parser.c revision 25230 
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 *	$Id: parser.c,v 1.19 1997/02/22 13:58:42 peter Exp $
37 */
38
39#ifndef lint
40static char const sccsid[] = "@(#)parser.c	8.7 (Berkeley) 5/16/95";
41#endif /* not lint */
42
43#include <stdlib.h>
44
45#include "shell.h"
46#include "parser.h"
47#include "nodes.h"
48#include "expand.h"	/* defines rmescapes() */
49#include "redir.h"	/* defines copyfd() */
50#include "syntax.h"
51#include "options.h"
52#include "input.h"
53#include "output.h"
54#include "var.h"
55#include "error.h"
56#include "memalloc.h"
57#include "mystring.h"
58#include "alias.h"
59#include "show.h"
60#ifndef NO_HISTORY
61#include "myhistedit.h"
62#endif
63
64/*
65 * Shell command parser.
66 */
67
68#define EOFMARKLEN 79
69
70/* values returned by readtoken */
71#include "token.h"
72
73
74
75struct heredoc {
76	struct heredoc *next;	/* next here document in list */
77	union node *here;		/* redirection node */
78	char *eofmark;		/* string indicating end of input */
79	int striptabs;		/* if set, strip leading tabs */
80};
81
82
83
84struct heredoc *heredoclist;	/* list of here documents to read */
85int parsebackquote;		/* nonzero if we are inside backquotes */
86int doprompt;			/* if set, prompt the user */
87int needprompt;			/* true if interactive and at start of line */
88int lasttoken;			/* last token read */
89MKINIT int tokpushback;		/* last token pushed back */
90char *wordtext;			/* text of last word returned by readtoken */
91MKINIT int checkkwd;            /* 1 == check for kwds, 2 == also eat newlines */
92struct nodelist *backquotelist;
93union node *redirnode;
94struct heredoc *heredoc;
95int quoteflag;			/* set if (part of) last token was quoted */
96int startlinno;			/* line # where last token started */
97
98/* XXX When 'noaliases' is set to one, no alias expansion takes place. */
99static int noaliases = 0;
100
101#define GDB_HACK 1 /* avoid local declarations which gdb can't handle */
102#ifdef GDB_HACK
103static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE, '@', '=', '\0'};
104static const char types[] = "}-+?=";
105#endif
106
107
108STATIC union node *list __P((int));
109STATIC union node *andor __P((void));
110STATIC union node *pipeline __P((void));
111STATIC union node *command __P((void));
112STATIC union node *simplecmd __P((union node **, union node *));
113STATIC union node *makename __P((void));
114STATIC void parsefname __P((void));
115STATIC void parseheredoc __P((void));
116STATIC int peektoken __P((void));
117STATIC int readtoken __P((void));
118STATIC int xxreadtoken __P((void));
119STATIC int readtoken1 __P((int, char const *, char *, int));
120STATIC int noexpand __P((char *));
121STATIC void synexpect __P((int));
122STATIC void synerror __P((char *));
123STATIC void setprompt __P((int));
124
125
126/*
127 * Read and parse a command.  Returns NEOF on end of file.  (NULL is a
128 * valid parse tree indicating a blank line.)
129 */
130
131union node *
132parsecmd(interact)
133	int interact;
134{
135	int t;
136
137	doprompt = interact;
138	if (doprompt)
139		setprompt(1);
140	else
141		setprompt(0);
142	needprompt = 0;
143	t = readtoken();
144	if (t == TEOF)
145		return NEOF;
146	if (t == TNL)
147		return NULL;
148	tokpushback++;
149	return list(1);
150}
151
152
153STATIC union node *
154list(nlflag)
155	int nlflag;
156{
157	union node *n1, *n2, *n3;
158	int tok;
159
160	checkkwd = 2;
161	if (nlflag == 0 && tokendlist[peektoken()])
162		return NULL;
163	n1 = NULL;
164	for (;;) {
165		n2 = andor();
166		tok = readtoken();
167		if (tok == TBACKGND) {
168			if (n2->type == NCMD || n2->type == NPIPE) {
169				n2->ncmd.backgnd = 1;
170			} else if (n2->type == NREDIR) {
171				n2->type = NBACKGND;
172			} else {
173				n3 = (union node *)stalloc(sizeof (struct nredir));
174				n3->type = NBACKGND;
175				n3->nredir.n = n2;
176				n3->nredir.redirect = NULL;
177				n2 = n3;
178			}
179		}
180		if (n1 == NULL) {
181			n1 = n2;
182		}
183		else {
184			n3 = (union node *)stalloc(sizeof (struct nbinary));
185			n3->type = NSEMI;
186			n3->nbinary.ch1 = n1;
187			n3->nbinary.ch2 = n2;
188			n1 = n3;
189		}
190		switch (tok) {
191		case TBACKGND:
192		case TSEMI:
193			tok = readtoken();
194			/* fall through */
195		case TNL:
196			if (tok == TNL) {
197				parseheredoc();
198				if (nlflag)
199					return n1;
200			} else {
201				tokpushback++;
202			}
203			checkkwd = 2;
204			if (tokendlist[peektoken()])
205				return n1;
206			break;
207		case TEOF:
208			if (heredoclist)
209				parseheredoc();
210			else
211				pungetc();		/* push back EOF on input */
212			return n1;
213		default:
214			if (nlflag)
215				synexpect(-1);
216			tokpushback++;
217			return n1;
218		}
219	}
220}
221
222
223
224STATIC union node *
225andor() {
226	union node *n1, *n2, *n3;
227	int t;
228
229	n1 = pipeline();
230	for (;;) {
231		if ((t = readtoken()) == TAND) {
232			t = NAND;
233		} else if (t == TOR) {
234			t = NOR;
235		} else {
236			tokpushback++;
237			return n1;
238		}
239		n2 = pipeline();
240		n3 = (union node *)stalloc(sizeof (struct nbinary));
241		n3->type = t;
242		n3->nbinary.ch1 = n1;
243		n3->nbinary.ch2 = n2;
244		n1 = n3;
245	}
246}
247
248
249
250STATIC union node *
251pipeline() {
252	union node *n1, *pipenode, *notnode;
253	struct nodelist *lp, *prev;
254	int negate = 0;
255
256	TRACE(("pipeline: entered\n"));
257	while (readtoken() == TNOT) {
258		TRACE(("pipeline: TNOT recognized\n"));
259		negate = !negate;
260	}
261	tokpushback++;
262	n1 = command();
263	if (readtoken() == TPIPE) {
264		pipenode = (union node *)stalloc(sizeof (struct npipe));
265		pipenode->type = NPIPE;
266		pipenode->npipe.backgnd = 0;
267		lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
268		pipenode->npipe.cmdlist = lp;
269		lp->n = n1;
270		do {
271			prev = lp;
272			lp = (struct nodelist *)stalloc(sizeof (struct nodelist));
273			lp->n = command();
274			prev->next = lp;
275		} while (readtoken() == TPIPE);
276		lp->next = NULL;
277		n1 = pipenode;
278	}
279	tokpushback++;
280	if (negate) {
281		notnode = (union node *)stalloc(sizeof(struct nnot));
282		notnode->type = NNOT;
283		notnode->nnot.com = n1;
284		n1 = notnode;
285	}
286	return n1;
287}
288
289
290
291STATIC union node *
292command() {
293	union node *n1, *n2;
294	union node *ap, **app;
295	union node *cp, **cpp;
296	union node *redir, **rpp;
297	int t;
298
299	checkkwd = 2;
300	redir = NULL;
301	n1 = NULL;
302	rpp = &redir;
303
304	/* Check for redirection which may precede command */
305	while (readtoken() == TREDIR) {
306		*rpp = n2 = redirnode;
307		rpp = &n2->nfile.next;
308		parsefname();
309	}
310	tokpushback++;
311
312	switch (readtoken()) {
313	case TIF:
314		n1 = (union node *)stalloc(sizeof (struct nif));
315		n1->type = NIF;
316		n1->nif.test = list(0);
317		if (readtoken() != TTHEN)
318			synexpect(TTHEN);
319		n1->nif.ifpart = list(0);
320		n2 = n1;
321		while (readtoken() == TELIF) {
322			n2->nif.elsepart = (union node *)stalloc(sizeof (struct nif));
323			n2 = n2->nif.elsepart;
324			n2->type = NIF;
325			n2->nif.test = list(0);
326			if (readtoken() != TTHEN)
327				synexpect(TTHEN);
328			n2->nif.ifpart = list(0);
329		}
330		if (lasttoken == TELSE)
331			n2->nif.elsepart = list(0);
332		else {
333			n2->nif.elsepart = NULL;
334			tokpushback++;
335		}
336		if (readtoken() != TFI)
337			synexpect(TFI);
338		checkkwd = 1;
339		break;
340	case TWHILE:
341	case TUNTIL: {
342		int got;
343		n1 = (union node *)stalloc(sizeof (struct nbinary));
344		n1->type = (lasttoken == TWHILE)? NWHILE : NUNTIL;
345		n1->nbinary.ch1 = list(0);
346		if ((got=readtoken()) != TDO) {
347TRACE(("expecting DO got %s %s\n", tokname[got], got == TWORD ? wordtext : ""));
348			synexpect(TDO);
349		}
350		n1->nbinary.ch2 = list(0);
351		if (readtoken() != TDONE)
352			synexpect(TDONE);
353		checkkwd = 1;
354		break;
355	}
356	case TFOR:
357		if (readtoken() != TWORD || quoteflag || ! goodname(wordtext))
358			synerror("Bad for loop variable");
359		n1 = (union node *)stalloc(sizeof (struct nfor));
360		n1->type = NFOR;
361		n1->nfor.var = wordtext;
362		if (readtoken() == TWORD && ! quoteflag && equal(wordtext, "in")) {
363			app = &ap;
364			while (readtoken() == TWORD) {
365				n2 = (union node *)stalloc(sizeof (struct narg));
366				n2->type = NARG;
367				n2->narg.text = wordtext;
368				n2->narg.backquote = backquotelist;
369				*app = n2;
370				app = &n2->narg.next;
371			}
372			*app = NULL;
373			n1->nfor.args = ap;
374			if (lasttoken != TNL && lasttoken != TSEMI)
375				synexpect(-1);
376		} else {
377#ifndef GDB_HACK
378			static const char argvars[5] = {CTLVAR, VSNORMAL|VSQUOTE,
379								   '@', '=', '\0'};
380#endif
381			n2 = (union node *)stalloc(sizeof (struct narg));
382			n2->type = NARG;
383			n2->narg.text = (char *)argvars;
384			n2->narg.backquote = NULL;
385			n2->narg.next = NULL;
386			n1->nfor.args = n2;
387			/*
388			 * Newline or semicolon here is optional (but note
389			 * that the original Bourne shell only allowed NL).
