xref: /freebsd/lib/libc/stdlib/strfmon.c (revision 6abee52e0d79f68fd725de748d7027ca8eef2294)
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 #include <ctype.h>
37 #include <errno.h>
38 #include <limits.h>
39 #include <locale.h>
40 #include <monetary.h>
41 #include <stdarg.h>
42 #include <stdio.h>
43 #include <stdlib.h>
44 #include <string.h>
45 
46 #include "xlocale_private.h"
47 
48 /* internal flags */
49 #define	NEED_GROUPING		0x01	/* print digits grouped (default) */
50 #define	SIGN_POSN_USED		0x02	/* '+' or '(' usage flag */
51 #define	LOCALE_POSN		0x04	/* use locale defined +/- (default) */
52 #define	PARENTH_POSN		0x08	/* enclose negative amount in () */
53 #define	SUPPRESS_CURR_SYMBOL	0x10	/* suppress the currency from output */
54 #define	LEFT_JUSTIFY		0x20	/* left justify */
55 #define	USE_INTL_CURRENCY	0x40	/* use international currency symbol */
56 #define	IS_NEGATIVE		0x80	/* is argument value negative ? */
57 
58 /* internal macros */
59 #define	PRINT(CH) do {						\
60 	if (dst >= s + maxsize)				\
61 		goto e2big_error;				\
62 	*dst++ = CH;						\
63 } while (0)
64 
65 #define	PRINTS(STR) do {					\
66 	char *tmps = STR;					\
67 	while (*tmps != '\0')					\
68 		PRINT(*tmps++);					\
69 } while (0)
70 
71 #define	GET_NUMBER(VAR, LOC) do {					\
72 	VAR = 0;						\
73 	while (isdigit_l((unsigned char)*fmt, LOC)) {			\
74 		if (VAR > INT_MAX / 10)				\
75 			goto e2big_error;			\
76 		VAR *= 10;					\
77 		VAR += *fmt - '0';				\
78 		if (VAR < 0)					\
79 			goto e2big_error;			\
80 		fmt++;						\
81 	}							\
82 } while (0)
83 
84 #define	GRPCPY(howmany) do {					\
85 	int i = howmany;					\
86 	while (i-- > 0) {					\
87 		avalue_size--;					\
88 		*--bufend = *(avalue + avalue_size + padded);	\
89 	}							\
90 } while (0)
91 
92 #define	GRPSEP do {						\
93 	bufend -= thousands_sep_size;				\
94 	memcpy(bufend, thousands_sep, thousands_sep_size);	\
95 	groups++;						\
96 } while (0)
97 
98 static void __setup_vars(int, char *, char *, char *, char **, struct lconv *);
99 static int __calc_left_pad(int, char *, struct lconv *);
100 static char *__format_grouped_double(double, int *, int, int, int,
101     struct lconv *, locale_t);
102 
103 static ssize_t
104 vstrfmon_l(char * __restrict s, size_t maxsize, locale_t loc,
105     const char * __restrict format, va_list ap)
106 {
107 	char		*dst;		/* output destination pointer */
108 	const char	*fmt;		/* current format poistion pointer */
109 	struct lconv	*lc;		/* pointer to lconv structure */
110 	char		*asciivalue;	/* formatted double pointer */
111 
112 	int		flags;		/* formatting options */
113 	int		pad_char;	/* padding character */
114 	int		pad_size;	/* pad size */
115 	int		width;		/* field width */
116 	int		left_prec;	/* left precision */
117 	int		right_prec;	/* right precision */
118 	double		value;		/* just value */
119 	char		space_char = ' '; /* space after currency */
120 
121 	char		cs_precedes,	/* values gathered from struct lconv */
122 			sep_by_space,
123 			sign_posn,
124 			*signstr,
125 			*currency_symbol;
126 
127 	char		*tmpptr;	/* temporary vars */
128 	int		sverrno;
129 	FIX_LOCALE(loc);
130 
131 	lc = localeconv_l(loc);
132 	dst = s;
133 	fmt = format;
134 	asciivalue = NULL;
135 	currency_symbol = NULL;
136 
137 	while (*fmt) {
138 		/* pass nonformating characters AS IS */
139 		if (*fmt != '%')
140 			goto literal;
141 
142 		/* '%' found ! */
143 
144 		/* "%%" mean just '%' */
145 		if (*(fmt + 1) == '%') {
146 			fmt++;
147 	literal:
148 			PRINT(*fmt++);
149 			continue;
150 		}
151 
