1/*	$OpenBSD: expr.c,v 1.24 2014/12/08 14:26:31 otto Exp $	*/
2
3/*-
4 * Copyright (c) 2003, 2004, 2005, 2006, 2007, 2008, 2009, 2010,
5 *		 2011, 2012, 2013, 2014, 2016
6 *	mirabilos <m@mirbsd.org>
7 *
8 * Provided that these terms and disclaimer and all copyright notices
9 * are retained or reproduced in an accompanying document, permission
10 * is granted to deal in this work without restriction, including un-
11 * limited rights to use, publicly perform, distribute, sell, modify,
12 * merge, give away, or sublicence.
13 *
14 * This work is provided "AS IS" and WITHOUT WARRANTY of any kind, to
15 * the utmost extent permitted by applicable law, neither express nor
16 * implied; without malicious intent or gross negligence. In no event
17 * may a licensor, author or contributor be held liable for indirect,
18 * direct, other damage, loss, or other issues arising in any way out
19 * of dealing in the work, even if advised of the possibility of such
20 * damage or existence of a defect, except proven that it results out
21 * of said person's immediate fault when using the work as intended.
22 */
23
24#include "sh.h"
25
26__RCSID("$MirOS: src/bin/mksh/expr.c,v 1.90 2016/11/07 16:58:48 tg Exp $");
27
28#define EXPRTOK_DEFNS
29#include "exprtok.h"
30
31/* precisions; used to be enum prec but we do arithmetics on it */
32#define P_PRIMARY	0	/* VAR, LIT, (), ! ~ ++ -- */
33#define P_MULT		1	/* * / % */
34#define P_ADD		2	/* + - */
35#define P_SHIFT		3	/* ^< ^> << >> */
36#define P_RELATION	4	/* < <= > >= */
37#define P_EQUALITY	5	/* == != */
38#define P_BAND		6	/* & */
39#define P_BXOR		7	/* ^ */
40#define P_BOR		8	/* | */
41#define P_LAND		9	/* && */
42#define P_LOR		10	/* || */
43#define P_TERN		11	/* ?: */
44	/* = += -= *= /= %= ^<= ^>= <<= >>= &= ^= |= */
45#define P_ASSIGN	12
46#define P_COMMA		13	/* , */
47#define MAX_PREC	P_COMMA
48
49enum token {
50#define EXPRTOK_ENUM
51#include "exprtok.h"
52};
53
54static const char opname[][4] = {
55#define EXPRTOK_NAME
56#include "exprtok.h"
57};
58
59static const uint8_t oplen[] = {
60#define EXPRTOK_LEN
61#include "exprtok.h"
62};
63
64static const uint8_t opprec[] = {
65#define EXPRTOK_PREC
66#include "exprtok.h"
67};
68
69typedef struct expr_state {
70	/* expression being evaluated */
71	const char *expression;
72	/* lexical position */
73	const char *tokp;
74	/* value from token() */
75	struct tbl *val;
76	/* variable that is being recursively expanded (EXPRINEVAL flag set) */
77	struct tbl *evaling;
78	/* token from token() */
79	enum token tok;
80	/* don't do assignments (for ?:, &&, ||) */
81	uint8_t noassign;
82	/* evaluating an $(()) expression? */
83	bool arith;
84	/* unsigned arithmetic calculation */
85	bool natural;
86} Expr_state;
87
88enum error_type {
89	ET_UNEXPECTED, ET_BADLIT, ET_RECURSIVE,
90	ET_LVALUE, ET_RDONLY, ET_STR
91};
92
93static void evalerr(Expr_state *, enum error_type, const char *)
94    MKSH_A_NORETURN;
95static struct tbl *evalexpr(Expr_state *, unsigned int);
96static void exprtoken(Expr_state *);
97static struct tbl *do_ppmm(Expr_state *, enum token, struct tbl *, bool);
98static void assign_check(Expr_state *, enum token, struct tbl *);
99static struct tbl *intvar(Expr_state *, struct tbl *);
100
101/*
102 * parse and evaluate expression
103 */
104int
105evaluate(const char *expr, mksh_ari_t *rval, int error_ok, bool arith)
106{
107	struct tbl v;
108	int ret;
109
110	v.flag = DEFINED | INTEGER;
111	v.type = 0;
112	ret = v_evaluate(&v, expr, error_ok, arith);
113	*rval = v.val.i;
114	return (ret);
115}
116
117/*
118 * parse and evaluate expression, storing result in vp.
