Home | History | Annotate | Line # | Download | only in tr
str.c revision 1.11.20.1
      1  1.11.20.1    jym /*	$NetBSD: str.c,v 1.11.20.1 2009/05/13 19:20:09 jym Exp $	*/
      2        1.6    jtc 
      3        1.1  glass /*-
      4        1.6    jtc  * Copyright (c) 1991, 1993
      5        1.6    jtc  *	The Regents of the University of California.  All rights reserved.
      6        1.1  glass  *
      7        1.1  glass  * Redistribution and use in source and binary forms, with or without
      8        1.1  glass  * modification, are permitted provided that the following conditions
      9        1.1  glass  * are met:
     10        1.1  glass  * 1. Redistributions of source code must retain the above copyright
     11        1.1  glass  *    notice, this list of conditions and the following disclaimer.
     12        1.1  glass  * 2. Redistributions in binary form must reproduce the above copyright
     13        1.1  glass  *    notice, this list of conditions and the following disclaimer in the
     14        1.1  glass  *    documentation and/or other materials provided with the distribution.
     15       1.10    agc  * 3. Neither the name of the University nor the names of its contributors
     16        1.1  glass  *    may be used to endorse or promote products derived from this software
     17        1.1  glass  *    without specific prior written permission.
     18        1.1  glass  *
     19        1.1  glass  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     20        1.1  glass  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     21        1.1  glass  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     22        1.1  glass  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     23        1.1  glass  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     24        1.1  glass  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     25        1.1  glass  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     26        1.1  glass  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     27        1.1  glass  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     28        1.1  glass  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     29        1.1  glass  * SUCH DAMAGE.
     30        1.1  glass  */
     31        1.1  glass 
     32        1.8  lukem #include <sys/cdefs.h>
     33        1.1  glass #ifndef lint
     34        1.6    jtc #if 0
     35        1.7    jtc static char sccsid[] = "@(#)str.c	8.2 (Berkeley) 4/28/95";
     36        1.6    jtc #endif
     37  1.11.20.1    jym __RCSID("$NetBSD: str.c,v 1.11.20.1 2009/05/13 19:20:09 jym Exp $");
     38        1.1  glass #endif /* not lint */
     39        1.1  glass 
     40        1.1  glass #include <sys/types.h>
     41        1.1  glass 
     42        1.8  lukem #include <err.h>
     43        1.1  glass #include <errno.h>
     44        1.1  glass #include <stddef.h>
     45        1.1  glass #include <stdio.h>
     46        1.1  glass #include <stdlib.h>
     47        1.1  glass #include <string.h>
     48        1.4    jtc #include <ctype.h>
     49        1.1  glass 
     50        1.1  glass #include "extern.h"
     51        1.1  glass 
     52        1.1  glass static int	backslash __P((STR *));
     53        1.1  glass static int	bracket __P((STR *));
     54        1.1  glass static int	c_class __P((const void *, const void *));
     55        1.1  glass static void	genclass __P((STR *));
     56        1.1  glass static void	genequiv __P((STR *));
     57        1.1  glass static int	genrange __P((STR *));
     58        1.1  glass static void	genseq __P((STR *));
     59        1.1  glass 
     60        1.1  glass int
     61        1.1  glass next(s)
     62        1.8  lukem 	STR *s;
     63        1.1  glass {
     64        1.8  lukem 	int ch;
     65        1.1  glass 
