1 /*- 2 * Copyright (c) 1991, 1993 3 * The Regents of the University of California. All rights reserved. 4 * 5 * This code is derived from software contributed to Berkeley by 6 * Kenneth Almquist. 7 * 8 * Redistribution and use in source and binary forms, with or without 9 * modification, are permitted provided that the following conditions 10 * are met: 11 * 1. Redistributions of source code must retain the above copyright 12 * notice, this list of conditions and the following disclaimer. 13 * 2. Redistributions in binary form must reproduce the above copyright 14 * notice, this list of conditions and the following disclaimer in the 15 * documentation and/or other materials provided with the distribution. 16 * 3. All advertising materials mentioning features or use of this software 17 * must display the following acknowledgement: 18 * This product includes software developed by the University of 19 * California, Berkeley and its contributors. 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 * @(#)parser.c 8.7 (Berkeley) 5/16/95 37 * $FreeBSD: src/bin/sh/parser.c,v 1.105 2011/02/05 15:02:19 jilles Exp $ 38 */ 39 40 #include <stdio.h> 41 #include <stdlib.h> 42 #include <unistd.h> 43 44 #include "shell.h" 45 #include "parser.h" 46 #include "nodes.h" 47 #include "expand.h" /* defines rmescapes() */ 48 #include "syntax.h" 49 #include "options.h" 50 #include "input.h" 51 #include "output.h" 52 #include "var.h" 53 #include "error.h" 54 #include "memalloc.h" 55 #include "mystring.h" 56 #include "alias.h" 57 #include "show.h" 58 #include "eval.h" 59 #include "exec.h" /* to check for special builtins */ 60 #ifndef NO_HISTORY 61 #include "myhistedit.h" 62 #endif 63 64 /* 65 * Shell command parser. 66 */ 67 68 #define EOFMARKLEN 79 69 #define PROMPTLEN 128 70 71 /* values of checkkwd variable */ 72 #define CHKALIAS 0x1 73 #define CHKKWD 0x2 74 #define CHKNL 0x4 75 76 /* values returned by readtoken */ 77 #include "token.h" 78 79 80 81 struct heredoc { 82 struct heredoc *next; /* next here document in list */ 83 union node *here; /* redirection node */ 84 char *eofmark; /* string indicating end of input */ 85 int striptabs; /* if set, strip leading tabs */ 86 }; 87 88 struct parser_temp { 89 struct parser_temp *next; 90 void *data; 91 }; 92 93 94 static struct heredoc *heredoclist; /* list of here documents to read */ 95 static int doprompt; /* if set, prompt the user */ 96 static int needprompt; /* true if interactive and at start of line */ 97 static int lasttoken; /* last token read */ 98 MKINIT int tokpushback; /* last token pushed back */ 99 static char *wordtext; /* text of last word returned by readtoken */ 100 MKINIT int checkkwd; /* 1 == check for kwds, 2 == also eat newlines */ 101 static struct nodelist *backquotelist; 102 static union node *redirnode; 103 static struct heredoc *heredoc; 104 static int quoteflag; /* set if (part of) last token was quoted */ 105 static int startlinno; /* line # where last token started */ 106 static int funclinno; /* line # where the current function started */ 107 static struct parser_temp *parser_temp; 108 109 110 static union node *list(int, int); 111 static union node *andor(void); 112 static union node *pipeline(void); 113 static union node *command(void); 114 static union node *simplecmd(union node **, union node *); 115 static union node *makename(void); 116 static void parsefname(void); 117 static void parseheredoc(void); 118 static int peektoken(void); 119 static int readtoken(void); 120 static int xxreadtoken(void); 121 static int readtoken1(int, char const *, char *, int); 122 static int noexpand(char *); 123 static void synexpect(int) __dead2; 124 static void synerror(const char *) __dead2; 125 static void setprompt(int); 126 127 128 static void * 129 parser_temp_alloc(size_t len) 130 { 131 struct parser_temp *t; 132 133 INTOFF; 134 t = ckmalloc(sizeof(*t)); 135 t->data = NULL; 136 t->next = parser_temp; 137 parser_temp = t; 138 t->data = ckmalloc(len); 139 INTON; 140 return t->data; 141 } 142 143 144 static void * 145 parser_temp_realloc(void *ptr, size_t len) 146 { 147 struct parser_temp *t; 148 149 INTOFF; 150 t = parser_temp; 151 if (ptr != t->data) 152 error("bug: parser_temp_realloc misused"); 153 t->data = ckrealloc(t->data, len); 154 INTON; 155 return t->data; 156 } 157 158 159 static void 160 parser_temp_free_upto(void *ptr) 161 { 162 struct parser_temp *t; 163 int done = 0; 164 165 INTOFF; 166 while (parser_temp != NULL && !done) { 167 t = parser_temp; 168 parser_temp = t->next; 169 done = t->data == ptr; 170 ckfree(t->data); 171 ckfree(t); 172 } 173 INTON; 174 if (!done) 175 error("bug: parser_temp_free_upto misused"); 176 } 177 178 179 static void 180 parser_temp_free_all(void) 181 { 182 struct parser_temp *t; 183 184 INTOFF; 185 while (parser_temp != NULL) { 186 t = parser_temp; 187 parser_temp = t->next; 188 ckfree(t->data); 189 ckfree(t); 190 } 191 INTON; 192 } 193 194 195 /* 196 * Read and parse a command. Returns NEOF on end of file. (NULL is a 197 * valid parse tree indicating a blank line.) 198 */ 199 200 union node * 201 parsecmd(int interact) 202 { 203 int t; 204 205 /* This assumes the parser is not re-entered, 206 * which could happen if we add command substitution on PS1/PS2. 