xref: /freebsd/lib/libc/stdlib/strfmon.c (revision cf85e7034ad5640b18a3b68d6b291b7bf89bfc80)
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause
3  *
4  * Copyright (c) 2001 Alexey Zelkin <phantom@FreeBSD.org>
5  * All rights reserved.
6  *
7  * Copyright (c) 2011 The FreeBSD Foundation
8  *
9  * Portions of this software were developed by David Chisnall
10  * under sponsorship from the FreeBSD Foundation.
11  *
12  * Redistribution and use in source and binary forms, with or without
13  * modification, are permitted provided that the following conditions
14  * are met:
15  * 1. Redistributions of source code must retain the above copyright
16  *    notice, this list of conditions and the following disclaimer.
17  * 2. Redistributions in binary form must reproduce the above copyright
18  *    notice, this list of conditions and the following disclaimer in the
19  *    documentation and/or other materials provided with the distribution.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR AND CONTRIBUTORS ``AS IS'' AND
22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR OR CONTRIBUTORS BE LIABLE
25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31  * SUCH DAMAGE.
32  *
33  */
34 
35 #include <sys/types.h>
36 
37 #include <ctype.h>
38 #include <errno.h>
39 #include <limits.h>
40 #include <locale.h>
41 #include <monetary.h>
42 #include <stdarg.h>
43 #include <stdio.h>
44 #include <stdlib.h>
45 #include <string.h>
46 
47 #include "xlocale_private.h"
48 
49 /* internal flags */
50 #define	NEED_GROUPING		0x01	/* print digits grouped (default) */
51 #define	SIGN_POSN_USED		0x02	/* '+' or '(' usage flag */
52 #define	LOCALE_POSN		0x04	/* use locale defined +/- (default) */
53 #define	PARENTH_POSN		0x08	/* enclose negative amount in () */
54 #define	SUPPRESS_CURR_SYMBOL	0x10	/* suppress the currency from output */
55 #define	LEFT_JUSTIFY		0x20	/* left justify */
56 #define	USE_INTL_CURRENCY	0x40	/* use international currency symbol */
57 #define	IS_NEGATIVE		0x80	/* is argument value negative ? */
58 
59 /* internal macros */
60 #define	PRINT(CH) do {						\
61 	if (dst >= s + maxsize)					\
62 		goto e2big_error;				\
63 	*dst++ = CH;						\
64 } while (0)
65 
66 #define	PRINTS(STR) do {					\
67 	char *tmps = STR;					\
68 	while (*tmps != '\0')					\
69 		PRINT(*tmps++);					\
70 } while (0)
71 
72 #define	GET_NUMBER(VAR, LOC) do {				\
73 	VAR = 0;						\
74 	while (isdigit_l((unsigned char)*fmt, LOC)) {		\
75 		if (VAR > INT_MAX / 10)				\
76 			goto e2big_error;			\
77 		VAR *= 10;					\
78 		VAR += *fmt - '0';				\
79 		if (VAR < 0)					\
80 			goto e2big_error;			\
81 		fmt++;						\
82 	}							\
83 } while (0)
84 
85 #define	GRPCPY(howmany) do {					\
86 	int i = howmany;					\
87 	while (i-- > 0) {					\
88 		avalue_size--;					\
89 		*--bufend = *(avalue + avalue_size + padded);	\
90 	}							\
91 } while (0)
92 
93 #define	GRPSEP do {						\
94 	bufend -= thousands_sep_size;				\
95 	memcpy(bufend, thousands_sep, thousands_sep_size);	\
96 	groups++;						\
97 } while (0)
98 
99 static void __setup_vars(int, char *, char *, char *, char **, struct lconv *);
100 static int __calc_left_pad(int, char *, struct lconv *);
