xref: /freebsd/lib/libc/stdlib/strfmon.c (revision d0b2dbfa)
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/cdefs.h>
36 #include <sys/types.h>
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
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 poistion 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 	FIX_LOCALE(loc);
131 
132 	lc = localeconv_l(loc);
133 	dst = s;
134 	fmt = format;
135 	asciivalue = NULL;
136 	currency_symbol = NULL;
137 
138 	while (*fmt) {
139 		/* pass nonformating characters AS IS */
140 		if (*fmt != '%')
141 			goto literal;
142 
143 		/* '%' found ! */
144 
145 		/* "%%" mean just '%' */
146 		if (*(fmt + 1) == '%') {
147 			fmt++;
148 	literal:
149 			PRINT(*fmt++);
150 			continue;
151 		}
152 
153 		/* set up initial values */
154 		flags = (NEED_GROUPING|LOCALE_POSN);
155 		pad_char = ' ';		/* padding character is "space" */
156 		pad_size = 0;		/* no padding initially */
157 		left_prec = -1;		/* no left precision specified */
158 		right_prec = -1;	/* no right precision specified */
159 		width = -1;		/* no width specified */
160 		value = 0;		/* we have no value to print now */
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 						goto format_error;
176 					flags |= (SIGN_POSN_USED|LOCALE_POSN);
177 					continue;
178 				case '(':	/* enclose negatives with () */
179 					if (flags & SIGN_POSN_USED)
180 						goto format_error;
181 					flags |= (SIGN_POSN_USED|PARENTH_POSN);
182 					continue;
183 				case '!':	/* suppress currency symbol */
184 					flags |= SUPPRESS_CURR_SYMBOL;
185 					continue;
186 				case '-':	/* alignment (left)  */
187 					flags |= LEFT_JUSTIFY;
188 					continue;
189 				default:
190 					break;
191 			}
192 			break;
193 		}
194 
195 		/* field Width */
196 		if (isdigit_l((unsigned char)*fmt, loc)) {
197 			GET_NUMBER(width, loc);
198 			/* Do we have enough space to put number with
199 			 * required width ?
200 			 */
201 			if ((unsigned int)width >= maxsize - (dst - s))
202 				goto e2big_error;
203 		}
204 
205 		/* Left precision */
206 		if (*fmt == '#') {
207 			if (!isdigit_l((unsigned char)*++fmt, loc))
208 				goto format_error;
209 			GET_NUMBER(left_prec, loc);
210 			if ((unsigned int)left_prec >= maxsize - (dst - s))
211 				goto e2big_error;
212 		}
213 
214 		/* Right precision */
215 		if (*fmt == '.') {
216 			if (!isdigit_l((unsigned char)*++fmt, loc))
217 				goto format_error;
218 			GET_NUMBER(right_prec, loc);
219 			if ((unsigned int)right_prec >= maxsize - (dst - s) -
220 			    left_prec)
221 				goto e2big_error;
222 		}
223 
224 		/* Conversion Characters */
225 		switch (*fmt++) {
226 			case 'i':	/* use international currency format */
227 				flags |= USE_INTL_CURRENCY;
228 				break;
229 			case 'n':	/* use national currency format */
230 				flags &= ~(USE_INTL_CURRENCY);
231 				break;
232 			default:	/* required character is missing or
233 					   premature EOS */
234 				goto format_error;
235 		}
236 
237 		if (currency_symbol != NULL)
238 			free(currency_symbol);
239 		if (flags & USE_INTL_CURRENCY) {
240 			currency_symbol = strdup(lc->int_curr_symbol);
241 			if (currency_symbol != NULL &&
242 			    strlen(currency_symbol) > 3) {
243 				space_char = currency_symbol[3];
244 				currency_symbol[3] = '\0';
245 			}
246 		} else
247 			currency_symbol = strdup(lc->currency_symbol);
248 
249 		if (currency_symbol == NULL)
250 			goto end_error;			/* ENOMEM. */
251 
252 		/* value itself */
253 		value = va_arg(ap, double);
254 
255 		/* detect sign */
256 		if (value < 0) {
257 			flags |= IS_NEGATIVE;
258 			value = -value;
259 		}
260 
261 		/* fill left_prec with amount of padding chars */
