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, 2017, 2018, 2019
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 1.107 2020/03/27 02:49:40 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
49 enum token {
50 #define EXPRTOK_ENUM
51 #include "exprtok.h"
52 };
53
54 static const char opname[][4] = {
55 #define EXPRTOK_NAME
56 #include "exprtok.h"
57 };
58
59 static const uint8_t oplen[] = {
60 #define EXPRTOK_LEN
61 #include "exprtok.h"
62 };
63
64 static const uint8_t opprec[] = {
65 #define EXPRTOK_PREC
66 #include "exprtok.h"
67 };
68
69 typedef 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
88 enum error_type {
89 ET_UNEXPECTED, ET_BADLIT, ET_RECURSIVE,
90 ET_LVALUE, ET_RDONLY, ET_STR
91 };
92
93 static void evalerr(Expr_state *, enum error_type, const char *)
94 MKSH_A_NORETURN;
95 static struct tbl *evalexpr(Expr_state *, unsigned int);
96 static void exprtoken(Expr_state *);
97 static struct tbl *do_ppmm(Expr_state *, enum token, struct tbl *, bool);
98 static void assign_check(Expr_state *, enum token, struct tbl *);
99 static struct tbl *intvar(Expr_state *, struct tbl *);
100
101 /*
102 * parse and evaluate expression
103 */
104 int
evaluate(const char * expr,mksh_ari_t * rval,int error_ok,bool arith)105 evaluate(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 */
120 int
v_evaluate(struct tbl * vp,const char * expr,volatile int error_ok,bool arith)121 v_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
173 static void
evalerr(Expr_state * es,enum error_type type,const char * str)174 evalerr(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 Tbadnum, 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 */
233 static struct tbl *
do_ppmm(Expr_state * es,enum token op,struct tbl * vasn,bool is_prefix)234 do_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
260 static struct tbl *
evalexpr(Expr_state * es,unsigned int prec)261 evalexpr(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
552 static void
exprtoken(Expr_state * es)553 exprtoken(Expr_state *es)
554 {
555 const char *cp = es->tokp;
556 int c;
557 char *tvar;
558
559 /* skip whitespace */
560 skip_spaces:
561 --cp;
562 do {
563 c = ord(*++cp);
564 } while (ctype(c, C_SPACE));
565 if (es->tokp == es->expression && (unsigned int)c == ORD('#')) {
566 /* expression begins with # */
567 /* switch to unsigned */
568 es->natural = true;
569 ++cp;
570 goto skip_spaces;
571 }
572 es->tokp = cp;
573
574 if (c == '\0')
575 es->tok = END;
576 else if (ctype(c, C_ALPHX)) {
577 do {
578 c = ord(*++cp);
579 } while (ctype(c, C_ALNUX));
580 if ((unsigned int)c == ORD('[')) {
581 size_t len;
582
583 len = array_ref_len(cp);
584 if (len == 0)
585 evalerr(es, ET_STR, "missing ]");
586 cp += len;
587 }
588 if (es->noassign) {
589 es->val = tempvar("");
590 es->val->flag |= EXPRLVALUE;
591 } else {
592 strndupx(tvar, es->tokp, cp - es->tokp, ATEMP);
593 es->val = global(tvar);
594 afree(tvar, ATEMP);
595 }
596 es->tok = VAR;
597 } else if (c == '1' && cp[1] == '#') {
598 cp += 2;
599 if (*cp)
600 cp += utf_ptradj(cp);
601 strndupx(tvar, es->tokp, cp - es->tokp, ATEMP);
602 goto process_tvar;
