Home | History | Annotate | Line # | Download | only in tr
str.c revision 1.16
      1  1.16  christos /*	$NetBSD: str.c,v 1.16 2011/09/08 01:18:05 christos 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  1.16  christos __RCSID("$NetBSD: str.c,v 1.16 2011/09/08 01:18:05 christos 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  1.13     joerg static int	backslash(STR *);
     53  1.13     joerg static int	bracket(STR *);
     54  1.13     joerg static int	c_class(const void *, const void *);
     55  1.13     joerg static void	genclass(STR *);
     56  1.13     joerg static void	genequiv(STR *);
     57  1.13     joerg static int	genrange(STR *);
     58  1.13     joerg static void	genseq(STR *);
     59   1.1     glass 
     60   1.1     glass int
     61  1.13     joerg next(STR *s)
     62   1.1     glass {
     63   1.8     lukem 	int ch;
     64   1.1     glass 
     65   1.1     glass 	switch (s->state) {
     66   1.1     glass 	case EOS:
     67  1.16  christos 		return 0;
     68   1.1     glass 	case INFINITE:
     69  1.16  christos 		return 1;
     70   1.1     glass 	case NORMAL:
     71   1.1     glass 		switch (ch = *s->str) {
     72   1.1     glass 		case '\0':
     73   1.1     glass 			s->state = EOS;
     74  1.16  christos 			return 0;
     75   1.1     glass 		case '\\':
     76   1.1     glass 			s->lastch = backslash(s);
     77   1.1     glass 			break;
     78   1.1     glass 		case '[':
     79   1.1     glass 			if (bracket(s))
     80  1.16  christos 				return next(s);
     81   1.1     glass 			/* FALLTHROUGH */
     82   1.1     glass 		default:
     83   1.1     glass 			++s->str;
     84   1.1     glass 			s->lastch = ch;
     85   1.1     glass 			break;
     86   1.1     glass 		}
     87   1.1     glass 
     88   1.1     glass 		/* We can start a range at any time. */
     89   1.1     glass 		if (s->str[0] == '-' && genrange(s))
     90  1.16  christos 			return next(s);
     91  1.16  christos 		return 1;
     92   1.1     glass 	case RANGE:
     93   1.1     glass 		if (s->cnt-- == 0) {
     94   1.1     glass 			s->state = NORMAL;
     95  1.16  christos 			return next(s);
     96   1.1     glass 		}
     97   1.1     glass 		++s->lastch;
     98  1.16  christos 		return 1;
     99   1.1     glass 	case SEQUENCE:
    100   1.1     glass 		if (s->cnt-- == 0) {
    101   1.1     glass 			s->state = NORMAL;
    102  1.16  christos 			return next(s);
    103   1.1     glass 		}
    104  1.16  christos 		return 1;
    105   1.1     glass 	case SET:
    106   1.1     glass 		if ((s->lastch = s->set[s->cnt++]) == OOBCH) {
    107   1.1     glass 			s->state = NORMAL;
    108  1.16  christos 			return next(s);
    109   1.1     glass 		}
    110  1.16  christos 		return 1;
    111   1.1     glass 	}
    112   1.1     glass 	/* NOTREACHED */
    113  1.16  christos 	return 0;
    114   1.1     glass }
    115   1.1     glass 
    116   1.1     glass static int
    117  1.13     joerg bracket(STR *s)
    118   1.1     glass {
    119   1.8     lukem 	char *p;
    120   1.1     glass 
    121   1.1     glass 	switch (s->str[1]) {
    122   1.1     glass 	case ':':				/* "[:class:]" */
    123   1.1     glass 		if ((p = strstr(s->str + 2, ":]")) == NULL)
    124  1.16  christos 			return 0;
    125   1.1     glass 		*p = '\0';
