xref: /netbsd-src/bin/csh/lex.c (revision 220b5c059a84c51ea44107ea8951a57ffaecdc8c)
1 /* $NetBSD: lex.c,v 1.20 2001/11/03 13:35:39 lukem Exp $ */
2 
3 /*-
4  * Copyright (c) 1980, 1991, 1993
5  *	The Regents of the University of California.  All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. All advertising materials mentioning features or use of this software
16  *    must display the following acknowledgement:
17  *	This product includes software developed by the University of
18  *	California, Berkeley and its contributors.
19  * 4. Neither the name of the University nor the names of its contributors
20  *    may be used to endorse or promote products derived from this software
21  *    without specific prior written permission.
22  *
23  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33  * SUCH DAMAGE.
34  */
35 
36 #include <sys/cdefs.h>
37 #ifndef lint
38 #if 0
39 static char sccsid[] = "@(#)lex.c	8.1 (Berkeley) 5/31/93";
40 #else
41 __RCSID("$NetBSD: lex.c,v 1.20 2001/11/03 13:35:39 lukem Exp $");
42 #endif
43 #endif /* not lint */
44 
45 #include <sys/ioctl.h>
46 #include <sys/types.h>
47 
48 #include <errno.h>
49 #include <stdlib.h>
50 #include <string.h>
51 #include <termios.h>
52 #include <unistd.h>
53 
54 #if __STDC__
55 # include <stdarg.h>
56 #else
57 # include <varargs.h>
58 #endif
59 
60 #include "csh.h"
61 #include "extern.h"
62 
63 /*
64  * These lexical routines read input and form lists of words.
65  * There is some involved processing here, because of the complications
66  * of input buffering, and especially because of history substitution.
67  */
68 
69 static Char *word(void);
70 static int getC1(int);
71 static void getdol(void);
72 static void getexcl(int);
73 static struct Hist *findev(Char *, bool);
74 static void setexclp(Char *);
75 static int bgetc(void);
76 static void bfree(void);
77 static struct wordent *gethent(int);
78 static int matchs(Char *, Char *);
79 static int getsel(int *, int *, int);
80 static struct wordent *getsub(struct wordent *);
81 static Char *subword(Char *, int, bool *);
82 static struct wordent *dosub(int, struct wordent *, bool);
83 
84 /*
85  * Peekc is a peek character for getC, peekread for readc.
86  * There is a subtlety here in many places... history routines
87  * will read ahead and then insert stuff into the input stream.
88  * If they push back a character then they must push it behind
89  * the text substituted by the history substitution.  On the other
90  * hand in several places we need 2 peek characters.  To make this
91  * all work, the history routines read with getC, and make use both
92  * of ungetC and unreadc.  The key observation is that the state
93  * of getC at the call of a history reference is such that calls
94  * to getC from the history routines will always yield calls of
95  * readc, unless this peeking is involved.  That is to say that during
96  * getexcl the variables lap, exclp, and exclnxt are all zero.
97  *
98  * Getdol invokes history substitution, hence the extra peek, peekd,
99  * which it can ungetD to be before history substitutions.
100  */
101 static Char peekc = 0, peekd = 0;
102 static Char peekread = 0;
103 
104 /* (Tail of) current word from ! subst */
105 static Char *exclp = NULL;
106 
107 /* The rest of the ! subst words */
108 static struct wordent *exclnxt = NULL;
109 
110 /* Count of remaining words in ! subst */
111 static int exclc = 0;
112 
113 /* "Globp" for alias resubstitution */
114 Char **alvec, *alvecp;
115 int aret = F_SEEK;
116 
117 /*
118  * Labuf implements a general buffer for lookahead during lexical operations.
119  * Text which is to be placed in the input stream can be stuck here.
120  * We stick parsed ahead $ constructs during initial input,
121  * process id's from `$$', and modified variable values (from qualifiers
122  * during expansion in sh.dol.c) here.
123  */
124 static Char labuf[BUFSIZE];
125 
126 /*
127  * Lex returns to its caller not only a wordlist (as a "var" parameter)
128  * but also whether a history substitution occurred.  This is used in
129  * the main (process) routine to determine whether to echo, and also
130  * when called by the alias routine to determine whether to keep the
131  * argument list.
132  */
133 static bool hadhist = 0;
134 
135 /*
136  * Avoid alias expansion recursion via \!#
137  */
138 int     hleft;
139 
140 static Char getCtmp;
141 
142 #define getC(f) ((getCtmp = peekc) ? (peekc = 0, getCtmp) : getC1(f))
143 #define	ungetC(c) peekc = c
144 #define	ungetD(c) peekd = c
145 
146 int
147 lex(struct wordent *hp)
148 {
149     struct wordent *wdp;
150     int c;
151 
152     btell(&lineloc);
153     hp->next = hp->prev = hp;
154     hp->word = STRNULL;
155     hadhist = 0;
156     do
157 	c = readc(0);
158     while (c == ' ' || c == '\t');
159     if (c == HISTSUB && intty)
160 	/* ^lef^rit	from tty is short !:s^lef^rit */
161 	getexcl(c);
162     else
163 	unreadc(c);
164     wdp = hp;
165     /*
166      * The following loop is written so that the links needed by freelex will
167      * be ready and rarin to go even if it is interrupted.
