xref: /freebsd/lib/libc/stdlib/strfmon.c (revision 7899f917b1c0ea178f1d2be0cfb452086d079d23)
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
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 
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 					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 			/*
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_l((unsigned char)*++fmt, loc))
209 				goto format_error;
210 			GET_NUMBER(left_prec, loc);
211 			if ((unsigned int)left_prec >= maxsize - (dst - s))
212 				goto e2big_error;
213 		}
214 
215 		/* Right precision */
216 		if (*fmt == '.') {
217 			if (!isdigit_l((unsigned char)*++fmt, loc))
218 				goto format_error;
219 			GET_NUMBER(right_prec, loc);
220 			if ((unsigned int)right_prec >=
221 			    maxsize - (dst - s) - 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:	/*
234 				 * required character is missing or
235 				 * premature EOS
236 				 */
237 			goto format_error;
238 		}
239 
240 		if (currency_symbol != NULL)
241 			free(currency_symbol);
242 		if (flags & USE_INTL_CURRENCY) {
243 			currency_symbol = strdup(lc->int_curr_symbol);
244 			if (currency_symbol != NULL &&
245 			    strlen(currency_symbol) > 3) {
246 				space_char = currency_symbol[3];
247 				currency_symbol[3] = '\0';
248 			}
249 		} else
250 			currency_symbol = strdup(lc->currency_symbol);
251 
252 		if (currency_symbol == NULL)
253 			goto end_error;	/* ENOMEM. */
254 
255 		/* value itself */
256 		value = va_arg(ap, double);
257 
258 		/* detect sign */
259 		if (value < 0) {
260 			flags |= IS_NEGATIVE;
261 			value = -value;
262 		}
263 
264 		/* fill left_prec with amount of padding chars */
265 		if (left_prec >= 0) {
266 			pad_size = __calc_left_pad((flags ^ IS_NEGATIVE),
267 			    currency_symbol, lc) -
268 			    __calc_left_pad(flags, currency_symbol, lc);
269 			if (pad_size < 0)
270 				pad_size = 0;
271 		}
272 
273 		if (asciivalue != NULL)
274 			free(asciivalue);
275 		asciivalue = __format_grouped_double(value, &flags, left_prec,
276 		    right_prec, pad_char, lc, loc);
277 		if (asciivalue == NULL)
278 			goto end_error;	/*
279 					 * errno already set to ENOMEM by
280 					 * malloc()
281 					 */
282 
283 		/* set some variables for later use */
284 		__setup_vars(flags, &cs_precedes, &sep_by_space, &sign_posn,
285 		    &signstr, lc);
286 
287 		/*
288 		 * Description of some LC_MONETARY's values:
289 		 *
290 		 * p_cs_precedes & n_cs_precedes
291 		 *
292 		 * = 1 - $currency_symbol precedes the value
293 		 *       for a monetary quantity with a non-negative value
294 		 * = 0 - symbol succeeds the value
295 		 *
296 		 * p_sep_by_space & n_sep_by_space
297 		 *
298 		 * = 0 - no space separates $currency_symbol
299 		 *       from the value for a monetary quantity with a
300 		 *       non-negative value
301 		 * = 1 - space separates the symbol from the value
302 		 * = 2 - space separates the symbol and the sign string,
303 		 *       if adjacent; otherwise, a space separates
304 		 *       the sign string from the value
305 		 *
306 		 * p_sign_posn & n_sign_posn
307 		 *
308 		 * = 0 - parentheses enclose the quantity and the
309 		 *       $currency_symbol
310 		 * = 1 - the sign string precedes the quantity and the
311 		 *       $currency_symbol
312 		 * = 2 - the sign string succeeds the quantity and the
313 		 *       $currency_symbol
314 		 * = 3 - the sign string precedes the $currency_symbol
315 		 * = 4 - the sign string succeeds the $currency_symbol
316 		 */
317 
318 		tmpptr = dst;
319 
320 		while (pad_size-- > 0)
321 			PRINT(' ');
322 
323 		if (sign_posn == 0 && (flags & IS_NEGATIVE))
324 			PRINT('(');
325 
326 		if (cs_precedes == 1) {
327 			if (sign_posn == 1 || sign_posn == 3) {
328 				PRINTS(signstr);
329 				if (sep_by_space == 2)
330 					PRINT(' ');
331 			}
332 
