parser.c revision 1.17 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
37 #ifndef lint
38 /*static char sccsid[] = "from: @(#)parser.c 8.1 (Berkeley) 5/31/93";*/
39 static char *rcsid = "$Id: parser.c,v 1.17 1994/06/11 16:12:28 mycroft Exp $";
40 #endif /* not lint */
41
42 #include "shell.h"
43 #include "parser.h"
44 #include "nodes.h"
45 #include "expand.h" /* defines rmescapes() */
46 #include "redir.h" /* defines copyfd() */
47 #include "syntax.h"
48 #include "options.h"
49 #include "input.h"
50 #include "output.h"
51 #include "var.h"
52 #include "error.h"
53 #include "memalloc.h"
54 #include "mystring.h"
55 #include "alias.h"
56 #ifndef NO_HISTORY
57 #include "myhistedit.h"
58 #endif
59
60 /*
61 * Shell command parser.
62 */
63
64 #define EOFMARKLEN 79
65
66 /* values returned by readtoken */
67 #include "token.def"
68
69
70
71 struct heredoc {
72 struct heredoc *next; /* next here document in list */
73 union node *here; /* redirection node */
74 char *eofmark; /* string indicating end of input */
75 int striptabs; /* if set, strip leading tabs */
76 };
77
78
79
80 struct heredoc *heredoclist; /* list of here documents to read */
81 int parsebackquote; /* nonzero if we are inside backquotes */
82 int doprompt; /* if set, prompt the user */
83 int needprompt; /* true if interactive and at start of line */
84 int lasttoken; /* last token read */
85 MKINIT int tokpushback; /* last token pushed back */
86 char *wordtext; /* text of last word returned by readtoken */
87 MKINIT int checkkwd; /* 1 == check for kwds, 2 == also eat newlines */
88 struct nodelist *backquotelist;
89 union node *redirnode;
90 struct heredoc *heredoc;
91 int quoteflag; /* set if (part of) last token was quoted */
92 int startlinno; /* line # where last token started */
93
94
95 #define GDB_HACK 1 /* avoid local declarations which gdb can't handle */
96 #ifdef GDB_HACK
97 static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0'};
98 static const char types[] = "}-+?=";
99 #endif
100
101
102 STATIC union node *list __P((int));
103 STATIC union node *andor __P((void));
104 STATIC union node *pipeline __P((void));
105 STATIC union node *command __P((void));
106 STATIC union node *simplecmd __P((union node **, union node *));
107 STATIC void parsefname __P((void));
108 STATIC void parseheredoc __P((void));
109 STATIC int readtoken __P((void));
110 STATIC int readtoken1 __P((int, char const *, char *, int));
111 STATIC void attyline __P((void));
112 STATIC int noexpand __P((char *));
113 STATIC void synexpect __P((int));
114 STATIC void synerror __P((char *));
115 STATIC void setprompt __P((int));
116
117 /*
118 * Read and parse a command. Returns NEOF on end of file. (NULL is a
119 * valid parse tree indicating a blank line.)
120 */
121
122 union node *
123 parsecmd(interact) {
124 int t;
125
126 doprompt = interact;
127 if (doprompt)
128 setprompt(1);
129 else
130 setprompt(0);
131 needprompt = 0;
132 t = readtoken();
133 if (t == TEOF)
134 return NEOF;
135 if (t == TNL)
136 return NULL;
137 tokpushback++;
138 return list(1);
139 }
140
141
142 STATIC union node *
143 list(nlflag) {
144 union node *n1, *n2, *n3;
145 int tok;
146
147 checkkwd = 2;
148 if (nlflag == 0 && tokendlist[peektoken()])
149 return NULL;
150 n1 = NULL;
151 for (;;) {
152 n2 = andor();
153 tok = readtoken();
154 if (tok == TBACKGND) {
155 if (n2->type == NCMD || n2->type == NPIPE) {
156 n2->ncmd.backgnd = 1;
157 } else if (n2->type == NREDIR) {
158 n2->type = NBACKGND;
159 } else {
160 n3 = (union node *)stalloc(sizeof (struct nredir));
161 n3->type = NBACKGND;
162 n3->nredir.n = n2;
163 n3->nredir.redirect = NULL;
164 n2 = n3;
165 }
166 }
167 if (n1 == NULL) {
168 n1 = n2;
169 }
170 else {
171 n3 = (union node *)stalloc(sizeof (struct nbinary));
172 n3->type = NSEMI;
173 n3->nbinary.ch1 = n1;
174 n3->nbinary.ch2 = n2;
175 n1 = n3;
176 }
177 switch (tok) {
178 case TBACKGND:
179 case TSEMI:
180 tok = readtoken();
181 /* fall through */
182 case TNL:
183 if (tok == TNL) {
184 parseheredoc();
185 if (nlflag)
186 return n1;
187 } else {
188 tokpushback++;
189 }
190 checkkwd = 2;
191 if (tokendlist[peektoken()])
192 return n1;
193 break;
194 case TEOF:
195 if (heredoclist)
196 parseheredoc();
197 else