    126   1.1     glass 		s->str += 2;
    127   1.1     glass 		genclass(s);
    128   1.1     glass 		s->str = p + 2;
    129  1.16  christos 		return 1;
    130   1.1     glass 	case '=':				/* "[=equiv=]" */
    131   1.1     glass 		if ((p = strstr(s->str + 2, "=]")) == NULL)
    132  1.16  christos 			return 0;
    133   1.1     glass 		s->str += 2;
    134   1.1     glass 		genequiv(s);
    135  1.16  christos 		return 1;
    136   1.1     glass 	default:				/* "[\###*n]" or "[#*n]" */
    137   1.1     glass 		if ((p = strpbrk(s->str + 2, "*]")) == NULL)
    138  1.16  christos 			return 0;
    139   1.8     lukem 		if (p[0] != '*' || strchr(p, ']') == NULL)
    140  1.16  christos 			return 0;
    141   1.1     glass 		s->str += 1;
    142   1.1     glass 		genseq(s);
    143  1.16  christos 		return 1;
    144   1.1     glass 	}
    145   1.1     glass 	/* NOTREACHED */
    146   1.1     glass }
    147   1.1     glass 
    148   1.1     glass typedef struct {
    149  1.12     lukem 	const char *name;
    150  1.13     joerg 	int (*func)(int);
    151   1.1     glass } CLASS;
    152   1.1     glass 
    153  1.16  christos static const CLASS classes[] = {
    154  1.16  christos 	{ "alnum",  isalnum  },
    155  1.16  christos 	{ "alpha",  isalpha  },
    156  1.16  christos 	{ "blank",  isblank  },
    157  1.16  christos 	{ "cntrl",  iscntrl  },
    158  1.16  christos 	{ "digit",  isdigit  },
    159  1.16  christos 	{ "graph",  isgraph  },
    160  1.16  christos 	{ "lower",  islower  },
    161  1.16  christos 	{ "print",  isprint  },
    162  1.16  christos 	{ "punct",  ispunct  },
    163  1.16  christos 	{ "space",  isspace  },
    164  1.16  christos 	{ "upper",  isupper  },
    165  1.16  christos 	{ "xdigit", isxdigit },
    166   1.1     glass };
    167   1.1     glass 
    168   1.1     glass static void
    169  1.13     joerg genclass(STR *s)
    170   1.1     glass {
    171  1.13     joerg 	int cnt, (*func)(int);
    172  1.16  christos 	const CLASS *cp;
    173  1.16  christos 	CLASS tmp;
    174   1.1     glass 	int *p;
    175   1.1     glass 
    176   1.1     glass 	tmp.name = s->str;
    177  1.16  christos 	if ((cp = bsearch(&tmp, classes, sizeof(classes) /
    178  1.16  christos 	    sizeof(*cp), sizeof(*cp), c_class)) == NULL)
    179   1.8     lukem 		errx(1, "unknown class %s", s->str);
    180   1.1     glass 
    181  1.16  christos 	if ((s->set = p = malloc((NCHARS + 1) * sizeof(*p))) == NULL)
    182   1.8     lukem 		err(1, "malloc");
    183   1.1     glass 	for (cnt = 0, func = cp->func; cnt < NCHARS; ++cnt)
    184   1.1     glass 		if ((func)(cnt))
    185   1.1     glass 			*p++ = cnt;
    186   1.1     glass 	*p = OOBCH;
    187   1.1     glass 
    188   1.1     glass 	s->cnt = 0;
    189   1.1     glass 	s->state = SET;
    190   1.1     glass }
    191   1.1     glass 
    192   1.1     glass static int
    193  1.13     joerg c_class(const void *a, const void *b)
    194   1.1     glass {
    195  1.16  christos 	return strcmp(((const CLASS *)a)->name, ((const CLASS *)b)->name);
    196   1.1     glass }
    197   1.1     glass 
    198   1.1     glass /*
    199   1.1     glass  * English doesn't have any equivalence classes, so for now
    200   1.1     glass  * we just syntax check and grab the character.