603 #ifndef MKSH_SMALL
604 } else if (c == '\'') {
605 if (*++cp == '\0') {
606 es->tok = END;
607 evalerr(es, ET_UNEXPECTED, NULL);
608 }
609 cp += utf_ptradj(cp);
610 if (*cp++ != '\'')
611 evalerr(es, ET_STR,
612 "multi-character character constant");
613 /* 'x' -> 1#x (x = one multibyte character) */
614 c = cp - es->tokp;
615 tvar = alloc(c + /* NUL */ 1, ATEMP);
616 tvar[0] = '1';
617 tvar[1] = '#';
618 memcpy(tvar + 2, es->tokp + 1, c - 2);
619 tvar[c] = '\0';
620 goto process_tvar;
621 #endif
622 } else if (ctype(c, C_DIGIT)) {
623 while (ctype(c, C_ALNUM | C_HASH))
624 c = ord(*cp++);
625 strndupx(tvar, es->tokp, --cp - es->tokp, ATEMP);
626 process_tvar:
627 es->val = tempvar("");
628 es->val->flag &= ~INTEGER;
629 es->val->type = 0;
630 es->val->val.s = tvar;
631 if (setint_v(es->val, es->val, es->arith) == NULL)
632 evalerr(es, ET_BADLIT, tvar);
633 afree(tvar, ATEMP);
634 es->tok = LIT;
635 } else {
636 int i, n0;
637
638 for (i = 0; (n0 = ord(opname[i][0])); i++)
639 if (c == n0 && strncmp(cp, opname[i],
640 (size_t)oplen[i]) == 0) {
641 es->tok = (enum token)i;
642 cp += oplen[i];
643 break;
644 }
645 if (!n0)
646 es->tok = BAD;
647 }
648 es->tokp = cp;
649 }
650
651 static void
assign_check(Expr_state * es,enum token op,struct tbl * vasn)652 assign_check(Expr_state *es, enum token op, struct tbl *vasn)
653 {
654 if (es->tok == END || !vasn ||
655 (vasn->name[0] == '\0' && !(vasn->flag & EXPRLVALUE)))
656 evalerr(es, ET_LVALUE, opname[(int)op]);
657 else if (vasn->flag & RDONLY)
658 evalerr(es, ET_RDONLY, opname[(int)op]);
659 }
660
661 struct tbl *
tempvar(const char * vname)662 tempvar(const char *vname)
663 {
664 struct tbl *vp;
665 size_t vsize;
666
667 vsize = strlen(vname) + 1;
668 vp = alloc(offsetof(struct tbl, name[0]) + vsize, ATEMP);
669 memcpy(vp->name, vname, vsize);
670 vp->flag = ISSET|INTEGER;
671 vp->type = 0;
672 vp->areap = ATEMP;
673 vp->ua.hval = 0;
674 vp->val.i = 0;
675 return (vp);
676 }
677
678 /* cast (string) variable to temporary integer variable */
679 static struct tbl *
intvar(Expr_state * es,struct tbl * vp)680 intvar(Expr_state *es, struct tbl *vp)
681 {
682 struct tbl *vq;
683
684 /* try to avoid replacing a temp var with another temp var */
685 if (vp->name[0] == '\0' &&
686 (vp->flag & (ISSET|INTEGER|EXPRLVALUE)) == (ISSET|INTEGER))
687 return (vp);
688
689 vq = tempvar("");
690 if (setint_v(vq, vp, es->arith) == NULL) {
691 if (vp->flag & EXPRINEVAL)
692 evalerr(es, ET_RECURSIVE, vp->name);
693 es->evaling = vp;
694 vp->flag |= EXPRINEVAL;
695 v_evaluate(vq, str_val(vp), KSH_UNWIND_ERROR, es->arith);
696 vp->flag &= ~EXPRINEVAL;
697 es->evaling = NULL;
698 }
699 return (vq);
700 }
701
702
703 /*
704 * UTF-8 support code: high-level functions
705 */
706
707 int
utf_widthadj(const char * src,const char ** dst)708 utf_widthadj(const char *src, const char **dst)
709 {
710 size_t len;
711 unsigned int wc;
712 int width;
713
714 if (!UTFMODE || (len = utf_mbtowc(&wc, src)) == (size_t)-1 ||
715 wc == 0)
716 len = width = 1;
717 else if ((width = utf_wcwidth(wc)) < 0)
718 /* XXX use 2 for x_zotc3 here? */
719 width = 1;