198 pungetc(); /* push back EOF on input */
199 return n1;
200 default:
201 if (nlflag)
202 synexpect(-1);
203 tokpushback++;
204 return n1;
205 }
206 }
207 }
208
209
210
211 STATIC union node *
212 andor() {
213 union node *n1, *n2, *n3;
214 int t;
215
216 n1 = pipeline();
217 for (;;) {
218 if ((t = readtoken()) == TAND) {
219 t = NAND;
220 } else if (t == TOR) {
221 t = NOR;
222 } else {
223 tokpushback++;
224 return n1;
225 }
226 n2 = pipeline();
227 n3 = (union node *)stalloc(sizeof (struct nbinary));
228 n3->type = t;
229 n3->nbinary.ch1 = n1;
230 n3->nbinary.ch2 = n2;
231 n1 = n3;
232 }
233 }
234
235
236
237 STATIC union node *
238 pipeline() {
239 union node *n1, *pipenode, *notnode;
240 struct nodelist *lp, *prev;
241 int negate = 0;
242
243 TRACE(("pipeline: entered\n"));
244 while (readtoken() == TNOT) {
245 TRACE(("pipeline: TNOT recognized\n"));
246 negate = !negate;
247 }
248 tokpushback++;
249 n1 = command();
250 if (readtoken() == TPIPE) {
251 pipenode = (union node *)stalloc(sizeof (struct npipe));
252 pipenode->type = NPIPE;
253 pipenode->npipe.backgnd = 0;
254 lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
255 pipenode->npipe.cmdlist = lp;
256 lp->n = n1;
257 do {
258 prev = lp;
259 lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
260 lp->n = command();
261 prev->next = lp;
262 } while (readtoken() == TPIPE);
263 lp->next = NULL;
264 n1 = pipenode;
265 }
266 tokpushback++;
267 if (negate) {
268 notnode = (union node *)stalloc(sizeof (struct nnot));
269 notnode->type = NNOT;
270 notnode->nnot.com = n1;
271 n1 = notnode;
272 }
273 return n1;
274 }
275
276
277
278 STATIC union node *
279 command() {
280 union node *n1, *n2;
281 union node *ap, **app;
282 union node *cp, **cpp;
283 union node *redir, **rpp;
284 int t;
285
286 checkkwd = 2;
287 redir = 0;
288 rpp = &redir;
289 /* Check for redirection which may precede command */
290 while (readtoken() == TREDIR) {
291 *rpp = n2 = redirnode;
292 rpp = &n2->nfile.next;
293 parsefname();
294 }
295 tokpushback++;
296
297 switch (readtoken()) {
298 case TIF:
299 n1 = (union node *)stalloc(sizeof (struct nif));
300 n1->type = NIF;
301 n1->nif.test = list(0);
302 if (readtoken() != TTHEN)
303 synexpect(TTHEN);
304 n1->nif.ifpart = list(0);
305 n2 = n1;
306 while (readtoken() == TELIF) {
307 n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif));
308 n2 = n2->nif.elsepart;
309 n2->type = NIF;
310 n2->nif.test = list(0);
311 if (readtoken() != TTHEN)
312 synexpect(TTHEN);
313 n2->nif.ifpart = list(0);
314 }
315 if (lasttoken == TELSE)
316 n2->nif.elsepart = list(0);
317 else {
318 n2->nif.elsepart = NULL;
319 tokpushback++;
320 }
321 if (readtoken() != TFI)
322 synexpect(TFI);
323 checkkwd = 1;
324 break;
325 case TWHILE:
326 case TUNTIL: {
327 int got;
328 n1 = (union node *)stalloc(sizeof (struct nbinary));
329 n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL;
330 n1->nbinary.ch1 = list(0);
331 if ((got=readtoken()) != TDO) {
332 TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : ""));
333 synexpect(TDO);
334 }
335 n1->nbinary.ch2 = list(0);
336 if (readtoken() != TDONE)
337 synexpect(TDONE);
338 checkkwd = 1;
339 break;
340 }
341 case TFOR:
342 if (readtoken() != TWORD || quoteflag || ! goodname(wordtext))
343 synerror("Bad for loop variable");
344 n1 = (union node *)stalloc(sizeof (struct nfor));
345 n1->type = NFOR;
346 n1->nfor.var = wordtext;
347 if (readtoken() == TWORD && ! quoteflag && equal(wordtext, "in")) {
348 app = ≈
349 while (readtoken() == TWORD) {
350 n2 = (union node *)stalloc(sizeof (struct narg));
351 n2->type = NARG;
352 n2->narg.text = wordtext;
353 n2->narg.backquote = backquotelist;
354 *app = n2;
355 app = &n2->narg.next;
356 }
357 *app = NULL;
358 n1->nfor.args = ap;
359 if (lasttoken != TNL && lasttoken != TSEMI)
360 synexpect(-1);
361 } else {
362 #ifndef GDB_HACK
363 static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE,
364 '@', '=', '\0'};
365 #endif
366 n2 = (union node *)stalloc(sizeof (struct narg));
367 n2->type = NARG;
368 n2->narg.text = (char *)argvars;
369 n2->narg.backquote = NULL;
370 n2->narg.next = NULL;
371 n1->nfor.args = n2;
372 /*
373 * Newline or semicolon here is optional (but note
374 * that the original Bourne shell only allowed NL).