168      */
169     do {
170 	struct wordent *new;
171 
172 	new = (struct wordent *)xmalloc((size_t)sizeof(*wdp));
173 	new->word = 0;
174 	new->prev = wdp;
175 	new->next = hp;
176 	wdp->next = new;
177 	wdp = new;
178 	wdp->word = word();
179     } while (wdp->word[0] != '\n');
180     hp->prev = wdp;
181     return (hadhist);
182 }
183 
184 void
185 prlex(FILE *fp, struct wordent *sp0)
186 {
187     struct wordent *sp;
188 
189     sp = sp0->next;
190     for (;;) {
191 	(void)fprintf(fp, "%s", vis_str(sp->word));
192 	sp = sp->next;
193 	if (sp == sp0)
194 	    break;
195 	if (sp->word[0] != '\n')
196 	    (void) fputc(' ', fp);
197     }
198 }
199 
200 void
201 copylex(struct wordent *hp, struct wordent *fp)
202 {
203     struct wordent *wdp;
204 
205     wdp = hp;
206     fp = fp->next;
207     do {
208 	struct wordent *new;
209 
210 	new = (struct wordent *)xmalloc((size_t)sizeof(*wdp));
211 	new->prev = wdp;
212 	new->next = hp;
213 	wdp->next = new;
214 	wdp = new;
215 	wdp->word = Strsave(fp->word);
216 	fp = fp->next;
217     } while (wdp->word[0] != '\n');
218     hp->prev = wdp;
219 }
220 
221 void
222 freelex(struct wordent *vp)
223 {
224     struct wordent *fp;
225 
226     while (vp->next != vp) {
227 	fp = vp->next;
228 	vp->next = fp->next;
229 	xfree((ptr_t) fp->word);
230 	xfree((ptr_t) fp);
231     }
232     vp->prev = vp;
233 }
234 
235 static Char *
236 word(void)
237 {
238     Char wbuf[BUFSIZE], *wp;
239     int i;
240     Char c, c1;
241     bool dolflg;
242 
243     wp = wbuf;
244     i = BUFSIZE - 4;
245 loop:
246     while ((c = getC(DOALL)) == ' ' || c == '\t')
247 	continue;
248     if (cmap(c, _META | _ESC))
249 	switch (c) {
250 	case '&':
251 	case '|':
252 	case '<':
253 	case '>':
254 	    *wp++ = c;
255 	    c1 = getC(DOALL);
256 	    if (c1 == c)
257 		*wp++ = c1;
258 	    else
259 		ungetC(c1);
260 	    goto ret;
261 
262 	case '#':
263 	    if (intty)
264 		break;
265 	    c = 0;
266 	    do {
267 		c1 = c;
268 		c = getC(0);
269 	    } while (c != '\n');
270 	    if (c1 == '\\')
271 		goto loop;
272 	    /* FALLTHROUGH */
273 
274 	case ';':
275 	case '(':
276 	case ')':
277 	case '\n':
278 	    *wp++ = c;
279 	    goto ret;
280 
281 	case '\\':
282 	    c = getC(0);
283 	    if (c == '\n') {
284 		if (onelflg == 1)
285 		    onelflg = 2;
286 		goto loop;
287 	    }
288 	    if (c != HIST)
289 		*wp++ = '\\', --i;
290 	    c |= QUOTE;
291 	    break;
292 	}
293     c1 = 0;
294     dolflg = DOALL;
295     for (;;) {
296 	if (c1) {
297 	    if (c == c1) {
298 		c1 = 0;
299 		dolflg = DOALL;
300 	    }
301 	    else if (c == '\\') {
302 		c = getC(0);
303 		if (c == HIST)
304 		    c |= QUOTE;
305 		else {
306 		    if (c == '\n')
307 			/*
308 			 * if (c1 == '`') c = ' '; else
309 			 */
310 			c |= QUOTE;
311 		    ungetC(c);
312 		    c = '\\';
313 		}
314 	    }
315 	    else if (c == '\n') {
316 		seterror(ERR_UNMATCHED, c1);
317 		ungetC(c);
318 		break;
319 	    }
320 	}
321 	else if (cmap(c, _META | _QF | _QB | _ESC)) {
322 	    if (c == '\\') {
323 		c = getC(0);
324 		if (c == '\n') {
325 		    if (onelflg == 1)
326 			onelflg = 2;
327 		    break;
328 		}
329 		if (c != HIST)
330 		    *wp++ = '\\', --i;
331 		c |= QUOTE;
332 	    }
333 	    else if (cmap(c, _QF | _QB)) {	/* '"` */
334 		c1 = c;
335 		dolflg = c == '"' ? DOALL : DOEXCL;
336 	    }
337 	    else if (c != '#' || !intty) {
338 		ungetC(c);
339 		break;
340 	    }
341 	}
342 	if (--i > 0) {
343 	    *wp++ = c;
344 	    c = getC(dolflg);
345 	}
346 	else {
347 	    seterror(ERR_WTOOLONG);
348 	    wp = &wbuf[1];
349 	    break;
350 	}
351     }
352 ret:
353     *wp = 0;
354     return (Strsave(wbuf));
355 }
356 
357 static int
358 getC1(int flag)
359 {
360     Char c;
361 
362     for (;;) {
363 	if ((c = peekc) != '\0') {
364 	    peekc = 0;
365 	    return (c);
366 	}
367 	if (lap) {
368 	    if ((c = *lap++) == 0)
369 		lap = 0;
370 	    else {
371 		if (cmap(c, _META | _QF | _QB))
372 		    c |= QUOTE;
373 		return (c);
374 	    }
375 	}
376 	if ((c = peekd) != '\0') {
377 	    peekd = 0;
378 	    return (c);
379 	}
380 	if (exclp) {
381 	    if ((c = *exclp++) != '\0')
382 		return (c);
383 	    if (exclnxt && --exclc >= 0) {
384 		exclnxt = exclnxt->next;
385 		setexclp(exclnxt->word);
386 		return (' ');
387 	    }
388 	    exclp = 0;
389 	    exclnxt = 0;
390 	}
391 	if (exclnxt) {
392 	    exclnxt = exclnxt->next;
393 	    if (--exclc < 0)
394 		exclnxt = 0;
395 	    else
396 		setexclp(exclnxt->word);
397 	    continue;
398 	}
399 	c = readc(0);
400 	if (c == '$' && (flag & DODOL)) {
401 	    getdol();
402 	    continue;
403 	}
404 	if (c == HIST && (flag & DOEXCL)) {
405 	    getexcl(0);
406 	    continue;
407 	}
408 	break;
409     }
410     return (c);
411 }
412 
413 static void
414 getdol(void)
415 {
416     Char name[4*MAXVARLEN+1], *ep, *np;
417     int c, sc;
418     bool special, toolong;
419 
420     special = 0;
421     np = name, *np++ = '$';
422     c = sc = getC(DOEXCL);
423     if (any("\t \n", c)) {
424 	ungetD(c);
425 	ungetC('$' | QUOTE);
426 	return;
427     }
428     if (c == '{')
429 	*np++ = c, c = getC(DOEXCL);
430     if (c == '#' || c == '?')
