parser.c revision 1.18 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 1.18 1994/06/14 05:49:25 jtc 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 union node *makename __P((void));
108 STATIC void parsefname __P((void));
109 STATIC void parseheredoc __P((void));
110 STATIC int readtoken __P((void));
111 STATIC int readtoken1 __P((int, char const *, char *, int));
112 STATIC void attyline __P((void));
113 STATIC int noexpand __P((char *));
114 STATIC void synexpect __P((int));
115 STATIC void synerror __P((char *));
116 STATIC void setprompt __P((int));
117
118 /*
119 * Read and parse a command. Returns NEOF on end of file. (NULL is a
120 * valid parse tree indicating a blank line.)
121 */
122
123 union node *
124 parsecmd(interact) {
125 int t;
126
127 doprompt = interact;
128 if (doprompt)
129 setprompt(1);
130 else
131 setprompt(0);
132 needprompt = 0;
133 t = readtoken();
134 if (t == TEOF)
135 return NEOF;
136 if (t == TNL)
137 return NULL;
138 tokpushback++;
139 return list(1);
140 }
141
142
143 STATIC union node *
144 list(nlflag) {
145 union node *n1, *n2, *n3;
146 int tok;
147
148 checkkwd = 2;
149 if (nlflag == 0 && tokendlist[peektoken()])
150 return NULL;
151 n1 = NULL;
152 for (;;) {
153 n2 = andor();
154 tok = readtoken();
155 if (tok == TBACKGND) {
156 if (n2->type == NCMD || n2->type == NPIPE) {
157 n2->ncmd.backgnd = 1;
158 } else if (n2->type == NREDIR) {
159 n2->type = NBACKGND;
160 } else {
161 n3 = (union node *)stalloc(sizeof (struct nredir));
162 n3->type = NBACKGND;
163 n3->nredir.n = n2;
164 n3->nredir.redirect = NULL;
165 n2 = n3;
166 }
167 }
168 if (n1 == NULL) {
169 n1 = n2;
170 }
171 else {
172 n3 = (union node *)stalloc(sizeof (struct nbinary));
173 n3->type = NSEMI;
174 n3->nbinary.ch1 = n1;
175 n3->nbinary.ch2 = n2;
176 n1 = n3;
177 }
178 switch (tok) {
179 case TBACKGND:
180 case TSEMI:
181 tok = readtoken();
182 /* fall through */
183 case TNL:
184 if (tok == TNL) {
185 parseheredoc();
186 if (nlflag)
187 return n1;
188 } else {
189 tokpushback++;
190 }
191 checkkwd = 2;
192 if (tokendlist[peektoken()])
193 return n1;
194 break;
195 case TEOF:
196 if (heredoclist)
197 parseheredoc();
198 else
199 pungetc(); /* push back EOF on input */
200 return n1;
201 default:
202 if (nlflag)
203 synexpect(-1);
204 tokpushback++;
205 return n1;
206 }
207 }
208 }
209
210
211
212 STATIC union node *
213 andor() {
214 union node *n1, *n2, *n3;
215 int t;
216
217 n1 = pipeline();
218 for (;;) {
219 if ((t = readtoken()) == TAND) {
220 t = NAND;
221 } else if (t == TOR) {
222 t = NOR;
223 } else {
224 tokpushback++;
225 return n1;
226 }
227 n2 = pipeline();
228 n3 = (union node *)stalloc(sizeof (struct nbinary));
229 n3->type = t;
230 n3->nbinary.ch1 = n1;
231 n3->nbinary.ch2 = n2;
232 n1 = n3;
233 }
234 }
235
236
237
238 STATIC union node *
239 pipeline() {
240 union node *n1, *pipenode, *notnode;
241 struct nodelist *lp, *prev;
242 int negate = 0;
243
244 TRACE(("pipeline: entered\n"));
245 while (readtoken() == TNOT) {
246 TRACE(("pipeline: TNOT recognized\n"));
247 negate = !negate;
248 }
249 tokpushback++;
250 n1 = command();
251 if (readtoken() == TPIPE) {
252 pipenode = (union node *)stalloc(sizeof (struct npipe));