101 static char *__format_grouped_double(double, int *, int, int, int,
102     struct lconv *, locale_t);
103 
104 static ssize_t
vstrfmon_l(char * __restrict s,size_t maxsize,locale_t loc,const char * __restrict format,va_list ap)105 vstrfmon_l(char *__restrict s, size_t maxsize, locale_t loc,
106     const char *__restrict format, va_list ap)
107 {
108 	char		*dst;		/* output destination pointer */
109 	const char	*fmt;		/* current format position pointer */
110 	struct lconv	*lc;		/* pointer to lconv structure */
111 	char		*asciivalue;	/* formatted double pointer */
112 
113 	int		flags;			/* formatting options */
114 	int		pad_char;		/* padding character */
115 	int		pad_size;		/* pad size */
116 	int		width;			/* field width */
117 	int		left_prec;		/* left precision */
118 	int		right_prec;		/* right precision */
119 	double		value;			/* just value */
120 	char		space_char = ' ';	/* space after currency */
121 
122 	char		cs_precedes,	/* values gathered from struct lconv */
123 			sep_by_space,
124 			sign_posn,
125 			*signstr,
126 			*currency_symbol;
127 
128 	char		*tmpptr;	/* temporary vars */
129 	int		sverrno;
130 
131 	FIX_LOCALE(loc);
132 
133 	lc = localeconv_l(loc);
134 	dst = s;
135 	fmt = format;
136 	asciivalue = NULL;
137 	currency_symbol = NULL;
138 
139 	while (*fmt != 0) {
140 		/* pass nonformating characters AS IS */
141 		if (*fmt != '%')
142 			goto literal;
143 
144 		/* '%' found ! */
145 
146 		/* "%%" mean just '%' */
147 		if (*(fmt + 1) == '%') {
148 			fmt++;
149 literal:
150 			PRINT(*fmt++);
151 			continue;
152 		}
153 
154 		/* set up initial values */
155 		flags = NEED_GROUPING | LOCALE_POSN;
156 		pad_char = ' ';		/* padding character is "space" */
157 		pad_size = 0;		/* no padding initially */
158 		left_prec = -1;		/* no left precision specified */
159 		right_prec = -1;	/* no right precision specified */
160 		width = -1;		/* no width specified */
161 
162 		/* Flags */
163 		while (1) {
164 			switch (*++fmt) {
165 			case '=':	/* fill character */
166 				pad_char = *++fmt;
167 				if (pad_char == '\0')
168 					goto format_error;
169 				continue;
170 			case '^':	/* not group currency */
171 				flags &= ~(NEED_GROUPING);
172 				continue;
173 			case '+':	/* use locale defined signs */
174 				if ((flags & SIGN_POSN_USED) ||
175 				    ((lc->positive_sign[0] == '\0') &&
176 				    (lc->negative_sign[0] == '\0')))
177 					goto format_error;
178 				flags |= (SIGN_POSN_USED | LOCALE_POSN);
179 				continue;
180 			case '(':	/* enclose negatives with () */
181 				if (flags & SIGN_POSN_USED)
182 					goto format_error;
183 				flags |= (SIGN_POSN_USED | PARENTH_POSN);
184 				continue;
185 			case '!':	/* suppress currency symbol */
186 				flags |= SUPPRESS_CURR_SYMBOL;
187 				continue;
188 			case '-':	/* alignment (left) */
189 				flags |= LEFT_JUSTIFY;
190 				continue;
191 			default:
192 				break;
193 			}
194 			break;
195 		}
196 
197 		/* field Width */
198 		if (isdigit_l((unsigned char)*fmt, loc)) {
199 			GET_NUMBER(width, loc);
200 			/*
201 			 * Do we have enough space to put number with
202 			 * required width ?
