parser.c revision 38887 
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	"$Id: parser.c,v 1.22 1998/05/18 06:44:12 charnier Exp $";
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		rmquotes0(wordtext);
623		if (here->striptabs) {
624			while (*wordtext == '\t')
625				wordtext++;
626		}
627		if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN)
628			synerror("Illegal eof marker for << redirection");
629		rmescapes(wordtext);
630		here->eofmark = wordtext;
631		here->next = NULL;
632		if (heredoclist == NULL)
633			heredoclist = here;
634		else {
635			for (p = heredoclist ; p->next ; p = p->next);
636			p->next = here;
637		}
638	} else if (n->type == NTOFD || n->type == NFROMFD) {
639		fixredir(n, wordtext, 0);
640	} else {
641		n->nfile.fname = makename();
642	}
643}
644
645
646/*
647 * Input any here documents.
648 */
649
650STATIC void
651parseheredoc() {
652	struct heredoc *here;
653	union node *n;
654
655	while (heredoclist) {
656		here = heredoclist;
657		heredoclist = here->next;
658		if (needprompt) {
659			setprompt(2);
660			needprompt = 0;
661		}
662		readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX,
663				here->eofmark, here->striptabs);
664		n = (union node *)stalloc(sizeof (struct narg));
665		n->narg.type = NARG;
666		n->narg.next = NULL;
667		n->narg.text = wordtext;
668		n->narg.backquote = backquotelist;
669		here->here->nhere.doc = n;
670	}
671}
672
673STATIC int
674peektoken() {
675	int t;
676
677	t = readtoken();
678	tokpushback++;
679	return (t);
680}
681
682STATIC int xxreadtoken();
683
684STATIC int
685readtoken() {
686	int t;
687	int savecheckkwd = checkkwd;
688	struct alias *ap;
689#ifdef DEBUG
690	int alreadyseen = tokpushback;
691#endif
692
693	top:
694	t = xxreadtoken();
695
696	if (checkkwd) {
697		/*
698		 * eat newlines
699		 */
700		if (checkkwd == 2) {
701			checkkwd = 0;
702			while (t == TNL) {
703				parseheredoc();
704				t = xxreadtoken();
705			}
706		} else
707			checkkwd = 0;
708		/*
709		 * check for keywords and aliases
710		 */
711		if (t == TWORD && !quoteflag)
712		{
713			char * const *pp;
714
715			for (pp = (char **)parsekwd; *pp; pp++) {
716				if (**pp == *wordtext && equal(*pp, wordtext))
717				{
718					lasttoken = t = pp - parsekwd + KWDOFFSET;
719					TRACE(("keyword %s recognized\n", tokname[t]));
720					goto out;
721				}
722			}
723			if (noaliases == 0 &&
724			    (ap = lookupalias(wordtext, 1)) != NULL) {
725				pushstring(ap->val, strlen(ap->val), ap);
726				checkkwd = savecheckkwd;
727				goto top;
728			}
729		}
730out:
731		checkkwd = 0;
732	}
733#ifdef DEBUG
734	if (!alreadyseen)
735	    TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
736	else
737	    TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
738#endif
739	return (t);
740}
741
742
743/*
744 * Read the next input token.
745 * If the token is a word, we set backquotelist to the list of cmds in
746 *	backquotes.  We set quoteflag to true if any part of the word was
747 *	quoted.
748 * If the token is TREDIR, then we set redirnode to a structure containing
749 *	the redirection.
750 * In all cases, the variable startlinno is set to the number of the line
751 *	on which the token starts.
752 *
753 * [Change comment:  here documents and internal procedures]
754 * [Readtoken shouldn't have any arguments.  Perhaps we should make the
755 *  word parsing code into a separate routine.  In this case, readtoken
756 *  doesn't need to have any internal procedures, but parseword does.
757 *  We could also make parseoperator in essence the main routine, and
758 *  have parseword (readtoken1?) handle both words and redirection.]
