expr.c
1/* $OpenBSD: expr.c,v 1.34 2019/02/20 23:59:17 schwarze Exp $ */
2
3/*
4 * Korn expression evaluation
5 */
6/*
7 * todo: better error handling: if in builtin, should be builtin error, etc.
8 */
9
10#include <ctype.h>
11#include <limits.h>
12#include <string.h>
13
14#include "sh.h"
15
16/* The order of these enums is constrained by the order of opinfo[] */
17enum token {
18 /* some (long) unary operators */
19 O_PLUSPLUS = 0, O_MINUSMINUS,
20 /* binary operators */
21 O_EQ, O_NE,
22 /* assignments are assumed to be in range O_ASN .. O_BORASN */
23 O_ASN, O_TIMESASN, O_DIVASN, O_MODASN, O_PLUSASN, O_MINUSASN,
24 O_LSHIFTASN, O_RSHIFTASN, O_BANDASN, O_BXORASN, O_BORASN,
25 O_LSHIFT, O_RSHIFT,
26 O_LE, O_GE, O_LT, O_GT,
27 O_LAND,
28 O_LOR,
29 O_TIMES, O_DIV, O_MOD,
30 O_PLUS, O_MINUS,
31 O_BAND,
32 O_BXOR,
33 O_BOR,
34 O_TERN,
35 O_COMMA,
36 /* things after this aren't used as binary operators */
37 /* unary that are not also binaries */
38 O_BNOT, O_LNOT,
39 /* misc */
40 OPEN_PAREN, CLOSE_PAREN, CTERN,
41 /* things that don't appear in the opinfo[] table */
42 VAR, LIT, END, BAD
43};
44#define IS_BINOP(op) (((int)op) >= (int)O_EQ && ((int)op) <= (int)O_COMMA)
45#define IS_ASSIGNOP(op) ((int)(op) >= (int)O_ASN && (int)(op) <= (int)O_BORASN)
46
47enum prec {
48 P_PRIMARY = 0, /* VAR, LIT, (), ~ ! - + */
49 P_MULT, /* * / % */
50 P_ADD, /* + - */
51 P_SHIFT, /* << >> */
52 P_RELATION, /* < <= > >= */
53 P_EQUALITY, /* == != */
54 P_BAND, /* & */
55 P_BXOR, /* ^ */
56 P_BOR, /* | */
57 P_LAND, /* && */
58 P_LOR, /* || */
59 P_TERN, /* ?: */
60 P_ASSIGN, /* = *= /= %= += -= <<= >>= &= ^= |= */
61 P_COMMA /* , */
62};
63#define MAX_PREC P_COMMA
64
65struct opinfo {
66 char name[4];
67 int len; /* name length */
68 enum prec prec; /* precedence: lower is higher */
69};
70
71/* Tokens in this table must be ordered so the longest are first
72 * (eg, += before +). If you change something, change the order
73 * of enum token too.
