xref: /freebsd/bin/expr/expr.y (revision c11e094d96120a2e0e726ed9705ae0ec08db49b6) 
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 		av = argv + 1;
283 		eflag = 1;
284 	} else {
285 		while ((c = getopt(argc, argv, "e")) != -1)
286 			switch (c) {
287 			case 'e':
288 				eflag = 1;
289 				break;
290 
291 			default:
292 				fprintf(stderr,
293 				    "usage: expr [-e] expression\n");
294 				exit(ERR_EXIT);
295 			}
296 		av = argv + optind;
297 	}
298 
299 	yyparse();
300 
301 	if (result->type == integer)
302 		printf("%jd\n", result->u.i);
303 	else
304 		printf("%s\n", result->u.s);
305 
306 	return (is_zero_or_null(result));
307 }
308 
309 int
310 yyerror(const char *s __unused)
311 {
312 	errx(ERR_EXIT, "syntax error");
313 }
314 
315 
316 struct val *
317 op_or(struct val *a, struct val *b)
318 {
319 	if (is_zero_or_null (a)) {
320 		free_value (a);
321 		return (b);
322 	} else {
323 		free_value (b);
324 		return (a);
325 	}
326 }
327 
328 struct val *
329 op_and(struct val *a, struct val *b)
330 {
331 	if (is_zero_or_null (a) || is_zero_or_null (b)) {
332 		free_value (a);
333 		free_value (b);
334 		return (make_integer ((intmax_t)0));
335 	} else {
336 		free_value (b);
337 		return (a);
338 	}
339 }
340 
341 struct val *
342 op_eq(struct val *a, struct val *b)
343 {
344 	struct val *r;
345 
346 	if (isstring (a) || isstring (b)) {
347 		to_string (a);
348 		to_string (b);
349 		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) == 0));
350 	} else {
351 		(void)to_integer(a);
352 		(void)to_integer(b);
353 		r = make_integer ((intmax_t)(a->u.i == b->u.i));
354 	}
355 
356 	free_value (a);
357 	free_value (b);
358 	return r;
359 }
360 
361 struct val *
362 op_gt(struct val *a, struct val *b)
363 {
364 	struct val *r;
365 
366 	if (isstring (a) || isstring (b)) {
367 		to_string (a);
368 		to_string (b);
369 		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) > 0));
370 	} else {
371 		(void)to_integer(a);
372 		(void)to_integer(b);
373 		r = make_integer ((intmax_t)(a->u.i > b->u.i));
374 	}
375 
376 	free_value (a);
377 	free_value (b);
378 	return r;
379 }
380 
381 struct val *
382 op_lt(struct val *a, struct val *b)
383 {
384 	struct val *r;
385 
386 	if (isstring (a) || isstring (b)) {
387 		to_string (a);
388 		to_string (b);
389 		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) < 0));
390 	} else {
391 		(void)to_integer(a);
392 		(void)to_integer(b);
393 		r = make_integer ((intmax_t)(a->u.i < b->u.i));
394 	}
395 
396 	free_value (a);
397 	free_value (b);
398 	return r;
399 }
400 
401 struct val *
402 op_ge(struct val *a, struct val *b)
403 {
404 	struct val *r;
405 
406 	if (isstring (a) || isstring (b)) {
407 		to_string (a);
408 		to_string (b);
409 		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) >= 0));
410 	} else {
411 		(void)to_integer(a);
412 		(void)to_integer(b);
413 		r = make_integer ((intmax_t)(a->u.i >= b->u.i));
414 	}
415 
416 	free_value (a);
417 	free_value (b);
418 	return r;
419 }
420 
421 struct val *
422 op_le(struct val *a, struct val *b)
423 {
424 	struct val *r;
425 
426 	if (isstring (a) || isstring (b)) {
427 		to_string (a);
428 		to_string (b);
429 		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) <= 0));
430 	} else {
431 		(void)to_integer(a);
432 		(void)to_integer(b);
433 		r = make_integer ((intmax_t)(a->u.i <= b->u.i));
434 	}
435 
436 	free_value (a);
437 	free_value (b);
438 	return r;
439 }
440 
441 struct val *
442 op_ne(struct val *a, struct val *b)
443 {
444 	struct val *r;
445 
446 	if (isstring (a) || isstring (b)) {
447 		to_string (a);
448 		to_string (b);
449 		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) != 0));
450 	} else {
451 		(void)to_integer(a);
452 		(void)to_integer(b);
453 		r = make_integer ((intmax_t)(a->u.i != b->u.i));
454 	}
455 
456 	free_value (a);
457 	free_value (b);
458 	return r;
459 }
460 
461 int
462 chk_plus(intmax_t a, intmax_t b, intmax_t r)
463 {
464 
465 	/* sum of two positive numbers must be positive */
466 	if (a > 0 && b > 0 && r <= 0)
467 		return 1;
468 	/* sum of two negative numbers must be negative */
469 	if (a < 0 && b < 0 && r >= 0)
470 		return 1;
471 	/* all other cases are OK */
472 	return 0;
473 }
474 
475 struct val *
476 op_plus(struct val *a, struct val *b)
477 {
478 	struct val *r;
479 
480 	if (!to_integer(a) || !to_integer(b)) {
