOTEMARK; 784 return p; 785 786 case VSASSIGN: 787 case VSQUESTION: 788 if (!set) { 789 p = subevalvar_misc(p, argbackq, var, subtype, 790 startloc, varflags); 791 /* assert(subtype == VSASSIGN); */ 792 val = lookupvar(var); 793 strtodest(val, flag, subtype, varflags & VSQUOTE, dst); 794 return p; 795 } 796 break; 797 798 case VSERROR: 799 c = p - var - 1; 800 error("${%.*s%s}: Bad substitution", c, var, 801 (c > 0 && *p != CTLENDVAR) ? "..." : ""); 802 803 default: 804 abort(); 805 } 806 807 { /* skip to end of alternative */ 808 int nesting = 1; 809 for (;;) { 810 if ((c = *p++) == CTLESC) 811 p++; 812 else if (c == CTLBACKQ || c == (CTLBACKQ|CTLQUOTE)) 813 *argbackq = (*argbackq)->next; 814 else if (c == CTLVAR) { 815 if ((*p++ & VSTYPE) != VSNORMAL) 816 nesting++; 817 } else if (c == CTLENDVAR) { 818 if (--nesting == 0) 819 break; 820 } 821 } 822 } 823 return p; 824 } 825 826 827 828 /* 829 * Test whether a specialized variable is set. 830 */ 831 832 static int 833 varisset(const char *name, int nulok) 834 { 835 836 if (*name == '!') 837 return backgndpidset(); 838 else if (*name == '@' || *name == '*') { 839 if (*shellparam.p == NULL) 840 return 0; 841 842 if (nulok) { 843 char **av; 844 845 for (av = shellparam.p; *av; av++) 846 if (**av != '\0') 847 return 1; 848 return 0; 849 } 850 } else if (is_digit(*name)) { 851 char *ap; 852 long num; 853 854 errno = 0; 855 num = strtol(name, NULL, 10); 856 if (errno != 0 || num > shellparam.nparam) 857 return 0; 858 859 if (num == 0) 860 ap = arg0; 861 else 862 ap = shellparam.p[num - 1]; 863 864 if (nulok && (ap == NULL || *ap == '\0')) 865 return 0; 866 } 867 return 1; 868 } 869 870 static void 871 strtodest(const char *p, int flag, int subtype, int quoted, 872 struct worddest *dst) 873 { 874 if (subtype == VSLENGTH || subtype == VSTRIMLEFT || 875 subtype == VSTRIMLEFTMAX || subtype == VSTRIMRIGHT || 876 subtype == VSTRIMRIGHTMAX) 877 STPUTS(p, expdest); 878 else if (flag & EXP_SPLIT && !quoted && dst != NULL) 879 STPUTS_SPLIT(p, BASESYNTAX, flag, expdest, dst); 880 else if (flag & (EXP_GLOB | EXP_CASE)) 881 STPUTS_QUOTES(p, quoted ? DQSYNTAX : BASESYNTAX, expdest); 882 else 883 STPUTS(p, expdest); 884 } 885 886 static void 887 reprocess(int startloc, int flag, int subtype, int quoted, 888 struct worddest *dst) 889 { 890 static char *buf = NULL; 891 static size_t buflen = 0; 892 char *startp; 893 size_t len, zpos, zlen; 894 895 startp = stackblock() + startloc; 896 len = expdest - startp; 897 if (len >= SIZE_MAX / 2) 898 abort(); 899 INTOFF; 900 if (len >= buflen) { 901 ckfree(buf); 902 buf = NULL; 903 } 904 if (buflen < 128) 905 buflen = 128; 906 while (len >= buflen) 907 buflen <<= 1; 908 if (buf == NULL) 909 buf = ckmalloc(buflen); 910 INTON; 911 memcpy(buf, startp, len); 912 buf[len] = '\0'; 913 STADJUST(-len, expdest); 914 for (zpos = 0;;) { 915 zlen = strlen(buf + zpos); 916 strtodest(buf + zpos, flag, subtype, quoted, dst); 917 zpos += zlen + 1; 918 if (zpos == len + 1) 919 break; 920 if (flag & EXP_SPLIT && (quoted || (zlen > 0 && zpos < len))) 921 NEXTWORD('\0', flag, expdest, dst); 922 } 923 } 924 925 /* 926 * Add the value of a specialized variable to the stack string. 