390			 */
391			if (lasttoken != TNL && lasttoken != TSEMI)
392				tokpushback++;
393		}
394		checkkwd = 2;
395		if ((t = readtoken()) == TDO)
396			t = TDONE;
397		else if (t == TBEGIN)
398			t = TEND;
399		else
400			synexpect(-1);
401		n1->nfor.body = list(0);
402		if (readtoken() != t)
403			synexpect(t);
404		checkkwd = 1;
405		break;
406	case TCASE:
407		n1 = (union node *)stalloc(sizeof (struct ncase));
408		n1->type = NCASE;
409		if (readtoken() != TWORD)
410			synexpect(TWORD);
411		n1->ncase.expr = n2 = (union node *)stalloc(sizeof (struct narg));
412		n2->type = NARG;
413		n2->narg.text = wordtext;
414		n2->narg.backquote = backquotelist;
415		n2->narg.next = NULL;
416		while (readtoken() == TNL);
417		if (lasttoken != TWORD || ! equal(wordtext, "in"))
418			synerror("expecting \"in\"");
419		cpp = &n1->ncase.cases;
420		noaliases = 1;	/* turn off alias expansion */
421		checkkwd = 2, readtoken();
422		do {
423			*cpp = cp = (union node *)stalloc(sizeof (struct nclist));
424			cp->type = NCLIST;
425			app = &cp->nclist.pattern;
426			for (;;) {
427				*app = ap = (union node *)stalloc(sizeof (struct narg));
428				ap->type = NARG;
429				ap->narg.text = wordtext;
430				ap->narg.backquote = backquotelist;
431				if (checkkwd = 2, readtoken() != TPIPE)
432					break;
433				app = &ap->narg.next;
434				readtoken();
435			}
436			ap->narg.next = NULL;
437			if (lasttoken != TRP)
438				noaliases = 0, synexpect(TRP);
439			cp->nclist.body = list(0);
440
441			checkkwd = 2;
442			if ((t = readtoken()) != TESAC) {
443				if (t != TENDCASE)
444					noaliases = 0, synexpect(TENDCASE);
445				else
446					checkkwd = 2, readtoken();
447			}
448			cpp = &cp->nclist.next;
449		} while(lasttoken != TESAC);
450		noaliases = 0;	/* reset alias expansion */
451		*cpp = NULL;
452		checkkwd = 1;
453		break;
454	case TLP:
455		n1 = (union node *)stalloc(sizeof (struct nredir));
456		n1->type = NSUBSHELL;
457		n1->nredir.n = list(0);
458		n1->nredir.redirect = NULL;
459		if (readtoken() != TRP)
460			synexpect(TRP);
461		checkkwd = 1;
462		break;
463	case TBEGIN:
464		n1 = list(0);
465		if (readtoken() != TEND)
466			synexpect(TEND);
467		checkkwd = 1;
468		break;
469	/* Handle an empty command like other simple commands.  */
470	case TSEMI:
471		/*
472		 * An empty command before a ; doesn't make much sense, and
473		 * should certainly be disallowed in the case of `if ;'.
474		 */
475		if (!redir)
476			synexpect(-1);
477	case TAND:
478	case TOR:
479	case TNL:
480	case TEOF:
481	case TWORD:
482	case TRP:
483		tokpushback++;
484		return simplecmd(rpp, redir);
485	default:
486		synexpect(-1);
487	}
488
489	/* Now check for redirection which may follow command */
490	while (readtoken() == TREDIR) {
491		*rpp = n2 = redirnode;
492		rpp = &n2->nfile.next;
493		parsefname();
494	}
495	tokpushback++;
496	*rpp = NULL;
497	if (redir) {
498		if (n1->type != NSUBSHELL) {
499			n2 = (union node *)stalloc(sizeof (struct nredir));
500			n2->type = NREDIR;
501			n2->nredir.n = n1;
502			n1 = n2;
503		}
504		n1->nredir.redirect = redir;
505	}
506	return n1;
507}
508
509
510STATIC union node *
511simplecmd(rpp, redir)
512	union node **rpp, *redir;
513	{
514	union node *args, **app;
515	union node **orig_rpp = rpp;
516	union node *n = NULL;
517
518	/* If we don't have any redirections already, then we must reset */
519	/* rpp to be the address of the local redir variable.  */
520	if (redir == 0)
521		rpp = &redir;
522
523	args = NULL;
524	app = &args;
525	/*
526	 * We save the incoming value, because we need this for shell
527	 * functions.  There can not be a redirect or an argument between
528	 * the function name and the open parenthesis.
