expand.c revision 229220
1218885Sdim/*- 2218885Sdim * Copyright (c) 1991, 1993 3218885Sdim * The Regents of the University of California. All rights reserved. 4218885Sdim * Copyright (c) 1997-2005 5218885Sdim * Herbert Xu <herbert@gondor.apana.org.au>. All rights reserved. 6218885Sdim * 7218885Sdim * This code is derived from software contributed to Berkeley by 8218885Sdim * Kenneth Almquist. 9218885Sdim * 10263508Sdim * Redistribution and use in source and binary forms, with or without 11263508Sdim * modification, are permitted provided that the following conditions 12263508Sdim * are met: 13218885Sdim * 1. Redistributions of source code must retain the above copyright 14218885Sdim * notice, this list of conditions and the following disclaimer. 15218885Sdim * 2. Redistributions in binary form must reproduce the above copyright 16263508Sdim * notice, this list of conditions and the following disclaimer in the 17263508Sdim * documentation and/or other materials provided with the distribution. 18263508Sdim * 4. Neither the name of the University nor the names of its contributors 19263508Sdim * may be used to endorse or promote products derived from this software 20263508Sdim * without specific prior written permission. 21263508Sdim * 22263508Sdim * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 23263508Sdim * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 24263508Sdim * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 25263508Sdim * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 26263508Sdim * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 27263508Sdim * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 28263508Sdim * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 29263508Sdim * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 30263508Sdim * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 31263508Sdim * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 32263508Sdim * SUCH DAMAGE. 33263508Sdim */ 34263508Sdim 35263508Sdim#ifndef lint 36263508Sdim#if 0 37263508Sdimstatic char sccsid[] = "@(#)expand.c 8.5 (Berkeley) 5/15/95"; 38263508Sdim#endif 39263508Sdim#endif /* not lint */ 40263508Sdim#include <sys/cdefs.h> 41263508Sdim__FBSDID("$FreeBSD: head/bin/sh/expand.c 229220 2012-01-01 22:17:12Z jilles $"); 42263508Sdim 43263508Sdim#include <sys/types.h> 44263508Sdim#include <sys/time.h> 45263508Sdim#include <sys/stat.h> 46263508Sdim#include <dirent.h> 47263508Sdim#include <errno.h> 48263508Sdim#include <inttypes.h> 49263508Sdim#include <limits.h> 50263508Sdim#include <pwd.h> 51263508Sdim#include <stdio.h> 52263508Sdim#include <stdlib.h> 53263508Sdim#include <string.h> 54263508Sdim#include <unistd.h> 55263508Sdim#include <wchar.h> 56263508Sdim#include <wctype.h> 57263508Sdim 58263508Sdim/* 59263508Sdim * Routines to expand arguments to commands. We have to deal with 60263508Sdim * backquotes, shell variables, and file metacharacters. 61263508Sdim */ 62263508Sdim 63263508Sdim#include "shell.h" 64263508Sdim#include "main.h" 65263508Sdim#include "nodes.h" 66263508Sdim#include "eval.h" 67263508Sdim#include "expand.h" 68263508Sdim#include "syntax.h" 69263508Sdim#include "parser.h" 70263508Sdim#include "jobs.h" 71263508Sdim#include "options.h" 72263508Sdim#include "var.h" 73263508Sdim#include "input.h" 74263508Sdim#include "output.h" 75263508Sdim#include "memalloc.h" 76263508Sdim#include "error.h" 77263508Sdim#include "mystring.h" 78263508Sdim#include "arith.h" 79263508Sdim#include "show.h" 80263508Sdim#include "builtins.h" 81263508Sdim 82263508Sdim/* 83263508Sdim * Structure specifying which parts of the string should be searched 84263508Sdim * for IFS characters. 85263508Sdim */ 86263508Sdim 87263508Sdimstruct ifsregion { 88263508Sdim struct ifsregion *next; /* next region in list */ 89263508Sdim int begoff; /* offset of start of region */ 90263508Sdim int endoff; /* offset of end of region */ 91263508Sdim int inquotes; /* search for nul bytes only */ 92263508Sdim}; 93263508Sdim 94263508Sdim 95263508Sdimstatic char *expdest; /* output of current string */ 96263508Sdimstatic struct nodelist *argbackq; /* list of back quote expressions */ 97263508Sdimstatic struct ifsregion ifsfirst; /* first struct in list of ifs regions */ 98263508Sdimstatic struct ifsregion *ifslastp; /* last struct in list */ 99263508Sdimstatic struct arglist exparg; /* holds expanded arg list */ 100263508Sdim 101263508Sdimstatic void argstr(char *, int); 102263508Sdimstatic char *exptilde(char *, int); 103263508Sdimstatic void expbackq(union node *, int, int); 104263508Sdimstatic int subevalvar(char *, char *, int, int, int, int, int); 105263508Sdimstatic char *evalvar(char *, int); 106263508Sdimstatic int varisset(char *, int); 107263508Sdimstatic void varvalue(char *, int, int, int); 108263508Sdimstatic void recordregion(int, int, int); 109263508Sdimstatic void removerecordregions(int); 110263508Sdimstatic void ifsbreakup(char *, struct arglist *); 111263508Sdimstatic void expandmeta(struct strlist *, int); 112263508Sdimstatic void expmeta(char *, char *); 113263508Sdimstatic void addfname(char *); 114263508Sdimstatic struct strlist *expsort(struct strlist *); 115263508Sdimstatic struct strlist *msort(struct strlist *, int); 116263508Sdimstatic int patmatch(const char *, const char *, int); 117263508Sdimstatic char *cvtnum(int, char *); 118263508Sdimstatic int collate_range_cmp(wchar_t, wchar_t); 119263508Sdim 120263508Sdimstatic int 121263508Sdimcollate_range_cmp(wchar_t c1, wchar_t c2) 122263508Sdim{ 123263508Sdim static wchar_t s1[2], s2[2]; 124263508Sdim 125263508Sdim s1[0] = c1; 126263508Sdim s2[0] = c2; 127263508Sdim return (wcscoll(s1, s2)); 128263508Sdim} 129263508Sdim 130263508Sdim/* 131263508Sdim * Expand shell variables and backquotes inside a here document. 