207 */ 208 parser_temp_free_all(); 209 heredoclist = NULL; 210 211 tokpushback = 0; 212 doprompt = interact; 213 if (doprompt) 214 setprompt(1); 215 else 216 setprompt(0); 217 needprompt = 0; 218 t = readtoken(); 219 if (t == TEOF) 220 return NEOF; 221 if (t == TNL) 222 return NULL; 223 tokpushback++; 224 return list(1, 1); 225 } 226 227 228 static union node * 229 list(int nlflag, int erflag) 230 { 231 union node *ntop, *n1, *n2, *n3; 232 int tok; 233 234 checkkwd = CHKNL | CHKKWD | CHKALIAS; 235 if (!nlflag && !erflag && tokendlist[peektoken()]) 236 return NULL; 237 ntop = n1 = NULL; 238 for (;;) { 239 n2 = andor(); 240 tok = readtoken(); 241 if (tok == TBACKGND) { 242 if (n2->type == NCMD || n2->type == NPIPE) { 243 n2->ncmd.backgnd = 1; 244 } else if (n2->type == NREDIR) { 245 n2->type = NBACKGND; 246 } else { 247 n3 = (union node *)stalloc(sizeof (struct nredir)); 248 n3->type = NBACKGND; 249 n3->nredir.n = n2; 250 n3->nredir.redirect = NULL; 251 n2 = n3; 252 } 253 } 254 if (ntop == NULL) 255 ntop = n2; 256 else if (n1 == NULL) { 257 n1 = (union node *)stalloc(sizeof (struct nbinary)); 258 n1->type = NSEMI; 259 n1->nbinary.ch1 = ntop; 260 n1->nbinary.ch2 = n2; 261 ntop = n1; 262 } 263 else { 264 n3 = (union node *)stalloc(sizeof (struct nbinary)); 265 n3->type = NSEMI; 266 n3->nbinary.ch1 = n1->nbinary.ch2; 267 n3->nbinary.ch2 = n2; 268 n1->nbinary.ch2 = n3; 269 n1 = n3; 270 } 271 switch (tok) { 272 case TBACKGND: 273 case TSEMI: 274 tok = readtoken(); 275 /* FALLTHROUGH */ 276 case TNL: 277 if (tok == TNL) { 278 parseheredoc(); 279 if (nlflag) 280 return ntop; 281 } else if (tok == TEOF && nlflag) { 282 parseheredoc(); 283 return ntop; 284 } else { 285 tokpushback++; 286 } 287 checkkwd = CHKNL | CHKKWD | CHKALIAS; 288 if (!nlflag && !erflag && tokendlist[peektoken()]) 289 return ntop; 290 break; 291 case TEOF: 292 if (heredoclist) 293 parseheredoc(); 294 else 295 pungetc(); /* push back EOF on input */ 296 return ntop; 297 default: 298 if (nlflag || erflag) 299 synexpect(-1); 300 tokpushback++; 301 return ntop; 302 } 303 } 304 } 305 306 307 308 static union node * 309 andor(void) 310 { 311 union node *n1, *n2, *n3; 312 int t; 313 314 n1 = pipeline(); 315 for (;;) { 316 if ((t = readtoken()) == TAND) { 317 t = NAND; 318 } else if (t == TOR) { 319 t = NOR; 320 } else { 321 tokpushback++; 322 return n1; 323 } 324 n2 = pipeline(); 325 n3 = (union node *)stalloc(sizeof (struct nbinary)); 326 n3->type = t; 327 n3->nbinary.ch1 = n1; 328 n3->nbinary.ch2 = n2; 329 n1 = n3; 330 } 331 } 332 333 334 335 static union node * 336 pipeline(void) 337 { 338 union node *n1, *n2, *pipenode; 339 struct nodelist *lp, *prev; 340 int negate, t; 341 342 negate = 0; 343 checkkwd = CHKNL | CHKKWD | CHKALIAS; 344 TRACE(("pipeline: entered\n")); 345 while (readtoken() == TNOT) 346 negate = !negate; 347 tokpushback++; 348 n1 = command(); 349 if (readtoken() == TPIPE) { 350 pipenode = (union node *)stalloc(sizeof (struct npipe)); 351 pipenode->type = NPIPE; 352 pipenode->npipe.backgnd = 0; 353 lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 354 pipenode->npipe.cmdlist = lp; 355 lp->n = n1; 356 do { 357 prev = lp; 358 lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 359 checkkwd = CHKNL | CHKKWD | CHKALIAS; 360 t = readtoken(); 361 tokpushback++; 362 if (t == TNOT) 363 lp->n = pipeline(); 364 else 365 lp->n = command(); 366 prev->next = lp; 367 } while (readtoken() == TPIPE); 368 lp->next = NULL; 369 n1 = pipenode; 370 } 371 tokpushback++; 372 if (negate) { 373 n2 = (union node *)stalloc(sizeof (struct nnot)); 374 n2->type = NNOT; 375 n2->nnot.com = n1; 376 return n2; 377 } else 378 return n1; 379 } 380 381 382 383 static union node * 384 command(void) 385 { 386 union node *n1, *n2; 387 union node *ap, **app; 388 union node *cp, **cpp; 389 union node *redir, **rpp; 390 int t; 391 int is_subshell; 392 393 checkkwd = CHKNL | CHKKWD | CHKALIAS; 394 is_subshell = 0; 395 redir = NULL; 396 n1 = NULL; 397 rpp = &redir; 398 399 /* Check for redirection which may precede command */ 400 while (readtoken() == TREDIR) { 401 *rpp = n2 = redirnode; 402 rpp = &n2->nfile.next; 403 parsefname(); 404 } 405 tokpushback++; 406 407 switch (readtoken()) { 408 case TIF: 409 n1 = (union node *)stalloc(sizeof (struct nif)); 410 n1->type = NIF; 411 if ((n1->nif.test = list(0, 0)) == NULL) 412 synexpect(-1); 413 if (readtoken() != TTHEN) 414 synexpect(TTHEN); 415 n1->nif.ifpart = list(0, 0); 416 n2 = n1; 417 while (readtoken() == TELIF) { 418 n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif)); 419 n2 = n2->nif.elsepart; 420 n2->type = NIF; 421 if ((n2->nif.test = list(0, 0)) == NULL) 422 synexpect(-1); 423 if (readtoken() != TTHEN) 424 synexpect(TTHEN); 425 n2->nif.ifpart = list(0, 0); 426 } 427 if (lasttoken == TELSE) 428 n2->nif.elsepart = list(0, 0); 429 else { 430 n2->nif.elsepart = NULL; 431 tokpushback++; 432 } 433 if (readtoken() != TFI) 434 synexpect(TFI); 435 checkkwd = CHKKWD | CHKALIAS; 436 break; 437 case TWHILE: 438 case TUNTIL: { 439 int got; 440 n1 = (union node *)stalloc(sizeof (struct nbinary)); 441 n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL; 442 if ((n1->nbinary.ch1 = list(0, 0)) == NULL) 443 synexpect(-1); 444 if ((got=readtoken()) != TDO) { 445 TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : "")); 446 synexpect(TDO); 447 } 448 n1->nbinary.ch2 = list(0, 0); 449 if (readtoken() != TDONE) 450 synexpect(TDONE); 451 checkkwd = CHKKWD | CHKALIAS; 452 break; 453 } 454 case TFOR: 455 if (readtoken() != TWORD || quoteflag || ! goodname(wordtext)) 456 synerror("Bad for loop variable"); 457 n1 = (union node *)stalloc(sizeof (struct nfor)); 