    201   1.1     glass  */
    202   1.1     glass static void
    203  1.13     joerg genequiv(STR *s)
    204   1.1     glass {
    205   1.1     glass 	if (*s->str == '\\') {
    206   1.1     glass 		s->equiv[0] = backslash(s);
    207   1.1     glass 		if (*s->str != '=')
    208   1.8     lukem 			errx(1, "misplaced equivalence equals sign");
    209   1.1     glass 	} else {
    210   1.1     glass 		s->equiv[0] = s->str[0];
    211   1.1     glass 		if (s->str[1] != '=')
    212   1.8     lukem 			errx(1, "misplaced equivalence equals sign");
    213   1.1     glass 	}
    214   1.1     glass 	s->str += 2;
    215   1.1     glass 	s->cnt = 0;
    216   1.1     glass 	s->state = SET;
    217   1.1     glass 	s->set = s->equiv;
    218   1.1     glass }
    219   1.1     glass 
    220   1.1     glass static int
    221  1.13     joerg genrange(STR *s)
    222   1.1     glass {
    223   1.1     glass 	int stopval;
    224   1.1     glass 	char *savestart;
    225   1.1     glass 
    226   1.1     glass 	savestart = s->str;
    227   1.5       jtc 	stopval = *++s->str == '\\' ? backslash(s) : *s->str++;
    228   1.7       jtc 	if (stopval < (u_char)s->lastch) {
    229   1.1     glass 		s->str = savestart;
    230  1.16  christos 		return 0;
    231   1.1     glass 	}
    232   1.1     glass 	s->cnt = stopval - s->lastch + 1;
    233   1.1     glass 	s->state = RANGE;
    234   1.1     glass 	--s->lastch;
    235  1.16  christos 	return 1;
    236   1.1     glass }
    237   1.1     glass 
    238   1.1     glass static void
    239  1.13     joerg genseq(STR *s)
    240   1.1     glass {
    241   1.1     glass 	char *ep;
    242   1.1     glass 
    243   1.1     glass 	if (s->which == STRING1)
    244   1.8     lukem 		errx(1, "sequences only valid in string2");
    245   1.1     glass 
    246   1.1     glass 	if (*s->str == '\\')
    247   1.1     glass 		s->lastch = backslash(s);
    248   1.1     glass 	else
    249   1.1     glass 		s->lastch = *s->str++;
    250   1.1     glass 	if (*s->str != '*')
    251   1.8     lukem 		errx(1, "misplaced sequence asterisk");
    252   1.1     glass 
    253   1.1     glass 	switch (*++s->str) {
    254   1.1     glass 	case '\\':
    255   1.1     glass 		s->cnt = backslash(s);
    256   1.1     glass 		break;
    257   1.1     glass 	case ']':
    258   1.1     glass 		s->cnt = 0;
    259   1.1     glass 		++s->str;
    260   1.1     glass 		break;
    261   1.1     glass 	default:
    262   1.1     glass 		if (isdigit(*s->str)) {
    263   1.1     glass 			s->cnt = strtol(s->str, &ep, 0);
    264   1.1     glass 			if (*ep == ']') {
    265   1.1     glass 				s->str = ep + 1;
    266   1.1     glass 				break;
    267   1.1     glass 			}
    268   1.1     glass 		}
    269   1.8     lukem 		errx(1, "illegal sequence count");
    270   1.1     glass 		/* NOTREACHED */
    271   1.1     glass 	}
    272   1.1     glass 
    273   1.1     glass 	s->state = s->cnt ? SEQUENCE : INFINITE;
    274   1.1     glass }
    275   1.1     glass 
    276   1.1     glass /*
    277   1.1     glass  * Translate \??? into a character.  Up to 3 octal digits, if no digits either
    278   1.1     glass  * an escape code or a literal character.
    279   1.1     glass  */
    280   1.1     glass static int
    281  1.13     joerg backslash(STR *s)
    282   1.1     glass {
    283   1.8     lukem 	int ch, cnt, val;
    284   1.1     glass 
    285   1.1     glass 	for (cnt = val = 0;;) {
    286   1.1     glass 		ch = *++s->str;
    287   1.1     glass 		if (!isascii(ch) || !isdigit(ch))
    288   1.1     glass 			break;
    289   1.1     glass 		val = val * 8 + ch - '0';
    290   1.1     glass 		if (++cnt == 3) {
    291   1.1     glass 			++s->str;
    292   1.1     glass 			break;
    293   1.1     glass 		}
    294   1.1     glass 	}
    295   1.1     glass 	if (cnt)
    296  1.16  christos 		return val;
    297   1.1     glass 	if (ch != '\0')
    298   1.1     glass 		++s->str;
    299   1.1     glass 	switch (ch) {
    300  1.16  christos 	    case 'a':			/* escape characters */
    301  1.16  christos 		    return '\7';
    302  1.16  christos 	    case 'b':
    303  1.16  christos 		    return '\b';
    304  1.16  christos 	    case 'e':
    305  1.16  christos 		    return '\033';
    306  1.16  christos 	    case 'f':
    307  1.16  christos 		    return '\f';
    308  1.16  christos 	    case 'n':
    309  1.16  christos 		    return '\n';
    310  1.16  christos 	    case 'r':
    311  1.16  christos 		    return '\r';
    312  1.16  christos 	    case 't':
    313  1.16  christos 		    return '\t';
    314  1.16  christos 	    case 'v':
    315  1.16  christos 		    return '\13';
    316  1.16  christos 	    case '\0':			/*  \" -> \ */
    317  1.16  christos 		    s->state = EOS;
    318  1.16  christos 		    return '\\';
    319  1.16  christos 	    default:			/* \x" -> x */
    320  1.16  christos 		    return ch;
    321   1.1     glass 	}
    322   1.1     glass }
    323