132263508Sdim * union node *arg the document 133263508Sdim * int fd; where to write the expanded version 134263508Sdim */ 135263508Sdim 136263508Sdimvoid 137263508Sdimexpandhere(union node *arg, int fd) 138263508Sdim{ 139263508Sdim expandarg(arg, (struct arglist *)NULL, 0); 140263508Sdim xwrite(fd, stackblock(), expdest - stackblock()); 141263508Sdim} 142263508Sdim 143263508Sdimstatic char * 144263508Sdimstputs_quotes(const char *data, const char *syntax, char *p) 145263508Sdim{ 146263508Sdim while (*data) { 147263508Sdim CHECKSTRSPACE(2, p); 148263508Sdim if (syntax[(int)*data] == CCTL) 149263508Sdim USTPUTC(CTLESC, p); 150263508Sdim USTPUTC(*data++, p); 151263508Sdim } 152263508Sdim return (p); 153263508Sdim} 154263508Sdim#define STPUTS_QUOTES(data, syntax, p) p = stputs_quotes((data), syntax, p) 155263508Sdim 156263508Sdim/* 157263508Sdim * Perform expansions on an argument, placing the resulting list of arguments 158263508Sdim * in arglist. Parameter expansion, command substitution and arithmetic 159263508Sdim * expansion are always performed; additional expansions can be requested 160263508Sdim * via flag (EXP_*). 161263508Sdim * The result is left in the stack string. 162263508Sdim * When arglist is NULL, perform here document expansion. 163263508Sdim * 164263508Sdim * Caution: this function uses global state and is not reentrant. 165263508Sdim * However, a new invocation after an interrupted invocation is safe 166263508Sdim * and will reset the global state for the new call. 167263508Sdim */ 168263508Sdimvoid 169263508Sdimexpandarg(union node *arg, struct arglist *arglist, int flag) 170263508Sdim{ 171263508Sdim struct strlist *sp; 172263508Sdim char *p; 173263508Sdim 174263508Sdim argbackq = arg->narg.backquote; 175263508Sdim STARTSTACKSTR(expdest); 176263508Sdim ifsfirst.next = NULL; 177263508Sdim ifslastp = NULL; 178263508Sdim argstr(arg->narg.text, flag); 179263508Sdim if (arglist == NULL) { 180263508Sdim STACKSTRNUL(expdest); 181263508Sdim return; /* here document expanded */ 182263508Sdim } 183263508Sdim STPUTC('\0', expdest); 184263508Sdim p = grabstackstr(expdest); 185263508Sdim exparg.lastp = &exparg.list; 186263508Sdim /* 187263508Sdim * TODO - EXP_REDIR 188263508Sdim */ 189263508Sdim if (flag & EXP_FULL) { 190263508Sdim ifsbreakup(p, &exparg); 191263508Sdim *exparg.lastp = NULL; 192263508Sdim exparg.lastp = &exparg.list; 193263508Sdim expandmeta(exparg.list, flag); 194263508Sdim } else { 195263508Sdim if (flag & EXP_REDIR) /*XXX - for now, just remove escapes */ 196263508Sdim rmescapes(p); 197263508Sdim sp = (struct strlist *)stalloc(sizeof (struct strlist)); 198263508Sdim sp->text = p; 199263508Sdim *exparg.lastp = sp; 200263508Sdim exparg.lastp = &sp->next; 201263508Sdim } 202263508Sdim while (ifsfirst.next != NULL) { 203263508Sdim struct ifsregion *ifsp; 204263508Sdim INTOFF; 205263508Sdim ifsp = ifsfirst.next->next; 206263508Sdim ckfree(ifsfirst.next); 207263508Sdim ifsfirst.next = ifsp; 208263508Sdim INTON; 209263508Sdim } 210263508Sdim *exparg.lastp = NULL; 211263508Sdim if (exparg.list) { 212263508Sdim *arglist->lastp = exparg.list; 213263508Sdim arglist->lastp = exparg.lastp; 214263508Sdim } 215263508Sdim} 216263508Sdim 217263508Sdim 218263508Sdim 219263508Sdim/* 220263508Sdim * Perform parameter expansion, command substitution and arithmetic 221263508Sdim * expansion, and tilde expansion if requested via EXP_TILDE/EXP_VARTILDE. 222263508Sdim * Processing ends at a CTLENDVAR character as well as '\0'. 223263508Sdim * This is used to expand word in ${var+word} etc. 224263508Sdim * If EXP_FULL, EXP_CASE or EXP_REDIR are set, keep and/or generate CTLESC 225263508Sdim * characters to allow for further processing. 226263508Sdim * If EXP_FULL is set, also preserve CTLQUOTEMARK characters. 227263508Sdim */ 228263508Sdimstatic void 229263508Sdimargstr(char *p, int flag) 230263508Sdim{ 231263508Sdim char c; 232263508Sdim int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR); /* do CTLESC */ 233263508Sdim int firsteq = 1; 234263508Sdim int split_lit; 235263508Sdim int lit_quoted; 236263508Sdim 237263508Sdim split_lit = flag & EXP_SPLIT_LIT; 238263508Sdim lit_quoted = flag & EXP_LIT_QUOTED; 239263508Sdim flag &= ~(EXP_SPLIT_LIT | EXP_LIT_QUOTED); 240263508Sdim if (*p == '~' && (flag & (EXP_TILDE | EXP_VARTILDE))) 241263508Sdim p = exptilde(p, flag); 242263508Sdim for (;;) { 243263508Sdim CHECKSTRSPACE(2, expdest); 244263508Sdim switch (c = *p++) { 245263508Sdim case '\0': 246263508Sdim case CTLENDVAR: 247263508Sdim goto breakloop; 248263508Sdim case CTLQUOTEMARK: 249263508Sdim lit_quoted = 1; 250263508Sdim /* "$@" syntax adherence hack */ 251263508Sdim if (p[0] == CTLVAR && p[2] == '@' && p[3] == '=') 252263508Sdim break; 253263508Sdim if ((flag & EXP_FULL) != 0) 254263508Sdim USTPUTC(c, expdest); 255263508Sdim break; 256263508Sdim case CTLQUOTEEND: 257263508Sdim lit_quoted = 0; 258263508Sdim break; 259263508Sdim case CTLESC: 260263508Sdim if (quotes) 261263508Sdim USTPUTC(c, expdest); 262263508Sdim c = *p++; 263263508Sdim USTPUTC(c, expdest); 264263508Sdim if (split_lit && !lit_quoted) 265263508Sdim recordregion(expdest - stackblock() - 266263508Sdim (quotes ? 