119 */
120int
121v_evaluate(struct tbl *vp, const char *expr, volatile int error_ok,
122    bool arith)
123{
124	struct tbl *v;
125	Expr_state curstate;
126	Expr_state * const es = &curstate;
127	int i;
128
129	/* save state to allow recursive calls */
130	memset(&curstate, 0, sizeof(curstate));
131	curstate.expression = curstate.tokp = expr;
132	curstate.tok = BAD;
133	curstate.arith = arith;
134
135	newenv(E_ERRH);
136	if ((i = kshsetjmp(e->jbuf))) {
137		/* Clear EXPRINEVAL in of any variables we were playing with */
138		if (curstate.evaling)
139			curstate.evaling->flag &= ~EXPRINEVAL;
140		quitenv(NULL);
141		if (i == LAEXPR) {
142			if (error_ok == KSH_RETURN_ERROR)
143				return (0);
144			errorfz();
145		}
146		unwind(i);
147		/* NOTREACHED */
148	}
149
150	exprtoken(es);
151	if (es->tok == END) {
152		es->tok = LIT;
153		es->val = tempvar("");
154	}
155	v = intvar(es, evalexpr(es, MAX_PREC));
156
157	if (es->tok != END)
158		evalerr(es, ET_UNEXPECTED, NULL);
159
160	if (es->arith && es->natural)
161		vp->flag |= INT_U;
162	if (vp->flag & INTEGER)
163		setint_v(vp, v, es->arith);
164	else
165		/* can fail if readonly */
166		setstr(vp, str_val(v), error_ok);
167
168	quitenv(NULL);
169
170	return (1);
171}
172
173static void
174evalerr(Expr_state *es, enum error_type type, const char *str)
175{
176	char tbuf[2];
177	const char *s;
178
179	es->arith = false;
180	switch (type) {
181	case ET_UNEXPECTED:
182		switch (es->tok) {
183		case VAR:
184			s = es->val->name;
185			break;
186		case LIT:
187			s = str_val(es->val);
188			break;
189		case END:
190			s = "end of expression";
191			break;
192		case BAD:
193			tbuf[0] = *es->tokp;
194			tbuf[1] = '\0';
195			s = tbuf;
196			break;
197		default:
198			s = opname[(int)es->tok];
199		}
200		warningf(true, Tf_sD_s_qs, es->expression,
201		    Tunexpected, s);
202		break;
203
204	case ET_BADLIT:
205		warningf(true, Tf_sD_s_qs, es->expression,
206		    "bad number", str);
207		break;
208
209	case ET_RECURSIVE:
210		warningf(true, Tf_sD_s_qs, es->expression,
211		    "expression recurses on parameter", str);
212		break;
213
214	case ET_LVALUE:
215		warningf(true, Tf_sD_s_s,
216		    es->expression, str, "requires lvalue");
217		break;
218
219	case ET_RDONLY:
220		warningf(true, Tf_sD_s_s,
221		    es->expression, str, "applied to read-only variable");
222		break;
223
224	default: /* keep gcc happy */
225	case ET_STR:
226		warningf(true, Tf_sD_s, es->expression, str);
227		break;
228	}
229	unwind(LAEXPR);
230}
231
232/* do a ++ or -- operation */
233static struct tbl *
234do_ppmm(Expr_state *es, enum token op, struct tbl *vasn, bool is_prefix)
235{
236	struct tbl *vl;
237	mksh_uari_t oval;
238
239	assign_check(es, op, vasn);
240
241	vl = intvar(es, vasn);
242	oval = vl->val.u;
243	if (op == O_PLUSPLUS)
244		++vl->val.u;
245	else
246		--vl->val.u;
247	if (!es->noassign) {
248		if (vasn->flag & INTEGER)
249			setint_v(vasn, vl, es->arith);
250		else
251			setint(vasn, vl->val.i);
252	}
253	if (!is_prefix)
254		/* undo the increment/decrement */
255		vl->val.u = oval;
256
257	return (vl);
258}
259
260static struct tbl *
261evalexpr(Expr_state *es, unsigned int prec)
262{
263	struct tbl *vl, *vr = NULL, *vasn;
264	enum token op;
265	mksh_uari_t res = 0, t1, t2, t3;
266
267	if (prec == P_PRIMARY) {
268		switch ((int)(op = es->tok)) {
269		case O_BNOT:
270		case O_LNOT:
271		case O_MINUS:
272		case O_PLUS:
273			exprtoken(es);
274			vl = intvar(es, evalexpr(es, P_PRIMARY));
275			switch ((int)op) {
276			case O_BNOT:
277				vl->val.u = ~vl->val.u;
278				break;
279			case O_LNOT:
280				vl->val.u = !vl->val.u;
281				break;
282			case O_MINUS:
283				vl->val.u = -vl->val.u;
284				break;
285			case O_PLUS:
286				/* nop */
287				break;
288			}
289			break;
290
291		case OPEN_PAREN:
292			exprtoken(es);
293			vl = evalexpr(es, MAX_PREC);
294			if (es->tok != CLOSE_PAREN)
295				evalerr(es, ET_STR, "missing )");
296			exprtoken(es);
297			break;
298
299		case O_PLUSPLUS:
300		case O_MINUSMINUS:
301			exprtoken(es);
