xref: /netbsd-src/usr.bin/printf/printf.c (revision 53b02e147d4ed531c0d2a5ca9b3e8026ba3e99b5)
1 /*	$NetBSD: printf.c,v 1.54 2021/05/20 02:01:07 christos 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 1.54 2021/05/20 02:01:07 christos 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 intmax_t	 wide_char(const char *);
78 static void      check_conversion(const char *, const char *);
79 static void	 usage(void);
80 
81 static void	b_count(int);
82 static void	b_output(int);
83 static size_t	b_length;
84 static char	*b_fmt;
85 
86 static int	rval;
87 static char  **gargv;
88 
89 #ifdef BUILTIN		/* csh builtin */
90 #define main progprintf
91 #endif
92 
93 #ifdef SHELL		/* sh (aka ash) builtin */
94 #define main printfcmd
95 #include "../../bin/sh/bltin/bltin.h"
96 #endif /* SHELL */
97 
98 #define PF(f, func) { \
99 	if (fieldwidth != -1) { \
100 		if (precision != -1) \
101 			error = printf(f, fieldwidth, precision, func); \
102 		else \
103 			error = printf(f, fieldwidth, func); \
104 	} else if (precision != -1) \
105 		error = printf(f, precision, func); \
106 	else \
107 		error = printf(f, func); \
108 }
109 
110 #define APF(cpp, f, func) { \
111 	if (fieldwidth != -1) { \
112 		if (precision != -1) \
113 			error = asprintf(cpp, f, fieldwidth, precision, func); \
114 		else \
115 			error = asprintf(cpp, f, fieldwidth, func); \
116 	} else if (precision != -1) \
117 		error = asprintf(cpp, f, precision, func); \
118 	else \
119 		error = asprintf(cpp, f, func); \
120 }
121 
122 #define isodigit(c)	((c) >= '0' && (c) <= '7')
123 #define octtobin(c)	((c) - '0')
124 #define check(c, a)	(c) >= (a) && (c) <= (a) + 5 ? (c) - (a) + 10
125 #define hextobin(c)	(check(c, 'a') : check(c, 'A') : (c) - '0')
126 #ifdef main
127 int main(int, char *[]);
128 #endif
129 
130 int
131 main(int argc, char *argv[])
132 {
133 	char *fmt, *start;
134 	int fieldwidth, precision;
135 	char nextch;
136 	char *format;
137 	char ch;
138 	int error;
139 
140 #if !defined(SHELL) && !defined(BUILTIN)
141 	(void)setlocale (LC_ALL, "");
142 #endif
143 
144 	rval = 0;	/* clear for builtin versions (avoid holdover) */
145 	clearerr(stdout);	/* for the builtin version */
146 
147 	/*
148 	 * printf does not comply with Posix XBD 12.2 - there are no opts,
149 	 * not even the -- end of options marker.   Do not run getoot().
150 	 */
151 	if (argc > 2 && strchr(argv[1], '%') == NULL) {
152 		int o;
153 
154 		/*
155 		 * except that if there are multiple args and
156 		 * the first (the nominal format) contains no '%'
157 		 * conversions (which we will approximate as no '%'
158 		 * characters at all, conversions or not) then the
159 		 * results are unspecified, and we can do what we
160 		 * like.   So in that case, for some backward compat
161 		 * to scripts which (stupidly) do:
162 		 *	printf -- format args
163 		 * process this case the old way.
164 		 */
165 
166 		while ((o = getopt(argc, argv, "")) != -1) {
167 			switch (o) {
168 			case '?':
169 			default:
170 				usage();
171 				return 1;
172 			}
173 		}
174 		argc -= optind;
175 		argv += optind;
176 	} else {
177 		argc -= 1;	/* drop argv[0] (the program name) */
178 		argv += 1;
179 	}
180 
181 	if (argc < 1) {
182 		usage();
183 		return 1;
184 	}
185 
186 	format = *argv;
187 	gargv = ++argv;
188 
189 #define SKIP1	"#-+ 0'"
190 #define SKIP2	"0123456789"
191 	do {
192 		/*
193 		 * Basic algorithm is to scan the format string for conversion
194 		 * specifications -- once one is found, find out if the field
195 		 * width or precision is a '*'; if it is, gather up value.
