xref: /freebsd/bin/expr/expr.y (revision 74bf4e164ba5851606a27d4feff27717452583e5) 
1 %{
2 /* Written by Pace Willisson (pace@blitz.com)
3  * and placed in the public domain.
4  *
5  * Largely rewritten by J.T. Conklin (jtc@wimsey.com)
6  *
7  * $FreeBSD$
8  */
9 
10 #include <sys/types.h>
11 
12 #include <ctype.h>
13 #include <err.h>
14 #include <errno.h>
15 #include <inttypes.h>
16 #include <limits.h>
17 #include <locale.h>
18 #include <stdio.h>
19 #include <stdlib.h>
20 #include <string.h>
21 #include <regex.h>
22 #include <unistd.h>
23 
24 /*
25  * POSIX specifies a specific error code for syntax errors.  We exit
26  * with this code for all errors.
27  */
28 #define	ERR_EXIT	2
29 
30 enum valtype {
31 	integer, numeric_string, string
32 } ;
33 
34 struct val {
35 	enum valtype type;
36 	union {
37 		char *s;
38 		intmax_t i;
39 	} u;
40 } ;
41 
42 struct val *result;
43 
44 int		chk_div(intmax_t, intmax_t);
45 int		chk_minus(intmax_t, intmax_t, intmax_t);
46 int		chk_plus(intmax_t, intmax_t, intmax_t);
47 int		chk_times(intmax_t, intmax_t, intmax_t);
48 void		free_value(struct val *);
49 int		is_zero_or_null(struct val *);
50 int		isstring(struct val *);
51 struct val	*make_integer(intmax_t);
52 struct val	*make_str(const char *);
53 struct val	*op_and(struct val *, struct val *);
54 struct val	*op_colon(struct val *, struct val *);
55 struct val	*op_div(struct val *, struct val *);
56 struct val	*op_eq(struct val *, struct val *);
57 struct val	*op_ge(struct val *, struct val *);
58 struct val	*op_gt(struct val *, struct val *);
59 struct val	*op_le(struct val *, struct val *);
60 struct val	*op_lt(struct val *, struct val *);
61 struct val	*op_minus(struct val *, struct val *);
62 struct val	*op_ne(struct val *, struct val *);
63 struct val	*op_or(struct val *, struct val *);
64 struct val	*op_plus(struct val *, struct val *);
65 struct val	*op_rem(struct val *, struct val *);
66 struct val	*op_times(struct val *, struct val *);
67 intmax_t	to_integer(struct val *);
68 void		to_string(struct val *);
69 int		yyerror(const char *);
70 int		yylex(void);
71 int		yyparse(void);
72 
73 static int	eflag;
74 char **av;
75 %}
76 
77 %union
78 {
79 	struct val *val;
80 }
81 
82 %left <val> '|'
83 %left <val> '&'
84 %left <val> '=' '>' '<' GE LE NE
85 %left <val> '+' '-'
86 %left <val> '*' '/' '%'
87 %left <val> ':'
88 
89 %token <val> TOKEN
90 %type <val> start expr
91 
92 %%
93 
94 start: expr { result = $$; }
95 
96 expr:	TOKEN
97 	| '(' expr ')' { $$ = $2; }
98 	| expr '|' expr { $$ = op_or ($1, $3); }
99 	| expr '&' expr { $$ = op_and ($1, $3); }
100 	| expr '=' expr { $$ = op_eq ($1, $3); }
101 	| expr '>' expr { $$ = op_gt ($1, $3); }
102 	| expr '<' expr { $$ = op_lt ($1, $3); }
103 	| expr GE expr  { $$ = op_ge ($1, $3); }
104 	| expr LE expr  { $$ = op_le ($1, $3); }