2 : 1), 267263508Sdim expdest - stackblock(), 0); 268263508Sdim break; 269263508Sdim case CTLVAR: 270263508Sdim p = evalvar(p, flag); 271263508Sdim break; 272263508Sdim case CTLBACKQ: 273263508Sdim case CTLBACKQ|CTLQUOTE: 274263508Sdim expbackq(argbackq->n, c & CTLQUOTE, flag); 275263508Sdim argbackq = argbackq->next; 276263508Sdim break; 277263508Sdim case CTLENDARI: 278263508Sdim expari(flag); 279263508Sdim break; 280263508Sdim case ':': 281263508Sdim case '=': 282263508Sdim /* 283263508Sdim * sort of a hack - expand tildes in variable 284263508Sdim * assignments (after the first '=' and after ':'s). 285263508Sdim */ 286263508Sdim USTPUTC(c, expdest); 287263508Sdim if (split_lit && !lit_quoted) 288263508Sdim recordregion(expdest - stackblock() - 1, 289263508Sdim expdest - stackblock(), 0); 290263508Sdim if (flag & EXP_VARTILDE && *p == '~' && 291263508Sdim (c != '=' || firsteq)) { 292263508Sdim if (c == '=') 293263508Sdim firsteq = 0; 294263508Sdim p = exptilde(p, flag); 295263508Sdim } 296263508Sdim break; 297263508Sdim default: 298263508Sdim USTPUTC(c, expdest); 299263508Sdim if (split_lit && !lit_quoted) 300263508Sdim recordregion(expdest - stackblock() - 1, 301263508Sdim expdest - stackblock(), 0); 302263508Sdim } 303263508Sdim } 304263508Sdimbreakloop:; 305263508Sdim} 306263508Sdim 307263508Sdim/* 308263508Sdim * Perform tilde expansion, placing the result in the stack string and 309263508Sdim * returning the next position in the input string to process. 310263508Sdim */ 311263508Sdimstatic char * 312263508Sdimexptilde(char *p, int flag) 313263508Sdim{ 314263508Sdim char c, *startp = p; 315263508Sdim struct passwd *pw; 316263508Sdim char *home; 317263508Sdim int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR); 318263508Sdim 319263508Sdim while ((c = *p) != '\0') { 320263508Sdim switch(c) { 321263508Sdim case CTLESC: /* This means CTL* are always considered quoted. */ 322263508Sdim case CTLVAR: 323263508Sdim case CTLBACKQ: 324263508Sdim case CTLBACKQ | CTLQUOTE: 325263508Sdim case CTLARI: 326263508Sdim case CTLENDARI: 327263508Sdim case CTLQUOTEMARK: 328263508Sdim return (startp); 329263508Sdim case ':': 330263508Sdim if (flag & EXP_VARTILDE) 331263508Sdim goto done; 332263508Sdim break; 333263508Sdim case '/': 334263508Sdim case CTLENDVAR: 335263508Sdim goto done; 336263508Sdim } 337263508Sdim p++; 338263508Sdim } 339263508Sdimdone: 340263508Sdim *p = '\0'; 341263508Sdim if (*(startp+1) == '\0') { 342263508Sdim if ((home = lookupvar("HOME")) == NULL) 343263508Sdim goto lose; 344263508Sdim } else { 345263508Sdim if ((pw = getpwnam(startp+1)) == NULL) 346263508Sdim goto lose; 347263508Sdim home = pw->pw_dir; 348263508Sdim } 349263508Sdim if (*home == '\0') 350263508Sdim goto lose; 351263508Sdim *p = c; 352263508Sdim if (quotes) 353263508Sdim STPUTS_QUOTES(home, SQSYNTAX, expdest); 354263508Sdim else 355263508Sdim STPUTS(home, expdest); 356263508Sdim return (p); 357263508Sdimlose: 358263508Sdim *p = c; 359263508Sdim return (startp); 360263508Sdim} 361263508Sdim 362263508Sdim 363263508Sdimstatic void 364263508Sdimremoverecordregions(int endoff) 365263508Sdim{ 366263508Sdim if (ifslastp == NULL) 367263508Sdim return; 368263508Sdim 369263508Sdim if (ifsfirst.endoff > endoff) { 370263508Sdim while (ifsfirst.next != NULL) { 371263508Sdim struct ifsregion *ifsp; 372263508Sdim INTOFF; 373263508Sdim ifsp = ifsfirst.next->next; 374263508Sdim ckfree(ifsfirst.next); 375263508Sdim ifsfirst.next = ifsp; 376263508Sdim INTON; 377263508Sdim } 378263508Sdim if (ifsfirst.begoff > endoff) 379263508Sdim ifslastp = NULL; 380263508Sdim else { 381263508Sdim ifslastp = &ifsfirst; 382263508Sdim ifsfirst.endoff = endoff; 383263508Sdim } 384263508Sdim return; 385263508Sdim } 386263508Sdim 387263508Sdim ifslastp = &ifsfirst; 388263508Sdim while (ifslastp->next && ifslastp->next->begoff < endoff) 389263508Sdim ifslastp=ifslastp->next; 390 while (ifslastp->next != NULL) { 391 struct ifsregion *ifsp; 392 INTOFF; 393 ifsp = ifslastp->next->next; 394 ckfree(ifslastp->next); 395 ifslastp->next = ifsp; 396 INTON; 397 } 398 if (ifslastp->endoff > endoff) 399 ifslastp->endoff = endoff; 400} 401 402/* 403 * Expand arithmetic expression. Backup to start of expression, 404 * evaluate, place result in (backed up) result, adjust string position. 405 */ 406void 407expari(int flag) 408{ 409 char *p, *q, *start; 410 arith_t result; 411 int begoff; 412 int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR); 413 int quoted; 414 415 /* 416 * This routine is slightly over-complicated for 417 * efficiency. First we make sure there is 418 * enough space for the result, which may be bigger 419 * than the expression. Next we 420 * scan backwards looking for the start of arithmetic. If the 421 * next previous character is a CTLESC character, then we 422 * have to rescan starting from the beginning since CTLESC 423 * characters have to be processed left to right. 424 */ 425 CHECKSTRSPACE(DIGITS(result) - 2, expdest); 426 USTPUTC('\0', expdest); 427 start = stackblock(); 428 p = expdest - 2; 429 while (p >= start && *p != CTLARI) 430 --p; 431 if (p < start || *p != CTLARI) 432 error("missing CTLARI (shouldn't happen)"); 433 if (p > start && *(p - 1) == CTLESC) 434 for (p = start; *p != CTLARI; p++) 435 if (*p == CTLESC) 436 p++; 437 438 if (p[1] == '"') 439 quoted=1; 440 else 441 quoted=0; 442 begoff = p - start; 443 removerecordregions(begoff); 444 if (quotes) 445 rmescapes(p+2); 446 q = grabstackstr(expdest); 447 result = arith(p+2); 448 ungrabstackstr(q, expdest); 449 fmtstr(p, DIGITS(result), ARITH_FORMAT_STR, result); 450 while (*p++) 451 ; 452 if (quoted == 0) 453 recordregion(begoff, p - 1 - start, 0); 454 result = expdest - p + 1; 455 STADJUST(-result, expdest); 456} 457 458 459/* 460 * Perform command substitution. 