375 */
376 if (lasttoken != TNL && lasttoken != TSEMI)
377 tokpushback++;
378 }
379 checkkwd = 2;
380 if ((t = readtoken()) == TDO)
381 t = TDONE;
382 else if (t == TBEGIN)
383 t = TEND;
384 else
385 synexpect(-1);
386 n1->nfor.body = list(0);
387 if (readtoken() != t)
388 synexpect(t);
389 checkkwd = 1;
390 break;
391 case TCASE:
392 n1 = (union node *)stalloc(sizeof (struct ncase));
393 n1->type = NCASE;
394 if (readtoken() != TWORD)
395 synexpect(TWORD);
396 n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg));
397 n2->type = NARG;
398 n2->narg.text = wordtext;
399 n2->narg.backquote = backquotelist;
400 n2->narg.next = NULL;
401 while (readtoken() == TNL);
402 if (lasttoken != TWORD || ! equal(wordtext, "in"))
403 synerror("expecting \"in\"");
404 cpp = &n1->ncase.cases;
405 checkkwd = 2, readtoken();
406 do {
407 *cpp = cp = (union node *)stalloc(sizeof (struct nclist));
408 cp->type = NCLIST;
409 app = &cp->nclist.pattern;
410 for (;;) {
411 *app = ap = (union node *)stalloc(sizeof (struct narg));
412 ap->type = NARG;
413 ap->narg.text = wordtext;
414 ap->narg.backquote = backquotelist;
415 if (checkkwd = 2, readtoken() != TPIPE)
416 break;
417 app = &ap->narg.next;
418 readtoken();
419 }
420 ap->narg.next = NULL;
421 if (lasttoken != TRP)
422 synexpect(TRP);
423 cp->nclist.body = list(0);
424
425 checkkwd = 2;
426 if ((t = readtoken()) != TESAC) {
427 if (t != TENDCASE)
428 synexpect(TENDCASE);
429 else
430 checkkwd = 2, readtoken();
431 }
432 cpp = &cp->nclist.next;
433 } while(lasttoken != TESAC);
434 *cpp = NULL;
435 checkkwd = 1;
436 break;
437 case TLP:
438 n1 = (union node *)stalloc(sizeof (struct nredir));
439 n1->type = NSUBSHELL;
440 n1->nredir.n = list(0);
441 n1->nredir.redirect = NULL;
442 if (readtoken() != TRP)
443 synexpect(TRP);
444 checkkwd = 1;
445 break;
446 case TBEGIN:
447 n1 = list(0);
448 if (readtoken() != TEND)
449 synexpect(TEND);
450 checkkwd = 1;
451 break;
452 /* Handle an empty command like other simple commands. */
453 case TNL:
454 case TWORD:
455 tokpushback++;
456 return simplecmd(rpp, redir);
457 default:
458 synexpect(-1);
459 }
460
461 /* Now check for redirection which may follow command */
462 while (readtoken() == TREDIR) {
463 *rpp = n2 = redirnode;
464 rpp = &n2->nfile.next;
465 parsefname();
466 }
467 tokpushback++;
468 *rpp = NULL;
469 if (redir) {
470 if (n1->type != NSUBSHELL) {
471 n2 = (union node *)stalloc(sizeof (struct nredir));
472 n2->type = NREDIR;
473 n2->nredir.n = n1;
474 n1 = n2;
475 }
476 n1->nredir.redirect = redir;
477 }
478 return n1;
479 }
480
481
482 STATIC union node *
483 simplecmd(rpp, redir)
484 union node **rpp, *redir;
485 {
486 union node *args, **app;
487 union node **orig_rpp = rpp;
488 union node *n;
489
490 /* If we don't have any redirections already, then we must reset */
491 /* rpp to be the address of the local redir variable. */
492 if (redir == 0)
493 rpp = &redir;
494
495 args = NULL;
496 app = &args;
497 /*
498 * We save the incoming value, because we need this for shell
499 * functions. There can not be a redirect or an argument between
500 * the function name and the open parenthesis.