529	 */
530	orig_rpp = rpp;
531
532	for (;;) {
533		if (readtoken() == TWORD) {
534			n = (union node *)stalloc(sizeof (struct narg));
535			n->type = NARG;
536			n->narg.text = wordtext;
537			n->narg.backquote = backquotelist;
538			*app = n;
539			app = &n->narg.next;
540		} else if (lasttoken == TREDIR) {
541			*rpp = n = redirnode;
542			rpp = &n->nfile.next;
543			parsefname();	/* read name of redirection file */
544		} else if (lasttoken == TLP && app == &args->narg.next
545					    && rpp == orig_rpp) {
546			/* We have a function */
547			if (readtoken() != TRP)
548				synexpect(TRP);
549#ifdef notdef
550			if (! goodname(n->narg.text))
551				synerror("Bad function name");
552#endif
553			n->type = NDEFUN;
554			n->narg.next = command();
555			return n;
556		} else {
557			tokpushback++;
558			break;
559		}
560	}
561	*app = NULL;
562	*rpp = NULL;
563	n = (union node *)stalloc(sizeof (struct ncmd));
564	n->type = NCMD;
565	n->ncmd.backgnd = 0;
566	n->ncmd.args = args;
567	n->ncmd.redirect = redir;
568	return n;
569}
570
571STATIC union node *
572makename() {
573	union node *n;
574
575	n = (union node *)stalloc(sizeof (struct narg));
576	n->type = NARG;
577	n->narg.next = NULL;
578	n->narg.text = wordtext;
579	n->narg.backquote = backquotelist;
580	return n;
581}
582
583void fixredir(n, text, err)
584	union node *n;
585	const char *text;
586	int err;
587	{
588	TRACE(("Fix redir %s %d\n", text, err));
589	if (!err)
590		n->ndup.vname = NULL;
591
592	if (is_digit(text[0]) && text[1] == '\0')
593		n->ndup.dupfd = digit_val(text[0]);
594	else if (text[0] == '-' && text[1] == '\0')
595		n->ndup.dupfd = -1;
596	else {
597
598		if (err)
599			synerror("Bad fd number");
600		else
601			n->ndup.vname = makename();
602	}
603}
604
605
606STATIC void
607parsefname() {
608	union node *n = redirnode;
609
610	if (readtoken() != TWORD)
611		synexpect(-1);
612	if (n->type == NHERE) {
613		struct heredoc *here = heredoc;
614		struct heredoc *p;
615		int i;
616
617		if (quoteflag == 0)
618			n->type = NXHERE;
619		TRACE(("Here document %d\n", n->type));
620		if (here->striptabs) {
621			while (*wordtext == '\t')
622				wordtext++;
623		}
624		if (! noexpand(wordtext) || (i = strlen(wordtext)) == 0 || i > EOFMARKLEN)
625			synerror("Illegal eof marker for << redirection");
626		rmescapes(wordtext);
627		here->eofmark = wordtext;
628		here->next = NULL;
629		if (heredoclist == NULL)
630			heredoclist = here;
631		else {
632			for (p = heredoclist ; p->next ; p = p->next);
633			p->next = here;
634		}
635	} else if (n->type == NTOFD || n->type == NFROMFD) {
636		fixredir(n, wordtext, 0);
637	} else {
638		n->nfile.fname = makename();
639	}
640}
641
642
643/*
644 * Input any here documents.
645 */
646
647STATIC void
648parseheredoc() {
649	struct heredoc *here;
650	union node *n;
651
652	while (heredoclist) {
653		here = heredoclist;
654		heredoclist = here->next;
655		if (needprompt) {
656			setprompt(2);
657			needprompt = 0;
658		}
659		readtoken1(pgetc(), here->here->type == NHERE? SQSYNTAX : DQSYNTAX,
660				here->eofmark, here->striptabs);
661		n = (union node *)stalloc(sizeof (struct narg));
662		n->narg.type = NARG;
663		n->narg.next = NULL;
664		n->narg.text = wordtext;
665		n->narg.backquote = backquotelist;
666		here->here->nhere.doc = n;
667	}
668}
669
670STATIC int
671peektoken() {
672	int t;
673
674	t = readtoken();
675	tokpushback++;
676	return (t);
677}
678
679STATIC int xxreadtoken();
680
681STATIC int
682readtoken() {
683	int t;
684	int savecheckkwd = checkkwd;
685	struct alias *ap;
686#ifdef DEBUG
687	int alreadyseen = tokpushback;
688#endif
689
690	top:
691	t = xxreadtoken();
692
693	if (checkkwd) {
694		/*
695		 * eat newlines
696		 */
697		if (checkkwd == 2) {
698			checkkwd = 0;
699			while (t == TNL) {
700				parseheredoc();
701				t = xxreadtoken();
702			}
703		} else
704			checkkwd = 0;
705		/*
706		 * check for keywords and aliases
707		 */
708		if (t == TWORD && !quoteflag)
709		{
710			char * const *pp;
711
712			for (pp = (char **)parsekwd; *pp; pp++) {
713				if (**pp == *wordtext && equal(*pp, wordtext))
714				{
715					lasttoken = t = pp - parsekwd + KWDOFFSET;
716					TRACE(("keyword %s recognized\n", tokname[t]));
717					goto out;
718				}
719			}
720			if (noaliases == 0 &&
721			    (ap = lookupalias(wordtext, 1)) != NULL) {
722				pushstring(ap->val, strlen(ap->val), ap);
723				checkkwd = savecheckkwd;
724				goto top;
725			}
726		}
727out:
728		checkkwd = 0;
729	}
730#ifdef DEBUG
731	if (!alreadyseen)
732	    TRACE(("token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
733	else
734	    TRACE(("reread token %s %s\n", tokname[t], t == TWORD ? wordtext : ""));
735#endif
736	return (t);
737}
738
739
740/*
741 * Read the next input token.
742 * If the token is a word, we set backquotelist to the list of cmds in
743 *	backquotes.  We set quoteflag to true if any part of the word was
744 *	quoted.
745 * If the token is TREDIR, then we set redirnode to a structure containing
746 *	the redirection.
747 * In all cases, the variable startlinno is set to the number of the line
748 *	on which the token starts.
749 *
750 * [Change comment:  here documents and internal procedures]
751 * [Readtoken shouldn't have any arguments.  Perhaps we should make the
752 *  word parsing code into a separate routine.  In this case, readtoken
753 *  doesn't need to have any internal procedures, but parseword does.
754 *  We could also make parseoperator in essence the main routine, and
755 *  have parseword (readtoken1?) handle both words and redirection.]
