Home | History | Annotate | Line # | Download | only in printf
printf.c revision 1.37.8.3
      1 /*	$NetBSD: printf.c,v 1.37.8.3 2018/09/23 17:16:33 martin Exp $	*/
      2 
      3 /*
      4  * Copyright (c) 1989, 1993
      5  *	The Regents of the University of California.  All rights reserved.
      6  *
      7  * Redistribution and use in source and binary forms, with or without
      8  * modification, are permitted provided that the following conditions
      9  * are met:
     10  * 1. Redistributions of source code must retain the above copyright
     11  *    notice, this list of conditions and the following disclaimer.
     12  * 2. Redistributions in binary form must reproduce the above copyright
     13  *    notice, this list of conditions and the following disclaimer in the
     14  *    documentation and/or other materials provided with the distribution.
     15  * 3. Neither the name of the University nor the names of its contributors
     16  *    may be used to endorse or promote products derived from this software
     17  *    without specific prior written permission.
     18  *
     19  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     29  * SUCH DAMAGE.
     30  */
     31 
     32 #include <sys/cdefs.h>
     33 #ifndef lint
     34 #if !defined(BUILTIN) && !defined(SHELL)
     35 __COPYRIGHT("@(#) Copyright (c) 1989, 1993\
     36  The Regents of the University of California.  All rights reserved.");
     37 #endif
     38 #endif
     39 
     40 #ifndef lint
     41 #if 0
     42 static char sccsid[] = "@(#)printf.c	8.2 (Berkeley) 3/22/95";
     43 #else
     44 __RCSID("$NetBSD: printf.c,v 1.37.8.3 2018/09/23 17:16:33 martin Exp $");
     45 #endif
     46 #endif /* not lint */
     47 
     48 #include <sys/types.h>
     49 
     50 #include <ctype.h>
     51 #include <err.h>
     52 #include <errno.h>
     53 #include <inttypes.h>
     54 #include <limits.h>
     55 #include <locale.h>
     56 #include <stdarg.h>
     57 #include <stdio.h>
     58 #include <stdlib.h>
     59 #include <string.h>
     60 #include <unistd.h>
     61 
     62 #ifdef __GNUC__
     63 #define ESCAPE '\e'
     64 #else
     65 #define ESCAPE 033
     66 #endif
     67 
     68 static void	 conv_escape_str(char *, void (*)(int), int);
     69 static char	*conv_escape(char *, char *, int);
     70 static char	*conv_expand(const char *);
     71 static char	 getchr(void);
     72 static double	 getdouble(void);
     73 static int	 getwidth(void);
     74 static intmax_t	 getintmax(void);
     75 static char	*getstr(void);
     76 static char	*mklong(const char *, char);
     77 static void      check_conversion(const char *, const char *);
     78 static void	 usage(void);
     79 
     80 static void	b_count(int);
     81 static void	b_output(int);
     82 static size_t	b_length;
     83 static char	*b_fmt;
     84 
     85 static int	rval;
     86 static char  **gargv;
     87 
     88 #ifdef BUILTIN		/* csh builtin */
     89 #define main progprintf
     90 #endif
     91 
     92 #ifdef SHELL		/* sh (aka ash) builtin */
     93 #define main printfcmd
     94 #include "../../bin/sh/bltin/bltin.h"
     95 #endif /* SHELL */
     96 
     97 #define PF(f, func) { \
     98 	if (fieldwidth != -1) { \
     99 		if (precision != -1) \
    100 			error = printf(f, fieldwidth, precision, func); \
    101 		else \
    102 			error = printf(f, fieldwidth, func); \
    103 	} else if (precision != -1) \
    104 		error = printf(f, precision, func); \
    105 	else \
    106 		error = printf(f, func); \
    107 }
    108 
    109 #define APF(cpp, f, func) { \
    110 	if (fieldwidth != -1) { \
    111 		if (precision != -1) \
    112 			error = asprintf(cpp, f, fieldwidth, precision, func); \
    113 		else \
    114 			error = asprintf(cpp, f, fieldwidth, func); \
    115 	} else if (precision != -1) \
    116 		error = asprintf(cpp, f, precision, func); \
    117 	else \
    118 		error = asprintf(cpp, f, func); \
    119 }
    120 
    121 #ifdef main
    122 int main(int, char *[]);
    123 #endif
    124 int main(int argc, char *argv[])
    125 {
    126 	char *fmt, *start;
    127 	int fieldwidth, precision;
    128 	char nextch;
    129 	char *format;
    130 	char ch;
    131 	int error, o;
    132 
    133 #if !defined(SHELL) && !defined(BUILTIN)
    134 	(void)setlocale (LC_ALL, "");
    135 #endif
    136 
    137 	rval = 0;	/* clear for builtin versions (avoid holdover) */
    138 
    139 	while ((o = getopt(argc, argv, "")) != -1) {
    140 		switch (o) {
    141 		case '?':
    142 		default:
    143 			usage();
    144 			return 1;
    145 		}
    146 	}
    147 	argc -= optind;
    148 	argv += optind;
    149 
    150 	if (argc < 1) {
    151 		usage();
    152 		return 1;
    153 	}
    154 
    155 	format = *argv;
    156 	gargv = ++argv;
    157 
    158 #define SKIP1	"#-+ 0'"
    159 #define SKIP2	"0123456789"
    160 	do {
    161 		/*
    162 		 * Basic algorithm is to scan the format string for conversion
    163 		 * specifications -- once one is found, find out if the field
    164 		 * width or precision is a '*'; if it is, gather up value.
