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 1.29.2.9 2002/10/18 11:24:04 tjr Exp $ 38 * $DragonFly: src/bin/sh/parser.c,v 1.3 2003/08/24 16:26:00 drhodus Exp $ 39 */ 40 41 #include <stdlib.h> 42 43 #include "shell.h" 44 #include "parser.h" 45 #include "nodes.h" 46 #include "expand.h" /* defines rmescapes() */ 47 #include "redir.h" /* defines copyfd() */ 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 #ifndef NO_HISTORY 60 #include "myhistedit.h" 61 #endif 62 63 /* 64 * Shell command parser. 65 */ 66 67 #define EOFMARKLEN 79 68 69 /* values returned by readtoken */ 70 #include "token.h" 71 72 73 74 struct heredoc { 75 struct heredoc *next; /* next here document in list */ 76 union node *here; /* redirection node */ 77 char *eofmark; /* string indicating end of input */ 78 int striptabs; /* if set, strip leading tabs */ 79 }; 80 81 82 83 STATIC struct heredoc *heredoclist; /* list of here documents to read */ 84 STATIC int parsebackquote; /* nonzero if we are inside backquotes */ 85 STATIC int doprompt; /* if set, prompt the user */ 86 STATIC int needprompt; /* true if interactive and at start of line */ 87 STATIC int lasttoken; /* last token read */ 88 MKINIT int tokpushback; /* last token pushed back */ 89 STATIC char *wordtext; /* text of last word returned by readtoken */ 90 MKINIT int checkkwd; /* 1 == check for kwds, 2 == also eat newlines */ 91 STATIC struct nodelist *backquotelist; 92 STATIC union node *redirnode; 93 STATIC struct heredoc *heredoc; 94 STATIC int quoteflag; /* set if (part of) last token was quoted */ 95 STATIC int startlinno; /* line # where last token started */ 96 97 /* XXX When 'noaliases' is set to one, no alias expansion takes place. */ 98 static int noaliases = 0; 99 100 #define GDB_HACK 1 /* avoid local declarations which gdb can't handle */ 101 #ifdef GDB_HACK 102 static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0'}; 103 static const char types[] = "}-+?="; 104 #endif 105 106 107 STATIC union node *list(int); 108 STATIC union node *andor(void); 109 STATIC union node *pipeline(void); 110 STATIC union node *command(void); 111 STATIC union node *simplecmd(union node **, union node *); 112 STATIC union node *makename(void); 113 STATIC void parsefname(void); 114 STATIC void parseheredoc(void); 115 STATIC int peektoken(void); 116 STATIC int readtoken(void); 117 STATIC int xxreadtoken(void); 118 STATIC int readtoken1(int, char const *, char *, int); 119 STATIC int noexpand(char *); 120 STATIC void synexpect(int); 121 STATIC void synerror(char *); 122 STATIC void setprompt(int); 123 124 125 /* 126 * Read and parse a command. Returns NEOF on end of file. (NULL is a 127 * valid parse tree indicating a blank line.) 128 */ 129 130 union node * 131 parsecmd(int interact) 132 { 133 int t; 134 135 tokpushback = 0; 136 doprompt = interact; 137 if (doprompt) 138 setprompt(1); 139 else 140 setprompt(0); 141 needprompt = 0; 142 t = readtoken(); 143 if (t == TEOF) 144 return NEOF; 145 if (t == TNL) 146 return NULL; 147 tokpushback++; 148 return list(1); 149 } 150 151 152 STATIC union node * 153 list(int nlflag) 154 { 155 union node *n1, *n2, *n3; 156 int tok; 157 158 checkkwd = 2; 159 if (nlflag == 0 && tokendlist[peektoken()]) 160 return NULL; 161 n1 = NULL; 162 for (;;) { 163 n2 = andor(); 164 tok = readtoken(); 165 if (tok == TBACKGND) { 166 if (n2->type == NCMD || n2->type == NPIPE) { 167 n2->ncmd.backgnd = 1; 168 } else if (n2->type == NREDIR) { 169 n2->type = NBACKGND; 170 } else { 171 n3 = (union node *)stalloc(sizeof (struct nredir)); 172 n3->type = NBACKGND; 173 n3->nredir.n = n2; 174 n3->nredir.redirect = NULL; 175 n2 = n3; 176 } 177 } 178 if (n1 == NULL) { 179 n1 = n2; 180 } 181 else { 182 n3 = (union node *)stalloc(sizeof (struct nbinary)); 183 n3->type = NSEMI; 184 n3->nbinary.ch1 = n1; 185 n3->nbinary.ch2 = n2; 186 n1 = n3; 187 } 188 switch (tok) { 189 case TBACKGND: 190 case TSEMI: 191 tok = readtoken(); 192 /* fall through */ 193 case TNL: 194 if (tok == TNL) { 195 parseheredoc(); 196 if (nlflag) 197 return n1; 198 } else { 199 tokpushback++; 200 } 201 checkkwd = 2; 202 if (tokendlist[peektoken()]) 203 return n1; 204 break; 205 case TEOF: 206 if (heredoclist) 207 parseheredoc(); 208 else 209 pungetc(); /* push back EOF on input */ 210 return n1; 211 default: 212 if (nlflag) 213 synexpect(-1); 214 tokpushback++; 215 return n1; 216 } 217 } 218 } 219 220 221 222 STATIC union node * 223 andor(void) 224 { 225 union node *n1, *n2, *n3; 