461 */ 462static void 463expbackq(union node *cmd, int quoted, int flag) 464{ 465 struct backcmd in; 466 int i; 467 char buf[128]; 468 char *p; 469 char *dest = expdest; 470 struct ifsregion saveifs, *savelastp; 471 struct nodelist *saveargbackq; 472 char lastc; 473 int startloc = dest - stackblock(); 474 char const *syntax = quoted? DQSYNTAX : BASESYNTAX; 475 int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR); 476 int nnl; 477 478 INTOFF; 479 saveifs = ifsfirst; 480 savelastp = ifslastp; 481 saveargbackq = argbackq; 482 p = grabstackstr(dest); 483 evalbackcmd(cmd, &in); 484 ungrabstackstr(p, dest); 485 ifsfirst = saveifs; 486 ifslastp = savelastp; 487 argbackq = saveargbackq; 488 489 p = in.buf; 490 lastc = '\0'; 491 nnl = 0; 492 /* Don't copy trailing newlines */ 493 for (;;) { 494 if (--in.nleft < 0) { 495 if (in.fd < 0) 496 break; 497 while ((i = read(in.fd, buf, sizeof buf)) < 0 && errno == EINTR); 498 TRACE(("expbackq: read returns %d\n", i)); 499 if (i <= 0) 500 break; 501 p = buf; 502 in.nleft = i - 1; 503 } 504 lastc = *p++; 505 if (lastc != '\0') { 506 if (lastc == '\n') { 507 nnl++; 508 } else { 509 CHECKSTRSPACE(nnl + 2, dest); 510 while (nnl > 0) { 511 nnl--; 512 USTPUTC('\n', dest); 513 } 514 if (quotes && syntax[(int)lastc] == CCTL) 515 USTPUTC(CTLESC, dest); 516 USTPUTC(lastc, dest); 517 } 518 } 519 } 520 521 if (in.fd >= 0) 522 close(in.fd); 523 if (in.buf) 524 ckfree(in.buf); 525 if (in.jp) 526 exitstatus = waitforjob(in.jp, (int *)NULL); 527 if (quoted == 0) 528 recordregion(startloc, dest - stackblock(), 0); 529 TRACE(("expbackq: size=%td: \"%.*s\"\n", 530 ((dest - stackblock()) - startloc), 531 (int)((dest - stackblock()) - startloc), 532 stackblock() + startloc)); 533 expdest = dest; 534 INTON; 535} 536 537 538 539static int 540subevalvar(char *p, char *str, int strloc, int subtype, int startloc, 541 int varflags, int quotes) 542{ 543 char *startp; 544 char *loc = NULL; 545 char *q; 546 int c = 0; 547 struct nodelist *saveargbackq = argbackq; 548 int amount; 549 550 argstr(p, (subtype == VSTRIMLEFT || subtype == VSTRIMLEFTMAX || 551 subtype == VSTRIMRIGHT || subtype == VSTRIMRIGHTMAX ? 552 EXP_CASE : 0) | EXP_TILDE); 553 STACKSTRNUL(expdest); 554 argbackq = saveargbackq; 555 startp = stackblock() + startloc; 556 if (str == NULL) 557 str = stackblock() + strloc; 558 559 switch (subtype) { 560 case VSASSIGN: 561 setvar(str, startp, 0); 562 amount = startp - expdest; 563 STADJUST(amount, expdest); 564 varflags &= ~VSNUL; 565 return 1; 566 567 case VSQUESTION: 568 if (*p != CTLENDVAR) { 569 outfmt(out2, "%s\n", startp); 570 error((char *)NULL); 571 } 572 error("%.*s: parameter %snot set", (int)(p - str - 1), 573 str, (varflags & VSNUL) ? "null or " 574 : nullstr); 575 return 0; 576 577 case VSTRIMLEFT: 578 for (loc = startp; loc < str; loc++) { 579 c = *loc; 580 *loc = '\0'; 581 if (patmatch(str, startp, quotes)) { 582 *loc = c; 583 goto recordleft; 584 } 585 *loc = c; 586 if (quotes && *loc == CTLESC) 587 loc++; 588 } 589 return 0; 590 591 case VSTRIMLEFTMAX: 592 for (loc = str - 1; loc >= startp;) { 593 c = *loc; 594 *loc = '\0'; 595 if (patmatch(str, startp, quotes)) { 596 *loc = c; 597 goto recordleft; 598 } 599 *loc = c; 600 loc--; 601 if (quotes && loc > startp && *(loc - 1) == CTLESC) { 602 for (q = startp; q < loc; q++) 603 if (*q == CTLESC) 604 q++; 605 if (q > loc) 606 loc--; 607 } 608 } 609 return 0; 610 611 case VSTRIMRIGHT: 612 for (loc = str - 1; loc >= startp;) { 613 if (patmatch(str, loc, quotes)) { 614 amount = loc - expdest; 615 STADJUST(amount, expdest); 616 return 1; 617 } 618 loc--; 619 if (quotes && loc > startp && *(loc - 1) == CTLESC) { 620 for (q = startp; q < loc; q++) 621 if (*q == CTLESC) 622 q++; 623 if (q > loc) 624 loc--; 625 } 626 } 627 return 0; 628 629 case VSTRIMRIGHTMAX: 630 for (loc = startp; loc < str - 1; loc++) { 631 if (patmatch(str, loc, quotes)) { 632 amount = loc - expdest; 633 STADJUST(amount, expdest); 634 return 1; 635 } 636 if (quotes && *loc == CTLESC) 637 loc++; 638 } 639 return 0; 640 641 642 default: 643 abort(); 644 } 645 646recordleft: 647 amount = ((str - 1) - (loc - startp)) - expdest; 648 STADJUST(amount, expdest); 649 while (loc != str - 1) 650 *startp++ = *loc++; 651 return 1; 652} 653 654 655/* 656 * Expand a variable, and return a pointer to the next character in the 657 * input string. 658 */ 659 660static char * 661evalvar(char *p, int flag) 662{ 663 int subtype; 664 int varflags; 665 char *var; 666 char *val; 667 int patloc; 668 int c; 669 int set; 670 int special; 671 int startloc; 672 int varlen; 673 int varlenb; 674 int easy; 675 int quotes = flag & (EXP_FULL | EXP_CASE | EXP_REDIR); 676 677 varflags = (unsigned char)*p++; 678 subtype = varflags & VSTYPE; 679 var = p; 680 special = 0; 681 if (! is_name(*p)) 682 special = 1; 683 p = strchr(p, '=') + 1; 684again: /* jump here after setting a variable with ${var=text} */ 685 if (varflags & VSLINENO) { 686 set = 1; 687 special = 0; 688 val = var; 689 p[-1] = '\0'; /* temporarily overwrite '=' to have \0 690 terminated string */ 691 } else if (special) { 692 set = varisset(var, varflags & VSNUL); 693 val = NULL; 694 } else { 695 val = bltinlookup(var, 1); 696 if (val == NULL || ((varflags & VSNUL) && val[0] == '\0')) { 697 val = NULL; 698 set = 0; 699 } else 700 set = 1; 701 } 702 varlen = 0; 703 startloc = expdest - stackblock(); 704 if (!set && uflag && *var != '@' && *var != '*') { 705 switch (subtype) { 706 case VSNORMAL: 707 case VSTRIMLEFT: 708 case VSTRIMLEFTMAX: 709 case VSTRIMRIGHT: 710 case VSTRIMRIGHTMAX: 711 case VSLENGTH: 712 error("%.