501 */
502 orig_rpp = rpp;
503
504 for (;;) {
505 if (readtoken() == TWORD) {
506 n = (union node *)stalloc(sizeof (struct narg));
507 n->type = NARG;
508 n->narg.text = wordtext;
509 n->narg.backquote = backquotelist;
510 *app = n;
511 app = &n->narg.next;
512 } else if (lasttoken == TREDIR) {
513 *rpp = n = redirnode;
514 rpp = &n->nfile.next;
515 parsefname(); /* read name of redirection file */
516 } else if (lasttoken == TLP && app == &args->narg.next
517 && rpp == orig_rpp) {
518 /* We have a function */
519 if (readtoken() != TRP)
520 synexpect(TRP);
521 #ifdef notdef
522 if (! goodname(n->narg.text))
523 synerror("Bad function name");
524 #endif
525 n->type = NDEFUN;
526 n->narg.next = command();
527 return n;
528 } else {
529 tokpushback++;
530 break;
531 }
532 }
533 *app = NULL;
534 *rpp = NULL;
535 n = (union node *)stalloc(sizeof (struct ncmd));
536 n->type = NCMD;
537 n->ncmd.backgnd = 0;
538 n->ncmd.args = args;
539 n->ncmd.redirect = redir;
540 return n;
541 }
542
543
544 STATIC void
545 parsefname() {
546 union node *n = redirnode;
547
548 if (readtoken() != TWORD)
549 synexpect(-1);
550 if (n->type == NHERE) {
551 struct heredoc *here = heredoc;
552 struct heredoc *p;
553 int i;
554
555 if (quoteflag == 0)
556 n->type = NXHERE;
557 TRACE(("Here document %d\n", n->type));
558 if (here->striptabs) {
559 while (*wordtext == '\t')
560 wordtext++;
561 }
562 if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN)
563 synerror("Illegal eof marker for << redirection");
564 rmescapes(wordtext);
565 here->eofmark = wordtext;
566 here->next = NULL;
567 if (heredoclist == NULL)
568 heredoclist = here;
569 else {
570 for (p = heredoclist ; p->next ; p = p->next);
571 p->next = here;
572 }
573 } else if (n->type == NTOFD || n->type == NFROMFD) {
574 if (is_digit(wordtext[0]))
575 n->ndup.dupfd = digit_val(wordtext[0]);
576 else if (wordtext[0] == '-')
577 n->ndup.dupfd = -1;
578 else
579 goto bad;
580 if (wordtext[1] != '\0') {
581 bad:
582 synerror("Bad fd number");
583 }
584 } else {
585 n->nfile.fname = (union node *)stalloc(sizeof (struct narg));
586 n = n->nfile.fname;
587 n->type = NARG;
588 n->narg.next = NULL;
589 n->narg.text = wordtext;
590 n->narg.backquote = backquotelist;
591 }
592 }
593
594
595 /*
596 * Input any here documents.
597 */
598
599 STATIC void
600 parseheredoc() {
601 struct heredoc *here;
602 union node *n;
603
604 while (heredoclist) {
605 here = heredoclist;
606 heredoclist = here->next;
607 if (needprompt) {
608 setprompt(2);
609 needprompt = 0;
610 }
611 readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX,
612 here->eofmark, here->striptabs);
613 n = (union node *)stalloc(sizeof (struct narg));
614 n->narg.type = NARG;
615 n->narg.next = NULL;
616 n->narg.text = wordtext;
617 n->narg.backquote = backquotelist;
618 here->here->nhere.doc = n;
619 }
620 }
621
622 STATIC int
623 peektoken() {
624 int t;
625
626 t = readtoken();
627 tokpushback++;
628 return (t);
629 }
630
631 STATIC int xxreadtoken();
632
633 STATIC int
634 readtoken() {
635 int t;
636 int savecheckkwd = checkkwd;
637 struct alias *ap;
638 #ifdef DEBUG
639 int alreadyseen = tokpushback;
640 #endif
641
642 top:
643 t = xxreadtoken();
644
645 if (checkkwd) {
646 /*
647 * eat newlines
648 */
649 if (checkkwd == 2) {
650 checkkwd = 0;
651 while (t == TNL) {
652 parseheredoc();
653 t = xxreadtoken();
654 }
655 } else
656 checkkwd = 0;
657 /*
658 * check for keywords and aliases
659 */
660 if (t == TWORD && !quoteflag) {
661 register char * const *pp, *s;
662
663 for (pp = (char **)parsekwd; *pp; pp++) {
664 if (**pp == *wordtext && equal(*pp, wordtext)) {
665 lasttoken = t = pp - parsekwd + KWDOFFSET;
666 TRACE(("keyword %s recognized\n", tokname[t]));
667 goto out;
668 }
669 }
670 if (ap = lookupalias(wordtext, 1)) {
671 pushstring(ap->val, strlen(ap->val), ap);
672 checkkwd = savecheckkwd;
673 goto top;
674 }
675 }
676 out:
677 checkkwd = 0;
678 }
679 #ifdef DEBUG
680 if (!alreadyseen)
681 TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
682 else
683 TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
684 #endif
685 return (t);
686 }
687
688
689 /*
690 * Read the next input token.
691 * If the token is a word, we set backquotelist to the list of cmds in
692 * backquotes. We set quoteflag to true if any part of the word was
693 * quoted.
694 * If the token is TREDIR, then we set redirnode to a structure containing
695 * the redirection.
696 * In all cases, the variable startlinno is set to the number of the line
697 * on which the token starts.