105 	| expr NE expr  { $$ = op_ne ($1, $3); }
106 	| expr '+' expr { $$ = op_plus ($1, $3); }
107 	| expr '-' expr { $$ = op_minus ($1, $3); }
108 	| expr '*' expr { $$ = op_times ($1, $3); }
109 	| expr '/' expr { $$ = op_div ($1, $3); }
110 	| expr '%' expr { $$ = op_rem ($1, $3); }
111 	| expr ':' expr { $$ = op_colon ($1, $3); }
112 	;
113 
114 
115 %%
116 
117 struct val *
118 make_integer(intmax_t i)
119 {
120 	struct val *vp;
121 
122 	vp = (struct val *) malloc (sizeof (*vp));
123 	if (vp == NULL) {
124 		errx(ERR_EXIT, "malloc() failed");
125 	}
126 
127 	vp->type = integer;
128 	vp->u.i  = i;
129 	return vp;
130 }
131 
132 struct val *
133 make_str(const char *s)
134 {
135 	struct val *vp;
136 	char *ep;
137 
138 	vp = (struct val *) malloc (sizeof (*vp));
139 	if (vp == NULL || ((vp->u.s = strdup (s)) == NULL)) {
140 		errx(ERR_EXIT, "malloc() failed");
141 	}
142 
143 	/*
144 	 * Previously we tried to scan the string to see if it ``looked like''
145 	 * an integer (erroneously, as it happened).  Let strtoimax() do the
146 	 * dirty work.  We could cache the value, except that we are using
147 	 * a union and need to preserve the original string form until we
148 	 * are certain that it is not needed.
149 	 *
150 	 * IEEE Std.1003.1-2001 says:
151 	 * /integer/ An argument consisting only of an (optional) unary minus
152 	 *	     followed by digits.
153 	 *
154 	 * This means that arguments which consist of digits followed by
155 	 * non-digits MUST NOT be considered integers.  strtoimax() will
156 	 * figure this out for us.
157 	 */
158 	if (eflag)
159 		(void)strtoimax(s, &ep, 10);
160 	else
161 		(void)strtol(s, &ep, 10);
162 
163 	if (*ep != '\0')
164 		vp->type = string;
165 	else
166 		vp->type = numeric_string;
167 
168 	return vp;
169 }
170 
171 
172 void
173 free_value(struct val *vp)
174 {
175 	if (vp->type == string || vp->type == numeric_string)
176 		free (vp->u.s);
177 }
178 
179 
180 intmax_t
181 to_integer(struct val *vp)
182 {
183 	intmax_t i;
184 
185 	if (vp->type == integer)
186 		return 1;
187 
188 	if (vp->type == string)
189 		return 0;
190 
191 	/* vp->type == numeric_string, make it numeric */
192 	errno = 0;
193 	if (eflag) {
194 		i  = strtoimax(vp->u.s, (char **)NULL, 10);
195 		if (errno == ERANGE)
196 			err(ERR_EXIT, NULL);
197 	} else {
198 		i = strtol(vp->u.s, (char **)NULL, 10);
199 	}
200 
201 	free (vp->u.s);
202 	vp->u.i = i;
203 	vp->type = integer;
204 	return 1;
205 }
206 
207 void
208 to_string(struct val *vp)
209 {
210 	char *tmp;
211 
212 	if (vp->type == string || vp->type == numeric_string)
213 		return;
214 
215 	/*
216 	 * log_10(x) ~= 0.3 * log_2(x).  Rounding up gives the number
217 	 * of digits; add one each for the sign and terminating null
218 	 * character, respectively.