458 n1->type = NFOR; 459 n1->nfor.var = wordtext; 460 while (readtoken() == TNL) 461 ; 462 if (lasttoken == TWORD && ! quoteflag && equal(wordtext, "in")) { 463 app = ≈ 464 while (readtoken() == TWORD) { 465 n2 = (union node *)stalloc(sizeof (struct narg)); 466 n2->type = NARG; 467 n2->narg.text = wordtext; 468 n2->narg.backquote = backquotelist; 469 *app = n2; 470 app = &n2->narg.next; 471 } 472 *app = NULL; 473 n1->nfor.args = ap; 474 if (lasttoken != TNL && lasttoken != TSEMI) 475 synexpect(-1); 476 } else { 477 static char argvars[5] = { 478 CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0' 479 }; 480 n2 = (union node *)stalloc(sizeof (struct narg)); 481 n2->type = NARG; 482 n2->narg.text = argvars; 483 n2->narg.backquote = NULL; 484 n2->narg.next = NULL; 485 n1->nfor.args = n2; 486 /* 487 * Newline or semicolon here is optional (but note 488 * that the original Bourne shell only allowed NL). 489 */ 490 if (lasttoken != TNL && lasttoken != TSEMI) 491 tokpushback++; 492 } 493 checkkwd = CHKNL | CHKKWD | CHKALIAS; 494 if ((t = readtoken()) == TDO) 495 t = TDONE; 496 else if (t == TBEGIN) 497 t = TEND; 498 else 499 synexpect(-1); 500 n1->nfor.body = list(0, 0); 501 if (readtoken() != t) 502 synexpect(t); 503 checkkwd = CHKKWD | CHKALIAS; 504 break; 505 case TCASE: 506 n1 = (union node *)stalloc(sizeof (struct ncase)); 507 n1->type = NCASE; 508 if (readtoken() != TWORD) 509 synexpect(TWORD); 510 n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg)); 511 n2->type = NARG; 512 n2->narg.text = wordtext; 513 n2->narg.backquote = backquotelist; 514 n2->narg.next = NULL; 515 while (readtoken() == TNL); 516 if (lasttoken != TWORD || ! equal(wordtext, "in")) 517 synerror("expecting \"in\""); 518 cpp = &n1->ncase.cases; 519 checkkwd = CHKNL | CHKKWD, readtoken(); 520 while (lasttoken != TESAC) { 521 *cpp = cp = (union node *)stalloc(sizeof (struct nclist)); 522 cp->type = NCLIST; 523 app = &cp->nclist.pattern; 524 if (lasttoken == TLP) 525 readtoken(); 526 for (;;) { 527 *app = ap = (union node *)stalloc(sizeof (struct narg)); 528 ap->type = NARG; 529 ap->narg.text = wordtext; 530 ap->narg.backquote = backquotelist; 531 checkkwd = CHKNL | CHKKWD; 532 if (readtoken() != TPIPE) 533 break; 534 app = &ap->narg.next; 535 readtoken(); 536 } 537 ap->narg.next = NULL; 538 if (lasttoken != TRP) 539 synexpect(TRP); 540 cp->nclist.body = list(0, 0); 541 542 checkkwd = CHKNL | CHKKWD | CHKALIAS; 543 if ((t = readtoken()) != TESAC) { 544 if (t != TENDCASE) 545 synexpect(TENDCASE); 546 else 547 checkkwd = CHKNL | CHKKWD, readtoken(); 548 } 549 cpp = &cp->nclist.next; 550 } 551 *cpp = NULL; 552 checkkwd = CHKKWD | CHKALIAS; 553 break; 554 case TLP: 555 n1 = (union node *)stalloc(sizeof (struct nredir)); 556 n1->type = NSUBSHELL; 557 n1->nredir.n = list(0, 0); 558 n1->nredir.redirect = NULL; 559 if (readtoken() != TRP) 560 synexpect(TRP); 561 checkkwd = CHKKWD | CHKALIAS; 562 is_subshell = 1; 563 break; 564 case TBEGIN: 565 n1 = list(0, 0); 566 if (readtoken() != TEND) 567 synexpect(TEND); 568 checkkwd = CHKKWD | CHKALIAS; 569 break; 570 /* Handle an empty command like other simple commands. */ 571 case TBACKGND: 572 case TSEMI: 573 case TAND: 574 case TOR: 575 /* 576 * An empty command before a ; doesn't make much sense, and 577 * should certainly be disallowed in the case of `if ;'. 578 */ 579 if (!redir) 580 synexpect(-1); 581 case TNL: 582 case TEOF: 583 case TWORD: 584 case TRP: 585 tokpushback++; 586 n1 = simplecmd(rpp, redir); 587 return n1; 588 default: 589 synexpect(-1); 590 } 591 592 /* Now check for redirection which may follow command */ 593 while (readtoken() == TREDIR) { 594 *rpp = n2 = redirnode; 595 rpp = &n2->nfile.next; 596 parsefname(); 597 } 598 tokpushback++; 599 *rpp = NULL; 600 if (redir) { 601 if (!is_subshell) { 602 n2 = (union node *)stalloc(sizeof (struct nredir)); 603 n2->type = NREDIR; 604 n2->nredir.n = n1; 605 n1 = n2; 606 } 607 n1->nredir.redirect = redir; 608 } 609 610 return n1; 611 } 612 613 614 static union node * 615 simplecmd(union node **rpp, union node *redir) 616 { 617 union node *args, **app; 618 union node **orig_rpp = rpp; 619 union node *n = NULL; 620 int special; 621 622 /* If we don't have any redirections already, then we must reset */ 623 /* rpp to be the address of the local redir variable. */ 624 if (redir == 0) 625 rpp = &redir; 626 627 args = NULL; 628 app = &args; 629 /* 630 * We save the incoming value, because we need this for shell 631 * functions. There can not be a redirect or an argument between 632 * the function name and the open parenthesis. 633 */ 634 orig_rpp = rpp; 635 636 for (;;) { 637 if (readtoken() == TWORD) { 638 n = (union node *)stalloc(sizeof (struct narg)); 639 n->type = NARG; 640 n->narg.text = wordtext; 641 n->narg.backquote = backquotelist; 642 *app = n; 643 app = &n->narg.next; 644 } else if (lasttoken == TREDIR) { 645 *rpp = n = redirnode; 646 rpp = &n->nfile.next; 647 parsefname(); /* read name of redirection file */ 648 } else if (lasttoken == TLP && app == &args->narg.next 649 && rpp == orig_rpp) { 650 /* We have a function */ 651 if (readtoken() != TRP) 652 synexpect(TRP); 653 funclinno = plinno; 654 /* 655 * - Require plain text. 656 * - Functions with '/' cannot be called. 657 * - Reject name=(). 658 * - Reject ksh extended glob patterns. 659 */ 660 if (!noexpand(n->narg.text) || quoteflag || 661 strchr(n->narg.text, '/') || 662 strchr("!%*+-=?