    165 		 * Note, format strings are reused as necessary to use up the
    166 		 * provided arguments, arguments of zero/null string are
    167 		 * provided to use up the format string.
    168 		 */
    169 
    170 		/* find next format specification */
    171 		for (fmt = format; (ch = *fmt++) != '\0';) {
    172 			if (ch == '\\') {
    173 				char c_ch;
    174 				fmt = conv_escape(fmt, &c_ch, 0);
    175 				putchar(c_ch);
    176 				continue;
    177 			}
    178 			if (ch != '%' || (*fmt == '%' && ++fmt)) {
    179 				(void)putchar(ch);
    180 				continue;
    181 			}
    182 
    183 			/* Ok - we've found a format specification,
    184 			   Save its address for a later printf(). */
    185 			start = fmt - 1;
    186 
    187 			/* skip to field width */
    188 			fmt += strspn(fmt, SKIP1);
    189 			if (*fmt == '*') {
    190 				fmt++;
    191 				fieldwidth = getwidth();
    192 			} else
    193 				fieldwidth = -1;
    194 
    195 			/* skip to possible '.', get following precision */
    196 			fmt += strspn(fmt, SKIP2);
    197 			if (*fmt == '.') {
    198 				fmt++;
    199 				if (*fmt == '*') {
    200 					fmt++;
    201 					precision = getwidth();
    202 				} else
    203 					precision = -1;
    204 			} else
    205 				precision = -1;
    206 
    207 			fmt += strspn(fmt, SKIP2);
    208 
    209 			ch = *fmt;
    210 			if (!ch) {
    211 				warnx("missing format character");
    212 				return (1);
    213 			}
    214 			/* null terminate format string to we can use it
    215 			   as an argument to printf. */
    216 			nextch = fmt[1];
    217 			fmt[1] = 0;
    218 			switch (ch) {
    219 
    220 			case 'B': {
    221 				const char *p = conv_expand(getstr());
    222 				if (p == NULL)
    223 					goto out;
    224 				*fmt = 's';
    225 				PF(start, p);
    226 				if (error < 0)
    227 					goto out;
    228 				break;
    229 			}
    230 			case 'b': {
    231 				/* There has to be a better way to do this,
    232 				 * but the string we generate might have
    233 				 * embedded nulls. */
    234 				static char *a, *t;
    235 				char *cp = getstr();
    236 				/* Free on entry in case shell longjumped out */
    237 				if (a != NULL)
    238 					free(a);
    239 				a = NULL;
    240 				if (t != NULL)
    241 					free(t);
    242 				t = NULL;
    243 				/* Count number of bytes we want to output */
    244 				b_length = 0;
    245 				conv_escape_str(cp, b_count, 0);
    246 				t = malloc(b_length + 1);
    247 				if (t == NULL)
    248 					goto out;
    249 				(void)memset(t, 'x', b_length);
    250 				t[b_length] = 0;
    251 				/* Get printf to calculate the lengths */
    252 				*fmt = 's';
    253 				APF(&a, start, t);
    254 				if (error == -1)
    255 					goto out;
    256 				b_fmt = a;
    257 				/* Output leading spaces and data bytes */
    258 				conv_escape_str(cp, b_output, 1);
    259 				/* Add any trailing spaces */
    260 				printf("%s", b_fmt);
    261 				break;
    262 			}
    263 			case 'c': {
    264 				char p = getchr();
    265 				PF(start, p);
    266 				if (error < 0)
    267 					goto out;
    268 				break;
    269 			}
    270 			case 's': {
    271 				char *p = getstr();
    272 				PF(start, p);
    273 				if (error < 0)
    274 					goto out;