*s: parameter not set", (int)(p - var - 1), 713 var); 714 } 715 } 716 if (set && subtype != VSPLUS) { 717 /* insert the value of the variable */ 718 if (special) { 719 varvalue(var, varflags & VSQUOTE, subtype, flag); 720 if (subtype == VSLENGTH) { 721 varlenb = expdest - stackblock() - startloc; 722 varlen = varlenb; 723 if (localeisutf8) { 724 val = stackblock() + startloc; 725 for (;val != expdest; val++) 726 if ((*val & 0xC0) == 0x80) 727 varlen--; 728 } 729 STADJUST(-varlenb, expdest); 730 } 731 } else { 732 char const *syntax = (varflags & VSQUOTE) ? DQSYNTAX 733 : BASESYNTAX; 734 735 if (subtype == VSLENGTH) { 736 for (;*val; val++) 737 if (!localeisutf8 || 738 (*val & 0xC0) != 0x80) 739 varlen++; 740 } 741 else { 742 if (quotes) 743 STPUTS_QUOTES(val, syntax, expdest); 744 else 745 STPUTS(val, expdest); 746 747 } 748 } 749 } 750 751 if (subtype == VSPLUS) 752 set = ! set; 753 754 easy = ((varflags & VSQUOTE) == 0 || 755 (*var == '@' && shellparam.nparam != 1)); 756 757 758 switch (subtype) { 759 case VSLENGTH: 760 expdest = cvtnum(varlen, expdest); 761 goto record; 762 763 case VSNORMAL: 764 if (!easy) 765 break; 766record: 767 recordregion(startloc, expdest - stackblock(), 768 varflags & VSQUOTE || (ifsset() && ifsval()[0] == '\0' && 769 (*var == '@' || *var == '*'))); 770 break; 771 772 case VSPLUS: 773 case VSMINUS: 774 if (!set) { 775 argstr(p, flag | (flag & EXP_FULL ? EXP_SPLIT_LIT : 0) | 776 (varflags & VSQUOTE ? EXP_LIT_QUOTED : 0)); 777 break; 778 } 779 if (easy) 780 goto record; 781 break; 782 783 case VSTRIMLEFT: 784 case VSTRIMLEFTMAX: 785 case VSTRIMRIGHT: 786 case VSTRIMRIGHTMAX: 787 if (!set) 788 break; 789 /* 790 * Terminate the string and start recording the pattern 791 * right after it 792 */ 793 STPUTC('\0', expdest); 794 patloc = expdest - stackblock(); 795 if (subevalvar(p, NULL, patloc, subtype, 796 startloc, varflags, quotes) == 0) { 797 int amount = (expdest - stackblock() - patloc) + 1; 798 STADJUST(-amount, expdest); 799 } 800 /* Remove any recorded regions beyond start of variable */ 801 removerecordregions(startloc); 802 goto record; 803 804 case VSASSIGN: 805 case VSQUESTION: 806 if (!set) { 807 if (subevalvar(p, var, 0, subtype, startloc, varflags, 808 quotes)) { 809 varflags &= ~VSNUL; 810 /* 811 * Remove any recorded regions beyond 812 * start of variable 813 */ 814 removerecordregions(startloc); 815 goto again; 816 } 817 break; 818 } 819 if (easy) 820 goto record; 821 break; 822 823 case VSERROR: 824 c = p - var - 1; 825 error("${%.*s%s}: Bad substitution", c, var, 826 (c > 0 && *p != CTLENDVAR) ? "..." : ""); 827 828 default: 829 abort(); 830 } 831 p[-1] = '='; /* recover overwritten '=' */ 832 833 if (subtype != VSNORMAL) { /* skip to end of alternative */ 834 int nesting = 1; 835 for (;;) { 836 if ((c = *p++) == CTLESC) 837 p++; 838 else if (c == CTLBACKQ || c == (CTLBACKQ|CTLQUOTE)) { 839 if (set) 840 argbackq = argbackq->next; 841 } else if (c == CTLVAR) { 842 if ((*p++ & VSTYPE) != VSNORMAL) 843 nesting++; 844 } else if (c == CTLENDVAR) { 845 if (--nesting == 0) 846 break; 847 } 848 } 849 } 850 return p; 851} 852 853 854 855/* 856 * Test whether a specialized variable is set. 857 */ 858 859static int 860varisset(char *name, int nulok) 861{ 862 863 if (*name == '!') 864 return backgndpidset(); 865 else if (*name == '@' || *name == '*') { 866 if (*shellparam.p == NULL) 867 return 0; 868 869 if (nulok) { 870 char **av; 871 872 for (av = shellparam.p; *av; av++) 873 if (**av != '\0') 874 return 1; 875 return 0; 876 } 877 } else if (is_digit(*name)) { 878 char *ap; 879 int num = atoi(name); 880 881 if (num > shellparam.nparam) 882 return 0; 883 884 if (num == 0) 885 ap = arg0; 886 else 887 ap = shellparam.p[num - 1]; 888 889 if (nulok && (ap == NULL || *ap == '\0')) 890 return 0; 891 } 892 return 1; 893} 894 895static void 896strtodest(const char *p, int flag, int subtype, int quoted) 897{ 898 if (flag & (EXP_FULL | EXP_CASE) && subtype != VSLENGTH) 899 STPUTS_QUOTES(p, quoted ? DQSYNTAX : BASESYNTAX, expdest); 900 else 901 STPUTS(p, expdest); 902} 903 904/* 905 * Add the value of a specialized variable to the stack string. 906 */ 907 908static void 909varvalue(char *name, int quoted, int subtype, int flag) 910{ 911 int num; 912 char *p; 913 int i; 914 char sep; 915 char **ap; 916 917 switch (*name) { 918 case '$': 919 num = rootpid; 920 goto numvar; 921 case '?': 922 num = oexitstatus; 923 goto numvar; 924 case '#': 925 num = shellparam.nparam; 926 goto numvar; 927 case '!': 928 num = backgndpidval(); 929numvar: 930 expdest = cvtnum(num, expdest); 931 break; 932 case '-': 933 for (i = 0 ; i < NOPTS ; i++) { 934 if (optlist[i].val) 935 STPUTC(optlist[i].letter, expdest); 936 } 937 break; 938 case '@': 939 if (flag & EXP_FULL && quoted) { 940 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 941 strtodest(p, flag, subtype, quoted); 942 if (*ap) 943 STPUTC('\0', expdest); 944 } 945 break; 946 } 947 /* FALLTHROUGH */ 948 case '*': 949 if (ifsset()) 950 sep = ifsval()[0]; 951 else 952 sep = ' '; 953 for (ap = shellparam.p ; (p = *ap++) != NULL ; ) { 954 strtodest(p, flag, subtype, quoted); 955 if (!