226 int t; 227 228 n1 = pipeline(); 229 for (;;) { 230 if ((t = readtoken()) == TAND) { 231 t = NAND; 232 } else if (t == TOR) { 233 t = NOR; 234 } else { 235 tokpushback++; 236 return n1; 237 } 238 n2 = pipeline(); 239 n3 = (union node *)stalloc(sizeof (struct nbinary)); 240 n3->type = t; 241 n3->nbinary.ch1 = n1; 242 n3->nbinary.ch2 = n2; 243 n1 = n3; 244 } 245 } 246 247 248 249 STATIC union node * 250 pipeline(void) 251 { 252 union node *n1, *n2, *pipenode; 253 struct nodelist *lp, *prev; 254 int negate; 255 256 negate = 0; 257 TRACE(("pipeline: entered\n")); 258 while (readtoken() == TNOT) 259 negate = !negate; 260 tokpushback++; 261 n1 = command(); 262 if (readtoken() == TPIPE) { 263 pipenode = (union node *)stalloc(sizeof (struct npipe)); 264 pipenode->type = NPIPE; 265 pipenode->npipe.backgnd = 0; 266 lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 267 pipenode->npipe.cmdlist = lp; 268 lp->n = n1; 269 do { 270 prev = lp; 271 lp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 272 lp->n = command(); 273 prev->next = lp; 274 } while (readtoken() == TPIPE); 275 lp->next = NULL; 276 n1 = pipenode; 277 } 278 tokpushback++; 279 if (negate) { 280 n2 = (union node *)stalloc(sizeof (struct nnot)); 281 n2->type = NNOT; 282 n2->nnot.com = n1; 283 return n2; 284 } else 285 return n1; 286 } 287 288 289 290 STATIC union node * 291 command(void) 292 { 293 union node *n1, *n2; 294 union node *ap, **app; 295 union node *cp, **cpp; 296 union node *redir, **rpp; 297 int t, negate = 0; 298 299 checkkwd = 2; 300 redir = NULL; 301 n1 = NULL; 302 rpp = &redir; 303 304 /* Check for redirection which may precede command */ 305 while (readtoken() == TREDIR) { 306 *rpp = n2 = redirnode; 307 rpp = &n2->nfile.next; 308 parsefname(); 309 } 310 tokpushback++; 311 312 while (readtoken() == TNOT) { 313 TRACE(("command: TNOT recognized\n")); 314 negate = !negate; 315 } 316 tokpushback++; 317 318 switch (readtoken()) { 319 case TIF: 320 n1 = (union node *)stalloc(sizeof (struct nif)); 321 n1->type = NIF; 322 n1->nif.test = list(0); 323 if (readtoken() != TTHEN) 324 synexpect(TTHEN); 325 n1->nif.ifpart = list(0); 326 n2 = n1; 327 while (readtoken() == TELIF) { 328 n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif)); 329 n2 = n2->nif.elsepart; 330 n2->type = NIF; 331 n2->nif.test = list(0); 332 if (readtoken() != TTHEN) 333 synexpect(TTHEN); 334 n2->nif.ifpart = list(0); 335 } 336 if (lasttoken == TELSE) 337 n2->nif.elsepart = list(0); 338 else { 339 n2->nif.elsepart = NULL; 340 tokpushback++; 341 } 342 if (readtoken() != TFI) 343 synexpect(TFI); 344 checkkwd = 1; 345 break; 346 case TWHILE: 347 case TUNTIL: { 348 int got; 349 n1 = (union node *)stalloc(sizeof (struct nbinary)); 350 n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL; 351 n1->nbinary.ch1 = list(0); 352 if ((got=readtoken()) != TDO) { 353 TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : "")); 354 synexpect(TDO); 355 } 356 n1->nbinary.ch2 = list(0); 357 if (readtoken() != TDONE) 358 synexpect(TDONE); 359 checkkwd = 1; 360 break; 361 } 362 case TFOR: 363 if (readtoken() != TWORD || quoteflag || ! goodname(wordtext)) 364 synerror("Bad for loop variable"); 365 n1 = (union node *)stalloc(sizeof (struct nfor)); 366 n1->type = NFOR; 367 n1->nfor.var = wordtext; 368 if (readtoken() == TWORD && ! quoteflag && equal(wordtext, "in")) { 369 app = ≈ 370 while (readtoken() == TWORD) { 371 n2 = (union node *)stalloc(sizeof (struct narg)); 372 n2->type = NARG; 373 n2->narg.text = wordtext; 374 n2->narg.backquote = backquotelist; 375 *app = n2; 376 app = &n2->narg.next; 377 } 378 *app = NULL; 379 n1->nfor.args = ap; 380 if (lasttoken != TNL && lasttoken != TSEMI) 381 synexpect(-1); 382 } else { 383 #ifndef GDB_HACK 384 static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE, 385 '@', '=', '\0'}; 386 #endif 387 n2 = (union node *)stalloc(sizeof (struct narg)); 388 n2->type = NARG; 389 n2->narg.text = (char *)argvars; 390 n2->narg.backquote = NULL; 391 n2->narg.next = NULL; 392 n1->nfor.args = n2; 393 /* 394 * Newline or semicolon here is optional (but note 395 * that the original Bourne shell only allowed NL). 