219 	 */
220 #define	NDIGITS(x) (3 * (sizeof(x) * CHAR_BIT) / 10 + 1 + 1 + 1)
221 	tmp = malloc(NDIGITS(vp->u.i));
222 	if (tmp == NULL)
223 		errx(ERR_EXIT, "malloc() failed");
224 
225 	sprintf(tmp, "%jd", vp->u.i);
226 	vp->type = string;
227 	vp->u.s  = tmp;
228 }
229 
230 
231 int
232 isstring(struct val *vp)
233 {
234 	/* only TRUE if this string is not a valid integer */
235 	return (vp->type == string);
236 }
237 
238 
239 int
240 yylex(void)
241 {
242 	char *p;
243 
244 	if (*av == NULL)
245 		return (0);
246 
247 	p = *av++;
248 
249 	if (strlen (p) == 1) {
250 		if (strchr ("|&=<>+-*/%:()", *p))
251 			return (*p);
252 	} else if (strlen (p) == 2 && p[1] == '=') {
253 		switch (*p) {
254 		case '>': return (GE);
255 		case '<': return (LE);
256 		case '!': return (NE);
257 		}
258 	}
259 
260 	yylval.val = make_str (p);
261 	return (TOKEN);
262 }
263 
264 int
265 is_zero_or_null(struct val *vp)
266 {
267 	if (vp->type == integer) {
268 		return (vp->u.i == 0);
269 	} else {
270 		return (*vp->u.s == 0 || (to_integer (vp) && vp->u.i == 0));
271 	}
272 	/* NOTREACHED */
273 }
274 
275 int
276 main(int argc, char *argv[])
277 {
278 	int c;
279 
280 	setlocale (LC_ALL, "");
281 	if (getenv("EXPR_COMPAT") != NULL
282 	    || check_utility_compat("expr")) {
283 		av = argv + 1;
284 		eflag = 1;
285 	} else {
286 		while ((c = getopt(argc, argv, "e")) != -1)
287 			switch (c) {
288 			case 'e':
289 				eflag = 1;
290 				break;
291 
292 			default:
293 				fprintf(stderr,
294 				    "usage: expr [-e] expression\n");
295 				exit(ERR_EXIT);
296 			}
297 		av = argv + optind;
298 	}
299 
300 	yyparse();
301 
302 	if (result->type == integer)
303 		printf("%jd\n", result->u.i);
304 	else
305 		printf("%s\n", result->u.s);
306 
307 	return (is_zero_or_null(result));
308 }
309 
310 int
311 yyerror(const char *s __unused)
312 {
313 	errx(ERR_EXIT, "syntax error");
314 }
315 
316 
317 struct val *
318 op_or(struct val *a, struct val *b)
319 {
320 	if (is_zero_or_null (a)) {
321 		free_value (a);
322 		return (b);
323 	} else {
324 		free_value (b);
325 		return (a);
326 	}
327 }
328 
329 struct val *
330 op_and(struct val *a, struct val *b)
331 {
332 	if (is_zero_or_null (a) || is_zero_or_null (b)) {
333 		free_value (a);
334 		free_value (b);
335 		return (make_integer ((intmax_t)0));
336 	} else {
337 		free_value (b);
338 		return (a);
339 	}
340 }
341 
342 struct val *
343 op_eq(struct val *a, struct val *b)
344 {
345 	struct val *r;
346 
347 	if (isstring (a) || isstring (b)) {
348 		to_string (a);
349 		to_string (b);
350 		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) == 0));
351 	} else {
352 		(void)to_integer(a);
353 		(void)to_integer(b);
354 		r = make_integer ((intmax_t)(a->u.i == b->u.i));
355 	}
356 
357 	free_value (a);
358 	free_value (b);
359 	return r;
360 }
361 
362 struct val *
363 op_gt(struct val *a, struct val *b)
364 {
365 	struct val *r;
366 
367 	if (isstring (a) || isstring (b)) {
368 		to_string (a);
369 		to_string (b);
370 		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) > 0));
371 	} else {
372 		(void)to_integer(a);
373 		(void)to_integer(b);
374 		r = make_integer ((intmax_t)(a->u.i > b->u.i));
375 	}
376 
377 	free_value (a);
378 	free_value (b);
379 	return r;
380 }
381 
382 struct val *