253 pipenode->type = NPIPE;
254 pipenode->npipe.backgnd = 0;
255 lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
256 pipenode->npipe.cmdlist = lp;
257 lp->n = n1;
258 do {
259 prev = lp;
260 lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
261 lp->n = command();
262 prev->next = lp;
263 } while (readtoken() == TPIPE);
264 lp->next = NULL;
265 n1 = pipenode;
266 }
267 tokpushback++;
268 if (negate) {
269 notnode = (union node *)stalloc(sizeof (struct nnot));
270 notnode->type = NNOT;
271 notnode->nnot.com = n1;
272 n1 = notnode;
273 }
274 return n1;
275 }
276
277
278
279 STATIC union node *
280 command() {
281 union node *n1, *n2;
282 union node *ap, **app;
283 union node *cp, **cpp;
284 union node *redir, **rpp;
285 int t;
286
287 checkkwd = 2;
288 redir = 0;
289 rpp = &redir;
290 /* Check for redirection which may precede command */
291 while (readtoken() == TREDIR) {
292 *rpp = n2 = redirnode;
293 rpp = &n2->nfile.next;
294 parsefname();
295 }
296 tokpushback++;
297
298 switch (readtoken()) {
299 case TIF:
300 n1 = (union node *)stalloc(sizeof (struct nif));
301 n1->type = NIF;
302 n1->nif.test = list(0);
303 if (readtoken() != TTHEN)
304 synexpect(TTHEN);
305 n1->nif.ifpart = list(0);
306 n2 = n1;
307 while (readtoken() == TELIF) {
308 n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif));
309 n2 = n2->nif.elsepart;
310 n2->type = NIF;
311 n2->nif.test = list(0);
312 if (readtoken() != TTHEN)
313 synexpect(TTHEN);
314 n2->nif.ifpart = list(0);
315 }
316 if (lasttoken == TELSE)
317 n2->nif.elsepart = list(0);
318 else {
319 n2->nif.elsepart = NULL;
320 tokpushback++;
321 }
322 if (readtoken() != TFI)
323 synexpect(TFI);
324 checkkwd = 1;
325 break;
326 case TWHILE:
327 case TUNTIL: {
328 int got;
329 n1 = (union node *)stalloc(sizeof (struct nbinary));
330 n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL;
331 n1->nbinary.ch1 = list(0);
332 if ((got=readtoken()) != TDO) {
333 TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : ""));
334 synexpect(TDO);
335 }
336 n1->nbinary.ch2 = list(0);
337 if (readtoken() != TDONE)
338 synexpect(TDONE);
339 checkkwd = 1;
340 break;
341 }
342 case TFOR:
343 if (readtoken() != TWORD || quoteflag || ! goodname(wordtext))
344 synerror("Bad for loop variable");
345 n1 = (union node *)stalloc(sizeof (struct nfor));
346 n1->type = NFOR;
347 n1->nfor.var = wordtext;
348 if (readtoken() == TWORD && ! quoteflag && equal(wordtext, "in")) {
349 app = ≈
350 while (readtoken() == TWORD) {
351 n2 = (union node *)stalloc(sizeof (struct narg));
352 n2->type = NARG;
353 n2->narg.text = wordtext;
354 n2->narg.backquote = backquotelist;
355 *app = n2;
356 app = &n2->narg.next;
357 }
358 *app = NULL;
359 n1->nfor.args = ap;
360 if (lasttoken != TNL && lasttoken != TSEMI)
361 synexpect(-1);
362 } else {
363 #ifndef GDB_HACK
364 static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE,
365 '@', '=', '\0'};
366 #endif
367 n2 = (union node *)stalloc(sizeof (struct narg));
368 n2->type = NARG;
369 n2->narg.text = (char *)argvars;
370 n2->narg.backquote = NULL;
371 n2->narg.next = NULL;
372 n1->nfor.args = n2;
373 /*
374 * Newline or semicolon here is optional (but note
375 * that the original Bourne shell only allowed NL).