431 	special++, *np++ = c, c = getC(DOEXCL);
432     *np++ = c;
433     switch (c) {
434     case '<':
435     case '$':
436     case '!':
437 	if (special)
438 	    seterror(ERR_SPDOLLT);
439 	*np = 0;
440 	addla(name);
441 	return;
442     case '\n':
443 	ungetD(c);
444 	np--;
445 	seterror(ERR_NEWLINE);
446 	*np = 0;
447 	addla(name);
448 	return;
449     case '*':
450 	if (special)
451 	    seterror(ERR_SPSTAR);
452 	*np = 0;
453 	addla(name);
454 	return;
455     default:
456 	toolong = 0;
457 	if (Isdigit(c)) {
458 #ifdef notdef
459 	    /* let $?0 pass for now */
460 	    if (special) {
461 		seterror(ERR_DIGIT);
462 		*np = 0;
463 		addla(name);
464 		return;
465 	    }
466 #endif
467 	    /* we know that np < &name[4] */
468 	    ep = &np[MAXVARLEN];
469 	    while ((c = getC(DOEXCL)) != '\0'){
470 		if (!Isdigit(c))
471 		    break;
472 		if (np < ep)
473 		    *np++ = c;
474 		else
475 		    toolong = 1;
476 	    }
477 	}
478 	else if (letter(c)) {
479 	    /* we know that np < &name[4] */
480 	    ep = &np[MAXVARLEN];
481 	    toolong = 0;
482 	    while ((c = getC(DOEXCL)) != '\0') {
483 		/* Bugfix for ${v123x} from Chris Torek, DAS DEC-90. */
484 		if (!letter(c) && !Isdigit(c))
485 		    break;
486 		if (np < ep)
487 		    *np++ = c;
488 		else
489 		    toolong = 1;
490 	    }
491 	}
492 	else {
493 	    *np = 0;
494 	    seterror(ERR_VARILL);
495 	    addla(name);
496 	    return;
497 	}
498 	if (toolong) {
499 	    seterror(ERR_VARTOOLONG);
500 	    *np = 0;
501 	    addla(name);
502 	    return;
503 	}
504 	break;
505     }
506     if (c == '[') {
507 	*np++ = c;
508 	/*
509 	 * Name up to here is a max of MAXVARLEN + 8.
510 	 */
511 	ep = &np[2 * MAXVARLEN + 8];
512 	do {
513 	    /*
514 	     * Michael Greim: Allow $ expansion to take place in selector
515 	     * expressions. (limits the number of characters returned)
516 	     */
517 	    c = getC(DOEXCL | DODOL);
518 	    if (c == '\n') {
519 		ungetD(c);
520 		np--;
521 		seterror(ERR_NLINDEX);
522 		*np = 0;
523 		addla(name);
524 		return;
525 	    }
526 	    if (np < ep)
527 		*np++ = c;
528 	} while (c != ']');
529 	*np = '\0';
530 	if (np >= ep) {
531 	    seterror(ERR_SELOVFL);
532 	    addla(name);
533 	    return;
534 	}
535 	c = getC(DOEXCL);
536     }
537     /*
538      * Name up to here is a max of 2 * MAXVARLEN + 8.
539      */
540     if (c == ':') {
541 	/*
542 	 * if the :g modifier is followed by a newline, then error right away!
543 	 * -strike
544 	 */
545 	int amodflag, gmodflag;
546 
547 	amodflag = 0;
548 	gmodflag = 0;
549 	do {
550 	    *np++ = c, c = getC(DOEXCL);
551 	    if (c == 'g' || c == 'a') {
552 		if (c == 'g')
553 		    gmodflag++;
554 		else
555 		    amodflag++;
556 		*np++ = c; c = getC(DOEXCL);
557 	    }
558 	    if ((c == 'g' && !gmodflag) || (c == 'a' && !amodflag)) {
559 		if (c == 'g')
560 		    gmodflag++;
561 		else
562 		    amodflag++;
563 		*np++ = c; c = getC(DOEXCL);
564 	    }
565 	    *np++ = c;
566 	    /* scan s// [eichin:19910926.0512EST] */
567 	    if (c == 's') {
568 		int delimcnt = 2;
569 		int delim = getC(0);
570 		*np++ = delim;
571 
572 		if (!delim || letter(delim)
573 		    || Isdigit(delim) || any(" \t\n", delim)) {
574 		    seterror(ERR_BADSUBST);
575 		    break;
576 		}
577 		while ((c = getC(0)) != (-1)) {
578 		    *np++ = c;
579 		    if(c == delim) delimcnt--;
580 		    if(!delimcnt) break;
581 		}
582 		if(delimcnt) {
583 		    seterror(ERR_BADSUBST);
584 		    break;
585 		}
586 		c = 's';
587 	    }
588 	    if (!any("htrqxes", c)) {
589 		if ((amodflag || gmodflag) && c == '\n')
590 		    stderror(ERR_VARSYN);	/* strike */
591 		seterror(ERR_VARMOD, c);
592 		*np = 0;
593 		addla(name);
594 		return;
595 	    }
596 	}
597 	while ((c = getC(DOEXCL)) == ':');
598 	ungetD(c);
599     }
600     else
601 	ungetD(c);
602     if (sc == '{') {
603 	c = getC(DOEXCL);
604 	if (c != '}') {
605 	    ungetD(c);
606 	    seterror(ERR_MISSING, '}');
607 	    *np = 0;
608 	    addla(name);
609 	    return;
610 	}
611 	*np++ = c;
612     }
613     *np = 0;
614     addla(name);
615     return;
616 }
617 
618 void
619 addla(Char *cp)
620 {
621     Char buf[BUFSIZE];
622 
623     if (Strlen(cp) + (lap ? Strlen(lap) : 0) >=
624 	(sizeof(labuf) - 4) / sizeof(Char)) {
625 	seterror(ERR_EXPOVFL);
626 	return;
627     }
628     if (lap)
629 	(void)Strcpy(buf, lap);
630     (void)Strcpy(labuf, cp);
