parser.c revision 50471 
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#if 0
39static char sccsid[] = "@(#)parser.c	8.7 (Berkeley) 5/16/95";
40#endif
41static const char rcsid[] =
42  "$FreeBSD: head/bin/sh/parser.c 50471 1999-08-27 23:15:48Z peter $";
43#endif /* not lint */
44
45#include <stdlib.h>
46
47#include "shell.h"
48#include "parser.h"
49#include "nodes.h"
50#include "expand.h"	/* defines rmescapes() */
51#include "redir.h"	/* defines copyfd() */
52#include "syntax.h"
53#include "options.h"
54#include "input.h"
55#include "output.h"
56#include "var.h"
57#include "error.h"
58#include "memalloc.h"
59#include "mystring.h"
60#include "alias.h"
61#include "show.h"
62#ifndef NO_HISTORY
63#include "myhistedit.h"
64#endif
65
66/*
67 * Shell command parser.
68 */
69
70#define EOFMARKLEN 79
71
72/* values returned by readtoken */
73#include "token.h"
74
75
76
77struct heredoc {
78	struct heredoc *next;	/* next here document in list */
79	union node *here;		/* redirection node */
80	char *eofmark;		/* string indicating end of input */
81	int striptabs;		/* if set, strip leading tabs */
82};
83
84
85
86struct heredoc *heredoclist;	/* list of here documents to read */
87int parsebackquote;		/* nonzero if we are inside backquotes */
88int doprompt;			/* if set, prompt the user */
89int needprompt;			/* true if interactive and at start of line */
90int lasttoken;			/* last token read */
91MKINIT int tokpushback;		/* last token pushed back */
92char *wordtext;			/* text of last word returned by readtoken */
93MKINIT int checkkwd;            /* 1 == check for kwds, 2 == also eat newlines */
94struct nodelist *backquotelist;
95union node *redirnode;
96struct heredoc *heredoc;
97int quoteflag;			/* set if (part of) last token was quoted */
98int startlinno;			/* line # where last token started */
99
100/* XXX When 'noaliases' is set to one, no alias expansion takes place. */
101static int noaliases = 0;
102
103#define GDB_HACK 1 /* avoid local declarations which gdb can't handle */
104#ifdef GDB_HACK
105static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0'};
106static const char types[] = "}-+?=";
107#endif
108
109
110STATIC union node *list __P((int));
111STATIC union node *andor __P((void));
112STATIC union node *pipeline __P((void));
113STATIC union node *command __P((void));
114STATIC union node *simplecmd __P((union node **, union node *));
115STATIC union node *makename __P((void));
116STATIC void parsefname __P((void));
117STATIC void parseheredoc __P((void));
118STATIC int peektoken __P((void));
119STATIC int readtoken __P((void));
120STATIC int xxreadtoken __P((void));
121STATIC int readtoken1 __P((int, char const *, char *, int));
122STATIC int noexpand __P((char *));
123STATIC void synexpect __P((int));
124STATIC void synerror __P((char *));
125STATIC void setprompt __P((int));
126
127
128/*
129 * Read and parse a command.  Returns NEOF on end of file.  (NULL is a
130 * valid parse tree indicating a blank line.)
131 */
132
133union node *
134parsecmd(interact)
135	int interact;
136{
137	int t;
138
139	doprompt = interact;
140	if (doprompt)
141		setprompt(1);
142	else
143		setprompt(0);
144	needprompt = 0;
145	t = readtoken();
146	if (t == TEOF)
147		return NEOF;
148	if (t == TNL)
149		return NULL;
150	tokpushback++;
151	return list(1);
152}
153
154
155STATIC union node *
156list(nlflag)
157	int nlflag;
158{
159	union node *n1, *n2, *n3;
160	int tok;
161
162	checkkwd = 2;
163	if (nlflag == 0 && tokendlist[peektoken()])
164		return NULL;
165	n1 = NULL;
166	for (;;) {
167		n2 = andor();
168		tok = readtoken();
169		if (tok == TBACKGND) {
170			if (n2->type == NCMD || n2->type == NPIPE) {
171				n2->ncmd.backgnd = 1;
172			} else if (n2->type == NREDIR) {
173				n2->type = NBACKGND;
174			} else {
175				n3 = (union node *)stalloc(sizeof (struct nredir));
176				n3->type = NBACKGND;
177				n3->nredir.n = n2;
178				n3->nredir.redirect = NULL;
179				n2 = n3;
180			}
181		}
182		if (n1 == NULL) {
183			n1 = n2;
184		}
185		else {
186			n3 = (union node *)stalloc(sizeof (struct nbinary));
187			n3->type = NSEMI;
188			n3->nbinary.ch1 = n1;
189			n3->nbinary.ch2 = n2;
190			n1 = n3;
191		}
192		switch (tok) {
193		case TBACKGND:
194		case TSEMI:
195			tok = readtoken();
196			/* fall through */
197		case TNL:
198			if (tok == TNL) {
199				parseheredoc();
200				if (nlflag)
201					return n1;
202			} else {
203				tokpushback++;
204			}
205			checkkwd = 2;
206			if (tokendlist[peektoken()])
207				return n1;
208			break;