196 		 * Note, format strings are reused as necessary to use up the
197 		 * provided arguments, arguments of zero/null string are
198 		 * provided to use up the format string.
199 		 */
200 
201 		/* find next format specification */
202 		for (fmt = format; (ch = *fmt++) != '\0';) {
203 			if (ch == '\\') {
204 				char c_ch;
205 				fmt = conv_escape(fmt, &c_ch, 0);
206 				putchar(c_ch);
207 				continue;
208 			}
209 			if (ch != '%' || (*fmt == '%' && ++fmt)) {
210 				(void)putchar(ch);
211 				continue;
212 			}
213 
214 			/*
215 			 * Ok - we've found a format specification,
216 			 * Save its address for a later printf().
217 			 */
218 			start = fmt - 1;
219 
220 			/* skip to field width */
221 			fmt += strspn(fmt, SKIP1);
222 			if (*fmt == '*') {
223 				fmt++;
224 				fieldwidth = getwidth();
225 			} else {
226 				fieldwidth = -1;
227 
228 				/* skip to possible '.' for precision */
229 				fmt += strspn(fmt, SKIP2);
230 			}
231 
232 			if (*fmt == '.') {
233 				 /* get following precision */
234 				fmt++;
235 				if (*fmt == '*') {
236 					fmt++;
237 					precision = getwidth();
238 				} else {
239 					precision = -1;
240 					fmt += strspn(fmt, SKIP2);
241 				}
242 			} else
243 				precision = -1;
244 
245 			ch = *fmt;
246 			if (!ch) {
247 				warnx("%s: missing format character", start);
248 				return 1;
249 			}
250 
251 			/*
252 			 * null terminate format string to we can use it
253 			 * as an argument to printf.
254 			 */
255 			nextch = fmt[1];
256 			fmt[1] = 0;
257 
258 			switch (ch) {
259 
260 			case 'B': {
261 				const char *p = conv_expand(getstr());
262 
263 				if (p == NULL)
264 					goto out;
265 				*fmt = 's';
266 				PF(start, p);
267 				if (error < 0)
268 					goto out;
269 				break;
270 			}
271 			case 'b': {
272 				/*
273 				 * There has to be a better way to do this,
274 				 * but the string we generate might have
275 				 * embedded nulls
276 				 */
277 				static char *a, *t;
278 				char *cp = getstr();
279 
280 				/* Free on entry in case shell longjumped out */
281 				if (a != NULL)
282 					free(a);
283 				a = NULL;
284 				if (t != NULL)
285 					free(t);
286 				t = NULL;
287 
288 				/* Count number of bytes we want to output */
289 				b_length = 0;
290 				conv_escape_str(cp, b_count, 0);
291 				t = malloc(b_length + 1);
292 				if (t == NULL)
293 					goto out;
294 				(void)memset(t, 'x', b_length);
295 				t[b_length] = 0;
296 
297 				/* Get printf to calculate the lengths */
298 				*fmt = 's';
299 				APF(&a, start, t);
300 				if (error == -1)
301 					goto out;
302 				b_fmt = a;
303 
304 				/* Output leading spaces and data bytes */
305 				conv_escape_str(cp, b_output, 1);
306 
307 				/* Add any trailing spaces */
308 				printf("%s", b_fmt);
309 				break;
310 			}
311 			case 'c': {
312 				char p = getchr();
313 
314 				PF(start, p);
315 				if (error < 0)
316 					goto out;
317 				break;
318 			}
319 			case 's': {
320 				char *p = getstr();
321 
322 				PF(start, p);
323 				if (error < 0)
324 					goto out;
325 				break;
326 			}
327 			case 'd':