    275 				break;
    276 			}
    277 			case 'd':
    278 			case 'i': {
    279 				intmax_t p = getintmax();
    280 				char *f = mklong(start, ch);
    281 				PF(f, p);
    282 				if (error < 0)
    283 					goto out;
    284 				break;
    285 			}
    286 			case 'o':
    287 			case 'u':
    288 			case 'x':
    289 			case 'X': {
    290 				uintmax_t p = (uintmax_t)getintmax();
    291 				char *f = mklong(start, ch);
    292 				PF(f, p);
    293 				if (error < 0)
    294 					goto out;
    295 				break;
    296 			}
    297 			case 'e':
    298 			case 'E':
    299 			case 'f':
    300 			case 'g':
    301 			case 'G': {
    302 				double p = getdouble();
    303 				PF(start, p);
    304 				if (error < 0)
    305 					goto out;
    306 				break;
    307 			}
    308 			default:
    309 				warnx("%s: invalid directive", start);
    310 				return 1;
    311 			}
    312 			*fmt++ = ch;
    313 			*fmt = nextch;
    314 			/* escape if a \c was encountered */
    315 			if (rval & 0x100)
    316 				return rval & ~0x100;
    317 		}
    318 	} while (gargv != argv && *gargv);
    319 
    320 	return rval & ~0x100;
    321 out:
    322 	warn("print failed");
    323 	return 1;
    324 }
    325 
    326 /* helper functions for conv_escape_str */
    327 
    328 static void
    329 /*ARGSUSED*/
    330 b_count(int ch)
    331 {
    332 	b_length++;
    333 }
    334 
    335 /* Output one converted character for every 'x' in the 'format' */
    336 
    337 static void
    338 b_output(int ch)
    339 {
    340 	for (;;) {
    341 		switch (*b_fmt++) {
    342 		case 0:
    343 			b_fmt--;
    344 			return;
    345 		case ' ':
    346 			putchar(' ');
    347 			break;
    348 		default:
    349 			putchar(ch);
    350 			return;
    351 		}
    352 	}
    353 }
    354 
    355 
    356 /*
    357  * Print SysV echo(1) style escape string
    358  *	Halts processing string if a \c escape is encountered.
    359  */
    360 static void
    361 conv_escape_str(char *str, void (*do_putchar)(int), int quiet)
    362 {
    363 	int value;
    364 	int ch;
    365 	char c;
    366 
    367 	while ((ch = *str++) != '\0') {
    368 		if (ch != '\\') {
    369 			do_putchar(ch);
    370 			continue;
    371 		}
    372 
    373 		ch = *str++;
    374 		if (ch == 'c') {
    375 			/* \c as in SYSV echo - abort all processing.... */
    376 			rval |= 0x100;
    377 			break;
    378 		}
    379 
    380 		/*
    381 		 * %b string octal constants are not like those in C.
    382 		 * They start with a \0, and are followed by 0, 1, 2,
    383 		 * or 3 octal digits.
    384 		 */
    385 		if (ch == '0') {
    386 			int octnum = 0, i;
    387 			for (i = 0; i < 3; i++) {
    388 				if (!isdigit((unsigned char)*str) || *str > '7')
    389 					break;
    390 				octnum = (octnum << 3) | (*str++ - '0');
    391 			}
    392 			do_putchar(octnum);
    393 			continue;
    394 		}
    395 
    396 		/* \[M][^|-]C as defined by vis(3) */
    397 		if (ch == 'M' && *str == '-') {
    398 			do_putchar(0200 | str[1]);
    399 			str += 2;
    400 			continue;
    401 		}
    402 		if (ch == 'M' && *str == '^') {
    403 			str++;
    404 			value = 0200;
    405 			ch = '^';
    406 		} else
    407 			value = 0;
    408 		if (ch == '^') {
    409 			ch = *str++;
    410 			if (ch == '?')