@}~", 663 n->narg.text[strlen(n->narg.text) - 1])) 664 synerror("Bad function name"); 665 rmescapes(n->narg.text); 666 if (find_builtin(n->narg.text, &special) >= 0 && 667 special) 668 synerror("Cannot override a special builtin with a function"); 669 n->type = NDEFUN; 670 n->narg.next = command(); 671 funclinno = 0; 672 return n; 673 } else { 674 tokpushback++; 675 break; 676 } 677 } 678 *app = NULL; 679 *rpp = NULL; 680 n = (union node *)stalloc(sizeof (struct ncmd)); 681 n->type = NCMD; 682 n->ncmd.backgnd = 0; 683 n->ncmd.args = args; 684 n->ncmd.redirect = redir; 685 return n; 686 } 687 688 static union node * 689 makename(void) 690 { 691 union node *n; 692 693 n = (union node *)stalloc(sizeof (struct narg)); 694 n->type = NARG; 695 n->narg.next = NULL; 696 n->narg.text = wordtext; 697 n->narg.backquote = backquotelist; 698 return n; 699 } 700 701 void 702 fixredir(union node *n, const char *text, int err) 703 { 704 TRACE(("Fix redir %s %d\n", text, err)); 705 if (!err) 706 n->ndup.vname = NULL; 707 708 if (is_digit(text[0]) && text[1] == '\0') 709 n->ndup.dupfd = digit_val(text[0]); 710 else if (text[0] == '-' && text[1] == '\0') 711 n->ndup.dupfd = -1; 712 else { 713 714 if (err) 715 synerror("Bad fd number"); 716 else 717 n->ndup.vname = makename(); 718 } 719 } 720 721 722 static void 723 parsefname(void) 724 { 725 union node *n = redirnode; 726 727 if (readtoken() != TWORD) 728 synexpect(-1); 729 if (n->type == NHERE) { 730 struct heredoc *here = heredoc; 731 struct heredoc *p; 732 int i; 733 734 if (quoteflag == 0) 735 n->type = NXHERE; 736 TRACE(("Here document %d\n", n->type)); 737 if (here->striptabs) { 738 while (*wordtext == '\t') 739 wordtext++; 740 } 741 if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN) 742 synerror("Illegal eof marker for << redirection"); 743 rmescapes(wordtext); 744 here->eofmark = wordtext; 745 here->next = NULL; 746 if (heredoclist == NULL) 747 heredoclist = here; 748 else { 749 for (p = heredoclist ; p->next ; p = p->next); 750 p->next = here; 751 } 752 } else if (n->type == NTOFD || n->type == NFROMFD) { 753 fixredir(n, wordtext, 0); 754 } else { 755 n->nfile.fname = makename(); 756 } 757 } 758 759 760 /* 761 * Input any here documents. 762 */ 763 764 static void 765 parseheredoc(void) 766 { 767 struct heredoc *here; 768 union node *n; 769 770 while (heredoclist) { 771 here = heredoclist; 772 heredoclist = here->next; 773 if (needprompt) { 774 setprompt(2); 775 needprompt = 0; 776 } 777 readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX, 778 here->eofmark, here->striptabs); 779 n = (union node *)stalloc(sizeof (struct narg)); 780 n->narg.type = NARG; 781 n->narg.next = NULL; 782 n->narg.text = wordtext; 783 n->narg.backquote = backquotelist; 784 here->here->nhere.doc = n; 785 } 786 } 787 788 static int 789 peektoken(void) 790 { 791 int t; 792 793 t = readtoken(); 794 tokpushback++; 795 return (t); 796 } 797 798 static int 799 readtoken(void) 800 { 801 int t; 802 struct alias *ap; 803 #ifdef DEBUG 804 int alreadyseen = tokpushback; 805 #endif 806 807 top: 808 t = xxreadtoken(); 809 810 /* 811 * eat newlines 812 */ 813 if (checkkwd & CHKNL) { 814 while (t == TNL) { 815 parseheredoc(); 816 t = xxreadtoken(); 817 } 818 } 819 820 /* 821 * check for keywords and aliases 822 */ 823 if (t == TWORD && !quoteflag) 824 { 825 const char * const *pp; 826 827 if (checkkwd & CHKKWD) 828 for (pp = parsekwd; *pp; pp++) { 829 if (**pp == *wordtext && equal(*pp, wordtext)) 830 { 831 lasttoken = t = pp - parsekwd + KWDOFFSET; 832 TRACE(("keyword %s recognized\n", tokname[t])); 833 goto out; 834 } 835 } 836 if (checkkwd & CHKALIAS && 837 (ap = lookupalias(wordtext, 1)) != NULL) { 838 pushstring(ap->val, strlen(ap->val), ap); 839 goto top; 840 } 841 } 842 out: 843 if (t != TNOT) 844 checkkwd = 0; 845 846 #ifdef DEBUG 847 if (!alreadyseen) 848 TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); 849 else 850 TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); 851 #endif 852 return (t); 853 } 854 855 856 /* 857 * Read the next input token. 858 * If the token is a word, we set backquotelist to the list of cmds in 859 * backquotes. We set quoteflag to true if any part of the word was 860 * quoted. 861 * If the token is TREDIR, then we set redirnode to a structure containing 862 * the redirection. 863 * In all cases, the variable startlinno is set to the number of the line 864 * on which the token starts. 865 * 866 * [Change comment: here documents and internal procedures] 867 * [Readtoken shouldn't have any arguments. Perhaps we should make the 868 * word parsing code into a separate routine. In this case, readtoken 869 * doesn't need to have any internal procedures, but parseword does. 870 * We could also make parseoperator in essence the main routine, and 871 * have parseword (readtoken1?) handle both words and redirection.] 