     66        1.1  glass 	switch (s->state) {
     67        1.1  glass 	case EOS:
     68        1.1  glass 		return (0);
     69        1.1  glass 	case INFINITE:
     70        1.1  glass 		return (1);
     71        1.1  glass 	case NORMAL:
     72        1.1  glass 		switch (ch = *s->str) {
     73        1.1  glass 		case '\0':
     74        1.1  glass 			s->state = EOS;
     75        1.1  glass 			return (0);
     76        1.1  glass 		case '\\':
     77        1.1  glass 			s->lastch = backslash(s);
     78        1.1  glass 			break;
     79        1.1  glass 		case '[':
     80        1.1  glass 			if (bracket(s))
     81        1.1  glass 				return (next(s));
     82        1.1  glass 			/* FALLTHROUGH */
     83        1.1  glass 		default:
     84        1.1  glass 			++s->str;
     85        1.1  glass 			s->lastch = ch;
     86        1.1  glass 			break;
     87        1.1  glass 		}
     88        1.1  glass 
     89        1.1  glass 		/* We can start a range at any time. */
     90        1.1  glass 		if (s->str[0] == '-' && genrange(s))
     91        1.1  glass 			return (next(s));
     92        1.1  glass 		return (1);
     93        1.1  glass 	case RANGE:
     94        1.1  glass 		if (s->cnt-- == 0) {
     95        1.1  glass 			s->state = NORMAL;
     96        1.1  glass 			return (next(s));
     97        1.1  glass 		}
     98        1.1  glass 		++s->lastch;
     99        1.1  glass 		return (1);
    100        1.1  glass 	case SEQUENCE:
    101        1.1  glass 		if (s->cnt-- == 0) {
    102        1.1  glass 			s->state = NORMAL;
    103        1.1  glass 			return (next(s));
    104        1.1  glass 		}
    105        1.1  glass 		return (1);
    106        1.1  glass 	case SET:
    107        1.1  glass 		if ((s->lastch = s->set[s->cnt++]) == OOBCH) {
    108        1.1  glass 			s->state = NORMAL;
    109        1.1  glass 			return (next(s));
    110        1.1  glass 		}
    111        1.1  glass 		return (1);
    112        1.1  glass 	}
    113        1.1  glass 	/* NOTREACHED */
    114        1.8  lukem 	return (0);
    115        1.1  glass }
    116        1.1  glass 
    117        1.1  glass static int
    118        1.1  glass bracket(s)
    119        1.8  lukem 	STR *s;
    120        1.1  glass {
    121        1.8  lukem 	char *p;
    122        1.1  glass 
    123        1.1  glass 	switch (s->str[1]) {
    124        1.1  glass 	case ':':				/* "[:class:]" */
    125        1.1  glass 		if ((p = strstr(s->str + 2, ":]")) == NULL)
    126        1.1  glass 			return (0);
    127        1.1  glass 		*p = '\0';
    128        1.1  glass 		s->str += 2;
    129        1.1  glass 		genclass(s);
    130        1.1  glass 		s->str = p + 2;
    131        1.1  glass 		return (1);
    132        1.1  glass 	case '=':				/* "[=equiv=]" */
    133        1.1  glass 		if ((p = strstr(s->str + 2, "=]")) == NULL)
    134        1.1  glass 			return (0);
    135        1.1  glass 		s->str += 2;
    136        1.1  glass 		genequiv(s);
    137        1.1  glass 		return (1);
    138        1.1  glass 	default:				/* "[\###*n]" or "[#*n]" */
    139        1.1  glass 		if ((p = strpbrk(s->str + 2, "*]")) == NULL)
    140        1.1  glass 			return (0);
    141        1.8  lukem 		if (p[0] != '*' || strchr(p, ']') == NULL)
    142        1.1  glass 			return (0);
    143        1.1  glass 		s->str += 1;
    144        1.1  glass 		genseq(s);
    145        1.1  glass 		return (1);
    146        1.1  glass 	}
    147        1.1  glass 	/* NOTREACHED */
    148        1.1  glass }
    149        1.1  glass 
    150        1.1  glass typedef struct {
    151  1.11.20.1    jym 	const char *name;
    152        1.1  glass 	int (*func) __P((int));
    153        1.1  glass 	int *set;
    154        1.1  glass } CLASS;
    155        1.1  glass 