152 		/* set up initial values */
153 		flags = (NEED_GROUPING|LOCALE_POSN);
154 		pad_char = ' ';		/* padding character is "space" */
155 		pad_size = 0;		/* no padding initially */
156 		left_prec = -1;		/* no left precision specified */
157 		right_prec = -1;	/* no right precision specified */
158 		width = -1;		/* no width specified */
159 
160 		/* Flags */
161 		while (1) {
162 			switch (*++fmt) {
163 				case '=':	/* fill character */
164 					pad_char = *++fmt;
165 					if (pad_char == '\0')
166 						goto format_error;
167 					continue;
168 				case '^':	/* not group currency  */
169 					flags &= ~(NEED_GROUPING);
170 					continue;
171 				case '+':	/* use locale defined signs */
172 					if (flags & SIGN_POSN_USED)
173 						goto format_error;
174 					flags |= (SIGN_POSN_USED|LOCALE_POSN);
175 					continue;
176 				case '(':	/* enclose negatives with () */
177 					if (flags & SIGN_POSN_USED)
178 						goto format_error;
179 					flags |= (SIGN_POSN_USED|PARENTH_POSN);
180 					continue;
181 				case '!':	/* suppress currency symbol */
182 					flags |= SUPPRESS_CURR_SYMBOL;
183 					continue;
184 				case '-':	/* alignment (left)  */
185 					flags |= LEFT_JUSTIFY;
186 					continue;
187 				default:
188 					break;
189 			}
190 			break;
191 		}
192 
193 		/* field Width */
194 		if (isdigit_l((unsigned char)*fmt, loc)) {
195 			GET_NUMBER(width, loc);
196 			/* Do we have enough space to put number with
197 			 * required width ?
198 			 */
199 			if ((unsigned int)width >= maxsize - (dst - s))
200 				goto e2big_error;
201 		}
202 
203 		/* Left precision */
204 		if (*fmt == '#') {
205 			if (!isdigit_l((unsigned char)*++fmt, loc))
206 				goto format_error;
207 			GET_NUMBER(left_prec, loc);
208 			if ((unsigned int)left_prec >= maxsize - (dst - s))
209 				goto e2big_error;
210 		}
211 
212 		/* Right precision */
213 		if (*fmt == '.') {
214 			if (!isdigit_l((unsigned char)*++fmt, loc))
215 				goto format_error;
216 			GET_NUMBER(right_prec, loc);
217 			if ((unsigned int)right_prec >= maxsize - (dst - s) -
218 			    left_prec)
219 				goto e2big_error;
220 		}
221 
222 		/* Conversion Characters */
223 		switch (*fmt++) {
224 			case 'i':	/* use international currency format */
225 				flags |= USE_INTL_CURRENCY;
226 				break;
227 			case 'n':	/* use national currency format */
228 				flags &= ~(USE_INTL_CURRENCY);
229 				break;
230 			default:	/* required character is missing or
231 					   premature EOS */
232 				goto format_error;
233 		}
234 
235 		if (currency_symbol != NULL)
236 			free(currency_symbol);
237 		if (flags & USE_INTL_CURRENCY) {
238 			currency_symbol = strdup(lc->int_curr_symbol);
239 			if (currency_symbol != NULL &&
240 			    strlen(currency_symbol) > 3) {
241 				space_char = currency_symbol[3];
242 				currency_symbol[3] = '\0';
243 			}
244 		} else
245 			currency_symbol = strdup(lc->currency_symbol);
246 
247 		if (currency_symbol == NULL)
248 			goto end_error;			/* ENOMEM. */
249 
250 		/* value itself */
251 		value = va_arg(ap, double);
252 
253 		/* detect sign */
254 		if (value < 0) {
255 			flags |= IS_NEGATIVE;
256 			value = -value;
257 		}
258 
259 		/* fill left_prec with amount of padding chars */
260 		if (left_prec >= 0) {
261 			pad_size = __calc_left_pad((flags ^ IS_NEGATIVE),
262 			    currency_symbol, lc) -
263 			    __calc_left_pad(flags, currency_symbol, lc);
264 			if (pad_size < 0)
265 				pad_size = 0;
266 		}
267 
268 		if (asciivalue != NULL)
269 			free(asciivalue);
270 		asciivalue = __format_grouped_double(value, &flags,