872 */ 873 874 #define RETURN(token) return lasttoken = token 875 876 static int 877 xxreadtoken(void) 878 { 879 int c; 880 881 if (tokpushback) { 882 tokpushback = 0; 883 return lasttoken; 884 } 885 if (needprompt) { 886 setprompt(2); 887 needprompt = 0; 888 } 889 startlinno = plinno; 890 for (;;) { /* until token or start of word found */ 891 c = pgetc_macro(); 892 switch (c) { 893 case ' ': case '\t': 894 continue; 895 case '#': 896 while ((c = pgetc()) != '\n' && c != PEOF); 897 pungetc(); 898 continue; 899 case '\\': 900 if (pgetc() == '\n') { 901 startlinno = ++plinno; 902 if (doprompt) 903 setprompt(2); 904 else 905 setprompt(0); 906 continue; 907 } 908 pungetc(); 909 goto breakloop; 910 case '\n': 911 plinno++; 912 needprompt = doprompt; 913 RETURN(TNL); 914 case PEOF: 915 RETURN(TEOF); 916 case '&': 917 if (pgetc() == '&') 918 RETURN(TAND); 919 pungetc(); 920 RETURN(TBACKGND); 921 case '|': 922 if (pgetc() == '|') 923 RETURN(TOR); 924 pungetc(); 925 RETURN(TPIPE); 926 case ';': 927 if (pgetc() == ';') 928 RETURN(TENDCASE); 929 pungetc(); 930 RETURN(TSEMI); 931 case '(': 932 RETURN(TLP); 933 case ')': 934 RETURN(TRP); 935 default: 936 goto breakloop; 937 } 938 } 939 breakloop: 940 return readtoken1(c, BASESYNTAX, NULL, 0); 941 #undef RETURN 942 } 943 944 945 #define MAXNEST_STATIC 8 946 struct tokenstate 947 { 948 const char *syntax; /* *SYNTAX */ 949 int parenlevel; /* levels of parentheses in arithmetic */ 950 enum tokenstate_category 951 { 952 TSTATE_TOP, 953 TSTATE_VAR_OLD, /* ${var+-=?}, inherits dquotes */ 954 TSTATE_VAR_NEW, /* other ${var...}, own dquote state */ 955 TSTATE_ARITH 956 } category; 957 }; 958 959 960 /* 961 * Called to parse command substitutions. 962 */ 963 964 static char * 965 parsebackq(char *out, struct nodelist **pbqlist, 966 int oldstyle, int dblquote, int quoted) 967 { 968 struct nodelist **nlpp; 969 union node *n; 970 char *volatile str; 971 struct jmploc jmploc; 972 struct jmploc *const savehandler = handler; 973 int savelen; 974 int saveprompt; 975 const int bq_startlinno = plinno; 976 char *volatile ostr = NULL; 977 struct parsefile *const savetopfile = getcurrentfile(); 978 struct heredoc *const saveheredoclist = heredoclist; 979 struct heredoc *here; 980 981 str = NULL; 982 if (setjmp(jmploc.loc)) { 983 popfilesupto(savetopfile); 984 if (str) 985 ckfree(str); 986 if (ostr) 987 ckfree(ostr); 988 heredoclist = saveheredoclist; 989 handler = savehandler; 990 if (exception == EXERROR) { 991 startlinno = bq_startlinno; 992 synerror("Error in command substitution"); 993 } 994 longjmp(handler->loc, 1); 995 } 996 INTOFF; 997 savelen = out - stackblock(); 998 if (savelen > 0) { 999 str = ckmalloc(savelen); 1000 memcpy(str, stackblock(), savelen); 1001 } 1002 handler = &jmploc; 1003 heredoclist = NULL; 1004 INTON; 1005 if (oldstyle) { 1006 /* 1007 * We must read until the closing backquote, giving special 1008 * treatment to some slashes, and then push the string and 1009 * reread it as input, interpreting it normally. 1010 */ 1011 char *oout; 1012 int c, olen; 1013 1014 STARTSTACKSTR(oout); 1015 for (;;) { 1016 if (needprompt) { 1017 setprompt(2); 1018 needprompt = 0; 1019 } 1020 CHECKSTRSPACE(2, oout); 1021 switch (c = pgetc()) { 1022 case '`': 1023 goto done; 1024 1025 case '\\': 1026 if ((c = pgetc()) == '\n') { 1027 plinno++; 1028 if (doprompt) 1029 setprompt(2); 1030 else 1031 setprompt(0); 1032 /* 1033 * If eating a newline, avoid putting 1034 * the newline into the new character 1035 * stream (via the USTPUTC after the 1036 * switch). 1037 */ 1038 continue; 1039 } 1040 if (c != '\\' && c != '`' && c != '$' 1041 && (!dblquote || c != '"')) 1042 USTPUTC('\\', oout); 1043 break; 1044 1045 case '\n': 1046 plinno++; 1047 needprompt = doprompt; 1048 break; 1049 1050 case PEOF: 1051 startlinno = plinno; 1052 synerror("EOF in backquote substitution"); 1053 break; 1054 1055 default: 1056 break; 1057 } 1058 USTPUTC(c, oout); 1059 } 1060 done: 1061 USTPUTC('\0', oout); 1062 olen = oout - stackblock(); 1063 INTOFF; 1064 ostr = ckmalloc(olen); 1065 memcpy(ostr, stackblock(), olen); 1066 setinputstring(ostr, 1); 1067 INTON; 1068 } 1069 nlpp = pbqlist; 1070 while (*nlpp) 1071 nlpp = &(*nlpp)->next; 1072 *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 1073 (*nlpp)->next = NULL; 1074 1075 if (oldstyle) { 1076 saveprompt = doprompt; 1077 doprompt = 0; 1078 } 1079 1080 n = list(0, oldstyle); 1081 1082 if (oldstyle) 1083 doprompt = saveprompt; 1084 else { 1085 if (readtoken() != TRP) 1086 synexpect(TRP); 1087 } 1088 1089 (*nlpp)->n = n; 1090 if (oldstyle) { 1091 /* 1092 * Start reading from old file again, ignoring any pushed back 1093 * tokens left from the backquote parsing 1094 */ 1095 popfile(); 1096 tokpushback = 0; 1097 } 1098 STARTSTACKSTR(out); 1099 CHECKSTRSPACE(savelen + 1, out); 1100 INTOFF; 1101 if (str) { 1102 memcpy(out, str, savelen); 1103 STADJUST(savelen, out); 1104 ckfree(str); 1105 str = NULL; 1106 } 1107 if (ostr) { 1108 ckfree(ostr); 1109 ostr = NULL; 1110 } 1111 here = saveheredoclist; 1112 if (here != NULL) { 1113 while (here->next != NULL) 1114 here = here->next; 1115 here->next = heredoclist; 1116 heredoclist = saveheredoclist; 1117 } 1118 handler = savehandler; 1119 INTON; 1120 if (quoted) 1121 USTPUTC(CTLBACKQ | CTLQUOTE, out); 1122 else 1123 USTPUTC(CTLBACKQ, out); 1124 return out; 1125 } 1126 1127 1128 /* 1129 * If eofmark is NULL, read a word or a redirection symbol. If eofmark 1130 * is not NULL, read a here document. In the latter case, eofmark is the 1131 * word which marks the end of the document and striptabs is true if 1132 * leading tabs should be stripped from the document. The argument firstc 1133 * is the first character of the input token or document. 1134 * 1135 * Because C does not have internal subroutines, I have simulated them 1136 * using goto's to implement the subroutine linkage. The following macros 1137 * will run code that appears at the end of readtoken1. 