698 *
699 * [Change comment: here documents and internal procedures]
700 * [Readtoken shouldn't have any arguments. Perhaps we should make the
701 * word parsing code into a separate routine. In this case, readtoken
702 * doesn't need to have any internal procedures, but parseword does.
703 * We could also make parseoperator in essence the main routine, and
704 * have parseword (readtoken1?) handle both words and redirection.]
705 */
706
707 #define RETURN(token) return lasttoken = token
708
709 STATIC int
710 xxreadtoken() {
711 register c;
712
713 if (tokpushback) {
714 tokpushback = 0;
715 return lasttoken;
716 }
717 if (needprompt) {
718 setprompt(2);
719 needprompt = 0;
720 }
721 startlinno = plinno;
722 for (;;) { /* until token or start of word found */
723 c = pgetc_macro();
724 if (c == ' ' || c == '\t')
725 continue; /* quick check for white space first */
726 switch (c) {
727 case ' ': case '\t':
728 continue;
729 case '#':
730 while ((c = pgetc()) != '\n' && c != PEOF);
731 pungetc();
732 continue;
733 case '\\':
734 if (pgetc() == '\n') {
735 startlinno = ++plinno;
736 if (doprompt)
737 setprompt(2);
738 else
739 setprompt(0);
740 continue;
741 }
742 pungetc();
743 goto breakloop;
744 case '\n':
745 plinno++;
746 needprompt = doprompt;
747 RETURN(TNL);
748 case PEOF:
749 RETURN(TEOF);
750 case '&':
751 if (pgetc() == '&')
752 RETURN(TAND);
753 pungetc();
754 RETURN(TBACKGND);
755 case '|':
756 if (pgetc() == '|')
757 RETURN(TOR);
758 pungetc();
759 RETURN(TPIPE);
760 case ';':
761 if (pgetc() == ';')
762 RETURN(TENDCASE);
763 pungetc();
764 RETURN(TSEMI);
765 case '(':
766 RETURN(TLP);
767 case ')':
768 RETURN(TRP);
769 default:
770 goto breakloop;
771 }
772 }
773 breakloop:
774 return readtoken1(c, BASESYNTAX, (char *)NULL, 0);
775 #undef RETURN
776 }
777
778
779
780 /*
781 * If eofmark is NULL, read a word or a redirection symbol. If eofmark
782 * is not NULL, read a here document. In the latter case, eofmark is the
783 * word which marks the end of the document and striptabs is true if
784 * leading tabs should be stripped from the document. The argument firstc
785 * is the first character of the input token or document.
786 *
787 * Because C does not have internal subroutines, I have simulated them
788 * using goto's to implement the subroutine linkage. The following macros
789 * will run code that appears at the end of readtoken1.
790 */
791
792 #define CHECKEND() {goto checkend; checkend_return:;}
793 #define PARSEREDIR() {goto parseredir; parseredir_return:;}
794 #define PARSESUB() {goto parsesub; parsesub_return:;}
795 #define PARSEBACKQOLD() {oldstyle = 1; goto parsebackq; parsebackq_oldreturn:;}
796 #define PARSEBACKQNEW() {oldstyle = 0; goto parsebackq; parsebackq_newreturn:;}
797 #define PARSEARITH() {goto parsearith; parsearith_return:;}
798
799 STATIC int
800 readtoken1(firstc, syntax, eofmark, striptabs)
801 int firstc;
802 char const *syntax;
803 char *eofmark;
804 int striptabs;
805 {
806 register c = firstc;
807 register char *out;
808 int len;
809 char line[EOFMARKLEN + 1];
810 struct nodelist *bqlist;
811 int quotef;
812 int dblquote;
813 int varnest; /* levels of variables expansion */
814 int arinest; /* levels of arithmetic expansion */
815 int parenlevel; /* levels of parens in arithmetic */
816 int oldstyle;
817 char const *prevsyntax; /* syntax before arithmetic */
818
819 startlinno = plinno;
820 dblquote = 0;
821 if (syntax == DQSYNTAX)
822 dblquote = 1;
823 quotef = 0;
824 bqlist = NULL;
825 varnest = 0;
826 arinest = 0;
827 parenlevel = 0;
828
829 STARTSTACKSTR(out);
830 loop: { /* for each line, until end of word */
831 #if ATTY
832 if (c == '\034' && doprompt
833 && attyset() && ! equal(termval(), "emacs")) {
834 attyline();
835 if (syntax == BASESYNTAX)
836 return readtoken();
837 c = pgetc();
838 goto loop;
839 }
840 #endif
841 CHECKEND(); /* set c to PEOF if at end of here document */
842 for (;;) { /* until end of line or end of word */