481 		errx(ERR_EXIT, "non-numeric argument");
482 	}
483 
484 	if (eflag) {
485 		r = make_integer(a->u.i + b->u.i);
486 		if (chk_plus(a->u.i, b->u.i, r->u.i)) {
487 			errx(ERR_EXIT, "overflow");
488 		}
489 	} else
490 		r = make_integer((long)a->u.i + (long)b->u.i);
491 
492 	free_value (a);
493 	free_value (b);
494 	return r;
495 }
496 
497 int
498 chk_minus(intmax_t a, intmax_t b, intmax_t r)
499 {
500 
501 	/* special case subtraction of INTMAX_MIN */
502 	if (b == INTMAX_MIN) {
503 		if (a >= 0)
504 			return 1;
505 		else
506 			return 0;
507 	}
508 	/* this is allowed for b != INTMAX_MIN */
509 	return chk_plus (a, -b, r);
510 }
511 
512 struct val *
513 op_minus(struct val *a, struct val *b)
514 {
515 	struct val *r;
516 
517 	if (!to_integer(a) || !to_integer(b)) {
518 		errx(ERR_EXIT, "non-numeric argument");
519 	}
520 
521 	if (eflag) {
522 		r = make_integer(a->u.i - b->u.i);
523 		if (chk_minus(a->u.i, b->u.i, r->u.i)) {
524 			errx(ERR_EXIT, "overflow");
525 		}
526 	} else
527 		r = make_integer((long)a->u.i - (long)b->u.i);
528 
529 	free_value (a);
530 	free_value (b);
531 	return r;
532 }
533 
534 int
535 chk_times(intmax_t a, intmax_t b, intmax_t r)
536 {
537 	/* special case: first operand is 0, no overflow possible */
538 	if (a == 0)
539 		return 0;
540 	/* cerify that result of division matches second operand */
541 	if (r / a != b)
542 		return 1;
543 	return 0;
544 }
545 
546 struct val *
547 op_times(struct val *a, struct val *b)
548 {
549 	struct val *r;
550 
551 	if (!to_integer(a) || !to_integer(b)) {
552 		errx(ERR_EXIT, "non-numeric argument");
553 	}
554 
555 	if (eflag) {
556 		r = make_integer(a->u.i * b->u.i);
557 		if (chk_times(a->u.i, b->u.i, r->u.i)) {
558 			errx(ERR_EXIT, "overflow");
559 		}
560 	} else
561 		r = make_integer((long)a->u.i * (long)b->u.i);
562 
563 	free_value (a);
564 	free_value (b);
565 	return (r);
566 }
567 
568 int
569 chk_div(intmax_t a, intmax_t b)
570 {
571 	/* div by zero has been taken care of before */
572 	/* only INTMAX_MIN / -1 causes overflow */
573 	if (a == INTMAX_MIN && b == -1)
574 		return 1;
575 	/* everything else is OK */
576 	return 0;
577 }
578 
579 struct val *
580 op_div(struct val *a, struct val *b)
581 {
582 	struct val *r;
583 
584 	if (!to_integer(a) || !to_integer(b)) {
585 		errx(ERR_EXIT, "non-numeric argument");
586 	}
587 
588 	if (b->u.i == 0) {
589 		errx(ERR_EXIT, "division by zero");
590 	}
591 
592 	if (eflag) {
593 		r = make_integer(a->u.i / b->u.i);
594 		if (chk_div(a->u.i, b->u.i)) {
595 			errx(ERR_EXIT, "overflow");
596 		}
597 	} else
598 		r = make_integer((long)a->u.i / (long)b->u.i);
599 
600 	free_value (a);
601 	free_value (b);
602 	return r;
603 }
604 
605 struct val *
606 op_rem(struct val *a, struct val *b)
607 {
608 	struct val *r;
609 
610 	if (!to_integer(a) || !to_integer(b)) {
611 		errx(ERR_EXIT, "non-numeric argument");
612 	}
613 
614 	if (b->u.i == 0) {
615 		errx(ERR_EXIT, "division by zero");
616 	}
617 
618 	if (eflag)
619 		r = make_integer(a->u.i % b->u.i);
620 	        /* chk_rem necessary ??? */
621 	else
622 		r = make_integer((long)a->u.i % (long)b->u.i);
623 
624 	free_value (a);
625 	free_value (b);
626 	return r;
627 }
628 
629 struct val *
630 op_colon(struct val *a, struct val *b)
631 {
632 	regex_t rp;
633 	regmatch_t rm[2];
634 	char errbuf[256];
635 	int eval;
636 	struct val *v;
637 
638 	/* coerce to both arguments to strings */
639 	to_string(a);
640 	to_string(b);
641 
642 	/* compile regular expression */
643 	if ((eval = regcomp (&rp, b->u.s, 0)) != 0) {
644 		regerror (eval, &rp, errbuf, sizeof(errbuf));
645 		errx(ERR_EXIT, "%s", errbuf);
646 	}
647 
648 	/* compare string against pattern */
649 	/* remember that patterns are anchored to the beginning of the line */
650 	if (regexec(&rp, a->u.s, (size_t)2, rm, 0) == 0 && rm[0].rm_so == 0) {
651 		if (rm[1].rm_so >= 0) {
652 			*(a->u.s + rm[1].rm_eo) = '\0';
653 			v = make_str (a->u.s + rm[1].rm_so);
654 
655 		} else {
656 			v = make_integer ((intmax_t)(rm[0].rm_eo - rm[0].rm_so));
657 		}
658 	} else {
659 		if (rp.re_nsub == 0) {
660 			v = make_integer ((intmax_t)0);
661 		} else {
662 			v = make_str ("");
663 		}
664 	}
665 
666 	/* free arguments and pattern buffer */
667 	free_value (a);
668 	free_value (b);
669 	regfree (&rp);
670 
671 	return v;
672 }
673