756 */
757
758#define RETURN(token)	return lasttoken = token
759
760STATIC int
761xxreadtoken() {
762	int c;
763
764	if (tokpushback) {
765		tokpushback = 0;
766		return lasttoken;
767	}
768	if (needprompt) {
769		setprompt(2);
770		needprompt = 0;
771	}
772	startlinno = plinno;
773	for (;;) {	/* until token or start of word found */
774		c = pgetc_macro();
775		if (c == ' ' || c == '\t')
776			continue;		/* quick check for white space first */
777		switch (c) {
778		case ' ': case '\t':
779			continue;
780		case '#':
781			while ((c = pgetc()) != '\n' && c != PEOF);
782			pungetc();
783			continue;
784		case '\\':
785			if (pgetc() == '\n') {
786				startlinno = ++plinno;
787				if (doprompt)
788					setprompt(2);
789				else
790					setprompt(0);
791				continue;
792			}
793			pungetc();
794			goto breakloop;
795		case '\n':
796			plinno++;
797			needprompt = doprompt;
798			RETURN(TNL);
799		case PEOF:
800			RETURN(TEOF);
801		case '&':
802			if (pgetc() == '&')
803				RETURN(TAND);
804			pungetc();
805			RETURN(TBACKGND);
806		case '|':
807			if (pgetc() == '|')
808				RETURN(TOR);
809			pungetc();
810			RETURN(TPIPE);
811		case ';':
812			if (pgetc() == ';')
813				RETURN(TENDCASE);
814			pungetc();
815			RETURN(TSEMI);
816		case '(':
817			RETURN(TLP);
818		case ')':
819			RETURN(TRP);
820		default:
821			goto breakloop;
822		}
823	}
824breakloop:
825	return readtoken1(c, BASESYNTAX, (char *)NULL, 0);
826#undef RETURN
827}
828
829
830
831/*
832 * If eofmark is NULL, read a word or a redirection symbol.  If eofmark
833 * is not NULL, read a here document.  In the latter case, eofmark is the
834 * word which marks the end of the document and striptabs is true if
835 * leading tabs should be stripped from the document.  The argument firstc
836 * is the first character of the input token or document.
837 *
838 * Because C does not have internal subroutines, I have simulated them
839 * using goto's to implement the subroutine linkage.  The following macros
840 * will run code that appears at the end of readtoken1.
841 */
842
843#define CHECKEND()	{goto checkend; checkend_return:;}
844#define PARSEREDIR()	{goto parseredir; parseredir_return:;}
845#define PARSESUB()	{goto parsesub; parsesub_return:;}
846#define PARSEBACKQOLD()	{oldstyle = 1; goto parsebackq; parsebackq_oldreturn:;}
847#define PARSEBACKQNEW()	{oldstyle = 0; goto parsebackq; parsebackq_newreturn:;}
848#define	PARSEARITH()	{goto parsearith; parsearith_return:;}
849
850STATIC int
851readtoken1(firstc, syntax, eofmark, striptabs)
852	int firstc;
853	char const *syntax;
854	char *eofmark;
855	int striptabs;
856	{
857	int c = firstc;
858	char *out;
859	int len;
860	char line[EOFMARKLEN + 1];
861	struct nodelist *bqlist;
862	int quotef;
863	int dblquote;
864	int varnest;	/* levels of variables expansion */
865	int arinest;	/* levels of arithmetic expansion */
866	int parenlevel;	/* levels of parens in arithmetic */
867	int oldstyle;
868	char const *prevsyntax;	/* syntax before arithmetic */
869#if __GNUC__
870	/* Avoid longjmp clobbering */
871	(void) &out;
872	(void) &quotef;
873	(void) &dblquote;
874	(void) &varnest;
875	(void) &arinest;
876	(void) &parenlevel;
877	(void) &oldstyle;
878	(void) &prevsyntax;
879	(void) &syntax;
880#endif
881
882	startlinno = plinno;
883	dblquote = 0;
884	if (syntax == DQSYNTAX)
885		dblquote = 1;
886	quotef = 0;
887	bqlist = NULL;
888	varnest = 0;
889	arinest = 0;
890	parenlevel = 0;
891
892	STARTSTACKSTR(out);
893	loop: {	/* for each line, until end of word */
894#if ATTY
895		if (c == '\034' && doprompt
896		 && attyset() && ! equal(termval(), "emacs")) {
897			attyline();
898			if (syntax == BASESYNTAX)
899				return readtoken();
900			c = pgetc();
901			goto loop;
902		}
903#endif
904		CHECKEND();	/* set c to PEOF if at end of here document */
905		for (;;) {	/* until end of line or end of word */
906			CHECKSTRSPACE(3, out);	/* permit 3 calls to USTPUTC */
907			switch(syntax[c]) {
908			case CNL:	/* '\n' */
909				if (syntax == BASESYNTAX)
910					goto endword;	/* exit outer loop */
911				USTPUTC(c, out);
912				plinno++;
913				if (doprompt)
914					setprompt(2);
915				else
916					setprompt(0);
917				c = pgetc();
918				goto loop;		/* continue outer loop */
919			case CWORD:
920				USTPUTC(c, out);
921				break;
922			case CCTL:
923				if (eofmark == NULL || dblquote)
924					USTPUTC(CTLESC, out);
925				USTPUTC(c, out);
926				break;
927			case CBACK:	/* backslash */
928				c = pgetc();
929				if (c == PEOF) {
930					USTPUTC('\\', out);
931					pungetc();
932				} else if (c == '\n') {
933					if (doprompt)
934						setprompt(2);
935					else
936						setprompt(0);
937				} else {
938					if (dblquote && c != '\\' && c != '`' && c != '$'