*ap) 956 break; 957 if (sep || (flag & EXP_FULL && !quoted && **ap != '\0')) 958 STPUTC(sep, expdest); 959 } 960 break; 961 case '0': 962 p = arg0; 963 strtodest(p, flag, subtype, quoted); 964 break; 965 default: 966 if (is_digit(*name)) { 967 num = atoi(name); 968 if (num > 0 && num <= shellparam.nparam) { 969 p = shellparam.p[num - 1]; 970 strtodest(p, flag, subtype, quoted); 971 } 972 } 973 break; 974 } 975} 976 977 978 979/* 980 * Record the fact that we have to scan this region of the 981 * string for IFS characters. 982 */ 983 984static void 985recordregion(int start, int end, int inquotes) 986{ 987 struct ifsregion *ifsp; 988 989 if (ifslastp == NULL) { 990 ifsp = &ifsfirst; 991 } else { 992 if (ifslastp->endoff == start 993 && ifslastp->inquotes == inquotes) { 994 /* extend previous area */ 995 ifslastp->endoff = end; 996 return; 997 } 998 ifsp = (struct ifsregion *)ckmalloc(sizeof (struct ifsregion)); 999 ifslastp->next = ifsp; 1000 } 1001 ifslastp = ifsp; 1002 ifslastp->next = NULL; 1003 ifslastp->begoff = start; 1004 ifslastp->endoff = end; 1005 ifslastp->inquotes = inquotes; 1006} 1007 1008 1009 1010/* 1011 * Break the argument string into pieces based upon IFS and add the 1012 * strings to the argument list. The regions of the string to be 1013 * searched for IFS characters have been stored by recordregion. 1014 * CTLESC characters are preserved but have little effect in this pass 1015 * other than escaping CTL* characters. In particular, they do not escape 1016 * IFS characters: that should be done with the ifsregion mechanism. 1017 * CTLQUOTEMARK characters are used to preserve empty quoted strings. 1018 * This pass treats them as a regular character, making the string non-empty. 1019 * Later, they are removed along with the other CTL* characters. 1020 */ 1021static void 1022ifsbreakup(char *string, struct arglist *arglist) 1023{ 1024 struct ifsregion *ifsp; 1025 struct strlist *sp; 1026 char *start; 1027 char *p; 1028 char *q; 1029 const char *ifs; 1030 const char *ifsspc; 1031 int had_param_ch = 0; 1032 1033 start = string; 1034 1035 if (ifslastp == NULL) { 1036 /* Return entire argument, IFS doesn't apply to any of it */ 1037 sp = (struct strlist *)stalloc(sizeof *sp); 1038 sp->text = start; 1039 *arglist->lastp = sp; 1040 arglist->lastp = &sp->next; 1041 return; 1042 } 1043 1044 ifs = ifsset() ? ifsval() : " \t\n"; 1045 1046 for (ifsp = &ifsfirst; ifsp != NULL; ifsp = ifsp->next) { 1047 p = string + ifsp->begoff; 1048 while (p < string + ifsp->endoff) { 1049 q = p; 1050 if (*p == CTLESC) 1051 p++; 1052 if (ifsp->inquotes) { 1053 /* Only NULs (should be from "$@") end args */ 1054 had_param_ch = 1; 1055 if (*p != 0) { 1056 p++; 1057 continue; 1058 } 1059 ifsspc = NULL; 1060 } else { 1061 if (!strchr(ifs, *p)) { 1062 had_param_ch = 1; 1063 p++; 1064 continue; 1065 } 1066 ifsspc = strchr(" \t\n", *p); 1067 1068 /* Ignore IFS whitespace at start */ 1069 if (q == start && ifsspc != NULL) { 1070 p++; 1071 start = p; 1072 continue; 1073 } 1074 had_param_ch = 0; 1075 } 1076 1077 /* Save this argument... */ 1078 *q = '\0'; 1079 sp = (struct strlist *)stalloc(sizeof *sp); 1080 sp->text = start; 1081 *arglist->lastp = sp; 1082 arglist->lastp = &sp->next; 1083 p++; 1084 1085 if (ifsspc != NULL) { 1086 /* Ignore further trailing IFS whitespace */ 1087 for (; p < string + ifsp->endoff; p++) { 1088 q = p; 1089 if (*p == CTLESC) 1090 p++; 1091 if (strchr(ifs, *p) == NULL) { 1092 p = q; 1093 break; 1094 } 1095 if (strchr(" \t\n", *p) == NULL) { 1096 p++; 1097 break; 1098 } 1099 } 1100 } 1101 start = p; 1102 } 1103 } 1104 1105 /* 1106 * Save anything left as an argument. 1107 * Traditionally we have treated 'IFS=':'; set -- x$IFS' as 1108 * generating 2 arguments, the second of which is empty. 1109 * Some recent clarification of the Posix spec say that it 1110 * should only generate one.... 1111 */ 1112 if (had_param_ch || *start != 0) { 1113 sp = (struct strlist *)stalloc(sizeof *sp); 1114 sp->text = start; 1115 *arglist->lastp = sp; 1116 arglist->lastp = &sp->next; 1117 } 1118} 1119 1120 1121static char expdir[PATH_MAX]; 1122#define expdir_end (expdir + sizeof(expdir)) 1123 1124/* 1125 * Perform pathname generation and remove control characters. 1126 * At this point, the only control characters should be CTLESC and CTLQUOTEMARK. 1127 * The results are stored in the list exparg. 1128 */ 1129static void 1130expandmeta(struct strlist *str, int flag __unused) 1131{ 1132 char *p; 1133 struct strlist **savelastp; 1134 struct strlist *sp; 1135 char c; 1136 /* TODO - EXP_REDIR */ 1137 1138 while (str) { 1139 if (fflag) 1140 goto nometa; 1141 p = str->text; 1142 for (;;) { /* fast check for meta chars */ 1143 if ((c = *p++) == '\0') 1144 goto nometa; 1145 if (c == '*' || c == '?' || c == '[') 1146 break; 1147 } 1148 savelastp = exparg.lastp; 1149 INTOFF; 1150 expmeta(expdir, str->text); 1151 INTON; 1152 if (exparg.lastp == savelastp) { 1153 /* 1154 * no matches 1155 */ 1156nometa: 1157 *exparg.lastp = str; 1158 rmescapes(str->text); 1159 exparg.lastp = &str->next; 1160 } else { 1161 *exparg.lastp = NULL; 1162 *savelastp = sp = expsort(*savelastp); 1163 while (sp->next != NULL) 1164 sp = sp->next; 1165 exparg.lastp = &sp->next; 1166 } 1167 str = str->next; 1168 } 1169} 1170 1171 1172/* 1173 * Do metacharacter (i.e. *, ?, [...]) expansion. 