xref: /freebsd/lib/libc/stdlib/strfmon.c (revision 9e03b903e377c75a60cbbb89ed78955769a1c804)
1 /*-
2  * SPDX-License-Identifier: BSD-2-Clause-FreeBSD
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 __FBSDID("$FreeBSD$");
37 
38 #include <sys/types.h>
39 #include <ctype.h>
40 #include <errno.h>
41 #include <limits.h>
42 #include <locale.h>
43 #include <monetary.h>
44 #include <stdarg.h>
45 #include <stdio.h>
46 #include <stdlib.h>
47 #include <string.h>
48 
49 #include "xlocale_private.h"
50 
51 /* internal flags */
52 #define	NEED_GROUPING		0x01	/* print digits grouped (default) */
53 #define	SIGN_POSN_USED		0x02	/* '+' or '(' usage flag */
54 #define	LOCALE_POSN		0x04	/* use locale defined +/- (default) */
55 #define	PARENTH_POSN		0x08	/* enclose negative amount in () */
56 #define	SUPPRESS_CURR_SYMBOL	0x10	/* suppress the currency from output */
57 #define	LEFT_JUSTIFY		0x20	/* left justify */
58 #define	USE_INTL_CURRENCY	0x40	/* use international currency symbol */
59 #define	IS_NEGATIVE		0x80	/* is argument value negative ? */
60 
61 /* internal macros */
62 #define	PRINT(CH) do {						\
63 	if (dst >= s + maxsize)				\
64 		goto e2big_error;				\
65 	*dst++ = CH;						\
66 } while (0)
67 
68 #define	PRINTS(STR) do {					\
69 	char *tmps = STR;					\
70 	while (*tmps != '\0')					\
71 		PRINT(*tmps++);					\
72 } while (0)
73 
74 #define	GET_NUMBER(VAR) do {					\
75 	VAR = 0;						\
76 	while (isdigit((unsigned char)*fmt)) {			\
77 		if (VAR > INT_MAX / 10)				\
78 			goto e2big_error;			\
79 		VAR *= 10;					\
80 		VAR += *fmt - '0';				\
81 		if (VAR < 0)					\
82 			goto e2big_error;			\
83 		fmt++;						\
84 	}							\
85 } while (0)
86 
87 #define	GRPCPY(howmany) do {					\
88 	int i = howmany;					\
89 	while (i-- > 0) {					\
90 		avalue_size--;					\
91 		*--bufend = *(avalue + avalue_size + padded);	\
92 	}							\
93 } while (0)
94 
95 #define	GRPSEP do {						\
96 	bufend -= thousands_sep_size;				\
97 	memcpy(bufend, thousands_sep, thousands_sep_size);	\
98 	groups++;						\
99 } while (0)
100 
101 static void __setup_vars(int, char *, char *, char *, char **);
102 static int __calc_left_pad(int, char *);
103 static char *__format_grouped_double(double, int *, int, int, int);
104 
105 static ssize_t
106 vstrfmon_l(char * __restrict s, size_t maxsize, locale_t loc,
107     const char * __restrict format, va_list ap)
108 {
109 	char		*dst;		/* output destination pointer */
110 	const char	*fmt;		/* current format poistion pointer */
111 	struct lconv	*lc;		/* pointer to lconv structure */
112 	char		*asciivalue;	/* formatted double pointer */
113 
114 	int		flags;		/* formatting options */
115 	int		pad_char;	/* padding character */
116 	int		pad_size;	/* pad size */
117 	int		width;		/* field width */
118 	int		left_prec;	/* left precision */
119 	int		right_prec;	/* right precision */
120 	double		value;		/* just value */
121 	char		space_char = ' '; /* space after currency */
122 
123 	char		cs_precedes,	/* values gathered from struct lconv */
124 			sep_by_space,
125 			sign_posn,
126 			*signstr,
127 			*currency_symbol;
128 
129 	char		*tmpptr;	/* temporary vars */
130 	int		sverrno;
131 	FIX_LOCALE(loc);
132 
133 	lc = localeconv_l(loc);
134 	dst = s;
135 	fmt = format;
136 	asciivalue = NULL;
137 	currency_symbol = NULL;