939							 && (c != '"' || eofmark != NULL))
940						USTPUTC('\\', out);
941					if (SQSYNTAX[c] == CCTL)
942						USTPUTC(CTLESC, out);
943					USTPUTC(c, out);
944					quotef++;
945				}
946				break;
947			case CSQUOTE:
948				syntax = SQSYNTAX;
949				break;
950			case CDQUOTE:
951				syntax = DQSYNTAX;
952				dblquote = 1;
953				break;
954			case CENDQUOTE:
955				if (eofmark) {
956					USTPUTC(c, out);
957				} else {
958					if (arinest)
959						syntax = ARISYNTAX;
960					else
961						syntax = BASESYNTAX;
962					quotef++;
963					dblquote = 0;
964				}
965				break;
966			case CVAR:	/* '$' */
967				PARSESUB();		/* parse substitution */
968				break;
969			case CENDVAR:	/* '}' */
970				if (varnest > 0) {
971					varnest--;
972					USTPUTC(CTLENDVAR, out);
973				} else {
974					USTPUTC(c, out);
975				}
976				break;
977			case CLP:	/* '(' in arithmetic */
978				parenlevel++;
979				USTPUTC(c, out);
980				break;
981			case CRP:	/* ')' in arithmetic */
982				if (parenlevel > 0) {
983					USTPUTC(c, out);
984					--parenlevel;
985				} else {
986					if (pgetc() == ')') {
987						if (--arinest == 0) {
988							USTPUTC(CTLENDARI, out);
989							syntax = prevsyntax;
990						} else
991							USTPUTC(')', out);
992					} else {
993						/*
994						 * unbalanced parens
995						 *  (don't 2nd guess - no error)
996						 */
997						pungetc();
998						USTPUTC(')', out);
999					}
1000				}
1001				break;
1002			case CBQUOTE:	/* '`' */
1003				PARSEBACKQOLD();
1004				break;
1005			case CEOF:
1006				goto endword;		/* exit outer loop */
1007			default:
1008				if (varnest == 0)
1009					goto endword;	/* exit outer loop */
1010				USTPUTC(c, out);
1011			}
1012			c = pgetc_macro();
1013		}
1014	}
1015endword:
1016	if (syntax == ARISYNTAX)
1017		synerror("Missing '))'");
1018	if (syntax != BASESYNTAX && ! parsebackquote && eofmark == NULL)
1019		synerror("Unterminated quoted string");
1020	if (varnest != 0) {
1021		startlinno = plinno;
1022		synerror("Missing '}'");
1023	}
1024	USTPUTC('\0', out);
1025	len = out - stackblock();
1026	out = stackblock();
1027	if (eofmark == NULL) {
1028		if ((c == '>' || c == '<')
1029		 && quotef == 0
1030		 && len <= 2
1031		 && (*out == '\0' || is_digit(*out))) {
1032			PARSEREDIR();
1033			return lasttoken = TREDIR;
1034		} else {
1035			pungetc();
1036		}
1037	}
1038	quoteflag = quotef;
1039	backquotelist = bqlist;
1040	grabstackblock(len);
1041	wordtext = out;
1042	return lasttoken = TWORD;
1043/* end of readtoken routine */
1044
1045
1046
1047/*
1048 * Check to see whether we are at the end of the here document.  When this
1049 * is called, c is set to the first character of the next input line.  If
1050 * we are at the end of the here document, this routine sets the c to PEOF.
1051 */
1052
1053checkend: {
1054	if (eofmark) {
1055		if (striptabs) {
1056			while (c == '\t')
1057				c = pgetc();
1058		}
1059		if (c == *eofmark) {
1060			if (pfgets(line, sizeof line) != NULL) {
1061				char *p, *q;
1062
1063				p = line;
1064				for (q = eofmark + 1 ; *q && *p == *q ; p++, q++);
1065				if (*p == '\n' && *q == '\0') {
1066					c = PEOF;
1067					plinno++;
1068					needprompt = doprompt;
1069				} else {
1070					pushstring(line, strlen(line), NULL);
1071				}
1072			}
1073		}
1074	}
1075	goto checkend_return;
1076}
1077
1078
1079/*
1080 * Parse a redirection operator.  The variable "out" points to a string
1081 * specifying the fd to be redirected.  The variable "c" contains the
1082 * first character of the redirection operator.
1083 */
1084
1085parseredir: {
1086	char fd = *out;
1087	union node *np;
1088
1089	np = (union node *)stalloc(sizeof (struct nfile));
1090	if (c == '>') {
1091		np->nfile.fd = 1;
1092		c = pgetc();
1093		if (c == '>')
1094			np->type = NAPPEND;
1095		else if (c == '&')
1096			np->type = NTOFD;
1097		else {
1098			np->type = NTO;
1099			pungetc();
1100		}
1101	} else {	/* c == '<' */
1102		np->nfile.fd = 0;
1103		c = pgetc();
1104		if (c == '<') {
1105			if (sizeof (struct nfile) != sizeof (struct nhere)) {
1106				np = (union node *)stalloc(sizeof (struct nhere));
1107				np->nfile.fd = 0;
1108			}
1109			np->type = NHERE;
1110			heredoc = (struct heredoc *)stalloc(sizeof (struct heredoc));
1111			heredoc->here = np;
1112			if ((c = pgetc()) == '-') {
1113				heredoc->striptabs = 1;
1114			} else {
1115				heredoc->striptabs = 0;
1116				pungetc();
1117			}
1118		} else if (c == '&')
1119			np->type = NFROMFD;
1120		else {
1121			np->type = NFROM;
1122			pungetc();
1123		}
1124	}
1125	if (fd != '\0')
1126		np->nfile.fd = digit_val(fd);
1127	redirnode = np;
1128	goto parseredir_return;
1129}
1130
1131
1132/*
1133 * Parse a substitution.  At this point, we have read the dollar sign
1134 * and nothing else.