631     if (lap)
632 	(void)Strcat(labuf, buf);
633     lap = labuf;
634 }
635 
636 static Char lhsb[32];
637 static Char slhs[32];
638 static Char rhsb[64];
639 static int quesarg;
640 
641 static void
642 getexcl(int sc)
643 {
644     struct wordent *hp, *ip;
645     int c, dol, left, right;
646 
647     if (sc == 0) {
648 	sc = getC(0);
649 	if (sc != '{') {
650 	    ungetC(sc);
651 	    sc = 0;
652 	}
653     }
654     quesarg = -1;
655     lastev = eventno;
656     hp = gethent(sc);
657     if (hp == 0)
658 	return;
659     hadhist = 1;
660     dol = 0;
661     if (hp == alhistp)
662 	for (ip = hp->next->next; ip != alhistt; ip = ip->next)
663 	    dol++;
664     else
665 	for (ip = hp->next->next; ip != hp->prev; ip = ip->next)
666 	    dol++;
667     left = 0, right = dol;
668     if (sc == HISTSUB) {
669 	ungetC('s'), unreadc(HISTSUB), c = ':';
670 	goto subst;
671     }
672     c = getC(0);
673     if (!any(":^$*-%", c))
674 	goto subst;
675     left = right = -1;
676     if (c == ':') {
677 	c = getC(0);
678 	unreadc(c);
679 	if (letter(c) || c == '&') {
680 	    c = ':';
681 	    left = 0, right = dol;
682 	    goto subst;
683 	}
684     }
685     else
686 	ungetC(c);
687     if (!getsel(&left, &right, dol))
688 	return;
689     c = getC(0);
690     if (c == '*')
691 	ungetC(c), c = '-';
692     if (c == '-') {
693 	if (!getsel(&left, &right, dol))
694 	    return;
695 	c = getC(0);
696     }
697 subst:
698     exclc = right - left + 1;
699     while (--left >= 0)
700 	hp = hp->next;
701     if (sc == HISTSUB || c == ':') {
702 	do {
703 	    hp = getsub(hp);
704 	    c = getC(0);
705 	} while (c == ':');
706     }
707     unreadc(c);
708     if (sc == '{') {
709 	c = getC(0);
710 	if (c != '}')
711 	    seterror(ERR_BADBANG);
712     }
713     exclnxt = hp;
714 }
715 
716 static struct wordent *
717 getsub(struct wordent *en)
718 {
719     Char orhsb[sizeof(rhsb) / sizeof(Char)];
720     Char *cp;
721     int c, delim, sc;
722     bool global;
723 
724     do {
725 	exclnxt = 0;
726 	global = 0;
727 	sc = c = getC(0);
728 	if (c == 'g' || c == 'a') {
729 	    global |= (c == 'g') ? 1 : 2;
730 	    sc = c = getC(0);
731 	}
732 	if (((c =='g') && !(global & 1)) || ((c == 'a') && !(global & 2))) {
733 	    global |= (c == 'g') ? 1 : 2;
734 	    sc = c = getC(0);
735 	}
736 
737 	switch (c) {
738 	case 'p':
739 	    justpr++;
740 	    return (en);
741 	case 'x':
742 	case 'q':
743 	    global |= 1;
744 	    /* FALLTHROUGH */
745 	case 'h':
746 	case 'r':
747 	case 't':
748 	case 'e':
749 	    break;
750 	case '&':
751 	    if (slhs[0] == 0) {
752 		seterror(ERR_NOSUBST);
753 		return (en);
754 	    }
755 	    (void) Strcpy(lhsb, slhs);
756 	    break;
757 #ifdef notdef
758 	case '~':
759 	    if (lhsb[0] == 0)
760 		goto badlhs;
761 	    break;
762 #endif
763 	case 's':
764 	    delim = getC(0);
765 	    if (letter(delim) || Isdigit(delim) || any(" \t\n", delim)) {
766 		unreadc(delim);
767 		lhsb[0] = 0;
768 		seterror(ERR_BADSUBST);
769 		return (en);
770 	    }
771 	    cp = lhsb;
772 	    for (;;) {
773 		c = getC(0);
774 		if (c == '\n') {
775 		    unreadc(c);
776 		    break;
777 		}
778 		if (c == delim)
779 		    break;
780 		if (cp > &lhsb[sizeof(lhsb) / sizeof(Char) - 2]) {
781 		    lhsb[0] = 0;
782 		    seterror(ERR_BADSUBST);
783 		    return (en);
784 		}
785 		if (c == '\\') {
786 		    c = getC(0);
787 		    if (c != delim && c != '\\')
788 			*cp++ = '\\';
789 		}
790 		*cp++ = c;
791 	    }
792 	    if (cp != lhsb)
793 		*cp++ = 0;
794 	    else if (lhsb[0] == 0) {
795 		seterror(ERR_LHS);
796 		return (en);
797 	    }
798 	    cp = rhsb;
799 	    (void)Strcpy(orhsb, cp);
800 	    for (;;) {
801 		c = getC(0);
802 		if (c == '\n') {
803 		    unreadc(c);
804 		    break;
805 		}
806 		if (c == delim)
807 		    break;
808 #ifdef notdef
809 		if (c == '~') {
810 		    if (&cp[Strlen(orhsb)] > &rhsb[sizeof(rhsb) /
811 						   sizeof(Char) - 2])
812 			goto toorhs;
813 		    (void)Strcpy(cp, orhsb);
814 		    cp = Strend(cp);
815 		    continue;
816 		}
817 #endif
818 		if (cp > &rhsb[sizeof(rhsb) / sizeof(Char) - 2]) {
819 		    seterror(ERR_RHSLONG);
820 		    return (en);
821 		}
822 		if (c == '\\') {
823 		    c = getC(0);
824 		    if (c != delim /* && c != '~' */ )
825 			*cp++ = '\\';
826 		}
827 		*cp++ = c;
828 	    }
829 	    *cp++ = 0;
830 	    break;
831 	default:
832 	    if (c == '\n')
833 		unreadc(c);
834 	    seterror(ERR_BADBANGMOD, c);
835 	    return (en);
836 	}
837 	(void)Strcpy(slhs, lhsb);
838 	if (exclc)
839 	    en = dosub(sc, en, global);
840     }
841     while ((c = getC(0)) == ':');
842     unreadc(c);
843     return (en);
844 }
845 
846 static struct wordent *
847 dosub(int sc, struct wordent *en, bool global)
848 {
849     struct wordent lexi, *hp, *wdp;
850     int i;
851     bool didone, didsub;
852 
853     didone = 0;
854     didsub = 0;
855     i = exclc;
856     hp = &lexi;
857 
858     wdp = hp;
859     while (--i >= 0) {
860 	struct wordent *new = (struct wordent *)xcalloc(1, sizeof *wdp);
861 
862 	new->word = 0;
863 	new->prev = wdp;
864 	new->next = hp;
865 	wdp->next = new;
866 	wdp = new;
867 	en = en->next;
868 	if (en->word) {
869 	    Char *tword, *otword;
870 
871 	    if ((global & 1) || didsub == 0) {
872 		tword = subword(en->word, sc, &didone);
873 		if (didone)
874 		    didsub = 1;
875 		if (global & 2) {
876 		    while (didone && tword != STRNULL) {
877 			otword = tword;
878 			tword = subword(otword, sc, &didone);
879 			if (Strcmp(tword, otword) == 0) {
880 			    xfree((ptr_t) otword);
881 			    break;
882 			}
883 			else
884 			    xfree((ptr_t)otword);
885 		    }
886 		}
887 	    }
888 	    else
889 		tword = Strsave(en->word);
890 	    wdp->word = tword;
891 	}
892     }
893     if (didsub == 0)
894 	seterror(ERR_MODFAIL);
895     hp->prev = wdp;
896     return (&enthist(-1000, &lexi, 0)->Hlex);
897 }
898 
899 static Char *
900 subword(Char *cp, int type, bool *adid)
901 {
902     Char wbuf[BUFSIZE];
903     Char *mp, *np, *wp;
904     int i;
905 
906     *adid = 0;
907     switch (type) {
908     case 'r':
909     case 'e':
910     case 'h':
911     case 't':
912     case 'q':
913     case 'x':
914 	wp = domod(cp, type);
915 	if (wp == 0)
916 	    return (Strsave(cp));
917 	*adid = 1;
918 	return (wp);
919     default:
920 	wp = wbuf;
921 	i = BUFSIZE - 4;
922 	for (mp = cp; *mp; mp++)
923 	    if (matchs(mp, lhsb)) {
924 		for (np = cp; np < mp;)
925 		    *wp++ = *np++, --i;
926 		for (np = rhsb; *np; np++)
927 		    switch (*np) {
928 		    case '\\':
929 			if (np[1] == '&')
930 			    np++;
931 			/* FALLTHROUGH */
932 		    default:
933 			if (--i < 0) {
934 			    seterror(ERR_SUBOVFL);
935 			    return (STRNULL);
936 			}
937 			*wp++ = *np;
938 			continue;
939 		    case '&':
940 			i -= Strlen(lhsb);
941 			if (i < 0) {
942 			    seterror(ERR_SUBOVFL);
943 			    return (STRNULL);
944 			}
945 			*wp = 0;
946 			(void) Strcat(wp, lhsb);
947 			wp = Strend(wp);
948 			continue;
949 		    }
950 		mp += Strlen(lhsb);
951 		i -= Strlen(mp);
952 		if (i < 0) {
953 		    seterror(ERR_SUBOVFL);
954 		    return (STRNULL);
955 		}
956 		*wp = 0;
957 		(void) Strcat(wp, mp);
958 		*adid = 1;
959 		return (Strsave(wbuf));
960 	    }
961 	return (Strsave(cp));
962     }
963 }
964 
965 Char *
966 domod(Char *cp, int type)
967 {
968     Char *wp, *xp;
969     int c;
970 
971     switch (type) {
972     case 'x':
973     case 'q':
974 	wp = Strsave(cp);
975 	for (xp = wp; (c = *xp) != '\0'; xp++)
976 	    if ((c != ' ' && c != '\t') || type == 'q')
977 		*xp |= QUOTE;
978 	return (wp);
979     case 'h':
980     case 't':
981 	if (!any(short2str(cp), '/'))
982 	    return (type == 't' ? Strsave(cp) : 0);
983 	wp = Strend(cp);
984 	while (*--wp != '/')
985 	    continue;
986 	if (type == 'h')
987 	    xp = Strsave(cp), xp[wp - cp] = 0;
988 	else
989 	    xp = Strsave(wp + 1);
990 	return (xp);
991     case 'e':
992     case 'r':
993 	wp = Strend(cp);
994 	for (wp--; wp >= cp && *wp != '/'; wp--)
995 	    if (*wp == '.') {
996 		if (type == 'e')
997 		    xp = Strsave(wp + 1);
998 		else
999 		    xp = Strsave(cp), xp[wp - cp] = 0;
1000 		return (xp);
1001 	    }
1002 	return (Strsave(type == 'e' ? STRNULL : cp));
1003     default:
1004 	break;
1005     }
1006     return (0);
1007 }
1008 
1009 static int
1010 matchs(Char *str, Char *pat)
1011 {
1012     while (*str && *pat && *str == *pat)
1013 	str++, pat++;
1014     return (*pat == 0);
1015 }
1016 
1017 static int
1018 getsel(int *al, int *ar, int dol)
1019 {
1020     int c, i;
1021     bool first;
1022 
1023     c = getC(0);
1024     first = *al < 0;
1025 
1026     switch (c) {
1027     case '%':
1028 	if (quesarg == -1) {
1029 	    seterror(ERR_BADBANGARG);