376 */
377 if (lasttoken != TNL && lasttoken != TSEMI)
378 tokpushback++;
379 }
380 checkkwd = 2;
381 if ((t = readtoken()) == TDO)
382 t = TDONE;
383 else if (t == TBEGIN)
384 t = TEND;
385 else
386 synexpect(-1);
387 n1->nfor.body = list(0);
388 if (readtoken() != t)
389 synexpect(t);
390 checkkwd = 1;
391 break;
392 case TCASE:
393 n1 = (union node *)stalloc(sizeof (struct ncase));
394 n1->type = NCASE;
395 if (readtoken() != TWORD)
396 synexpect(TWORD);
397 n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg));
398 n2->type = NARG;
399 n2->narg.text = wordtext;
400 n2->narg.backquote = backquotelist;
401 n2->narg.next = NULL;
402 while (readtoken() == TNL);
403 if (lasttoken != TWORD || ! equal(wordtext, "in"))
404 synerror("expecting \"in\"");
405 cpp = &n1->ncase.cases;
406 checkkwd = 2, readtoken();
407 do {
408 *cpp = cp = (union node *)stalloc(sizeof (struct nclist));
409 cp->type = NCLIST;
410 app = &cp->nclist.pattern;
411 for (;;) {
412 *app = ap = (union node *)stalloc(sizeof (struct narg));
413 ap->type = NARG;
414 ap->narg.text = wordtext;
415 ap->narg.backquote = backquotelist;
416 if (checkkwd = 2, readtoken() != TPIPE)
417 break;
418 app = &ap->narg.next;
419 readtoken();
420 }
421 ap->narg.next = NULL;
422 if (lasttoken != TRP)
423 synexpect(TRP);
424 cp->nclist.body = list(0);
425
426 checkkwd = 2;
427 if ((t = readtoken()) != TESAC) {
428 if (t != TENDCASE)
429 synexpect(TENDCASE);
430 else
431 checkkwd = 2, readtoken();
432 }
433 cpp = &cp->nclist.next;
434 } while(lasttoken != TESAC);
435 *cpp = NULL;
436 checkkwd = 1;
437 break;
438 case TLP:
439 n1 = (union node *)stalloc(sizeof (struct nredir));
440 n1->type = NSUBSHELL;
441 n1->nredir.n = list(0);
442 n1->nredir.redirect = NULL;
443 if (readtoken() != TRP)
444 synexpect(TRP);
445 checkkwd = 1;
446 break;
447 case TBEGIN:
448 n1 = list(0);
449 if (readtoken() != TEND)
450 synexpect(TEND);
451 checkkwd = 1;
452 break;
453 /* Handle an empty command like other simple commands. */
454 case TNL:
455 case TWORD:
456 tokpushback++;
457 return simplecmd(rpp, redir);
458 default:
459 synexpect(-1);
460 }
461
462 /* Now check for redirection which may follow command */
463 while (readtoken() == TREDIR) {
464 *rpp = n2 = redirnode;
465 rpp = &n2->nfile.next;
466 parsefname();
467 }
468 tokpushback++;
469 *rpp = NULL;
470 if (redir) {
471 if (n1->type != NSUBSHELL) {
472 n2 = (union node *)stalloc(sizeof (struct nredir));
473 n2->type = NREDIR;
474 n2->nredir.n = n1;
475 n1 = n2;
476 }
477 n1->nredir.redirect = redir;
478 }
479 return n1;
480 }
481
482
483 STATIC union node *
484 simplecmd(rpp, redir)
485 union node **rpp, *redir;
486 {
487 union node *args, **app;
488 union node **orig_rpp = rpp;
489 union node *n;
490
491 /* If we don't have any redirections already, then we must reset */
492 /* rpp to be the address of the local redir variable. */
493 if (redir == 0)
494 rpp = &redir;
495
496 args = NULL;
497 app = &args;
498 /*
499 * We save the incoming value, because we need this for shell
500 * functions. There can not be a redirect or an argument between
501 * the function name and the open parenthesis.
502 */
503 orig_rpp = rpp;
504
505 for (;;) {
506 if (readtoken() == TWORD) {
507 n = (union node *)stalloc(sizeof (struct narg));
508 n->type = NARG;
509 n->narg.text = wordtext;
510 n->narg.backquote = backquotelist;
511 *app = n;
512 app = &n->narg.next;
513 } else if (lasttoken == TREDIR) {
514 *rpp = n = redirnode;
515 rpp = &n->nfile.next;
516 parsefname(); /* read name of redirection file */
517 } else if (lasttoken == TLP && app == &args->narg.next
518 && rpp == orig_rpp) {
519 /* We have a function */
520 if (readtoken() != TRP)
521 synexpect(TRP);
522 #ifdef notdef
523 if (! goodname(n->narg.text))
524 synerror("Bad function name");
525 #endif
526 n->type = NDEFUN;
527 n->narg.next = command();
528 return n;
529 } else {
530 tokpushback++;
531 break;
532 }
533 }
534 *app = NULL;
535 *rpp = NULL;
536 n = (union node *)stalloc(sizeof (struct ncmd));
537 n->type = NCMD;
538 n->ncmd.backgnd = 0;
539 n->ncmd.args = args;
540 n->ncmd.redirect = redir;
541 return n;
542 }
543
544 STATIC union node *
545 makename() {
546 union node *n;
547
548 n = (union node *)stalloc(sizeof (struct narg));
549 n->type = NARG;
550 n->narg.next = NULL;
551 n->narg.text = wordtext;
552 n->narg.backquote = backquotelist;
553 return n;
554 }
555
556 void fixredir(n, text, err)
557 union node *n;
558 const char *text;
559 int err;
560 {
561 TRACE(("Fix redir %s %d\n", text, err));
562 if (!err)
563 n->ndup.vname = NULL;
564
565 if (is_digit(text[0]) && text[1] == '\0')
566 n->ndup.dupfd = digit_val(text[0]);
567 else if (text[0] == '-' && text[1] == '\0')
568 n->ndup.dupfd = -1;
569 else {
570
571 if (err)
572 synerror("Bad fd number");
573 else
574 n->ndup.vname = makename();
575 }
576 }
577
578
579 STATIC void
580 parsefname() {
581 union node *n = redirnode;
582
583 if (readtoken() != TWORD)
584 synexpect(-1);
585 if (n->type == NHERE) {
586 struct heredoc *here = heredoc;
587 struct heredoc *p;
588 int i;
589
590 if (quoteflag == 0)
591 n->type = NXHERE;
592 TRACE(("Here document %d\n", n->type));
593 if (here->striptabs) {
594 while (*wordtext == '\t')
595 wordtext++;
596 }
597 if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN)
598 synerror("Illegal eof marker for << redirection");
599 rmescapes(wordtext);
600 here->eofmark = wordtext;
601 here->next = NULL;
602 if (heredoclist == NULL)
603 heredoclist = here;
604 else {
605 for (p = heredoclist ; p->next ; p = p->next);
606 p->next = here;
607 }
608 } else if (n->type == NTOFD || n->type == NFROMFD) {
609 fixredir(n, wordtext, 0);
610 } else {
611 n->nfile.fname = makename();
612 }
613 }
614
615
616 /*
617 * Input any here documents.