720
721 if (dst)
722 *dst = src + len;
723 return (width);
724 }
725
726 size_t
utf_mbswidth(const char * s)727 utf_mbswidth(const char *s)
728 {
729 size_t len, width = 0;
730 unsigned int wc;
731 int cw;
732
733 if (!UTFMODE)
734 return (strlen(s));
735
736 while (*s)
737 if (((len = utf_mbtowc(&wc, s)) == (size_t)-1) ||
738 ((cw = utf_wcwidth(wc)) == -1)) {
739 s++;
740 width += 1;
741 } else {
742 s += len;
743 width += cw;
744 }
745 return (width);
746 }
747
748 const char *
utf_skipcols(const char * p,int cols,int * colp)749 utf_skipcols(const char *p, int cols, int *colp)
750 {
751 int c = 0;
752 const char *q;
753
754 while (c < cols) {
755 if (!*p) {
756 /* end of input; special handling for edit.c */
757 if (!colp)
758 return (p + cols - c);
759 *colp = c;
760 return (p);
761 }
762 c += utf_widthadj(p, &p);
763 }
764 if (UTFMODE)
765 while (utf_widthadj(p, &q) == 0)
766 p = q;
767 if (colp)
768 *colp = c;
769 return (p);
770 }
771
772 size_t
utf_ptradj(const char * src)773 utf_ptradj(const char *src)
774 {
775 register size_t n;
776
777 if (!UTFMODE || rtt2asc(*src) < 0xC2 ||
778 (n = utf_mbtowc(NULL, src)) == (size_t)-1)
779 n = 1;
780 return (n);
781 }
782
783 /*
784 * UTF-8 support code: low-level functions
785 */
786
787 /* CESU-8 multibyte and wide character conversion crafted for mksh */
788
789 size_t
utf_mbtowc(unsigned int * dst,const char * src)790 utf_mbtowc(unsigned int *dst, const char *src)
791 {
792 const unsigned char *s = (const unsigned char *)src;
793 unsigned int c, wc;
794
795 if ((wc = ord(rtt2asc(*s++))) < 0x80) {
796 out:
797 if (dst != NULL)
798 *dst = wc;
799 return (wc ? ((const char *)s - src) : 0);
800 }
801 if (wc < 0xC2 || wc >= 0xF0)
802 /* < 0xC0: spurious second byte */
803 /* < 0xC2: non-minimalistic mapping error in 2-byte seqs */
804 /* > 0xEF: beyond BMP */
805 goto ilseq;
806
807 if (wc < 0xE0) {
808 wc = (wc & 0x1F) << 6;
809 if (((c = ord(rtt2asc(*s++))) & 0xC0) != 0x80)
810 goto ilseq;
811 wc |= c & 0x3F;
812 goto out;
813 }
814
815 wc = (wc & 0x0F) << 12;
816
817 if (((c = ord(rtt2asc(*s++))) & 0xC0) != 0x80)
818 goto ilseq;
819 wc |= (c & 0x3F) << 6;
820
821 if (((c = ord(rtt2asc(*s++))) & 0xC0) != 0x80)
822 goto ilseq;
823 wc |= c & 0x3F;
824
825 /* Check for non-minimalistic mapping error in 3-byte seqs */
826 if (wc >= 0x0800 && wc <= 0xFFFD)
827 goto out;
828 ilseq:
829 return ((size_t)(-1));
830 }
831
832 size_t
utf_wctomb(char * dst,unsigned int wc)833 utf_wctomb(char *dst, unsigned int wc)
834 {
835 unsigned char *d;
836
837 if (wc < 0x80) {
838 *dst = asc2rtt(wc);
839 return (1);
840 }
841
842 d = (unsigned char *)dst;
843 if (wc < 0x0800)
844 *d++ = asc2rtt((wc >> 6) | 0xC0);
845 else {
846 *d++ = asc2rtt(((wc = wc > 0xFFFD ? 0xFFFD : wc) >> 12) | 0xE0);
847 *d++ = asc2rtt(((wc >> 6) & 0x3F) | 0x80);
848 }
849 *d++ = asc2rtt((wc & 0x3F) | 0x80);
850 return ((char *)d - dst);
851 }
852
853 /*
854 * Wrapper around access(2) because it says root can execute everything
855 * on some operating systems. Does not set errno, no user needs it. Use
856 * this iff mode can have the X_OK bit set, access otherwise.