302			vl = do_ppmm(es, op, es->val, true);
303			exprtoken(es);
304			break;
305
306		case VAR:
307		case LIT:
308			vl = es->val;
309			exprtoken(es);
310			break;
311
312		default:
313			evalerr(es, ET_UNEXPECTED, NULL);
314			/* NOTREACHED */
315		}
316
317		if (es->tok == O_PLUSPLUS || es->tok == O_MINUSMINUS) {
318			vl = do_ppmm(es, es->tok, vl, false);
319			exprtoken(es);
320		}
321
322		return (vl);
323		/* prec == P_PRIMARY */
324	}
325
326	vl = evalexpr(es, prec - 1);
327	while ((int)(op = es->tok) >= (int)O_EQ && (int)op <= (int)O_COMMA &&
328	    opprec[(int)op] == prec) {
329		switch ((int)op) {
330		case O_TERN:
331		case O_LAND:
332		case O_LOR:
333			break;
334		default:
335			exprtoken(es);
336		}
337
338		vasn = vl;
339		if (op != O_ASN)
340			/* vl may not have a value yet */
341			vl = intvar(es, vl);
342		if (IS_ASSIGNOP(op)) {
343			if (!es->noassign)
344				assign_check(es, op, vasn);
345			vr = intvar(es, evalexpr(es, P_ASSIGN));
346		} else if (op == O_TERN) {
347			bool ev = vl->val.u != 0;
348
349			if (!ev)
350				es->noassign++;
351			exprtoken(es);
352			vl = evalexpr(es, MAX_PREC);
353			if (!ev)
354				es->noassign--;
355			if (es->tok != CTERN)
356				evalerr(es, ET_STR, "missing :");
357			if (ev)
358				es->noassign++;
359			exprtoken(es);
360			vr = evalexpr(es, P_TERN);
361			if (ev)
362				es->noassign--;
363			vl = ev ? vl : vr;
364			continue;
365		} else if (op != O_LAND && op != O_LOR)
366			vr = intvar(es, evalexpr(es, prec - 1));
367
368		/* common ops setup */
369		switch ((int)op) {
370		case O_DIV:
371		case O_DIVASN:
372		case O_MOD:
373		case O_MODASN:
374			if (vr->val.u == 0) {
375				if (!es->noassign)
376					evalerr(es, ET_STR, "zero divisor");
377				vr->val.u = 1;
378			}
379			/* calculate the absolute values */
380			t1 = vl->val.i < 0 ? -vl->val.u : vl->val.u;
381			t2 = vr->val.i < 0 ? -vr->val.u : vr->val.u;
382			break;
383#ifndef MKSH_LEGACY_MODE
384		case O_LSHIFT:
385		case O_LSHIFTASN:
386		case O_RSHIFT:
387		case O_RSHIFTASN:
388		case O_ROL:
389		case O_ROLASN:
390		case O_ROR:
391		case O_RORASN:
392			t1 = vl->val.u;
393			t2 = vr->val.u & 31;
394			break;
395#endif
396		case O_LAND:
397		case O_LOR:
398			t1 = vl->val.u;
399			t2 = 0;	/* gcc */
400			break;
401		default:
402			t1 = vl->val.u;
403			t2 = vr->val.u;
404			break;
405		}
406
407#define cmpop(op)	(es->natural ?			\
408	(mksh_uari_t)(vl->val.u op vr->val.u) :		\
409	(mksh_uari_t)(vl->val.i op vr->val.i)		\
410)
411
412		/* op calculation */
413		switch ((int)op) {
414		case O_TIMES:
415		case O_TIMESASN:
416			res = t1 * t2;
417			break;
418		case O_MOD:
419		case O_MODASN:
420			if (es->natural) {
421				res = vl->val.u % vr->val.u;
422				break;
423			}
424			goto signed_division;
425		case O_DIV:
426		case O_DIVASN:
427			if (es->natural) {
428				res = vl->val.u / vr->val.u;
429				break;
430			}
431 signed_division:
432			/*
433			 * a / b = abs(a) / abs(b) * sgn((u)a^(u)b)
434			 */
435			t3 = t1 / t2;
436#ifndef MKSH_LEGACY_MODE
437			res = ((vl->val.u ^ vr->val.u) & 0x80000000) ? -t3 : t3;
438#else
439			res = ((t1 == vl->val.u ? 0 : 1) ^
440			    (t2 == vr->val.u ? 0 : 1)) ? -t3 : t3;
441#endif
442			if (op == O_MOD || op == O_MODASN) {
443				/*
444				 * primitive modulo, to get the sign of
445				 * the result correct:
446				 * (a % b) = a - ((a / b) * b)
447				 * the subtraction and multiplication
448				 * are, amazingly enough, sign ignorant
449				 */
450				res = vl->val.u - (res * vr->val.u);
451			}
452			break;
453		case O_PLUS:
454		case O_PLUSASN:
455			res = t1 + t2;
456			break;
457		case O_MINUS:
458		case O_MINUSASN:
459			res = t1 - t2;
460			break;
461#ifndef MKSH_LEGACY_MODE
462		case O_ROL:
463		case O_ROLASN:
464			res = (t1 << t2) | (t1 >> (32 - t2));
465			break;
466		case O_ROR:
467		case O_RORASN:
468			res = (t1 >> t2) | (t1 << (32 - t2));
469			break;
470#endif
471		case O_LSHIFT:
472		case O_LSHIFTASN:
473			res = t1 << t2;
474			break;
475		case O_RSHIFT:
476		case O_RSHIFTASN:
477			res = es->natural || vl->val.i >= 0 ?