1138 */ 1139 1140 #define CHECKEND() {goto checkend; checkend_return:;} 1141 #define PARSEREDIR() {goto parseredir; parseredir_return:;} 1142 #define PARSESUB() {goto parsesub; parsesub_return:;} 1143 #define PARSEARITH() {goto parsearith; parsearith_return:;} 1144 1145 static int 1146 readtoken1(int firstc, char const *initialsyntax, char *eofmark, int striptabs) 1147 { 1148 int c = firstc; 1149 char * volatile out; 1150 int len; 1151 char line[EOFMARKLEN + 1]; 1152 struct nodelist *bqlist; 1153 volatile int quotef; 1154 int newvarnest; 1155 int level; 1156 int synentry; 1157 struct tokenstate state_static[MAXNEST_STATIC]; 1158 int maxnest = MAXNEST_STATIC; 1159 struct tokenstate *state = state_static; 1160 1161 startlinno = plinno; 1162 quotef = 0; 1163 bqlist = NULL; 1164 newvarnest = 0; 1165 level = 0; 1166 state[level].syntax = initialsyntax; 1167 state[level].parenlevel = 0; 1168 state[level].category = TSTATE_TOP; 1169 1170 STARTSTACKSTR(out); 1171 loop: { /* for each line, until end of word */ 1172 CHECKEND(); /* set c to PEOF if at end of here document */ 1173 for (;;) { /* until end of line or end of word */ 1174 CHECKSTRSPACE(4, out); /* permit 4 calls to USTPUTC */ 1175 1176 synentry = state[level].syntax[c]; 1177 1178 switch(synentry) { 1179 case CNL: /* '\n' */ 1180 if (state[level].syntax == BASESYNTAX) 1181 goto endword; /* exit outer loop */ 1182 USTPUTC(c, out); 1183 plinno++; 1184 if (doprompt) 1185 setprompt(2); 1186 else 1187 setprompt(0); 1188 c = pgetc(); 1189 goto loop; /* continue outer loop */ 1190 case CWORD: 1191 USTPUTC(c, out); 1192 break; 1193 case CCTL: 1194 if (eofmark == NULL || initialsyntax != SQSYNTAX) 1195 USTPUTC(CTLESC, out); 1196 USTPUTC(c, out); 1197 break; 1198 case CBACK: /* backslash */ 1199 c = pgetc(); 1200 if (c == PEOF) { 1201 USTPUTC('\\', out); 1202 pungetc(); 1203 } else if (c == '\n') { 1204 plinno++; 1205 if (doprompt) 1206 setprompt(2); 1207 else 1208 setprompt(0); 1209 } else { 1210 if (state[level].syntax == DQSYNTAX && 1211 c != '\\' && c != '`' && c != '$' && 1212 (c != '"' || (eofmark != NULL && 1213 newvarnest == 0)) && 1214 (c != '}' || state[level].category != TSTATE_VAR_OLD)) 1215 USTPUTC('\\', out); 1216 if ((eofmark == NULL || 1217 newvarnest > 0) && 1218 state[level].syntax == BASESYNTAX) 1219 USTPUTC(CTLQUOTEMARK, out); 1220 if (SQSYNTAX[c] == CCTL) 1221 USTPUTC(CTLESC, out); 1222 USTPUTC(c, out); 1223 if ((eofmark == NULL || 1224 newvarnest > 0) && 1225 state[level].syntax == BASESYNTAX && 1226 state[level].category == TSTATE_VAR_OLD) 1227 USTPUTC(CTLQUOTEEND, out); 1228 quotef++; 1229 } 1230 break; 1231 case CSQUOTE: 1232 USTPUTC(CTLQUOTEMARK, out); 1233 state[level].syntax = SQSYNTAX; 1234 break; 1235 case CDQUOTE: 1236 USTPUTC(CTLQUOTEMARK, out); 1237 state[level].syntax = DQSYNTAX; 1238 break; 1239 case CENDQUOTE: 1240 if (eofmark != NULL && newvarnest == 0) 1241 USTPUTC(c, out); 1242 else { 1243 if (state[level].category == TSTATE_VAR_OLD) 1244 USTPUTC(CTLQUOTEEND, out); 1245 state[level].syntax = BASESYNTAX; 1246 quotef++; 1247 } 1248 break; 1249 case CVAR: /* '$' */ 1250 PARSESUB(); /* parse substitution */ 1251 break; 1252 case CENDVAR: /* '}' */ 1253 if (level > 0 && 1254 ((state[level].category == TSTATE_VAR_OLD && 1255 state[level].syntax == 1256 state[level - 1].syntax) || 1257 (state[level].category == TSTATE_VAR_NEW && 1258 state[level].syntax == BASESYNTAX))) { 1259 if (state[level].category == TSTATE_VAR_NEW) 1260 newvarnest--; 1261 level--; 1262 USTPUTC(CTLENDVAR, out); 1263 } else { 1264 USTPUTC(c, out); 1265 } 1266 break; 1267 case CLP: /* '(' in arithmetic */ 1268 state[level].parenlevel++; 1269 USTPUTC(c, out); 1270 break; 1271 case CRP: /* ')' in arithmetic */ 1272 if (state[level].parenlevel > 0) { 1273 USTPUTC(c, out); 1274 --state[level].parenlevel; 1275 } else { 1276 if (pgetc() == ')') { 1277 if (level > 0 && 1278 state[level].category == TSTATE_ARITH) { 1279 level--; 1280 USTPUTC(CTLENDARI, out); 1281 } else 1282 USTPUTC(')', out); 1283 } else { 1284 /* 1285 * unbalanced parens 1286 * (don't 2nd guess - no error) 1287 */ 1288 pungetc(); 1289 USTPUTC(')', out); 1290 } 1291 } 1292 break; 1293 case CBQUOTE: /* '`' */ 1294 out = parsebackq(out, &bqlist, 1, 1295 state[level].syntax == DQSYNTAX && 1296 (eofmark == NULL || newvarnest > 0), 1297 state[level].syntax == DQSYNTAX || state[level].syntax == ARISYNTAX); 1298 break; 1299 case CEOF: 1300 goto endword; /* exit outer loop */ 1301 case CIGN: 1302 break; 1303 default: 1304 if (level == 0) 1305 goto endword; /* exit outer loop */ 1306 USTPUTC(c, out); 1307 } 1308 c = pgetc_macro(); 1309 } 1310 } 1311 endword: 1312 if (state[level].syntax == ARISYNTAX) 1313 synerror("Missing '))'"); 1314 if (state[level].syntax != BASESYNTAX && eofmark == NULL) 1315 synerror("Unterminated quoted string"); 1316 if (state[level].category == TSTATE_VAR_OLD || 1317 state[level].category == TSTATE_VAR_NEW) { 1318 startlinno = plinno; 1319 synerror("Missing '}'"); 1320 } 1321 if (state != state_static) 1322 parser_temp_free_upto(state); 1323 USTPUTC('\0', out); 1324 len = out - stackblock(); 1325 out = stackblock(); 1326 if (eofmark == NULL) { 1327 if ((c == '>' || c == '<') 1328 && quotef == 0 1329 && len <= 2 1330 && (*out == '\0' || is_digit(*out))) { 1331 PARSEREDIR(); 1332 return lasttoken = TREDIR; 1333 } else { 1334 pungetc(); 1335 } 1336 } 1337 quoteflag = quotef; 1338 backquotelist = bqlist; 1339 grabstackblock(len); 1340 wordtext = out; 1341 return lasttoken = TWORD; 1342 /* end of readtoken routine */ 1343 1344 1345 /* 1346 * Check