383 op_lt(struct val *a, struct val *b)
384 {
385 	struct val *r;
386 
387 	if (isstring (a) || isstring (b)) {
388 		to_string (a);
389 		to_string (b);
390 		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) < 0));
391 	} else {
392 		(void)to_integer(a);
393 		(void)to_integer(b);
394 		r = make_integer ((intmax_t)(a->u.i < b->u.i));
395 	}
396 
397 	free_value (a);
398 	free_value (b);
399 	return r;
400 }
401 
402 struct val *
403 op_ge(struct val *a, struct val *b)
404 {
405 	struct val *r;
406 
407 	if (isstring (a) || isstring (b)) {
408 		to_string (a);
409 		to_string (b);
410 		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) >= 0));
411 	} else {
412 		(void)to_integer(a);
413 		(void)to_integer(b);
414 		r = make_integer ((intmax_t)(a->u.i >= b->u.i));
415 	}
416 
417 	free_value (a);
418 	free_value (b);
419 	return r;
420 }
421 
422 struct val *
423 op_le(struct val *a, struct val *b)
424 {
425 	struct val *r;
426 
427 	if (isstring (a) || isstring (b)) {
428 		to_string (a);
429 		to_string (b);
430 		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) <= 0));
431 	} else {
432 		(void)to_integer(a);
433 		(void)to_integer(b);
434 		r = make_integer ((intmax_t)(a->u.i <= b->u.i));
435 	}
436 
437 	free_value (a);
438 	free_value (b);
439 	return r;
440 }
441 
442 struct val *
443 op_ne(struct val *a, struct val *b)
444 {
445 	struct val *r;
446 
447 	if (isstring (a) || isstring (b)) {
448 		to_string (a);
449 		to_string (b);
450 		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) != 0));
451 	} else {
452 		(void)to_integer(a);
453 		(void)to_integer(b);
454 		r = make_integer ((intmax_t)(a->u.i != b->u.i));
455 	}
456 
457 	free_value (a);
458 	free_value (b);
459 	return r;
460 }
461 
462 int
463 chk_plus(intmax_t a, intmax_t b, intmax_t r)
464 {
465 
466 	/* sum of two positive numbers must be positive */
467 	if (a > 0 && b > 0 && r <= 0)
468 		return 1;
469 	/* sum of two negative numbers must be negative */
470 	if (a < 0 && b < 0 && r >= 0)
471 		return 1;
472 	/* all other cases are OK */
473 	return 0;
474 }
475 
476 struct val *
477 op_plus(struct val *a, struct val *b)
478 {
479 	struct val *r;
480 
481 	if (!to_integer(a) || !to_integer(b)) {
482 		errx(ERR_EXIT, "non-numeric argument");
483 	}
484 
485 	if (eflag) {
486 		r = make_integer(a->u.i + b->u.i);
487 		if (chk_plus(a->u.i, b->u.i, r->u.i)) {
488 			errx(ERR_EXIT, "overflow");
489 		}
490 	} else
491 		r = make_integer((long)a->u.i + (long)b->u.i);
492 
493 	free_value (a);
494 	free_value (b);
495 	return r;
496 }
497 
498 int
499 chk_minus(intmax_t a, intmax_t b, intmax_t r)
500 {
501 
502 	/* special case subtraction of INTMAX_MIN */
503 	if (b == INTMAX_MIN) {
504 		if (a >= 0)
505 			return 1;
506 		else
507 			return 0;
508 	}
509 	/* this is allowed for b != INTMAX_MIN */
510 	return chk_plus (a, -b, r);
511 }
512 
513 struct val *
514 op_minus(struct val *a, struct val *b)
515 {
516 	struct val *r;
517 
518 	if (!to_integer(a) || !to_integer(b)) {
519 		errx(ERR_EXIT, "non-numeric argument");
520 	}
521 
522 	if (eflag) {
523 		r = make_integer(a->u.i - b->u.i);
524 		if (chk_minus(a->u.i, b->u.i, r->u.i)) {
525 			errx(ERR_EXIT, "overflow");
526 		}
527 	} else
528 		r = make_integer((long)a->u.i - (long)b->u.i);