209		case TEOF:
210			if (heredoclist)
211				parseheredoc();
212			else
213				pungetc();		/* push back EOF on input */
214			return n1;
215		default:
216			if (nlflag)
217				synexpect(-1);
218			tokpushback++;
219			return n1;
220		}
221	}
222}
223
224
225
226STATIC union node *
227andor() {
228	union node *n1, *n2, *n3;
229	int t;
230
231	n1 = pipeline();
232	for (;;) {
233		if ((t = readtoken()) == TAND) {
234			t = NAND;
235		} else if (t == TOR) {
236			t = NOR;
237		} else {
238			tokpushback++;
239			return n1;
240		}
241		n2 = pipeline();
242		n3 = (union node *)stalloc(sizeof (struct nbinary));
243		n3->type = t;
244		n3->nbinary.ch1 = n1;
245		n3->nbinary.ch2 = n2;
246		n1 = n3;
247	}
248}
249
250
251
252STATIC union node *
253pipeline() {
254	union node *n1, *pipenode, *notnode;
255	struct nodelist *lp, *prev;
256	int negate = 0;
257
258	TRACE(("pipeline: entered\n"));
259	while (readtoken() == TNOT) {
260		TRACE(("pipeline: TNOT recognized\n"));
261		negate = !negate;
262	}
263	tokpushback++;
264	n1 = command();
265	if (readtoken() == TPIPE) {
266		pipenode = (union node *)stalloc(sizeof (struct npipe));
267		pipenode->type = NPIPE;
268		pipenode->npipe.backgnd = 0;
269		lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
270		pipenode->npipe.cmdlist = lp;
271		lp->n = n1;
272		do {
273			prev = lp;
274			lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
275			lp->n = command();
276			prev->next = lp;
277		} while (readtoken() == TPIPE);
278		lp->next = NULL;
279		n1 = pipenode;
280	}
281	tokpushback++;
282	if (negate) {
283		notnode = (union node *)stalloc(sizeof(struct nnot));
284		notnode->type = NNOT;
285		notnode->nnot.com = n1;
286		n1 = notnode;
287	}
288	return n1;
289}
290
291
292
293STATIC union node *
294command() {
295	union node *n1, *n2;
296	union node *ap, **app;
297	union node *cp, **cpp;
298	union node *redir, **rpp;
299	int t;
300
301	checkkwd = 2;
302	redir = NULL;
303	n1 = NULL;
304	rpp = &redir;
305
306	/* Check for redirection which may precede command */
307	while (readtoken() == TREDIR) {
308		*rpp = n2 = redirnode;
309		rpp = &n2->nfile.next;
310		parsefname();
311	}
312	tokpushback++;
313
314	switch (readtoken()) {
315	case TIF:
316		n1 = (union node *)stalloc(sizeof (struct nif));
317		n1->type = NIF;
318		n1->nif.test = list(0);
319		if (readtoken() != TTHEN)
320			synexpect(TTHEN);
321		n1->nif.ifpart = list(0);
322		n2 = n1;
323		while (readtoken() == TELIF) {
324			n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif));
325			n2 = n2->nif.elsepart;
326			n2->type = NIF;
327			n2->nif.test = list(0);
328			if (readtoken() != TTHEN)
329				synexpect(TTHEN);
330			n2->nif.ifpart = list(0);
331		}
332		if (lasttoken == TELSE)
333			n2->nif.elsepart = list(0);
334		else {
335			n2->nif.elsepart = NULL;
336			tokpushback++;
337		}
338		if (readtoken() != TFI)
339			synexpect(TFI);
340		checkkwd = 1;
341		break;
342	case TWHILE:
343	case TUNTIL: {
344		int got;
345		n1 = (union node *)stalloc(sizeof (struct nbinary));
346		n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL;
347		n1->nbinary.ch1 = list(0);
348		if ((got=readtoken()) != TDO) {
349TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : ""));
350			synexpect(TDO);
351		}
352		n1->nbinary.ch2 = list(0);
353		if (readtoken() != TDONE)
354			synexpect(TDONE);
355		checkkwd = 1;
356		break;
357	}
358	case TFOR:
359		if (readtoken() != TWORD || quoteflag || ! goodname(wordtext))
360			synerror("Bad for loop variable");
361		n1 = (union node *)stalloc(sizeof (struct nfor));
362		n1->type = NFOR;
363		n1->nfor.var = wordtext;
364		if (readtoken() == TWORD && ! quoteflag && equal(wordtext, "in")) {
365			app = &ap;
366			while (readtoken() == TWORD) {
367				n2 = (union node *)stalloc(sizeof (struct narg));
368				n2->type = NARG;
369				n2->narg.text = wordtext;
370				n2->narg.backquote = backquotelist;
371				*app = n2;
372				app = &n2->narg.next;
373			}
374			*app = NULL;
375			n1->nfor.args = ap;
376			if (lasttoken != TNL && lasttoken != TSEMI)
377				synexpect(-1);
378		} else {
379#ifndef GDB_HACK
380			static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE,
381								   '@', '=', '\0'};
382#endif
383			n2 = (union node *)stalloc(sizeof (struct narg));
384			n2->type = NARG;
385			n2->narg.text = (char *)argvars;
386			n2->narg.backquote = NULL;
387			n2->narg.next = NULL;
388			n1->nfor.args = n2;
389			/*
390			 * Newline or semicolon here is optional (but note
391			 * that the original Bourne shell only allowed NL).