138 	pad_size = 0;
139 
140 	while (*fmt) {
141 		/* pass nonformating characters AS IS */
142 		if (*fmt != '%')
143 			goto literal;
144 
145 		/* '%' found ! */
146 
147 		/* "%%" mean just '%' */
148 		if (*(fmt + 1) == '%') {
149 			fmt++;
150 	literal:
151 			PRINT(*fmt++);
152 			continue;
153 		}
154 
155 		/* set up initial values */
156 		flags = (NEED_GROUPING|LOCALE_POSN);
157 		pad_char = ' ';		/* padding character is "space" */
158 		left_prec = -1;		/* no left precision specified */
159 		right_prec = -1;	/* no right precision specified */
160 		width = -1;		/* no width specified */
161 		value = 0;		/* we have no value to print now */
162 
163 		/* Flags */
164 		while (1) {
165 			switch (*++fmt) {
166 				case '=':	/* fill character */
167 					pad_char = *++fmt;
168 					if (pad_char == '\0')
169 						goto format_error;
170 					continue;
171 				case '^':	/* not group currency  */
172 					flags &= ~(NEED_GROUPING);
173 					continue;
174 				case '+':	/* use locale defined signs */
175 					if (flags & SIGN_POSN_USED)
176 						goto format_error;
177 					flags |= (SIGN_POSN_USED|LOCALE_POSN);
178 					continue;
179 				case '(':	/* enclose negatives with () */
180 					if (flags & SIGN_POSN_USED)
181 						goto format_error;
182 					flags |= (SIGN_POSN_USED|PARENTH_POSN);
183 					continue;
184 				case '!':	/* suppress currency symbol */
185 					flags |= SUPPRESS_CURR_SYMBOL;
186 					continue;
187 				case '-':	/* alignment (left)  */
188 					flags |= LEFT_JUSTIFY;
189 					continue;
190 				default:
191 					break;
192 			}
193 			break;
194 		}
195 
196 		/* field Width */
197 		if (isdigit((unsigned char)*fmt)) {
198 			GET_NUMBER(width);
199 			/* Do we have enough space to put number with
200 			 * required width ?
201 			 */
202 			if ((unsigned int)width >= maxsize - (dst - s))
203 				goto e2big_error;
204 		}
205 
206 		/* Left precision */
207 		if (*fmt == '#') {
208 			if (!isdigit((unsigned char)*++fmt))
209 				goto format_error;
210 			GET_NUMBER(left_prec);
211 			if ((unsigned int)left_prec >= maxsize - (dst - s))
212 				goto e2big_error;
213 		}
214 
215 		/* Right precision */
216 		if (*fmt == '.') {
217 			if (!isdigit((unsigned char)*++fmt))
218 				goto format_error;
219 			GET_NUMBER(right_prec);
220 			if ((unsigned int)right_prec >= maxsize - (dst - s) -
221 			    left_prec)
222 				goto e2big_error;
223 		}
224 
225 		/* Conversion Characters */
226 		switch (*fmt++) {
227 			case 'i':	/* use international currency format */
228 				flags |= USE_INTL_CURRENCY;
229 				break;
230 			case 'n':	/* use national currency format */
231 				flags &= ~(USE_INTL_CURRENCY);
232 				break;
233 			default:	/* required character is missing or
234 					   premature EOS */
235 				goto format_error;
236 		}
237 
238 		if (currency_symbol != NULL)
239 			free(currency_symbol);
240 		if (flags & USE_INTL_CURRENCY) {
241 			currency_symbol = strdup(lc->int_curr_symbol);
242 			if (currency_symbol != NULL &&
243 			    strlen(currency_symbol) > 3)
244 				space_char = currency_symbol[3];
245 		} else
246 			currency_symbol = strdup(lc->currency_symbol);
247 
248 		if (currency_symbol == NULL)
249 			goto end_error;			/* ENOMEM. */
250 
251 		/* value itself */
252 		value = va_arg(ap, double);