927 */ 928 929 static void 930 varvalue(const char *name, int quoted, int subtype, int flag, 931 struct worddest *dst) 932 { 933 int num; 934 char *p; 935 int i; 936 int splitlater; 937 char sep[2]; 938 char **ap; 939 char buf[(NSHORTOPTS > 10 ? NSHORTOPTS : 10) + 1]; 940 941 if (subtype == VSLENGTH) 942 flag &= ~EXP_FULL; 943 splitlater = subtype == VSTRIMLEFT || subtype == VSTRIMLEFTMAX || 944 subtype == VSTRIMRIGHT || subtype == VSTRIMRIGHTMAX; 945 946 switch (*name) { 947 case '$': 948 num = rootpid; 949 break; 950 case '?': 951 num = oexitstatus; 952 break; 953 case '#': 954 num = shellparam.nparam; 955 break; 956 case '!': 957 num = backgndpidval(); 958 break; 959 case '-': 960 p = buf; 961 for (i = 0 ; i < NSHORTOPTS ; i++) { 962 if (optval[i]) 963 *p++ = optletter[i]; 964 } 965 *p = '\0'; 966 strtodest(buf, flag, subtype, quoted, dst); 967 return; 968 case '@': 969 if (flag & EXP_SPLIT && quoted) { 970 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 971 strtodest(p, flag, subtype, quoted, dst); 972 if (*ap) { 973 if (splitlater) 974 STPUTC('\0', expdest); 975 else 976 NEXTWORD('\0', flag, expdest, 977 dst); 978 } 979 } 980 if (shellparam.nparam > 0) 981 dst->state = WORD_QUOTEMARK; 982 return; 983 } 984 /* FALLTHROUGH */ 985 case '*': 986 if (ifsset()) 987 sep[0] = ifsval()[0]; 988 else 989 sep[0] = ' '; 990 sep[1] = '\0'; 991 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 992 strtodest(p, flag, subtype, quoted, dst); 993 if (!*ap) 994 break; 995 if (sep[0]) 996 strtodest(sep, flag, subtype, quoted, dst); 997 else if (flag & EXP_SPLIT && !quoted && **ap != '\0') { 998 if (splitlater) 999 STPUTC('\0', expdest); 1000 else 1001 NEXTWORD('\0', flag, expdest, dst); 1002 } 1003 } 1004 return; 1005 default: 1006 if (is_digit(*name)) { 1007 num = atoi(name); 1008 if (num == 0) 1009 p = arg0; 1010 else if (num > 0 && num <= shellparam.nparam) 1011 p = shellparam.p[num - 1]; 1012 else 1013 return; 1014 strtodest(p, flag, subtype, quoted, dst); 1015 } 1016 return; 1017 } 1018 cvtnum(num, buf); 1019 strtodest(buf, flag, subtype, quoted, dst); 1020 } 1021 1022 1023 1024 static char expdir[PATH_MAX]; 1025 #define expdir_end (expdir + sizeof(expdir)) 1026 1027 /* 1028 * Perform pathname generation and remove control characters. 1029 * At this point, the only control characters should be CTLESC. 1030 * The results are stored in the list dstlist. 1031 */ 1032 static void 1033 expandmeta(char *pattern, struct arglist *dstlist) 1034 { 1035 char *p; 1036 int firstmatch; 1037 char c; 1038 1039 firstmatch = dstlist->count; 1040 p = pattern; 1041 for (; (c = *p) != '\0'; p++) { 1042 /* fast check for meta chars */ 1043 if (c == '*' || c == '?' || c == '[') { 1044 INTOFF; 1045 expmeta(expdir, pattern, dstlist); 1046 INTON; 1047 break; 1048 } 1049 } 1050 if (dstlist->count == firstmatch) { 1051 /* 1052 * no matches 1053 */ 1054 rmescapes(pattern); 1055 appendarglist(dstlist, pattern); 1056 } else { 1057 qsort(&dstlist->args[firstmatch], 1058 dstlist->count - firstmatch, 1059 sizeof(dstlist->args[0]), expsortcmp); 1060 } 1061 } 1062 1063 1064 /* 1065 * Do metacharacter (i.e. *, ?, [...]) expansion. 1066 */ 1067 1068 static void 1069 expmeta(char *enddir, char *name, struct arglist *arglist) 1070 { 1071 const char *p; 1072 const char *q; 1073 const char *start; 1074 char *endname; 1075 int metaflag; 1076 struct stat statb; 1077 DIR *dirp; 1078 struct dirent *dp; 1079 int atend; 1080 int matchdot; 1081 int esc; 1082 int namlen; 1083 1084 metaflag = 0; 1085 start = name; 1086 for (p = name; esc = 0, *p; p += esc + 1) { 1087 if (*p == '*' || *p == '?') 