396 */ 397 if (lasttoken != TNL && lasttoken != TSEMI) 398 tokpushback++; 399 } 400 checkkwd = 2; 401 if ((t = readtoken()) == TDO) 402 t = TDONE; 403 else if (t == TBEGIN) 404 t = TEND; 405 else 406 synexpect(-1); 407 n1->nfor.body = list(0); 408 if (readtoken() != t) 409 synexpect(t); 410 checkkwd = 1; 411 break; 412 case TCASE: 413 n1 = (union node *)stalloc(sizeof (struct ncase)); 414 n1->type = NCASE; 415 if (readtoken() != TWORD) 416 synexpect(TWORD); 417 n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg)); 418 n2->type = NARG; 419 n2->narg.text = wordtext; 420 n2->narg.backquote = backquotelist; 421 n2->narg.next = NULL; 422 while (readtoken() == TNL); 423 if (lasttoken != TWORD || ! equal(wordtext, "in")) 424 synerror("expecting \"in\""); 425 cpp = &n1->ncase.cases; 426 noaliases = 1; /* turn off alias expansion */ 427 checkkwd = 2, readtoken(); 428 while (lasttoken != TESAC) { 429 *cpp = cp = (union node *)stalloc(sizeof (struct nclist)); 430 cp->type = NCLIST; 431 app = &cp->nclist.pattern; 432 if (lasttoken == TLP) 433 readtoken(); 434 for (;;) { 435 *app = ap = (union node *)stalloc(sizeof (struct narg)); 436 ap->type = NARG; 437 ap->narg.text = wordtext; 438 ap->narg.backquote = backquotelist; 439 if (checkkwd = 2, readtoken() != TPIPE) 440 break; 441 app = &ap->narg.next; 442 readtoken(); 443 } 444 ap->narg.next = NULL; 445 if (lasttoken != TRP) 446 noaliases = 0, synexpect(TRP); 447 cp->nclist.body = list(0); 448 449 checkkwd = 2; 450 if ((t = readtoken()) != TESAC) { 451 if (t != TENDCASE) 452 noaliases = 0, synexpect(TENDCASE); 453 else 454 checkkwd = 2, readtoken(); 455 } 456 cpp = &cp->nclist.next; 457 } 458 noaliases = 0; /* reset alias expansion */ 459 *cpp = NULL; 460 checkkwd = 1; 461 break; 462 case TLP: 463 n1 = (union node *)stalloc(sizeof (struct nredir)); 464 n1->type = NSUBSHELL; 465 n1->nredir.n = list(0); 466 n1->nredir.redirect = NULL; 467 if (readtoken() != TRP) 468 synexpect(TRP); 469 checkkwd = 1; 470 break; 471 case TBEGIN: 472 n1 = list(0); 473 if (readtoken() != TEND) 474 synexpect(TEND); 475 checkkwd = 1; 476 break; 477 /* Handle an empty command like other simple commands. */ 478 case TSEMI: 479 case TAND: 480 case TOR: 481 /* 482 * An empty command before a ; doesn't make much sense, and 483 * should certainly be disallowed in the case of `if ;'. 484 */ 485 if (!redir) 486 synexpect(-1); 487 case TNL: 488 case TEOF: 489 case TWORD: 490 case TRP: 491 tokpushback++; 492 n1 = simplecmd(rpp, redir); 493 goto checkneg; 494 default: 495 synexpect(-1); 496 } 497 498 /* Now check for redirection which may follow command */ 499 while (readtoken() == TREDIR) { 500 *rpp = n2 = redirnode; 501 rpp = &n2->nfile.next; 502 parsefname(); 503 } 504 tokpushback++; 505 *rpp = NULL; 506 if (redir) { 507 if (n1->type != NSUBSHELL) { 508 n2 = (union node *)stalloc(sizeof (struct nredir)); 509 n2->type = NREDIR; 510 n2->nredir.n = n1; 511 n1 = n2; 512 } 513 n1->nredir.redirect = redir; 514 } 515 516 checkneg: 517 if (negate) { 518 n2 = (union node *)stalloc(sizeof (struct nnot)); 519 n2->type = NNOT; 520 n2->nnot.com = n1; 521 return n2; 522 } 523 else 524 return n1; 525 } 526 527 528 STATIC union node * 529 simplecmd(union node **rpp, union node *redir) 530 { 531 union node *args, **app; 532 union node **orig_rpp = rpp; 533 union node *n = NULL, *n2; 534 int negate = 0; 535 536 /* If we don't have any redirections already, then we must reset */ 537 /* rpp to be the address of the local redir variable. */ 538 if (redir == 0) 539 rpp = &redir; 540 541 args = NULL; 542 app = &args; 543 /* 544 * We save the incoming value, because we need this for shell 545 * functions. There can not be a redirect or an argument between 546 * the function name and the open parenthesis. 547 */ 548 orig_rpp = rpp; 549 550 while (readtoken() == TNOT) { 551 TRACE(("command: TNOT recognized\n")); 552 negate = !negate; 553 } 554 tokpushback++; 555 556 for (;;) { 557 if (readtoken() == TWORD) { 558 n = (union node *)stalloc(sizeof (struct narg)); 559 n->type = NARG; 560 n->narg.text = wordtext; 561 n->narg.backquote = backquotelist; 562 *app = n; 563 app = &n->narg.next; 564 } else if (lasttoken == TREDIR) { 565 *rpp = n = redirnode; 566 rpp = &n->nfile.next; 567 parsefname(); /* read name of redirection file */ 568 } else if (lasttoken == TLP && app == &args->narg.next 569 && rpp == orig_rpp) { 570 /* We have a function */ 571 if (readtoken() != TRP) 572 synexpect(TRP); 573 #ifdef notdef 574 if (! goodname(n->narg.text)) 575 synerror("Bad function name"); 576 #endif 577 n->type = NDEFUN; 578 n->narg.next = command(); 579 goto checkneg; 580 } else { 581 tokpushback++; 582 break; 583 } 584 } 585 *app = NULL; 586 *rpp = NULL; 587 n = (union node *)stalloc(sizeof (struct ncmd)); 588 n->type = NCMD; 589 