74 */
75static const struct opinfo opinfo[] = {
76 { "++", 2, P_PRIMARY }, /* before + */
77 { "--", 2, P_PRIMARY }, /* before - */
78 { "==", 2, P_EQUALITY }, /* before = */
79 { "!=", 2, P_EQUALITY }, /* before ! */
80 { "=", 1, P_ASSIGN }, /* keep assigns in a block */
81 { "*=", 2, P_ASSIGN },
82 { "/=", 2, P_ASSIGN },
83 { "%=", 2, P_ASSIGN },
84 { "+=", 2, P_ASSIGN },
85 { "-=", 2, P_ASSIGN },
86 { "<<=", 3, P_ASSIGN },
87 { ">>=", 3, P_ASSIGN },
88 { "&=", 2, P_ASSIGN },
89 { "^=", 2, P_ASSIGN },
90 { "|=", 2, P_ASSIGN },
91 { "<<", 2, P_SHIFT },
92 { ">>", 2, P_SHIFT },
93 { "<=", 2, P_RELATION },
94 { ">=", 2, P_RELATION },
95 { "<", 1, P_RELATION },
96 { ">", 1, P_RELATION },
97 { "&&", 2, P_LAND },
98 { "||", 2, P_LOR },
99 { "*", 1, P_MULT },
100 { "/", 1, P_MULT },
101 { "%", 1, P_MULT },
102 { "+", 1, P_ADD },
103 { "-", 1, P_ADD },
104 { "&", 1, P_BAND },
105 { "^", 1, P_BXOR },
106 { "|", 1, P_BOR },
107 { "?", 1, P_TERN },
108 { ",", 1, P_COMMA },
109 { "~", 1, P_PRIMARY },
110 { "!", 1, P_PRIMARY },
111 { "(", 1, P_PRIMARY },
112 { ")", 1, P_PRIMARY },
113 { ":", 1, P_PRIMARY },
114 { "", 0, P_PRIMARY } /* end of table */
115};
116
117
118typedef struct expr_state Expr_state;
119struct expr_state {
120 const char *expression; /* expression being evaluated */
121 const char *tokp; /* lexical position */
122 enum token tok; /* token from token() */
123 int noassign; /* don't do assigns (for ?:,&&,||) */
124 bool arith; /* true if evaluating an $(())
125 * expression
126 */
127 struct tbl *val; /* value from token() */
128 struct tbl *evaling; /* variable that is being recursively
129 * expanded (EXPRINEVAL flag set)
130 */
131};
132
133enum error_type {
134 ET_UNEXPECTED, ET_BADLIT, ET_RECURSIVE,
135 ET_LVALUE, ET_RDONLY, ET_STR
136};
137
138static void evalerr(Expr_state *, enum error_type, const char *)
139 __attribute__((__noreturn__));
140static struct tbl *evalexpr(Expr_state *, enum prec);
141static void token(Expr_state *);
142static struct tbl *do_ppmm(Expr_state *, enum token, struct tbl *, bool);
143static void assign_check(Expr_state *, enum token, struct tbl *);
144static struct tbl *tempvar(void);
145static struct tbl *intvar(Expr_state *, struct tbl *);
146
147/*
148 * parse and evaluate expression
149 */
150int
151evaluate(const char *expr, int64_t *rval, int error_ok, bool arith)
152{
153 struct tbl v;
154 int ret;
155
156 v.flag = DEFINED|INTEGER;
157 v.type = 0;
158 ret = v_evaluate(&v, expr, error_ok, arith);
159 *rval = v.val.i;
160 return ret;
161}
162
163/*
164 * parse and evaluate expression, storing result in vp.
165 */
166int
167v_evaluate(struct tbl *vp, const char *expr, volatile int error_ok,
168 bool arith)
169{
170 struct tbl *v;
171 Expr_state curstate;
172 Expr_state * const es = &curstate;
173 int save_disable_subst;
174 int i;
175
176 /* save state to allow recursive calls */
177 curstate.expression = curstate.tokp = expr;
178 curstate.noassign = 0;
179 curstate.arith = arith;
180 curstate.evaling = NULL;
181 curstate.val = NULL;
182
183 newenv(E_ERRH);
184 save_disable_subst = disable_subst;
185 i = sigsetjmp(genv->jbuf, 0);
186 if (i) {
187 disable_subst = save_disable_subst;