618 */
619
620 STATIC void
621 parseheredoc() {
622 struct heredoc *here;
623 union node *n;
624
625 while (heredoclist) {
626 here = heredoclist;
627 heredoclist = here->next;
628 if (needprompt) {
629 setprompt(2);
630 needprompt = 0;
631 }
632 readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX,
633 here->eofmark, here->striptabs);
634 n = (union node *)stalloc(sizeof (struct narg));
635 n->narg.type = NARG;
636 n->narg.next = NULL;
637 n->narg.text = wordtext;
638 n->narg.backquote = backquotelist;
639 here->here->nhere.doc = n;
640 }
641 }
642
643 STATIC int
644 peektoken() {
645 int t;
646
647 t = readtoken();
648 tokpushback++;
649 return (t);
650 }
651
652 STATIC int xxreadtoken();
653
654 STATIC int
655 readtoken() {
656 int t;
657 int savecheckkwd = checkkwd;
658 struct alias *ap;
659 #ifdef DEBUG
660 int alreadyseen = tokpushback;
661 #endif
662
663 top:
664 t = xxreadtoken();
665
666 if (checkkwd) {
667 /*
668 * eat newlines
669 */
670 if (checkkwd == 2) {
671 checkkwd = 0;
672 while (t == TNL) {
673 parseheredoc();
674 t = xxreadtoken();
675 }
676 } else
677 checkkwd = 0;
678 /*
679 * check for keywords and aliases
680 */
681 if (t == TWORD && !quoteflag) {
682 register char * const *pp, *s;
683
684 for (pp = (char **)parsekwd; *pp; pp++) {
685 if (**pp == *wordtext && equal(*pp, wordtext)) {
686 lasttoken = t = pp - parsekwd + KWDOFFSET;
687 TRACE(("keyword %s recognized\n", tokname[t]));
688 goto out;
689 }
690 }
691 if (ap = lookupalias(wordtext, 1)) {
692 pushstring(ap->val, strlen(ap->val), ap);
693 checkkwd = savecheckkwd;
694 goto top;
695 }
696 }
697 out:
698 checkkwd = 0;
699 }
700 #ifdef DEBUG
701 if (!alreadyseen)
702 TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
703 else
704 TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
705 #endif
706 return (t);
707 }
708
709
710 /*
711 * Read the next input token.
712 * If the token is a word, we set backquotelist to the list of cmds in
713 * backquotes. We set quoteflag to true if any part of the word was
714 * quoted.
715 * If the token is TREDIR, then we set redirnode to a structure containing
716 * the redirection.
717 * In all cases, the variable startlinno is set to the number of the line
718 * on which the token starts.
719 *
720 * [Change comment: here documents and internal procedures]
721 * [Readtoken shouldn't have any arguments. Perhaps we should make the
722 * word parsing code into a separate routine. In this case, readtoken
723 * doesn't need to have any internal procedures, but parseword does.
724 * We could also make parseoperator in essence the main routine, and
725 * have parseword (readtoken1?) handle both words and redirection.]