n->ncmd.backgnd = 0; 590 n->ncmd.args = args; 591 n->ncmd.redirect = redir; 592 593 checkneg: 594 if (negate) { 595 n2 = (union node *)stalloc(sizeof (struct nnot)); 596 n2->type = NNOT; 597 n2->nnot.com = n; 598 return n2; 599 } 600 else 601 return n; 602 } 603 604 STATIC union node * 605 makename(void) 606 { 607 union node *n; 608 609 n = (union node *)stalloc(sizeof (struct narg)); 610 n->type = NARG; 611 n->narg.next = NULL; 612 n->narg.text = wordtext; 613 n->narg.backquote = backquotelist; 614 return n; 615 } 616 617 void fixredir(union node *n, const char *text, int err) 618 { 619 TRACE(("Fix redir %s %d\n", text, err)); 620 if (!err) 621 n->ndup.vname = NULL; 622 623 if (is_digit(text[0]) && text[1] == '\0') 624 n->ndup.dupfd = digit_val(text[0]); 625 else if (text[0] == '-' && text[1] == '\0') 626 n->ndup.dupfd = -1; 627 else { 628 629 if (err) 630 synerror("Bad fd number"); 631 else 632 n->ndup.vname = makename(); 633 } 634 } 635 636 637 STATIC void 638 parsefname(void) 639 { 640 union node *n = redirnode; 641 642 if (readtoken() != TWORD) 643 synexpect(-1); 644 if (n->type == NHERE) { 645 struct heredoc *here = heredoc; 646 struct heredoc *p; 647 int i; 648 649 if (quoteflag == 0) 650 n->type = NXHERE; 651 TRACE(("Here document %d\n", n->type)); 652 if (here->striptabs) { 653 while (*wordtext == '\t') 654 wordtext++; 655 } 656 if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN) 657 synerror("Illegal eof marker for << redirection"); 658 rmescapes(wordtext); 659 here->eofmark = wordtext; 660 here->next = NULL; 661 if (heredoclist == NULL) 662 heredoclist = here; 663 else { 664 for (p = heredoclist ; p->next ; p = p->next); 665 p->next = here; 666 } 667 } else if (n->type == NTOFD || n->type == NFROMFD) { 668 fixredir(n, wordtext, 0); 669 } else { 670 n->nfile.fname = makename(); 671 } 672 } 673 674 675 /* 676 * Input any here documents. 677 */ 678 679 STATIC void 680 parseheredoc(void) 681 { 682 struct heredoc *here; 683 union node *n; 684 685 while (heredoclist) { 686 here = heredoclist; 687 heredoclist = here->next; 688 if (needprompt) { 689 setprompt(2); 690 needprompt = 0; 691 } 692 readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX, 693 here->eofmark, here->striptabs); 694 n = (union node *)stalloc(sizeof (struct narg)); 695 n->narg.type = NARG; 696 n->narg.next = NULL; 697 n->narg.text = wordtext; 698 n->narg.backquote = backquotelist; 699 here->here->nhere.doc = n; 700 } 701 } 702 703 STATIC int 704 peektoken(void) 705 { 706 int t; 707 708 t = readtoken(); 709 tokpushback++; 710 return (t); 711 } 712 713 STATIC int 714 readtoken(void) 715 { 716 int t; 717 int savecheckkwd = checkkwd; 718 struct alias *ap; 719 #ifdef DEBUG 720 int alreadyseen = tokpushback; 721 #endif 722 723 top: 724 t = xxreadtoken(); 725 726 if (checkkwd) { 727 /* 728 * eat newlines 729 */ 730 if (checkkwd == 2) { 731 checkkwd = 0; 732 while (t == TNL) { 733 parseheredoc(); 734 t = xxreadtoken(); 735 } 736 } else 737 checkkwd = 0; 738 /* 739 * check for keywords and aliases 740 */ 741 if (t == TWORD && !quoteflag) 742 { 743 const char * const *pp; 744 745 for (pp = parsekwd; *pp; pp++) { 746 if (**pp == *wordtext && equal(*pp, wordtext)) 747 { 748 lasttoken = t = pp - parsekwd + KWDOFFSET; 749 TRACE(("keyword %s recognized\n", tokname[t])); 750 goto out; 751 } 752 } 753 if (noaliases == 0 && 754 (ap = lookupalias(wordtext, 1)) != NULL) { 755 pushstring(ap->val, strlen(ap->val), ap); 756 checkkwd = savecheckkwd; 757 goto top; 758 } 759 } 760 out: 761 checkkwd = (t == TNOT) ? savecheckkwd : 0; 762 } 763 #ifdef DEBUG 764 if (!alreadyseen) 765 TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); 766 else 767 TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : "")); 768 #endif 769 return (t); 770 } 771 772 773 /* 774 * Read the next input token. 775 * If the token is a word, we set backquotelist to the list of cmds in 776 * backquotes. We set quoteflag to true if any part of the word was 777 * quoted. 778 * If the token is TREDIR, then we set redirnode to a structure containing 779 * the redirection. 780 * In all cases, the variable startlinno is set to the number of the line 781 * on which the token starts. 782 * 783 * [Change comment: here documents and internal procedures] 784 * [Readtoken shouldn't have any arguments. Perhaps we should make the 785 * word parsing code into a separate routine. In this case, readtoken 786 * doesn't need to have any internal procedures, but parseword does. 787 * We could also make parseoperator in essence the main routine, and 788 * have parseword (readtoken1?) handle both words and redirection.] 