Home | History | Annotate | Line # | Download | only in tr
str.c revision 1.1
      1  1.1  glass /*-
      2  1.1  glass  * Copyright (c) 1991 The Regents of the University of California.
      3  1.1  glass  * All rights reserved.
      4  1.1  glass  *
      5  1.1  glass  * Redistribution and use in source and binary forms, with or without
      6  1.1  glass  * modification, are permitted provided that the following conditions
      7  1.1  glass  * are met:
      8  1.1  glass  * 1. Redistributions of source code must retain the above copyright
      9  1.1  glass  *    notice, this list of conditions and the following disclaimer.
     10  1.1  glass  * 2. Redistributions in binary form must reproduce the above copyright
     11  1.1  glass  *    notice, this list of conditions and the following disclaimer in the
     12  1.1  glass  *    documentation and/or other materials provided with the distribution.
     13  1.1  glass  * 3. All advertising materials mentioning features or use of this software
     14  1.1  glass  *    must display the following acknowledgement:
     15  1.1  glass  *	This product includes software developed by the University of
     16  1.1  glass  *	California, Berkeley and its contributors.
     17  1.1  glass  * 4. Neither the name of the University nor the names of its contributors
     18  1.1  glass  *    may be used to endorse or promote products derived from this software
     19  1.1  glass  *    without specific prior written permission.
     20  1.1  glass  *
     21  1.1  glass  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
     22  1.1  glass  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
     23  1.1  glass  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
     24  1.1  glass  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
     25  1.1  glass  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
     26  1.1  glass  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
     27  1.1  glass  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
     28  1.1  glass  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
     29  1.1  glass  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
     30  1.1  glass  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
     31  1.1  glass  * SUCH DAMAGE.
     32  1.1  glass  */
     33  1.1  glass 
     34  1.1  glass #ifndef lint
     35  1.1  glass static char sccsid[] = "@(#)str.c	5.9 (Berkeley) 3/4/93";
     36  1.1  glass #endif /* not lint */
     37  1.1  glass 
     38  1.1  glass #include <sys/cdefs.h>
     39  1.1  glass #include <sys/types.h>
     40  1.1  glass 
     41  1.1  glass #include <errno.h>
     42  1.1  glass #include <stddef.h>
     43  1.1  glass #include <stdio.h>
     44  1.1  glass #include <stdlib.h>
     45  1.1  glass #include <string.h>
     46  1.1  glass 
     47  1.1  glass #include "extern.h"
     48  1.1  glass 
     49  1.1  glass static int	backslash __P((STR *));
     50  1.1  glass static int	bracket __P((STR *));
     51  1.1  glass static int	c_class __P((const void *, const void *));
     52  1.1  glass static void	genclass __P((STR *));
     53  1.1  glass static void	genequiv __P((STR *));
     54  1.1  glass static int	genrange __P((STR *));
     55  1.1  glass static void	genseq __P((STR *));
     56  1.1  glass 
     57  1.1  glass int
     58  1.1  glass next(s)
     59  1.1  glass 	register STR *s;
     60  1.1  glass {
     61  1.1  glass 	register int ch;
     62  1.1  glass 
     63  1.1  glass 	switch (s->state) {
     64  1.1  glass 	case EOS:
     65  1.1  glass 		return (0);
     66  1.1  glass 	case INFINITE:
     67  1.1  glass 		return (1);
     68  1.1  glass 	case NORMAL:
     69  1.1  glass 		switch (ch = *s->str) {
     70  1.1  glass 		case '\0':
     71  1.1  glass 			s->state = EOS;
     72  1.1  glass 			return (0);
     73  1.1  glass 		case '\\':
     74  1.1  glass 			s->lastch = backslash(s);
     75  1.1  glass 			break;
     76  1.1  glass 		case '[':
     77  1.1  glass 			if (bracket(s))