1088 metaflag = 1; 1089 else if (*p == '[') { 1090 q = p + 1; 1091 if (*q == '!' || *q == '^') 1092 q++; 1093 for (;;) { 1094 if (*q == CTLESC) 1095 q++; 1096 if (*q == '/' || *q == '\0') 1097 break; 1098 if (*++q == ']') { 1099 metaflag = 1; 1100 break; 1101 } 1102 } 1103 } else if (*p == '\0') 1104 break; 1105 else { 1106 if (*p == CTLESC) 1107 esc++; 1108 if (p[esc] == '/') { 1109 if (metaflag) 1110 break; 1111 start = p + esc + 1; 1112 } 1113 } 1114 } 1115 if (metaflag == 0) { /* we've reached the end of the file name */ 1116 if (enddir != expdir) 1117 metaflag++; 1118 for (p = name ; ; p++) { 1119 if (*p == CTLESC) 1120 p++; 1121 *enddir++ = *p; 1122 if (*p == '\0') 1123 break; 1124 if (enddir == expdir_end) 1125 return; 1126 } 1127 if (metaflag == 0 || lstat(expdir, &statb) >= 0) 1128 appendarglist(arglist, stsavestr(expdir)); 1129 return; 1130 } 1131 endname = name + (p - name); 1132 if (start != name) { 1133 p = name; 1134 while (p < start) { 1135 if (*p == CTLESC) 1136 p++; 1137 *enddir++ = *p++; 1138 if (enddir == expdir_end) 1139 return; 1140 } 1141 } 1142 if (enddir == expdir) { 1143 p = "."; 1144 } else if (enddir == expdir + 1 && *expdir == '/') { 1145 p = "/"; 1146 } else { 1147 p = expdir; 1148 enddir[-1] = '\0'; 1149 } 1150 if ((dirp = opendir(p)) == NULL) 1151 return; 1152 if (enddir != expdir) 1153 enddir[-1] = '/'; 1154 if (*endname == 0) { 1155 atend = 1; 1156 } else { 1157 atend = 0; 1158 *endname = '\0'; 1159 endname += esc + 1; 1160 } 1161 matchdot = 0; 1162 p = start; 1163 if (*p == CTLESC) 1164 p++; 1165 if (*p == '.') 1166 matchdot++; 1167 while (! int_pending() && (dp = readdir(dirp)) != NULL) { 1168 if (dp->d_name[0] == '.' && ! matchdot) 1169 continue; 1170 if (patmatch(start, dp->d_name)) { 1171 namlen = dp->d_namlen; 1172 if (enddir + namlen + 1 > expdir_end) 1173 continue; 1174 memcpy(enddir, dp->d_name, namlen + 1); 1175 if (atend) 1176 appendarglist(arglist, stsavestr(expdir)); 1177 else { 1178 if (dp->d_type != DT_UNKNOWN && 1179 dp->d_type != DT_DIR && 1180 dp->d_type != DT_LNK) 1181 continue; 1182 if (enddir + namlen + 2 > expdir_end) 1183 continue; 1184 enddir[namlen] = '/'; 1185 enddir[namlen + 1] = '\0'; 1186 expmeta(enddir + namlen + 1, endname, arglist); 1187 } 1188 } 1189 } 1190 closedir(dirp); 1191 if (! atend) 1192 endname[-esc - 1] = esc ? CTLESC : '/'; 1193 } 1194 1195 1196 static int 1197 expsortcmp(const void *p1, const void *p2) 1198 { 1199 const char *s1 = *(const char * const *)p1; 1200 const char *s2 = *(const char * const *)p2; 1201 1202 return (strcoll(s1, s2)); 1203 } 1204 1205 1206 1207 static wchar_t 1208 get_wc(const char **p) 1209 { 1210 wchar_t c; 1211 int chrlen; 1212 1213 chrlen = mbtowc(&c, *p, 4); 1214 if (chrlen == 0) 1215 return 0; 1216 else if (chrlen == -1) 1217 c = 0; 1218 else 1219 *p += chrlen; 1220 return c; 1221 } 1222 1223 1224 /* 1225 * See if a character matches a character class, starting at the first colon 1226 * of "[:class:]". 