857 */
858 int
ksh_access(const char * fn,int mode)859 ksh_access(const char *fn, int mode)
860 {
861 #ifdef __OS2__
862 return (access_ex(access, fn, mode));
863 #else
864 int rv;
865 struct stat sb;
866
867 if ((rv = access(fn, mode)) == 0 && (mode & X_OK) &&
868 (kshuid == 0 || ksheuid == 0) &&
869 (rv = stat(fn, &sb)) == 0 && !S_ISDIR(sb.st_mode) &&
870 (sb.st_mode & (S_IXUSR|S_IXGRP|S_IXOTH)) == 0)
871 rv = -1;
872
873 return (rv);
874 #endif
875 }
876
877 #ifndef MIRBSD_BOOTFLOPPY
878 /* From: X11/xc/programs/xterm/wcwidth.c,v 1.10 */
879
880 struct mb_ucsrange {
881 unsigned short beg;
882 unsigned short end;
883 };
884
885 static int mb_ucsbsearch(const struct mb_ucsrange arr[], size_t elems,
886 unsigned int val) MKSH_A_PURE;
887
888 /*
889 * Generated from the UCD 13.0.0 by
890 * MirOS: contrib/code/Snippets/eawparse,v 1.14 2020/03/27 01:33:21 tg Exp $
891 */
892
893 /*-
894 * Parts Copyright © 1991–2020 Unicode, Inc. All rights reserved.
895 * Distributed under the Terms of Use in
896 * https://www.unicode.org/copyright.html.
897 *
898 * Permission is hereby granted, free of charge, to any person obtaining
899 * a copy of the Unicode data files and any associated documentation
900 * (the "Data Files") or Unicode software and any associated documentation
901 * (the "Software") to deal in the Data Files or Software
902 * without restriction, including without limitation the rights to use,
903 * copy, modify, merge, publish, distribute, and/or sell copies of
904 * the Data Files or Software, and to permit persons to whom the Data Files
905 * or Software are furnished to do so, provided that either
906 * (a) this copyright and permission notice appear with all copies
907 * of the Data Files or Software, or
908 * (b) this copyright and permission notice appear in associated
909 * Documentation.
910 *
911 * THE DATA FILES AND SOFTWARE ARE PROVIDED "AS IS", WITHOUT WARRANTY OF
912 * ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE
913 * WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
914 * NONINFRINGEMENT OF THIRD PARTY RIGHTS.
915 * IN NO EVENT SHALL THE COPYRIGHT HOLDER OR HOLDERS INCLUDED IN THIS
916 * NOTICE BE LIABLE FOR ANY CLAIM, OR ANY SPECIAL INDIRECT OR CONSEQUENTIAL
917 * DAMAGES, OR ANY DAMAGES WHATSOEVER RESULTING FROM LOSS OF USE,
918 * DATA OR PROFITS, WHETHER IN AN ACTION OF CONTRACT, NEGLIGENCE OR OTHER
919 * TORTIOUS ACTION, ARISING OUT OF OR IN CONNECTION WITH THE USE OR
920 * PERFORMANCE OF THE DATA FILES OR SOFTWARE.
921 *
922 * Except as contained in this notice, the name of a copyright holder
923 * shall not be used in advertising or otherwise to promote the sale,
