parser.c revision 1.40 1 /* $NetBSD: parser.c,v 1.40 1998/09/26 20:56:33 itohy Exp $ */
2
3 /*-
4 * Copyright (c) 1991, 1993
5 * The Regents of the University of California. All rights reserved.
6 *
7 * This code is derived from software contributed to Berkeley by
8 * Kenneth Almquist.
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 * 3. All advertising materials mentioning features or use of this software
19 * must display the following acknowledgement:
20 * This product includes software developed by the University of
21 * California, Berkeley and its contributors.
22 * 4. Neither the name of the University nor the names of its contributors
23 * may be used to endorse or promote products derived from this software
24 * without specific prior written permission.
25 *
26 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
27 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
29 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
30 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
31 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
32 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
33 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
34 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
35 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36 * SUCH DAMAGE.
37 */
38
39 #include <sys/cdefs.h>
40 #ifndef lint
41 #if 0
42 static char sccsid[] = "@(#)parser.c 8.7 (Berkeley) 5/16/95";
43 #else
44 __RCSID("$NetBSD: parser.c,v 1.40 1998/09/26 20:56:33 itohy Exp $");
45 #endif
46 #endif /* not lint */
47
48 #include <stdlib.h>
49
50 #include "shell.h"
51 #include "parser.h"
52 #include "nodes.h"
53 #include "expand.h" /* defines rmescapes() */
54 #include "redir.h" /* defines copyfd() */
55 #include "syntax.h"
56 #include "options.h"
57 #include "input.h"
58 #include "output.h"
59 #include "var.h"
60 #include "error.h"
61 #include "memalloc.h"
62 #include "mystring.h"
63 #include "alias.h"
64 #include "show.h"
65 #ifndef SMALL
66 #include "myhistedit.h"
67 #endif
68
69 /*
70 * Shell command parser.
71 */
72
73 #define EOFMARKLEN 79
74
75 /* values returned by readtoken */
76 #include "token.h"
77
78
79
80 struct heredoc {
81 struct heredoc *next; /* next here document in list */
82 union node *here; /* redirection node */
83 char *eofmark; /* string indicating end of input */
84 int striptabs; /* if set, strip leading tabs */
85 };
86
87
88
89 struct heredoc *heredoclist; /* list of here documents to read */
90 int parsebackquote; /* nonzero if we are inside backquotes */
91 int doprompt; /* if set, prompt the user */
92 int needprompt; /* true if interactive and at start of line */
93 int lasttoken; /* last token read */
94 MKINIT int tokpushback; /* last token pushed back */
95 char *wordtext; /* text of last word returned by readtoken */
96 MKINIT int checkkwd; /* 1 == check for kwds, 2 == also eat newlines */
97 struct nodelist *backquotelist;
98 union node *redirnode;
99 struct heredoc *heredoc;
100 int quoteflag; /* set if (part of) last token was quoted */
101 int startlinno; /* line # where last token started */
102
103
104 #define GDB_HACK 1 /* avoid local declarations which gdb can't handle */
105 #ifdef GDB_HACK
106 static const char argvars[5] = {(char)CTLVAR, (char)(VSNORMAL|VSQUOTE),
107 '@', '=', '\0'};
108 static const char types[] = "}-+?=";
109 #endif
110
111
112 STATIC union node *list __P((int));
113 STATIC union node *andor __P((void));
114 STATIC union node *pipeline __P((void));
115 STATIC union node *command __P((void));
116 STATIC union node *simplecmd __P((union node **, union node *));
117 STATIC union node *makename __P((void));
118 STATIC void parsefname __P((void));
119 STATIC void parseheredoc __P((void));
120 STATIC int peektoken __P((void));
121 STATIC int readtoken __P((void));
122 STATIC int xxreadtoken __P((void));
123 STATIC int readtoken1 __P((int, char const *, char *, int));
124 STATIC int noexpand __P((char *));
125 STATIC void synexpect __P((int)) __attribute__((noreturn));
126 STATIC void synerror __P((char *)) __attribute__((noreturn));
127 STATIC void setprompt __P((int));
128
129
130 /*
131 * Read and parse a command. Returns NEOF on end of file. (NULL is a
132 * valid parse tree indicating a blank line.)