253 
254 		/* detect sign */
255 		if (value < 0) {
256 			flags |= IS_NEGATIVE;
257 			value = -value;
258 		}
259 
260 		/* fill left_prec with amount of padding chars */
261 		if (left_prec >= 0) {
262 			pad_size = __calc_left_pad((flags ^ IS_NEGATIVE),
263 							currency_symbol) -
264 				   __calc_left_pad(flags, currency_symbol);
265 			if (pad_size < 0)
266 				pad_size = 0;
267 		}
268 
269 		if (asciivalue != NULL)
270 			free(asciivalue);
271 		asciivalue = __format_grouped_double(value, &flags,
272 				left_prec, right_prec, pad_char);
273 		if (asciivalue == NULL)
274 			goto end_error;		/* errno already set     */
275 						/* to ENOMEM by malloc() */
276 
277 		/* set some variables for later use */
278 		__setup_vars(flags, &cs_precedes, &sep_by_space,
279 				&sign_posn, &signstr);
280 
281 		/*
282 		 * Description of some LC_MONETARY's values:
283 		 *
284 		 * p_cs_precedes & n_cs_precedes
285 		 *
286 		 * = 1 - $currency_symbol precedes the value
287 		 *       for a monetary quantity with a non-negative value
288 		 * = 0 - symbol succeeds the value
289 		 *
290 		 * p_sep_by_space & n_sep_by_space
291 		 *
292 		 * = 0 - no space separates $currency_symbol
293 		 *       from the value for a monetary quantity with a
294 		 *       non-negative value
295 		 * = 1 - space separates the symbol from the value
296 		 * = 2 - space separates the symbol and the sign string,
297 		 *       if adjacent.
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)		/* XXX: ? */
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 
342 		PRINTS(asciivalue);
343 
344 		if (cs_precedes == 0) {
345 			if (sign_posn == 3) {
346 				if (sep_by_space == 1)
347 					PRINT(' ');
348 				PRINTS(signstr);
349 			}
350 
351 			if (!(flags & SUPPRESS_CURR_SYMBOL)) {
352 				if ((sign_posn == 3 && sep_by_space == 2)
353 				    || (sep_by_space == 1
354 				    && (sign_posn == 0
355 				    || sign_posn == 1
356 				    || sign_posn == 2
357 				    || sign_posn == 4)))
358 					PRINT(space_char);
359 				PRINTS(currency_symbol); /* XXX: len */
360 				if (sign_posn == 4) {
361 					if (sep_by_space == 2)
362 						PRINT(' ');
363 					PRINTS(signstr);
364 				}
365 			}
366 		}
367 
368 		if (sign_posn == 2) {
369 			if (sep_by_space == 2)
370 				PRINT(' ');
371 			PRINTS(signstr);
372 		}
373 
374 		if (sign_posn == 0 && (flags & IS_NEGATIVE))
375 			PRINT(')');
376 
377 		if (dst - tmpptr < width) {
378 			if (flags & LEFT_JUSTIFY) {
379 				while (dst - tmpptr < width)
380 					PRINT(' ');
381 			} else {
382 				pad_size = dst - tmpptr;
383 				memmove(tmpptr + width - pad_size, tmpptr,
384 				    pad_size);
385 				memset(tmpptr, ' ', width - pad_size);
386 				dst += width - pad_size;
387 			}
388 		}
389 	}
390 
391 	PRINT('\0');
392 	free(asciivalue);
393 	free(currency_symbol);
394 	return (dst - s - 1);	/* return size of put data except trailing '\0' */
395 
396 e2big_error:
397 	errno = E2BIG;
398 	goto end_error;
399 
400 format_error:
401 	errno = EINVAL;
402 
403 end_error:
404 	sverrno = errno;
405 	if (asciivalue != NULL)
406 		free(asciivalue);
407 	if (currency_symbol != NULL)
408 		free(currency_symbol);
409 	errno = sverrno;
410 	return (-1);
411 }
412 
413 static void
414 __setup_vars(int flags, char *cs_precedes, char *sep_by_space,