203 			 */
204 			if ((unsigned int)width >= maxsize - (dst - s))
205 				goto e2big_error;
206 		}
207 
208 		/* Left precision */
209 		if (*fmt == '#') {
210 			if (!isdigit_l((unsigned char)*++fmt, loc))
211 				goto format_error;
212 			GET_NUMBER(left_prec, loc);
213 			if ((unsigned int)left_prec >= maxsize - (dst - s))
214 				goto e2big_error;
215 		}
216 
217 		/* Right precision */
218 		if (*fmt == '.') {
219 			if (!isdigit_l((unsigned char)*++fmt, loc))
220 				goto format_error;
221 			GET_NUMBER(right_prec, loc);
222 			if ((unsigned int)right_prec >=
223 			    maxsize - (dst - s) - left_prec)
224 				goto e2big_error;
225 		}
226 
227 		/* Conversion Characters */
228 		switch (*fmt++) {
229 		case 'i':	/* use international currency format */
230 			flags |= USE_INTL_CURRENCY;
231 			break;
232 		case 'n':	/* use national currency format */
233 			flags &= ~(USE_INTL_CURRENCY);
234 			break;
235 		default:	/*
236 				 * required character is missing or
237 				 * premature EOS
238 				 */
239 			goto format_error;
240 		}
241 
242 		if (currency_symbol != NULL)
243 			free(currency_symbol);
244 		if (flags & USE_INTL_CURRENCY) {
245 			currency_symbol = strdup(lc->int_curr_symbol);
246 			if (currency_symbol != NULL &&
247 			    strlen(currency_symbol) > 3) {
248 				space_char = currency_symbol[3];
249 				currency_symbol[3] = '\0';
250 			}
251 		} else
252 			currency_symbol = strdup(lc->currency_symbol);
253 
254 		if (currency_symbol == NULL)
255 			goto end_error;	/* ENOMEM. */
256 
257 		/* value itself */
258 		value = va_arg(ap, double);
259 
260 		/* detect sign */
261 		if (value < 0) {
262 			flags |= IS_NEGATIVE;
263 			value = -value;
264 		}
265 
266 		/* fill left_prec with amount of padding chars */
267 		if (left_prec >= 0) {
268 			pad_size = __calc_left_pad((flags ^ IS_NEGATIVE),
269 			    currency_symbol, lc) -
270 			    __calc_left_pad(flags, currency_symbol, lc);
271 			if (pad_size < 0)
272 				pad_size = 0;
273 		}
274 
275 		if (asciivalue != NULL)
276 			free(asciivalue);
277 		asciivalue = __format_grouped_double(value, &flags, left_prec,
278 		    right_prec, pad_char, lc, loc);
279 		if (asciivalue == NULL)
280 			goto end_error;	/*
281 					 * errno already set to ENOMEM by
282 					 * malloc()
283 					 */
284 
285 		/* set some variables for later use */
286 		__setup_vars(flags, &cs_precedes, &sep_by_space, &sign_posn,
287 		    &signstr, lc);
288 
289 		/*
290 		 * Description of some LC_MONETARY's values:
291 		 *
292 		 * p_cs_precedes & n_cs_precedes
293 		 *
294 		 * = 1 - $currency_symbol precedes the value
295 		 *       for a monetary quantity with a non-negative value
296 		 * = 0 - symbol succeeds the value
297 		 *
298 		 * p_sep_by_space & n_sep_by_space
299 		 *
300 		 * = 0 - no space separates $currency_symbol
301 		 *       from the value for a monetary quantity with a
302 		 *       non-negative value
303 		 * = 1 - space separates the symbol from the value
304 		 * = 2 - space separates the symbol and the sign string,
305 		 *       if adjacent; otherwise, a space separates
306 		 *       the sign string from the value
307 		 *
308 		 * p_sign_posn & n_sign_posn
309 		 *
310 		 * = 0 - parentheses enclose the quantity and the
311 		 *       $currency_symbol