924 * use or other dealings in these Data Files or Software without prior
925 * written authorization of the copyright holder.
926 */
927
928 static const struct mb_ucsrange mb_ucs_combining[] = {
929 { 0x0300, 0x036F },
930 { 0x0483, 0x0489 },
931 { 0x0591, 0x05BD },
932 { 0x05BF, 0x05BF },
933 { 0x05C1, 0x05C2 },
934 { 0x05C4, 0x05C5 },
935 { 0x05C7, 0x05C7 },
936 { 0x0610, 0x061A },
937 { 0x061C, 0x061C },
938 { 0x064B, 0x065F },
939 { 0x0670, 0x0670 },
940 { 0x06D6, 0x06DC },
941 { 0x06DF, 0x06E4 },
942 { 0x06E7, 0x06E8 },
943 { 0x06EA, 0x06ED },
944 { 0x0711, 0x0711 },
945 { 0x0730, 0x074A },
946 { 0x07A6, 0x07B0 },
947 { 0x07EB, 0x07F3 },
948 { 0x07FD, 0x07FD },
949 { 0x0816, 0x0819 },
950 { 0x081B, 0x0823 },
951 { 0x0825, 0x0827 },
952 { 0x0829, 0x082D },
953 { 0x0859, 0x085B },
954 { 0x08D3, 0x08E1 },
955 { 0x08E3, 0x0902 },
956 { 0x093A, 0x093A },
957 { 0x093C, 0x093C },
958 { 0x0941, 0x0948 },
959 { 0x094D, 0x094D },
960 { 0x0951, 0x0957 },
961 { 0x0962, 0x0963 },
962 { 0x0981, 0x0981 },
963 { 0x09BC, 0x09BC },
964 { 0x09C1, 0x09C4 },
965 { 0x09CD, 0x09CD },
966 { 0x09E2, 0x09E3 },
967 { 0x09FE, 0x09FE },
968 { 0x0A01, 0x0A02 },
969 { 0x0A3C, 0x0A3C },
970 { 0x0A41, 0x0A42 },
971 { 0x0A47, 0x0A48 },
972 { 0x0A4B, 0x0A4D },
973 { 0x0A51, 0x0A51 },
974 { 0x0A70, 0x0A71 },
975 { 0x0A75, 0x0A75 },
976 { 0x0A81, 0x0A82 },
977 { 0x0ABC, 0x0ABC },
978 { 0x0AC1, 0x0AC5 },
979 { 0x0AC7, 0x0AC8 },
980 { 0x0ACD, 0x0ACD },
981 { 0x0AE2, 0x0AE3 },
982 { 0x0AFA, 0x0AFF },
983 { 0x0B01, 0x0B01 },
984 { 0x0B3C, 0x0B3C },
985 { 0x0B3F, 0x0B3F },
986 { 0x0B41, 0x0B44 },
987 { 0x0B4D, 0x0B4D },
988 { 0x0B55, 0x0B56 },
989 { 0x0B62, 0x0B63 },
990 { 0x0B82, 0x0B82 },
991 { 0x0BC0, 0x0BC0 },
992 { 0x0BCD, 0x0BCD },
993 { 0x0C00, 0x0C00 },
994 { 0x0C04, 0x0C04 },
995 { 0x0C3E, 0x0C40 },
996 { 0x0C46, 0x0C48 },
997 { 0x0C4A, 0x0C4D },
998 { 0x0C55, 0x0C56 },
999 { 0x0C62, 0x0C63 },
1000 { 0x0C81, 0x0C81 },
1001 { 0x0CBC, 0x0CBC },
1002 { 0x0CBF, 0x0CBF },
1003 { 0x0CC6, 0x0CC6 },
1004 { 0x0CCC, 0x0CCD },
1005 { 0x0CE2, 0x0CE3 },
1006 { 0x0D00, 0x0D01 },
1007 { 0x0D3B, 0x0D3C },
1008 { 0x0D41, 0x0D44 },
1009 { 0x0D4D, 0x0D4D },
1010 { 0x0D62, 0x0D63 },
1011 { 0x0D81, 0x0D81 },
1012 { 0x0DCA, 0x0DCA },
1013 { 0x0DD2, 0x0DD4 },
1014 { 0x0DD6, 0x0DD6 },
1015 { 0x0E31, 0x0E31 },
1016 { 0x0E34, 0x0E3A },
1017 { 0x0E47, 0x0E4E },
1018 { 0x0EB1, 0x0EB1 },
1019 { 0x0EB4, 0x0EBC },
1020 { 0x0EC8, 0x0ECD },
1021 { 0x0F18, 0x0F19 },
1022 { 0x0F35, 0x0F35 },
1023 { 0x0F37, 0x0F37 },
1024 { 0x0F39, 0x0F39 },
1025 { 0x0F71, 0x0F7E },
1026 { 0x0F80, 0x0F84 },
1027 { 0x0F86, 0x0F87 },
1028 { 0x0F8D, 0x0F97 },
1029 { 0x0F99, 0x0FBC },