328 			case 'i': {
329 				intmax_t p = getintmax();
330 				char *f = mklong(start, ch);
331 
332 				PF(f, p);
333 				if (error < 0)
334 					goto out;
335 				break;
336 			}
337 			case 'o':
338 			case 'u':
339 			case 'x':
340 			case 'X': {
341 				uintmax_t p = (uintmax_t)getintmax();
342 				char *f = mklong(start, ch);
343 
344 				PF(f, p);
345 				if (error < 0)
346 					goto out;
347 				break;
348 			}
349 			case 'a':
350 			case 'A':
351 			case 'e':
352 			case 'E':
353 			case 'f':
354 			case 'F':
355 			case 'g':
356 			case 'G': {
357 				double p = getdouble();
358 
359 				PF(start, p);
360 				if (error < 0)
361 					goto out;
362 				break;
363 			}
364 			case '%':
365 				/* Don't ask, but this is useful ... */
366 				if (fieldwidth == 'N' && precision == 'B')
367 					return 0;
368 				/* FALLTHROUGH */
369 			default:
370 				warnx("%s: invalid directive", start);
371 				return 1;
372 			}
373 			*fmt++ = ch;
374 			*fmt = nextch;
375 			/* escape if a \c was encountered */
376 			if (rval & 0x100)
377 				goto done;
378 		}
379 	} while (gargv != argv && *gargv);
380 
381   done:
382 	(void)fflush(stdout);
383 	if (ferror(stdout)) {
384 		clearerr(stdout);
385 		err(1, "write error");
386 	}
387 	return rval & ~0x100;
388   out:
389 	warn("print failed");
390 	return 1;
391 }
392 
393 /* helper functions for conv_escape_str */
394 
395 static void
396 /*ARGSUSED*/
397 b_count(int ch)
398 {
399 	b_length++;
400 }
401 
402 /* Output one converted character for every 'x' in the 'format' */
403 
404 static void
405 b_output(int ch)
406 {
407 	for (;;) {
408 		switch (*b_fmt++) {
409 		case 0:
410 			b_fmt--;
411 			return;
412 		case ' ':
413 			putchar(' ');
414 			break;
415 		default:
416 			putchar(ch);
417 			return;
418 		}
419 	}
420 }
421 
422 
423 /*
424  * Print SysV echo(1) style escape string
425  *	Halts processing string if a \c escape is encountered.
426  */
427 static void
428 conv_escape_str(char *str, void (*do_putchar)(int), int quiet)
429 {
430 	int value;
431 	int ch;
432 	char c;
433 
434 	while ((ch = *str++) != '\0') {
435 		if (ch != '\\') {
436 			do_putchar(ch);
437 			continue;
438 		}
439 
440 		ch = *str++;
441 		if (ch == 'c') {
442 			/* \c as in SYSV echo - abort all processing.... */
443 			rval |= 0x100;
444 			break;
445 		}
446 
447 		/*
448 		 * %b string octal constants are not like those in C.
449 		 * They start with a \0, and are followed by 0, 1, 2,
450 		 * or 3 octal digits.
451 		 */
452 		if (ch == '0') {
453 			int octnum = 0, i;
454 			for (i = 0; i < 3; i++) {
455 				if (!isdigit((unsigned char)*str) || *str > '7')
456 					break;
457 				octnum = (octnum << 3) | (*str++ - '0');
458 			}
459 			do_putchar(octnum);
460 			continue;
461 		}
462 
463 		/* \[M][^|-]C as defined by vis(3) */
464 		if (ch == 'M' && *str == '-') {
465 			do_putchar(0200 | str[1]);
466 			str += 2;
467 			continue;
468 		}
469 		if (ch == 'M' && *str == '^') {
470 			str++;
471 			value = 0200;
472 			ch = '^';
473 		} else
474 			value = 0;
475 		if (ch == '^') {
476 			ch = *str++;
477 			if (ch == '?')