789 */ 790 791 #define RETURN(token) return lasttoken = token 792 793 STATIC int 794 xxreadtoken(void) 795 { 796 int c; 797 798 if (tokpushback) { 799 tokpushback = 0; 800 return lasttoken; 801 } 802 if (needprompt) { 803 setprompt(2); 804 needprompt = 0; 805 } 806 startlinno = plinno; 807 for (;;) { /* until token or start of word found */ 808 c = pgetc_macro(); 809 if (c == ' ' || c == '\t') 810 continue; /* quick check for white space first */ 811 switch (c) { 812 case ' ': case '\t': 813 continue; 814 case '#': 815 while ((c = pgetc()) != '\n' && c != PEOF); 816 pungetc(); 817 continue; 818 case '\\': 819 if (pgetc() == '\n') { 820 startlinno = ++plinno; 821 if (doprompt) 822 setprompt(2); 823 else 824 setprompt(0); 825 continue; 826 } 827 pungetc(); 828 goto breakloop; 829 case '\n': 830 plinno++; 831 needprompt = doprompt; 832 RETURN(TNL); 833 case PEOF: 834 RETURN(TEOF); 835 case '&': 836 if (pgetc() == '&') 837 RETURN(TAND); 838 pungetc(); 839 RETURN(TBACKGND); 840 case '|': 841 if (pgetc() == '|') 842 RETURN(TOR); 843 pungetc(); 844 RETURN(TPIPE); 845 case ';': 846 if (pgetc() == ';') 847 RETURN(TENDCASE); 848 pungetc(); 849 RETURN(TSEMI); 850 case '(': 851 RETURN(TLP); 852 case ')': 853 RETURN(TRP); 854 default: 855 goto breakloop; 856 } 857 } 858 breakloop: 859 return readtoken1(c, BASESYNTAX, (char *)NULL, 0); 860 #undef RETURN 861 } 862 863 864 865 /* 866 * If eofmark is NULL, read a word or a redirection symbol. If eofmark 867 * is not NULL, read a here document. In the latter case, eofmark is the 868 * word which marks the end of the document and striptabs is true if 869 * leading tabs should be stripped from the document. The argument firstc 870 * is the first character of the input token or document. 871 * 872 * Because C does not have internal subroutines, I have simulated them 873 * using goto's to implement the subroutine linkage. The following macros 874 * will run code that appears at the end of readtoken1. 875 */ 876 877 #define CHECKEND() {goto checkend; checkend_return:;} 878 #define PARSEREDIR() {goto parseredir; parseredir_return:;} 879 #define PARSESUB() {goto parsesub; parsesub_return:;} 880 #define PARSEBACKQOLD() {oldstyle = 1; goto parsebackq; parsebackq_oldreturn:;} 881 #define PARSEBACKQNEW() {oldstyle = 0; goto parsebackq; parsebackq_newreturn:;} 882 #define PARSEARITH() {goto parsearith; parsearith_return:;} 883 884 STATIC int 885 readtoken1(int firstc, char const *syntax, char *eofmark, int striptabs) 886 { 887 int c = firstc; 888 char *out; 889 int len; 890 char line[EOFMARKLEN + 1]; 891 struct nodelist *bqlist; 892 int quotef; 893 int dblquote; 894 int varnest; /* levels of variables expansion */ 895 int arinest; /* levels of arithmetic expansion */ 896 int parenlevel; /* levels of parens in arithmetic */ 897 int oldstyle; 898 char const *prevsyntax; /* syntax before arithmetic */ 899 int synentry; 900 #if __GNUC__ 901 /* Avoid longjmp clobbering */ 902 (void) &out; 903 (void) "ef; 904 (void) &dblquote; 905 (void) &varnest; 906 (void) &arinest; 907 (void) &parenlevel; 908 (void) &oldstyle; 909 (void) &prevsyntax; 910 (void) &syntax; 911 (void) &synentry; 912 #endif 913 914 startlinno = plinno; 915 dblquote = 0; 916 if (syntax == DQSYNTAX) 917 dblquote = 1; 918 quotef = 0; 919 bqlist = NULL; 920 varnest = 0; 921 arinest = 0; 922 parenlevel = 0; 923 924 STARTSTACKSTR(out); 925 loop: { /* for each line, until end of word */ 926 #if ATTY 927 if (c == '\034' && doprompt 928 && attyset() && ! equal(termval(), "emacs")) { 929 attyline(); 930 if (syntax == BASESYNTAX) 931 return readtoken(); 932 c = pgetc(); 933 goto loop; 934 } 935 #endif 936 CHECKEND(); /* set c to PEOF if at end of here document */ 937 for (;;) { /* until end of line or end of word */ 938 CHECKSTRSPACE(3, out); /* permit 3 calls to USTPUTC */ 939 940 synentry = syntax[c]; 941 942 switch(synentry) { 943 case CNL: /* '\n' */ 944 if (syntax == BASESYNTAX) 945 goto endword; /* exit outer loop */ 946 USTPUTC(c, out); 947 plinno++; 948 if (doprompt) 949 setprompt(2); 950 else 951 setprompt(0); 952 c = pgetc(); 953 goto loop; /* continue outer loop */ 954 case CWORD: 955 USTPUTC(c, out); 956 break; 957 case CCTL: 958 if (eofmark == NULL || dblquote) 959 USTPUTC(CTLESC, out); 960 USTPUTC(c, out); 961 break; 962 case CBACK: /* backslash */ 963 c = pgetc(); 964 if (c == PEOF) { 965 USTPUTC('\\', out); 966 pungetc(); 967 } else if (c == '\n') { 968 if (doprompt) 969 setprompt(2); 970 else 971 setprompt(0); 972 } else { 973 if (dblquote && c != '\\' && 974 c != '`' && c != '$' && 975 (c != '"' || eofmark != NULL)) 976 USTPUTC('\\', out); 977 if (SQSYNTAX[c] == CCTL) 978 USTPUTC(CTLESC, out); 979 else if (eofmark == NULL) 980 USTPUTC(CTLQUOTEMARK, out); 981 USTPUTC(c, out); 982 quotef++; 983 } 984 break; 985 case CSQUOTE: 986 if (eofmark == NULL) 987 USTPUTC(CTLQUOTEMARK, out); 988 syntax = SQSYNTAX; 989 break; 990 case CDQUOTE: 991 if (eofmark == NULL) 992 USTPUTC(CTLQUOTEMARK, out); 993 syntax = DQSYNTAX; 994 dblquote = 1; 995 break; 996 case CENDQUOTE: 997 if (eofmark != NULL && arinest == 0 && 998 varnest == 0) { 999 USTPUTC(c, out); 1000 } else { 1001 if (arinest) { 1002 syntax = ARISYNTAX; 1003 dblquote = 0; 1004 } else if (eofmark == NULL) { 1005 syntax = BASESYNTAX; 1006 dblquote = 0; 1007 } 1008 quotef++; 1009 } 1010 break; 1011 case CVAR: /* '$' */ 1012 PARSESUB(); /* parse substitution */ 1013 break; 1014 case CENDVAR: /* '}' */ 1015 if (varnest > 0) { 1016 varnest--; 1017 USTPUTC(CTLENDVAR, out); 1018 } else { 1019 USTPUTC(c, out); 1020 } 1021 break; 1022 case CLP: /* '(' in arithmetic */ 1023 parenlevel++; 1024 USTPUTC(c, out); 1025 break; 1026 case CRP: /* ')' in arithmetic */ 1027 if (parenlevel > 0) { 1028 USTPUTC(c, out); 1029 --parenlevel; 1030 } else { 1031 if (pgetc() == ')') { 1032 if (--arinest == 0) { 1033 USTPUTC(CTLENDARI, out); 1034 syntax = prevsyntax; 1035 if (syntax == DQSYNTAX) 1036 dblquote = 1; 1037 else 1038 dblquote = 0; 1039 } else 1040 USTPUTC(')', out); 1041 } else { 1042 /* 1043 * unbalanced parens 1044 * (don't 2nd guess - no error) 1045 */ 1046 pungetc(); 1047 USTPUTC(')', out); 1048 } 1049 } 1050 break; 1051 case CBQUOTE: /* '`' */ 1052 PARSEBACKQOLD(); 1053 break; 1054 case CEOF: 1055 goto endword; /* exit outer loop */ 1056 default: 1057 if (varnest == 0) 1058 goto endword; /* exit outer loop */ 1059 USTPUTC(c, out); 1060 } 1061 c = pgetc_macro(); 1062 } 1063 } 1064 endword: 1065 if (syntax == ARISYNTAX) 1066 synerror("Missing '))'"); 1067 if (syntax != BASESYNTAX && ! parsebackquote && eofmark == NULL) 1068 synerror("Unterminated quoted string"); 1069 if (varnest != 0) { 1070 startlinno = plinno; 1071 synerror("Missing '}'"); 1072 } 1073 USTPUTC('\0', out); 1074 len = out - stackblock(); 1075 out = stackblock(); 1076 if (eofmark == NULL) { 1077 if ((c == '>' || c == '<') 1078 && quotef == 0 1079 && len <= 2 1080 && (*out == '\0' || is_digit(*out))) { 1081 PARSEREDIR(); 1082 return lasttoken = TREDIR; 1083 } else { 1084 pungetc(); 1085 } 1086 } 1087 quoteflag = quotef; 1088 backquotelist = bqlist; 1089 grabstackblock(len); 1090 wordtext = out; 1091 return lasttoken = TWORD; 1092 /* end of readtoken routine */ 1093 1094 1095 1096 /* 1097 * Check to see whether we are at the end of the here document. When this 1098 * is called, c is set to the first character of the next input line. If 1099 * we are at the end of the here document, this routine sets the c to PEOF. 1100 */ 1101 1102 checkend: { 1103 if (eofmark) { 1104 if (striptabs) { 1105 while (c == '\t') 1106 c = pgetc(); 1107 } 1108 if (c == *eofmark) { 1109 if (pfgets(line, sizeof line) != NULL) { 1110 char *p, *q; 1111 1112 p = line; 1113 for (q = eofmark + 1 ; *q && *p == *q ; p++, q++); 1114 if (*p == '\n' && *q == '\0') { 1115 c = PEOF; 1116 plinno++; 1117 needprompt = doprompt; 1118 } else { 1119 pushstring(line, strlen(line), NULL); 1120 } 1121 } 1122 } 1123 } 1124 goto checkend_return; 1125 } 1126 1127 1128 /* 1129 * Parse a redirection operator. The variable "out" points to a string 1130 * specifying the fd to be redirected. The variable "c" contains the 1131 * first character of the redirection operator. 