    411 				value |= 0177;
    412 			else
    413 				value |= ch & 037;
    414 			do_putchar(value);
    415 			continue;
    416 		}
    417 
    418 		/* Finally test for sequences valid in the format string */
    419 		str = conv_escape(str - 1, &c, quiet);
    420 		do_putchar(c);
    421 	}
    422 }
    423 
    424 /*
    425  * Print "standard" escape characters
    426  */
    427 static char *
    428 conv_escape(char *str, char *conv_ch, int quiet)
    429 {
    430 	char value;
    431 	char ch;
    432 	char num_buf[4], *num_end;
    433 
    434 	ch = *str++;
    435 
    436 	switch (ch) {
    437 	case '\0':
    438 		if (!quiet)
    439 			warnx("incomplete escape sequence");
    440 		rval = 1;
    441 		value = '\\';
    442 		--str;
    443 		break;
    444 
    445 	case '0': case '1': case '2': case '3':
    446 	case '4': case '5': case '6': case '7':
    447 		num_buf[0] = ch;
    448 		ch = str[0];
    449 		num_buf[1] = ch;
    450 		num_buf[2] = (char)(ch != '\0' ? str[1] : '\0');
    451 		num_buf[3] = '\0';
    452 		value = (char)strtoul(num_buf, &num_end, 8);
    453 		str += num_end  - (num_buf + 1);
    454 		break;
    455 
    456 	case 'x':
    457 		/* Hexadecimal character constants are not required to be
    458 		   supported (by SuS v1) because there is no consistent
    459 		   way to detect the end of the constant.
    460 		   Supporting 2 byte constants is a compromise. */
    461 		ch = str[0];
    462 		num_buf[0] = ch;
    463 		num_buf[1] = (char)(ch != '\0' ? str[1] : '\0');
    464 		num_buf[2] = '\0';
    465 		value = (char)strtoul(num_buf, &num_end, 16);
    466 		str += num_end - num_buf;
    467 		break;
    468 
    469 	case '\\':	value = '\\';	break;	/* backslash */
    470 	case '\'':	value = '\'';	break;	/* single quote */
    471 	case '"':	value = '"';	break;	/* double quote */
    472 	case 'a':	value = '\a';	break;	/* alert */
    473 	case 'b':	value = '\b';	break;	/* backspace */
    474 	case 'e':	value = ESCAPE;	break;	/* escape */
    475 	case 'f':	value = '\f';	break;	/* form-feed */
    476 	case 'n':	value = '\n';	break;	/* newline */
    477 	case 'r':	value = '\r';	break;	/* carriage-return */
    478 	case 't':	value = '\t';	break;	/* tab */
    479 	case 'v':	value = '\v';	break;	/* vertical-tab */
    480 
    481 	default:
    482 		if (!quiet)
    483 			warnx("unknown escape sequence `\\%c'", ch);
    484 		rval = 1;
    485 		value = ch;
    486 		break;
    487 	}
    488 
    489 	*conv_ch = value;
    490 	return str;
    491 }
    492 
    493 /* expand a string so that everything is printable */
    494 
    495 static char *
    496 conv_expand(const char *str)
    497 {
    498 	static char *conv_str;
    499 	char *cp;
    500 	char ch;
    501 
    502 	if (conv_str)
    503 		free(conv_str);
    504 	/* get a buffer that is definitely large enough.... */
    505 	conv_str = malloc(4 * strlen(str) + 1);
    506 	if (!conv_str)
    507 		return NULL;
    508 	cp = conv_str;
    509 
    510 	while ((ch = *(const char *)str++) != '\0') {
    511 		switch (ch) {
    512 		/* Use C escapes for expected control characters */
    513 		case '\\':	ch = '\\';	break;	/* backslash */
    514 		case '\'':	ch = '\'';	break;	/* single quote */
    515 		case '"':	ch = '"';	break;	/* double quote */
    516 		case '\a':	ch = 'a';	break;	/* alert */
    517 		case '\b':	ch = 'b';	break;	/* backspace */
    518 		case ESCAPE:	ch = 'e';	break;	/* escape */
    519 		case '\f':	ch = 'f';	break;	/* form-feed */
    520 		case '\n':	ch = 'n';	break;	/* newline */
