expr.y revision 92979 
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: head/bin/expr/expr.y 92979 2002-03-22 20:18:26Z wollman $
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
30enum valtype {
31	integer, numeric_string, string
32} ;
33
34struct val {
35	enum valtype type;
36	union {
37		char *s;
38		intmax_t i;
39	} u;
40} ;
41
42struct val *result;
43
44int		chk_div(intmax_t, intmax_t);
45int		chk_minus(intmax_t, intmax_t, intmax_t);
46int		chk_plus(intmax_t, intmax_t, intmax_t);
47int		chk_times(intmax_t, intmax_t, intmax_t);
48void		free_value(struct val *);
49int		is_zero_or_null(struct val *);
50int		isstring(struct val *);
51struct val	*make_integer(intmax_t);
52struct val	*make_str(const char *);
53struct val	*op_and(struct val *, struct val *);
54struct val	*op_colon(struct val *, struct val *);
55struct val	*op_div(struct val *, struct val *);
56struct val	*op_eq(struct val *, struct val *);
57struct val	*op_ge(struct val *, struct val *);
58struct val	*op_gt(struct val *, struct val *);
59struct val	*op_le(struct val *, struct val *);
60struct val	*op_lt(struct val *, struct val *);
61struct val	*op_minus(struct val *, struct val *);
62struct val	*op_ne(struct val *, struct val *);
63struct val	*op_or(struct val *, struct val *);
64struct val	*op_plus(struct val *, struct val *);
65struct val	*op_rem(struct val *, struct val *);
66struct val	*op_times(struct val *, struct val *);
67intmax_t	to_integer(struct val *);
68void		to_string(struct val *);
69int		yyerror(const char *);
70int		yylex(void);
71int		yyparse(void);
72
73char **av;
74%}
75
76%union
77{
78	struct val *val;
79}
80
81%left <val> '|'
82%left <val> '&'
83%left <val> '=' '>' '<' GE LE NE
84%left <val> '+' '-'
85%left <val> '*' '/' '%'
86%left <val> ':'
87
88%token <val> TOKEN
89%type <val> start expr
90
91%%
92
93start: expr { result = $$; }
94
95expr:	TOKEN
96	| '(' expr ')' { $$ = $2; }
97	| expr '|' expr { $$ = op_or ($1, $3); }
98	| expr '&' expr { $$ = op_and ($1, $3); }
99	| expr '=' expr { $$ = op_eq ($1, $3); }
100	| expr '>' expr { $$ = op_gt ($1, $3); }
101	| expr '<' expr { $$ = op_lt ($1, $3); }
102	| expr GE expr  { $$ = op_ge ($1, $3); }
103	| expr LE expr  { $$ = op_le ($1, $3); }
104	| expr NE expr  { $$ = op_ne ($1, $3); }
105	| expr '+' expr { $$ = op_plus ($1, $3); }
106	| expr '-' expr { $$ = op_minus ($1, $3); }
107	| expr '*' expr { $$ = op_times ($1, $3); }
108	| expr '/' expr { $$ = op_div ($1, $3); }
109	| expr '%' expr { $$ = op_rem ($1, $3); }
110	| expr ':' expr { $$ = op_colon ($1, $3); }
111	;
112
113
114%%
115
116struct val *
117make_integer(intmax_t i)
118{
119	struct val *vp;
120
121	vp = (struct val *) malloc (sizeof (*vp));
122	if (vp == NULL) {
123		errx(ERR_EXIT, "malloc() failed");
124	}
125
126	vp->type = integer;
127	vp->u.i  = i;
128	return vp;
129}
130
131struct val *
132make_str(const char *s)
133{
134	struct val *vp;
135	char *ep;
136
137	vp = (struct val *) malloc (sizeof (*vp));
138	if (vp == NULL || ((vp->u.s = strdup (s)) == NULL)) {
139		errx(ERR_EXIT, "malloc() failed");
140	}
141
142	/*
143	 * Previously we tried to scan the string to see if it ``looked like''
144	 * an integer (erroneously, as it happened).  Let strtoimax() do the
145	 * dirty work.  We could cache the value, except that we are using
146	 * a union and need to preserve the original string form until we
147	 * are certain that it is not needed.