188 /* Clear EXPRINEVAL in of any variables we were playing with */
189 if (curstate.evaling)
190 curstate.evaling->flag &= ~EXPRINEVAL;
191 quitenv(NULL);
192 if (i == LAEXPR) {
193 if (error_ok == KSH_RETURN_ERROR)
194 return 0;
195 errorf(NULL);
196 }
197 unwind(i);
198 /* NOTREACHED */
199 }
200
201 token(es);
202#if 1 /* ifdef-out to disallow empty expressions to be treated as 0 */
203 if (es->tok == END) {
204 es->tok = LIT;
205 es->val = tempvar();
206 }
207#endif /* 0 */
208 v = intvar(es, evalexpr(es, MAX_PREC));
209
210 if (es->tok != END)
211 evalerr(es, ET_UNEXPECTED, NULL);
212
213 if (vp->flag & INTEGER)
214 setint_v(vp, v, es->arith);
215 else
216 /* can fail if readonly */
217 setstr(vp, str_val(v), error_ok);
218
219 quitenv(NULL);
220
221 return 1;
222}
223
224static void
225evalerr(Expr_state *es, enum error_type type, const char *str)
226{
227 char tbuf[2];
228 const char *s;
229
230 es->arith = false;
231 switch (type) {
232 case ET_UNEXPECTED:
233 switch (es->tok) {
234 case VAR:
235 s = es->val->name;
236 break;
237 case LIT:
238 s = str_val(es->val);
239 break;
240 case END:
241 s = "end of expression";
242 break;
243 case BAD:
244 tbuf[0] = *es->tokp;
245 tbuf[1] = '\0';
246 s = tbuf;
247 break;
248 default:
249 s = opinfo[(int)es->tok].name;
250 }
251 warningf(true, "%s: unexpected `%s'", es->expression, s);
252 break;
253
254 case ET_BADLIT:
255 warningf(true, "%s: bad number `%s'", es->expression, str);
256 break;
257
258 case ET_RECURSIVE:
259 warningf(true, "%s: expression recurses on parameter `%s'",
260 es->expression, str);
261 break;
262
263 case ET_LVALUE:
264 warningf(true, "%s: %s requires lvalue",
265 es->expression, str);
266 break;
267
268 case ET_RDONLY:
269 warningf(true, "%s: %s applied to read only variable",
270 es->expression, str);
271 break;
272
273 default: /* keep gcc happy */
274 case ET_STR:
275 warningf(true, "%s: %s", es->expression, str);
276 break;
277 }
278 unwind(LAEXPR);
279}
280
281static struct tbl *
282evalexpr(Expr_state *es, enum prec prec)
283{
284 struct tbl *vl, *vr = NULL, *vasn;
285 enum token op;
286 int64_t res = 0;
287
288 if (prec == P_PRIMARY) {
289 op = es->tok;
290 if (op == O_BNOT || op == O_LNOT || op == O_MINUS ||
291 op == O_PLUS) {
292 token(es);
293 vl = intvar(es, evalexpr(es, P_PRIMARY));
294 if (op == O_BNOT)
295 vl->val.i = ~vl->val.i;
296 else if (op == O_LNOT)
297 vl->val.i = !vl->val.i;
298 else if (op == O_MINUS)
299 vl->val.i = -vl->val.i;
300 /* op == O_PLUS is a no-op */
301 } else if (op == OPEN_PAREN) {
302 token(es);
303 vl = evalexpr(es, MAX_PREC);
304 if (es->tok != CLOSE_PAREN)
305 evalerr(es, ET_STR, "missing )");
306 token(es);
307 } else if (op == O_PLUSPLUS || op == O_MINUSMINUS) {
308 token(es);
309 vl = do_ppmm(es, op, es->val, true);
310 token(es);
311 } else if (op == VAR || op == LIT) {
312 vl = es->val;
313 token(es);
314 } else {
315 evalerr(es, ET_UNEXPECTED, NULL);
316 /* NOTREACHED */
317 }
318 if (es->tok == O_PLUSPLUS || es->tok == O_MINUSMINUS) {
319 vl = do_ppmm(es, es->tok, vl, false);
320 token(es);
321 }
322 return vl;
323 }
324 vl = evalexpr(es, ((int) prec) - 1);