262 		if (left_prec >= 0) {
263 			pad_size = __calc_left_pad((flags ^ IS_NEGATIVE),
264 			    currency_symbol, lc) -
265 			    __calc_left_pad(flags, currency_symbol, lc);
266 			if (pad_size < 0)
267 				pad_size = 0;
268 		}
269 
270 		if (asciivalue != NULL)
271 			free(asciivalue);
272 		asciivalue = __format_grouped_double(value, &flags,
273 		    left_prec, right_prec, pad_char, lc, loc);
274 		if (asciivalue == NULL)
275 			goto end_error;		/* errno already set     */
276 						/* to ENOMEM by malloc() */
277 
278 		/* set some variables for later use */
279 		__setup_vars(flags, &cs_precedes, &sep_by_space,
280 		    &sign_posn, &signstr, lc);
281 
282 		/*
283 		 * Description of some LC_MONETARY's values:
284 		 *
285 		 * p_cs_precedes & n_cs_precedes
286 		 *
287 		 * = 1 - $currency_symbol precedes the value
288 		 *       for a monetary quantity with a non-negative value
289 		 * = 0 - symbol succeeds the value
290 		 *
291 		 * p_sep_by_space & n_sep_by_space
292 		 *
293 		 * = 0 - no space separates $currency_symbol
294 		 *       from the value for a monetary quantity with a
295 		 *       non-negative value
296 		 * = 1 - space separates the symbol from the value
297 		 * = 2 - space separates the symbol and the sign string,
298 		 *       if adjacent; otherwise, a space separates
299 		 *       the sign string from the value
300 		 *
301 		 * p_sign_posn & n_sign_posn
302 		 *
303 		 * = 0 - parentheses enclose the quantity and the
304 		 *       $currency_symbol
305 		 * = 1 - the sign string precedes the quantity and the
306 		 *       $currency_symbol
307 		 * = 2 - the sign string succeeds the quantity and the
308 		 *       $currency_symbol
309 		 * = 3 - the sign string precedes the $currency_symbol
310 		 * = 4 - the sign string succeeds the $currency_symbol
311 		 *
312 		 */
313 
314 		tmpptr = dst;
315 
316 		while (pad_size-- > 0)
317 			PRINT(' ');
318 
319 		if (sign_posn == 0 && (flags & IS_NEGATIVE))
320 			PRINT('(');
321 
322 		if (cs_precedes == 1) {
323 			if (sign_posn == 1 || sign_posn == 3) {
324 				PRINTS(signstr);
325 				if (sep_by_space == 2)
326 					PRINT(' ');
327 			}
328 
329 			if (!(flags & SUPPRESS_CURR_SYMBOL)) {
330 				PRINTS(currency_symbol);
331 
332 				if (sign_posn == 4) {
333 					if (sep_by_space == 2)
334 						PRINT(space_char);
335 					PRINTS(signstr);
336 					if (sep_by_space == 1)
337 						PRINT(' ');
338 				} else if (sep_by_space == 1)
339 					PRINT(space_char);
340 			}
341 		} else if (sign_posn == 1) {
342 			PRINTS(signstr);
343 			if (sep_by_space == 2)
344 				PRINT(' ');
345 		}
346 
347 		PRINTS(asciivalue);
348 
349 		if (cs_precedes == 0) {
350 			if (sign_posn == 3) {
351 				if (sep_by_space == 1)
352 					PRINT(' ');
353 				PRINTS(signstr);
354 			}
355 
356 			if (!(flags & SUPPRESS_CURR_SYMBOL)) {
357 				if ((sign_posn == 3 && sep_by_space == 2)
358 				    || (sep_by_space == 1
359 				    && (sign_posn == 0
360 				    || sign_posn == 1
361 				    || sign_posn == 2
362 				    || sign_posn == 4)))
363 					PRINT(space_char);
364 				PRINTS(currency_symbol);
365 				if (sign_posn == 4) {
366 					if (sep_by_space == 2)
367 						PRINT(' ');
368 					PRINTS(signstr);
369 				}
370 			}
371 		}
372 
373 		if (sign_posn == 2) {
374 			if (sep_by_space == 2)
375 				PRINT(' ');
376 			PRINTS(signstr);
377 		}
378 
379 		if (sign_posn == 0) {
380 			if (flags & IS_NEGATIVE)
381 				PRINT(')');
382 			else if (left_prec >= 0)
383 				PRINT(' ');
384 		}
385 
386 		if (dst - tmpptr < width) {
387 			if (flags & LEFT_JUSTIFY) {
388 				while (dst - tmpptr < width)
389 					PRINT(' ');
390 			} else {
391 				pad_size = dst - tmpptr;