1227 * If a valid character class is recognized, a pointer to the next character 1228 * after the final closing bracket is stored into *end, otherwise a null 1229 * pointer is stored into *end. 1230 */ 1231 static int 1232 match_charclass(const char *p, wchar_t chr, const char **end) 1233 { 1234 char name[20]; 1235 const char *nameend; 1236 wctype_t cclass; 1237 1238 *end = NULL; 1239 p++; 1240 nameend = strstr(p, ":]"); 1241 if (nameend == NULL || (size_t)(nameend - p) >= sizeof(name) || 1242 nameend == p) 1243 return 0; 1244 memcpy(name, p, nameend - p); 1245 name[nameend - p] = '\0'; 1246 *end = nameend + 2; 1247 cclass = wctype(name); 1248 /* An unknown class matches nothing but is valid nevertheless. */ 1249 if (cclass == 0) 1250 return 0; 1251 return iswctype(chr, cclass); 1252 } 1253 1254 1255 /* 1256 * Returns true if the pattern matches the string. 1257 */ 1258 1259 static int 1260 patmatch(const char *pattern, const char *string) 1261 { 1262 const char *p, *q, *end; 1263 const char *bt_p, *bt_q; 1264 char c; 1265 wchar_t wc, wc2; 1266 1267 p = pattern; 1268 q = string; 1269 bt_p = NULL; 1270 bt_q = NULL; 1271 for (;;) { 1272 switch (c = *p++) { 1273 case '\0': 1274 if (*q != '\0') 1275 goto backtrack; 1276 return 1; 1277 case CTLESC: 1278 if (*q++ != *p++) 1279 goto backtrack; 1280 break; 1281 case '?': 1282 if (*q == '\0') 1283 return 0; 1284 if (localeisutf8) { 1285 wc = get_wc(&q); 1286 /* 1287 * A '?' does not match invalid UTF-8 but a 1288 * '*' does, so backtrack. 1289 */ 1290 if (wc == 0) 1291 goto backtrack; 1292 } else 1293 q++; 1294 break; 1295 case '*': 1296 c = *p; 1297 while (c == '*') 1298 c = *++p; 1299 /* 1300 * If the pattern ends here, we know the string 1301 * matches without needing to look at the rest of it. 1302 */ 1303 if (c == '\0') 1304 return 1; 1305 /* 1306 * First try the shortest match for the '*' that 1307 * could work. We can forget any earlier '*' since 1308 * there is no way having it match more characters 1309 * can help us, given that we are already here. 1310 */ 1311 bt_p = p; 1312 bt_q = q; 1313 break; 1314 case '[': { 1315 const char *savep, *saveq; 1316 int invert, found; 1317 wchar_t chr; 1318 1319 savep = p, saveq = q; 1320 invert = 0; 1321 if (*p == '!' || *p == '^') { 1322 invert++; 1323 p++; 1324 } 1325 found = 0; 1326 if (*q == '\0') 1327 return 0; 1328 if (localeisutf8) { 1329 chr = get_wc(&q); 1330 if (chr == 0) 1331 goto backtrack; 1332 } else 1333 chr = (unsigned char)*q++; 1334 c = *p++; 1335 do { 1336 if (c == '\0') { 1337 p = savep, q = saveq; 1338 c = '['; 1339 goto dft; 1340 } 1341 if (c == '[' && *p == ':') { 1342 found |= match_charclass(p, chr, &end); 1343 if (end != NULL) 1344 p = end; 1345 } 1346 if (c == CTLESC) 1347 c = *p++; 1348 if (localeisutf8 && c & 0x80) { 1349 p--; 1350 wc = get_wc(&p); 1351 if (wc == 0) /* bad utf-8 */ 1352 return 0; 1353 } else 1354 wc = (unsigned char)c; 1355 if (*p == '-' && p[1] != ']') { 1356 p++; 1357 if (*p == CTLESC) 1358 p++; 1359 if (localeisutf8) { 1360 wc2 = get_wc(&p); 1361 if (wc2 == 0) /* bad utf-8 */ 1362 return 0; 1363 } else 1364 wc2 = (unsigned char)*p++; 1365 if ( collate_range_cmp(chr, wc) >= 0 1366 && collate_range_cmp(chr, wc2) <= 0 1367 ) 1368 found = 1; 1369 } else { 1370 if (chr == wc) 1371 found = 1; 1372 } 1373 } while ((c = *p++) != ']'); 1374 if (found == invert) 1375 goto backtrack; 1376 break; 1377 } 1378 dft: default: 1379 if (*q == '\0') 1380 return 0; 1381 if (*q++ == c) 1382 break; 1383 backtrack: 1384 /* 1385 * If we have a mismatch (other than hitting the end 1386 * of the string), go back to the last '*' seen and 1387 * have it match one additional character. 