759 */
760
761#define RETURN(token)	return lasttoken = token
762
763STATIC int
764xxreadtoken() {
765	int c;
766
767	if (tokpushback) {
768		tokpushback = 0;
769		return lasttoken;
770	}
771	if (needprompt) {
772		setprompt(2);
773		needprompt = 0;
774	}
775	startlinno = plinno;
776	for (;;) {	/* until token or start of word found */
777		c = pgetc_macro();
778		if (c == ' ' || c == '\t')
779			continue;		/* quick check for white space first */
780		switch (c) {
781		case ' ': case '\t':
782			continue;
783		case '#':
784			while ((c = pgetc()) != '\n' && c != PEOF);
785			pungetc();
786			continue;
787		case '\\':
788			if (pgetc() == '\n') {
789				startlinno = ++plinno;
790				if (doprompt)
791					setprompt(2);
792				else
793					setprompt(0);
794				continue;
795			}
796			pungetc();
797			goto breakloop;
798		case '\n':
799			plinno++;
800			needprompt = doprompt;
801			RETURN(TNL);
802		case PEOF:
803			RETURN(TEOF);
804		case '&':
805			if (pgetc() == '&')
806				RETURN(TAND);
807			pungetc();
808			RETURN(TBACKGND);
809		case '|':
810			if (pgetc() == '|')
811				RETURN(TOR);
812			pungetc();
813			RETURN(TPIPE);
814		case ';':
815			if (pgetc() == ';')
816				RETURN(TENDCASE);
817			pungetc();
818			RETURN(TSEMI);
819		case '(':
820			RETURN(TLP);
821		case ')':
822			RETURN(TRP);
823		default:
824			goto breakloop;
825		}
826	}
827breakloop:
828	return readtoken1(c, BASESYNTAX, (char *)NULL, 0);
829#undef RETURN
830}
831
832
833
834/*
835 * If eofmark is NULL, read a word or a redirection symbol.  If eofmark
836 * is not NULL, read a here document.  In the latter case, eofmark is the
837 * word which marks the end of the document and striptabs is true if
838 * leading tabs should be stripped from the document.  The argument firstc
839 * is the first character of the input token or document.
840 *
841 * Because C does not have internal subroutines, I have simulated them
842 * using goto's to implement the subroutine linkage.  The following macros
843 * will run code that appears at the end of readtoken1.
844 */
845
846#define CHECKEND()	{goto checkend; checkend_return:;}
847#define PARSEREDIR()	{goto parseredir; parseredir_return:;}
848#define PARSESUB()	{goto parsesub; parsesub_return:;}
849#define PARSEBACKQOLD()	{oldstyle = 1; goto parsebackq; parsebackq_oldreturn:;}
850#define PARSEBACKQNEW()	{oldstyle = 0; goto parsebackq; parsebackq_newreturn:;}
851#define	PARSEARITH()	{goto parsearith; parsearith_return:;}
852
853STATIC int
854readtoken1(firstc, syntax, eofmark, striptabs)
855	int firstc;
856	char const *syntax;
857	char *eofmark;
858	int striptabs;
859	{
860	int c = firstc;
861	char *out;
862	int len;
863	char line[EOFMARKLEN + 1];
864	struct nodelist *bqlist;
865	int quotef;
866	int dblquote;
867	int varnest;	/* levels of variables expansion */
868	int arinest;	/* levels of arithmetic expansion */
869	int parenlevel;	/* levels of parens in arithmetic */
870	int oldstyle;
871	char const *prevsyntax;	/* syntax before arithmetic */
872#if __GNUC__
873	/* Avoid longjmp clobbering */
874	(void) &out;
875	(void) &quotef;
876	(void) &dblquote;
877	(void) &varnest;
878	(void) &arinest;
879	(void) &parenlevel;
880	(void) &oldstyle;
881	(void) &prevsyntax;
882	(void) &syntax;
883#endif
884
885	startlinno = plinno;
886	dblquote = 0;
887	if (syntax == DQSYNTAX)
888		dblquote = 1;
889	quotef = 0;
890	bqlist = NULL;
891	varnest = 0;
892	arinest = 0;
893	parenlevel = 0;
894
895	STARTSTACKSTR(out);
896	loop: {	/* for each line, until end of word */
897#if ATTY
898		if (c == '\034' && doprompt
899		 && attyset() && ! equal(termval(), "emacs")) {
900			attyline();
901			if (syntax == BASESYNTAX)
902				return readtoken();
903			c = pgetc();
904			goto loop;
905		}
906#endif
907		CHECKEND();	/* set c to PEOF if at end of here document */