1132 */ 1133 1134 parseredir: { 1135 char fd = *out; 1136 union node *np; 1137 1138 np = (union node *)stalloc(sizeof (struct nfile)); 1139 if (c == '>') { 1140 np->nfile.fd = 1; 1141 c = pgetc(); 1142 if (c == '>') 1143 np->type = NAPPEND; 1144 else if (c == '&') 1145 np->type = NTOFD; 1146 else if (c == '|') 1147 np->type = NCLOBBER; 1148 else { 1149 np->type = NTO; 1150 pungetc(); 1151 } 1152 } else { /* c == '<' */ 1153 np->nfile.fd = 0; 1154 c = pgetc(); 1155 if (c == '<') { 1156 if (sizeof (struct nfile) != sizeof (struct nhere)) { 1157 np = (union node *)stalloc(sizeof (struct nhere)); 1158 np->nfile.fd = 0; 1159 } 1160 np->type = NHERE; 1161 heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc)); 1162 heredoc->here = np; 1163 if ((c = pgetc()) == '-') { 1164 heredoc->striptabs = 1; 1165 } else { 1166 heredoc->striptabs = 0; 1167 pungetc(); 1168 } 1169 } else if (c == '&') 1170 np->type = NFROMFD; 1171 else if (c == '>') 1172 np->type = NFROMTO; 1173 else { 1174 np->type = NFROM; 1175 pungetc(); 1176 } 1177 } 1178 if (fd != '\0') 1179 np->nfile.fd = digit_val(fd); 1180 redirnode = np; 1181 goto parseredir_return; 1182 } 1183 1184 1185 /* 1186 * Parse a substitution. At this point, we have read the dollar sign 1187 * and nothing else. 1188 */ 1189 1190 parsesub: { 1191 int subtype; 1192 int typeloc; 1193 int flags; 1194 char *p; 1195 #ifndef GDB_HACK 1196 static const char types[] = "}-+?="; 1197 #endif 1198 int bracketed_name = 0; /* used to handle ${[0-9]*} variables */ 1199 1200 c = pgetc(); 1201 if (c != '(' && c != '{' && !is_name(c) && !is_special(c)) { 1202 USTPUTC('$', out); 1203 pungetc(); 1204 } else if (c == '(') { /* $(command) or $((arith)) */ 1205 if (pgetc() == '(') { 1206 PARSEARITH(); 1207 } else { 1208 pungetc(); 1209 PARSEBACKQNEW(); 1210 } 1211 } else { 1212 USTPUTC(CTLVAR, out); 1213 typeloc = out - stackblock(); 1214 USTPUTC(VSNORMAL, out); 1215 subtype = VSNORMAL; 1216 if (c == '{') { 1217 bracketed_name = 1; 1218 c = pgetc(); 1219 if (c == '#') { 1220 if ((c = pgetc()) == '}') 1221 c = '#'; 1222 else 1223 subtype = VSLENGTH; 1224 } 1225 else 1226 subtype = 0; 1227 } 1228 if (is_name(c)) { 1229 do { 1230 STPUTC(c, out); 1231 c = pgetc(); 1232 } while (is_in_name(c)); 1233 } else if (is_digit(c)) { 1234 if (bracketed_name) { 1235 do { 1236 STPUTC(c, out); 1237 c = pgetc(); 1238 } while (is_digit(c)); 1239 } else { 1240 STPUTC(c, out); 1241 c = pgetc(); 1242 } 1243 } else { 1244 if (! is_special(c)) 1245 badsub: synerror("Bad substitution"); 1246 USTPUTC(c, out); 1247 c = pgetc(); 1248 } 1249 STPUTC('=', out); 1250 flags = 0; 1251 if (subtype == 0) { 1252 switch (c) { 1253 case ':': 1254 flags = VSNUL; 1255 c = pgetc(); 1256 /*FALLTHROUGH*/ 1257 default: 1258 p = strchr(types, c); 1259 if (p == NULL) 1260 goto badsub; 1261 subtype = p - types + VSNORMAL; 1262 break; 1263 case '%': 1264 case '#': 1265 { 1266 int cc = c; 1267 subtype = c == '#' ? VSTRIMLEFT : 1268 VSTRIMRIGHT; 1269 c = pgetc(); 1270 if (c == cc) 1271 subtype++; 1272 else 1273 pungetc(); 1274 break; 1275 } 1276 } 1277 } else { 1278 pungetc(); 1279 } 1280 if (subtype != VSLENGTH && (dblquote || arinest)) 1281 flags |= VSQUOTE; 1282 *(stackblock() + typeloc) = subtype | flags; 1283 if (subtype != VSNORMAL) 1284 varnest++; 1285 } 1286 goto parsesub_return; 1287 } 1288 1289 1290 /* 1291 * Called to parse command substitutions. Newstyle is set if the command 1292 * is enclosed inside $(...); nlpp is a pointer to the head of the linked 1293 * list of commands (passed by reference), and savelen is the number of 1294 * characters on the top of the stack which must be preserved. 1295 */ 1296 1297 parsebackq: { 1298 struct nodelist **nlpp; 1299 int savepbq; 1300 union node *n; 1301 char *volatile str; 1302 struct jmploc jmploc; 1303 struct jmploc *volatile savehandler; 1304 int savelen; 1305 int saveprompt; 1306 #if __GNUC__ 1307 /* Avoid longjmp clobbering */ 1308 (void) &saveprompt; 1309 #endif 1310 1311 savepbq = parsebackquote; 1312 if (setjmp(jmploc.loc)) { 1313 if (str) 1314 ckfree(str); 1315 parsebackquote = 0; 1316 handler = savehandler; 1317 longjmp(handler->loc, 1); 1318 } 1319 INTOFF; 1320 str = NULL; 1321 savelen = out - stackblock(); 1322 if (savelen > 0) { 1323 str = ckmalloc(savelen); 1324 memcpy(str, stackblock(), savelen); 1325 } 1326 savehandler = handler; 1327 handler = &jmploc; 1328 INTON; 1329 if (oldstyle) { 1330 /* We must read until the closing backquote, giving special 1331 treatment to some slashes, and then push the string and 1332 reread it as input, interpreting it normally. */ 1333 char *out; 1334 int c; 1335 int savelen; 1336 char *str; 1337 1338 1339 STARTSTACKSTR(out); 1340 for (;;) { 1341 if (needprompt) { 1342 setprompt(2); 1343 needprompt = 0; 1344 } 1345 switch (c = pgetc()) { 1346 case '`': 1347 goto done; 1348 1349 case '\\': 1350 if ((c = pgetc()) == '\n') { 1351 plinno++; 1352 if (doprompt) 1353 setprompt(2); 1354 else 1355 setprompt(0); 1356 /* 1357 * If eating a newline, avoid putting 1358 * the newline into the new character 1359 * stream (via the STPUTC after the 1360 * switch). 