325 for (op = es->tok; IS_BINOP(op) && opinfo[(int) op].prec == prec;
326 op = es->tok) {
327 token(es);
328 vasn = vl;
329 if (op != O_ASN) /* vl may not have a value yet */
330 vl = intvar(es, vl);
331 if (IS_ASSIGNOP(op)) {
332 assign_check(es, op, vasn);
333 vr = intvar(es, evalexpr(es, P_ASSIGN));
334 } else if (op != O_TERN && op != O_LAND && op != O_LOR)
335 vr = intvar(es, evalexpr(es, ((int) prec) - 1));
336 if ((op == O_DIV || op == O_MOD || op == O_DIVASN ||
337 op == O_MODASN) && vr->val.i == 0) {
338 if (es->noassign)
339 vr->val.i = 1;
340 else
341 evalerr(es, ET_STR, "zero divisor");
342 }
343 switch ((int) op) {
344 case O_TIMES:
345 case O_TIMESASN:
346 res = vl->val.i * vr->val.i;
347 break;
348 case O_DIV:
349 case O_DIVASN:
350 if (vl->val.i == LONG_MIN && vr->val.i == -1)
351 res = LONG_MIN;
352 else
353 res = vl->val.i / vr->val.i;
354 break;
355 case O_MOD:
356 case O_MODASN:
357 if (vl->val.i == LONG_MIN && vr->val.i == -1)
358 res = 0;
359 else
360 res = vl->val.i % vr->val.i;
361 break;
362 case O_PLUS:
363 case O_PLUSASN:
364 res = vl->val.i + vr->val.i;
365 break;
366 case O_MINUS:
367 case O_MINUSASN:
368 res = vl->val.i - vr->val.i;
369 break;
370 case O_LSHIFT:
371 case O_LSHIFTASN:
372 res = vl->val.i << vr->val.i;
373 break;
374 case O_RSHIFT:
375 case O_RSHIFTASN:
376 res = vl->val.i >> vr->val.i;
377 break;
378 case O_LT:
379 res = vl->val.i < vr->val.i;
380 break;
381 case O_LE:
382 res = vl->val.i <= vr->val.i;
383 break;
384 case O_GT:
385 res = vl->val.i > vr->val.i;
386 break;
387 case O_GE:
388 res = vl->val.i >= vr->val.i;
389 break;
390 case O_EQ:
391 res = vl->val.i == vr->val.i;
392 break;
393 case O_NE:
394 res = vl->val.i != vr->val.i;
395 break;
396 case O_BAND:
397 case O_BANDASN:
398 res = vl->val.i & vr->val.i;
399 break;
400 case O_BXOR:
401 case O_BXORASN:
402 res = vl->val.i ^ vr->val.i;
403 break;
404 case O_BOR:
405 case O_BORASN:
406 res = vl->val.i | vr->val.i;
407 break;
408 case O_LAND:
409 if (!vl->val.i)
410 es->noassign++;
411 vr = intvar(es, evalexpr(es, ((int) prec) - 1));
412 res = vl->val.i && vr->val.i;
413 if (!vl->val.i)
414 es->noassign--;
415 break;
416 case O_LOR:
417 if (vl->val.i)
418 es->noassign++;
419 vr = intvar(es, evalexpr(es, ((int) prec) - 1));
420 res = vl->val.i || vr->val.i;
421 if (vl->val.i)
422 es->noassign--;
423 break;
424 case O_TERN:
425 {
426 int e = vl->val.i != 0;
427
428 if (!e)
429 es->noassign++;
430 vl = evalexpr(es, MAX_PREC);
431 if (!e)
432 es->noassign--;
433 if (es->tok != CTERN)
434 evalerr(es, ET_STR, "missing :");
435 token(es);
436 if (e)
437 es->noassign++;
438 vr = evalexpr(es, P_TERN);
439 if (e)
440 es->noassign--;
441 vl = e ? vl : vr;
442 }
443 break;
444 case O_ASN:
445 res = vr->val.i;
446 break;
447 case O_COMMA:
448 res = vr->val.i;
449 break;
450 }
451 if (IS_ASSIGNOP(op)) {
452 vr->val.i = res;
453 if (vasn->flag & INTEGER)
454 setint_v(vasn, vr, es->arith);
455 else
456 setint(vasn, res);
457 vl = vr;
458 } else if (op != O_TERN)
459 vl->val.i = res;
460 }
461 return vl;
462}
463
464static void
465token(Expr_state *es)
466{
467 const char *cp;
468 int c;
469 char *tvar;
470
471 /* skip white space */