843 CHECKSTRSPACE(3, out); /* permit 3 calls to USTPUTC */
844 switch(syntax[c]) {
845 case CNL: /* '\n' */
846 if (syntax == BASESYNTAX)
847 goto endword; /* exit outer loop */
848 USTPUTC(c, out);
849 plinno++;
850 if (doprompt)
851 setprompt(2);
852 else
853 setprompt(0);
854 c = pgetc();
855 goto loop; /* continue outer loop */
856 case CWORD:
857 USTPUTC(c, out);
858 break;
859 case CCTL:
860 if (eofmark == NULL || dblquote)
861 USTPUTC(CTLESC, out);
862 USTPUTC(c, out);
863 break;
864 case CBACK: /* backslash */
865 c = pgetc();
866 if (c == PEOF) {
867 USTPUTC('\\', out);
868 pungetc();
869 } else if (c == '\n') {
870 if (doprompt)
871 setprompt(2);
872 else
873 setprompt(0);
874 } else {
875 if (dblquote && c != '\\' && c != '`' && c != '$'
876 && (c != '"' || eofmark != NULL))
877 USTPUTC('\\', out);
878 if (SQSYNTAX[c] == CCTL)
879 USTPUTC(CTLESC, out);
880 USTPUTC(c, out);
881 quotef++;
882 }
883 break;
884 case CSQUOTE:
885 syntax = SQSYNTAX;
886 break;
887 case CDQUOTE:
888 syntax = DQSYNTAX;
889 dblquote = 1;
890 break;
891 case CENDQUOTE:
892 if (eofmark) {
893 USTPUTC(c, out);
894 } else {
895 if (arinest)
896 syntax = ARISYNTAX;
897 else
898 syntax = BASESYNTAX;
899 quotef++;
900 dblquote = 0;
901 }
902 break;
903 case CVAR: /* '$' */
904 PARSESUB(); /* parse substitution */
905 break;
906 case CENDVAR: /* '}' */
907 if (varnest > 0) {
908 varnest--;
909 USTPUTC(CTLENDVAR, out);
910 } else {
911 USTPUTC(c, out);
912 }
913 break;
914 case CLP: /* '(' in arithmetic */
915 parenlevel++;
916 USTPUTC(c, out);
917 break;
918 case CRP: /* ')' in arithmetic */
919 if (parenlevel > 0) {
920 USTPUTC(c, out);
921 --parenlevel;
922 } else {
923 if (pgetc() == ')') {
924 if (--arinest == 0) {
925 USTPUTC(CTLENDARI, out);
926 syntax = prevsyntax;
927 } else
928 USTPUTC(')', out);
929 } else {
930 /*
931 * unbalanced parens
932 * (don't 2nd guess - no error)
933 */
934 pungetc();
935 USTPUTC(')', out);
936 }
937 }
938 break;
939 case CBQUOTE: /* '`' */
940 PARSEBACKQOLD();
941 break;
942 case CEOF:
943 goto endword; /* exit outer loop */
944 default:
945 if (varnest == 0)
946 goto endword; /* exit outer loop */
947 USTPUTC(c, out);
948 }
949 c = pgetc_macro();
950 }
951 }
952 endword:
953 if (syntax == ARISYNTAX)
954 synerror("Missing '))'");
955 if (syntax != BASESYNTAX && ! parsebackquote && eofmark == NULL)
956 synerror("Unterminated quoted string");
957 if (varnest != 0) {
958 startlinno = plinno;
959 synerror("Missing '}'");
960 }
961 USTPUTC('\0', out);
962 len = out - stackblock();
963 out = stackblock();
964 if (eofmark == NULL) {
965 if ((c == '>' || c == '<')
966 && quotef == 0
967 && len <= 2
968 && (*out == '\0' || is_digit(*out))) {
969 PARSEREDIR();
970 return lasttoken = TREDIR;
971 } else {
972 pungetc();
973 }
974 }
975 quoteflag = quotef;
976 backquotelist = bqlist;
977 grabstackblock(len);
978 wordtext = out;
979 return lasttoken = TWORD;
980 /* end of readtoken routine */
981
982
983
984 /*
985 * Check to see whether we are at the end of the here document. When this
986 * is called, c is set to the first character of the next input line. If
987 * we are at the end of the here document, this routine sets the c to PEOF.
988 */
989
990 checkend: {
991 if (eofmark) {
992 if (striptabs) {
993 while (c == '\t')
994 c = pgetc();
995 }
996 if (c == *eofmark) {
997 if (pfgets(line, sizeof line) != NULL) {
998 register char *p, *q;
999
1000 p = line;
1001 for (q = eofmark + 1 ; *q && *p == *q ; p++, q++);
1002 if (*p == '\n' && *q == '\0') {
1003 c = PEOF;
1004 plinno++;
1005 needprompt = doprompt;
1006 } else {
1007 pushstring(line, strlen(line), NULL);
1008 }
1009 }
1010 }
1011 }
1012 goto checkend_return;
1013 }
1014
1015
1016 /*
1017 * Parse a redirection operator. The variable "out" points to a string
1018 * specifying the fd to be redirected. The variable "c" contains the
1019 * first character of the redirection operator.
