expr.y revision 96367 
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 96367 2002-05-10 22:59:29Z 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
73static int	eflag;
74char **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
94start: expr { result = $$; }
95
96expr:	TOKEN
97	| '(' expr ')' { $$ = $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
117struct val *
118make_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
132struct val *
133make_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
172void
173free_value(struct val *vp)
174{
175	if (vp->type == string || vp->type == numeric_string)
176		free (vp->u.s);
177}
178
179
180intmax_t
181to_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
207void
208to_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
231int
232isstring(struct val *vp)
233{
234	/* only TRUE if this string is not a valid integer */
235	return (vp->type == string);
236}
237
238
239int
240yylex(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
264int
265is_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
275int
276main(int argc, char *argv[])
277{
278	int c;
279
280	setlocale (LC_ALL, "");
281	if (getenv("EXPR_COMPAT") != NULL) {
282		av = argv + 1;
283	} else {
284		while ((c = getopt(argc, argv, "e")) != -1)
285			switch (c) {
286			case 'e':
287				eflag = 1;
288				break;
289
290			default:
291				fprintf(stderr,
292				    "usage: expr [-e] expression\n");
293				exit(ERR_EXIT);
294			}
295		av = argv + optind;
296	}
297
298	yyparse();
299
300	if (result->type == integer)
301		printf("%jd\n", result->u.i);
302	else
303		printf("%s\n", result->u.s);
304
305	return (is_zero_or_null(result));
306}
307
308int
309yyerror(const char *s __unused)
310{
311	errx(ERR_EXIT, "syntax error");
312}
313
314
315struct val *
316op_or(struct val *a, struct val *b)
317{
318	if (is_zero_or_null (a)) {
319		free_value (a);
320		return (b);
321	} else {
322		free_value (b);
323		return (a);
324	}
325}
326
327struct val *
328op_and(struct val *a, struct val *b)
329{
330	if (is_zero_or_null (a) || is_zero_or_null (b)) {
331		free_value (a);
332		free_value (b);
333		return (make_integer ((intmax_t)0));
334	} else {
335		free_value (b);
336		return (a);
337	}
338}
339
340struct val *
341op_eq(struct val *a, struct val *b)
342{
343	struct val *r;
344
345	if (isstring (a) || isstring (b)) {
346		to_string (a);
347		to_string (b);
348		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) == 0));
349	} else {
350		(void)to_integer(a);
351		(void)to_integer(b);
352		r = make_integer ((intmax_t)(a->u.i == b->u.i));
353	}
354
355	free_value (a);
356	free_value (b);
357	return r;
358}
359
360struct val *
361op_gt(struct val *a, struct val *b)
362{
363	struct val *r;
364
365	if (isstring (a) || isstring (b)) {
366		to_string (a);
367		to_string (b);
368		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) > 0));
369	} else {
370		(void)to_integer(a);
371		(void)to_integer(b);
372		r = make_integer ((intmax_t)(a->u.i > b->u.i));
373	}
374
375	free_value (a);
376	free_value (b);
377	return r;
378}
379
380struct val *
381op_lt(struct val *a, struct val *b)
382{
383	struct val *r;
384
385	if (isstring (a) || isstring (b)) {
386		to_string (a);
387		to_string (b);
388		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) < 0));
389	} else {
390		(void)to_integer(a);
391		(void)to_integer(b);
392		r = make_integer ((intmax_t)(a->u.i < b->u.i));
393	}
394
395	free_value (a);
396	free_value (b);
397	return r;
398}
399
400struct val *
401op_ge(struct val *a, struct val *b)
402{
403	struct val *r;
404
405	if (isstring (a) || isstring (b)) {
406		to_string (a);
407		to_string (b);
408		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) >= 0));
409	} else {
410		(void)to_integer(a);
411		(void)to_integer(b);
412		r = make_integer ((intmax_t)(a->u.i >= b->u.i));
413	}
414
415	free_value (a);
416	free_value (b);
417	return r;
418}
419
420struct val *
421op_le(struct val *a, struct val *b)
422{
423	struct val *r;
424
425	if (isstring (a) || isstring (b)) {
426		to_string (a);
427		to_string (b);
428		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) <= 0));
429	} else {
430		(void)to_integer(a);
431		(void)to_integer(b);
432		r = make_integer ((intmax_t)(a->u.i <= b->u.i));
433	}
434
435	free_value (a);
436	free_value (b);
437	return r;
438}
439
440struct val *
441op_ne(struct val *a, struct val *b)
442{
443	struct val *r;
444
445	if (isstring (a) || isstring (b)) {
446		to_string (a);
447		to_string (b);
448		r = make_integer ((intmax_t)(strcoll (a->u.s, b->u.s) != 0));
449	} else {
450		(void)to_integer(a);
451		(void)to_integer(b);
452		r = make_integer ((intmax_t)(a->u.i != b->u.i));