478			    t1 >> t2 :
479			    ~(~t1 >> t2);
480			break;
481		case O_LT:
482			res = cmpop(<);
483			break;
484		case O_LE:
485			res = cmpop(<=);
486			break;
487		case O_GT:
488			res = cmpop(>);
489			break;
490		case O_GE:
491			res = cmpop(>=);
492			break;
493		case O_EQ:
494			res = t1 == t2;
495			break;
496		case O_NE:
497			res = t1 != t2;
498			break;
499		case O_BAND:
500		case O_BANDASN:
501			res = t1 & t2;
502			break;
503		case O_BXOR:
504		case O_BXORASN:
505			res = t1 ^ t2;
506			break;
507		case O_BOR:
508		case O_BORASN:
509			res = t1 | t2;
510			break;
511		case O_LAND:
512			if (!t1)
513				es->noassign++;
514			exprtoken(es);
515			vr = intvar(es, evalexpr(es, prec - 1));
516			res = t1 && vr->val.u;
517			if (!t1)
518				es->noassign--;
519			break;
520		case O_LOR:
521			if (t1)
522				es->noassign++;
523			exprtoken(es);
524			vr = intvar(es, evalexpr(es, prec - 1));
525			res = t1 || vr->val.u;
526			if (t1)
527				es->noassign--;
528			break;
529		case O_ASN:
530		case O_COMMA:
531			res = t2;
532			break;
533		}
534
535#undef cmpop
536
537		if (IS_ASSIGNOP(op)) {
538			vr->val.u = res;
539			if (!es->noassign) {
540				if (vasn->flag & INTEGER)
541					setint_v(vasn, vr, es->arith);
542				else
543					setint(vasn, vr->val.i);
544			}
545			vl = vr;
546		} else
547			vl->val.u = res;
548	}
549	return (vl);
550}
551
552static void
553exprtoken(Expr_state *es)
554{
555	const char *cp = es->tokp;
556	int c;
557	char *tvar;
558
559	/* skip whitespace */
560 skip_spaces:
561	while ((c = *cp), ksh_isspace(c))
562		++cp;
563	if (es->tokp == es->expression && c == '#') {
564		/* expression begins with # */
565		/* switch to unsigned */
566		es->natural = true;
567		++cp;
568		goto skip_spaces;
569	}
570	es->tokp = cp;
571
572	if (c == '\0')
573		es->tok = END;
574	else if (ksh_isalphx(c)) {
575		for (; ksh_isalnux(c); c = *cp)
576			cp++;
577		if (c == '[') {
578			size_t len;
579
580			len = array_ref_len(cp);
581			if (len == 0)
582				evalerr(es, ET_STR, "missing ]");
583			cp += len;
584		}
585		if (es->noassign) {
586			es->val = tempvar("");
587			es->val->flag |= EXPRLVALUE;
588		} else {
589			strndupx(tvar, es->tokp, cp - es->tokp, ATEMP);
590			es->val = global(tvar);
591			afree(tvar, ATEMP);
592		}
593		es->tok = VAR;
594	} else if (c == '1' && cp[1] == '#') {
595		cp += 2;
596		if (*cp)
597			cp += utf_ptradj(cp);
598		strndupx(tvar, es->tokp, cp - es->tokp, ATEMP);
599		goto process_tvar;
600#ifndef MKSH_SMALL
601	} else if (c == '\'') {
602		if (*++cp == '\0') {
603			es->tok = END;
604			evalerr(es, ET_UNEXPECTED, NULL);
605		}
606		cp += utf_ptradj(cp);
607		if (*cp++ != '\'')
608			evalerr(es, ET_STR,
609			    "multi-character character constant");
610		/* 'x' -> 1#x (x = one multibyte character) */
611		c = cp - es->tokp;
612		tvar = alloc(c + /* NUL */ 1, ATEMP);
613		tvar[0] = '1';
614		tvar[1] = '#';
615		memcpy(tvar + 2, es->tokp + 1, c - 2);
616		tvar[c] = '\0';
617		goto process_tvar;
618#endif
619	} else if (ksh_isdigit(c)) {
620		while (c != '_' && (ksh_isalnux(c) || c == '#'))
621			c = *cp++;
622		strndupx(tvar, es->tokp, --cp - es->tokp, ATEMP);
623 process_tvar:
624		es->val = tempvar("");
625		es->val->flag &= ~INTEGER;
626		es->val->type = 0;
627		es->val->val.s = tvar;
628		if (setint_v(es->val, es->val, es->arith) == NULL)
629			evalerr(es, ET_BADLIT, tvar);
630		afree(tvar, ATEMP);
631		es->tok = LIT;
632	} else {
633		int i, n0;
634
635		for (i = 0; (n0 = opname[i][0]); i++)
636			if (c == n0 && strncmp(cp, opname[i],
637			    (size_t)oplen[i]) == 0) {
638				es->tok = (enum token)i;
639				cp += oplen[i];
640				break;
641			}
642		if (!n0)
643			es->tok = BAD;
644	}
645	es->tokp = cp;
646}
647
648static void
649assign_check(Expr_state *es, enum token op, struct tbl *vasn)
650{
651	if (es->tok == END || !vasn ||
652	    (vasn->name[0] == '\0' && !(vasn->flag & EXPRLVALUE)))
653		evalerr(es, ET_LVALUE, opname[(int)op]);
654	else if (vasn->flag & RDONLY)
655		evalerr(es, ET_RDONLY, opname[(int)op]);
656}
657
658struct tbl *
659tempvar(const char *vname)
660{
661	struct tbl *vp;
662	size_t vsize;
663
664	vsize = strlen(vname) + 1;