148	 *
149	 * IEEE Std.1003.1-2001 says:
150	 * /integer/ An argument consisting only of an (optional) unary minus
151	 *	     followed by digits.
152	 *
153	 * This means that arguments which consist of digits followed by
154	 * non-digits MUST NOT be considered integers.  strtoimax() will
155	 * figure this out for us.
156	 */
157	(void)strtoimax(s, &ep, 10);
158
159	if (*ep != '\0')
160		vp->type = string;
161	else
162		vp->type = numeric_string;
163
164	return vp;
165}
166
167
168void
169free_value(struct val *vp)
170{
171	if (vp->type == string || vp->type == numeric_string)
172		free (vp->u.s);
173}
174
175
176intmax_t
177to_integer(struct val *vp)
178{
179	intmax_t i;
180
181	if (vp->type == integer)
182		return 1;
183
184	if (vp->type == string)
185		return 0;
186
187	/* vp->type == numeric_string, make it numeric */
188	errno = 0;
189	i  = strtoimax(vp->u.s, (char **)NULL, 10);
190	if (errno == ERANGE)
191		err(ERR_EXIT, NULL);
192
193	free (vp->u.s);
194	vp->u.i = i;
195	vp->type = integer;
196	return 1;
197}
198
199void
200to_string(struct val *vp)
201{
202	char *tmp;
203
204	if (vp->type == string || vp->type == numeric_string)
205		return;
206
207	/*
208	 * log_10(x) ~= 0.3 * log_2(x).  Rounding up gives the number
209	 * of digits; add one each for the sign and terminating null
210	 * character, respectively.
211	 */
212#define	NDIGITS(x) (3 * (sizeof(x) * CHAR_BIT) / 10 + 1 + 1 + 1)
213	tmp = malloc(NDIGITS(vp->u.i));
214	if (tmp == NULL)
215		errx(ERR_EXIT, "malloc() failed");
216
217	sprintf(tmp, "%jd", vp->u.i);
218	vp->type = string;
219	vp->u.s  = tmp;
220}
221
222
223int
224isstring(struct val *vp)
225{
226	/* only TRUE if this string is not a valid integer */
227	return (vp->type == string);
228}
229
230
231int
232yylex(void)
233{
234	char *p;
235
236	if (*av == NULL)
237		return (0);
238
239	p = *av++;
240
241	if (strlen (p) == 1) {
242		if (strchr ("|&=<>+-*/%:()", *p))
243			return (*p);
244	} else if (strlen (p) == 2 && p[1] == '=') {
245		switch (*p) {
246		case '>': return (GE);
247		case '<': return (LE);
248		case '!': return (NE);
249		}
250	}
251
252	yylval.val = make_str (p);
253	return (TOKEN);
254}
255
256int
257is_zero_or_null(struct val *vp)
258{
259	if (vp->type == integer) {
260		return (vp->u.i == 0);
261	} else {
262		return (*vp->u.s == 0 || (to_integer (vp) && vp->u.i == 0));
263	}
264	/* NOTREACHED */
265}
266
267int
268main(int argc, char *argv[])
269{
270	int c;
271
272	setlocale (LC_ALL, "");
273	while ((c = getopt(argc, argv, "")) != -1)
274		switch (c) {
275		default:
276			fprintf(stderr, "usage: expr [--] expression\n");
277			exit(ERR_EXIT);
278		}
279
280	av = argv + optind;
281
282	yyparse();
283
284	if (result->type == integer)
285		printf("%jd\n", result->u.i);
286	else
287		printf("%s\n", result->u.s);
288
289	return (is_zero_or_null(result));
290}
291
292int
293yyerror(const char *s __unused)
294{
295	errx(ERR_EXIT, "syntax error");
296}
297
298
299struct val *
300op_or(struct val *a, struct val *b)
301{
302	if (is_zero_or_null (a)) {
303		free_value (a);
304		return (b);
305	} else {
306		free_value (b);
307		return (a);
308	}
309}
310
311struct val *
312op_and(struct val *a, struct val *b)
313{
314	if (is_zero_or_null (a) || is_zero_or_null (b)) {
315		free_value (a);
316		free_value (b);
317		return (make_integer ((intmax_t)0));
318	} else {
319		free_value (b);
320		return (a);
321	}