1174 */ 1175 1176static void 1177expmeta(char *enddir, char *name) 1178{ 1179 char *p; 1180 char *q; 1181 char *start; 1182 char *endname; 1183 int metaflag; 1184 struct stat statb; 1185 DIR *dirp; 1186 struct dirent *dp; 1187 int atend; 1188 int matchdot; 1189 int esc; 1190 int namlen; 1191 1192 metaflag = 0; 1193 start = name; 1194 for (p = name; esc = 0, *p; p += esc + 1) { 1195 if (*p == '*' || *p == '?') 1196 metaflag = 1; 1197 else if (*p == '[') { 1198 q = p + 1; 1199 if (*q == '!' || *q == '^') 1200 q++; 1201 for (;;) { 1202 while (*q == CTLQUOTEMARK) 1203 q++; 1204 if (*q == CTLESC) 1205 q++; 1206 if (*q == '/' || *q == '\0') 1207 break; 1208 if (*++q == ']') { 1209 metaflag = 1; 1210 break; 1211 } 1212 } 1213 } else if (*p == '\0') 1214 break; 1215 else if (*p == CTLQUOTEMARK) 1216 continue; 1217 else { 1218 if (*p == CTLESC) 1219 esc++; 1220 if (p[esc] == '/') { 1221 if (metaflag) 1222 break; 1223 start = p + esc + 1; 1224 } 1225 } 1226 } 1227 if (metaflag == 0) { /* we've reached the end of the file name */ 1228 if (enddir != expdir) 1229 metaflag++; 1230 for (p = name ; ; p++) { 1231 if (*p == CTLQUOTEMARK) 1232 continue; 1233 if (*p == CTLESC) 1234 p++; 1235 *enddir++ = *p; 1236 if (*p == '\0') 1237 break; 1238 if (enddir == expdir_end) 1239 return; 1240 } 1241 if (metaflag == 0 || lstat(expdir, &statb) >= 0) 1242 addfname(expdir); 1243 return; 1244 } 1245 endname = p; 1246 if (start != name) { 1247 p = name; 1248 while (p < start) { 1249 while (*p == CTLQUOTEMARK) 1250 p++; 1251 if (*p == CTLESC) 1252 p++; 1253 *enddir++ = *p++; 1254 if (enddir == expdir_end) 1255 return; 1256 } 1257 } 1258 if (enddir == expdir) { 1259 p = "."; 1260 } else if (enddir == expdir + 1 && *expdir == '/') { 1261 p = "/"; 1262 } else { 1263 p = expdir; 1264 enddir[-1] = '\0'; 1265 } 1266 if ((dirp = opendir(p)) == NULL) 1267 return; 1268 if (enddir != expdir) 1269 enddir[-1] = '/'; 1270 if (*endname == 0) { 1271 atend = 1; 1272 } else { 1273 atend = 0; 1274 *endname = '\0'; 1275 endname += esc + 1; 1276 } 1277 matchdot = 0; 1278 p = start; 1279 while (*p == CTLQUOTEMARK) 1280 p++; 1281 if (*p == CTLESC) 1282 p++; 1283 if (*p == '.') 1284 matchdot++; 1285 while (! int_pending() && (dp = readdir(dirp)) != NULL) { 1286 if (dp->d_name[0] == '.' && ! matchdot) 1287 continue; 1288 if (patmatch(start, dp->d_name, 0)) { 1289 namlen = dp->d_namlen; 1290 if (enddir + namlen + 1 > expdir_end) 1291 continue; 1292 memcpy(enddir, dp->d_name, namlen + 1); 1293 if (atend) 1294 addfname(expdir); 1295 else { 1296 if (dp->d_type != DT_UNKNOWN && 1297 dp->d_type != DT_DIR && 1298 dp->d_type != DT_LNK) 1299 continue; 1300 if (enddir + namlen + 2 > expdir_end) 1301 continue; 1302 enddir[namlen] = '/'; 1303 enddir[namlen + 1] = '\0'; 1304 expmeta(enddir + namlen + 1, endname); 1305 } 1306 } 1307 } 1308 closedir(dirp); 1309 if (! atend) 1310 endname[-esc - 1] = esc ? CTLESC : '/'; 1311} 1312 1313 1314/* 1315 * Add a file name to the list. 1316 */ 1317 1318static void 1319addfname(char *name) 1320{ 1321 char *p; 1322 struct strlist *sp; 1323 1324 p = stalloc(strlen(name) + 1); 1325 scopy(name, p); 1326 sp = (struct strlist *)stalloc(sizeof *sp); 1327 sp->text = p; 1328 *exparg.lastp = sp; 1329 exparg.lastp = &sp->next; 1330} 1331 1332 1333/* 1334 * Sort the results of file name expansion. It calculates the number of 1335 * strings to sort and then calls msort (short for merge sort) to do the 1336 * work. 1337 */ 1338 1339static struct strlist * 1340expsort(struct strlist *str) 1341{ 1342 int len; 1343 struct strlist *sp; 1344 1345 len = 0; 1346 for (sp = str ; sp ; sp = sp->next) 1347 len++; 1348 return msort(str, len); 1349} 1350 1351 1352static struct strlist * 1353msort(struct strlist *list, int len) 1354{ 1355 struct strlist *p, *q = NULL; 1356 struct strlist **lpp; 1357 int half; 1358 int n; 1359 1360 if (len <= 1) 1361 return list; 1362 half = len >> 1; 1363 p = list; 1364 for (n = half ; --n >= 0 ; ) { 1365 q = p; 1366 p = p->next; 1367 } 1368 q->next = NULL; /* terminate first half of list */ 1369 q = msort(list, half); /* sort first half of list */ 1370 p = msort(p, len - half); /* sort second half */ 1371 lpp = &list; 1372 for (;;) { 1373 if (strcmp(p->text, q->text) < 0) { 1374 *lpp = p; 1375 lpp = &p->next; 1376 if ((p = *lpp) == NULL) { 1377 *lpp = q; 1378 break; 1379 } 1380 } else { 1381 *lpp = q; 1382 lpp = &q->next; 1383 if ((q = *lpp) == NULL) { 1384 *lpp = p; 1385 break; 1386 } 1387 } 1388 } 1389 return list; 1390} 1391 1392 1393 1394static wchar_t 1395get_wc(const char **p) 1396{ 1397 wchar_t c; 1398 int chrlen; 1399 1400 chrlen = mbtowc(&c, *p, 4); 1401 if (chrlen == 0) 1402 return 0; 1403 else if (chrlen == -1) 1404 c = 0; 1405 else 1406 *p += chrlen; 1407 return c; 1408} 1409 1410 1411/* 1412 * See if a character matches a character class, starting at the first colon 1413 * of "[:class:]". 1414 * If a valid character class is recognized, a pointer to the next character 1415 * after the final closing bracket is stored into *end, otherwise a null 1416 * pointer is stored into *end. 1417 */ 1418static int 1419match_charclass(const char *p, wchar_t chr, const char **end) 1420{ 1421 char name[20]; 1422 const char *nameend; 1423 wctype_t cclass; 1424 1425 *end = NULL; 1426 p++; 1427 nameend = strstr(p, ":]"); 1428 if (nameend == NULL || nameend - p >= sizeof(name) || nameend == p) 1429 return 0; 1430 memcpy(name, p, nameend - p); 1431 name[nameend - p] = '\0'; 1432 *end = nameend + 2; 1433 cclass = wctype(name); 1434 /* An unknown class matches nothing but is valid nevertheless. */ 1435 if (cclass == 0) 1436 return 0; 1437 return iswctype(chr, cclass); 1438} 1439 1440 1441/* 1442 * Returns true if the pattern matches the string. 