908		for (;;) {	/* until end of line or end of word */
909			CHECKSTRSPACE(3, out);	/* permit 3 calls to USTPUTC */
910			switch(syntax[c]) {
911			case CNL:	/* '\n' */
912				if (syntax == BASESYNTAX)
913					goto endword;	/* exit outer loop */
914				USTPUTC(c, out);
915				plinno++;
916				if (doprompt)
917					setprompt(2);
918				else
919					setprompt(0);
920				c = pgetc();
921				goto loop;		/* continue outer loop */
922			case CWORD:
923				USTPUTC(c, out);
924				break;
925			case CCTL:
926				if (eofmark == NULL || dblquote)
927					USTPUTC(CTLESC, out);
928				USTPUTC(c, out);
929				break;
930			case CBACK:	/* backslash */
931				c = pgetc();
932				if (c == PEOF) {
933					USTPUTC('\\', out);
934					pungetc();
935				} else if (c == '\n') {
936					if (doprompt)
937						setprompt(2);
938					else
939						setprompt(0);
940				} else {
941					if (dblquote && c != '\\' && c != '`' && c != '$'
942							 && (c != '"' || eofmark != NULL))
943						USTPUTC('\\', out);
944					if (SQSYNTAX[c] == CCTL)
945						USTPUTC(CTLESC, out);
946					else
947						USTPUTC(CTLQUOTEMARK, out);
948					USTPUTC(c, out);
949					quotef++;
950				}
951				break;
952			case CSQUOTE:
953				USTPUTC(CTLQUOTEMARK, out);
954				syntax = SQSYNTAX;
955				break;
956			case CDQUOTE:
957				USTPUTC(CTLQUOTEMARK, out);
958				syntax = DQSYNTAX;
959				dblquote = 1;
960				break;
961			case CENDQUOTE:
962				if (eofmark) {
963					USTPUTC(c, out);
964				} else {
965					if (arinest)
966						syntax = ARISYNTAX;
967					else
968						syntax = BASESYNTAX;
969					quotef++;
970					dblquote = 0;
971				}
972				break;
973			case CVAR:	/* '$' */
974				PARSESUB();		/* parse substitution */
975				break;
976			case CENDVAR:	/* '}' */
977				if (varnest > 0) {
978					varnest--;
979					USTPUTC(CTLENDVAR, out);
980				} else {
981					USTPUTC(c, out);
982				}
983				break;
984			case CLP:	/* '(' in arithmetic */
985				parenlevel++;
986				USTPUTC(c, out);
987				break;
988			case CRP:	/* ')' in arithmetic */
989				if (parenlevel > 0) {
990					USTPUTC(c, out);
991					--parenlevel;
992				} else {
993					if (pgetc() == ')') {
994						if (--arinest == 0) {
995							USTPUTC(CTLENDARI, out);
996							syntax = prevsyntax;
997						} else
998							USTPUTC(')', out);
999					} else {
1000						/*
1001						 * unbalanced parens
1002						 *  (don't 2nd guess - no error)
1003						 */
1004						pungetc();
1005						USTPUTC(')', out);
1006					}
1007				}
1008				break;
1009			case CBQUOTE:	/* '`' */
1010				PARSEBACKQOLD();
1011				break;
1012			case CEOF:
1013				goto endword;		/* exit outer loop */
1014			default:
1015				if (varnest == 0)
1016					goto endword;	/* exit outer loop */
1017				USTPUTC(c, out);
1018			}
1019			c = pgetc_macro();
1020		}
1021	}
1022endword:
1023	if (syntax == ARISYNTAX)
1024		synerror("Missing '))'");
1025	if (syntax != BASESYNTAX && ! parsebackquote && eofmark == NULL)
1026		synerror("Unterminated quoted string");
1027	if (varnest != 0) {
1028		startlinno = plinno;
1029		synerror("Missing '}'");
1030	}
1031	USTPUTC('\0', out);
1032	len = out - stackblock();
1033	out = stackblock();
1034	if (eofmark == NULL) {
1035		if ((c == '>' || c == '<')
1036		 && quotef == 0
1037		 && len <= 2
1038		 && (*out == '\0' || is_digit(*out))) {
1039			PARSEREDIR();
1040			return lasttoken = TREDIR;
1041		} else {
1042			pungetc();
1043		}
1044	}
1045	quoteflag = quotef;
1046	backquotelist = bqlist;
1047	grabstackblock(len);
1048	wordtext = out;
1049	return lasttoken = TWORD;
1050/* end of readtoken routine */
1051
1052
1053
1054/*
1055 * Check to see whether we are at the end of the here document.  When this
1056 * is called, c is set to the first character of the next input line.  If
1057 * we are at the end of the here document, this routine sets the c to PEOF.