333 			if (!(flags & SUPPRESS_CURR_SYMBOL)) {
334 				PRINTS(currency_symbol);
335 
336 				if (sign_posn == 4) {
337 					if (sep_by_space == 2)
338 						PRINT(space_char);
339 					PRINTS(signstr);
340 					if (sep_by_space == 1)
341 						PRINT(' ');
342 				} else if (sep_by_space == 1)
343 					PRINT(space_char);
344 			}
345 		} else if (sign_posn == 1) {
346 			PRINTS(signstr);
347 			if (sep_by_space == 2)
348 				PRINT(' ');
349 		}
350 
351 		PRINTS(asciivalue);
352 
353 		if (cs_precedes == 0) {
354 			if (sign_posn == 3) {
355 				if (sep_by_space == 1)
356 					PRINT(' ');
357 				PRINTS(signstr);
358 			}
359 
360 			if (!(flags & SUPPRESS_CURR_SYMBOL)) {
361 				if ((sign_posn == 3 && sep_by_space == 2) ||
362 				    (sep_by_space == 1 &&
363 				     (sign_posn == 0 || sign_posn == 1 ||
364 				      sign_posn == 2 || sign_posn == 4)))
365 					PRINT(space_char);
366 				PRINTS(currency_symbol);
367 				if (sign_posn == 4) {
368 					if (sep_by_space == 2)
369 						PRINT(' ');
370 					PRINTS(signstr);
371 				}
372 			}
373 		}
374 
375 		if (sign_posn == 2) {
376 			if (sep_by_space == 2)
377 				PRINT(' ');
378 			PRINTS(signstr);
379 		}
380 
381 		if (sign_posn == 0) {
382 			if (flags & IS_NEGATIVE)
383 				PRINT(')');
384 			else if (left_prec >= 0)
385 				PRINT(' ');
386 		}
387 
388 		if (dst - tmpptr < width) {
389 			if (flags & LEFT_JUSTIFY) {
390 				while (dst - tmpptr < width)
391 					PRINT(' ');
392 			} else {
393 				pad_size = dst - tmpptr;
394 				memmove(tmpptr + width - pad_size, tmpptr,
395 				    pad_size);
396 				memset(tmpptr, ' ', width - pad_size);
397 				dst += width - pad_size;
398 			}
399 		}
400 	}
401 
402 	PRINT('\0');
403 	free(asciivalue);
404 	free(currency_symbol);
405 	return (dst - s - 1);	/* size of put data except trailing '\0' */
406 
407 e2big_error:
408 	errno = E2BIG;
409 	goto end_error;
410 
411 format_error:
412 	errno = EINVAL;
413 
414 end_error:
415 	sverrno = errno;
416 	if (asciivalue != NULL)
417 		free(asciivalue);
418 	if (currency_symbol != NULL)
419 		free(currency_symbol);
420 	errno = sverrno;
421 	return (-1);
422 }
423 
424 static void
425 __setup_vars(int flags, char *cs_precedes, char *sep_by_space, char *sign_posn,
426     char **signstr, struct lconv *lc)
427 {
428 	if ((flags & IS_NEGATIVE) && (flags & USE_INTL_CURRENCY)) {
429 		*cs_precedes = lc->int_n_cs_precedes;
430 		*sep_by_space = lc->int_n_sep_by_space;
431 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->int_n_sign_posn;
432 		*signstr = (lc->negative_sign[0] == '\0') ? "-" :
433 		    lc->negative_sign;
434 	} else if (flags & USE_INTL_CURRENCY) {
435 		*cs_precedes = lc->int_p_cs_precedes;
436 		*sep_by_space = lc->int_p_sep_by_space;
437 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->int_p_sign_posn;
438 		*signstr = lc->positive_sign;
439 	} else if (flags & IS_NEGATIVE) {
440 		*cs_precedes = lc->n_cs_precedes;
441 		*sep_by_space = lc->n_sep_by_space;
442 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->n_sign_posn;
443 		*signstr = (lc->negative_sign[0] == '\0') ? "-" :
444 		    lc->negative_sign;
445 	} else {
446 		*cs_precedes = lc->p_cs_precedes;
447 		*sep_by_space = lc->p_sep_by_space;
448 		*sign_posn = (flags & PARENTH_POSN) ? 0 : lc->p_sign_posn;
449 		*signstr = lc->positive_sign;
450 	}
451 
452 	/* Set default values for unspecified information. */
453 	if (*cs_precedes != 0)
454 		*cs_precedes = 1;
455 	if (*sep_by_space == CHAR_MAX)
456 		*sep_by_space = 0;
457 	if (*sign_posn == CHAR_MAX)
458 		*sign_posn = 0;
459 }
460 
461 static int
462 __calc_left_pad(int flags, char *cur_symb, struct lconv *lc)
463 {
464 	char cs_precedes, sep_by_space, sign_posn, *signstr;
465 	int left_chars = 0;
466 
467 	__setup_vars(flags, &cs_precedes, &sep_by_space, &sign_posn, &signstr,
468 	    lc);
469 
470 	if (cs_precedes != 0) {
471 		left_chars += strlen(cur_symb);