to see whether we are at the end of the here document. When this 1347 * is called, c is set to the first character of the next input line. If 1348 * we are at the end of the here document, this routine sets the c to PEOF. 1349 */ 1350 1351 checkend: { 1352 if (eofmark) { 1353 if (striptabs) { 1354 while (c == '\t') 1355 c = pgetc(); 1356 } 1357 if (c == *eofmark) { 1358 if (pfgets(line, sizeof line) != NULL) { 1359 char *p, *q; 1360 1361 p = line; 1362 for (q = eofmark + 1 ; *q && *p == *q ; p++, q++); 1363 if (*p == '\n' && *q == '\0') { 1364 c = PEOF; 1365 plinno++; 1366 needprompt = doprompt; 1367 } else { 1368 pushstring(line, strlen(line), NULL); 1369 } 1370 } 1371 } 1372 } 1373 goto checkend_return; 1374 } 1375 1376 1377 /* 1378 * Parse a redirection operator. The variable "out" points to a string 1379 * specifying the fd to be redirected. The variable "c" contains the 1380 * first character of the redirection operator. 1381 */ 1382 1383 parseredir: { 1384 char fd = *out; 1385 union node *np; 1386 1387 np = (union node *)stalloc(sizeof (struct nfile)); 1388 if (c == '>') { 1389 np->nfile.fd = 1; 1390 c = pgetc(); 1391 if (c == '>') 1392 np->type = NAPPEND; 1393 else if (c == '&') 1394 np->type = NTOFD; 1395 else if (c == '|') 1396 np->type = NCLOBBER; 1397 else { 1398 np->type = NTO; 1399 pungetc(); 1400 } 1401 } else { /* c == '<' */ 1402 np->nfile.fd = 0; 1403 c = pgetc(); 1404 if (c == '<') { 1405 if (sizeof (struct nfile) != sizeof (struct nhere)) { 1406 np = (union node *)stalloc(sizeof (struct nhere)); 1407 np->nfile.fd = 0; 1408 } 1409 np->type = NHERE; 1410 heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc)); 1411 heredoc->here = np; 1412 if ((c = pgetc()) == '-') { 1413 heredoc->striptabs = 1; 1414 } else { 1415 heredoc->striptabs = 0; 1416 pungetc(); 1417 } 1418 } else if (c == '&') 1419 np->type = NFROMFD; 1420 else if (c == '>') 1421 np->type = NFROMTO; 1422 else { 1423 np->type = NFROM; 1424 pungetc(); 1425 } 1426 } 1427 if (fd != '\0') 1428 np->nfile.fd = digit_val(fd); 1429 redirnode = np; 1430 goto parseredir_return; 1431 } 1432 1433 1434 /* 1435 * Parse a substitution. At this point, we have read the dollar sign 1436 * and nothing else. 1437 */ 1438 1439 parsesub: { 1440 char buf[10]; 1441 int subtype; 1442 int typeloc; 1443 int flags; 1444 char *p; 1445 static const char types[] = "}-+?="; 1446 int bracketed_name = 0; /* used to handle ${[0-9]*} variables */ 1447 int linno; 1448 int length; 1449 1450 c = pgetc(); 1451 if (c != '(' && c != '{' && (is_eof(c) || !is_name(c)) && 1452 !is_special(c)) { 1453 USTPUTC('$', out); 1454 pungetc(); 1455 } else if (c == '(') { /* $(command) or $((arith)) */ 1456 if (pgetc() == '(') { 1457 PARSEARITH(); 1458 } else { 1459 pungetc(); 1460 out = parsebackq(out, &bqlist, 0, 1461 state[level].syntax == DQSYNTAX && 1462 (eofmark == NULL || newvarnest > 0), 1463 state[level].syntax == DQSYNTAX || 1464 state[level].syntax == ARISYNTAX); 1465 } 1466 } else { 1467 USTPUTC(CTLVAR, out); 1468 typeloc = out - stackblock(); 1469 USTPUTC(VSNORMAL, out); 1470 subtype = VSNORMAL; 1471 flags = 0; 1472 if (c == '{') { 1473 bracketed_name = 1; 1474 c = pgetc(); 1475 if (c == '#') { 1476 if ((c = pgetc()) == '}') 1477 c = '#'; 1478 else 1479 subtype = VSLENGTH; 1480 } 1481 else 1482 subtype = 0; 1483 } 1484 if (!is_eof(c) && is_name(c)) { 1485 length = 0; 1486 do { 1487 STPUTC(c, out); 1488 c = pgetc(); 1489 length++; 1490 } while (!is_eof(c) && is_in_name(c)); 1491 if (length == 6 && 1492 strncmp(out - length, "LINENO", length) == 0) { 1493 /* Replace the variable name with the 1494 * current line number. */ 1495 linno = plinno; 1496 if (funclinno != 0) 1497 linno -= funclinno - 1; 1498 snprintf(buf, sizeof(buf), "%d", linno); 1499 STADJUST(-6, out); 1500 STPUTS(buf, out); 1501 flags |= VSLINENO; 1502 } 1503 } else if (is_digit(c)) { 1504 if (bracketed_name) { 1505 do { 1506 STPUTC(c, out); 1507 c = pgetc(); 1508 } while (is_digit(c)); 1509 } else { 1510 STPUTC(c, out); 1511 c = pgetc(); 1512 } 1513 } else { 1514 if (! is_special(c)) { 1515 subtype = VSERROR; 1516 if (c == '}') 1517 pungetc(); 1518 else if (c == '\n' || c == PEOF) 1519 synerror("Unexpected end of line in substitution"); 1520 else 1521 USTPUTC(c, out); 1522 } else { 1523 USTPUTC(c, out); 1524 c = pgetc(); 1525 } 1526 } 1527 if (subtype == 0) { 1528 switch (c) { 1529 case ':': 1530 flags |= VSNUL; 1531 c = pgetc(); 1532 /*FALLTHROUGH*/ 1533 default: 1534 p = strchr(types, c); 1535 if (p == NULL) { 1536 if (c == '\n' || c == PEOF) 1537 synerror("Unexpected end of line in substitution"); 1538 if (flags == VSNUL) 1539 STPUTC(':', out); 1540 STPUTC(c, out); 1541 subtype = VSERROR; 1542 } else 1543 subtype = p - types + VSNORMAL; 1544 break; 1545 case '%': 1546 case '#': 1547 { 1548 int cc = c; 1549 subtype = c == '#' ? VSTRIMLEFT : 1550 VSTRIMRIGHT; 1551 c = pgetc(); 1552 if (c == cc) 1553 subtype++; 1554 else 1555 pungetc(); 1556 break; 1557 } 1558 } 1559 } else if (subtype != VSERROR) { 1560 pungetc(); 1561 } 1562 STPUTC('=', out); 1563 if (subtype != VSLENGTH && (state[level].syntax == DQSYNTAX || 1564 state[level].syntax == ARISYNTAX)) 1565 flags |= VSQUOTE; 1566 *(stackblock() + typeloc) = subtype | flags; 1567 if (subtype != VSNORMAL) { 1568 if (level + 1 >= maxnest) { 1569 maxnest *= 2; 1570 if (state == state_static) { 1571 state = parser_temp_alloc( 1572 maxnest * sizeof(*state)); 1573 memcpy(state, state_static, 1574 MAXNEST_STATIC * sizeof(*state)); 1575 } else 1576 state = parser_temp_realloc(state, 1577 maxnest * sizeof(*state)); 1578 } 1579 level++; 1580 state[level].parenlevel = 0; 1581 if (subtype == VSMINUS || subtype == VSPLUS || 1582 subtype == VSQUESTION || subtype == VSASSIGN) { 1583 /* 1584 * For operators that were in the Bourne shell, 1585 * inherit the double-quote state. 