271 		    left_prec, right_prec, pad_char, lc, loc);
272 		if (asciivalue == NULL)
273 			goto end_error;		/* errno already set     */
274 						/* to ENOMEM by malloc() */
275 
276 		/* set some variables for later use */
277 		__setup_vars(flags, &cs_precedes, &sep_by_space,
278 		    &sign_posn, &signstr, lc);
279 
280 		/*
281 		 * Description of some LC_MONETARY's values:
282 		 *
283 		 * p_cs_precedes & n_cs_precedes
284 		 *
285 		 * = 1 - $currency_symbol precedes the value
286 		 *       for a monetary quantity with a non-negative value
287 		 * = 0 - symbol succeeds the value
288 		 *
289 		 * p_sep_by_space & n_sep_by_space
290 		 *
291 		 * = 0 - no space separates $currency_symbol
292 		 *       from the value for a monetary quantity with a
293 		 *       non-negative value
294 		 * = 1 - space separates the symbol from the value
295 		 * = 2 - space separates the symbol and the sign string,
296 		 *       if adjacent; otherwise, a space separates
297 		 *       the sign string from the value
298 		 *
299 		 * p_sign_posn & n_sign_posn
300 		 *
301 		 * = 0 - parentheses enclose the quantity and the
302 		 *       $currency_symbol
303 		 * = 1 - the sign string precedes the quantity and the
304 		 *       $currency_symbol
305 		 * = 2 - the sign string succeeds the quantity and the
306 		 *       $currency_symbol
307 		 * = 3 - the sign string precedes the $currency_symbol
308 		 * = 4 - the sign string succeeds the $currency_symbol
309 		 *
310 		 */
311 
312 		tmpptr = dst;
313 
314 		while (pad_size-- > 0)
315 			PRINT(' ');
316 
317 		if (sign_posn == 0 && (flags & IS_NEGATIVE))
318 			PRINT('(');
319 
320 		if (cs_precedes == 1) {
321 			if (sign_posn == 1 || sign_posn == 3) {
322 				PRINTS(signstr);
323 				if (sep_by_space == 2)
324 					PRINT(' ');
325 			}
326 
327 			if (!(flags & SUPPRESS_CURR_SYMBOL)) {
328 				PRINTS(currency_symbol);
329 
330 				if (sign_posn == 4) {
331 					if (sep_by_space == 2)
332 						PRINT(space_char);
333 					PRINTS(signstr);
334 					if (sep_by_space == 1)
335 						PRINT(' ');
336 				} else if (sep_by_space == 1)
337 					PRINT(space_char);
338 			}
339 		} else if (sign_posn == 1) {
340 			PRINTS(signstr);
341 			if (sep_by_space == 2)
342 				PRINT(' ');
343 		}
344 
345 		PRINTS(asciivalue);
346 
347 		if (cs_precedes == 0) {
348 			if (sign_posn == 3) {
349 				if (sep_by_space == 1)
350 					PRINT(' ');
351 				PRINTS(signstr);
352 			}
353 
354 			if (!(flags & SUPPRESS_CURR_SYMBOL)) {
355 				if ((sign_posn == 3 && sep_by_space == 2)
356 				    || (sep_by_space == 1
357 				    && (sign_posn == 0
358 				    || sign_posn == 1
359 				    || sign_posn == 2
360 				    || sign_posn == 4)))
361 					PRINT(space_char);
362 				PRINTS(currency_symbol);
363 				if (sign_posn == 4) {
364 					if (sep_by_space == 2)
365 						PRINT(' ');
366 					PRINTS(signstr);
367 				}
368 			}
369 		}
370 
371 		if (sign_posn == 2) {
372 			if (sep_by_space == 2)
373 				PRINT(' ');
374 			PRINTS(signstr);
375 		}
376 
377 		if (sign_posn == 0) {
378 			if (flags & IS_NEGATIVE)
379 				PRINT(')');
380 			else if (left_prec >= 0)
381 				PRINT(' ');
382 		}
383 
384 		if (dst - tmpptr < width) {
385 			if (flags & LEFT_JUSTIFY) {
386 				while (dst - tmpptr < width)
387 					PRINT(' ');
388 			} else {
389 				pad_size = dst - tmpptr;
390 				memmove(tmpptr + width - pad_size, tmpptr,
391 				    pad_size);
392 				memset(tmpptr, ' ', width - pad_size);
393 				dst += width - pad_size;
394 			}
395 		}
396 	}
397 
398 	PRINT('\0');
399 	free(asciivalue);