1030 	    return (0);
1031 	}
1032 	if (*al < 0)
1033 	    *al = quesarg;
1034 	*ar = quesarg;
1035 	break;
1036     case '-':
1037 	if (*al < 0) {
1038 	    *al = 0;
1039 	    *ar = dol - 1;
1040 	    unreadc(c);
1041 	}
1042 	return (1);
1043     case '^':
1044 	if (*al < 0)
1045 	    *al = 1;
1046 	*ar = 1;
1047 	break;
1048     case '$':
1049 	if (*al < 0)
1050 	    *al = dol;
1051 	*ar = dol;
1052 	break;
1053     case '*':
1054 	if (*al < 0)
1055 	    *al = 1;
1056 	*ar = dol;
1057 	if (*ar < *al) {
1058 	    *ar = 0;
1059 	    *al = 1;
1060 	    return (1);
1061 	}
1062 	break;
1063     default:
1064 	if (Isdigit(c)) {
1065 	    i = 0;
1066 	    while (Isdigit(c)) {
1067 		i = i * 10 + c - '0';
1068 		c = getC(0);
1069 	    }
1070 	    if (i < 0)
1071 		i = dol + 1;
1072 	    if (*al < 0)
1073 		*al = i;
1074 	    *ar = i;
1075 	}
1076 	else if (*al < 0)
1077 	    *al = 0, *ar = dol;
1078 	else
1079 	    *ar = dol - 1;
1080 	unreadc(c);
1081 	break;
1082     }
1083     if (first) {
1084 	c = getC(0);
1085 	unreadc(c);
1086 	if (any("-$*", c))
1087 	    return (1);
1088     }
1089     if (*al > *ar || *ar > dol) {
1090 	seterror(ERR_BADBANGARG);
1091 	return (0);
1092     }
1093     return (1);
1094 
1095 }
1096 
1097 static struct wordent *
1098 gethent(int sc)
1099 {
1100     struct Hist *hp;
1101     Char *np;
1102     int c, event;
1103     bool back;
1104 
1105     back = 0;
1106     c = sc == HISTSUB ? HIST : getC(0);
1107     if (c == HIST) {
1108 	if (alhistp)
1109 	    return (alhistp);
1110 	event = eventno;
1111     }
1112     else
1113 	switch (c) {
1114 	case ':':
1115 	case '^':
1116 	case '$':
1117 	case '*':
1118 	case '%':
1119 	    ungetC(c);
1120 	    if (lastev == eventno && alhistp)
1121 		return (alhistp);
1122 	    event = lastev;
1123 	    break;
1124 	case '#':		/* !# is command being typed in (mrh) */
1125 	    if (--hleft == 0) {
1126 		seterror(ERR_HISTLOOP);
1127 		return (0);
1128 	    }
1129 	    else
1130 		return (&paraml);
1131 	    /* NOTREACHED */
1132 	case '-':
1133 	    back = 1;
1134 	    c = getC(0);
1135 	    /* FALLTHROUGH */
1136 	default:
1137 	    if (any("(=~", c)) {
1138 		unreadc(c);
1139 		ungetC(HIST);
1140 		return (0);
1141 	    }
1142 	    np = lhsb;
1143 	    event = 0;
1144 	    while (!cmap(c, _ESC | _META | _QF | _QB) && !any("${}:", c)) {
1145 		if (event != -1 && Isdigit(c))
1146 		    event = event * 10 + c - '0';
1147 		else
1148 		    event = -1;
1149 		if (np < &lhsb[sizeof(lhsb) / sizeof(Char) - 2])
1150 		    *np++ = c;
1151 		c = getC(0);
1152 	    }
1153 	    unreadc(c);
1154 	    if (np == lhsb) {
1155 		ungetC(HIST);
1156 		return (0);
1157 	    }
1158 	    *np++ = 0;
1159 	    if (event != -1) {
1160 		/*
1161 		 * History had only digits
1162 		 */
1163 		if (back)
1164 		    event = eventno + (alhistp == 0) - (event ? event : 0);
1165 		break;
1166 	    }
1167 	    hp = findev(lhsb, 0);
1168 	    if (hp)
1169 		lastev = hp->Hnum;
1170 	    return (&hp->Hlex);
1171 	case '?':
1172 	    np = lhsb;
1173 	    for (;;) {
1174 		c = getC(0);
1175 		if (c == '\n') {
1176 		    unreadc(c);
1177 		    break;
1178 		}
1179 		if (c == '?')
1180 		    break;
1181 		if (np < &lhsb[sizeof(lhsb) / sizeof(Char) - 2])
1182 		    *np++ = c;
1183 	    }
1184 	    if (np == lhsb) {
1185 		if (lhsb[0] == 0) {
1186 		    seterror(ERR_NOSEARCH);
1187 		    return (0);
1188 		}
1189 	    }
1190 	    else
1191 		*np++ = 0;
1192 	    hp = findev(lhsb, 1);
1193 	    if (hp)
1194 		lastev = hp->Hnum;
1195 	    return (&hp->Hlex);
1196 	}
1197 
1198     for (hp = Histlist.Hnext; hp; hp = hp->Hnext)
1199 	if (hp->Hnum == event) {
1200 	    hp->Href = eventno;
1201 	    lastev = hp->Hnum;
1202 	    return (&hp->Hlex);
1203 	}
1204     np = putn(event);
1205     seterror(ERR_NOEVENT, vis_str(np));
1206     return (0);
1207 }
1208 
1209 static struct Hist *
1210 findev(Char *cp, bool anyarg)
1211 {
1212     struct Hist *hp;
1213 
1214     for (hp = Histlist.Hnext; hp; hp = hp->Hnext) {
1215 	Char *dp, *p, *q;
1216 	struct wordent *lp;
1217 	int argno;
1218 
1219 	lp = hp->Hlex.next;
1220 	argno = 0;
1221 
1222 	/*
1223 	 * The entries added by alias substitution don't have a newline but do
1224 	 * have a negative event number. Savehist() trims off these entries,
1225 	 * but it happens before alias expansion, too early to delete those
1226 	 * from the previous command.