     78  1.1  glass 				return (next(s));
     79  1.1  glass 			/* FALLTHROUGH */
     80  1.1  glass 		default:
     81  1.1  glass 			++s->str;
     82  1.1  glass 			s->lastch = ch;
     83  1.1  glass 			break;
     84  1.1  glass 		}
     85  1.1  glass 
     86  1.1  glass 		/* We can start a range at any time. */
     87  1.1  glass 		if (s->str[0] == '-' && genrange(s))
     88  1.1  glass 			return (next(s));
     89  1.1  glass 		return (1);
     90  1.1  glass 	case RANGE:
     91  1.1  glass 		if (s->cnt-- == 0) {
     92  1.1  glass 			s->state = NORMAL;
     93  1.1  glass 			return (next(s));
     94  1.1  glass 		}
     95  1.1  glass 		++s->lastch;
     96  1.1  glass 		return (1);
     97  1.1  glass 	case SEQUENCE:
     98  1.1  glass 		if (s->cnt-- == 0) {
     99  1.1  glass 			s->state = NORMAL;
    100  1.1  glass 			return (next(s));
    101  1.1  glass 		}
    102  1.1  glass 		return (1);
    103  1.1  glass 	case SET:
    104  1.1  glass 		if ((s->lastch = s->set[s->cnt++]) == OOBCH) {
    105  1.1  glass 			s->state = NORMAL;
    106  1.1  glass 			return (next(s));
    107  1.1  glass 		}
    108  1.1  glass 		return (1);
    109  1.1  glass 	}
    110  1.1  glass 	/* NOTREACHED */
    111  1.1  glass }
    112  1.1  glass 
    113  1.1  glass static int
    114  1.1  glass bracket(s)
    115  1.1  glass 	register STR *s;
    116  1.1  glass {
    117  1.1  glass 	register char *p;
    118  1.1  glass 
    119  1.1  glass 	switch (s->str[1]) {
    120  1.1  glass 	case ':':				/* "[:class:]" */
    121  1.1  glass 		if ((p = strstr(s->str + 2, ":]")) == NULL)
    122  1.1  glass 			return (0);
    123  1.1  glass 		*p = '\0';
    124  1.1  glass 		s->str += 2;
    125  1.1  glass 		genclass(s);
    126  1.1  glass 		s->str = p + 2;
    127  1.1  glass 		return (1);
    128  1.1  glass 	case '=':				/* "[=equiv=]" */
    129  1.1  glass 		if ((p = strstr(s->str + 2, "=]")) == NULL)
    130  1.1  glass 			return (0);
    131  1.1  glass 		s->str += 2;
    132  1.1  glass 		genequiv(s);
    133  1.1  glass 		return (1);
    134  1.1  glass 	default:				/* "[\###*n]" or "[#*n]" */
    135  1.1  glass 		if ((p = strpbrk(s->str + 2, "*]")) == NULL)
    136  1.1  glass 			return (0);
    137  1.1  glass 		if (p[0] != '*' || index(p, ']') == NULL)
    138  1.1  glass 			return (0);
    139  1.1  glass 		s->str += 1;
    140  1.1  glass 		genseq(s);
    141  1.1  glass 		return (1);
    142  1.1  glass 	}
    143  1.1  glass 	/* NOTREACHED */
    144  1.1  glass }
    145  1.1  glass 
    146  1.1  glass int isalnum __P((int)),
    147  1.1  glass     isalpha __P((int)),
    148  1.1  glass     isblank __P((int)),
    149  1.1  glass     isspace __P((int)),
    150  1.1  glass     iscntrl __P((int)),
    151  1.1  glass     isdigit __P((int)),
    152  1.1  glass     isgraph __P((int)),
    153  1.1  glass     islower __P((int)),
    154  1.1  glass     isprint __P((int)),
    155  1.1  glass     ispunct __P((int)),
    156  1.1  glass     isupper __P((int)),
    157  1.1  glass     isxdigit __P((int));
    158  1.1  glass 
    159  1.1  glass 
    160  1.1  glass static int isblank(x) /* until 4.4 */
    161  1.1  glass      int x;
    162  1.1  glass {
    163  1.1  glass     if ((x == ' ') || (x== '\t')) return 1;
    164  1.1  glass     return 0;
    165  1.1  glass }
    166  1.1  glass 
    167  1.1  glass 
    168  1.1  glass typedef struct {
    169  1.1  glass 	char *name;
    170  1.1  glass 	int (*func) __P((int));
    171  1.1  glass 	int *set;
    172  1.1  glass } CLASS;
    173  1.1  glass 
    174  1.1  glass static CLASS classes[] = {
    175  1.1  glass 	{ "alnum",  isalnum,  },
    176  1.1  glass 	{ "alpha",  isalpha,  },
    177  1.1  glass 	{ "blank",  isblank,  },
    178  1.1  glass 	{ "cntrl",  iscntrl,  },
    179  1.1  glass 	{ "digit",  isdigit,  },
    180  1.1  glass 	{ "graph",  isgraph,  },
    181  1.1  glass 	{ "lower",  islower,  },
    182  1.1  glass 	{ "print",  isupper,  },
    183  1.1  glass 	{ "punct",  ispunct,  },
    184  1.1  glass 	{ "space",  isspace,  },