392			 */
393			if (lasttoken != TNL && lasttoken != TSEMI)
394				tokpushback++;
395		}
396		checkkwd = 2;
397		if ((t = readtoken()) == TDO)
398			t = TDONE;
399		else if (t == TBEGIN)
400			t = TEND;
401		else
402			synexpect(-1);
403		n1->nfor.body = list(0);
404		if (readtoken() != t)
405			synexpect(t);
406		checkkwd = 1;
407		break;
408	case TCASE:
409		n1 = (union node *)stalloc(sizeof (struct ncase));
410		n1->type = NCASE;
411		if (readtoken() != TWORD)
412			synexpect(TWORD);
413		n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg));
414		n2->type = NARG;
415		n2->narg.text = wordtext;
416		n2->narg.backquote = backquotelist;
417		n2->narg.next = NULL;
418		while (readtoken() == TNL);
419		if (lasttoken != TWORD || ! equal(wordtext, "in"))
420			synerror("expecting \"in\"");
421		cpp = &n1->ncase.cases;
422		noaliases = 1;	/* turn off alias expansion */
423		checkkwd = 2, readtoken();
424		do {
425			*cpp = cp = (union node *)stalloc(sizeof (struct nclist));
426			cp->type = NCLIST;
427			app = &cp->nclist.pattern;
428			for (;;) {
429				*app = ap = (union node *)stalloc(sizeof (struct narg));
430				ap->type = NARG;
431				ap->narg.text = wordtext;
432				ap->narg.backquote = backquotelist;
433				if (checkkwd = 2, readtoken() != TPIPE)
434					break;
435				app = &ap->narg.next;
436				readtoken();
437			}
438			ap->narg.next = NULL;
439			if (lasttoken != TRP)
440				noaliases = 0, synexpect(TRP);
441			cp->nclist.body = list(0);
442
443			checkkwd = 2;
444			if ((t = readtoken()) != TESAC) {
445				if (t != TENDCASE)
446					noaliases = 0, synexpect(TENDCASE);
447				else
448					checkkwd = 2, readtoken();
449			}
450			cpp = &cp->nclist.next;
451		} while(lasttoken != TESAC);
452		noaliases = 0;	/* reset alias expansion */
453		*cpp = NULL;
454		checkkwd = 1;
455		break;
456	case TLP:
457		n1 = (union node *)stalloc(sizeof (struct nredir));
458		n1->type = NSUBSHELL;
459		n1->nredir.n = list(0);
460		n1->nredir.redirect = NULL;
461		if (readtoken() != TRP)
462			synexpect(TRP);
463		checkkwd = 1;
464		break;
465	case TBEGIN:
466		n1 = list(0);
467		if (readtoken() != TEND)
468			synexpect(TEND);
469		checkkwd = 1;
470		break;
471	/* Handle an empty command like other simple commands.  */
472	case TSEMI:
473		/*
474		 * An empty command before a ; doesn't make much sense, and
475		 * should certainly be disallowed in the case of `if ;'.
476		 */
477		if (!redir)
478			synexpect(-1);
479	case TAND:
480	case TOR:
481	case TNL:
482	case TEOF:
483	case TWORD:
484	case TRP:
485		tokpushback++;
486		return simplecmd(rpp, redir);
487	default:
488		synexpect(-1);
489	}
490
491	/* Now check for redirection which may follow command */
492	while (readtoken() == TREDIR) {
493		*rpp = n2 = redirnode;
494		rpp = &n2->nfile.next;
495		parsefname();
496	}
497	tokpushback++;
498	*rpp = NULL;
499	if (redir) {
500		if (n1->type != NSUBSHELL) {
501			n2 = (union node *)stalloc(sizeof (struct nredir));
502			n2->type = NREDIR;
503			n2->nredir.n = n1;
504			n1 = n2;
505		}
506		n1->nredir.redirect = redir;
507	}
508	return n1;
509}
510
511
512STATIC union node *
513simplecmd(rpp, redir)
514	union node **rpp, *redir;
515	{
516	union node *args, **app;
517	union node **orig_rpp = rpp;
518	union node *n = NULL;
519
520	/* If we don't have any redirections already, then we must reset */
521	/* rpp to be the address of the local redir variable.  */
522	if (redir == 0)
523		rpp = &redir;
524
525	args = NULL;
526	app = &args;
527	/*
528	 * We save the incoming value, because we need this for shell
529	 * functions.  There can not be a redirect or an argument between
530	 * the function name and the open parenthesis.