1030 { 0x0FC6, 0x0FC6 },
1031 { 0x102D, 0x1030 },
1032 { 0x1032, 0x1037 },
1033 { 0x1039, 0x103A },
1034 { 0x103D, 0x103E },
1035 { 0x1058, 0x1059 },
1036 { 0x105E, 0x1060 },
1037 { 0x1071, 0x1074 },
1038 { 0x1082, 0x1082 },
1039 { 0x1085, 0x1086 },
1040 { 0x108D, 0x108D },
1041 { 0x109D, 0x109D },
1042 { 0x1160, 0x11FF },
1043 { 0x135D, 0x135F },
1044 { 0x1712, 0x1714 },
1045 { 0x1732, 0x1734 },
1046 { 0x1752, 0x1753 },
1047 { 0x1772, 0x1773 },
1048 { 0x17B4, 0x17B5 },
1049 { 0x17B7, 0x17BD },
1050 { 0x17C6, 0x17C6 },
1051 { 0x17C9, 0x17D3 },
1052 { 0x17DD, 0x17DD },
1053 { 0x180B, 0x180E },
1054 { 0x1885, 0x1886 },
1055 { 0x18A9, 0x18A9 },
1056 { 0x1920, 0x1922 },
1057 { 0x1927, 0x1928 },
1058 { 0x1932, 0x1932 },
1059 { 0x1939, 0x193B },
1060 { 0x1A17, 0x1A18 },
1061 { 0x1A1B, 0x1A1B },
1062 { 0x1A56, 0x1A56 },
1063 { 0x1A58, 0x1A5E },
1064 { 0x1A60, 0x1A60 },
1065 { 0x1A62, 0x1A62 },
1066 { 0x1A65, 0x1A6C },
1067 { 0x1A73, 0x1A7C },
1068 { 0x1A7F, 0x1A7F },
1069 { 0x1AB0, 0x1AC0 },
1070 { 0x1B00, 0x1B03 },
1071 { 0x1B34, 0x1B34 },
1072 { 0x1B36, 0x1B3A },
1073 { 0x1B3C, 0x1B3C },
1074 { 0x1B42, 0x1B42 },
1075 { 0x1B6B, 0x1B73 },
1076 { 0x1B80, 0x1B81 },
1077 { 0x1BA2, 0x1BA5 },
1078 { 0x1BA8, 0x1BA9 },
1079 { 0x1BAB, 0x1BAD },
1080 { 0x1BE6, 0x1BE6 },
1081 { 0x1BE8, 0x1BE9 },
1082 { 0x1BED, 0x1BED },
1083 { 0x1BEF, 0x1BF1 },
1084 { 0x1C2C, 0x1C33 },
1085 { 0x1C36, 0x1C37 },
1086 { 0x1CD0, 0x1CD2 },
1087 { 0x1CD4, 0x1CE0 },
1088 { 0x1CE2, 0x1CE8 },
1089 { 0x1CED, 0x1CED },
1090 { 0x1CF4, 0x1CF4 },
1091 { 0x1CF8, 0x1CF9 },
1092 { 0x1DC0, 0x1DF9 },
1093 { 0x1DFB, 0x1DFF },
1094 { 0x200B, 0x200F },
1095 { 0x202A, 0x202E },
1096 { 0x2060, 0x2064 },
1097 { 0x2066, 0x206F },
1098 { 0x20D0, 0x20F0 },
1099 { 0x2CEF, 0x2CF1 },
1100 { 0x2D7F, 0x2D7F },
1101 { 0x2DE0, 0x2DFF },
1102 { 0x302A, 0x302D },
1103 { 0x3099, 0x309A },
1104 { 0xA66F, 0xA672 },
1105 { 0xA674, 0xA67D },
1106 { 0xA69E, 0xA69F },
1107 { 0xA6F0, 0xA6F1 },
1108 { 0xA802, 0xA802 },
1109 { 0xA806, 0xA806 },
1110 { 0xA80B, 0xA80B },
1111 { 0xA825, 0xA826 },
1112 { 0xA82C, 0xA82C },
1113 { 0xA8C4, 0xA8C5 },
1114 { 0xA8E0, 0xA8F1 },
1115 { 0xA8FF, 0xA8FF },
1116 { 0xA926, 0xA92D },
1117 { 0xA947, 0xA951 },
1118 { 0xA980, 0xA982 },
1119 { 0xA9B3, 0xA9B3 },
1120 { 0xA9B6, 0xA9B9 },
1121 { 0xA9BC, 0xA9BD },
1122 { 0xA9E5, 0xA9E5 },
1123 { 0xAA29, 0xAA2E },
1124 { 0xAA31, 0xAA32 },
1125 { 0xAA35, 0xAA36 },
1126 { 0xAA43, 0xAA43 },
1127 { 0xAA4C, 0xAA4C },
1128 { 0xAA7C, 0xAA7C },
1129 { 0xAAB0, 0xAAB0 },
1130 { 0xAAB2, 0xAAB4 },
1131 { 0xAAB7, 0xAAB8 },
1132 { 0xAABE, 0xAABF },
1133 { 0xAAC1, 0xAAC1 },
1134 { 0xAAEC, 0xAAED },
1135 { 0xAAF6, 0xAAF6 },