665	vp = alloc(offsetof(struct tbl, name[0]) + vsize, ATEMP);
666	memcpy(vp->name, vname, vsize);
667	vp->flag = ISSET|INTEGER;
668	vp->type = 0;
669	vp->areap = ATEMP;
670	vp->ua.hval = 0;
671	vp->val.i = 0;
672	return (vp);
673}
674
675/* cast (string) variable to temporary integer variable */
676static struct tbl *
677intvar(Expr_state *es, struct tbl *vp)
678{
679	struct tbl *vq;
680
681	/* try to avoid replacing a temp var with another temp var */
682	if (vp->name[0] == '\0' &&
683	    (vp->flag & (ISSET|INTEGER|EXPRLVALUE)) == (ISSET|INTEGER))
684		return (vp);
685
686	vq = tempvar("");
687	if (setint_v(vq, vp, es->arith) == NULL) {
688		if (vp->flag & EXPRINEVAL)
689			evalerr(es, ET_RECURSIVE, vp->name);
690		es->evaling = vp;
691		vp->flag |= EXPRINEVAL;
692		v_evaluate(vq, str_val(vp), KSH_UNWIND_ERROR, es->arith);
693		vp->flag &= ~EXPRINEVAL;
694		es->evaling = NULL;
695	}
696	return (vq);
697}
698
699
700/*
701 * UTF-8 support code: high-level functions
702 */
703
704int
705utf_widthadj(const char *src, const char **dst)
706{
707	size_t len;
708	unsigned int wc;
709	int width;
710
711	if (!UTFMODE || (len = utf_mbtowc(&wc, src)) == (size_t)-1 ||
712	    wc == 0)
713		len = width = 1;
714	else if ((width = utf_wcwidth(wc)) < 0)
715		/* XXX use 2 for x_zotc3 here? */
716		width = 1;
717
718	if (dst)
719		*dst = src + len;
720	return (width);
721}
722
723size_t
724utf_mbswidth(const char *s)
725{
726	size_t len, width = 0;
727	unsigned int wc;
728	int cw;
729
730	if (!UTFMODE)
731		return (strlen(s));
732
733	while (*s)
734		if (((len = utf_mbtowc(&wc, s)) == (size_t)-1) ||
735		    ((cw = utf_wcwidth(wc)) == -1)) {
736			s++;
737			width += 1;
738		} else {
739			s += len;
740			width += cw;
741		}
742	return (width);
743}
744
745const char *
746utf_skipcols(const char *p, int cols, int *colp)
747{
748	int c = 0;
749	const char *q;
750
751	while (c < cols) {
752		if (!*p) {
753			/* end of input; special handling for edit.c */
754			if (!colp)
755				return (p + cols - c);
756			*colp = c;
757			return (p);
758		}
759		c += utf_widthadj(p, &p);
760	}
761	if (UTFMODE)
762		while (utf_widthadj(p, &q) == 0)
763			p = q;
764	if (colp)
765		*colp = c;
766	return (p);
767}
768
769size_t
770utf_ptradj(const char *src)
771{
772	register size_t n;
773
774	if (!UTFMODE ||
775	    *(const unsigned char *)(src) < 0xC2 ||
776	    (n = utf_mbtowc(NULL, src)) == (size_t)-1)
777		n = 1;
778	return (n);
779}
780
781/*
782 * UTF-8 support code: low-level functions
783 */
784
785/* CESU-8 multibyte and wide character conversion crafted for mksh */
786
787size_t
788utf_mbtowc(unsigned int *dst, const char *src)
789{
790	const unsigned char *s = (const unsigned char *)src;
791	unsigned int c, wc;
792
793	if ((wc = *s++) < 0x80) {
794 out:
795		if (dst != NULL)
796			*dst = wc;
797		return (wc ? ((const char *)s - src) : 0);
798	}
799	if (wc < 0xC2 || wc >= 0xF0)
800		/* < 0xC0: spurious second byte */
801		/* < 0xC2: non-minimalistic mapping error in 2-byte seqs */
802		/* > 0xEF: beyond BMP */
803		goto ilseq;
804
805	if (wc < 0xE0) {
806		wc = (wc & 0x1F) << 6;
807		if (((c = *s++) & 0xC0) != 0x80)
808			goto ilseq;
809		wc |= c & 0x3F;
810		goto out;
811	}
812
813	wc = (wc & 0x0F) << 12;
814
815	if (((c = *s++) & 0xC0) != 0x80)
816		goto ilseq;
817	wc |= (c & 0x3F) << 6;
818
819	if (((c = *s++) & 0xC0) != 0x80)
820		goto ilseq;
821	wc |= c & 0x3F;
822
823	/* Check for non-minimalistic mapping error in 3-byte seqs */
824	if (wc >= 0x0800 && wc <= 0xFFFD)
825		goto out;
826 ilseq:
827	return ((size_t)(-1));
828}
829
830size_t
831utf_wctomb(char *dst, unsigned int wc)
832{
833	unsigned char *d;
834
835	if (wc < 0x80) {
836		*dst = wc;
837		return (1);
838	}
839
840	d = (unsigned char *)dst;
841	if (wc < 0x0800)
842		*d++ = (wc >> 6) | 0xC0;
843	else {
844		*d++ = ((wc = wc > 0xFFFD ? 0xFFFD : wc) >> 12) | 0xE0;
845		*d++ = ((wc >> 6) & 0x3F) | 0x80;
846	}
847	*d++ = (wc & 0x3F) | 0x80;
848	return ((char *)d - dst);
849}
850
851/*
852 * Wrapper around access(2) because it says root can execute everything
853 * on some operating systems. Does not set errno, no user needs it. Use
854 * this iff mode can have the X_OK bit set, access otherwise.