531	 */
532	orig_rpp = rpp;
533
534	for (;;) {
535		if (readtoken() == TWORD) {
536			n = (union node *)stalloc(sizeof (struct narg));
537			n->type = NARG;
538			n->narg.text = wordtext;
539			n->narg.backquote = backquotelist;
540			*app = n;
541			app = &n->narg.next;
542		} else if (lasttoken == TREDIR) {
543			*rpp = n = redirnode;
544			rpp = &n->nfile.next;
545			parsefname();	/* read name of redirection file */
546		} else if (lasttoken == TLP && app == &args->narg.next
547					    && rpp == orig_rpp) {
548			/* We have a function */
549			if (readtoken() != TRP)
550				synexpect(TRP);
551#ifdef notdef
552			if (! goodname(n->narg.text))
553				synerror("Bad function name");
554#endif
555			n->type = NDEFUN;
556			n->narg.next = command();
557			return n;
558		} else {
559			tokpushback++;
560			break;
561		}
562	}
563	*app = NULL;
564	*rpp = NULL;
565	n = (union node *)stalloc(sizeof (struct ncmd));
566	n->type = NCMD;
567	n->ncmd.backgnd = 0;
568	n->ncmd.args = args;
569	n->ncmd.redirect = redir;
570	return n;
571}
572
573STATIC union node *
574makename() {
575	union node *n;
576
577	n = (union node *)stalloc(sizeof (struct narg));
578	n->type = NARG;
579	n->narg.next = NULL;
580	n->narg.text = wordtext;
581	n->narg.backquote = backquotelist;
582	return n;
583}
584
585void fixredir(n, text, err)
586	union node *n;
587	const char *text;
588	int err;
589	{
590	TRACE(("Fix redir %s %d\n", text, err));
591	if (!err)
592		n->ndup.vname = NULL;
593
594	if (is_digit(text[0]) && text[1] == '\0')
595		n->ndup.dupfd = digit_val(text[0]);
596	else if (text[0] == '-' && text[1] == '\0')
597		n->ndup.dupfd = -1;
598	else {
599
600		if (err)
601			synerror("Bad fd number");
602		else
603			n->ndup.vname = makename();
604	}
605}
606
607
608STATIC void
609parsefname() {
610	union node *n = redirnode;
611
612	if (readtoken() != TWORD)
613		synexpect(-1);
614	if (n->type == NHERE) {
615		struct heredoc *here = heredoc;
616		struct heredoc *p;
617		int i;
618
619		if (quoteflag == 0)
620			n->type = NXHERE;
621		TRACE(("Here document %d\n", n->type));
622		if (here->striptabs) {
623			while (*wordtext == '\t')
624				wordtext++;
625		}
626		if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN)
627			synerror("Illegal eof marker for << redirection");
628		rmescapes(wordtext);
629		here->eofmark = wordtext;
630		here->next = NULL;
631		if (heredoclist == NULL)
632			heredoclist = here;
633		else {
634			for (p = heredoclist ; p->next ; p = p->next);
635			p->next = here;
636		}
637	} else if (n->type == NTOFD || n->type == NFROMFD) {
638		fixredir(n, wordtext, 0);
639	} else {
640		n->nfile.fname = makename();
641	}
642}
643
644
645/*
646 * Input any here documents.
647 */
648
649STATIC void
650parseheredoc() {
651	struct heredoc *here;
652	union node *n;
653
654	while (heredoclist) {
655		here = heredoclist;
656		heredoclist = here->next;
657		if (needprompt) {
658			setprompt(2);
659			needprompt = 0;
660		}
661		readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX,
662				here->eofmark, here->striptabs);
663		n = (union node *)stalloc(sizeof (struct narg));
664		n->narg.type = NARG;
665		n->narg.next = NULL;
666		n->narg.text = wordtext;
667		n->narg.backquote = backquotelist;
668		here->here->nhere.doc = n;
669	}
670}
671
672STATIC int
673peektoken() {
674	int t;
675
676	t = readtoken();
677	tokpushback++;
678	return (t);
679}
680
681STATIC int xxreadtoken();
682
683STATIC int
684readtoken() {
685	int t;
686	int savecheckkwd = checkkwd;
687	struct alias *ap;
688#ifdef DEBUG
689	int alreadyseen = tokpushback;
690#endif
691
692	top:
693	t = xxreadtoken();
694
695	if (checkkwd) {
696		/*
697		 * eat newlines
698		 */
699		if (checkkwd == 2) {
700			checkkwd = 0;
701			while (t == TNL) {
702				parseheredoc();
703				t = xxreadtoken();
704			}
705		} else
706			checkkwd = 0;
707		/*
708		 * check for keywords and aliases
709		 */
710		if (t == TWORD && !quoteflag)
711		{
712			char * const *pp;
713
714			for (pp = (char **)parsekwd; *pp; pp++) {
715				if (**pp == *wordtext && equal(*pp, wordtext))
716				{
717					lasttoken = t = pp - parsekwd + KWDOFFSET;
718					TRACE(("keyword %s recognized\n", tokname[t]));
719					goto out;
720				}
721			}
722			if (noaliases == 0 &&
723			    (ap = lookupalias(wordtext, 1)) != NULL) {
724				pushstring(ap->val, strlen(ap->val), ap);
725				checkkwd = savecheckkwd;
726				goto top;
727			}
728		}
729out:
730		checkkwd = 0;
731	}
732#ifdef DEBUG
733	if (!alreadyseen)
734	    TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
735	else
736	    TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
737#endif
738	return (t);
739}
740
741
742/*
743 * Read the next input token.
744 * If the token is a word, we set backquotelist to the list of cmds in
745 *	backquotes.  We set quoteflag to true if any part of the word was
746 *	quoted.
747 * If the token is TREDIR, then we set redirnode to a structure containing
748 *	the redirection.
749 * In all cases, the variable startlinno is set to the number of the line
750 *	on which the token starts.
751 *
752 * [Change comment:  here documents and internal procedures]
753 * [Readtoken shouldn't have any arguments.  Perhaps we should make the
754 *  word parsing code into a separate routine.  In this case, readtoken
755 *  doesn't need to have any internal procedures, but parseword does.
756 *  We could also make parseoperator in essence the main routine, and
757 *  have parseword (readtoken1?) handle both words and redirection.]