478 				value |= 0177;
479 			else
480 				value |= ch & 037;
481 			do_putchar(value);
482 			continue;
483 		}
484 
485 		/* Finally test for sequences valid in the format string */
486 		str = conv_escape(str - 1, &c, quiet);
487 		do_putchar(c);
488 	}
489 }
490 
491 /*
492  * Print "standard" escape characters
493  */
494 static char *
495 conv_escape(char *str, char *conv_ch, int quiet)
496 {
497 	int value = 0;
498 	char ch, *begin;
499 	int c;
500 
501 	ch = *str++;
502 
503 	switch (ch) {
504 	case '\0':
505 		if (!quiet)
506 			warnx("incomplete escape sequence");
507 		rval = 1;
508 		value = '\\';
509 		--str;
510 		break;
511 
512 	case '0': case '1': case '2': case '3':
513 	case '4': case '5': case '6': case '7':
514 		str--;
515 		for (c = 3; c-- && isodigit(*str); str++) {
516 			value <<= 3;
517 			value += octtobin(*str);
518 		}
519 		break;
520 
521 	case 'x':
522 		/*
523 		 * Hexadecimal character constants are not required to be
524 		 * supported (by SuS v1) because there is no consistent
525 		 * way to detect the end of the constant.
526 		 * Supporting 2 byte constants is a compromise.
527 		 */
528 		begin = str;
529 		for (c = 2; c-- && isxdigit((unsigned char)*str); str++) {
530 			value <<= 4;
531 			value += hextobin(*str);
532 		}
533 		if (str == begin) {
534 			if (!quiet)
535 				warnx("\\x%s: missing hexadecimal number "
536 				    "in escape", begin);
537 			rval = 1;
538 		}
539 		break;
540 
541 	case '\\':	value = '\\';	break;	/* backslash */
542 	case '\'':	value = '\'';	break;	/* single quote */
543 	case '"':	value = '"';	break;	/* double quote */
544 	case 'a':	value = '\a';	break;	/* alert */
545 	case 'b':	value = '\b';	break;	/* backspace */
546 	case 'e':	value = ESCAPE;	break;	/* escape */
547 	case 'E':	value = ESCAPE;	break;	/* escape */
548 	case 'f':	value = '\f';	break;	/* form-feed */
549 	case 'n':	value = '\n';	break;	/* newline */
550 	case 'r':	value = '\r';	break;	/* carriage-return */
551 	case 't':	value = '\t';	break;	/* tab */
552 	case 'v':	value = '\v';	break;	/* vertical-tab */
553 
554 	default:
555 		if (!quiet)
556 			warnx("unknown escape sequence `\\%c'", ch);
557 		rval = 1;
558 		value = ch;
559 		break;
560 	}
561 
562 	*conv_ch = (char)value;
563 	return str;
564 }
565 
566 /* expand a string so that everything is printable */
567 
568 static char *
569 conv_expand(const char *str)
570 {
571 	static char *conv_str;
572 	char *cp;
573 	char ch;
574 
575 	if (conv_str)
576 		free(conv_str);
577 	/* get a buffer that is definitely large enough.... */
578 	conv_str = malloc(4 * strlen(str) + 1);
579 	if (!conv_str)
580 		return NULL;
581 	cp = conv_str;
582 
583 	while ((ch = *(const char *)str++) != '\0') {
584 		switch (ch) {
585 		/* Use C escapes for expected control characters */
586 		case '\\':	ch = '\\';	break;	/* backslash */
587 		case '\'':	ch = '\'';	break;	/* single quote */
588 		case '"':	ch = '"';	break;	/* double quote */
589 		case '\a':	ch = 'a';	break;	/* alert */
590 		case '\b':	ch = 'b';	break;	/* backspace */
591 		case ESCAPE:	ch = 'e';	break;	/* escape */
592 		case '\f':	ch = 'f';	break;	/* form-feed */
593 		case '\n':	ch = 'n';	break;	/* newline */
594 		case '\r':	ch = 'r';	break;	/* carriage-return */
595 		case '\t':	ch = 't';	break;	/* tab */
596 		case '\v':	ch = 'v';	break;	/* vertical-tab */
597 		default:
598 			/* Copy anything printable */
599 			if (isprint((unsigned char)ch)) {
600 				*cp++ = ch;
601 				continue;
602 			}