415     char *sign_posn, char **signstr)
416 {
417 	struct lconv *lc = localeconv();
418 
419 	if ((flags & IS_NEGATIVE) && (flags & USE_INTL_CURRENCY)) {
420 		*cs_precedes = lc->int_n_cs_precedes;
421 		*sep_by_space = lc->int_n_sep_by_space;
422 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->int_n_sign_posn;
423 		*signstr = (lc->negative_sign[0] == '\0') ? "-"
424 		    : lc->negative_sign;
425 	} else if (flags & USE_INTL_CURRENCY) {
426 		*cs_precedes = lc->int_p_cs_precedes;
427 		*sep_by_space = lc->int_p_sep_by_space;
428 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->int_p_sign_posn;
429 		*signstr = lc->positive_sign;
430 	} else if (flags & IS_NEGATIVE) {
431 		*cs_precedes = lc->n_cs_precedes;
432 		*sep_by_space = lc->n_sep_by_space;
433 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->n_sign_posn;
434 		*signstr = (lc->negative_sign[0] == '\0') ? "-"
435 		    : lc->negative_sign;
436 	} else {
437 		*cs_precedes = lc->p_cs_precedes;
438 		*sep_by_space = lc->p_sep_by_space;
439 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->p_sign_posn;
440 		*signstr = lc->positive_sign;
441 	}
442 
443 	/* Set default values for unspecified information. */
444 	if (*cs_precedes != 0)
445 		*cs_precedes = 1;
446 	if (*sep_by_space == CHAR_MAX)
447 		*sep_by_space = 0;
448 	if (*sign_posn == CHAR_MAX)
449 		*sign_posn = 0;
450 }
451 
452 static int
453 __calc_left_pad(int flags, char *cur_symb)
454 {
455 	char cs_precedes, sep_by_space, sign_posn, *signstr;
456 	int left_chars = 0;
457 
458 	__setup_vars(flags, &cs_precedes, &sep_by_space, &sign_posn, &signstr);
459 
460 	if (cs_precedes != 0) {
461 		left_chars += strlen(cur_symb);
462 		if (sep_by_space != 0)
463 			left_chars++;
464 	}
465 
466 	switch (sign_posn) {
467 		case 1:
468 			left_chars += strlen(signstr);
469 			break;
470 		case 3:
471 		case 4:
472 			if (cs_precedes != 0)
473 				left_chars += strlen(signstr);
474 	}
475 	return (left_chars);
476 }
477 
478 static int
479 get_groups(int size, char *grouping)
480 {
481 	int	chars = 0;
482 
483 	if (*grouping == CHAR_MAX || *grouping <= 0)	/* no grouping ? */
484 		return (0);
485 
486 	while (size > (int)*grouping) {
487 		chars++;
488 		size -= (int)*grouping++;
489 		/* no more grouping ? */
490 		if (*grouping == CHAR_MAX)
491 			break;
492 		/* rest grouping with same value ? */
493 		if (*grouping == 0) {
494 			chars += (size - 1) / *(grouping - 1);
495 			break;
496 		}
497 	}
498 	return (chars);
499 }
500 
501 /* convert double to locale-encoded string */
502 static char *
503 __format_grouped_double(double value, int *flags,
504     int left_prec, int right_prec, int pad_char)
505 {
506 
507 	char		*rslt;
508 	char		*avalue;
509 	int		avalue_size;
510 
511 	size_t		bufsize;
512 	char		*bufend;
513 
514 	int		padded;
515 
516 	struct lconv	*lc = localeconv();
517 	char		*grouping;
518 	const char	*decimal_point;
519 	const char	*thousands_sep;
520 	size_t		decimal_point_size;
521 	size_t		thousands_sep_size;
522 
523 	int groups = 0;
524 
525 	grouping = lc->mon_grouping;
526 	decimal_point = lc->mon_decimal_point;
527 	if (*decimal_point == '\0')
528 		decimal_point = lc->decimal_point;
529 	thousands_sep = lc->mon_thousands_sep;
530 	if (*thousands_sep == '\0')