472 		if (sep_by_space != 0)
473 			left_chars++;
474 	}
475 
476 	switch (sign_posn) {
477 	case 0:
478 		if (flags & IS_NEGATIVE)
479 			left_chars++;
480 		break;
481 	case 1:
482 		left_chars += strlen(signstr);
483 		break;
484 	case 3:
485 	case 4:
486 		if (cs_precedes != 0)
487 			left_chars += strlen(signstr);
488 	}
489 	return (left_chars);
490 }
491 
492 static int
493 get_groups(int size, char *grouping)
494 {
495 	int	chars = 0;
496 
497 	if (*grouping == CHAR_MAX || *grouping <= 0)	/* no grouping ? */
498 		return (0);
499 
500 	while (size > (int)*grouping) {
501 		chars++;
502 		size -= (int)*grouping++;
503 		/* no more grouping ? */
504 		if (*grouping == CHAR_MAX)
505 			break;
506 		/* rest grouping with same value ? */
507 		if (*grouping == 0) {
508 			chars += (size - 1) / *(grouping - 1);
509 			break;
510 		}
511 	}
512 	return (chars);
513 }
514 
515 /* convert double to locale-encoded string */
516 static char *
517 __format_grouped_double(double value, int *flags, int left_prec, int right_prec,
518     int pad_char, struct lconv *lc, locale_t loc)
519 {
520 
521 	char		*rslt;
522 	char		*avalue;
523 	int		avalue_size;
524 
525 	size_t		bufsize;
526 	char		*bufend;
527 
528 	int		padded;
529 
530 	char		*grouping;
531 	const char	*decimal_point;
532 	const char	*thousands_sep;
533 	size_t		decimal_point_size;
534 	size_t		thousands_sep_size;
535 
536 	int		groups = 0;
537 
538 	grouping = lc->mon_grouping;
539 	decimal_point = lc->mon_decimal_point;
540 	if (*decimal_point == '\0')
541 		decimal_point = lc->decimal_point;
542 	thousands_sep = lc->mon_thousands_sep;
543 	if (*thousands_sep == '\0')
544 		thousands_sep = lc->thousands_sep;
545 
546 	decimal_point_size = strlen(decimal_point);
547 	thousands_sep_size = strlen(thousands_sep);
548 
549 	/* fill left_prec with default value */
550 	if (left_prec == -1)
551 		left_prec = 0;
552 
553 	/* fill right_prec with default value */
554 	if (right_prec == -1) {
555 		if (*flags & USE_INTL_CURRENCY)
556 			right_prec = lc->int_frac_digits;
557 		else
558 			right_prec = lc->frac_digits;
559 
560 		if (right_prec == CHAR_MAX)	/* POSIX locale ? */
561 			right_prec = 2;
562 	}
563 
564 	if (*flags & NEED_GROUPING)
565 		left_prec += get_groups(left_prec, grouping);
566 
567 	/* convert to string */
568 	avalue_size = asprintf_l(&avalue, loc, "%*.*f",
569 	    left_prec + right_prec + 1, right_prec, value);
570 	if (avalue_size < 0)
571 		return (NULL);
572 
573 	/* make sure that we've enough space for result string */
574 	bufsize = avalue_size * (1 + thousands_sep_size) + decimal_point_size +
575 	    1;
576 	rslt = calloc(1, bufsize);
577 	if (rslt == NULL) {
578 		free(avalue);
579 		return (NULL);
580 	}
581 	bufend = rslt + bufsize - 1;	/* reserve space for trailing '\0' */
582 
583 	/* skip spaces at beginning */
584 	padded = 0;
585 	while (avalue[padded] == ' ') {
586 		padded++;
587 		avalue_size--;
588 	}
589 
590 	if (right_prec > 0) {
591 		bufend -= right_prec;
592 		memcpy(bufend, avalue + avalue_size + padded - right_prec,
593 		    right_prec);
594 		bufend -= decimal_point_size;
595 		memcpy(bufend, decimal_point, decimal_point_size);
596 		avalue_size -= (right_prec + 1);
597 	}
598 
599 	if ((*flags & NEED_GROUPING) && thousands_sep_size > 0 &&
600 	    *grouping != CHAR_MAX && *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 	ssize_t ret;
646 	va_list ap;
647 
648 	va_start(ap, format);
649 	ret = vstrfmon_l(s, maxsize, __get_locale(), format, ap);
650 	va_end(ap);
651 
652 	return (ret);
653 }
654 
655 ssize_t
656 strfmon_l(char *restrict s, size_t maxsize, locale_t loc,
657     const char *restrict format, ...)
658 {
659 	ssize_t ret;
660 	va_list ap;
661 
662 	va_start(ap, format);
663 	ret = vstrfmon_l(s, maxsize, loc, format, ap);
664 	va_end(ap);
665 
666 	return (ret);
667 }
668