400 	free(currency_symbol);
401 	return (dst - s - 1);	/* return size of put data except trailing '\0' */
402 
403 e2big_error:
404 	errno = E2BIG;
405 	goto end_error;
406 
407 format_error:
408 	errno = EINVAL;
409 
410 end_error:
411 	sverrno = errno;
412 	if (asciivalue != NULL)
413 		free(asciivalue);
414 	if (currency_symbol != NULL)
415 		free(currency_symbol);
416 	errno = sverrno;
417 	return (-1);
418 }
419 
420 static void
421 __setup_vars(int flags, char *cs_precedes, char *sep_by_space,
422     char *sign_posn, char **signstr, struct lconv *lc)
423 {
424 	if ((flags & IS_NEGATIVE) && (flags & USE_INTL_CURRENCY)) {
425 		*cs_precedes = lc->int_n_cs_precedes;
426 		*sep_by_space = lc->int_n_sep_by_space;
427 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->int_n_sign_posn;
428 		*signstr = (lc->negative_sign[0] == '\0') ? "-"
429 		    : lc->negative_sign;
430 	} else if (flags & USE_INTL_CURRENCY) {
431 		*cs_precedes = lc->int_p_cs_precedes;
432 		*sep_by_space = lc->int_p_sep_by_space;
433 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->int_p_sign_posn;
434 		*signstr = lc->positive_sign;
435 	} else if (flags & IS_NEGATIVE) {
436 		*cs_precedes = lc->n_cs_precedes;
437 		*sep_by_space = lc->n_sep_by_space;
438 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->n_sign_posn;
439 		*signstr = (lc->negative_sign[0] == '\0') ? "-"
440 		    : lc->negative_sign;
441 	} else {
442 		*cs_precedes = lc->p_cs_precedes;
443 		*sep_by_space = lc->p_sep_by_space;
444 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->p_sign_posn;
445 		*signstr = lc->positive_sign;
446 	}
447 
448 	/* Set default values for unspecified information. */
449 	if (*cs_precedes != 0)
450 		*cs_precedes = 1;
451 	if (*sep_by_space == CHAR_MAX)
452 		*sep_by_space = 0;
453 	if (*sign_posn == CHAR_MAX)
454 		*sign_posn = 0;
455 }
456 
457 static int
458 __calc_left_pad(int flags, char *cur_symb, struct lconv *lc)
459 {
460 	char cs_precedes, sep_by_space, sign_posn, *signstr;
461 	int left_chars = 0;
462 
463 	__setup_vars(flags, &cs_precedes, &sep_by_space, &sign_posn,
464 	    &signstr, lc);
465 
466 	if (cs_precedes != 0) {
467 		left_chars += strlen(cur_symb);
468 		if (sep_by_space != 0)
469 			left_chars++;
470 	}
471 
472 	switch (sign_posn) {
473 		case 0:
474 			if (flags & IS_NEGATIVE)
475 				left_chars++;
476 			break;
477 		case 1:
478 			left_chars += strlen(signstr);
479 			break;
480 		case 3:
481 		case 4:
482 			if (cs_precedes != 0)
483 				left_chars += strlen(signstr);
484 	}
485 	return (left_chars);
486 }
487 
488 static int
489 get_groups(int size, char *grouping)
490 {
491 	int	chars = 0;
492 
493 	if (*grouping == CHAR_MAX || *grouping <= 0)	/* no grouping ? */
494 		return (0);
495 
496 	while (size > (int)*grouping) {
497 		chars++;
498 		size -= (int)*grouping++;
499 		/* no more grouping ? */
500 		if (*grouping == CHAR_MAX)
501 			break;
502 		/* rest grouping with same value ? */
503 		if (*grouping == 0) {
504 			chars += (size - 1) / *(grouping - 1);
505 			break;
506 		}
507 	}
508 	return (chars);
509 }
510 
511 /* convert double to locale-encoded string */
512 static char *
513 __format_grouped_double(double value, int *flags,
514     int left_prec, int right_prec, int pad_char, struct lconv *lc, locale_t loc)
515 {
516 
517 	char		*rslt;
518 	char		*avalue;
519 	int		avalue_size;
520 
521 	size_t		bufsize;
522 	char		*bufend;
523 
524 	int		padded;
525 
526 	char		*grouping;
527 	const char	*decimal_point;
528 	const char	*thousands_sep;
529 	size_t		decimal_point_size;