133 */
134
135 union node *
136 parsecmd(interact)
137 int interact;
138 {
139 int t;
140
141 doprompt = interact;
142 if (doprompt)
143 setprompt(1);
144 else
145 setprompt(0);
146 needprompt = 0;
147 t = readtoken();
148 if (t == TEOF)
149 return NEOF;
150 if (t == TNL)
151 return NULL;
152 tokpushback++;
153 return list(1);
154 }
155
156
157 STATIC union node *
158 list(nlflag)
159 int nlflag;
160 {
161 union node *n1, *n2, *n3;
162 int tok;
163
164 checkkwd = 2;
165 if (nlflag == 0 && tokendlist[peektoken()])
166 return NULL;
167 n1 = NULL;
168 for (;;) {
169 n2 = andor();
170 tok = readtoken();
171 if (tok == TBACKGND) {
172 if (n2->type == NCMD || n2->type == NPIPE) {
173 n2->ncmd.backgnd = 1;
174 } else if (n2->type == NREDIR) {
175 n2->type = NBACKGND;
176 } else {
177 n3 = (union node *)stalloc(sizeof (struct nredir));
178 n3->type = NBACKGND;
179 n3->nredir.n = n2;
180 n3->nredir.redirect = NULL;
181 n2 = n3;
182 }
183 }
184 if (n1 == NULL) {
185 n1 = n2;
186 }
187 else {
188 n3 = (union node *)stalloc(sizeof (struct nbinary));
189 n3->type = NSEMI;
190 n3->nbinary.ch1 = n1;
191 n3->nbinary.ch2 = n2;
192 n1 = n3;
193 }
194 switch (tok) {
195 case TBACKGND:
196 case TSEMI:
197 tok = readtoken();
198 /* fall through */
199 case TNL:
200 if (tok == TNL) {
201 parseheredoc();
202 if (nlflag)
203 return n1;
204 } else {
205 tokpushback++;
206 }
207 checkkwd = 2;
208 if (tokendlist[peektoken()])
209 return n1;
210 break;
211 case TEOF:
212 if (heredoclist)
213 parseheredoc();
214 else
215 pungetc(); /* push back EOF on input */
216 return n1;
217 default:
218 if (nlflag)
219 synexpect(-1);
220 tokpushback++;
221 return n1;
222 }
223 }
224 }
225
226
227
228 STATIC union node *
229 andor() {
230 union node *n1, *n2, *n3;
231 int t;
232
233 n1 = pipeline();
234 for (;;) {
235 if ((t = readtoken()) == TAND) {
236 t = NAND;
237 } else if (t == TOR) {
238 t = NOR;
239 } else {
240 tokpushback++;
241 return n1;
242 }
243 n2 = pipeline();
244 n3 = (union node *)stalloc(sizeof (struct nbinary));
245 n3->type = t;
246 n3->nbinary.ch1 = n1;
247 n3->nbinary.ch2 = n2;
248 n1 = n3;
249 }
250 }
251
252
253
254 STATIC union node *
255 pipeline() {
256 union node *n1, *pipenode;
257 struct nodelist *lp, *prev;
258
259 TRACE(("pipeline: entered\n"));
260 n1 = command();
261 if (readtoken() == TPIPE) {
262 pipenode = (union node *)stalloc(sizeof (struct npipe));
263 pipenode->type = NPIPE;
264 pipenode->npipe.backgnd = 0;
265 lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
266 pipenode->npipe.cmdlist = lp;
267 lp->n = n1;
268 do {
269 prev = lp;
270 lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
271 lp->n = command();
272 prev->next = lp;
273 } while (readtoken() == TPIPE);
274 lp->next = NULL;
275 n1 = pipenode;
276 }
277 tokpushback++;
278 return n1;
279 }
280
281
282
283 STATIC union node *
284 command() {
285 union node *n1, *n2;
286 union node *ap, **app;
287 union node *cp, **cpp;
288 union node *redir, **rpp;
289 int t, negate = 0;
290
291 checkkwd = 2;
292 redir = NULL;
293 n1 = NULL;
294 rpp = &redir;
295
296 /* Check for redirection which may precede command */
297 while (readtoken() == TREDIR) {
298 *rpp = n2 = redirnode;
299 rpp = &n2->nfile.next;
300 parsefname();
301 }
302 tokpushback++;
303
304 while (readtoken() == TNOT) {
305 TRACE(("command: TNOT recognized\n"));
306 negate = !negate;
307 }
308 tokpushback++;
309
310 switch (readtoken()) {
311 case TIF:
312 n1 = (union node *)stalloc(sizeof (struct nif));
313 n1->type = NIF;
314 n1->nif.test = list(0);
315 if (readtoken() != TTHEN)
316 synexpect(TTHEN);
317 n1->nif.ifpart = list(0);
318 n2 = n1;
319 while (readtoken() == TELIF) {
320 n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif));
321 n2 = n2->nif.elsepart;
322 n2->type = NIF;
323 n2->nif.test = list(0);
324 if (readtoken() != TTHEN)
325 synexpect(TTHEN);
326 n2->nif.ifpart = list(0);
327 }
328 if (lasttoken == TELSE)
329 n2->nif.elsepart = list(0);
330 else {
331 n2->nif.elsepart = NULL;
332 tokpushback++;
333 }
334 if (readtoken() != TFI)
335 synexpect(TFI);
336 checkkwd = 1;
337 break;
338 case TWHILE:
339 case TUNTIL: {
340 int got;
341 n1 = (union node *)stalloc(sizeof (struct nbinary));
342 n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL;
343 n1->nbinary.ch1 = list(0);
344 if ((got=readtoken()) != TDO) {
345 TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : ""));
346 synexpect(TDO);
347 }
348 n1->nbinary.ch2 = list(0);
349 if (readtoken() != TDONE)
350 synexpect(TDONE);
351 checkkwd = 1;
352 break;
353 }
354 case TFOR:
355 if (readtoken() != TWORD || quoteflag || ! goodname(wordtext))
356 synerror("Bad for loop variable");
357 n1 = (union node *)stalloc(sizeof (struct nfor));
358 n1->type = NFOR;
359 n1->nfor.var = wordtext;
360 if (readtoken() == TWORD && ! quoteflag && equal(wordtext, "in")) {
361 app = ≈
362 while (readtoken() == TWORD) {
363 n2 = (union node *)stalloc(sizeof (struct narg));
364 n2->type = NARG;
365 n2->narg.text = wordtext;
366 n2->narg.backquote = backquotelist;
367 *app = n2;
368 app = &n2->narg.next;
369 }
370 *app = NULL;
371 n1->nfor.args = ap;
372 if (lasttoken != TNL && lasttoken != TSEMI)
373 synexpect(-1);
374 } else {
375 #ifndef GDB_HACK
376 static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE,
377 '@', '=', '\0'};
378 #endif
379 n2 = (union node *)stalloc(sizeof (struct narg));
380 n2->type = NARG;
381 n2->narg.text = (char *)argvars;
382 n2->narg.backquote = NULL;
383 n2->narg.next = NULL;
384 n1->nfor.args = n2;
385 /*
386 * Newline or semicolon here is optional (but note
387 * that the original Bourne shell only allowed NL).