1020 */
1021
1022 parseredir: {
1023 char fd = *out;
1024 union node *np;
1025
1026 np = (union node *)stalloc(sizeof (struct nfile));
1027 if (c == '>') {
1028 np->nfile.fd = 1;
1029 c = pgetc();
1030 if (c == '>')
1031 np->type = NAPPEND;
1032 else if (c == '&')
1033 np->type = NTOFD;
1034 else {
1035 np->type = NTO;
1036 pungetc();
1037 }
1038 } else { /* c == '<' */
1039 np->nfile.fd = 0;
1040 c = pgetc();
1041 if (c == '<') {
1042 if (sizeof (struct nfile) != sizeof (struct nhere)) {
1043 np = (union node *)stalloc(sizeof (struct nhere));
1044 np->nfile.fd = 0;
1045 }
1046 np->type = NHERE;
1047 heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc));
1048 heredoc->here = np;
1049 if ((c = pgetc()) == '-') {
1050 heredoc->striptabs = 1;
1051 } else {
1052 heredoc->striptabs = 0;
1053 pungetc();
1054 }
1055 } else if (c == '&')
1056 np->type = NFROMFD;
1057 else {
1058 np->type = NFROM;
1059 pungetc();
1060 }
1061 }
1062 if (fd != '\0')
1063 np->nfile.fd = digit_val(fd);
1064 redirnode = np;
1065 goto parseredir_return;
1066 }
1067
1068
1069 /*
1070 * Parse a substitution. At this point, we have read the dollar sign
1071 * and nothing else.
1072 */
1073
1074 parsesub: {
1075 int subtype;
1076 int typeloc;
1077 int flags;
1078 char *p;
1079 #ifndef GDB_HACK
1080 static const char types[] = "}-+?=";
1081 #endif
1082
1083 c = pgetc();
1084 if (c != '(' && c != '{' && !is_name(c) && !is_special(c)) {
1085 USTPUTC('$', out);
1086 pungetc();
1087 } else if (c == '(') { /* $(command) or $((arith)) */
1088 if (pgetc() == '(') {
1089 PARSEARITH();
1090 } else {
1091 pungetc();
1092 PARSEBACKQNEW();
1093 }
1094 } else {
1095 USTPUTC(CTLVAR, out);
1096 typeloc = out - stackblock();
1097 USTPUTC(VSNORMAL, out);
1098 subtype = VSNORMAL;
1099 if (c == '{') {
1100 c = pgetc();
1101 subtype = 0;
1102 }
1103 if (is_name(c)) {
1104 do {
1105 STPUTC(c, out);
1106 c = pgetc();
1107 } while (is_in_name(c));
1108 } else {
1109 if (! is_special(c))
1110 badsub: synerror("Bad substitution");
1111 USTPUTC(c, out);
1112 c = pgetc();
1113 }
1114 STPUTC('=', out);
1115 flags = 0;
1116 if (subtype == 0) {
1117 if (c == ':') {
1118 flags = VSNUL;
1119 c = pgetc();
1120 }
1121 p = strchr(types, c);
1122 if (p == NULL)
1123 goto badsub;
1124 subtype = p - types + VSNORMAL;
1125 } else {
1126 pungetc();
1127 }
1128 if (dblquote || arinest)
1129 flags |= VSQUOTE;
1130 *(stackblock() + typeloc) = subtype | flags;
1131 if (subtype != VSNORMAL)
1132 varnest++;
1133 }
1134 goto parsesub_return;
1135 }
1136
1137
1138 /*
1139 * Called to parse command substitutions. Newstyle is set if the command
1140 * is enclosed inside $(...); nlpp is a pointer to the head of the linked
1141 * list of commands (passed by reference), and savelen is the number of
1142 * characters on the top of the stack which must be preserved.