1227 	 */
1228 	if (hp->Hnum < 0)
1229 	    continue;
1230 	if (lp->word[0] == '\n')
1231 	    continue;
1232 	if (!anyarg) {
1233 	    p = cp;
1234 	    q = lp->word;
1235 	    do
1236 		if (!*p)
1237 		    return (hp);
1238 	    while (*p++ == *q++);
1239 	    continue;
1240 	}
1241 	do {
1242 	    for (dp = lp->word; *dp; dp++) {
1243 		p = cp;
1244 		q = dp;
1245 		do
1246 		    if (!*p) {
1247 			quesarg = argno;
1248 			return (hp);
1249 		    }
1250 		while (*p++ == *q++);
1251 	    }
1252 	    lp = lp->next;
1253 	    argno++;
1254 	} while (lp->word[0] != '\n');
1255     }
1256     seterror(ERR_NOEVENT, vis_str(cp));
1257     return (0);
1258 }
1259 
1260 
1261 static void
1262 setexclp(Char *cp)
1263 {
1264     if (cp && cp[0] == '\n')
1265 	return;
1266     exclp = cp;
1267 }
1268 
1269 void
1270 unreadc(int c)
1271 {
1272     peekread = c;
1273 }
1274 
1275 int
1276 readc(bool wanteof)
1277 {
1278     static int sincereal;
1279     int c;
1280 
1281     aret = F_SEEK;
1282     if ((c = peekread) != '\0') {
1283 	peekread = 0;
1284 	return (c);
1285     }
1286 top:
1287     aret = F_SEEK;
1288     if (alvecp) {
1289 	aret = A_SEEK;
1290 	if ((c = *alvecp++) != '\0')
1291 	    return (c);
1292 	if (alvec && *alvec) {
1293 		alvecp = *alvec++;
1294 		return (' ');
1295 	}
1296 	else {
1297 	    aret = F_SEEK;
1298 	    alvecp = NULL;
1299 	    return('\n');
1300 	}
1301     }
1302     if (alvec) {
1303 	if ((alvecp = *alvec) != '\0') {
1304 	    alvec++;
1305 	    goto top;
1306 	}
1307 	/* Infinite source! */
1308 	return ('\n');
1309     }
1310     if (evalp) {
1311 	aret = E_SEEK;
1312 	if ((c = *evalp++) != '\0')
1313 	    return (c);
1314 	if (evalvec && *evalvec) {
1315 	    evalp = *evalvec++;
1316 	    return (' ');
1317 	}
1318 	aret = F_SEEK;
1319 	evalp = 0;
1320     }
1321     if (evalvec) {
1322 	if (evalvec == (Char **) 1) {
1323 	    doneinp = 1;
1324 	    reset();
1325 	}
1326 	if ((evalp = *evalvec) != '\0') {
1327 	    evalvec++;
1328 	    goto top;
1329 	}
1330 	evalvec = (Char **) 1;
1331 	return ('\n');
1332     }
1333     do {
1334 	if (arginp == (Char *) 1 || onelflg == 1) {
1335 	    if (wanteof)
1336 		return (-1);
1337 	    exitstat();
1338 	}
1339 	if (arginp) {
1340 	    if ((c = *arginp++) == 0) {
1341 		arginp = (Char *) 1;
1342 		return ('\n');
1343 	    }
1344 	    return (c);
1345 	}
1346 reread:
1347 	c = bgetc();
1348 	if (c < 0) {
1349 	    struct termios tty;
1350 	    if (wanteof)
1351 		return (-1);
1352 	    /* was isatty but raw with ignoreeof yields problems */
1353 	    if (tcgetattr(SHIN, &tty) == 0 && (tty.c_lflag & ICANON))
1354 	    {
1355 		/* was 'short' for FILEC */
1356 		pid_t     ctpgrp;
1357 
1358 		if (++sincereal > 25)
1359 		    goto oops;
1360 		if (tpgrp != -1 &&
1361 		    (ctpgrp = tcgetpgrp(FSHTTY)) != -1 &&
1362 		    tpgrp != ctpgrp) {
1363 		    (void)tcsetpgrp(FSHTTY, tpgrp);
1364 		    (void)kill(-ctpgrp, SIGHUP);
1365 		    (void)fprintf(csherr, "Reset tty pgrp from %ld to %ld\n",
1366 				   (long)ctpgrp, (long)tpgrp);
1367 		    goto reread;
1368 		}
1369 		if (adrof(STRignoreeof)) {
1370 		    if (loginsh)
1371 			(void)fprintf(csherr,"\nUse \"logout\" to logout.\n");
1372 		    else
1373 			(void)fprintf(csherr,"\nUse \"exit\" to leave csh.\n");
1374 		    reset();
1375 		}
1376 		if (chkstop == 0)
1377 		    panystop(1);
1378 	    }
1379     oops:
1380 	    doneinp = 1;
1381 	    reset();
1382 	}
1383 	sincereal = 0;
1384 	if (c == '\n' && onelflg)
1385 	    onelflg--;
1386     } while (c == 0);
1387     return (c);
1388 }
1389 
1390 static int
1391 bgetc(void)
1392 {
1393 #ifdef FILEC
1394     char tbuf[BUFSIZE + 1];
1395     Char ttyline[BUFSIZE];
1396     int c, buf, numleft, off, roomleft;
1397 
1398     numleft = 0;
1399 #else /* FILEC */
1400     char tbuf[BUFSIZE + 1];
1401     int c, buf, off;
1402 #endif /* !FILEC */
1403 
1404     if (cantell) {
1405 	if (fseekp < fbobp || fseekp > feobp) {
1406 	    fbobp = feobp = fseekp;
1407 	    (void)lseek(SHIN, fseekp, SEEK_SET);
1408 	}
1409 	if (fseekp == feobp) {
1410 	    int i;
1411 
1412 	    fbobp = feobp;
1413 	    do