312 		 * = 1 - the sign string precedes the quantity and the
313 		 *       $currency_symbol
314 		 * = 2 - the sign string succeeds the quantity and the
315 		 *       $currency_symbol
316 		 * = 3 - the sign string precedes the $currency_symbol
317 		 * = 4 - the sign string succeeds the $currency_symbol
318 		 */
319 
320 		tmpptr = dst;
321 
322 		while (pad_size-- > 0)
323 			PRINT(' ');
324 
325 		if (sign_posn == 0 && (flags & IS_NEGATIVE))
326 			PRINT('(');
327 
328 		if (cs_precedes == 1) {
329 			if (sign_posn == 1 || sign_posn == 3) {
330 				PRINTS(signstr);
331 				if (sep_by_space == 2)
332 					PRINT(' ');
333 			}
334 
335 			if (!(flags & SUPPRESS_CURR_SYMBOL)) {
336 				PRINTS(currency_symbol);
337 
338 				if (sign_posn == 4) {
339 					if (sep_by_space == 2)
340 						PRINT(space_char);
341 					PRINTS(signstr);
342 					if (sep_by_space == 1)
343 						PRINT(' ');
344 				} else if (sep_by_space == 1)
345 					PRINT(space_char);
346 			}
347 		} else if (sign_posn == 1) {
348 			PRINTS(signstr);
349 			if (sep_by_space == 2)
350 				PRINT(' ');
351 		}
352 
353 		PRINTS(asciivalue);
354 
355 		if (cs_precedes == 0) {
356 			if (sign_posn == 3) {
357 				if (sep_by_space == 1)
358 					PRINT(' ');
359 				PRINTS(signstr);
360 			}
361 
362 			if (!(flags & SUPPRESS_CURR_SYMBOL)) {
363 				if ((sign_posn == 3 && sep_by_space == 2) ||
364 				    (sep_by_space == 1 &&
365 				     (sign_posn == 0 || sign_posn == 1 ||
366 				      sign_posn == 2 || sign_posn == 4)))
367 					PRINT(space_char);
368 				PRINTS(currency_symbol);
369 				if (sign_posn == 4) {
370 					if (sep_by_space == 2)
371 						PRINT(' ');
372 					PRINTS(signstr);
373 				}
374 			}
375 		}
376 
377 		if (sign_posn == 2) {
378 			if (sep_by_space == 2)
379 				PRINT(' ');
380 			PRINTS(signstr);
381 		}
382 
383 		if (sign_posn == 0) {
384 			if (flags & IS_NEGATIVE)
385 				PRINT(')');
386 			else if (left_prec >= 0)
387 				PRINT(' ');
388 		}
389 
390 		if (dst - tmpptr < width) {
391 			if (flags & LEFT_JUSTIFY) {
392 				while (dst - tmpptr < width)
393 					PRINT(' ');
394 			} else {
395 				pad_size = dst - tmpptr;
396 				memmove(tmpptr + width - pad_size, tmpptr,
397 				    pad_size);
398 				memset(tmpptr, ' ', width - pad_size);
399 				dst += width - pad_size;
400 			}
401 		}
402 	}
403 
404 	PRINT('\0');
405 	free(asciivalue);
406 	free(currency_symbol);
407 	return (dst - s - 1);	/* size of put data except trailing '\0' */
408 
409 e2big_error:
410 	errno = E2BIG;
411 	goto end_error;
412 
413 format_error:
414 	errno = EINVAL;
415 
416 end_error:
417 	sverrno = errno;
418 	if (asciivalue != NULL)
419 		free(asciivalue);
420 	if (currency_symbol != NULL)
421 		free(currency_symbol);
422 	errno = sverrno;
423 	return (-1);
424 }
425 
426 static void
__setup_vars(int flags,char * cs_precedes,char * sep_by_space,char * sign_posn,char ** signstr,struct lconv * lc)427 __setup_vars(int flags, char *cs_precedes, char *sep_by_space, char *sign_posn,
428     char **signstr, struct lconv *lc)
429 {
430 	if ((flags & IS_NEGATIVE) && (flags & USE_INTL_CURRENCY)) {
431 		*cs_precedes = lc->int_n_cs_precedes;
432 		*sep_by_space = lc->int_n_sep_by_space;
433 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->int_n_sign_posn;
434 		*signstr = (lc->negative_sign[0] == '\0') ? "-" :