726 */
727
728 #define RETURN(token) return lasttoken = token
729
730 STATIC int
731 xxreadtoken() {
732 register c;
733
734 if (tokpushback) {
735 tokpushback = 0;
736 return lasttoken;
737 }
738 if (needprompt) {
739 setprompt(2);
740 needprompt = 0;
741 }
742 startlinno = plinno;
743 for (;;) { /* until token or start of word found */
744 c = pgetc_macro();
745 if (c == ' ' || c == '\t')
746 continue; /* quick check for white space first */
747 switch (c) {
748 case ' ': case '\t':
749 continue;
750 case '#':
751 while ((c = pgetc()) != '\n' && c != PEOF);
752 pungetc();
753 continue;
754 case '\\':
755 if (pgetc() == '\n') {
756 startlinno = ++plinno;
757 if (doprompt)
758 setprompt(2);
759 else
760 setprompt(0);
761 continue;
762 }
763 pungetc();
764 goto breakloop;
765 case '\n':
766 plinno++;
767 needprompt = doprompt;
768 RETURN(TNL);
769 case PEOF:
770 RETURN(TEOF);
771 case '&':
772 if (pgetc() == '&')
773 RETURN(TAND);
774 pungetc();
775 RETURN(TBACKGND);
776 case '|':
777 if (pgetc() == '|')
778 RETURN(TOR);
779 pungetc();
780 RETURN(TPIPE);
781 case ';':
782 if (pgetc() == ';')
783 RETURN(TENDCASE);
784 pungetc();
785 RETURN(TSEMI);
786 case '(':
787 RETURN(TLP);
788 case ')':
789 RETURN(TRP);
790 default:
791 goto breakloop;
792 }
793 }
794 breakloop:
795 return readtoken1(c, BASESYNTAX, (char *)NULL, 0);
796 #undef RETURN
797 }
798
799
800
801 /*
802 * If eofmark is NULL, read a word or a redirection symbol. If eofmark
803 * is not NULL, read a here document. In the latter case, eofmark is the
804 * word which marks the end of the document and striptabs is true if
805 * leading tabs should be stripped from the document. The argument firstc
806 * is the first character of the input token or document.
807 *
808 * Because C does not have internal subroutines, I have simulated them
809 * using goto's to implement the subroutine linkage. The following macros
810 * will run code that appears at the end of readtoken1.
811 */
812
813 #define CHECKEND() {goto checkend; checkend_return:;}
814 #define PARSEREDIR() {goto parseredir; parseredir_return:;}
815 #define PARSESUB() {goto parsesub; parsesub_return:;}
816 #define PARSEBACKQOLD() {oldstyle = 1; goto parsebackq; parsebackq_oldreturn:;}
817 #define PARSEBACKQNEW() {oldstyle = 0; goto parsebackq; parsebackq_newreturn:;}
818 #define PARSEARITH() {goto parsearith; parsearith_return:;}
819
820 STATIC int
821 readtoken1(firstc, syntax, eofmark, striptabs)
822 int firstc;
823 char const *syntax;
824 char *eofmark;
825 int striptabs;
826 {
827 register c = firstc;
828 register char *out;
829 int len;
830 char line[EOFMARKLEN + 1];
831 struct nodelist *bqlist;
832 int quotef;
833 int dblquote;
834 int varnest; /* levels of variables expansion */
835 int arinest; /* levels of arithmetic expansion */
836 int parenlevel; /* levels of parens in arithmetic */
837 int oldstyle;
838 char const *prevsyntax; /* syntax before arithmetic */
839
840 startlinno = plinno;
841 dblquote = 0;
842 if (syntax == DQSYNTAX)
843 dblquote = 1;
844 quotef = 0;
845 bqlist = NULL;
846 varnest = 0;
847 arinest = 0;
848 parenlevel = 0;
849
850 STARTSTACKSTR(out);
851 loop: { /* for each line, until end of word */
852 #if ATTY
853 if (c == '\034' && doprompt
854 && attyset() && ! equal(termval(), "emacs")) {
855 attyline();
856 if (syntax == BASESYNTAX)
857 return readtoken();
858 c = pgetc();
859 goto loop;
860 }
861 #endif
862 CHECKEND(); /* set c to PEOF if at end of here document */
863 for (;;) { /* until end of line or end of word */
864 CHECKSTRSPACE(3, out); /* permit 3 calls to USTPUTC */
865 switch(syntax[c]) {
866 case CNL: /* '\n' */
867 if (syntax == BASESYNTAX)
868 goto endword; /* exit outer loop */
869 USTPUTC(c, out);