    156        1.1  glass static CLASS classes[] = {
    157  1.11.20.1    jym 	{ "alnum",  isalnum,  NULL, },
    158  1.11.20.1    jym 	{ "alpha",  isalpha,  NULL, },
    159  1.11.20.1    jym 	{ "blank",  isblank,  NULL, },
    160  1.11.20.1    jym 	{ "cntrl",  iscntrl,  NULL, },
    161  1.11.20.1    jym 	{ "digit",  isdigit,  NULL, },
    162  1.11.20.1    jym 	{ "graph",  isgraph,  NULL, },
    163  1.11.20.1    jym 	{ "lower",  islower,  NULL, },
    164  1.11.20.1    jym 	{ "print",  isprint,  NULL, },
    165  1.11.20.1    jym 	{ "punct",  ispunct,  NULL, },
    166  1.11.20.1    jym 	{ "space",  isspace,  NULL, },
    167  1.11.20.1    jym 	{ "upper",  isupper,  NULL, },
    168  1.11.20.1    jym 	{ "xdigit", isxdigit, NULL, },
    169        1.1  glass };
    170        1.1  glass 
    171        1.1  glass static void
    172        1.1  glass genclass(s)
    173        1.1  glass 	STR *s;
    174        1.1  glass {
    175        1.8  lukem 	int cnt, (*func) __P((int));
    176        1.1  glass 	CLASS *cp, tmp;
    177        1.1  glass 	int *p;
    178        1.1  glass 
    179        1.1  glass 	tmp.name = s->str;
    180        1.1  glass 	if ((cp = (CLASS *)bsearch(&tmp, classes, sizeof(classes) /
    181        1.1  glass 	    sizeof(CLASS), sizeof(CLASS), c_class)) == NULL)
    182        1.8  lukem 		errx(1, "unknown class %s", s->str);
    183        1.1  glass 
    184        1.1  glass 	if ((cp->set = p = malloc((NCHARS + 1) * sizeof(int))) == NULL)
    185        1.8  lukem 		err(1, "malloc");
    186        1.8  lukem 	memset(p, 0, (NCHARS + 1) * sizeof(int));
    187        1.1  glass 	for (cnt = 0, func = cp->func; cnt < NCHARS; ++cnt)
    188        1.1  glass 		if ((func)(cnt))
    189        1.1  glass 			*p++ = cnt;
    190        1.1  glass 	*p = OOBCH;
    191        1.1  glass 
    192        1.1  glass 	s->cnt = 0;
    193        1.1  glass 	s->state = SET;
    194        1.1  glass 	s->set = cp->set;
    195        1.1  glass }
    196        1.1  glass 
    197        1.1  glass static int
    198        1.1  glass c_class(a, b)
    199        1.1  glass 	const void *a, *b;
    200        1.1  glass {
    201  1.11.20.1    jym 	return (strcmp(((const CLASS *)a)->name, ((const CLASS *)b)->name));
    202        1.1  glass }
    203        1.1  glass 
    204        1.1  glass /*
    205        1.1  glass  * English doesn't have any equivalence classes, so for now
    206        1.1  glass  * we just syntax check and grab the character.
    207        1.1  glass  */
    208        1.1  glass static void
    209        1.1  glass genequiv(s)
    210        1.1  glass 	STR *s;
    211        1.1  glass {
    212        1.1  glass 	if (*s->str == '\\') {
    213        1.1  glass 		s->equiv[0] = backslash(s);
    214        1.1  glass 		if (*s->str != '=')
    215        1.8  lukem 			errx(1, "misplaced equivalence equals sign");
    216        1.1  glass 	} else {
    217        1.1  glass 		s->equiv[0] = s->str[0];
    218        1.1  glass 		if (s->str[1] != '=')
    219        1.8  lukem 			errx(1, "misplaced equivalence equals sign");
    220        1.1  glass 	}
    221        1.1  glass 	s->str += 2;
    222        1.1  glass 	s->cnt = 0;
    223        1.1  glass 	s->state = SET;
    224        1.1  glass 	s->set = s->equiv;
    225        1.1  glass }
    226        1.1  glass 
    227        1.1  glass static int
    228        1.1  glass genrange(s)
    229        1.1  glass 	STR *s;
    230        1.1  glass {
    231        1.1  glass 	int stopval;
    232        1.1  glass 	char *savestart;
    233        1.1  glass 
    234        1.1  glass 	savestart = s->str;
    235        1.5    jtc 	stopval = *++s->str == '\\' ? backslash(s) : *s->str++;
    236        1.7    jtc 	if (stopval < (u_char)s->lastch) {
    237        1.1  glass 		s->str = savestart;
    238        1.1  glass 		return (0);