1361 */ 1362 continue; 1363 } 1364 if (c != '\\' && c != '`' && c != '$' 1365 && (!dblquote || c != '"')) 1366 STPUTC('\\', out); 1367 break; 1368 1369 case '\n': 1370 plinno++; 1371 needprompt = doprompt; 1372 break; 1373 1374 case PEOF: 1375 startlinno = plinno; 1376 synerror("EOF in backquote substitution"); 1377 break; 1378 1379 default: 1380 break; 1381 } 1382 STPUTC(c, out); 1383 } 1384 done: 1385 STPUTC('\0', out); 1386 savelen = out - stackblock(); 1387 if (savelen > 0) { 1388 str = ckmalloc(savelen); 1389 memcpy(str, stackblock(), savelen); 1390 setinputstring(str, 1); 1391 } 1392 } 1393 nlpp = &bqlist; 1394 while (*nlpp) 1395 nlpp = &(*nlpp)->next; 1396 *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist)); 1397 (*nlpp)->next = NULL; 1398 parsebackquote = oldstyle; 1399 1400 if (oldstyle) { 1401 saveprompt = doprompt; 1402 doprompt = 0; 1403 } 1404 1405 n = list(0); 1406 1407 if (oldstyle) 1408 doprompt = saveprompt; 1409 else { 1410 if (readtoken() != TRP) 1411 synexpect(TRP); 1412 } 1413 1414 (*nlpp)->n = n; 1415 if (oldstyle) { 1416 /* 1417 * Start reading from old file again, ignoring any pushed back 1418 * tokens left from the backquote parsing 1419 */ 1420 popfile(); 1421 tokpushback = 0; 1422 } 1423 while (stackblocksize() <= savelen) 1424 growstackblock(); 1425 STARTSTACKSTR(out); 1426 if (str) { 1427 memcpy(out, str, savelen); 1428 STADJUST(savelen, out); 1429 INTOFF; 1430 ckfree(str); 1431 str = NULL; 1432 INTON; 1433 } 1434 parsebackquote = savepbq; 1435 handler = savehandler; 1436 if (arinest || dblquote) 1437 USTPUTC(CTLBACKQ | CTLQUOTE, out); 1438 else 1439 USTPUTC(CTLBACKQ, out); 1440 if (oldstyle) 1441 goto parsebackq_oldreturn; 1442 else 1443 goto parsebackq_newreturn; 1444 } 1445 1446 /* 1447 * Parse an arithmetic expansion (indicate start of one and set state) 1448 */ 1449 parsearith: { 1450 1451 if (++arinest == 1) { 1452 prevsyntax = syntax; 1453 syntax = ARISYNTAX; 1454 USTPUTC(CTLARI, out); 1455 if (dblquote) 1456 USTPUTC('"',out); 1457 else 1458 USTPUTC(' ',out); 1459 } else { 1460 /* 1461 * we collapse embedded arithmetic expansion to 1462 * parenthesis, which should be equivalent 1463 */ 1464 USTPUTC('(', out); 1465 } 1466 goto parsearith_return; 1467 } 1468 1469 } /* end of readtoken */ 1470 1471 1472 1473 #ifdef mkinit 1474 RESET { 1475 tokpushback = 0; 1476 checkkwd = 0; 1477 } 1478 #endif 1479 1480 /* 1481 * Returns true if the text contains nothing to expand (no dollar signs 1482 * or backquotes). 1483 */ 1484 1485 STATIC int 1486 noexpand(char *text) 1487 { 1488 char *p; 1489 char c; 1490 1491 p = text; 1492 while ((c = *p++) != '\0') { 1493 if ( c == CTLQUOTEMARK) 1494 continue; 1495 if (c == CTLESC) 1496 p++; 1497 else if (BASESYNTAX[(int)c] == CCTL) 1498 return 0; 1499 } 1500 return 1; 1501 } 1502 1503 1504 /* 1505 * Return true if the argument is a legal variable name (a letter or 1506 * underscore followed by zero or more letters, underscores, and digits). 1507 */ 1508 1509 int 1510 goodname(char *name) 1511 { 1512 char *p; 1513 1514 p = name; 1515 if (! is_name(*p)) 1516 return 0; 1517 while (*++p) { 1518 if (! is_in_name(*p)) 1519 return 0; 1520 } 1521 return 1; 1522 } 1523 1524 1525 /* 1526 * Called when an unexpected token is read during the parse. The argument 1527 * is the token that is expected, or -1 if more than one type of token can 1528 * occur at this point. 1529 */ 1530 1531 STATIC void 1532 synexpect(int token) 1533 { 1534 char msg[64]; 1535 1536 if (token >= 0) { 1537 fmtstr(msg, 64, "%s unexpected (expecting %s)", 1538 tokname[lasttoken], tokname[token]); 1539 } else { 1540 fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]); 1541 } 1542 synerror(msg); 1543 } 1544 1545 1546 STATIC void 1547 synerror(char *msg) 1548 { 1549 if (commandname) 1550 outfmt(&errout, "%s: %d: ", commandname, startlinno); 1551 outfmt(&errout, "Syntax error: %s\n", msg); 1552 error((char *)NULL); 1553 } 1554 1555 STATIC void 1556 setprompt(int which) 1557 { 1558 whichprompt = which; 1559 1560 #ifndef NO_HISTORY 1561 if (!el) 1562 #endif 1563 out2str(getprompt(NULL)); 1564 } 1565 1566 /* 1567 * called by editline -- any expansions to the prompt 1568 * should be added here. 1569 */ 1570 char * 1571 getprompt(void *unused __unused) 1572 { 1573 switch (whichprompt) { 1574 case 0: 1575 return ""; 1576 case 1: 1577 return ps1val(); 1578 case 2: 1579 return ps2val(); 1580 default: 1581 return "<internal prompt error>"; 1582 } 1583 } 1584