392 				memmove(tmpptr + width - pad_size, tmpptr,
393 				    pad_size);
394 				memset(tmpptr, ' ', width - pad_size);
395 				dst += width - pad_size;
396 			}
397 		}
398 	}
399 
400 	PRINT('\0');
401 	free(asciivalue);
402 	free(currency_symbol);
403 	return (dst - s - 1);	/* return size of put data except trailing '\0' */
404 
405 e2big_error:
406 	errno = E2BIG;
407 	goto end_error;
408 
409 format_error:
410 	errno = EINVAL;
411 
412 end_error:
413 	sverrno = errno;
414 	if (asciivalue != NULL)
415 		free(asciivalue);
416 	if (currency_symbol != NULL)
417 		free(currency_symbol);
418 	errno = sverrno;
419 	return (-1);
420 }
421 
422 static void
423 __setup_vars(int flags, char *cs_precedes, char *sep_by_space,
424     char *sign_posn, char **signstr, struct lconv *lc)
425 {
426 	if ((flags & IS_NEGATIVE) && (flags & USE_INTL_CURRENCY)) {
427 		*cs_precedes = lc->int_n_cs_precedes;
428 		*sep_by_space = lc->int_n_sep_by_space;
429 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->int_n_sign_posn;
430 		*signstr = (lc->negative_sign[0] == '\0') ? "-"
431 		    : lc->negative_sign;
432 	} else if (flags & USE_INTL_CURRENCY) {
433 		*cs_precedes = lc->int_p_cs_precedes;
434 		*sep_by_space = lc->int_p_sep_by_space;
435 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->int_p_sign_posn;
436 		*signstr = lc->positive_sign;
437 	} else if (flags & IS_NEGATIVE) {
438 		*cs_precedes = lc->n_cs_precedes;
439 		*sep_by_space = lc->n_sep_by_space;
440 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->n_sign_posn;
441 		*signstr = (lc->negative_sign[0] == '\0') ? "-"
442 		    : lc->negative_sign;
443 	} else {
444 		*cs_precedes = lc->p_cs_precedes;
445 		*sep_by_space = lc->p_sep_by_space;
446 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->p_sign_posn;
447 		*signstr = lc->positive_sign;
448 	}
449 
450 	/* Set default values for unspecified information. */
451 	if (*cs_precedes != 0)
452 		*cs_precedes = 1;
453 	if (*sep_by_space == CHAR_MAX)
454 		*sep_by_space = 0;
455 	if (*sign_posn == CHAR_MAX)
456 		*sign_posn = 0;
457 }
458 
459 static int
460 __calc_left_pad(int flags, char *cur_symb, struct lconv *lc)
461 {
462 	char cs_precedes, sep_by_space, sign_posn, *signstr;
463 	int left_chars = 0;
464 
465 	__setup_vars(flags, &cs_precedes, &sep_by_space, &sign_posn,
466 	    &signstr, lc);
467 
468 	if (cs_precedes != 0) {
469 		left_chars += strlen(cur_symb);
470 		if (sep_by_space != 0)
471 			left_chars++;
472 	}
473 
474 	switch (sign_posn) {
475 		case 0:
476 			if (flags & IS_NEGATIVE)
477 				left_chars++;
478 			break;
479 		case 1:
480 			left_chars += strlen(signstr);
481 			break;
482 		case 3:
483 		case 4:
484 			if (cs_precedes != 0)
485 				left_chars += strlen(signstr);
486 	}
487 	return (left_chars);
488 }
489 
490 static int
491 get_groups(int size, char *grouping)
492 {
493 	int	chars = 0;
494 
495 	if (*grouping == CHAR_MAX || *grouping <= 0)	/* no grouping ? */
496 		return (0);
497 
498 	while (size > (int)*grouping) {
499 		chars++;
500 		size -= (int)*grouping++;
501 		/* no more grouping ? */
502 		if (*grouping == CHAR_MAX)
503 			break;
504 		/* rest grouping with same value ? */
505 		if (*grouping == 0) {
506 			chars += (size - 1) / *(grouping - 1);
507 			break;
508 		}
509 	}
510 	return (chars);
511 }
512 
513 /* convert double to locale-encoded string */
514 static char *
515 __format_grouped_double(double value, int *flags,
516     int left_prec, int right_prec, int pad_char, struct lconv *lc, locale_t loc)
517 {
518 
519 	char		*rslt;
520 	char		*avalue;
521 	int		avalue_size;
522 
523 	size_t		bufsize;
524 	char		*bufend;
525 
526 	int		padded;