    521 		case '\r':	ch = 'r';	break;	/* carriage-return */
    522 		case '\t':	ch = 't';	break;	/* tab */
    523 		case '\v':	ch = 'v';	break;	/* vertical-tab */
    524 		default:
    525 			/* Copy anything printable */
    526 			if (isprint((unsigned char)ch)) {
    527 				*cp++ = ch;
    528 				continue;
    529 			}
    530 			/* Use vis(3) encodings for the rest */
    531 			*cp++ = '\\';
    532 			if (ch & 0200) {
    533 				*cp++ = 'M';
    534 				ch &= (char)~0200;
    535 			}
    536 			if (ch == 0177) {
    537 				*cp++ = '^';
    538 				*cp++ = '?';
    539 				continue;
    540 			}
    541 			if (ch < 040) {
    542 				*cp++ = '^';
    543 				*cp++ = ch | 0100;
    544 				continue;
    545 			}
    546 			*cp++ = '-';
    547 			*cp++ = ch;
    548 			continue;
    549 		}
    550 		*cp++ = '\\';
    551 		*cp++ = ch;
    552 	}
    553 
    554 	*cp = 0;
    555 	return conv_str;
    556 }
    557 
    558 static char *
    559 mklong(const char *str, char ch)
    560 {
    561 	static char copy[64];
    562 	size_t len;
    563 
    564 	len = strlen(str) + 2;
    565 	if (len > sizeof copy) {
    566 		warnx("format %s too complex", str);
    567 		len = 4;
    568 	}
    569 	(void)memmove(copy, str, len - 3);
    570 	copy[len - 3] = 'j';
    571 	copy[len - 2] = ch;
    572 	copy[len - 1] = '\0';
    573 	return copy;
    574 }
    575 
    576 static char
    577 getchr(void)
    578 {
    579 	if (!*gargv)
    580 		return 0;
    581 	return **gargv++;
    582 }
    583 
    584 static char *
    585 getstr(void)
    586 {
    587 	static char empty[] = "";
    588 	if (!*gargv)
    589 		return empty;
    590 	return *gargv++;
    591 }
    592 
    593 static int
    594 getwidth(void)
    595 {
    596 	unsigned long val;
    597 	char *s, *ep;
    598 
    599 	s = *gargv;
    600 	if (!*gargv)
    601 		return (0);
    602 	gargv++;
    603 
    604 	errno = 0;
    605 	val = strtoul(s, &ep, 0);
    606 	check_conversion(s, ep);
    607 
    608 	/* Arbitrarily 'restrict' field widths to 1Mbyte */
    609 	if (val > 1 << 20) {
    610 		warnx("%s: invalid field width", s);
    611 		return 0;
    612 	}
    613 
    614 	return (int)val;
    615 }
    616 
    617 static intmax_t
    618 getintmax(void)
    619 {
    620 	intmax_t val;
    621 	char *cp, *ep;
    622 
    623 	cp = *gargv;
    624 	if (cp == NULL)
    625 		return 0;
    626 	gargv++;
    627 
    628 	if (*cp == '\"' || *cp == '\'')
    629 		return *(cp + 1);
    630 
    631 	errno = 0;
    632 	val = strtoimax(cp, &ep, 0);
    633 	check_conversion(cp, ep);
    634 	return val;
    635 }
    636 
    637 static double
    638 getdouble(void)
    639 {
    640 	double val;
    641 	char *ep;
    642 
    643 	if (!*gargv)
    644 		return (0.0);
    645 
    646 	if (**gargv == '\"' || **gargv == '\'')
    647 		return (double) *((*gargv++)+1);
    648 
    649 	errno = 0;
    650 	val = strtod(*gargv, &ep);
    651 	check_conversion(*gargv++, ep);
    652 	return val;
    653 }
    654 
    655 static void
    656 check_conversion(const char *s, const char *ep)
    657 {
    658 	if (*ep) {
    659 		if (ep == s)
    660 			warnx("%s: expected numeric value", s);
    661 		else
    662 			warnx("%s: not completely converted", s);
    663 		rval = 1;
    664 	} else if (errno == ERANGE) {
    665 		warnx("%s: %s", s, strerror(ERANGE));
    666 		rval = 1;
    667 	}
    668 }
    669 
    670 static void
    671 usage(void)
    672 {
    673 	(void)fprintf(stderr, "Usage: %s format [arg ...]\n", getprogname());
    674 }
    675