1058 */
1059
1060checkend: {
1061	if (eofmark) {
1062		if (striptabs) {
1063			while (c == '\t')
1064				c = pgetc();
1065		}
1066		if (c == *eofmark) {
1067			if (pfgets(line, sizeof line) != NULL) {
1068				char *p, *q;
1069
1070				p = line;
1071				for (q = eofmark + 1 ; *q && *p == *q ; p++, q++);
1072				if (*p == '\n' && *q == '\0') {
1073					c = PEOF;
1074					plinno++;
1075					needprompt = doprompt;
1076				} else {
1077					pushstring(line, strlen(line), NULL);
1078				}
1079			}
1080		}
1081	}
1082	goto checkend_return;
1083}
1084
1085
1086/*
1087 * Parse a redirection operator.  The variable "out" points to a string
1088 * specifying the fd to be redirected.  The variable "c" contains the
1089 * first character of the redirection operator.
1090 */
1091
1092parseredir: {
1093	char fd = *out;
1094	union node *np;
1095
1096	np = (union node *)stalloc(sizeof (struct nfile));
1097	if (c == '>') {
1098		np->nfile.fd = 1;
1099		c = pgetc();
1100		if (c == '>')
1101			np->type = NAPPEND;
1102		else if (c == '&')
1103			np->type = NTOFD;
1104		else {
1105			np->type = NTO;
1106			pungetc();
1107		}
1108	} else {	/* c == '<' */
1109		np->nfile.fd = 0;
1110		c = pgetc();
1111		if (c == '<') {
1112			if (sizeof (struct nfile) != sizeof (struct nhere)) {
1113				np = (union node *)stalloc(sizeof (struct nhere));
1114				np->nfile.fd = 0;
1115			}
1116			np->type = NHERE;
1117			heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc));
1118			heredoc->here = np;
1119			if ((c = pgetc()) == '-') {
1120				heredoc->striptabs = 1;
1121			} else {
1122				heredoc->striptabs = 0;
1123				pungetc();
1124			}
1125		} else if (c == '&')
1126			np->type = NFROMFD;
1127		else {
1128			np->type = NFROM;
1129			pungetc();
1130		}
1131	}
1132	if (fd != '\0')
1133		np->nfile.fd = digit_val(fd);
1134	redirnode = np;
1135	goto parseredir_return;
1136}
1137
1138
1139/*
1140 * Parse a substitution.  At this point, we have read the dollar sign
1141 * and nothing else.
1142 */
1143
1144parsesub: {
1145	int subtype;
1146	int typeloc;
1147	int flags;
1148	char *p;
1149#ifndef GDB_HACK
1150	static const char types[] = "}-+?=";
1151#endif
1152       int bracketed_name = 0; /* used to handle ${[0-9]*} variables */
1153
1154	c = pgetc();
1155	if (c != '(' && c != '{' && !is_name(c) && !is_special(c)) {
1156		USTPUTC('$', out);
1157		pungetc();
1158	} else if (c == '(') {	/* $(command) or $((arith)) */
1159		if (pgetc() == '(') {
1160			PARSEARITH();
1161		} else {
1162			pungetc();
1163			PARSEBACKQNEW();
1164		}
1165	} else {
1166		USTPUTC(CTLVAR, out);
1167		typeloc = out - stackblock();
1168		USTPUTC(VSNORMAL, out);
1169		subtype = VSNORMAL;
1170		if (c == '{') {
1171			bracketed_name = 1;
1172			c = pgetc();
1173			if (c == '#') {
1174				if ((c = pgetc()) == '}')
1175					c = '#';
1176				else
1177					subtype = VSLENGTH;
1178			}
1179			else
1180				subtype = 0;
1181		}
1182		if (is_name(c)) {
1183			do {
1184				STPUTC(c, out);
1185				c = pgetc();
1186			} while (is_in_name(c));
1187		} else if (is_digit(c)) {
1188			if (bracketed_name) {
1189				do {
1190					STPUTC(c, out);
1191					c = pgetc();
1192				} while (is_digit(c));
1193			} else {
1194				STPUTC(c, out);
1195				c = pgetc();
1196			}
1197		} else {
1198			if (! is_special(c))
1199badsub:				synerror("Bad substitution");
1200			USTPUTC(c, out);
1201			c = pgetc();
1202		}
1203		STPUTC('=', out);
1204		flags = 0;
1205		if (subtype == 0) {
1206			switch (c) {
1207			case ':':
1208				flags = VSNUL;
1209				c = pgetc();
1210				/*FALLTHROUGH*/
1211			default:
1212				p = strchr(types, c);
1213				if (p == NULL)
1214					goto badsub;
1215				subtype = p - types + VSNORMAL;
1216				break;
1217			case '%':
1218			case '#':
1219				{
1220					int cc = c;
1221					subtype = c == '#' ? VSTRIMLEFT :
1222							     VSTRIMRIGHT;
1223					c = pgetc();
1224					if (c == cc)
1225						subtype++;
1226					else
1227						pungetc();
1228					break;
1229				}
1230			}
1231		} else {
1232			pungetc();
1233		}
1234		if (dblquote || arinest)
1235			flags |= VSQUOTE;
1236		*(stackblock() + typeloc) = subtype | flags;
1237		if (subtype != VSNORMAL)
1238			varnest++;
1239	}
1240	goto parsesub_return;
1241}
1242
1243
1244/*
1245 * Called to parse command substitutions.  Newstyle is set if the command
1246 * is enclosed inside $(...); nlpp is a pointer to the head of the linked
1247 * list of commands (passed by reference), and savelen is the number of
1248 * characters on the top of the stack which must be preserved.