870 plinno++;
871 if (doprompt)
872 setprompt(2);
873 else
874 setprompt(0);
875 c = pgetc();
876 goto loop; /* continue outer loop */
877 case CWORD:
878 USTPUTC(c, out);
879 break;
880 case CCTL:
881 if (eofmark == NULL || dblquote)
882 USTPUTC(CTLESC, out);
883 USTPUTC(c, out);
884 break;
885 case CBACK: /* backslash */
886 c = pgetc();
887 if (c == PEOF) {
888 USTPUTC('\\', out);
889 pungetc();
890 } else if (c == '\n') {
891 if (doprompt)
892 setprompt(2);
893 else
894 setprompt(0);
895 } else {
896 if (dblquote && c != '\\' && c != '`' && c != '$'
897 && (c != '"' || eofmark != NULL))
898 USTPUTC('\\', out);
899 if (SQSYNTAX[c] == CCTL)
900 USTPUTC(CTLESC, out);
901 USTPUTC(c, out);
902 quotef++;
903 }
904 break;
905 case CSQUOTE:
906 syntax = SQSYNTAX;
907 break;
908 case CDQUOTE:
909 syntax = DQSYNTAX;
910 dblquote = 1;
911 break;
912 case CENDQUOTE:
913 if (eofmark) {
914 USTPUTC(c, out);
915 } else {
916 if (arinest)
917 syntax = ARISYNTAX;
918 else
919 syntax = BASESYNTAX;
920 quotef++;
921 dblquote = 0;
922 }
923 break;
924 case CVAR: /* '$' */
925 PARSESUB(); /* parse substitution */
926 break;
927 case CENDVAR: /* '}' */
928 if (varnest > 0) {
929 varnest--;
930 USTPUTC(CTLENDVAR, out);
931 } else {
932 USTPUTC(c, out);
933 }
934 break;
935 case CLP: /* '(' in arithmetic */
936 parenlevel++;
937 USTPUTC(c, out);
938 break;
939 case CRP: /* ')' in arithmetic */
940 if (parenlevel > 0) {
941 USTPUTC(c, out);
942 --parenlevel;
943 } else {
944 if (pgetc() == ')') {
945 if (--arinest == 0) {
946 USTPUTC(CTLENDARI, out);
947 syntax = prevsyntax;
948 } else
949 USTPUTC(')', out);
950 } else {
951 /*
952 * unbalanced parens
953 * (don't 2nd guess - no error)
954 */
955 pungetc();
956 USTPUTC(')', out);
957 }
958 }
959 break;
960 case CBQUOTE: /* '`' */
961 PARSEBACKQOLD();
962 break;
963 case CEOF:
964 goto endword; /* exit outer loop */
965 default:
966 if (varnest == 0)
967 goto endword; /* exit outer loop */
968 USTPUTC(c, out);
969 }
970 c = pgetc_macro();
971 }
972 }
973 endword:
974 if (syntax == ARISYNTAX)
975 synerror("Missing '))'");
976 if (syntax != BASESYNTAX && ! parsebackquote && eofmark == NULL)
977 synerror("Unterminated quoted string");
978 if (varnest != 0) {
979 startlinno = plinno;
980 synerror("Missing '}'");
981 }
982 USTPUTC('\0', out);
983 len = out - stackblock();
984 out = stackblock();
985 if (eofmark == NULL) {
986 if ((c == '>' || c == '<')
987 && quotef == 0
988 && len <= 2
989 && (*out == '\0' || is_digit(*out))) {
990 PARSEREDIR();
991 return lasttoken = TREDIR;
992 } else {
993 pungetc();
994 }
995 }
996 quoteflag = quotef;
997 backquotelist = bqlist;
998 grabstackblock(len);
999 wordtext = out;
1000 return lasttoken = TWORD;
1001 /* end of readtoken routine */
1002
1003
1004
1005 /*
1006 * Check to see whether we are at the end of the here document. When this
1007 * is called, c is set to the first character of the next input line. If
1008 * we are at the end of the here document, this routine sets the c to PEOF.
1009 */
1010
1011 checkend: {
1012 if (eofmark) {
1013 if (striptabs) {
1014 while (c == '\t')
1015 c = pgetc();
1016 }
1017 if (c == *eofmark) {
1018 if (pfgets(line, sizeof line) != NULL) {
1019 register char *p, *q;
1020
1021 p = line;
1022 for (q = eofmark + 1 ; *q && *p == *q ; p++, q++);
1023 if (*p == '\n' && *q == '\0') {
1024 c = PEOF;
1025 plinno++;
1026 needprompt = doprompt;
1027 } else {
1028 pushstring(line, strlen(line), NULL);
1029 }
1030 }
1031 }
1032 }
1033 goto checkend_return;
1034 }
1035
1036
1037 /*
1038 * Parse a redirection operator. The variable "out" points to a string
1039 * specifying the fd to be redirected. The variable "c" contains the
1040 * first character of the redirection operator.