527 
528 	char		*grouping;
529 	const char	*decimal_point;
530 	const char	*thousands_sep;
531 	size_t		decimal_point_size;
532 	size_t		thousands_sep_size;
533 
534 	int groups = 0;
535 
536 	grouping = lc->mon_grouping;
537 	decimal_point = lc->mon_decimal_point;
538 	if (*decimal_point == '\0')
539 		decimal_point = lc->decimal_point;
540 	thousands_sep = lc->mon_thousands_sep;
541 	if (*thousands_sep == '\0')
542 		thousands_sep = lc->thousands_sep;
543 
544 	decimal_point_size = strlen(decimal_point);
545 	thousands_sep_size = strlen(thousands_sep);
546 
547 	/* fill left_prec with default value */
548 	if (left_prec == -1)
549 		left_prec = 0;
550 
551 	/* fill right_prec with default value */
552 	if (right_prec == -1) {
553 		if (*flags & USE_INTL_CURRENCY)
554 			right_prec = lc->int_frac_digits;
555 		else
556 			right_prec = lc->frac_digits;
557 
558 		if (right_prec == CHAR_MAX)	/* POSIX locale ? */
559 			right_prec = 2;
560 	}
561 
562 	if (*flags & NEED_GROUPING)
563 		left_prec += get_groups(left_prec, grouping);
564 
565 	/* convert to string */
566 	avalue_size = asprintf_l(&avalue, loc, "%*.*f",
567 	    left_prec + right_prec + 1, right_prec, value);
568 	if (avalue_size < 0)
569 		return (NULL);
570 
571 	/* make sure that we've enough space for result string */
572 	bufsize = avalue_size * (1 + thousands_sep_size) + decimal_point_size +
573 	    1;
574 	rslt = calloc(1, bufsize);
575 	if (rslt == NULL) {
576 		free(avalue);
577 		return (NULL);
578 	}
579 	bufend = rslt + bufsize - 1;	/* reserve space for trailing '\0' */
580 
581 	/* skip spaces at beginning */
582 	padded = 0;
583 	while (avalue[padded] == ' ') {
584 		padded++;
585 		avalue_size--;
586 	}
587 
588 	if (right_prec > 0) {
589 		bufend -= right_prec;
590 		memcpy(bufend, avalue + avalue_size + padded - right_prec,
591 		    right_prec);
592 		bufend -= decimal_point_size;
593 		memcpy(bufend, decimal_point, decimal_point_size);
594 		avalue_size -= (right_prec + 1);
595 	}
596 
597 	if ((*flags & NEED_GROUPING) &&
598 	    thousands_sep_size > 0 &&
599 	    *grouping != CHAR_MAX &&
600 	    *grouping > 0) {
601 		while (avalue_size > (int)*grouping) {
602 			GRPCPY(*grouping);
603 			GRPSEP;
604 			grouping++;
605 
606 			/* no more grouping ? */
607 			if (*grouping == CHAR_MAX)
608 				break;
609 
610 			/* rest grouping with same value ? */
611 			if (*grouping == 0) {
612 				grouping--;
613 				while (avalue_size > *grouping) {
614 					GRPCPY(*grouping);
615 					GRPSEP;
616 				}
617 			}
618 		}
619 		if (avalue_size != 0)
620 			GRPCPY(avalue_size);
621 		padded -= groups;
622 	} else {
623 		bufend -= avalue_size;
624 		memcpy(bufend, avalue + padded, avalue_size);
625 		/* decrease assumed $decimal_point */
626 		if (right_prec == 0)
627 			padded -= decimal_point_size;
628 	}
629 
630 	/* do padding with pad_char */
631 	if (padded > 0) {
632 		bufend -= padded;
633 		memset(bufend, pad_char, padded);
634 	}
635 
636 	bufsize = rslt + bufsize - bufend;
637 	memmove(rslt, bufend, bufsize);
638 	free(avalue);
639 	return (rslt);
640 }
641 
642 ssize_t
643 strfmon(char * __restrict s, size_t maxsize, const char * __restrict format,
644     ...)
645 {
646 	ssize_t ret;
647 	va_list ap;
648 
649 	va_start(ap, format);
650 	ret = vstrfmon_l(s, maxsize, __get_locale(), format, ap);
651 	va_end(ap);
652 
653 	return (ret);
654 }
655 
656 ssize_t
657 strfmon_l(char * __restrict s, size_t maxsize, locale_t loc,
658     const char * __restrict format, ...)
659 {
660 	ssize_t ret;
661 	va_list ap;
662 
663 	va_start(ap, format);
664 	ret = vstrfmon_l(s, maxsize, loc, format, ap);
665 	va_end(ap);
666 
667 	return (ret);
668 }
669