1249 */
1250
1251parsebackq: {
1252	struct nodelist **nlpp;
1253	int savepbq;
1254	union node *n;
1255	char *volatile str;
1256	struct jmploc jmploc;
1257	struct jmploc *volatile savehandler;
1258	int savelen;
1259	int saveprompt;
1260#if __GNUC__
1261	/* Avoid longjmp clobbering */
1262	(void) &saveprompt;
1263#endif
1264
1265	savepbq = parsebackquote;
1266	if (setjmp(jmploc.loc)) {
1267		if (str)
1268			ckfree(str);
1269		parsebackquote = 0;
1270		handler = savehandler;
1271		longjmp(handler->loc, 1);
1272	}
1273	INTOFF;
1274	str = NULL;
1275	savelen = out - stackblock();
1276	if (savelen > 0) {
1277		str = ckmalloc(savelen);
1278		memcpy(str, stackblock(), savelen);
1279	}
1280	savehandler = handler;
1281	handler = &jmploc;
1282	INTON;
1283        if (oldstyle) {
1284                /* We must read until the closing backquote, giving special
1285                   treatment to some slashes, and then push the string and
1286                   reread it as input, interpreting it normally.  */
1287                char *out;
1288                int c;
1289                int savelen;
1290                char *str;
1291
1292
1293                STARTSTACKSTR(out);
1294		for (;;) {
1295			if (needprompt) {
1296				setprompt(2);
1297				needprompt = 0;
1298			}
1299			switch (c = pgetc()) {
1300			case '`':
1301				goto done;
1302
1303			case '\\':
1304                                if ((c = pgetc()) == '\n') {
1305					plinno++;
1306					if (doprompt)
1307						setprompt(2);
1308					else
1309						setprompt(0);
1310					/*
1311					 * If eating a newline, avoid putting
1312					 * the newline into the new character
1313					 * stream (via the STPUTC after the
1314					 * switch).
1315					 */
1316					continue;
1317				}
1318                                if (c != '\\' && c != '`' && c != '$'
1319                                    && (!dblquote || c != '"'))
1320                                        STPUTC('\\', out);
1321				break;
1322
1323			case '\n':
1324				plinno++;
1325				needprompt = doprompt;
1326				break;
1327
1328			case PEOF:
1329			        startlinno = plinno;
1330				synerror("EOF in backquote substitution");
1331 				break;
1332
1333			default:
1334				break;
1335			}
1336			STPUTC(c, out);
1337                }
1338done:
1339                STPUTC('\0', out);
1340                savelen = out - stackblock();
1341                if (savelen > 0) {
1342                        str = ckmalloc(savelen);
1343                        memcpy(str, stackblock(), savelen);
1344			setinputstring(str, 1);
1345                }
1346        }
1347	nlpp = &bqlist;
1348	while (*nlpp)
1349		nlpp = &(*nlpp)->next;
1350	*nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist));
1351	(*nlpp)->next = NULL;
1352	parsebackquote = oldstyle;
1353
1354	if (oldstyle) {
1355		saveprompt = doprompt;
1356		doprompt = 0;
1357	}
1358
1359	n = list(0);
1360
1361	if (oldstyle)
1362		doprompt = saveprompt;
1363	else {
1364		if (readtoken() != TRP)
1365			synexpect(TRP);
1366	}
1367
1368	(*nlpp)->n = n;
1369        if (oldstyle) {
1370		/*
1371		 * Start reading from old file again, ignoring any pushed back
1372		 * tokens left from the backquote parsing
1373		 */
1374                popfile();
1375		tokpushback = 0;