388 */
389 if (lasttoken != TNL && lasttoken != TSEMI)
390 tokpushback++;
391 }
392 checkkwd = 2;
393 if ((t = readtoken()) == TDO)
394 t = TDONE;
395 else if (t == TBEGIN)
396 t = TEND;
397 else
398 synexpect(-1);
399 n1->nfor.body = list(0);
400 if (readtoken() != t)
401 synexpect(t);
402 checkkwd = 1;
403 break;
404 case TCASE:
405 n1 = (union node *)stalloc(sizeof (struct ncase));
406 n1->type = NCASE;
407 if (readtoken() != TWORD)
408 synexpect(TWORD);
409 n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg));
410 n2->type = NARG;
411 n2->narg.text = wordtext;
412 n2->narg.backquote = backquotelist;
413 n2->narg.next = NULL;
414 while (readtoken() == TNL);
415 if (lasttoken != TWORD || ! equal(wordtext, "in"))
416 synerror("expecting \"in\"");
417 cpp = &n1->ncase.cases;
418 checkkwd = 2, readtoken();
419 do {
420 *cpp = cp = (union node *)stalloc(sizeof (struct nclist));
421 cp->type = NCLIST;
422 app = &cp->nclist.pattern;
423 for (;;) {
424 *app = ap = (union node *)stalloc(sizeof (struct narg));
425 ap->type = NARG;
426 ap->narg.text = wordtext;
427 ap->narg.backquote = backquotelist;
428 if (checkkwd = 2, readtoken() != TPIPE)
429 break;
430 app = &ap->narg.next;
431 readtoken();
432 }
433 ap->narg.next = NULL;
434 if (lasttoken != TRP)
435 synexpect(TRP);
436 cp->nclist.body = list(0);
437
438 checkkwd = 2;
439 if ((t = readtoken()) != TESAC) {
440 if (t != TENDCASE)
441 synexpect(TENDCASE);
442 else
443 checkkwd = 2, readtoken();
444 }
445 cpp = &cp->nclist.next;
446 } while(lasttoken != TESAC);
447 *cpp = NULL;
448 checkkwd = 1;
449 break;
450 case TLP:
451 n1 = (union node *)stalloc(sizeof (struct nredir));
452 n1->type = NSUBSHELL;
453 n1->nredir.n = list(0);
454 n1->nredir.redirect = NULL;
455 if (readtoken() != TRP)
456 synexpect(TRP);
457 checkkwd = 1;
458 break;
459 case TBEGIN:
460 n1 = list(0);
461 if (readtoken() != TEND)
462 synexpect(TEND);
463 checkkwd = 1;
464 break;
465 /* Handle an empty command like other simple commands. */
466 case TSEMI:
467 /*
468 * An empty command before a ; doesn't make much sense, and
469 * should certainly be disallowed in the case of `if ;'.
470 */
471 if (!redir)
472 synexpect(-1);
473 case TAND:
474 case TOR:
475 case TNL:
476 case TEOF:
477 case TWORD:
478 case TRP:
479 tokpushback++;
480 n1 = simplecmd(rpp, redir);
481 goto checkneg;
482 default:
483 synexpect(-1);
484 /* NOTREACHED */
485 }
486
487 /* Now check for redirection which may follow command */
488 while (readtoken() == TREDIR) {
489 *rpp = n2 = redirnode;
490 rpp = &n2->nfile.next;
491 parsefname();
492 }
493 tokpushback++;
494 *rpp = NULL;
495 if (redir) {
496 if (n1->type != NSUBSHELL) {
497 n2 = (union node *)stalloc(sizeof (struct nredir));
498 n2->type = NREDIR;
499 n2->nredir.n = n1;
500 n1 = n2;
501 }
502 n1->nredir.redirect = redir;
503 }
504
505 checkneg:
506 if (negate) {
507 n2 = (union node *)stalloc(sizeof (struct nnot));
508 n2->type = NNOT;
509 n2->nnot.com = n1;
510 return n2;
511 }
512 else
513 return n1;
514 }
515
516
517 STATIC union node *
518 simplecmd(rpp, redir)
519 union node **rpp, *redir;
520 {
521 union node *args, **app;
522 union node **orig_rpp = rpp;
523 union node *n = NULL, *n2;
524 int negate = 0;
525
526 /* If we don't have any redirections already, then we must reset */
527 /* rpp to be the address of the local redir variable. */
528 if (redir == 0)
529 rpp = &redir;
530
531 args = NULL;
532 app = &args;
533 /*
534 * We save the incoming value, because we need this for shell
535 * functions. There can not be a redirect or an argument between
536 * the function name and the open parenthesis.