1414 		c = read(SHIN, tbuf, BUFSIZE);
1415 	    while (c < 0 && errno == EINTR);
1416 	    if (c <= 0)
1417 		return (-1);
1418 	    for (i = 0; i < c; i++)
1419 		fbuf[0][i] = (unsigned char) tbuf[i];
1420 	    feobp += c;
1421 	}
1422 	c = fbuf[0][fseekp - fbobp];
1423 	fseekp++;
1424 	return (c);
1425     }
1426 
1427 again:
1428     buf = (int) fseekp / BUFSIZE;
1429     if (buf >= fblocks) {
1430 	Char **nfbuf;
1431 
1432 	nfbuf = (Char **)xcalloc((size_t) (fblocks + 2), sizeof(char **));
1433 	if (fbuf) {
1434 	    (void)blkcpy(nfbuf, fbuf);
1435 	    xfree((ptr_t) fbuf);
1436 	}
1437 	fbuf = nfbuf;
1438 	fbuf[fblocks] = (Char *)xcalloc(BUFSIZE, sizeof(Char));
1439 	fblocks++;
1440 	if (!intty)
1441 	    goto again;
1442     }
1443     if (fseekp >= feobp) {
1444 	buf = (int) feobp / BUFSIZE;
1445 	off = (int) feobp % BUFSIZE;
1446 	roomleft = BUFSIZE - off;
1447 
1448 #ifdef FILEC
1449 	roomleft = BUFSIZE - off;
1450 	for (;;) {
1451 	    if (filec && intty) {
1452 		c = numleft ? numleft : tenex(ttyline, BUFSIZE);
1453 		if (c > roomleft) {
1454 		    /* start with fresh buffer */
1455 		    feobp = fseekp = fblocks * BUFSIZE;
1456 		    numleft = c;
1457 		    goto again;
1458 		}
1459 		if (c > 0)
1460 		    (void)memcpy(fbuf[buf] + off, ttyline, c * sizeof(Char));
1461 		numleft = 0;
1462 	    }
1463 	    else {
1464 #endif
1465 		c = read(SHIN, tbuf, roomleft);
1466 		if (c > 0) {
1467 		    int     i;
1468 		    Char   *ptr = fbuf[buf] + off;
1469 
1470 		    for (i = 0; i < c; i++)
1471 			ptr[i] = (unsigned char) tbuf[i];
1472 		}
1473 #ifdef FILEC
1474 	    }
1475 #endif
1476 	    if (c >= 0)
1477 		break;
1478 	    if (errno == EWOULDBLOCK) {
1479 		int     iooff = 0;
1480 
1481 		(void)ioctl(SHIN, FIONBIO, (ioctl_t) & iooff);
1482 	    }
1483 	    else if (errno != EINTR)
1484 		break;
1485 	}
1486 	if (c <= 0)
1487 	    return (-1);
1488 	feobp += c;
1489 #ifndef FILEC
1490 	goto again;
1491 #else
1492 	if (filec && !intty)
1493 	    goto again;
1494 #endif
1495     }
1496     c = fbuf[buf][(int)fseekp % BUFSIZE];
1497     fseekp++;
1498     return (c);
1499 }
1500 
1501 static void
1502 bfree(void)
1503 {
1504     int i, sb;
1505 
1506     if (cantell)
1507 	return;
1508     if (whyles)
1509 	return;
1510     sb = (int)(fseekp - 1) / BUFSIZE;
1511     if (sb > 0) {
1512 	for (i = 0; i < sb; i++)
1513 	    xfree((ptr_t) fbuf[i]);
1514 	(void)blkcpy(fbuf, &fbuf[sb]);
1515 	fseekp -= BUFSIZE * sb;
1516 	feobp -= BUFSIZE * sb;
1517 	fblocks -= sb;
1518     }
1519 }
1520 
1521 void
1522 bseek(struct Ain *l)
1523 {
1524     switch (aret = l->type) {
1525     case A_SEEK:
1526 	alvec = l->a_seek;
1527 	alvecp = l->c_seek;
1528 	return;
1529     case E_SEEK:
1530 	evalvec = l->a_seek;
1531 	evalp = l->c_seek;
1532 	return;
1533     case F_SEEK:
1534 	fseekp = l->f_seek;
1535 	return;
1536     default:
1537 	(void)fprintf(csherr, "Bad seek type %d\n", aret);
1538 	abort();
1539     }
1540 }
1541 
1542 void
1543 btell(struct Ain *l)
1544 {
1545     switch (l->type = aret) {
1546     case A_SEEK:
1547 	l->a_seek = alvec;
1548 	l->c_seek = alvecp;
1549 	return;
1550     case E_SEEK:
1551 	l->a_seek = evalvec;
1552 	l->c_seek = evalp;
1553 	return;
1554     case F_SEEK:
1555 	l->f_seek = fseekp;
1556 	l->a_seek = NULL;
1557 	return;
1558     default:
1559 	(void)fprintf(csherr, "Bad seek type %d\n", aret);
1560 	abort();
1561     }
1562 }
1563 
1564 void
1565 btoeof(void)
1566 {
1567     (void)lseek(SHIN, (off_t) 0, SEEK_END);
1568     aret = F_SEEK;
1569     fseekp = feobp;
1570     alvec = NULL;
1571     alvecp = NULL;
1572     evalvec = NULL;
1573     evalp = NULL;
1574     wfree();
1575     bfree();
1576 }
1577 
1578 void
1579 settell(void)
1580 {
1581     cantell = 0;
1582     if (arginp || onelflg || intty)
1583 	return;
1584     if (lseek(SHIN, (off_t) 0, SEEK_CUR) < 0 || errno == ESPIPE)
1585 	return;
1586     fbuf = (Char **)xcalloc(2, sizeof(Char **));
1587     fblocks = 1;
1588     fbuf[0] = (Char *)xcalloc(BUFSIZE, sizeof(Char));
1589     fseekp = fbobp = feobp = lseek(SHIN, (off_t) 0, SEEK_CUR);
1590     cantell = 1;
1591 }
1592