855 */
856int
857ksh_access(const char *fn, int mode)
858{
859	int rv;
860	struct stat sb;
861
862	if ((rv = access(fn, mode)) == 0 && kshuid == 0 && (mode & X_OK) &&
863	    (rv = stat(fn, &sb)) == 0 && !S_ISDIR(sb.st_mode) &&
864	    (sb.st_mode & (S_IXUSR|S_IXGRP|S_IXOTH)) == 0)
865		rv = -1;
866
867	return (rv);
868}
869
870#ifndef MIRBSD_BOOTFLOPPY
871/* From: X11/xc/programs/xterm/wcwidth.c,v 1.9 */
872
873struct mb_ucsrange {
874	unsigned short beg;
875	unsigned short end;
876};
877
878static int mb_ucsbsearch(const struct mb_ucsrange arr[], size_t elems,
879    unsigned int val) MKSH_A_PURE;
880
881/*
882 * Generated from the Unicode Character Database, Version 9.0.0, by
883 * MirOS: contrib/code/Snippets/eawparse,v 1.3 2014/11/16 12:16:24 tg Exp $
884 */
885
886static const struct mb_ucsrange mb_ucs_combining[] = {
887	{ 0x0300, 0x036F },
888	{ 0x0483, 0x0489 },
889	{ 0x0591, 0x05BD },
890	{ 0x05BF, 0x05BF },
891	{ 0x05C1, 0x05C2 },
892	{ 0x05C4, 0x05C5 },
893	{ 0x05C7, 0x05C7 },
894	{ 0x0600, 0x0605 },
895	{ 0x0610, 0x061A },
896	{ 0x061C, 0x061C },
897	{ 0x064B, 0x065F },
898	{ 0x0670, 0x0670 },
899	{ 0x06D6, 0x06DD },
900	{ 0x06DF, 0x06E4 },
901	{ 0x06E7, 0x06E8 },
902	{ 0x06EA, 0x06ED },
903	{ 0x070F, 0x070F },
904	{ 0x0711, 0x0711 },
905	{ 0x0730, 0x074A },
906	{ 0x07A6, 0x07B0 },
907	{ 0x07EB, 0x07F3 },
908	{ 0x0816, 0x0819 },
909	{ 0x081B, 0x0823 },
910	{ 0x0825, 0x0827 },
911	{ 0x0829, 0x082D },
912	{ 0x0859, 0x085B },
913	{ 0x08D4, 0x0902 },
914	{ 0x093A, 0x093A },
915	{ 0x093C, 0x093C },
916	{ 0x0941, 0x0948 },
917	{ 0x094D, 0x094D },
918	{ 0x0951, 0x0957 },
919	{ 0x0962, 0x0963 },
920	{ 0x0981, 0x0981 },
921	{ 0x09BC, 0x09BC },
922	{ 0x09C1, 0x09C4 },
923	{ 0x09CD, 0x09CD },
924	{ 0x09E2, 0x09E3 },
925	{ 0x0A01, 0x0A02 },
926	{ 0x0A3C, 0x0A3C },
927	{ 0x0A41, 0x0A42 },
928	{ 0x0A47, 0x0A48 },
929	{ 0x0A4B, 0x0A4D },
930	{ 0x0A51, 0x0A51 },
931	{ 0x0A70, 0x0A71 },
932	{ 0x0A75, 0x0A75 },
933	{ 0x0A81, 0x0A82 },
934	{ 0x0ABC, 0x0ABC },
935	{ 0x0AC1, 0x0AC5 },
936	{ 0x0AC7, 0x0AC8 },
937	{ 0x0ACD, 0x0ACD },
938	{ 0x0AE2, 0x0AE3 },
939	{ 0x0B01, 0x0B01 },
940	{ 0x0B3C, 0x0B3C },
941	{ 0x0B3F, 0x0B3F },
942	{ 0x0B41, 0x0B44 },
943	{ 0x0B4D, 0x0B4D },
944	{ 0x0B56, 0x0B56 },
945	{ 0x0B62, 0x0B63 },
946	{ 0x0B82, 0x0B82 },
947	{ 0x0BC0, 0x0BC0 },
948	{ 0x0BCD, 0x0BCD },
949	{ 0x0C00, 0x0C00 },
950	{ 0x0C3E, 0x0C40 },
951	{ 0x0C46, 0x0C48 },
952	{ 0x0C4A, 0x0C4D },
953	{ 0x0C55, 0x0C56 },
954	{ 0x0C62, 0x0C63 },
955	{ 0x0C81, 0x0C81 },
956	{ 0x0CBC, 0x0CBC },
957	{ 0x0CBF, 0x0CBF },
958	{ 0x0CC6, 0x0CC6 },
959	{ 0x0CCC, 0x0CCD },
960	{ 0x0CE2, 0x0CE3 },
961	{ 0x0D01, 0x0D01 },
962	{ 0x0D41, 0x0D44 },
963	{ 0x0D4D, 0x0D4D },
964	{ 0x0D62, 0x0D63 },
965	{ 0x0DCA, 0x0DCA },
966	{ 0x0DD2, 0x0DD4 },