537 */
538 orig_rpp = rpp;
539
540 while (readtoken() == TNOT) {
541 TRACE(("command: TNOT recognized\n"));
542 negate = !negate;
543 }
544 tokpushback++;
545
546 for (;;) {
547 if (readtoken() == TWORD) {
548 n = (union node *)stalloc(sizeof (struct narg));
549 n->type = NARG;
550 n->narg.text = wordtext;
551 n->narg.backquote = backquotelist;
552 *app = n;
553 app = &n->narg.next;
554 } else if (lasttoken == TREDIR) {
555 *rpp = n = redirnode;
556 rpp = &n->nfile.next;
557 parsefname(); /* read name of redirection file */
558 } else if (lasttoken == TLP && app == &args->narg.next
559 && rpp == orig_rpp) {
560 /* We have a function */
561 if (readtoken() != TRP)
562 synexpect(TRP);
563 #ifdef notdef
564 if (! goodname(n->narg.text))
565 synerror("Bad function name");
566 #endif
567 n->type = NDEFUN;
568 n->narg.next = command();
569 goto checkneg;
570 } else {
571 tokpushback++;
572 break;
573 }
574 }
575 *app = NULL;
576 *rpp = NULL;
577 n = (union node *)stalloc(sizeof (struct ncmd));
578 n->type = NCMD;
579 n->ncmd.backgnd = 0;
580 n->ncmd.args = args;
581 n->ncmd.redirect = redir;
582
583 checkneg:
584 if (negate) {
585 n2 = (union node *)stalloc(sizeof (struct nnot));
586 n2->type = NNOT;
587 n2->nnot.com = n;
588 return n2;
589 }
590 else
591 return n;
592 }
593
594 STATIC union node *
595 makename() {
596 union node *n;
597
598 n = (union node *)stalloc(sizeof (struct narg));
599 n->type = NARG;
600 n->narg.next = NULL;
601 n->narg.text = wordtext;
602 n->narg.backquote = backquotelist;
603 return n;
604 }
605
606 void fixredir(n, text, err)
607 union node *n;
608 const char *text;
609 int err;
610 {
611 TRACE(("Fix redir %s %d\n", text, err));
612 if (!err)
613 n->ndup.vname = NULL;
614
615 if (is_digit(text[0]) && text[1] == '\0')
616 n->ndup.dupfd = digit_val(text[0]);
617 else if (text[0] == '-' && text[1] == '\0')
618 n->ndup.dupfd = -1;
619 else {
620
621 if (err)
622 synerror("Bad fd number");
623 else
624 n->ndup.vname = makename();
625 }
626 }
627
628
629 STATIC void
630 parsefname() {
631 union node *n = redirnode;
632
633 if (readtoken() != TWORD)
634 synexpect(-1);
635 if (n->type == NHERE) {
636 struct heredoc *here = heredoc;
637 struct heredoc *p;
638 int i;
639
640 if (quoteflag == 0)
641 n->type = NXHERE;
642 TRACE(("Here document %d\n", n->type));
643 if (here->striptabs) {
644 while (*wordtext == '\t')
645 wordtext++;
646 }
647 if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN)
648 synerror("Illegal eof marker for << redirection");
649 rmescapes(wordtext);
650 here->eofmark = wordtext;
651 here->next = NULL;
652 if (heredoclist == NULL)
653 heredoclist = here;
654 else {
655 for (p = heredoclist ; p->next ; p = p->next);
656 p->next = here;
657 }
658 } else if (n->type == NTOFD || n->type == NFROMFD) {
659 fixredir(n, wordtext, 0);
660 } else {
661 n->nfile.fname = makename();
662 }
663 }
664
665
666 /*
667 * Input any here documents.
668 */
669
670 STATIC void
671 parseheredoc() {
672 struct heredoc *here;
673 union node *n;
674
675 while (heredoclist) {
676 here = heredoclist;
677 heredoclist = here->next;
678 if (needprompt) {
679 setprompt(2);
680 needprompt = 0;
681 }
682 readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX,
683 here->eofmark, here->striptabs);
684 n = (union node *)stalloc(sizeof (struct narg));
685 n->narg.type = NARG;
686 n->narg.next = NULL;
687 n->narg.text = wordtext;
688 n->narg.backquote = backquotelist;
689 here->here->nhere.doc = n;
690 }
691 }
692
693 STATIC int
694 peektoken() {
695 int t;
696
697 t = readtoken();
698 tokpushback++;
699 return (t);
700 }
701
702 STATIC int
703 readtoken() {
704 int t;
705 int savecheckkwd = checkkwd;
706 struct alias *ap;
707 #ifdef DEBUG
708 int alreadyseen = tokpushback;
709 #endif
710
711 top:
712 t = xxreadtoken();
713
714 if (checkkwd) {
715 /*
716 * eat newlines
717 */
718 if (checkkwd == 2) {
719 checkkwd = 0;
720 while (t == TNL) {
721 parseheredoc();
722 t = xxreadtoken();
723 }
724 } else
725 checkkwd = 0;
726 /*
727 * check for keywords and aliases
728 */
729 if (t == TWORD && !quoteflag)
730 {
731 char * const *pp;
732
733 for (pp = (char **)parsekwd; *pp; pp++) {
734 if (**pp == *wordtext && equal(*pp, wordtext))
735 {
736 lasttoken = t = pp - parsekwd + KWDOFFSET;
737 TRACE(("keyword %s recognized\n", tokname[t]));
738 goto out;
739 }
740 }
741 if ((ap = lookupalias(wordtext, 1)) != NULL) {
742 pushstring(ap->val, strlen(ap->val), ap);
743 checkkwd = savecheckkwd;
744 goto top;
745 }
746 }
747 out:
748 checkkwd = (t == TNOT) ? savecheckkwd : 0;
749 }
750 #ifdef DEBUG
751 if (!alreadyseen)
752 TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
753 else
754 TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
755 #endif
756 return (t);
757 }
758
759
760 /*
761 * Read the next input token.
762 * If the token is a word, we set backquotelist to the list of cmds in
763 * backquotes. We set quoteflag to true if any part of the word was
764 * quoted.
765 * If the token is TREDIR, then we set redirnode to a structure containing
766 * the redirection.
767 * In all cases, the variable startlinno is set to the number of the line
768 * on which the token starts.