1388 */ 1389 if (bt_p == NULL) 1390 return 0; 1391 if (*bt_q == '\0') 1392 return 0; 1393 bt_q++; 1394 p = bt_p; 1395 q = bt_q; 1396 break; 1397 } 1398 } 1399 } 1400 1401 1402 1403 /* 1404 * Remove any CTLESC and CTLQUOTEMARK characters from a string. 1405 */ 1406 1407 void 1408 rmescapes(char *str) 1409 { 1410 char *p, *q; 1411 1412 p = str; 1413 while (*p != CTLESC && *p != CTLQUOTEMARK && *p != CTLQUOTEEND) { 1414 if (*p++ == '\0') 1415 return; 1416 } 1417 q = p; 1418 while (*p) { 1419 if (*p == CTLQUOTEMARK || *p == CTLQUOTEEND) { 1420 p++; 1421 continue; 1422 } 1423 if (*p == CTLESC) 1424 p++; 1425 *q++ = *p++; 1426 } 1427 *q = '\0'; 1428 } 1429 1430 1431 1432 /* 1433 * See if a pattern matches in a case statement. 1434 */ 1435 1436 int 1437 casematch(union node *pattern, const char *val) 1438 { 1439 struct stackmark smark; 1440 struct nodelist *argbackq; 1441 int result; 1442 char *p; 1443 1444 setstackmark(&smark); 1445 argbackq = pattern->narg.backquote; 1446 STARTSTACKSTR(expdest); 1447 argstr(pattern->narg.text, &argbackq, EXP_TILDE | EXP_CASE, NULL); 1448 STPUTC('\0', expdest); 1449 p = grabstackstr(expdest); 1450 result = patmatch(p, val); 1451 popstackmark(&smark); 1452 return result; 1453 } 1454 1455 /* 1456 * Our own itoa(). 1457 */ 1458 1459 static void 1460 cvtnum(int num, char *buf) 1461 { 1462 char temp[32]; 1463 int neg = num < 0; 1464 char *p = temp + 31; 1465 1466 temp[31] = '\0'; 1467 1468 do { 1469 *--p = num % 10 + '0'; 1470 } while ((num /= 10) != 0); 1471 1472 if (neg) 1473 *--p = '-'; 1474 1475 memcpy(buf, p, temp + 32 - p); 1476 } 1477 1478 /* 1479 * Do most of the work for wordexp(3). 1480 */ 1481 1482 int 1483 wordexpcmd(int argc, char **argv) 1484 { 1485 size_t len; 1486 int i; 1487 1488 out1fmt("%08x", argc - 1); 1489 for (i = 1, len = 0; i < argc; i++) 1490 len += strlen(argv[i]); 1491 out1fmt("%08x", (int)len); 1492 for (i = 1; i < argc; i++) 1493 outbin(argv[i], strlen(argv[i]) + 1, out1); 1494 return (0); 1495 } 1496 1497 /* 1498 * Do most of the work for wordexp(3), new version. 1499 */ 1500 1501 int 1502 freebsd_wordexpcmd(int argc __unused, char **argv __unused) 1503 { 1504 struct arglist arglist; 1505 union node *args, *n; 1506 size_t len; 1507 int ch; 1508 int protected = 0; 1509 int fd = -1; 1510 int i; 1511 1512 while ((ch = nextopt("f:p")) != '\0') { 1513 switch (ch) { 1514 case 'f': 1515 fd = number(shoptarg); 1516 break; 1517 case 'p': 1518 protected = 1; 1519 break; 1520 } 1521 } 1522 if (*argptr != NULL) 1523 error("wrong number of arguments"); 1524 if (fd < 0) 1525 error("missing fd"); 1526 INTOFF; 1527 setinputfd(fd, 1); 1528 INTON; 1529 args = parsewordexp(); 1530 popfile(); /* will also close fd */ 1531 if (protected) 1532 for (n = args; n != NULL; n = n->narg.next) { 1533 if (n->narg.backquote != NULL) { 1534 outcslow('C', out1); 1535 error("command substitution disabled"); 1536 } 1537 } 1538 outcslow(' ', out1); 1539 emptyarglist(&arglist); 1540 for (n = args; n != NULL; n = n->narg.next) 1541 expandarg(n, &arglist, EXP_FULL | EXP_TILDE); 1542 for (i = 0, len = 0; i < arglist.count; i++) 1543 len += strlen(arglist.args[i]); 1544 out1fmt("%016x %016zx", arglist.count, len); 1545 for (i = 0; i < arglist.count; i++) 1546 outbin(arglist.args[i], strlen(arglist.args[i]) + 1, out1); 1547 return (0); 1548 } 1549