967	{ 0x0DD6, 0x0DD6 },
968	{ 0x0E31, 0x0E31 },
969	{ 0x0E34, 0x0E3A },
970	{ 0x0E47, 0x0E4E },
971	{ 0x0EB1, 0x0EB1 },
972	{ 0x0EB4, 0x0EB9 },
973	{ 0x0EBB, 0x0EBC },
974	{ 0x0EC8, 0x0ECD },
975	{ 0x0F18, 0x0F19 },
976	{ 0x0F35, 0x0F35 },
977	{ 0x0F37, 0x0F37 },
978	{ 0x0F39, 0x0F39 },
979	{ 0x0F71, 0x0F7E },
980	{ 0x0F80, 0x0F84 },
981	{ 0x0F86, 0x0F87 },
982	{ 0x0F8D, 0x0F97 },
983	{ 0x0F99, 0x0FBC },
984	{ 0x0FC6, 0x0FC6 },
985	{ 0x102D, 0x1030 },
986	{ 0x1032, 0x1037 },
987	{ 0x1039, 0x103A },
988	{ 0x103D, 0x103E },
989	{ 0x1058, 0x1059 },
990	{ 0x105E, 0x1060 },
991	{ 0x1071, 0x1074 },
992	{ 0x1082, 0x1082 },
993	{ 0x1085, 0x1086 },
994	{ 0x108D, 0x108D },
995	{ 0x109D, 0x109D },
996	{ 0x1160, 0x11FF },
997	{ 0x135D, 0x135F },
998	{ 0x1712, 0x1714 },
999	{ 0x1732, 0x1734 },
1000	{ 0x1752, 0x1753 },
1001	{ 0x1772, 0x1773 },
1002	{ 0x17B4, 0x17B5 },
1003	{ 0x17B7, 0x17BD },
1004	{ 0x17C6, 0x17C6 },
1005	{ 0x17C9, 0x17D3 },
1006	{ 0x17DD, 0x17DD },
1007	{ 0x180B, 0x180E },
1008	{ 0x1885, 0x1886 },
1009	{ 0x18A9, 0x18A9 },
1010	{ 0x1920, 0x1922 },
1011	{ 0x1927, 0x1928 },
1012	{ 0x1932, 0x1932 },
1013	{ 0x1939, 0x193B },
1014	{ 0x1A17, 0x1A18 },
1015	{ 0x1A1B, 0x1A1B },
1016	{ 0x1A56, 0x1A56 },
1017	{ 0x1A58, 0x1A5E },
1018	{ 0x1A60, 0x1A60 },
1019	{ 0x1A62, 0x1A62 },
1020	{ 0x1A65, 0x1A6C },
1021	{ 0x1A73, 0x1A7C },
1022	{ 0x1A7F, 0x1A7F },
1023	{ 0x1AB0, 0x1ABE },
1024	{ 0x1B00, 0x1B03 },
1025	{ 0x1B34, 0x1B34 },
1026	{ 0x1B36, 0x1B3A },
1027	{ 0x1B3C, 0x1B3C },
1028	{ 0x1B42, 0x1B42 },
1029	{ 0x1B6B, 0x1B73 },
1030	{ 0x1B80, 0x1B81 },
1031	{ 0x1BA2, 0x1BA5 },
1032	{ 0x1BA8, 0x1BA9 },
1033	{ 0x1BAB, 0x1BAD },
1034	{ 0x1BE6, 0x1BE6 },
1035	{ 0x1BE8, 0x1BE9 },
1036	{ 0x1BED, 0x1BED },
1037	{ 0x1BEF, 0x1BF1 },
1038	{ 0x1C2C, 0x1C33 },
1039	{ 0x1C36, 0x1C37 },
1040	{ 0x1CD0, 0x1CD2 },
1041	{ 0x1CD4, 0x1CE0 },
1042	{ 0x1CE2, 0x1CE8 },
1043	{ 0x1CED, 0x1CED },
1044	{ 0x1CF4, 0x1CF4 },
1045	{ 0x1CF8, 0x1CF9 },
1046	{ 0x1DC0, 0x1DF5 },
1047	{ 0x1DFB, 0x1DFF },
1048	{ 0x200B, 0x200F },
1049	{ 0x202A, 0x202E },
1050	{ 0x2060, 0x2064 },
1051	{ 0x2066, 0x206F },
1052	{ 0x20D0, 0x20F0 },
1053	{ 0x2CEF, 0x2CF1 },
1054	{ 0x2D7F, 0x2D7F },
1055	{ 0x2DE0, 0x2DFF },
1056	{ 0x302A, 0x302D },
1057	{ 0x3099, 0x309A },
1058	{ 0xA66F, 0xA672 },
1059	{ 0xA674, 0xA67D },
1060	{ 0xA69E, 0xA69F },
1061	{ 0xA6F0, 0xA6F1 },
1062	{ 0xA802, 0xA802 },
1063	{ 0xA806, 0xA806 },
1064	{ 0xA80B, 0xA80B },
1065	{ 0xA825, 0xA826 },
1066	{ 0xA8C4, 0xA8C5 },
1067	{ 0xA8E0, 0xA8F1 },
1068	{ 0xA926, 0xA92D },
1069	{ 0xA947, 0xA951 },
1070	{ 0xA980, 0xA982 },
1071	{ 0xA9B3, 0xA9B3 },
1072	{ 0xA9B6, 0xA9B9 },
1073	{ 0xA9BC, 0xA9BC },
1074	{ 0xA9E5, 0xA9E5 },
1075	{ 0xAA29, 0xAA2E },
1076	{ 0xAA31, 0xAA32 },
1077	{ 0xAA35, 0xAA36 },