435 		    lc->negative_sign;
436 	} else if (flags & USE_INTL_CURRENCY) {
437 		*cs_precedes = lc->int_p_cs_precedes;
438 		*sep_by_space = lc->int_p_sep_by_space;
439 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->int_p_sign_posn;
440 		*signstr = lc->positive_sign;
441 	} else if (flags & IS_NEGATIVE) {
442 		*cs_precedes = lc->n_cs_precedes;
443 		*sep_by_space = lc->n_sep_by_space;
444 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->n_sign_posn;
445 		*signstr = (lc->negative_sign[0] == '\0') ? "-" :
446 		    lc->negative_sign;
447 	} else {
448 		*cs_precedes = lc->p_cs_precedes;
449 		*sep_by_space = lc->p_sep_by_space;
450 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->p_sign_posn;
451 		*signstr = lc->positive_sign;
452 	}
453 
454 	/* Set default values for unspecified information. */
455 	if (*cs_precedes != 0)
456 		*cs_precedes = 1;
457 	if (*sep_by_space == CHAR_MAX)
458 		*sep_by_space = 0;
459 	if (*sign_posn == CHAR_MAX)
460 		*sign_posn = 1;
461 }
462 
463 static int
__calc_left_pad(int flags,char * cur_symb,struct lconv * lc)464 __calc_left_pad(int flags, char *cur_symb, struct lconv *lc)
465 {
466 	char cs_precedes, sep_by_space, sign_posn, *signstr;
467 	int left_chars = 0;
468 
469 	__setup_vars(flags, &cs_precedes, &sep_by_space, &sign_posn, &signstr,
470 	    lc);
471 
472 	if (cs_precedes != 0) {
473 		left_chars += strlen(cur_symb);
474 		if (sep_by_space != 0)
475 			left_chars++;
476 	}
477 
478 	switch (sign_posn) {
479 	case 0:
480 		if (flags & IS_NEGATIVE)
481 			left_chars++;
482 		break;
483 	case 1:
484 		left_chars += strlen(signstr);
485 		break;
486 	case 3:
487 	case 4:
488 		if (cs_precedes != 0)
489 			left_chars += strlen(signstr);
490 	}
491 	return (left_chars);
492 }
493 
494 static int
get_groups(int size,char * grouping)495 get_groups(int size, char *grouping)
496 {
497 	int	chars = 0;
498 
499 	if (*grouping == CHAR_MAX || *grouping <= 0)	/* no grouping ? */
500 		return (0);
501 
502 	while (size > (int)*grouping) {
503 		chars++;
504 		size -= (int)*grouping++;
505 		/* no more grouping ? */
506 		if (*grouping == CHAR_MAX)
507 			break;
508 		/* rest grouping with same value ? */
509 		if (*grouping == 0) {
510 			chars += (size - 1) / *(grouping - 1);
511 			break;
512 		}
513 	}
514 	return (chars);
515 }
516 
517 /* convert double to locale-encoded string */
518 static char *
__format_grouped_double(double value,int * flags,int left_prec,int right_prec,int pad_char,struct lconv * lc,locale_t loc)519 __format_grouped_double(double value, int *flags, int left_prec, int right_prec,
520     int pad_char, struct lconv *lc, locale_t loc)
521 {
522 
523 	char		*rslt;
524 	char		*avalue;
525 	int		avalue_size;
526 
527 	size_t		bufsize;
528 	char		*bufend;
529 
530 	int		padded;
531 
532 	char		*grouping;
533 	const char	*decimal_point;
534 	const char	*thousands_sep;
535 	size_t		decimal_point_size;
536 	size_t		thousands_sep_size;
537 
538 	int		groups = 0;
539 
540 	grouping = lc->mon_grouping;
541 	decimal_point = lc->mon_decimal_point;
542 	if (*decimal_point == '\0')
543 		decimal_point = lc->decimal_point;
544 	thousands_sep = lc->mon_thousands_sep;
545 	if (*thousands_sep == '\0')
546 		thousands_sep = lc->thousands_sep;
547 
548 	decimal_point_size = strlen(decimal_point);