1586 */ 1587 state[level].syntax = state[level - 1].syntax; 1588 state[level].category = TSTATE_VAR_OLD; 1589 } else { 1590 /* 1591 * The other operators take a pattern, 1592 * so go to BASESYNTAX. 1593 * Also, ' and " are now special, even 1594 * in here documents. 1595 */ 1596 state[level].syntax = BASESYNTAX; 1597 state[level].category = TSTATE_VAR_NEW; 1598 newvarnest++; 1599 } 1600 } 1601 } 1602 goto parsesub_return; 1603 } 1604 1605 1606 /* 1607 * Parse an arithmetic expansion (indicate start of one and set state) 1608 */ 1609 parsearith: { 1610 1611 if (level + 1 >= maxnest) { 1612 maxnest *= 2; 1613 if (state == state_static) { 1614 state = parser_temp_alloc( 1615 maxnest * sizeof(*state)); 1616 memcpy(state, state_static, 1617 MAXNEST_STATIC * sizeof(*state)); 1618 } else 1619 state = parser_temp_realloc(state, 1620 maxnest * sizeof(*state)); 1621 } 1622 level++; 1623 state[level].syntax = ARISYNTAX; 1624 state[level].parenlevel = 0; 1625 state[level].category = TSTATE_ARITH; 1626 USTPUTC(CTLARI, out); 1627 if (state[level - 1].syntax == DQSYNTAX) 1628 USTPUTC('"',out); 1629 else 1630 USTPUTC(' ',out); 1631 goto parsearith_return; 1632 } 1633 1634 } /* end of readtoken */ 1635 1636 1637 1638 #ifdef mkinit 1639 RESET { 1640 tokpushback = 0; 1641 checkkwd = 0; 1642 } 1643 #endif 1644 1645 /* 1646 * Returns true if the text contains nothing to expand (no dollar signs 1647 * or backquotes). 1648 */ 1649 1650 static int 1651 noexpand(char *text) 1652 { 1653 char *p; 1654 char c; 1655 1656 p = text; 1657 while ((c = *p++) != '\0') { 1658 if ( c == CTLQUOTEMARK) 1659 continue; 1660 if (c == CTLESC) 1661 p++; 1662 else if (BASESYNTAX[(int)c] == CCTL) 1663 return 0; 1664 } 1665 return 1; 1666 } 1667 1668 1669 /* 1670 * Return true if the argument is a legal variable name (a letter or 1671 * underscore followed by zero or more letters, underscores, and digits). 1672 */ 1673 1674 int 1675 goodname(const char *name) 1676 { 1677 const char *p; 1678 1679 p = name; 1680 if (! is_name(*p)) 1681 return 0; 1682 while (*++p) { 1683 if (! is_in_name(*p)) 1684 return 0; 1685 } 1686 return 1; 1687 } 1688 1689 1690 /* 1691 * Called when an unexpected token is read during the parse. The argument 1692 * is the token that is expected, or -1 if more than one type of token can 1693 * occur at this point. 1694 */ 1695 1696 static void 1697 synexpect(int token) 1698 { 1699 char msg[64]; 1700 1701 if (token >= 0) { 1702 fmtstr(msg, 64, "%s unexpected (expecting %s)", 1703 tokname[lasttoken], tokname[token]); 1704 } else { 1705 fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]); 1706 } 1707 synerror(msg); 1708 } 1709 1710 1711 static void 1712 synerror(const char *msg) 1713 { 1714 if (commandname) 1715 outfmt(out2, "%s: %d: ", commandname, startlinno); 1716 outfmt(out2, "Syntax error: %s\n", msg); 1717 error(NULL); 1718 } 1719 1720 static void 1721 setprompt(int which) 1722 { 1723 whichprompt = which; 1724 1725 #ifndef NO_HISTORY 1726 if (!el) 1727 #endif 1728 { 1729 out2str(getprompt(NULL)); 1730 flushout(out2); 1731 } 1732 } 1733 1734 /* 1735 * called by editline -- any expansions to the prompt 1736 * should be added here. 1737 */ 1738 const char * 1739 getprompt(void *unused __unused) 1740 { 1741 static char ps[PROMPTLEN]; 1742 const char *fmt; 1743 const char *pwd; 1744 int i, trim; 1745 1746 /* 1747 * Select prompt format. 1748 */ 1749 switch (whichprompt) { 1750 case 0: 1751 fmt = ""; 1752 break; 1753 case 1: 1754 fmt = ps1val(); 1755 break; 1756 case 2: 1757 fmt = ps2val(); 1758 break; 1759 default: 1760 return "??"; 1761 } 1762 1763 /* 1764 * Format prompt string. 1765 */ 1766 for (i = 0; (i < 127) && (*fmt != '\0'); i++, fmt++) 1767 if (*fmt == '\\') 1768 switch (*++fmt) { 1769 1770 /* 1771 * Hostname. 1772 * 1773 * \h specifies just the local hostname, 1774 * \H specifies fully-qualified hostname. 1775 */ 1776 case 'h': 1777 case 'H': 1778 ps[i] = '\0'; 1779 gethostname(&ps[i], PROMPTLEN - i); 1780 /* Skip to end of hostname. */ 1781 trim = (*fmt == 'h') ? '.' : '\0'; 1782 while ((ps[i+1] != '\0') && (ps[i+1] != trim)) 1783 i++; 1784 break; 1785 1786 /* 1787 * Working directory. 1788 * 1789 * \W specifies just the final component, 1790 * \w specifies the entire path. 1791 */ 1792 case 'W': 1793 case 'w': 1794 pwd = lookupvar("PWD"); 1795 if (pwd == NULL) 1796 pwd = "?"; 1797 if (*fmt == 'W' && 1798 *pwd == '/' && pwd[1] != '\0') 1799 strlcpy(&ps[i], strrchr(pwd, '/') + 1, 1800 PROMPTLEN - i); 1801 else 1802 strlcpy(&ps[i], pwd, PROMPTLEN - i); 1803 /* Skip to end of path. */ 1804 while (ps[i + 1] != '\0') 1805 i++; 1806 break; 1807 1808 /* 1809 * Superuser status. 1810 * 1811 * '$' for normal users, '#' for root. 1812 */ 1813 case '$': 1814 ps[i] = (geteuid() != 0) ? '$' : '#'; 1815 break; 1816 1817 /* 1818 * A literal \. 1819 */ 1820 case '\\': 1821 ps[i] = '\\'; 1822 break; 1823 1824 /* 1825 * Emit unrecognized formats verbatim. 1826 */ 1827 default: 1828 ps[i++] = '\\'; 1829 ps[i] = *fmt; 1830 break; 1831 } 1832 else 1833 ps[i] = *fmt; 1834 ps[i] = '\0'; 1835 return (ps); 1836 } 1837