322}
323
324struct val *
325op_eq(struct val *a, struct val *b)
326{
327	struct val *r;
328
329	if (isstring (a) || isstring (b)) {
330		to_string (a);
331		to_string (b);
332		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) == 0));
333	} else {
334		(void)to_integer(a);
335		(void)to_integer(b);
336		r = make_integer ((intmax_t)(a->u.i == b->u.i));
337	}
338
339	free_value (a);
340	free_value (b);
341	return r;
342}
343
344struct val *
345op_gt(struct val *a, struct val *b)
346{
347	struct val *r;
348
349	if (isstring (a) || isstring (b)) {
350		to_string (a);
351		to_string (b);
352		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) > 0));
353	} else {
354		(void)to_integer(a);
355		(void)to_integer(b);
356		r = make_integer ((intmax_t)(a->u.i > b->u.i));
357	}
358
359	free_value (a);
360	free_value (b);
361	return r;
362}
363
364struct val *
365op_lt(struct val *a, struct val *b)
366{
367	struct val *r;
368
369	if (isstring (a) || isstring (b)) {
370		to_string (a);
371		to_string (b);
372		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) < 0));
373	} else {
374		(void)to_integer(a);
375		(void)to_integer(b);
376		r = make_integer ((intmax_t)(a->u.i < b->u.i));
377	}
378
379	free_value (a);
380	free_value (b);
381	return r;
382}
383
384struct val *
385op_ge(struct val *a, struct val *b)
386{
387	struct val *r;
388
389	if (isstring (a) || isstring (b)) {
390		to_string (a);
391		to_string (b);
392		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) >= 0));
393	} else {
394		(void)to_integer(a);
395		(void)to_integer(b);
396		r = make_integer ((intmax_t)(a->u.i >= b->u.i));
397	}
398
399	free_value (a);
400	free_value (b);
401	return r;
402}
403
404struct val *
405op_le(struct val *a, struct val *b)
406{
407	struct val *r;
408
409	if (isstring (a) || isstring (b)) {
410		to_string (a);
411		to_string (b);
412		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) <= 0));
413	} else {
414		(void)to_integer(a);
415		(void)to_integer(b);
416		r = make_integer ((intmax_t)(a->u.i <= b->u.i));
417	}
418
419	free_value (a);
420	free_value (b);
421	return r;
422}
423
424struct val *
425op_ne(struct val *a, struct val *b)
426{
427	struct val *r;
428
429	if (isstring (a) || isstring (b)) {
430		to_string (a);
431		to_string (b);
432		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) != 0));
433	} else {
434		(void)to_integer(a);
435		(void)to_integer(b);
436		r = make_integer ((intmax_t)(a->u.i != b->u.i));
437	}
438
439	free_value (a);
440	free_value (b);
441	return r;
442}
443
444int
445chk_plus(intmax_t a, intmax_t b, intmax_t r)
446{
447	/* sum of two positive numbers must be positive */
448	if (a > 0 && b > 0 && r <= 0)
449		return 1;
450	/* sum of two negative numbers must be negative */
451	if (a < 0 && b < 0 && r >= 0)
452		return 1;
453	/* all other cases are OK */
454	return 0;
455}
456
457struct val *
458op_plus(struct val *a, struct val *b)
459{
460	struct val *r;
461
462	if (!to_integer (a) || !to_integer (b)) {
463		errx(ERR_EXIT, "non-numeric argument");
464	}
465
466	r = make_integer (/*(intmax_t)*/(a->u.i + b->u.i));
467	if (chk_plus (a->u.i, b->u.i, r->u.i)) {
468		errx(ERR_EXIT, "overflow");
469	}
470	free_value (a);
471	free_value (b);
472	return r;
473}
474
475int
476chk_minus(intmax_t a, intmax_t b, intmax_t r)
477{
478	/* special case subtraction of INTMAX_MIN */
479	if (b == INTMAX_MIN) {
480		if (a >= 0)