1443 */ 1444 1445static int 1446patmatch(const char *pattern, const char *string, int squoted) 1447{ 1448 const char *p, *q, *end; 1449 const char *bt_p, *bt_q; 1450 char c; 1451 wchar_t wc, wc2; 1452 1453 p = pattern; 1454 q = string; 1455 bt_p = NULL; 1456 bt_q = NULL; 1457 for (;;) { 1458 switch (c = *p++) { 1459 case '\0': 1460 if (*q != '\0') 1461 goto backtrack; 1462 return 1; 1463 case CTLESC: 1464 if (squoted && *q == CTLESC) 1465 q++; 1466 if (*q++ != *p++) 1467 goto backtrack; 1468 break; 1469 case CTLQUOTEMARK: 1470 continue; 1471 case '?': 1472 if (squoted && *q == CTLESC) 1473 q++; 1474 if (*q == '\0') 1475 return 0; 1476 if (localeisutf8) { 1477 wc = get_wc(&q); 1478 /* 1479 * A '?' does not match invalid UTF-8 but a 1480 * '*' does, so backtrack. 1481 */ 1482 if (wc == 0) 1483 goto backtrack; 1484 } else 1485 wc = (unsigned char)*q++; 1486 break; 1487 case '*': 1488 c = *p; 1489 while (c == CTLQUOTEMARK || c == '*') 1490 c = *++p; 1491 /* 1492 * If the pattern ends here, we know the string 1493 * matches without needing to look at the rest of it. 1494 */ 1495 if (c == '\0') 1496 return 1; 1497 /* 1498 * First try the shortest match for the '*' that 1499 * could work. We can forget any earlier '*' since 1500 * there is no way having it match more characters 1501 * can help us, given that we are already here. 1502 */ 1503 bt_p = p; 1504 bt_q = q; 1505 break; 1506 case '[': { 1507 const char *endp; 1508 int invert, found; 1509 wchar_t chr; 1510 1511 endp = p; 1512 if (*endp == '!' || *endp == '^') 1513 endp++; 1514 for (;;) { 1515 while (*endp == CTLQUOTEMARK) 1516 endp++; 1517 if (*endp == 0) 1518 goto dft; /* no matching ] */ 1519 if (*endp == CTLESC) 1520 endp++; 1521 if (*++endp == ']') 1522 break; 1523 } 1524 invert = 0; 1525 if (*p == '!' || *p == '^') { 1526 invert++; 1527 p++; 1528 } 1529 found = 0; 1530 if (squoted && *q == CTLESC) 1531 q++; 1532 if (*q == '\0') 1533 return 0; 1534 if (localeisutf8) { 1535 chr = get_wc(&q); 1536 if (chr == 0) 1537 goto backtrack; 1538 } else 1539 chr = (unsigned char)*q++; 1540 c = *p++; 1541 do { 1542 if (c == CTLQUOTEMARK) 1543 continue; 1544 if (c == '[' && *p == ':') { 1545 found |= match_charclass(p, chr, &end); 1546 if (end != NULL) 1547 p = end; 1548 } 1549 if (c == CTLESC) 1550 c = *p++; 1551 if (localeisutf8 && c & 0x80) { 1552 p--; 1553 wc = get_wc(&p); 1554 if (wc == 0) /* bad utf-8 */ 1555 return 0; 1556 } else 1557 wc = (unsigned char)c; 1558 if (*p == '-' && p[1] != ']') { 1559 p++; 1560 while (*p == CTLQUOTEMARK) 1561 p++; 1562 if (*p == CTLESC) 1563 p++; 1564 if (localeisutf8) { 1565 wc2 = get_wc(&p); 1566 if (wc2 == 0) /* bad utf-8 */ 1567 return 0; 1568 } else 1569 wc2 = (unsigned char)*p++; 1570 if ( collate_range_cmp(chr, wc) >= 0 1571 && collate_range_cmp(chr, wc2) <= 0 1572 ) 1573 found = 1; 1574 } else { 1575 if (chr == wc) 1576 found = 1; 1577 } 1578 } while ((c = *p++) != ']'); 1579 if (found == invert) 1580 goto backtrack; 1581 break; 1582 } 1583dft: default: 1584 if (squoted && *q == CTLESC) 1585 q++; 1586 if (*q == '\0') 1587 return 0; 1588 if (*q++ == c) 1589 break; 1590backtrack: 1591 /* 1592 * If we have a mismatch (other than hitting the end 1593 * of the string), go back to the last '*' seen and 1594 * have it match one additional character. 1595 */ 1596 if (bt_p == NULL) 1597 return 0; 1598 if (squoted && *bt_q == CTLESC) 1599 bt_q++; 1600 if (*bt_q == '\0') 1601 return 0; 1602 bt_q++; 1603 p = bt_p; 1604 q = bt_q; 1605 break; 1606 } 1607 } 1608} 1609 1610 1611 1612/* 1613 * Remove any CTLESC and CTLQUOTEMARK characters from a string. 1614 */ 1615 1616void 1617rmescapes(char *str) 1618{ 1619 char *p, *q; 1620 1621 p = str; 1622 while (*p != CTLESC && *p != CTLQUOTEMARK && *p != CTLQUOTEEND) { 1623 if (*p++ == '\0') 1624 return; 1625 } 1626 q = p; 1627 while (*p) { 1628 if (*p == CTLQUOTEMARK || *p == CTLQUOTEEND) { 1629 p++; 1630 continue; 1631 } 1632 if (*p == CTLESC) 1633 p++; 1634 *q++ = *p++; 1635 } 1636 *q = '\0'; 1637} 1638 1639 1640 1641/* 1642 * See if a pattern matches in a case statement. 1643 */ 1644 1645int 1646casematch(union node *pattern, const char *val) 1647{ 1648 struct stackmark smark; 1649 int result; 1650 char *p; 1651 1652 setstackmark(&smark); 1653 argbackq = pattern->narg.backquote; 1654 STARTSTACKSTR(expdest); 1655 ifslastp = NULL; 1656 argstr(pattern->narg.text, EXP_TILDE | EXP_CASE); 1657 STPUTC('\0', expdest); 1658 p = grabstackstr(expdest); 1659 result = patmatch(p, val, 0); 1660 popstackmark(&smark); 1661 return result; 1662} 1663 1664/* 1665 * Our own itoa(). 1666 */ 1667 1668static char * 1669cvtnum(int num, char *buf) 1670{ 1671 char temp[32]; 1672 int neg = num < 0; 1673 char *p = temp + 31; 1674 1675 temp[31] = '\0'; 1676 1677 do { 1678 *--p = num % 10 + '0'; 1679 } while ((num /= 10) != 0); 1680 1681 if (neg) 1682 *--p = '-'; 1683 1684 STPUTS(p, buf); 1685 return buf; 1686} 1687 1688/* 1689 * Do most of the work for wordexp(3). 1690 */ 1691 1692int 1693wordexpcmd(int argc, char **argv) 1694{ 1695 size_t len; 1696 int i; 1697 1698 out1fmt("%08x", argc - 1); 1699 for (i = 1, len = 0; i < argc; i++) 1700 len += strlen(argv[i]); 1701 out1fmt("%08x", (int)len); 1702 for (i = 1; i < argc; i++) 1703 outbin(argv[i], strlen(argv[i]) + 1, out1); 1704 return (0); 1705} 1706