769 *
770 * [Change comment: here documents and internal procedures]
771 * [Readtoken shouldn't have any arguments. Perhaps we should make the
772 * word parsing code into a separate routine. In this case, readtoken
773 * doesn't need to have any internal procedures, but parseword does.
774 * We could also make parseoperator in essence the main routine, and
775 * have parseword (readtoken1?) handle both words and redirection.]
776 */
777
778 #define RETURN(token) return lasttoken = token
779
780 STATIC int
781 xxreadtoken() {
782 int c;
783
784 if (tokpushback) {
785 tokpushback = 0;
786 return lasttoken;
787 }
788 if (needprompt) {
789 setprompt(2);
790 needprompt = 0;
791 }
792 startlinno = plinno;
793 for (;;) { /* until token or start of word found */
794 c = pgetc_macro();
795 if (c == ' ' || c == '\t')
796 continue; /* quick check for white space first */
797 switch (c) {
798 case ' ': case '\t':
799 continue;
800 case '#':
801 while ((c = pgetc()) != '\n' && c != PEOF);
802 pungetc();
803 continue;
804 case '\\':
805 if (pgetc() == '\n') {
806 startlinno = ++plinno;
807 if (doprompt)
808 setprompt(2);
809 else
810 setprompt(0);
811 continue;
812 }
813 pungetc();
814 goto breakloop;
815 case '\n':
816 plinno++;
817 needprompt = doprompt;
818 RETURN(TNL);
819 case PEOF:
820 RETURN(TEOF);
821 case '&':
822 if (pgetc() == '&')
823 RETURN(TAND);
824 pungetc();
825 RETURN(TBACKGND);
826 case '|':
827 if (pgetc() == '|')
828 RETURN(TOR);
829 pungetc();
830 RETURN(TPIPE);
831 case ';':
832 if (pgetc() == ';')
833 RETURN(TENDCASE);
834 pungetc();
835 RETURN(TSEMI);
836 case '(':
837 RETURN(TLP);
838 case ')':
839 RETURN(TRP);
840 default:
841 goto breakloop;
842 }
843 }
844 breakloop:
845 return readtoken1(c, BASESYNTAX, (char *)NULL, 0);
846 #undef RETURN
847 }
848
849
850
851 /*
852 * If eofmark is NULL, read a word or a redirection symbol. If eofmark
853 * is not NULL, read a here document. In the latter case, eofmark is the
854 * word which marks the end of the document and striptabs is true if
855 * leading tabs should be stripped from the document. The argument firstc
856 * is the first character of the input token or document.
857 *
858 * Because C does not have internal subroutines, I have simulated them
859 * using goto's to implement the subroutine linkage. The following macros
860 * will run code that appears at the end of readtoken1.
861 */
862
863 #define CHECKEND() {goto checkend; checkend_return:;}
864 #define PARSEREDIR() {goto parseredir; parseredir_return:;}
865 #define PARSESUB() {goto parsesub; parsesub_return:;}
866 #define PARSEBACKQOLD() {oldstyle = 1; goto parsebackq; parsebackq_oldreturn:;}
867 #define PARSEBACKQNEW() {oldstyle = 0; goto parsebackq; parsebackq_newreturn:;}
868 #define PARSEARITH() {goto parsearith; parsearith_return:;}
869
870 STATIC int
871 readtoken1(firstc, syntax, eofmark, striptabs)
872 int firstc;
873 char const *syntax;
874 char *eofmark;
875 int striptabs;
876 {
877 int c = firstc;
878 char *out;
879 int len;
880 char line[EOFMARKLEN + 1];
881 struct nodelist *bqlist;
882 int quotef;
883 int dblquote;
884 int varnest; /* levels of variables expansion */
885 int arinest; /* levels of arithmetic expansion */
886 int parenlevel; /* levels of parens in arithmetic */
887 int oldstyle;
888 char const *prevsyntax; /* syntax before arithmetic */
889 #if __GNUC__
890 /* Avoid longjmp clobbering */
891 (void) &out;
892 (void) "ef;
893 (void) &dblquote;
894 (void) &varnest;
895 (void) &arinest;
896 (void) &parenlevel;
897 (void) &oldstyle;
898 (void) &prevsyntax;
899 (void) &syntax;
900 #endif
901
902 startlinno = plinno;
903 dblquote = 0;
904 if (syntax == DQSYNTAX)
905 dblquote = 1;
906 quotef = 0;
907 bqlist = NULL;
908 varnest = 0;
909 arinest = 0;
910 parenlevel = 0;
911
912 STARTSTACKSTR(out);
913 loop: { /* for each line, until end of word */
914 #if ATTY
915 if (c == '\034' && doprompt
916 && attyset() && ! equal(termval(), "emacs")) {
917 attyline();
918 if (syntax == BASESYNTAX)
919 return readtoken();
920 c = pgetc();
921 goto loop;
922 }
923 #endif
924 CHECKEND(); /* set c to PEOF if at end of here document */
925 for (;;) { /* until end of line or end of word */
926 CHECKSTRSPACE(3, out); /* permit 3 calls to USTPUTC */
927 switch(syntax[c]) {
928 case CNL: /* '\n' */
929 if (syntax == BASESYNTAX)
930 goto endword; /* exit outer loop */
931 USTPUTC(c, out);
932 plinno++;
933 if (doprompt)
934 setprompt(2);
935 else
936 setprompt(0);
937 c = pgetc();
938 goto loop; /* continue outer loop */
939 case CWORD:
940 USTPUTC(c, out);
941 break;
942 case CCTL:
943 if (eofmark == NULL || dblquote)
944 USTPUTC(CTLESC, out);
945 USTPUTC(c, out);