1376	}
1377	while (stackblocksize() <= savelen)
1378		growstackblock();
1379	STARTSTACKSTR(out);
1380	if (str) {
1381		memcpy(out, str, savelen);
1382		STADJUST(savelen, out);
1383		INTOFF;
1384		ckfree(str);
1385		str = NULL;
1386		INTON;
1387	}
1388	parsebackquote = savepbq;
1389	handler = savehandler;
1390	if (arinest || dblquote)
1391		USTPUTC(CTLBACKQ | CTLQUOTE, out);
1392	else
1393		USTPUTC(CTLBACKQ, out);
1394	if (oldstyle)
1395		goto parsebackq_oldreturn;
1396	else
1397		goto parsebackq_newreturn;
1398}
1399
1400/*
1401 * Parse an arithmetic expansion (indicate start of one and set state)
1402 */
1403parsearith: {
1404
1405	if (++arinest == 1) {
1406		prevsyntax = syntax;
1407		syntax = ARISYNTAX;
1408		USTPUTC(CTLARI, out);
1409		if (dblquote)
1410			USTPUTC('"',out);
1411		else
1412			USTPUTC(' ',out);
1413	} else {
1414		/*
1415		 * we collapse embedded arithmetic expansion to
1416		 * parenthesis, which should be equivalent
1417		 */
1418		USTPUTC('(', out);
1419	}
1420	goto parsearith_return;
1421}
1422
1423} /* end of readtoken */
1424
1425
1426
1427#ifdef mkinit
1428RESET {
1429	tokpushback = 0;
1430	checkkwd = 0;
1431}
1432#endif
1433
1434/*
1435 * Returns true if the text contains nothing to expand (no dollar signs
1436 * or backquotes).
1437 */
1438
1439STATIC int
1440noexpand(text)
1441	char *text;
1442	{
1443	char *p;
1444	char c;
1445
1446	p = text;
1447	while ((c = *p++) != '\0') {
1448		if (c == CTLESC)
1449			p++;
1450		else if (BASESYNTAX[c] == CCTL)
1451			return 0;
1452	}
1453	return 1;
1454}
1455
1456
1457/*
1458 * Return true if the argument is a legal variable name (a letter or
1459 * underscore followed by zero or more letters, underscores, and digits).
1460 */
1461
1462int
1463goodname(name)
1464	char *name;
1465	{
1466	char *p;
1467
1468	p = name;
1469	if (! is_name(*p))
1470		return 0;
1471	while (*++p) {
1472		if (! is_in_name(*p))
1473			return 0;
1474	}
1475	return 1;
1476}
1477
1478
1479/*
1480 * Called when an unexpected token is read during the parse.  The argument
1481 * is the token that is expected, or -1 if more than one type of token can
1482 * occur at this point.
1483 */
1484
1485STATIC void
1486synexpect(token)
1487	int token;
1488{
1489	char msg[64];
1490
1491	if (token >= 0) {
1492		fmtstr(msg, 64, "%s unexpected (expecting %s)",
1493			tokname[lasttoken], tokname[token]);
1494	} else {
1495		fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]);
1496	}
1497	synerror(msg);
1498}
1499
1500
1501STATIC void
1502synerror(msg)
1503	char *msg;
1504	{
1505	if (commandname)
1506		outfmt(&errout, "%s: %d: ", commandname, startlinno);
1507	outfmt(&errout, "Syntax error: %s\n", msg);
1508	error((char *)NULL);
1509}
1510
1511STATIC void
1512setprompt(which)
1513	int which;
1514	{
1515	whichprompt = which;
1516
1517#ifndef NO_HISTORY
1518	if (!el)
1519#endif
1520		out2str(getprompt(NULL));
1521}
1522
1523/*
1524 * called by editline -- any expansions to the prompt
1525 *    should be added here.
1526 */
1527char *
1528getprompt(unused)
1529	void *unused __unused;
1530{
1531	switch (whichprompt) {
1532	case 0:
1533		return "";
1534	case 1:
1535		return ps1val();
1536	case 2:
1537		return ps2val();
1538	default:
1539		return "<internal prompt error>";
1540	}
1541}
1542