531 		thousands_sep = lc->thousands_sep;
532 
533 	decimal_point_size = strlen(decimal_point);
534 	thousands_sep_size = strlen(thousands_sep);
535 
536 	/* fill left_prec with default value */
537 	if (left_prec == -1)
538 		left_prec = 0;
539 
540 	/* fill right_prec with default value */
541 	if (right_prec == -1) {
542 		if (*flags & USE_INTL_CURRENCY)
543 			right_prec = lc->int_frac_digits;
544 		else
545 			right_prec = lc->frac_digits;
546 
547 		if (right_prec == CHAR_MAX)	/* POSIX locale ? */
548 			right_prec = 2;
549 	}
550 
551 	if (*flags & NEED_GROUPING)
552 		left_prec += get_groups(left_prec, grouping);
553 
554 	/* convert to string */
555 	avalue_size = asprintf(&avalue, "%*.*f", left_prec + right_prec + 1,
556 	    right_prec, value);
557 	if (avalue_size < 0)
558 		return (NULL);
559 
560 	/* make sure that we've enough space for result string */
561 	bufsize = avalue_size * (1 + thousands_sep_size) + decimal_point_size +
562 	    1;
563 	rslt = calloc(1, bufsize);
564 	if (rslt == NULL) {
565 		free(avalue);
566 		return (NULL);
567 	}
568 	bufend = rslt + bufsize - 1;	/* reserve space for trailing '\0' */
569 
570 	/* skip spaces at beginning */
571 	padded = 0;
572 	while (avalue[padded] == ' ') {
573 		padded++;
574 		avalue_size--;
575 	}
576 
577 	if (right_prec > 0) {
578 		bufend -= right_prec;
579 		memcpy(bufend, avalue + avalue_size + padded - right_prec,
580 		    right_prec);
581 		bufend -= decimal_point_size;
582 		memcpy(bufend, decimal_point, decimal_point_size);
583 		avalue_size -= (right_prec + 1);
584 	}
585 
586 	if ((*flags & NEED_GROUPING) &&
587 	    thousands_sep_size > 0 &&	/* XXX: need investigation */
588 	    *grouping != CHAR_MAX &&
589 	    *grouping > 0) {
590 		while (avalue_size > (int)*grouping) {
591 			GRPCPY(*grouping);
592 			GRPSEP;
593 			grouping++;
594 
595 			/* no more grouping ? */
596 			if (*grouping == CHAR_MAX)
597 				break;
598 
599 			/* rest grouping with same value ? */
600 			if (*grouping == 0) {
601 				grouping--;
602 				while (avalue_size > *grouping) {
603 					GRPCPY(*grouping);
604 					GRPSEP;
605 				}
606 			}
607 		}
608 		if (avalue_size != 0)
609 			GRPCPY(avalue_size);
610 		padded -= groups;
611 	} else {
612 		bufend -= avalue_size;
613 		memcpy(bufend, avalue + padded, avalue_size);
614 		/* decrease assumed $decimal_point */
615 		if (right_prec == 0)
616 			padded -= decimal_point_size;
617 	}
618 
619 	/* do padding with pad_char */
620 	if (padded > 0) {
621 		bufend -= padded;
622 		memset(bufend, pad_char, padded);
623 	}
624 
625 	bufsize = rslt + bufsize - bufend;
626 	memmove(rslt, bufend, bufsize);
627 	free(avalue);
628 	return (rslt);
629 }
630 
631 ssize_t
632 strfmon(char * __restrict s, size_t maxsize, const char * __restrict format,
633     ...)
634 {
635 	size_t ret;
636 	va_list ap;
637 
638 	va_start(ap, format);
639 	ret = vstrfmon_l(s, maxsize, __get_locale(), format, ap);
640 	va_end(ap);
641 
642 	return (ret);
643 }
644 
645 ssize_t
646 strfmon_l(char * __restrict s, size_t maxsize, locale_t loc,
647     const char * __restrict format, ...)
648 {
649 	size_t ret;
650 	va_list ap;
651 
652 	va_start(ap, format);
653 	ret = vstrfmon_l(s, maxsize, loc, format, ap);
654 	va_end(ap);
655 
656 	return (ret);
657 }
658