549 	thousands_sep_size = strlen(thousands_sep);
550 
551 	/* fill left_prec with default value */
552 	if (left_prec == -1)
553 		left_prec = 0;
554 
555 	/* fill right_prec with default value */
556 	if (right_prec == -1) {
557 		if (*flags & USE_INTL_CURRENCY)
558 			right_prec = lc->int_frac_digits;
559 		else
560 			right_prec = lc->frac_digits;
561 
562 		if (right_prec == CHAR_MAX)	/* POSIX locale ? */
563 			right_prec = 2;
564 	}
565 
566 	if (*flags & NEED_GROUPING)
567 		left_prec += get_groups(left_prec, grouping);
568 
569 	/* convert to string */
570 	avalue_size = asprintf_l(&avalue, loc, "%*.*f",
571 	    left_prec + right_prec + 1, right_prec, value);
572 	if (avalue_size < 0)
573 		return (NULL);
574 
575 	/* make sure that we've enough space for result string */
576 	bufsize = avalue_size * (1 + thousands_sep_size) + decimal_point_size +
577 	    1;
578 	rslt = calloc(1, bufsize);
579 	if (rslt == NULL) {
580 		free(avalue);
581 		return (NULL);
582 	}
583 	bufend = rslt + bufsize - 1;	/* reserve space for trailing '\0' */
584 
585 	/* skip spaces at beginning */
586 	padded = 0;
587 	while (avalue[padded] == ' ') {
588 		padded++;
589 		avalue_size--;
590 	}
591 
592 	if (right_prec > 0) {
593 		bufend -= right_prec;
594 		memcpy(bufend, avalue + avalue_size + padded - right_prec,
595 		    right_prec);
596 		bufend -= decimal_point_size;
597 		memcpy(bufend, decimal_point, decimal_point_size);
598 		avalue_size -= (right_prec + 1);
599 	}
600 
601 	if ((*flags & NEED_GROUPING) && thousands_sep_size > 0 &&
602 	    *grouping != CHAR_MAX && *grouping > 0) {
603 		while (avalue_size > (int)*grouping) {
604 			GRPCPY(*grouping);
605 			GRPSEP;
606 			grouping++;
607 
608 			/* no more grouping ? */
609 			if (*grouping == CHAR_MAX)
610 				break;
611 
612 			/* rest grouping with same value ? */
613 			if (*grouping == 0) {
614 				grouping--;
615 				while (avalue_size > *grouping) {
616 					GRPCPY(*grouping);
617 					GRPSEP;
618 				}
619 			}
620 		}
621 		if (avalue_size != 0)
622 			GRPCPY(avalue_size);
623 		padded -= groups;
624 	} else {
625 		bufend -= avalue_size;
626 		memcpy(bufend, avalue + padded, avalue_size);
627 		/* decrease assumed $decimal_point */
628 		if (right_prec == 0)
629 			padded -= decimal_point_size;
630 	}
631 
632 	/* do padding with pad_char */
633 	if (padded > 0) {
634 		bufend -= padded;
635 		memset(bufend, pad_char, padded);
636 	}
637 
638 	bufsize = rslt + bufsize - bufend;
639 	memmove(rslt, bufend, bufsize);
640 	free(avalue);
641 	return (rslt);
642 }
643 
644 ssize_t
strfmon(char * restrict s,size_t maxsize,const char * restrict format,...)645 strfmon(char *restrict s, size_t maxsize, const char *restrict format, ...)
646 {
647 	ssize_t ret;
648 	va_list ap;
649 
650 	va_start(ap, format);
651 	ret = vstrfmon_l(s, maxsize, __get_locale(), format, ap);
652 	va_end(ap);
653 
654 	return (ret);
655 }
656 
657 ssize_t
strfmon_l(char * restrict s,size_t maxsize,locale_t loc,const char * restrict format,...)658 strfmon_l(char *restrict s, size_t maxsize, locale_t loc,
659     const char *restrict format, ...)
660 {
661 	ssize_t ret;
662 	va_list ap;
663 
664 	va_start(ap, format);
665 	ret = vstrfmon_l(s, maxsize, loc, format, ap);
666 	va_end(ap);
667 
668 	return (ret);
669 }
670