758 */
759
760#define RETURN(token)	return lasttoken = token
761
762STATIC int
763xxreadtoken() {
764	int c;
765
766	if (tokpushback) {
767		tokpushback = 0;
768		return lasttoken;
769	}
770	if (needprompt) {
771		setprompt(2);
772		needprompt = 0;
773	}
774	startlinno = plinno;
775	for (;;) {	/* until token or start of word found */
776		c = pgetc_macro();
777		if (c == ' ' || c == '\t')
778			continue;		/* quick check for white space first */
779		switch (c) {
780		case ' ': case '\t':
781			continue;
782		case '#':
783			while ((c = pgetc()) != '\n' && c != PEOF);
784			pungetc();
785			continue;
786		case '\\':
787			if (pgetc() == '\n') {
788				startlinno = ++plinno;
789				if (doprompt)
790					setprompt(2);
791				else
792					setprompt(0);
793				continue;
794			}
795			pungetc();
796			goto breakloop;
797		case '\n':
798			plinno++;
799			needprompt = doprompt;
800			RETURN(TNL);
801		case PEOF:
802			RETURN(TEOF);
803		case '&':
804			if (pgetc() == '&')
805				RETURN(TAND);
806			pungetc();
807			RETURN(TBACKGND);
808		case '|':
809			if (pgetc() == '|')
810				RETURN(TOR);
811			pungetc();
812			RETURN(TPIPE);
813		case ';':
814			if (pgetc() == ';')
815				RETURN(TENDCASE);
816			pungetc();
817			RETURN(TSEMI);
818		case '(':
819			RETURN(TLP);
820		case ')':
821			RETURN(TRP);
822		default:
823			goto breakloop;
824		}
825	}
826breakloop:
827	return readtoken1(c, BASESYNTAX, (char *)NULL, 0);
828#undef RETURN
829}
830
831
832
833/*
834 * If eofmark is NULL, read a word or a redirection symbol.  If eofmark
835 * is not NULL, read a here document.  In the latter case, eofmark is the
836 * word which marks the end of the document and striptabs is true if
837 * leading tabs should be stripped from the document.  The argument firstc
838 * is the first character of the input token or document.
839 *
840 * Because C does not have internal subroutines, I have simulated them
841 * using goto's to implement the subroutine linkage.  The following macros
842 * will run code that appears at the end of readtoken1.
843 */
844
845#define CHECKEND()	{goto checkend; checkend_return:;}
846#define PARSEREDIR()	{goto parseredir; parseredir_return:;}
847#define PARSESUB()	{goto parsesub; parsesub_return:;}
848#define PARSEBACKQOLD()	{oldstyle = 1; goto parsebackq; parsebackq_oldreturn:;}
849#define PARSEBACKQNEW()	{oldstyle = 0; goto parsebackq; parsebackq_newreturn:;}
850#define	PARSEARITH()	{goto parsearith; parsearith_return:;}
851
852STATIC int
853readtoken1(firstc, syntax, eofmark, striptabs)
854	int firstc;
855	char const *syntax;
856	char *eofmark;
857	int striptabs;
858	{
859	int c = firstc;
860	char *out;
861	int len;
862	char line[EOFMARKLEN + 1];
863	struct nodelist *bqlist;
864	int quotef;
865	int dblquote;
866	int varnest;	/* levels of variables expansion */
867	int arinest;	/* levels of arithmetic expansion */
868	int parenlevel;	/* levels of parens in arithmetic */
869	int oldstyle;
870	char const *prevsyntax;	/* syntax before arithmetic */
871#if __GNUC__
872	/* Avoid longjmp clobbering */
873	(void) &out;
874	(void) &quotef;
875	(void) &dblquote;
876	(void) &varnest;
877	(void) &arinest;
878	(void) &parenlevel;
879	(void) &oldstyle;
880	(void) &prevsyntax;
881	(void) &syntax;
882#endif
883
884	startlinno = plinno;
885	dblquote = 0;
886	if (syntax == DQSYNTAX)
887		dblquote = 1;
888	quotef = 0;
889	bqlist = NULL;
890	varnest = 0;
891	arinest = 0;
892	parenlevel = 0;
893
894	STARTSTACKSTR(out);
895	loop: {	/* for each line, until end of word */
896#if ATTY
897		if (c == '\034' && doprompt
898		 && attyset() && ! equal(termval(), "emacs")) {
899			attyline();
900			if (syntax == BASESYNTAX)
901				return readtoken();
902			c = pgetc();
903			goto loop;
904		}
905#endif
906		CHECKEND();	/* set c to PEOF if at end of here document */
907		for (;;) {	/* until end of line or end of word */
908			CHECKSTRSPACE(3, out);	/* permit 3 calls to USTPUTC */
909			switch(syntax[c]) {
910			case CNL:	/* '\n' */
911				if (syntax == BASESYNTAX)
912					goto endword;	/* exit outer loop */
913				USTPUTC(c, out);
914				plinno++;
915				if (doprompt)
916					setprompt(2);
917				else
918					setprompt(0);
919				c = pgetc();
920				goto loop;		/* continue outer loop */
921			case CWORD:
922				USTPUTC(c, out);
923				break;
924			case CCTL:
925				if (eofmark == NULL || dblquote)