481			return 1;
482		else
483			return 0;
484	}
485	/* this is allowed for b != INTMAX_MIN */
486	return chk_plus (a, -b, r);
487}
488
489struct val *
490op_minus(struct val *a, struct val *b)
491{
492	struct val *r;
493
494	if (!to_integer (a) || !to_integer (b)) {
495		errx(ERR_EXIT, "non-numeric argument");
496	}
497
498	r = make_integer (/*(intmax_t)*/(a->u.i - b->u.i));
499	if (chk_minus (a->u.i, b->u.i, r->u.i)) {
500		errx(ERR_EXIT, "overflow");
501	}
502	free_value (a);
503	free_value (b);
504	return r;
505}
506
507int
508chk_times(intmax_t a, intmax_t b, intmax_t r)
509{
510	/* special case: first operand is 0, no overflow possible */
511	if (a == 0)
512		return 0;
513	/* cerify that result of division matches second operand */
514	if (r / a != b)
515		return 1;
516	return 0;
517}
518
519struct val *
520op_times(struct val *a, struct val *b)
521{
522	struct val *r;
523
524	if (!to_integer (a) || !to_integer (b)) {
525		errx(ERR_EXIT, "non-numeric argument");
526	}
527
528	r = make_integer (/*(intmax_t)*/(a->u.i * b->u.i));
529	if (chk_times (a->u.i, b->u.i, r->u.i)) {
530		errx(ERR_EXIT, "overflow");
531	}
532	free_value (a);
533	free_value (b);
534	return (r);
535}
536
537int
538chk_div(intmax_t a, intmax_t b)
539{
540	/* div by zero has been taken care of before */
541	/* only INTMAX_MIN / -1 causes overflow */
542	if (a == INTMAX_MIN && b == -1)
543		return 1;
544	/* everything else is OK */
545	return 0;
546}
547
548struct val *
549op_div(struct val *a, struct val *b)
550{
551	struct val *r;
552
553	if (!to_integer (a) || !to_integer (b)) {
554		errx(ERR_EXIT, "non-numeric argument");
555	}
556
557	if (b->u.i == 0) {
558		errx(ERR_EXIT, "division by zero");
559	}
560
561	r = make_integer (/*(intmax_t)*/(a->u.i / b->u.i));
562	if (chk_div (a->u.i, b->u.i)) {
563		errx(ERR_EXIT, "overflow");
564	}
565	free_value (a);
566	free_value (b);
567	return r;
568}
569
570struct val *
571op_rem(struct val *a, struct val *b)
572{
573	struct val *r;
574
575	if (!to_integer (a) || !to_integer (b)) {
576		errx(ERR_EXIT, "non-numeric argument");
577	}
578
579	if (b->u.i == 0) {
580		errx(ERR_EXIT, "division by zero");
581	}
582
583	r = make_integer (/*(intmax_t)*/(a->u.i % b->u.i));
584	/* chk_rem necessary ??? */
585	free_value (a);
586	free_value (b);
587	return r;
588}
589
590struct val *
591op_colon(struct val *a, struct val *b)
592{
593	regex_t rp;
594	regmatch_t rm[2];
595	char errbuf[256];
596	int eval;
597	struct val *v;
598
599	/* coerce to both arguments to strings */
600	to_string(a);
601	to_string(b);
602
603	/* compile regular expression */
604	if ((eval = regcomp (&rp, b->u.s, 0)) != 0) {
605		regerror (eval, &rp, errbuf, sizeof(errbuf));
606		errx(ERR_EXIT, "%s", errbuf);
607	}
608
609	/* compare string against pattern */
610	/* remember that patterns are anchored to the beginning of the line */
611	if (regexec(&rp, a->u.s, (size_t)2, rm, 0) == 0 && rm[0].rm_so == 0) {
612		if (rm[1].rm_so >= 0) {
613			*(a->u.s + rm[1].rm_eo) = '\0';
614			v = make_str (a->u.s + rm[1].rm_so);
615
616		} else {
617			v = make_integer ((intmax_t)(rm[0].rm_eo - rm[0].rm_so));
618		}
619	} else {
620		if (rp.re_nsub == 0) {
621			v = make_integer ((intmax_t)0);
622		} else {
623			v = make_str ("");
624		}
625	}
626
627	/* free arguments and pattern buffer */
628	free_value (a);
629	free_value (b);
630	regfree (&rp);
631
632	return v;
633}
634