1136 { 0xABE5, 0xABE5 },
1137 { 0xABE8, 0xABE8 },
1138 { 0xABED, 0xABED },
1139 { 0xFB1E, 0xFB1E },
1140 { 0xFE00, 0xFE0F },
1141 { 0xFE20, 0xFE2F },
1142 { 0xFEFF, 0xFEFF },
1143 { 0xFFF9, 0xFFFB }
1144 };
1145
1146 static const struct mb_ucsrange mb_ucs_fullwidth[] = {
1147 { 0x1100, 0x115F },
1148 { 0x231A, 0x231B },
1149 { 0x2329, 0x232A },
1150 { 0x23E9, 0x23EC },
1151 { 0x23F0, 0x23F0 },
1152 { 0x23F3, 0x23F3 },
1153 { 0x25FD, 0x25FE },
1154 { 0x2614, 0x2615 },
1155 { 0x2648, 0x2653 },
1156 { 0x267F, 0x267F },
1157 { 0x2693, 0x2693 },
1158 { 0x26A1, 0x26A1 },
1159 { 0x26AA, 0x26AB },
1160 { 0x26BD, 0x26BE },
1161 { 0x26C4, 0x26C5 },
1162 { 0x26CE, 0x26CE },
1163 { 0x26D4, 0x26D4 },
1164 { 0x26EA, 0x26EA },
1165 { 0x26F2, 0x26F3 },
1166 { 0x26F5, 0x26F5 },
1167 { 0x26FA, 0x26FA },
1168 { 0x26FD, 0x26FD },
1169 { 0x2705, 0x2705 },
1170 { 0x270A, 0x270B },
1171 { 0x2728, 0x2728 },
1172 { 0x274C, 0x274C },
1173 { 0x274E, 0x274E },
1174 { 0x2753, 0x2755 },
1175 { 0x2757, 0x2757 },
1176 { 0x2795, 0x2797 },
1177 { 0x27B0, 0x27B0 },
1178 { 0x27BF, 0x27BF },
1179 { 0x2B1B, 0x2B1C },
1180 { 0x2B50, 0x2B50 },
1181 { 0x2B55, 0x2B55 },
1182 { 0x2E80, 0x3029 },
1183 { 0x302E, 0x303E },
1184 { 0x3040, 0x3098 },
1185 { 0x309B, 0xA4CF },
1186 { 0xA960, 0xA97F },
1187 { 0xAC00, 0xD7A3 },
1188 { 0xF900, 0xFAFF },
1189 { 0xFE10, 0xFE19 },
1190 { 0xFE30, 0xFE6F },
1191 { 0xFF01, 0xFF60 },
1192 { 0xFFE0, 0xFFE6 }
1193 };
1194
1195 /* simple binary search in ranges, with bounds optimisation */
1196 static int
mb_ucsbsearch(const struct mb_ucsrange arr[],size_t elems,unsigned int val)1197 mb_ucsbsearch(const struct mb_ucsrange arr[], size_t elems, unsigned int val)
1198 {
1199 size_t min = 0, mid, max = elems;
1200
1201 if (val < arr[min].beg || val > arr[max - 1].end)
1202 return (0);
1203
1204 while (min < max) {
1205 mid = (min + max) / 2;
1206
1207 if (val < arr[mid].beg)
1208 max = mid;
1209 else if (val > arr[mid].end)
1210 min = mid + 1;
1211 else
1212 return (1);
1213 }
1214 return (0);
1215 }
1216
1217 /* Unix column width of a wide character (UCS code point, really) */
1218 int
utf_wcwidth(unsigned int wc)1219 utf_wcwidth(unsigned int wc)
1220 {
1221 /* except NUL, C0/C1 control characters and DEL yield -1 */
1222 if (wc < 0x20 || (wc >= 0x7F && wc < 0xA0))
1223 return (wc ? -1 : 0);
1224
1225 /* combining characters use 0 screen columns */
1226 if (mb_ucsbsearch(mb_ucs_combining, NELEM(mb_ucs_combining), wc))
1227 return (0);
1228
1229 /* all others use 1 or 2 screen columns */
1230 if (mb_ucsbsearch(mb_ucs_fullwidth, NELEM(mb_ucs_fullwidth), wc))
1231 return (2);
1232 return (1);
1233 }
1234 #endif
1235