926					USTPUTC(CTLESC, out);
927				USTPUTC(c, out);
928				break;
929			case CBACK:	/* backslash */
930				c = pgetc();
931				if (c == PEOF) {
932					USTPUTC('\\', out);
933					pungetc();
934				} else if (c == '\n') {
935					if (doprompt)
936						setprompt(2);
937					else
938						setprompt(0);
939				} else {
940					if (dblquote && c != '\\' && c != '`' && c != '$'
941							 && (c != '"' || eofmark != NULL))
942						USTPUTC('\\', out);
943					if (SQSYNTAX[c] == CCTL)
944						USTPUTC(CTLESC, out);
945					else if (eofmark == NULL)
946						USTPUTC(CTLQUOTEMARK, out);
947					USTPUTC(c, out);
948					quotef++;
949				}
950				break;
951			case CSQUOTE:
952				if (eofmark == NULL)
953					USTPUTC(CTLQUOTEMARK, out);
954				syntax = SQSYNTAX;
955				break;
956			case CDQUOTE:
957				if (eofmark == NULL)
958					USTPUTC(CTLQUOTEMARK, out);
959				syntax = DQSYNTAX;
960				dblquote = 1;
961				break;
962			case CENDQUOTE:
963				if (eofmark != NULL && arinest == 0 &&
964				    varnest == 0) {
965					USTPUTC(c, out);
966				} else {
967					if (arinest) {
968						syntax = ARISYNTAX;
969						dblquote = 0;
970					} else if (eofmark == NULL) {
971						syntax = BASESYNTAX;
972						dblquote = 0;
973					}
974					quotef++;
975				}
976				break;
977			case CVAR:	/* '$' */
978				PARSESUB();		/* parse substitution */
979				break;
980			case CENDVAR:	/* '}' */
981				if (varnest > 0) {
982					varnest--;
983					USTPUTC(CTLENDVAR, out);
984				} else {
985					USTPUTC(c, out);
986				}
987				break;
988			case CLP:	/* '(' in arithmetic */
989				parenlevel++;
990				USTPUTC(c, out);
991				break;
992			case CRP:	/* ')' in arithmetic */
993				if (parenlevel > 0) {
994					USTPUTC(c, out);
995					--parenlevel;
996				} else {
997					if (pgetc() == ')') {
998						if (--arinest == 0) {
999							USTPUTC(CTLENDARI, out);
1000							syntax = prevsyntax;
1001							if (syntax == DQSYNTAX)
1002								dblquote = 1;
1003							else
1004								dblquote = 0;
1005						} else
1006							USTPUTC(')', out);
1007					} else {
1008						/*
1009						 * unbalanced parens
1010						 *  (don't 2nd guess - no error)
1011						 */
1012						pungetc();
1013						USTPUTC(')', out);
1014					}
1015				}
1016				break;
1017			case CBQUOTE:	/* '`' */
1018				PARSEBACKQOLD();
1019				break;
1020			case CEOF:
1021				goto endword;		/* exit outer loop */
1022			default:
1023				if (varnest == 0)
1024					goto endword;	/* exit outer loop */
1025				USTPUTC(c, out);
1026			}
1027			c = pgetc_macro();
1028		}
1029	}
1030endword:
1031	if (syntax == ARISYNTAX)
1032		synerror("Missing '))'");
1033	if (syntax != BASESYNTAX && ! parsebackquote && eofmark == NULL)
1034		synerror("Unterminated quoted string");
1035	if (varnest != 0) {
1036		startlinno = plinno;
1037		synerror("Missing '}'");
1038	}
1039	USTPUTC('\0', out);
1040	len = out - stackblock();
1041	out = stackblock();
1042	if (eofmark == NULL) {
1043		if ((c == '>' || c == '<')
1044		 && quotef == 0
1045		 && len <= 2
1046		 && (*out == '\0' || is_digit(*out))) {
1047			PARSEREDIR();
1048			return lasttoken = TREDIR;
1049		} else {
1050			pungetc();
1051		}
1052	}
1053	quoteflag = quotef;
1054	backquotelist = bqlist;
1055	grabstackblock(len);
1056	wordtext = out;
1057	return lasttoken = TWORD;
1058/* end of readtoken routine */
1059
1060
1061
1062/*
1063 * Check to see whether we are at the end of the here document.  When this
1064 * is called, c is set to the first character of the next input line.  If
1065 * we are at the end of the here document, this routine sets the c to PEOF.
1066 */
1067
1068checkend: {
1069	if (eofmark) {
1070		if (striptabs) {
1071			while (c == '\t')
1072				c = pgetc();
1073		}
1074		if (c == *eofmark) {
1075			if (pfgets(line, sizeof line) != NULL) {
1076				char *p, *q;
1077
1078				p = line;
1079				for (q = eofmark + 1 ; *q && *p == *q ; p++, q++);
1080				if (*p == '\n' && *q == '\0') {
1081					c = PEOF;
1082					plinno++;
1083					needprompt = doprompt;
1084				} else {
1085					pushstring(line, strlen(line), NULL);
1086				}
1087			}
1088		}
1089	}
1090	goto checkend_return;
1091}
1092
1093
1094/*
1095 * Parse a redirection operator.  The variable "out" points to a string
1096 * specifying the fd to be redirected.  The variable "c" contains the
1097 * first character of the redirection operator.