1135 */
1136
1137parsesub: {
1138	int subtype;
1139	int typeloc;
1140	int flags;
1141	char *p;
1142#ifndef GDB_HACK
1143	static const char types[] = "}-+?=";
1144#endif
1145       int bracketed_name = 0; /* used to handle ${[0-9]*} variables */
1146
1147	c = pgetc();
1148	if (c != '(' && c != '{' && !is_name(c) && !is_special(c)) {
1149		USTPUTC('$', out);
1150		pungetc();
1151	} else if (c == '(') {	/* $(command) or $((arith)) */
1152		if (pgetc() == '(') {
1153			PARSEARITH();
1154		} else {
1155			pungetc();
1156			PARSEBACKQNEW();
1157		}
1158	} else {
1159		USTPUTC(CTLVAR, out);
1160		typeloc = out - stackblock();
1161		USTPUTC(VSNORMAL, out);
1162		subtype = VSNORMAL;
1163		if (c == '{') {
1164			bracketed_name = 1;
1165			c = pgetc();
1166			if (c == '#') {
1167				if ((c = pgetc()) == '}')
1168					c = '#';
1169				else
1170					subtype = VSLENGTH;
1171			}
1172			else
1173				subtype = 0;
1174		}
1175		if (is_name(c)) {
1176			do {
1177				STPUTC(c, out);
1178				c = pgetc();
1179			} while (is_in_name(c));
1180		} else if (is_digit(c)) {
1181			if (bracketed_name) {
1182				do {
1183					STPUTC(c, out);
1184					c = pgetc();
1185				} while (is_digit(c));
1186			} else {
1187				STPUTC(c, out);
1188				c = pgetc();
1189			}
1190		} else {
1191			if (! is_special(c))
1192badsub:				synerror("Bad substitution");
1193			USTPUTC(c, out);
1194			c = pgetc();
1195		}
1196		STPUTC('=', out);
1197		flags = 0;
1198		if (subtype == 0) {
1199			switch (c) {
1200			case ':':
1201				flags = VSNUL;
1202				c = pgetc();
1203				/*FALLTHROUGH*/
1204			default:
1205				p = strchr(types, c);
1206				if (p == NULL)
1207					goto badsub;
1208				subtype = p - types + VSNORMAL;
1209				break;
1210			case '%':
1211			case '#':
1212				{
1213					int cc = c;
1214					subtype = c == '#' ? VSTRIMLEFT :
1215							     VSTRIMRIGHT;
1216					c = pgetc();
1217					if (c == cc)
1218						subtype++;
1219					else
1220						pungetc();
1221					break;
1222				}
1223			}
1224		} else {
1225			pungetc();
1226		}
1227		if (dblquote || arinest)
1228			flags |= VSQUOTE;
1229		*(stackblock() + typeloc) = subtype | flags;
1230		if (subtype != VSNORMAL)
1231			varnest++;
1232	}
1233	goto parsesub_return;
1234}
1235
1236
1237/*
1238 * Called to parse command substitutions.  Newstyle is set if the command
1239 * is enclosed inside $(...); nlpp is a pointer to the head of the linked
1240 * list of commands (passed by reference), and savelen is the number of
1241 * characters on the top of the stack which must be preserved.
1242 */
1243
1244parsebackq: {
1245	struct nodelist **nlpp;
1246	int savepbq;
1247	union node *n;
1248	char *volatile str;
1249	struct jmploc jmploc;
1250	struct jmploc *volatile savehandler;
1251	int savelen;
1252	int saveprompt;
1253#if __GNUC__
1254	/* Avoid longjmp clobbering */
1255	(void) &saveprompt;
1256#endif
1257
1258	savepbq = parsebackquote;
1259	if (setjmp(jmploc.loc)) {
1260		if (str)
1261			ckfree(str);
1262		parsebackquote = 0;
1263		handler = savehandler;
1264		longjmp(handler->loc, 1);
1265	}
1266	INTOFF;
1267	str = NULL;
1268	savelen = out - stackblock();
1269	if (savelen > 0) {
1270		str = ckmalloc(savelen);
1271		memcpy(str, stackblock(), savelen);
1272	}
1273	savehandler = handler;
1274	handler = &jmploc;
1275	INTON;
1276        if (oldstyle) {
1277                /* We must read until the closing backquote, giving special
1278                   treatment to some slashes, and then push the string and
1279                   reread it as input, interpreting it normally.  */
1280                char *out;
1281                int c;
1282                int savelen;
1283                char *str;
1284
1285
1286                STARTSTACKSTR(out);
1287		for (;;) {
1288			if (needprompt) {
1289				setprompt(2);
1290				needprompt = 0;
1291			}
1292			switch (c = pgetc()) {
1293			case '`':
1294				goto done;
1295
1296			case '\\':
1297                                if ((c = pgetc()) == '\n') {
1298					plinno++;
1299					if (doprompt)
1300						setprompt(2);
1301					else
1302						setprompt(0);
1303					/*
1304					 * If eating a newline, avoid putting
1305					 * the newline into the new character
1306					 * stream (via the STPUTC after the
1307					 * switch).