    239        1.1  glass 	}
    240        1.1  glass 	s->cnt = stopval - s->lastch + 1;
    241        1.1  glass 	s->state = RANGE;
    242        1.1  glass 	--s->lastch;
    243        1.1  glass 	return (1);
    244        1.1  glass }
    245        1.1  glass 
    246        1.1  glass static void
    247        1.1  glass genseq(s)
    248        1.1  glass 	STR *s;
    249        1.1  glass {
    250        1.1  glass 	char *ep;
    251        1.1  glass 
    252        1.1  glass 	if (s->which == STRING1)
    253        1.8  lukem 		errx(1, "sequences only valid in string2");
    254        1.1  glass 
    255        1.1  glass 	if (*s->str == '\\')
    256        1.1  glass 		s->lastch = backslash(s);
    257        1.1  glass 	else
    258        1.1  glass 		s->lastch = *s->str++;
    259        1.1  glass 	if (*s->str != '*')
    260        1.8  lukem 		errx(1, "misplaced sequence asterisk");
    261        1.1  glass 
    262        1.1  glass 	switch (*++s->str) {
    263        1.1  glass 	case '\\':
    264        1.1  glass 		s->cnt = backslash(s);
    265        1.1  glass 		break;
    266        1.1  glass 	case ']':
    267        1.1  glass 		s->cnt = 0;
    268        1.1  glass 		++s->str;
    269        1.1  glass 		break;
    270        1.1  glass 	default:
    271        1.1  glass 		if (isdigit(*s->str)) {
    272        1.1  glass 			s->cnt = strtol(s->str, &ep, 0);
    273        1.1  glass 			if (*ep == ']') {
    274        1.1  glass 				s->str = ep + 1;
    275        1.1  glass 				break;
    276        1.1  glass 			}
    277        1.1  glass 		}
    278        1.8  lukem 		errx(1, "illegal sequence count");
    279        1.1  glass 		/* NOTREACHED */
    280        1.1  glass 	}
    281        1.1  glass 
    282        1.1  glass 	s->state = s->cnt ? SEQUENCE : INFINITE;
    283        1.1  glass }
    284        1.1  glass 
    285        1.1  glass /*
    286        1.1  glass  * Translate \??? into a character.  Up to 3 octal digits, if no digits either
    287        1.1  glass  * an escape code or a literal character.
    288        1.1  glass  */
    289        1.1  glass static int
    290        1.1  glass backslash(s)
    291        1.8  lukem 	STR *s;
    292        1.1  glass {
    293        1.8  lukem 	int ch, cnt, val;
    294        1.1  glass 
    295        1.1  glass 	for (cnt = val = 0;;) {
    296        1.1  glass 		ch = *++s->str;
    297        1.1  glass 		if (!isascii(ch) || !isdigit(ch))
    298        1.1  glass 			break;
    299        1.1  glass 		val = val * 8 + ch - '0';
    300        1.1  glass 		if (++cnt == 3) {
    301        1.1  glass 			++s->str;
    302        1.1  glass 			break;
    303        1.1  glass 		}
    304        1.1  glass 	}
    305        1.1  glass 	if (cnt)
    306        1.1  glass 		return (val);
    307        1.1  glass 	if (ch != '\0')
    308        1.1  glass 		++s->str;
    309        1.1  glass 	switch (ch) {
    310        1.1  glass 		case 'a':			/* escape characters */
    311        1.1  glass 			return ('\7');
    312        1.1  glass 		case 'b':
    313        1.1  glass 			return ('\b');
    314        1.1  glass 		case 'f':
    315        1.1  glass 			return ('\f');
    316        1.1  glass 		case 'n':
    317        1.1  glass 			return ('\n');
    318        1.1  glass 		case 'r':
    319        1.1  glass 			return ('\r');
    320        1.1  glass 		case 't':
    321        1.1  glass 			return ('\t');
    322        1.1  glass 		case 'v':
    323        1.1  glass 			return ('\13');
    324        1.1  glass 		case '\0':			/*  \" -> \ */
    325        1.1  glass 			s->state = EOS;
    326        1.1  glass 			return ('\\');
    327        1.1  glass 		default:			/* \x" -> x */
    328        1.1  glass 			return (ch);
    329        1.1  glass 	}
    330        1.1  glass }
    331