946 break;
947 case CBACK: /* backslash */
948 c = pgetc();
949 if (c == PEOF) {
950 USTPUTC('\\', out);
951 pungetc();
952 } else if (c == '\n') {
953 if (doprompt)
954 setprompt(2);
955 else
956 setprompt(0);
957 } else {
958 if (dblquote && c != '\\' && c != '`' && c != '$'
959 && (c != '"' || eofmark != NULL))
960 USTPUTC('\\', out);
961 if (SQSYNTAX[c] == CCTL)
962 USTPUTC(CTLESC, out);
963 USTPUTC(c, out);
964 quotef++;
965 }
966 break;
967 case CSQUOTE:
968 syntax = SQSYNTAX;
969 break;
970 case CDQUOTE:
971 syntax = DQSYNTAX;
972 dblquote = 1;
973 break;
974 case CENDQUOTE:
975 if (eofmark) {
976 USTPUTC(c, out);
977 } else {
978 if (arinest)
979 syntax = ARISYNTAX;
980 else
981 syntax = BASESYNTAX;
982 quotef++;
983 dblquote = 0;
984 }
985 break;
986 case CVAR: /* '$' */
987 PARSESUB(); /* parse substitution */
988 break;
989 case CENDVAR: /* '}' */
990 if (varnest > 0) {
991 varnest--;
992 USTPUTC(CTLENDVAR, out);
993 } else {
994 USTPUTC(c, out);
995 }
996 break;
997 case CLP: /* '(' in arithmetic */
998 parenlevel++;
999 USTPUTC(c, out);
1000 break;
1001 case CRP: /* ')' in arithmetic */
1002 if (parenlevel > 0) {
1003 USTPUTC(c, out);
1004 --parenlevel;
1005 } else {
1006 if (pgetc() == ')') {
1007 if (--arinest == 0) {
1008 USTPUTC(CTLENDARI, out);
1009 syntax = prevsyntax;
1010 } else
1011 USTPUTC(')', out);
1012 } else {
1013 /*
1014 * unbalanced parens
1015 * (don't 2nd guess - no error)
1016 */
1017 pungetc();
1018 USTPUTC(')', out);
1019 }
1020 }
1021 break;
1022 case CBQUOTE: /* '`' */
1023 PARSEBACKQOLD();
1024 break;
1025 case CEOF:
1026 goto endword; /* exit outer loop */
1027 default:
1028 if (varnest == 0)
1029 goto endword; /* exit outer loop */
1030 USTPUTC(c, out);
1031 }
1032 c = pgetc_macro();
1033 }
1034 }
1035 endword:
1036 if (syntax == ARISYNTAX)
1037 synerror("Missing '))'");
1038 if (syntax != BASESYNTAX && ! parsebackquote && eofmark == NULL)
1039 synerror("Unterminated quoted string");
1040 if (varnest != 0) {
1041 startlinno = plinno;
1042 synerror("Missing '}'");
1043 }
1044 USTPUTC('\0', out);
1045 len = out - stackblock();
1046 out = stackblock();
1047 if (eofmark == NULL) {
1048 if ((c == '>' || c == '<')
1049 && quotef == 0
1050 && len <= 2
1051 && (*out == '\0' || is_digit(*out))) {
1052 PARSEREDIR();
1053 return lasttoken = TREDIR;
1054 } else {
1055 pungetc();
1056 }
1057 }
1058 quoteflag = quotef;
1059 backquotelist = bqlist;
1060 grabstackblock(len);
1061 wordtext = out;
1062 return lasttoken = TWORD;
1063 /* end of readtoken routine */
1064
1065
1066
1067 /*
1068 * Check to see whether we are at the end of the here document. When this
1069 * is called, c is set to the first character of the next input line. If
1070 * we are at the end of the here document, this routine sets the c to PEOF.
1071 */
1072
1073 checkend: {
1074 if (eofmark) {
1075 if (striptabs) {
1076 while (c == '\t')
1077 c = pgetc();
1078 }
1079 if (c == *eofmark) {
1080 if (pfgets(line, sizeof line) != NULL) {
1081 char *p, *q;
1082
1083 p = line;
1084 for (q = eofmark + 1 ; *q && *p == *q ; p++, q++);
1085 if (*p == '\n' && *q == '\0') {
1086 c = PEOF;
1087 plinno++;
1088 needprompt = doprompt;
1089 } else {
1090 pushstring(line, strlen(line), NULL);
1091 }
1092 }
1093 }
1094 }
1095 goto checkend_return;
1096 }
1097
1098
1099 /*
1100 * Parse a redirection operator. The variable "out" points to a string
1101 * specifying the fd to be redirected. The variable "c" contains the
1102 * first character of the redirection operator.
1103 */
1104
1105 parseredir: {
1106 char fd = *out;
1107 union node *np;
1108
1109 np = (union node *)stalloc(sizeof (struct nfile));
1110 if (c == '>') {
1111 np->nfile.fd = 1;
1112 c = pgetc();
1113 if (c == '>')
1114 np->type = NAPPEND;
1115 else if (c == '&')
1116 np->type = NTOFD;
1117 else {
1118 np->type = NTO;
1119 pungetc();
1120 }
1121 } else { /* c == '<' */
1122 np->nfile.fd = 0;
1123 c = pgetc();
1124 if (c == '<') {
1125 if (sizeof (struct nfile) != sizeof (struct nhere)) {
1126 np = (union node *)stalloc(sizeof (struct nhere));
1127 np->nfile.fd = 0;
1128 }
1129 np->type = NHERE;
1130 heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc));
1131 heredoc->here = np;
1132 if ((c = pgetc()) == '-') {
1133 heredoc->striptabs = 1;
1134 } else {
1135 heredoc->striptabs = 0;
1136 pungetc();
1137 }
1138 } else if (c == '&')
1139 np->type = NFROMFD;
1140 else {
1141 np->type = NFROM;
1142 pungetc();
1143 }
1144 }
1145 if (fd != '\0')
1146 np->nfile.fd = digit_val(fd);
1147 redirnode = np;
1148 goto parseredir_return;
1149 }
1150
1151
1152 /*
1153 * Parse a substitution. At this point, we have read the dollar sign
1154 * and nothing else.
