xref: /freebsd/lib/libc/stdlib/strfmon.c (revision dbd5678d)
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, LOC) do {					\
75 	VAR = 0;						\
76 	while (isdigit_l((unsigned char)*fmt, LOC)) {			\
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 **, struct lconv *);
102 static int __calc_left_pad(int, char *, struct lconv *);
103 static char *__format_grouped_double(double, int *, int, int, int,
104     struct lconv *, locale_t);
105 
106 static ssize_t
107 vstrfmon_l(char * __restrict s, size_t maxsize, locale_t loc,
108     const char * __restrict format, va_list ap)
109 {
110 	char		*dst;		/* output destination pointer */
111 	const char	*fmt;		/* current format poistion pointer */
112 	struct lconv	*lc;		/* pointer to lconv structure */
113 	char		*asciivalue;	/* formatted double pointer */
114 
115 	int		flags;		/* formatting options */
116 	int		pad_char;	/* padding character */
117 	int		pad_size;	/* pad size */
118 	int		width;		/* field width */
119 	int		left_prec;	/* left precision */
120 	int		right_prec;	/* right precision */
121 	double		value;		/* just value */
122 	char		space_char = ' '; /* space after currency */
123 
124 	char		cs_precedes,	/* values gathered from struct lconv */
125 			sep_by_space,
126 			sign_posn,
127 			*signstr,
128 			*currency_symbol;
129 
130 	char		*tmpptr;	/* temporary vars */
131 	int		sverrno;
132 	FIX_LOCALE(loc);
133 
134 	lc = localeconv_l(loc);
135 	dst = s;
136 	fmt = format;
137 	asciivalue = NULL;
138 	currency_symbol = NULL;
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 		pad_size = 0;		/* no padding initially */
159 		left_prec = -1;		/* no left precision specified */
160 		right_prec = -1;	/* no right precision specified */
161 		width = -1;		/* no width specified */
162 		value = 0;		/* we have no value to print now */
163 
164 		/* Flags */
165 		while (1) {
166 			switch (*++fmt) {
167 				case '=':	/* fill character */
168 					pad_char = *++fmt;
169 					if (pad_char == '\0')
170 						goto format_error;
171 					continue;
172 				case '^':	/* not group currency  */
173 					flags &= ~(NEED_GROUPING);
174 					continue;
175 				case '+':	/* use locale defined signs */
176 					if (flags & SIGN_POSN_USED)
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 			/* Do we have enough space to put number with
201 			 * required width ?
202 			 */
203 			if ((unsigned int)width >= maxsize - (dst - s))
204 				goto e2big_error;
205 		}
206 
207 		/* Left precision */
208 		if (*fmt == '#') {
209 			if (!isdigit_l((unsigned char)*++fmt, loc))
210 				goto format_error;
211 			GET_NUMBER(left_prec, loc);
212 			if ((unsigned int)left_prec >= maxsize - (dst - s))
213 				goto e2big_error;
214 		}
215 
216 		/* Right precision */
217 		if (*fmt == '.') {
218 			if (!isdigit_l((unsigned char)*++fmt, loc))
219 				goto format_error;
220 			GET_NUMBER(right_prec, loc);
221 			if ((unsigned int)right_prec >= maxsize - (dst - s) -
222 			    left_prec)
223 				goto e2big_error;
224 		}
225 
226 		/* Conversion Characters */
227 		switch (*fmt++) {
228 			case 'i':	/* use international currency format */
229 				flags |= USE_INTL_CURRENCY;
230 				break;
231 			case 'n':	/* use national currency format */
232 				flags &= ~(USE_INTL_CURRENCY);
233 				break;
234 			default:	/* required character is missing or
235 					   premature EOS */
236 				goto format_error;
237 		}
238 
239 		if (currency_symbol != NULL)
240 			free(currency_symbol);
241 		if (flags & USE_INTL_CURRENCY) {
242 			currency_symbol = strdup(lc->int_curr_symbol);
243 			if (currency_symbol != NULL &&
244 			    strlen(currency_symbol) > 3) {
245 				space_char = currency_symbol[3];
246 				currency_symbol[3] = '\0';
247 			}
248 		} else
249 			currency_symbol = strdup(lc->currency_symbol);
250 
251 		if (currency_symbol == NULL)
252 			goto end_error;			/* ENOMEM. */
253 
254 		/* value itself */
255 		value = va_arg(ap, double);
256 
257 		/* detect sign */
258 		if (value < 0) {
259 			flags |= IS_NEGATIVE;
260 			value = -value;
261 		}
262 
263 		/* fill left_prec with amount of padding chars */