1041 */
1042
1043 parseredir: {
1044 char fd = *out;
1045 union node *np;
1046
1047 np = (union node *)stalloc(sizeof (struct nfile));
1048 if (c == '>') {
1049 np->nfile.fd = 1;
1050 c = pgetc();
1051 if (c == '>')
1052 np->type = NAPPEND;
1053 else if (c == '&')
1054 np->type = NTOFD;
1055 else {
1056 np->type = NTO;
1057 pungetc();
1058 }
1059 } else { /* c == '<' */
1060 np->nfile.fd = 0;
1061 c = pgetc();
1062 if (c == '<') {
1063 if (sizeof (struct nfile) != sizeof (struct nhere)) {
1064 np = (union node *)stalloc(sizeof (struct nhere));
1065 np->nfile.fd = 0;
1066 }
1067 np->type = NHERE;
1068 heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc));
1069 heredoc->here = np;
1070 if ((c = pgetc()) == '-') {
1071 heredoc->striptabs = 1;
1072 } else {
1073 heredoc->striptabs = 0;
1074 pungetc();
1075 }
1076 } else if (c == '&')
1077 np->type = NFROMFD;
1078 else {
1079 np->type = NFROM;
1080 pungetc();
1081 }
1082 }
1083 if (fd != '\0')
1084 np->nfile.fd = digit_val(fd);
1085 redirnode = np;
1086 goto parseredir_return;
1087 }
1088
1089
1090 /*
1091 * Parse a substitution. At this point, we have read the dollar sign
1092 * and nothing else.
1093 */
1094
1095 parsesub: {
1096 int subtype;
1097 int typeloc;
1098 int flags;
1099 char *p;
1100 #ifndef GDB_HACK
1101 static const char types[] = "}-+?=";
1102 #endif
1103
1104 c = pgetc();
1105 if (c != '(' && c != '{' && !is_name(c) && !is_special(c)) {
1106 USTPUTC('$', out);
1107 pungetc();
1108 } else if (c == '(') { /* $(command) or $((arith)) */
1109 if (pgetc() == '(') {
1110 PARSEARITH();
1111 } else {
1112 pungetc();
1113 PARSEBACKQNEW();
1114 }
1115 } else {
1116 USTPUTC(CTLVAR, out);
1117 typeloc = out - stackblock();
1118 USTPUTC(VSNORMAL, out);
1119 subtype = VSNORMAL;
1120 if (c == '{') {
1121 c = pgetc();
1122 subtype = 0;
1123 }
1124 if (is_name(c)) {
1125 do {
1126 STPUTC(c, out);
1127 c = pgetc();
1128 } while (is_in_name(c));
1129 } else {
1130 if (! is_special(c))
1131 badsub: synerror("Bad substitution");
1132 USTPUTC(c, out);
1133 c = pgetc();
1134 }
1135 STPUTC('=', out);
1136 flags = 0;
1137 if (subtype == 0) {
1138 if (c == ':') {
1139 flags = VSNUL;
1140 c = pgetc();
1141 }
1142 p = strchr(types, c);
1143 if (p == NULL)
1144 goto badsub;
1145 subtype = p - types + VSNORMAL;
1146 } else {
1147 pungetc();
1148 }
1149 if (dblquote || arinest)
1150 flags |= VSQUOTE;
1151 *(stackblock() + typeloc) = subtype | flags;
1152 if (subtype != VSNORMAL)
1153 varnest++;
1154 }
1155 goto parsesub_return;
1156 }
1157
1158
1159 /*
1160 * Called to parse command substitutions. Newstyle is set if the command
1161 * is enclosed inside $(...); nlpp is a pointer to the head of the linked
1162 * list of commands (passed by reference), and savelen is the number of
1163 * characters on the top of the stack which must be preserved.
1164 */
1165
1166 parsebackq: {
1167 struct nodelist **nlpp;
1168 int savepbq;
1169 union node *n;
1170 char *volatile str;
1171 struct jmploc jmploc;
1172 struct jmploc *volatile savehandler;
1173 int savelen;
1174
1175 savepbq = parsebackquote;
1176 if (setjmp(jmploc.loc)) {
1177 if (str)
1178 ckfree(str);
1179 parsebackquote = 0;
1180 handler = savehandler;
1181 longjmp(handler->loc, 1);
1182 }
1183 INTOFF;
1184 str = NULL;
1185 savelen = out - stackblock();
1186 if (savelen > 0) {
1187 str = ckmalloc(savelen);
1188 bcopy(stackblock(), str, savelen);
1189 }
1190 savehandler = handler;
1191 handler = &jmploc;
1192 INTON;
1193 if (oldstyle) {
1194 /* We must read until the closing backquote, giving special
1195 treatment to some slashes, and then push the string and
1196 reread it as input, interpreting it normally. */
1197 register char *out;
1198 register c;
1199 int savelen;
1200 char *str;
1201
1202 STARTSTACKSTR(out);
1203 while ((c = pgetc ()) != '`') {
1204 if (c == '\\') {
1205 c = pgetc ();