1155 */
1156
1157 parsesub: {
1158 int subtype;
1159 int typeloc;
1160 int flags;
1161 char *p;
1162 #ifndef GDB_HACK
1163 static const char types[] = "}-+?=";
1164 #endif
1165
1166 c = pgetc();
1167 if (c != '(' && c != '{' && !is_name(c) && !is_special(c)) {
1168 USTPUTC('$', out);
1169 pungetc();
1170 } else if (c == '(') { /* $(command) or $((arith)) */
1171 if (pgetc() == '(') {
1172 PARSEARITH();
1173 } else {
1174 pungetc();
1175 PARSEBACKQNEW();
1176 }
1177 } else {
1178 USTPUTC(CTLVAR, out);
1179 typeloc = out - stackblock();
1180 USTPUTC(VSNORMAL, out);
1181 subtype = VSNORMAL;
1182 if (c == '{') {
1183 c = pgetc();
1184 if (c == '#') {
1185 if ((c = pgetc()) == '}')
1186 c = '#';
1187 else
1188 subtype = VSLENGTH;
1189 }
1190 else
1191 subtype = 0;
1192 }
1193 if (is_name(c)) {
1194 do {
1195 STPUTC(c, out);
1196 c = pgetc();
1197 } while (is_in_name(c));
1198 } else if (is_digit(c)) {
1199 do {
1200 USTPUTC(c, out);
1201 c = pgetc();
1202 } while (is_digit(c));
1203 }
1204 else if (is_special(c)) {
1205 USTPUTC(c, out);
1206 c = pgetc();
1207 }
1208 else
1209 badsub: synerror("Bad substitution");
1210
1211 STPUTC('=', out);
1212 flags = 0;
1213 if (subtype == 0) {
1214 switch (c) {
1215 case ':':
1216 flags = VSNUL;
1217 c = pgetc();
1218 /*FALLTHROUGH*/
1219 default:
1220 p = strchr(types, c);
1221 if (p == NULL)
1222 goto badsub;
1223 subtype = p - types + VSNORMAL;
1224 break;
1225 case '%':
1226 case '#':
1227 {
1228 int cc = c;
1229 subtype = c == '#' ? VSTRIMLEFT :
1230 VSTRIMRIGHT;
1231 c = pgetc();
1232 if (c == cc)
1233 subtype++;
1234 else
1235 pungetc();
1236 break;
1237 }
1238 }
1239 } else {
1240 pungetc();
1241 }
1242 if (dblquote || arinest)
1243 flags |= VSQUOTE;
1244 *(stackblock() + typeloc) = subtype | flags;
1245 if (subtype != VSNORMAL)
1246 varnest++;
1247 }
1248 goto parsesub_return;
1249 }
1250
1251
1252 /*
1253 * Called to parse command substitutions. Newstyle is set if the command
1254 * is enclosed inside $(...); nlpp is a pointer to the head of the linked
1255 * list of commands (passed by reference), and savelen is the number of
1256 * characters on the top of the stack which must be preserved.
1257 */
1258
1259 parsebackq: {
1260 struct nodelist **nlpp;
1261 int savepbq;
1262 union node *n;
1263 char *volatile str;
1264 struct jmploc jmploc;
1265 struct jmploc *volatile savehandler;
1266 int savelen;
1267 int saveprompt;
1268 #ifdef __GNUC__
1269 (void) &saveprompt;
1270 #endif
1271
1272 savepbq = parsebackquote;
1273 if (setjmp(jmploc.loc)) {
1274 if (str)
1275 ckfree(str);
1276 parsebackquote = 0;
1277 handler = savehandler;
1278 longjmp(handler->loc, 1);
1279 }
1280 INTOFF;
1281 str = NULL;
1282 savelen = out - stackblock();
1283 if (savelen > 0) {
1284 str = ckmalloc(savelen);
1285 memcpy(str, stackblock(), savelen);
1286 }
1287 savehandler = handler;
1288 handler = &jmploc;
1289 INTON;
1290 if (oldstyle) {
1291 /* We must read until the closing backquote, giving special
1292 treatment to some slashes, and then push the string and
1293 reread it as input, interpreting it normally. */
1294 char *out;
1295 int c;
1296 int savelen;
1297 char *str;
1298
1299
1300 STARTSTACKSTR(out);
1301 for (;;) {
1302 if (needprompt) {
1303 setprompt(2);
1304 needprompt = 0;
1305 }
1306 switch (c = pgetc()) {
1307 case '`':
1308 goto done;
1309
1310 case '\\':
1311 if ((c = pgetc()) == '\n') {
1312 plinno++;
1313 if (doprompt)
1314 setprompt(2);
1315 else
1316 setprompt(0);
1317 /*
1318 * If eating a newline, avoid putting
1319 * the newline into the new character
1320 * stream (via the STPUTC after the
1321 * switch).