1143 */
1144
1145 parsebackq: {
1146 struct nodelist **nlpp;
1147 int savepbq;
1148 union node *n;
1149 char *volatile str;
1150 struct jmploc jmploc;
1151 struct jmploc *volatile savehandler;
1152 int savelen;
1153
1154 savepbq = parsebackquote;
1155 if (setjmp(jmploc.loc)) {
1156 if (str)
1157 ckfree(str);
1158 parsebackquote = 0;
1159 handler = savehandler;
1160 longjmp(handler->loc, 1);
1161 }
1162 INTOFF;
1163 str = NULL;
1164 savelen = out - stackblock();
1165 if (savelen > 0) {
1166 str = ckmalloc(savelen);
1167 bcopy(stackblock(), str, savelen);
1168 }
1169 savehandler = handler;
1170 handler = &jmploc;
1171 INTON;
1172 if (oldstyle) {
1173 /* We must read until the closing backquote, giving special
1174 treatment to some slashes, and then push the string and
1175 reread it as input, interpreting it normally. */
1176 register char *out;
1177 register c;
1178 int savelen;
1179 char *str;
1180
1181 STARTSTACKSTR(out);
1182 while ((c = pgetc ()) != '`') {
1183 if (c == '\\') {
1184 c = pgetc ();
1185 if (c != '\\' && c != '`' && c != '$'
1186 && (!dblquote || c != '"'))
1187 STPUTC('\\', out);
1188 }
1189 STPUTC(c, out);
1190 }
1191 STPUTC('\0', out);
1192 savelen = out - stackblock();
1193 if (savelen > 0) {
1194 str = ckmalloc(savelen);
1195 bcopy(stackblock(), str, savelen);
1196 }
1197 setinputstring(str, 1);
1198 }
1199 nlpp = &bqlist;
1200 while (*nlpp)
1201 nlpp = &(*nlpp)->next;
1202 *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist));
1203 (*nlpp)->next = NULL;
1204 parsebackquote = oldstyle;
1205 n = list(0);
1206 if (!oldstyle && (readtoken() != TRP))
1207 synexpect(TRP);
1208 (*nlpp)->n = n;
1209 /* Start reading from old file again. */
1210 if (oldstyle)
1211 popfile();
1212 while (stackblocksize() <= savelen)
1213 growstackblock();
1214 STARTSTACKSTR(out);
1215 if (str) {
1216 bcopy(str, out, savelen);
1217 STADJUST(savelen, out);
1218 INTOFF;
1219 ckfree(str);
1220 str = NULL;
1221 INTON;
1222 }
1223 parsebackquote = savepbq;
1224 handler = savehandler;
1225 if (arinest || dblquote)
1226 USTPUTC(CTLBACKQ | CTLQUOTE, out);
1227 else
1228 USTPUTC(CTLBACKQ, out);
1229 if (oldstyle)
1230 goto parsebackq_oldreturn;
1231 else
1232 goto parsebackq_newreturn;
1233 }
1234
1235 /*
1236 * Parse an arithmetic expansion (indicate start of one and set state)
1237 */
1238 parsearith: {
1239
1240 if (++arinest == 1) {
1241 prevsyntax = syntax;
1242 syntax = ARISYNTAX;
1243 USTPUTC(CTLARI, out);
1244 } else {
1245 /*
1246 * we collapse embedded arithmetic expansion to
1247 * parenthesis, which should be equivalent
1248 */
1249 USTPUTC('(', out);
1250 }
1251 goto parsearith_return;
1252 }
1253
1254 } /* end of readtoken */
1255
1256
1257
1258 #ifdef mkinit
1259 RESET {
1260 tokpushback = 0;
1261 checkkwd = 0;
1262 }
1263 #endif
1264
1265 /*
1266 * Returns true if the text contains nothing to expand (no dollar signs
1267 * or backquotes).
1268 */
1269
1270 STATIC int
1271 noexpand(text)
1272 char *text;
1273 {
1274 register char *p;
1275 register char c;
1276
1277 p = text;
1278 while ((c = *p++) != '\0') {
1279 if (c == CTLESC)
1280 p++;
1281 else if (BASESYNTAX[c] == CCTL)
1282 return 0;
1283 }
1284 return 1;
1285 }
1286
1287
1288 /*
1289 * Return true if the argument is a legal variable name (a letter or
1290 * underscore followed by zero or more letters, underscores, and digits).
1291 */
1292
1293 int
1294 goodname(name)
1295 char *name;
1296 {
1297 register char *p;
1298
1299 p = name;
1300 if (! is_name(*p))
1301 return 0;
1302 while (*++p) {
1303 if (! is_in_name(*p))
1304 return 0;
1305 }
1306 return 1;
1307 }
1308
1309
1310 /*
1311 * Called when an unexpected token is read during the parse. The argument
1312 * is the token that is expected, or -1 if more than one type of token can
1313 * occur at this point.
1314 */
1315
1316 STATIC void
1317 synexpect(token) {
1318 char msg[64];
1319
1320 if (token >= 0) {
1321 fmtstr(msg, 64, "%s unexpected (expecting %s)",
1322 tokname[lasttoken], tokname[token]);
1323 } else {
1324 fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]);
1325 }
1326 synerror(msg);
1327 }
1328
1329
1330 STATIC void
1331 synerror(msg)
1332 char *msg;
1333 {
1334 if (commandname)
1335 outfmt(&errout, "%s: %d: ", commandname, startlinno);
1336 outfmt(&errout, "Syntax error: %s\n", msg);
1337 error((char *)NULL);
1338 }
1339
1340 STATIC void
1341 setprompt(which)
1342 int which;
1343 {
1344 whichprompt = which;
1345
1346 #ifndef NO_HISTORY
1347 if (!el)
1348 #endif
1349 out2str(getprompt(NULL));
1350 }
1351
1352 /*
1353 * called by editline -- any expansions to the prompt
1354 * should be added here.
1355 */
1356 char *
1357 getprompt(unused)
1358 void *unused;
1359 {
1360 switch (whichprompt) {
1361 case 0:
1362 return "";
1363 case 1:
1364 return ps1val();
1365 case 2:
1366 return ps2val();
1367 default:
1368 return "<internal prompt error>";
1369 }
1370 }
1371