1206 if (c != '\\' && c != '`' && c != '$'
1207 && (!dblquote || c != '"'))
1208 STPUTC('\\', out);
1209 }
1210 STPUTC(c, out);
1211 }
1212 STPUTC('\0', out);
1213 savelen = out - stackblock();
1214 if (savelen > 0) {
1215 str = ckmalloc(savelen);
1216 bcopy(stackblock(), str, savelen);
1217 }
1218 setinputstring(str, 1);
1219 }
1220 nlpp = &bqlist;
1221 while (*nlpp)
1222 nlpp = &(*nlpp)->next;
1223 *nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist));
1224 (*nlpp)->next = NULL;
1225 parsebackquote = oldstyle;
1226 n = list(0);
1227 if (!oldstyle && (readtoken() != TRP))
1228 synexpect(TRP);
1229 (*nlpp)->n = n;
1230 /* Start reading from old file again. */
1231 if (oldstyle)
1232 popfile();
1233 while (stackblocksize() <= savelen)
1234 growstackblock();
1235 STARTSTACKSTR(out);
1236 if (str) {
1237 bcopy(str, out, savelen);
1238 STADJUST(savelen, out);
1239 INTOFF;
1240 ckfree(str);
1241 str = NULL;
1242 INTON;
1243 }
1244 parsebackquote = savepbq;
1245 handler = savehandler;
1246 if (arinest || dblquote)
1247 USTPUTC(CTLBACKQ | CTLQUOTE, out);
1248 else
1249 USTPUTC(CTLBACKQ, out);
1250 if (oldstyle)
1251 goto parsebackq_oldreturn;
1252 else
1253 goto parsebackq_newreturn;
1254 }
1255
1256 /*
1257 * Parse an arithmetic expansion (indicate start of one and set state)
1258 */
1259 parsearith: {
1260
1261 if (++arinest == 1) {
1262 prevsyntax = syntax;
1263 syntax = ARISYNTAX;
1264 USTPUTC(CTLARI, out);
1265 } else {
1266 /*
1267 * we collapse embedded arithmetic expansion to
1268 * parenthesis, which should be equivalent
1269 */
1270 USTPUTC('(', out);
1271 }
1272 goto parsearith_return;
1273 }
1274
1275 } /* end of readtoken */
1276
1277
1278
1279 #ifdef mkinit
1280 RESET {
1281 tokpushback = 0;
1282 checkkwd = 0;
1283 }
1284 #endif
1285
1286 /*
1287 * Returns true if the text contains nothing to expand (no dollar signs
1288 * or backquotes).
1289 */
1290
1291 STATIC int
1292 noexpand(text)
1293 char *text;
1294 {
1295 register char *p;
1296 register char c;
1297
1298 p = text;
1299 while ((c = *p++) != '\0') {
1300 if (c == CTLESC)
1301 p++;
1302 else if (BASESYNTAX[c] == CCTL)
1303 return 0;
1304 }
1305 return 1;
1306 }
1307
1308
1309 /*
1310 * Return true if the argument is a legal variable name (a letter or
1311 * underscore followed by zero or more letters, underscores, and digits).
1312 */
1313
1314 int
1315 goodname(name)
1316 char *name;
1317 {
1318 register char *p;
1319
1320 p = name;
1321 if (! is_name(*p))
1322 return 0;
1323 while (*++p) {
1324 if (! is_in_name(*p))
1325 return 0;
1326 }
1327 return 1;
1328 }
1329
1330
1331 /*
1332 * Called when an unexpected token is read during the parse. The argument
1333 * is the token that is expected, or -1 if more than one type of token can
1334 * occur at this point.
1335 */
1336
1337 STATIC void
1338 synexpect(token) {
1339 char msg[64];
1340
1341 if (token >= 0) {
1342 fmtstr(msg, 64, "%s unexpected (expecting %s)",
1343 tokname[lasttoken], tokname[token]);
1344 } else {
1345 fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]);
1346 }
1347 synerror(msg);
1348 }
1349
1350
1351 STATIC void
1352 synerror(msg)
1353 char *msg;
1354 {
1355 if (commandname)
1356 outfmt(&errout, "%s: %d: ", commandname, startlinno);
1357 outfmt(&errout, "Syntax error: %s\n", msg);
1358 error((char *)NULL);
1359 }
1360
1361 STATIC void
1362 setprompt(which)
1363 int which;
1364 {
1365 whichprompt = which;
1366
1367 #ifndef NO_HISTORY
1368 if (!el)
1369 #endif
1370 out2str(getprompt(NULL));
1371 }
1372
1373 /*
1374 * called by editline -- any expansions to the prompt
1375 * should be added here.
1376 */
1377 char *
1378 getprompt(unused)
1379 void *unused;
1380 {
1381 switch (whichprompt) {
1382 case 0:
1383 return "";
1384 case 1:
1385 return ps1val();
1386 case 2:
1387 return ps2val();
1388 default:
1389 return "<internal prompt error>";
1390 }
1391 }
1392