529 
530 	free_value (a);
531 	free_value (b);
532 	return r;
533 }
534 
535 int
536 chk_times(intmax_t a, intmax_t b, intmax_t r)
537 {
538 	/* special case: first operand is 0, no overflow possible */
539 	if (a == 0)
540 		return 0;
541 	/* cerify that result of division matches second operand */
542 	if (r / a != b)
543 		return 1;
544 	return 0;
545 }
546 
547 struct val *
548 op_times(struct val *a, struct val *b)
549 {
550 	struct val *r;
551 
552 	if (!to_integer(a) || !to_integer(b)) {
553 		errx(ERR_EXIT, "non-numeric argument");
554 	}
555 
556 	if (eflag) {
557 		r = make_integer(a->u.i * b->u.i);
558 		if (chk_times(a->u.i, b->u.i, r->u.i)) {
559 			errx(ERR_EXIT, "overflow");
560 		}
561 	} else
562 		r = make_integer((long)a->u.i * (long)b->u.i);
563 
564 	free_value (a);
565 	free_value (b);
566 	return (r);
567 }
568 
569 int
570 chk_div(intmax_t a, intmax_t b)
571 {
572 	/* div by zero has been taken care of before */
573 	/* only INTMAX_MIN / -1 causes overflow */
574 	if (a == INTMAX_MIN && b == -1)
575 		return 1;
576 	/* everything else is OK */
577 	return 0;
578 }
579 
580 struct val *
581 op_div(struct val *a, struct val *b)
582 {
583 	struct val *r;
584 
585 	if (!to_integer(a) || !to_integer(b)) {
586 		errx(ERR_EXIT, "non-numeric argument");
587 	}
588 
589 	if (b->u.i == 0) {
590 		errx(ERR_EXIT, "division by zero");
591 	}
592 
593 	if (eflag) {
594 		r = make_integer(a->u.i / b->u.i);
595 		if (chk_div(a->u.i, b->u.i)) {
596 			errx(ERR_EXIT, "overflow");
597 		}
598 	} else
599 		r = make_integer((long)a->u.i / (long)b->u.i);
600 
601 	free_value (a);
602 	free_value (b);
603 	return r;
604 }
605 
606 struct val *
607 op_rem(struct val *a, struct val *b)
608 {
609 	struct val *r;
610 
611 	if (!to_integer(a) || !to_integer(b)) {
612 		errx(ERR_EXIT, "non-numeric argument");
613 	}
614 
615 	if (b->u.i == 0) {
616 		errx(ERR_EXIT, "division by zero");
617 	}
618 
619 	if (eflag)
620 		r = make_integer(a->u.i % b->u.i);
621 	        /* chk_rem necessary ??? */
622 	else
623 		r = make_integer((long)a->u.i % (long)b->u.i);
624 
625 	free_value (a);
626 	free_value (b);
627 	return r;
628 }
629 
630 struct val *
631 op_colon(struct val *a, struct val *b)
632 {
633 	regex_t rp;
634 	regmatch_t rm[2];
635 	char errbuf[256];
636 	int eval;
637 	struct val *v;
638 
639 	/* coerce to both arguments to strings */
640 	to_string(a);
641 	to_string(b);
642 
643 	/* compile regular expression */
644 	if ((eval = regcomp (&rp, b->u.s, 0)) != 0) {
645 		regerror (eval, &rp, errbuf, sizeof(errbuf));
646 		errx(ERR_EXIT, "%s", errbuf);
647 	}
648 
649 	/* compare string against pattern */
650 	/* remember that patterns are anchored to the beginning of the line */
651 	if (regexec(&rp, a->u.s, (size_t)2, rm, 0) == 0 && rm[0].rm_so == 0) {
652 		if (rm[1].rm_so >= 0) {
653 			*(a->u.s + rm[1].rm_eo) = '\0';
654 			v = make_str (a->u.s + rm[1].rm_so);
655 
656 		} else {
657 			v = make_integer ((intmax_t)(rm[0].rm_eo - rm[0].rm_so));
658 		}
659 	} else {
660 		if (rp.re_nsub == 0) {
661 			v = make_integer ((intmax_t)0);
662 		} else {
663 			v = make_str ("");
664 		}
665 	}
666 
667 	/* free arguments and pattern buffer */
668 	free_value (a);
669 	free_value (b);
670 	regfree (&rp);
671 
672 	return v;
673 }
674