530 	size_t		thousands_sep_size;
531 
532 	int groups = 0;
533 
534 	grouping = lc->mon_grouping;
535 	decimal_point = lc->mon_decimal_point;
536 	if (*decimal_point == '\0')
537 		decimal_point = lc->decimal_point;
538 	thousands_sep = lc->mon_thousands_sep;
539 	if (*thousands_sep == '\0')
540 		thousands_sep = lc->thousands_sep;
541 
542 	decimal_point_size = strlen(decimal_point);
543 	thousands_sep_size = strlen(thousands_sep);
544 
545 	/* fill left_prec with default value */
546 	if (left_prec == -1)
547 		left_prec = 0;
548 
549 	/* fill right_prec with default value */
550 	if (right_prec == -1) {
551 		if (*flags & USE_INTL_CURRENCY)
552 			right_prec = lc->int_frac_digits;
553 		else
554 			right_prec = lc->frac_digits;
555 
556 		if (right_prec == CHAR_MAX)	/* POSIX locale ? */
557 			right_prec = 2;
558 	}
559 
560 	if (*flags & NEED_GROUPING)
561 		left_prec += get_groups(left_prec, grouping);
562 
563 	/* convert to string */
564 	avalue_size = asprintf_l(&avalue, loc, "%*.*f",
565 	    left_prec + right_prec + 1, right_prec, value);
566 	if (avalue_size < 0)
567 		return (NULL);
568 
569 	/* make sure that we've enough space for result string */
570 	bufsize = avalue_size * (1 + thousands_sep_size) + decimal_point_size +
571 	    1;
572 	rslt = calloc(1, bufsize);
573 	if (rslt == NULL) {
574 		free(avalue);
575 		return (NULL);
576 	}
577 	bufend = rslt + bufsize - 1;	/* reserve space for trailing '\0' */
578 
579 	/* skip spaces at beginning */
580 	padded = 0;
581 	while (avalue[padded] == ' ') {
582 		padded++;
583 		avalue_size--;
584 	}
585 
586 	if (right_prec > 0) {
587 		bufend -= right_prec;
588 		memcpy(bufend, avalue + avalue_size + padded - right_prec,
589 		    right_prec);
590 		bufend -= decimal_point_size;
591 		memcpy(bufend, decimal_point, decimal_point_size);
592 		avalue_size -= (right_prec + 1);
593 	}
594 
595 	if ((*flags & NEED_GROUPING) &&
596 	    thousands_sep_size > 0 &&
597 	    *grouping != CHAR_MAX &&
598 	    *grouping > 0) {
599 		while (avalue_size > (int)*grouping) {
600 			GRPCPY(*grouping);
601 			GRPSEP;
602 			grouping++;
603 
604 			/* no more grouping ? */
605 			if (*grouping == CHAR_MAX)
606 				break;
607 
608 			/* rest grouping with same value ? */
609 			if (*grouping == 0) {
610 				grouping--;
611 				while (avalue_size > *grouping) {
612 					GRPCPY(*grouping);
613 					GRPSEP;
614 				}
615 			}
616 		}
617 		if (avalue_size != 0)
618 			GRPCPY(avalue_size);
619 		padded -= groups;
620 	} else {
621 		bufend -= avalue_size;
622 		memcpy(bufend, avalue + padded, avalue_size);
623 		/* decrease assumed $decimal_point */
624 		if (right_prec == 0)
625 			padded -= decimal_point_size;
626 	}
627 
628 	/* do padding with pad_char */
629 	if (padded > 0) {
630 		bufend -= padded;
631 		memset(bufend, pad_char, padded);
632 	}
633 
634 	bufsize = rslt + bufsize - bufend;
635 	memmove(rslt, bufend, bufsize);
636 	free(avalue);
637 	return (rslt);
638 }
639 
640 ssize_t
641 strfmon(char * __restrict s, size_t maxsize, const char * __restrict format,
642     ...)
643 {
644 	ssize_t ret;
645 	va_list ap;
646 
647 	va_start(ap, format);
648 	ret = vstrfmon_l(s, maxsize, __get_locale(), format, ap);
649 	va_end(ap);
650 
651 	return (ret);
652 }
653 
654 ssize_t
655 strfmon_l(char * __restrict s, size_t maxsize, locale_t loc,
656     const char * __restrict format, ...)
657 {
658 	ssize_t ret;
659 	va_list ap;
660 
661 	va_start(ap, format);
662 	ret = vstrfmon_l(s, maxsize, loc, format, ap);
663 	va_end(ap);
664 
665 	return (ret);
666 }
667