1308					 */
1309					continue;
1310				}
1311                                if (c != '\\' && c != '`' && c != '$'
1312                                    && (!dblquote || c != '"'))
1313                                        STPUTC('\\', out);
1314				break;
1315
1316			case '\n':
1317				plinno++;
1318				needprompt = doprompt;
1319				break;
1320
1321			case PEOF:
1322			        startlinno = plinno;
1323				synerror("EOF in backquote substitution");
1324 				break;
1325
1326			default:
1327				break;
1328			}
1329			STPUTC(c, out);
1330                }
1331done:
1332                STPUTC('\0', out);
1333                savelen = out - stackblock();
1334                if (savelen > 0) {
1335                        str = ckmalloc(savelen);
1336                        memcpy(str, stackblock(), savelen);
1337			setinputstring(str, 1);
1338                }
1339        }
1340	nlpp = &bqlist;
1341	while (*nlpp)
1342		nlpp = &(*nlpp)->next;
1343	*nlpp = (struct nodelist *)stalloc(sizeof (struct nodelist));
1344	(*nlpp)->next = NULL;
1345	parsebackquote = oldstyle;
1346
1347	if (oldstyle) {
1348		saveprompt = doprompt;
1349		doprompt = 0;
1350	}
1351
1352	n = list(0);
1353
1354	if (oldstyle)
1355		doprompt = saveprompt;
1356	else {
1357		if (readtoken() != TRP)
1358			synexpect(TRP);
1359	}
1360
1361	(*nlpp)->n = n;
1362        if (oldstyle) {
1363		/*
1364		 * Start reading from old file again, ignoring any pushed back
1365		 * tokens left from the backquote parsing
1366		 */
1367                popfile();
1368		tokpushback = 0;
1369	}
1370	while (stackblocksize() <= savelen)
1371		growstackblock();
1372	STARTSTACKSTR(out);
1373	if (str) {
1374		memcpy(out, str, savelen);
1375		STADJUST(savelen, out);
1376		INTOFF;
1377		ckfree(str);
1378		str = NULL;
1379		INTON;
1380	}
1381	parsebackquote = savepbq;
1382	handler = savehandler;
1383	if (arinest || dblquote)
1384		USTPUTC(CTLBACKQ | CTLQUOTE, out);
1385	else
1386		USTPUTC(CTLBACKQ, out);
1387	if (oldstyle)
1388		goto parsebackq_oldreturn;
1389	else
1390		goto parsebackq_newreturn;
1391}
1392
1393/*
1394 * Parse an arithmetic expansion (indicate start of one and set state)
1395 */
1396parsearith: {
1397
1398	if (++arinest == 1) {
1399		prevsyntax = syntax;
1400		syntax = ARISYNTAX;
1401		USTPUTC(CTLARI, out);
1402	} else {
1403		/*
1404		 * we collapse embedded arithmetic expansion to
1405		 * parenthesis, which should be equivalent
1406		 */
1407		USTPUTC('(', out);
1408	}
1409	goto parsearith_return;
1410}
1411
1412} /* end of readtoken */
1413
1414
1415
1416#ifdef mkinit
1417RESET {
1418	tokpushback = 0;
1419	checkkwd = 0;
1420}
1421#endif
1422
1423/*
1424 * Returns true if the text contains nothing to expand (no dollar signs
1425 * or backquotes).
1426 */
1427
1428STATIC int
1429noexpand(text)
1430	char *text;
1431	{
1432	char *p;
1433	char c;
1434
1435	p = text;
1436	while ((c = *p++) != '\0') {
1437		if (c == CTLESC)
1438			p++;
1439		else if (BASESYNTAX[c] == CCTL)
1440			return 0;
1441	}
1442	return 1;
1443}
1444
1445
1446/*
1447 * Return true if the argument is a legal variable name (a letter or
1448 * underscore followed by zero or more letters, underscores, and digits).
1449 */
1450
1451int
1452goodname(name)
1453	char *name;
1454	{
1455	char *p;
1456
1457	p = name;
1458	if (! is_name(*p))
1459		return 0;
1460	while (*++p) {
1461		if (! is_in_name(*p))
1462			return 0;
1463	}
1464	return 1;
1465}
1466
1467
1468/*
1469 * Called when an unexpected token is read during the parse.  The argument
1470 * is the token that is expected, or -1 if more than one type of token can
1471 * occur at this point.
1472 */
1473
1474STATIC void
1475synexpect(token)
1476	int token;
1477{
1478	char msg[64];
1479
1480	if (token >= 0) {
1481		fmtstr(msg, 64, "%s unexpected (expecting %s)",
1482			tokname[lasttoken], tokname[token]);
1483	} else {
1484		fmtstr(msg, 64, "%s unexpected", tokname[lasttoken]);
1485	}
1486	synerror(msg);
1487}
1488
1489
1490STATIC void
1491synerror(msg)
1492	char *msg;
1493	{
1494	if (commandname)
1495		outfmt(&errout, "%s: %d: ", commandname, startlinno);
1496	outfmt(&errout, "Syntax error: %s\n", msg);
1497	error((char *)NULL);
1498}
1499
1500STATIC void
1501setprompt(which)
1502	int which;
1503	{
1504	whichprompt = which;
1505
1506#ifndef NO_HISTORY
1507	if (!el)
1508#endif
1509		out2str(getprompt(NULL));
1510}
1511
1512/*
1513 * called by editline -- any expansions to the prompt
1514 *    should be added here.
1515 */
1516char *
1517getprompt(unused)
1518	void *unused;
1519	{
1520	switch (whichprompt) {
1521	case 0:
1522		return "";
1523	case 1:
1524		return ps1val();
1525	case 2:
1526		return ps2val();
1527	default:
1528		return "<internal prompt error>";
1529	}
1530}
1531