453	}
454
455	free_value (a);
456	free_value (b);
457	return r;
458}
459
460int
461chk_plus(intmax_t a, intmax_t b, intmax_t r)
462{
463
464	/* sum of two positive numbers must be positive */
465	if (a > 0 && b > 0 && r <= 0)
466		return 1;
467	/* sum of two negative numbers must be negative */
468	if (a < 0 && b < 0 && r >= 0)
469		return 1;
470	/* all other cases are OK */
471	return 0;
472}
473
474struct val *
475op_plus(struct val *a, struct val *b)
476{
477	struct val *r;
478
479	if (!to_integer(a) || !to_integer(b)) {
480		errx(ERR_EXIT, "non-numeric argument");
481	}
482
483	if (eflag) {
484		r = make_integer(a->u.i + b->u.i);
485		if (chk_plus(a->u.i, b->u.i, r->u.i)) {
486			errx(ERR_EXIT, "overflow");
487		}
488	} else
489		r = make_integer((long)a->u.i + (long)b->u.i);
490
491	free_value (a);
492	free_value (b);
493	return r;
494}
495
496int
497chk_minus(intmax_t a, intmax_t b, intmax_t r)
498{
499
500	/* special case subtraction of INTMAX_MIN */
501	if (b == INTMAX_MIN) {
502		if (a >= 0)
503			return 1;
504		else
505			return 0;
506	}
507	/* this is allowed for b != INTMAX_MIN */
508	return chk_plus (a, -b, r);
509}
510
511struct val *
512op_minus(struct val *a, struct val *b)
513{
514	struct val *r;
515
516	if (!to_integer(a) || !to_integer(b)) {
517		errx(ERR_EXIT, "non-numeric argument");
518	}
519
520	if (eflag) {
521		r = make_integer(a->u.i - b->u.i);
522		if (chk_minus(a->u.i, b->u.i, r->u.i)) {
523			errx(ERR_EXIT, "overflow");
524		}
525	} else
526		r = make_integer((long)a->u.i - (long)b->u.i);
527
528	free_value (a);
529	free_value (b);
530	return r;
531}
532
533int
534chk_times(intmax_t a, intmax_t b, intmax_t r)
535{
536	/* special case: first operand is 0, no overflow possible */
537	if (a == 0)
538		return 0;
539	/* cerify that result of division matches second operand */
540	if (r / a != b)
541		return 1;
542	return 0;
543}
544
545struct val *
546op_times(struct val *a, struct val *b)
547{
548	struct val *r;
549
550	if (!to_integer(a) || !to_integer(b)) {
551		errx(ERR_EXIT, "non-numeric argument");
552	}
553
554	if (eflag) {
555		r = make_integer(a->u.i * b->u.i);
556		if (chk_times(a->u.i, b->u.i, r->u.i)) {
557			errx(ERR_EXIT, "overflow");
558		}
559	} else
560		r = make_integer((long)a->u.i * (long)b->u.i);
561
562	free_value (a);
563	free_value (b);
564	return (r);
565}
566
567int
568chk_div(intmax_t a, intmax_t b)
569{
570	/* div by zero has been taken care of before */
571	/* only INTMAX_MIN / -1 causes overflow */
572	if (a == INTMAX_MIN && b == -1)
573		return 1;
574	/* everything else is OK */
575	return 0;
576}
577
578struct val *
579op_div(struct val *a, struct val *b)
580{
581	struct val *r;
582
583	if (!to_integer(a) || !to_integer(b)) {
584		errx(ERR_EXIT, "non-numeric argument");
585	}
586
587	if (b->u.i == 0) {
588		errx(ERR_EXIT, "division by zero");
589	}
590
591	if (eflag) {
592		r = make_integer(a->u.i / b->u.i);
593		if (chk_div(a->u.i, b->u.i)) {
594			errx(ERR_EXIT, "overflow");
595		}
596	} else
597		r = make_integer((long)a->u.i / (long)b->u.i);
598
599	free_value (a);
600	free_value (b);
601	return r;
602}
603
604struct val *
605op_rem(struct val *a, struct val *b)
606{
607	struct val *r;
608
609	if (!to_integer(a) || !to_integer(b)) {
610		errx(ERR_EXIT, "non-numeric argument");
611	}
612
613	if (b->u.i == 0) {
614		errx(ERR_EXIT, "division by zero");
615	}
616
617	if (eflag)
618		r = make_integer(a->u.i % b->u.i);
619	        /* chk_rem necessary ??? */
620	else
621		r = make_integer((long)a->u.i % (long)b->u.i);
622
623	free_value (a);
624	free_value (b);
625	return r;
626}
627
628struct val *
629op_colon(struct val *a, struct val *b)
630{
631	regex_t rp;
632	regmatch_t rm[2];
633	char errbuf[256];
634	int eval;
635	struct val *v;
636
637	/* coerce to both arguments to strings */
638	to_string(a);
639	to_string(b);
640
641	/* compile regular expression */
642	if ((eval = regcomp (&rp, b->u.s, 0)) != 0) {
643		regerror (eval, &rp, errbuf, sizeof(errbuf));
644		errx(ERR_EXIT, "%s", errbuf);
645	}
646
647	/* compare string against pattern */
648	/* remember that patterns are anchored to the beginning of the line */
649	if (regexec(&rp, a->u.s, (size_t)2, rm, 0) == 0 && rm[0].rm_so == 0) {
650		if (rm[1].rm_so >= 0) {
651			*(a->u.s + rm[1].rm_eo) = '\0';
652			v = make_str (a->u.s + rm[1].rm_so);
653
654		} else {
655			v = make_integer ((intmax_t)(rm[0].rm_eo - rm[0].rm_so));
656		}
657	} else {
658		if (rp.re_nsub == 0) {
659			v = make_integer ((intmax_t)0);
660		} else {
661			v = make_str ("");
662		}
663	}
664
665	/* free arguments and pattern buffer */
666	free_value (a);
667	free_value (b);
668	regfree (&rp);
669
670	return v;
671}
672