1078	{ 0xAA43, 0xAA43 },
1079	{ 0xAA4C, 0xAA4C },
1080	{ 0xAA7C, 0xAA7C },
1081	{ 0xAAB0, 0xAAB0 },
1082	{ 0xAAB2, 0xAAB4 },
1083	{ 0xAAB7, 0xAAB8 },
1084	{ 0xAABE, 0xAABF },
1085	{ 0xAAC1, 0xAAC1 },
1086	{ 0xAAEC, 0xAAED },
1087	{ 0xAAF6, 0xAAF6 },
1088	{ 0xABE5, 0xABE5 },
1089	{ 0xABE8, 0xABE8 },
1090	{ 0xABED, 0xABED },
1091	{ 0xFB1E, 0xFB1E },
1092	{ 0xFE00, 0xFE0F },
1093	{ 0xFE20, 0xFE2F },
1094	{ 0xFEFF, 0xFEFF },
1095	{ 0xFFF9, 0xFFFB }
1096};
1097
1098static const struct mb_ucsrange mb_ucs_fullwidth[] = {
1099	{ 0x1100, 0x115F },
1100	{ 0x231A, 0x231B },
1101	{ 0x2329, 0x232A },
1102	{ 0x23E9, 0x23EC },
1103	{ 0x23F0, 0x23F0 },
1104	{ 0x23F3, 0x23F3 },
1105	{ 0x25FD, 0x25FE },
1106	{ 0x2614, 0x2615 },
1107	{ 0x2648, 0x2653 },
1108	{ 0x267F, 0x267F },
1109	{ 0x2693, 0x2693 },
1110	{ 0x26A1, 0x26A1 },
1111	{ 0x26AA, 0x26AB },
1112	{ 0x26BD, 0x26BE },
1113	{ 0x26C4, 0x26C5 },
1114	{ 0x26CE, 0x26CE },
1115	{ 0x26D4, 0x26D4 },
1116	{ 0x26EA, 0x26EA },
1117	{ 0x26F2, 0x26F3 },
1118	{ 0x26F5, 0x26F5 },
1119	{ 0x26FA, 0x26FA },
1120	{ 0x26FD, 0x26FD },
1121	{ 0x2705, 0x2705 },
1122	{ 0x270A, 0x270B },
1123	{ 0x2728, 0x2728 },
1124	{ 0x274C, 0x274C },
1125	{ 0x274E, 0x274E },
1126	{ 0x2753, 0x2755 },
1127	{ 0x2757, 0x2757 },
1128	{ 0x2795, 0x2797 },
1129	{ 0x27B0, 0x27B0 },
1130	{ 0x27BF, 0x27BF },
1131	{ 0x2B1B, 0x2B1C },
1132	{ 0x2B50, 0x2B50 },
1133	{ 0x2B55, 0x2B55 },
1134	{ 0x2E80, 0x303E },
1135	{ 0x3040, 0xA4CF },
1136	{ 0xA960, 0xA97F },
1137	{ 0xAC00, 0xD7A3 },
1138	{ 0xF900, 0xFAFF },
1139	{ 0xFE10, 0xFE19 },
1140	{ 0xFE30, 0xFE6F },
1141	{ 0xFF00, 0xFF60 },
1142	{ 0xFFE0, 0xFFE6 }
1143};
1144
1145/* simple binary search in ranges, with bounds optimisation */
1146static int
1147mb_ucsbsearch(const struct mb_ucsrange arr[], size_t elems, unsigned int val)
1148{
1149	size_t min = 0, mid, max = elems;
1150
1151	if (val < arr[min].beg || val > arr[max - 1].end)
1152		return (0);
1153
1154	while (min < max) {
1155		mid = (min + max) / 2;
1156
1157		if (val < arr[mid].beg)
1158			max = mid;
1159		else if (val > arr[mid].end)
1160			min = mid + 1;
1161		else
1162			return (1);
1163	}
1164	return (0);
1165}
1166
1167/* Unix column width of a wide character (Unicode code point, really) */
1168int
1169utf_wcwidth(unsigned int wc)
1170{
1171	/* except NUL, C0/C1 control characters and DEL yield -1 */
1172	if (wc < 0x20 || (wc >= 0x7F && wc < 0xA0))
1173		return (wc ? -1 : 0);
1174
1175	/* combining characters use 0 screen columns */
1176	if (mb_ucsbsearch(mb_ucs_combining, NELEM(mb_ucs_combining), wc))
1177		return (0);
1178
1179	/* all others use 1 or 2 screen columns */
1180	if (mb_ucsbsearch(mb_ucs_fullwidth, NELEM(mb_ucs_fullwidth), wc))
1181		return (2);
1182	return (1);
1183}
1184#endif
1185