603 			/* Use vis(3) encodings for the rest */
604 			*cp++ = '\\';
605 			if (ch & 0200) {
606 				*cp++ = 'M';
607 				ch &= (char)~0200;
608 			}
609 			if (ch == 0177) {
610 				*cp++ = '^';
611 				*cp++ = '?';
612 				continue;
613 			}
614 			if (ch < 040) {
615 				*cp++ = '^';
616 				*cp++ = ch | 0100;
617 				continue;
618 			}
619 			*cp++ = '-';
620 			*cp++ = ch;
621 			continue;
622 		}
623 		*cp++ = '\\';
624 		*cp++ = ch;
625 	}
626 
627 	*cp = 0;
628 	return conv_str;
629 }
630 
631 static char *
632 mklong(const char *str, char ch)
633 {
634 	static char copy[64];
635 	size_t len;
636 
637 	len = strlen(str) + 2;
638 	if (len > sizeof copy) {
639 		warnx("format \"%s\" too complex", str);
640 		len = 4;
641 		rval = 1;
642 	}
643 	(void)memmove(copy, str, len - 3);
644 	copy[len - 3] = 'j';
645 	copy[len - 2] = ch;
646 	copy[len - 1] = '\0';
647 	return copy;
648 }
649 
650 static char
651 getchr(void)
652 {
653 	if (!*gargv)
654 		return 0;
655 	return **gargv++;
656 }
657 
658 static char *
659 getstr(void)
660 {
661 	static char empty[] = "";
662 	if (!*gargv)
663 		return empty;
664 	return *gargv++;
665 }
666 
667 static int
668 getwidth(void)
669 {
670 	unsigned long val;
671 	char *s, *ep;
672 
673 	s = *gargv;
674 	if (s == NULL)
675 		return 0;
676 	gargv++;
677 
678 	errno = 0;
679 	val = strtoul(s, &ep, 0);
680 	check_conversion(s, ep);
681 
682 	/* Arbitrarily 'restrict' field widths to 1Mbyte */
683 	if (val > 1 << 20) {
684 		warnx("%s: invalid field width", s);
685 		return 0;
686 	}
687 
688 	return (int)val;
689 }
690 
691 static intmax_t
692 getintmax(void)
693 {
694 	intmax_t val;
695 	char *cp, *ep;
696 
697 	cp = *gargv;
698 	if (cp == NULL)
699 		return 0;
700 	gargv++;
701 
702 	if (*cp == '\"' || *cp == '\'')
703 		return wide_char(cp);
704 
705 	errno = 0;
706 	val = strtoimax(cp, &ep, 0);
707 	check_conversion(cp, ep);
708 	return val;
709 }
710 
711 static double
712 getdouble(void)
713 {
714 	double val;
715 	char *ep;
716 
717 	if (!*gargv)
718 		return 0.0;
719 
720 	/* This is a NetBSD extension, not required by POSIX (it is useless) */
721 	if (*(ep = *gargv) == '\"' || *ep == '\'')
722 		return (double)wide_char(ep);
723 
724 	errno = 0;
725 	val = strtod(*gargv, &ep);
726 	check_conversion(*gargv++, ep);
727 	return val;
728 }
729 
730 /*
731  * XXX This is just a placeholder for a later version which
732  *     will do mbtowc() on p+1 (and after checking that all of the
733  *     string has been consumed) return that value.
734  *
735  * This (mbtowc) behaviour is required by POSIX (as is the check
736  * that the whole arg is consumed).
737  *
738  * What follows is actually correct if we assume that LC_CTYPE=C
739  * (or something else similar that is a single byte charset).
740  */
741 static intmax_t
742 wide_char(const char *p)
743 {
744 	intmax_t ch = (intmax_t)(unsigned char)p[1];
745 
746 	if (ch != 0 && p[2] != '\0') {
747 		warnx("%s: not completely converted", p);
748 		rval = 1;
749 	}
750 
751 	return ch;
752 }
753 
754 static void
755 check_conversion(const char *s, const char *ep)
756 {
757 	if (*ep) {
758 		if (ep == s)
759 			warnx("%s: expected numeric value", s);
760 		else
761 			warnx("%s: not completely converted", s);
762 		rval = 1;
763 	} else if (errno == ERANGE) {
764 		warnx("%s: %s", s, strerror(ERANGE));
765 		rval = 1;
766 	}
767 }
768 
769 static void
770 usage(void)
771 {
772 	(void)fprintf(stderr, "Usage: %s format [arg ...]\n", getprogname());
773 }
774