1098 */
1099
1100parseredir: {
1101	char fd = *out;
1102	union node *np;
1103
1104	np = (union node *)stalloc(sizeof (struct nfile));
1105	if (c == '>') {
1106		np->nfile.fd = 1;
1107		c = pgetc();
1108		if (c == '>')
1109			np->type = NAPPEND;
1110		else if (c == '&')
1111			np->type = NTOFD;
1112		else {
1113			np->type = NTO;
1114			pungetc();
1115		}
1116	} else {	/* c == '<' */
1117		np->nfile.fd = 0;
1118		c = pgetc();
1119		if (c == '<') {
1120			if (sizeof (struct nfile) != sizeof (struct nhere)) {
1121				np = (union node *)stalloc(sizeof (struct nhere));
1122				np->nfile.fd = 0;
1123			}
1124			np->type = NHERE;
1125			heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc));
1126			heredoc->here = np;
1127			if ((c = pgetc()) == '-') {
1128				heredoc->striptabs = 1;
1129			} else {
1130				heredoc->striptabs = 0;
1131				pungetc();
1132			}
1133		} else if (c == '&')
1134			np->type = NFROMFD;
1135		else {
1136			np->type = NFROM;
1137			pungetc();
1138		}
1139	}
1140	if (fd != '\0')
1141		np->nfile.fd = digit_val(fd);
1142	redirnode = np;
1143	goto parseredir_return;
1144}
1145
1146
1147/*
1148 * Parse a substitution.  At this point, we have read the dollar sign
1149 * and nothing else.
1150 */
1151
1152parsesub: {
1153	int subtype;
1154	int typeloc;
1155	int flags;
1156	char *p;
1157#ifndef GDB_HACK
1158	static const char types[] = "}-+?=";
1159#endif
1160       int bracketed_name = 0; /* used to handle ${[0-9]*} variables */
1161
1162	c = pgetc();
1163	if (c != '(' && c != '{' && !is_name(c) && !is_special(c)) {
1164		USTPUTC('$', out);
1165		pungetc();
1166	} else if (c == '(') {	/* $(command) or $((arith)) */
1167		if (pgetc() == '(') {
1168			PARSEARITH();
1169		} else {
1170			pungetc();
1171			PARSEBACKQNEW();
1172		}
1173	} else {
1174		USTPUTC(CTLVAR, out);
1175		typeloc = out - stackblock();
1176		USTPUTC(VSNORMAL, out);
1177		subtype = VSNORMAL;
1178		if (c == '{') {
1179			bracketed_name = 1;
1180			c = pgetc();
1181			if (c == '#') {
1182				if ((c = pgetc()) == '}')
1183					c = '#';
1184				else
1185					subtype = VSLENGTH;
1186			}
1187			else
1188				subtype = 0;
1189		}
1190		if (is_name(c)) {
1191			do {
1192				STPUTC(c, out);
1193				c = pgetc();
1194			} while (is_in_name(c));
1195		} else if (is_digit(c)) {
1196			if (bracketed_name) {
1197				do {
1198					STPUTC(c, out);
1199					c = pgetc();
1200				} while (is_digit(c));
1201			} else {
1202				STPUTC(c, out);
1203				c = pgetc();
1204			}
1205		} else {
1206			if (! is_special(c))
1207badsub:				synerror("Bad substitution");
1208			USTPUTC(c, out);
1209			c = pgetc();
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
1259parsebackq: {
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#if __GNUC__
1269	/* Avoid longjmp clobbering */
1270	(void) &saveprompt;
1271#endif
1272
1273	savepbq = parsebackquote;
1274	if (setjmp(jmploc.loc)) {
1275		if (str)
1276			ckfree(str);
1277		parsebackquote = 0;
1278		handler = savehandler;
1279		longjmp(handler->loc, 1);
1280	}
1281	INTOFF;
1282	str = NULL;
1283	savelen = out - stackblock();
1284	if (savelen > 0) {
1285		str = ckmalloc(savelen);
1286		memcpy(str, stackblock(), savelen);
1287	}
1288	savehandler = handler;
1289	handler = &jmploc;
1290	INTON;
1291        if (oldstyle) {
1292                /* We must read until the closing backquote, giving special
1293                   treatment to some slashes, and then push the string and
1294                   reread it as input, interpreting it normally.  */
1295                char *out;
1296                int c;
1297                int savelen;
1298                char *str;
1299
1300
1301                STARTSTACKSTR(out);
1302		for (;;) {
1303			if (needprompt) {
1304				setprompt(2);
1305				needprompt = 0;
1306			}
1307			switch (c = pgetc()) {
1308			case '`':
1309				goto done;
1310
1311			case '\\':
1312                                if ((c = pgetc()) == '\n') {
1313					plinno++;
1314					if (doprompt)
1315						setprompt(2);
1316					else
1317						setprompt(0);
1318					/*
1319					 * If eating a newline, avoid putting
1320					 * the newline into the new character
1321					 * stream (via the STPUTC after the
1322					 * switch).