264 		if (left_prec >= 0) {
265 			pad_size = __calc_left_pad((flags ^ IS_NEGATIVE),
266 			    currency_symbol, lc) -
267 			    __calc_left_pad(flags, currency_symbol, lc);
268 			if (pad_size < 0)
269 				pad_size = 0;
270 		}
271 
272 		if (asciivalue != NULL)
273 			free(asciivalue);
274 		asciivalue = __format_grouped_double(value, &flags,
275 		    left_prec, right_prec, pad_char, lc, loc);
276 		if (asciivalue == NULL)
277 			goto end_error;		/* errno already set     */
278 						/* to ENOMEM by malloc() */
279 
280 		/* set some variables for later use */
281 		__setup_vars(flags, &cs_precedes, &sep_by_space,
282 		    &sign_posn, &signstr, lc);
283 
284 		/*
285 		 * Description of some LC_MONETARY's values:
286 		 *
287 		 * p_cs_precedes & n_cs_precedes
288 		 *
289 		 * = 1 - $currency_symbol precedes the value
290 		 *       for a monetary quantity with a non-negative value
291 		 * = 0 - symbol succeeds the value
292 		 *
293 		 * p_sep_by_space & n_sep_by_space
294 		 *
295 		 * = 0 - no space separates $currency_symbol
296 		 *       from the value for a monetary quantity with a
297 		 *       non-negative value
298 		 * = 1 - space separates the symbol from the value
299 		 * = 2 - space separates the symbol and the sign string,
300 		 *       if adjacent; otherwise, a space separates
301 		 *       the sign string from the value
302 		 *
303 		 * p_sign_posn & n_sign_posn
304 		 *
305 		 * = 0 - parentheses enclose the quantity and the
306 		 *       $currency_symbol
307 		 * = 1 - the sign string precedes the quantity and the
308 		 *       $currency_symbol
309 		 * = 2 - the sign string succeeds the quantity and the
310 		 *       $currency_symbol
311 		 * = 3 - the sign string precedes the $currency_symbol
312 		 * = 4 - the sign string succeeds the $currency_symbol
313 		 *
314 		 */
315 
316 		tmpptr = dst;
317 
318 		while (pad_size-- > 0)
319 			PRINT(' ');
320 
321 		if (sign_posn == 0 && (flags & IS_NEGATIVE))
322 			PRINT('(');
323 
324 		if (cs_precedes == 1) {
325 			if (sign_posn == 1 || sign_posn == 3) {
326 				PRINTS(signstr);
327 				if (sep_by_space == 2)
328 					PRINT(' ');
329 			}
330 
331 			if (!(flags & SUPPRESS_CURR_SYMBOL)) {
332 				PRINTS(currency_symbol);
333 
334 				if (sign_posn == 4) {
335 					if (sep_by_space == 2)
336 						PRINT(space_char);
337 					PRINTS(signstr);
338 					if (sep_by_space == 1)
339 						PRINT(' ');
340 				} else if (sep_by_space == 1)
341 					PRINT(space_char);
342 			}
343 		} else if (sign_posn == 1) {
344 			PRINTS(signstr);
345 			if (sep_by_space == 2)
346 				PRINT(' ');
347 		}
348 
349 		PRINTS(asciivalue);
350 
351 		if (cs_precedes == 0) {
352 			if (sign_posn == 3) {
353 				if (sep_by_space == 1)
354 					PRINT(' ');
355 				PRINTS(signstr);
356 			}
357 
358 			if (!(flags & SUPPRESS_CURR_SYMBOL)) {
359 				if ((sign_posn == 3 && sep_by_space == 2)
360 				    || (sep_by_space == 1
361 				    && (sign_posn == 0
362 				    || sign_posn == 1
363 				    || sign_posn == 2
364 				    || 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);	/* return 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,
426     char *sign_posn, 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,
468 	    &signstr, 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,
518     int left_prec, int right_prec, 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) &&
600 	    thousands_sep_size > 0 &&
601 	    *grouping != CHAR_MAX &&
602 	    *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
645 strfmon(char * __restrict s, size_t maxsize, const char * __restrict format,
646     ...)
647 {
648 	size_t ret;
649 	va_list ap;
650 
651 	va_start(ap, format);
652 	ret = vstrfmon_l(s, maxsize, __get_locale(), format, ap);
653 	va_end(ap);
654 
655 	return (ret);
656 }
657 
658 ssize_t
659 strfmon_l(char * __restrict s, size_t maxsize, locale_t loc,
660     const char * __restrict format, ...)
661 {
662 	size_t ret;
663 	va_list ap;
664 
665 	va_start(ap, format);
666 	ret = vstrfmon_l(s, maxsize, loc, format, ap);
667 	va_end(ap);
668 
669 	return (ret);
670 }
671