472 for (cp = es->tokp; (c = *cp), isspace((unsigned char)c); cp++)
473 ;
474 es->tokp = cp;
475
476 if (c == '\0')
477 es->tok = END;
478 else if (letter(c)) {
479 for (; letnum(c); c = *cp)
480 cp++;
481 if (c == '[') {
482 int len;
483
484 len = array_ref_len(cp);
485 if (len == 0)
486 evalerr(es, ET_STR, "missing ]");
487 cp += len;
488 } else if (c == '(' /*)*/ ) {
489 /* todo: add math functions (all take single argument):
490 * abs acos asin atan cos cosh exp int log sin sinh sqrt
491 * tan tanh
492 */
493 ;
494 }
495 if (es->noassign) {
496 es->val = tempvar();
497 es->val->flag |= EXPRLVALUE;
498 } else {
499 tvar = str_nsave(es->tokp, cp - es->tokp, ATEMP);
500 es->val = global(tvar);
501 afree(tvar, ATEMP);
502 }
503 es->tok = VAR;
504 } else if (digit(c)) {
505 for (; c != '_' && (letnum(c) || c == '#'); c = *cp++)
506 ;
507 tvar = str_nsave(es->tokp, --cp - es->tokp, ATEMP);
508 es->val = tempvar();
509 es->val->flag &= ~INTEGER;
510 es->val->type = 0;
511 es->val->val.s = tvar;
512 if (setint_v(es->val, es->val, es->arith) == NULL)
513 evalerr(es, ET_BADLIT, tvar);
514 afree(tvar, ATEMP);
515 es->tok = LIT;
516 } else {
517 int i, n0;
518
519 for (i = 0; (n0 = opinfo[i].name[0]); i++)
520 if (c == n0 &&
521 strncmp(cp, opinfo[i].name, opinfo[i].len) == 0) {
522 es->tok = (enum token) i;
523 cp += opinfo[i].len;
524 break;
525 }
526 if (!n0)
527 es->tok = BAD;
528 }
529 es->tokp = cp;
530}
531
532/* Do a ++ or -- operation */
533static struct tbl *
534do_ppmm(Expr_state *es, enum token op, struct tbl *vasn, bool is_prefix)
535{
536 struct tbl *vl;
537 int oval;
538
539 assign_check(es, op, vasn);
540
541 vl = intvar(es, vasn);
542 oval = op == O_PLUSPLUS ? vl->val.i++ : vl->val.i--;
543 if (vasn->flag & INTEGER)
544 setint_v(vasn, vl, es->arith);
545 else
546 setint(vasn, vl->val.i);
547 if (!is_prefix) /* undo the inc/dec */
548 vl->val.i = oval;
549
550 return vl;
551}
552
553static void
554assign_check(Expr_state *es, enum token op, struct tbl *vasn)
555{
556 if (es->tok == END || vasn == NULL ||
557 (vasn->name[0] == '\0' && !(vasn->flag & EXPRLVALUE)))
558 evalerr(es, ET_LVALUE, opinfo[(int) op].name);
559 else if (vasn->flag & RDONLY)
560 evalerr(es, ET_RDONLY, opinfo[(int) op].name);
561}
562
563static struct tbl *
564tempvar(void)
565{
566 struct tbl *vp;
567
568 vp = alloc(sizeof(struct tbl), ATEMP);
569 vp->flag = ISSET|INTEGER;
570 vp->type = 0;
571 vp->areap = ATEMP;
572 vp->val.i = 0;
573 vp->name[0] = '\0';
574 return vp;
575}
576
577/* cast (string) variable to temporary integer variable */
578static struct tbl *
579intvar(Expr_state *es, struct tbl *vp)
580{
581 struct tbl *vq;
582
583 /* try to avoid replacing a temp var with another temp var */
584 if (vp->name[0] == '\0' &&
585 (vp->flag & (ISSET|INTEGER|EXPRLVALUE)) == (ISSET|INTEGER))
586 return vp;
587
588 vq = tempvar();
589 if (setint_v(vq, vp, es->arith) == NULL) {
590 if (vp->flag & EXPRINEVAL)
591 evalerr(es, ET_RECURSIVE, vp->name);
592 es->evaling = vp;
593 vp->flag |= EXPRINEVAL;
594 disable_subst++;
595 v_evaluate(vq, str_val(vp), KSH_UNWIND_ERROR, es->arith);
596 disable_subst--;
597 vp->flag &= ~EXPRINEVAL;
598 es->evaling = NULL;
599 }
600 return vq;
601}