1322 */
1323 continue;
1324 }
1325 if (c != '\\' && c != '`' && c != '$'
1326 && (!dblquote || c != '"'))
1327 STPUTC('\\', out);
1328 break;
1329
1330 case '\n':
1331 plinno++;
1332 needprompt = doprompt;
1333 break;
1334
1335 case PEOF:
1336 startlinno = plinno;
1337 synerror("EOF in backquote substitution");
1338 break;
1339
1340 default:
1341 break;
1342 }
1343 STPUTC(c, out);
1344 }
1345 done:
1346 STPUTC('\0', out);
1347 savelen = out - stackblock();
1348 if (savelen > 0) {
1349 str = grabstackstr(out);
1350 setinputstring(str, 1);
1351 }
1352 }
1353 nlpp = &bqlist;
1354 while (*nlpp)
1355 nlpp = &(*nlpp)->next;
1356 *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist));
1357 (*nlpp)->next = NULL;
1358 parsebackquote = oldstyle;
1359
1360 if (oldstyle) {
1361 saveprompt = doprompt;
1362 doprompt = 0;
1363 }
1364
1365 n = list(0);
1366
1367 if (oldstyle)
1368 doprompt = saveprompt;
1369 else {
1370 if (readtoken() != TRP)
1371 synexpect(TRP);
1372 }
1373
1374 (*nlpp)->n = n;
1375 if (oldstyle) {
1376 /*
1377 * Start reading from old file again, ignoring any pushed back
1378 * tokens left from the backquote parsing
1379 */
1380 popfile();
1381 tokpushback = 0;
1382 }
1383 while (stackblocksize() <= savelen)
1384 growstackblock();
1385 STARTSTACKSTR(out);
1386 if (str) {
1387 memcpy(out, str, savelen);
1388 STADJUST(savelen, out);
1389 INTOFF;
1390 ckfree(str);
1391 str = NULL;
1392 INTON;
1393 }
1394 parsebackquote = savepbq;
1395 handler = savehandler;
1396 if (arinest || dblquote)
1397 USTPUTC(CTLBACKQ | CTLQUOTE, out);
1398 else
1399 USTPUTC(CTLBACKQ, out);
1400 if (oldstyle)
1401 goto parsebackq_oldreturn;
1402 else
1403 goto parsebackq_newreturn;
1404 }
1405
1406 /*
1407 * Parse an arithmetic expansion (indicate start of one and set state)
1408 */
1409 parsearith: {
1410
1411 if (++arinest == 1) {
1412 prevsyntax = syntax;
1413 syntax = ARISYNTAX;
1414 USTPUTC(CTLARI, out);
1415 } else {
1416 /*
1417 * we collapse embedded arithmetic expansion to
1418 * parenthesis, which should be equivalent
1419 */
1420 USTPUTC('(', out);
1421 }
1422 goto parsearith_return;
1423 }
1424
1425 } /* end of readtoken */
1426
1427
1428
1429 #ifdef mkinit
1430 RESET {
1431 tokpushback = 0;
1432 checkkwd = 0;
1433 }
1434 #endif
1435
1436 /*
1437 * Returns true if the text contains nothing to expand (no dollar signs
1438 * or backquotes).
1439 */
1440
1441 STATIC int
1442 noexpand(text)
1443 char *text;
1444 {
1445 char *p;
1446 char c;
1447
1448 p = text;
1449 while ((c = *p++) != '\0') {
1450 if (c == CTLESC)
1451 p++;
1452 else if (BASESYNTAX[(int)c] == CCTL)
1453 return 0;
1454 }
1455 return 1;
1456 }
1457
1458
1459 /*
1460 * Return true if the argument is a legal variable name (a letter or
1461 * underscore followed by zero or more letters, underscores, and digits).
1462 */
1463
1464 int
1465 goodname(name)
1466 char *name;
1467 {
1468 char *p;
1469
1470 p = name;
1471 if (! is_name(*p))
1472 return 0;
1473 while (*++p) {
1474 if (! is_in_name(*p))
1475 return 0;
1476 }
1477 return 1;
1478 }
1479
1480
1481 /*
1482 * Called when an unexpected token is read during the parse. The argument
1483 * is the token that is expected, or -1 if more than one type of token can
1484 * occur at this point.
1485 */
1486
1487 STATIC void
1488 synexpect(token)
1489 int token;
1490 {
1491 char msg[64];
1492
1493 if (token >= 0) {
1494 fmtstr(msg, 64, "%s unexpected (expecting %s)",
1495 tokname[lasttoken], tokname[token]);
1496 } else {
1497 fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]);
1498 }
1499 synerror(msg);
1500 /* NOTREACHED */
1501 }
1502
1503
1504 STATIC void
1505 synerror(msg)
1506 char *msg;
1507 {
1508 if (commandname)
1509 outfmt(&errout, "%s: %d: ", commandname, startlinno);
1510 outfmt(&errout, "Syntax error: %s\n", msg);
1511 error((char *)NULL);
1512 /* NOTREACHED */
1513 }
1514
1515 STATIC void
1516 setprompt(which)
1517 int which;
1518 {
1519 whichprompt = which;
1520
1521 #ifndef SMALL
1522 if (!el)
1523 #endif
1524 out2str(getprompt(NULL));
1525 }
1526
1527 /*
1528 * called by editline -- any expansions to the prompt
1529 * should be added here.
1530 */
1531 char *
1532 getprompt(unused)
1533 void *unused;
1534 {
1535 switch (whichprompt) {
1536 case 0:
1537 return "";
1538 case 1:
1539 return ps1val();
1540 case 2:
1541 return ps2val();
1542 default:
1543 return "<internal prompt error>";
1544 }
1545 }
1546