1323					 */
1324					continue;
1325				}
1326                                if (c != '\\' && c != '`' && c != '$'
1327                                    && (!dblquote || c != '"'))
1328                                        STPUTC('\\', out);
1329				break;
1330
1331			case '\n':
1332				plinno++;
1333				needprompt = doprompt;
1334				break;
1335
1336			case PEOF:
1337			        startlinno = plinno;
1338				synerror("EOF in backquote substitution");
1339 				break;
1340
1341			default:
1342				break;
1343			}
1344			STPUTC(c, out);
1345                }
1346done:
1347                STPUTC('\0', out);
1348                savelen = out - stackblock();
1349                if (savelen > 0) {
1350                        str = ckmalloc(savelen);
1351                        memcpy(str, stackblock(), savelen);
1352			setinputstring(str, 1);
1353                }
1354        }
1355	nlpp = &bqlist;
1356	while (*nlpp)
1357		nlpp = &(*nlpp)->next;
1358	*nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist));
1359	(*nlpp)->next = NULL;
1360	parsebackquote = oldstyle;
1361
1362	if (oldstyle) {
1363		saveprompt = doprompt;
1364		doprompt = 0;
1365	}
1366
1367	n = list(0);
1368
1369	if (oldstyle)
1370		doprompt = saveprompt;
1371	else {
1372		if (readtoken() != TRP)
1373			synexpect(TRP);
1374	}
1375
1376	(*nlpp)->n = n;
1377        if (oldstyle) {
1378		/*
1379		 * Start reading from old file again, ignoring any pushed back
1380		 * tokens left from the backquote parsing
1381		 */
1382                popfile();
1383		tokpushback = 0;
1384	}
1385	while (stackblocksize() <= savelen)
1386		growstackblock();
1387	STARTSTACKSTR(out);
1388	if (str) {
1389		memcpy(out, str, savelen);
1390		STADJUST(savelen, out);
1391		INTOFF;
1392		ckfree(str);
1393		str = NULL;
1394		INTON;
1395	}
1396	parsebackquote = savepbq;
1397	handler = savehandler;
1398	if (arinest || dblquote)
1399		USTPUTC(CTLBACKQ | CTLQUOTE, out);
1400	else
1401		USTPUTC(CTLBACKQ, out);
1402	if (oldstyle)
1403		goto parsebackq_oldreturn;
1404	else
1405		goto parsebackq_newreturn;
1406}
1407
1408/*
1409 * Parse an arithmetic expansion (indicate start of one and set state)
1410 */
1411parsearith: {
1412
1413	if (++arinest == 1) {
1414		prevsyntax = syntax;
1415		syntax = ARISYNTAX;
1416		USTPUTC(CTLARI, out);
1417		if (dblquote)
1418			USTPUTC('"',out);
1419		else
1420			USTPUTC(' ',out);
1421	} else {
1422		/*
1423		 * we collapse embedded arithmetic expansion to
1424		 * parenthesis, which should be equivalent
1425		 */
1426		USTPUTC('(', out);
1427	}
1428	goto parsearith_return;
1429}
1430
1431} /* end of readtoken */
1432
1433
1434
1435#ifdef mkinit
1436RESET {
1437	tokpushback = 0;
1438	checkkwd = 0;
1439}
1440#endif
1441
1442/*
1443 * Returns true if the text contains nothing to expand (no dollar signs
1444 * or backquotes).
1445 */
1446
1447STATIC int
1448noexpand(text)
1449	char *text;
1450	{
1451	char *p;
1452	char c;
1453
1454	p = text;
1455	while ((c = *p++) != '\0') {
1456		if ( c == CTLQUOTEMARK)
1457			continue;
1458		if (c == CTLESC)
1459			p++;
1460		else if (BASESYNTAX[c] == CCTL)
1461			return 0;
1462	}
1463	return 1;
1464}
1465
1466
1467/*
1468 * Return true if the argument is a legal variable name (a letter or
1469 * underscore followed by zero or more letters, underscores, and digits).
1470 */
1471
1472int
1473goodname(name)
1474	char *name;
1475	{
1476	char *p;
1477
1478	p = name;
1479	if (! is_name(*p))
1480		return 0;
1481	while (*++p) {
1482		if (! is_in_name(*p))
1483			return 0;
1484	}
1485	return 1;
1486}
1487
1488
1489/*
1490 * Called when an unexpected token is read during the parse.  The argument
1491 * is the token that is expected, or -1 if more than one type of token can
1492 * occur at this point.
1493 */
1494
1495STATIC void
1496synexpect(token)
1497	int token;
1498{
1499	char msg[64];
1500
1501	if (token >= 0) {
1502		fmtstr(msg, 64, "%s unexpected (expecting %s)",
1503			tokname[lasttoken], tokname[token]);
1504	} else {
1505		fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]);
1506	}
1507	synerror(msg);
1508}
1509
1510
1511STATIC void
1512synerror(msg)
1513	char *msg;
1514	{
1515	if (commandname)
1516		outfmt(&errout, "%s: %d: ", commandname, startlinno);
1517	outfmt(&errout, "Syntax error: %s\n", msg);
1518	error((char *)NULL);
1519}
1520
1521STATIC void
1522setprompt(which)
1523	int which;
1524	{
1525	whichprompt = which;
1526
1527#ifndef NO_HISTORY
1528	if (!el)
1529#endif
1530		out2str(getprompt(NULL));
1531}
1532
1533/*
1534 * called by editline -- any expansions to the prompt
1535 *    should be added here.
1536 */
1537char *
1538getprompt(unused)
1539	void *unused __unused;
1540{
1541	switch (whichprompt) {
1542	case 0:
1543		return "";
1544	case 1:
1545		return ps1val();
1546	case 2:
1547		return ps2val();
1548	default:
1549		return "<internal prompt error>";
1550	}
1551}
1552