    185  1.1  glass 	{ "upper",  isupper,  },
    186  1.1  glass 	{ "xdigit", isxdigit, },
    187  1.1  glass };
    188  1.1  glass 
    189  1.1  glass static void
    190  1.1  glass genclass(s)
    191  1.1  glass 	STR *s;
    192  1.1  glass {
    193  1.1  glass 	register int cnt, (*func) __P((int));
    194  1.1  glass 	CLASS *cp, tmp;
    195  1.1  glass 	int *p;
    196  1.1  glass 
    197  1.1  glass 	tmp.name = s->str;
    198  1.1  glass 	if ((cp = (CLASS *)bsearch(&tmp, classes, sizeof(classes) /
    199  1.1  glass 	    sizeof(CLASS), sizeof(CLASS), c_class)) == NULL)
    200  1.1  glass 		err("unknown class %s", s->str);
    201  1.1  glass 
    202  1.1  glass 	if ((cp->set = p = malloc((NCHARS + 1) * sizeof(int))) == NULL)
    203  1.1  glass 		err("%s", strerror(errno));
    204  1.1  glass 	bzero(p, (NCHARS + 1) * sizeof(int));
    205  1.1  glass 	for (cnt = 0, func = cp->func; cnt < NCHARS; ++cnt)
    206  1.1  glass 		if ((func)(cnt))
    207  1.1  glass 			*p++ = cnt;
    208  1.1  glass 	*p = OOBCH;
    209  1.1  glass 
    210  1.1  glass 	s->cnt = 0;
    211  1.1  glass 	s->state = SET;
    212  1.1  glass 	s->set = cp->set;
    213  1.1  glass }
    214  1.1  glass 
    215  1.1  glass static int
    216  1.1  glass c_class(a, b)
    217  1.1  glass 	const void *a, *b;
    218  1.1  glass {
    219  1.1  glass 	return (strcmp(((CLASS *)a)->name, ((CLASS *)b)->name));
    220  1.1  glass }
    221  1.1  glass 
    222  1.1  glass /*
    223  1.1  glass  * English doesn't have any equivalence classes, so for now
    224  1.1  glass  * we just syntax check and grab the character.
    225  1.1  glass  */
    226  1.1  glass static void
    227  1.1  glass genequiv(s)
    228  1.1  glass 	STR *s;
    229  1.1  glass {
    230  1.1  glass 	if (*s->str == '\\') {
    231  1.1  glass 		s->equiv[0] = backslash(s);
    232  1.1  glass 		if (*s->str != '=')
    233  1.1  glass 			err("misplaced equivalence equals sign");
    234  1.1  glass 	} else {
    235  1.1  glass 		s->equiv[0] = s->str[0];
    236  1.1  glass 		if (s->str[1] != '=')
    237  1.1  glass 			err("misplaced equivalence equals sign");
    238  1.1  glass 	}
    239  1.1  glass 	s->str += 2;
    240  1.1  glass 	s->cnt = 0;
    241  1.1  glass 	s->state = SET;
    242  1.1  glass 	s->set = s->equiv;
    243  1.1  glass }
    244  1.1  glass 
    245  1.1  glass static int
    246  1.1  glass genrange(s)
    247  1.1  glass 	STR *s;
    248  1.1  glass {
    249  1.1  glass 	int stopval;
    250  1.1  glass 	char *savestart;
    251  1.1  glass 
    252  1.1  glass 	savestart = s->str;
    253  1.1  glass 	stopval = *++s->str == '\\' ? backslash(s) : *s->str;
    254  1.1  glass 	if (stopval < s->lastch) {
    255  1.1  glass 		s->str = savestart;
    256  1.1  glass 		return (0);
    257  1.1  glass 	}
    258  1.1  glass 	s->cnt = stopval - s->lastch + 1;
    259  1.1  glass 	s->state = RANGE;
    260  1.1  glass 	--s->lastch;
    261  1.1  glass 	return (1);
    262  1.1  glass }
    263  1.1  glass 
    264  1.1  glass static void
    265  1.1  glass genseq(s)
    266  1.1  glass 	STR *s;
    267  1.1  glass {
    268  1.1  glass 	char *ep;
    269  1.1  glass 
    270  1.1  glass 	if (s->which == STRING1)
    271  1.1  glass 		err("sequences only valid in string2");
    272  1.1  glass 
    273  1.1  glass 	if (*s->str == '\\')
    274  1.1  glass 		s->lastch = backslash(s);
    275  1.1  glass 	else
    276  1.1  glass 		s->lastch = *s->str++;
    277  1.1  glass 	if (*s->str != '*')
    278  1.1  glass 		err("misplaced sequence asterisk");
    279  1.1  glass 
    280  1.1  glass 	switch (*++s->str) {
    281  1.1  glass 	case '\\':
    282  1.1  glass 		s->cnt = backslash(s);
    283  1.1  glass 		break;
    284  1.1  glass 	case ']':
    285  1.1  glass 		s->cnt = 0;
    286  1.1  glass 		++s->str;
    287  1.1  glass 		break;
    288  1.1  glass 	default:
    289  1.1  glass 		if (isdigit(*s->str)) {
    290  1.1  glass 			s->cnt = strtol(s->str, &ep, 0);
    291  1.1  glass 			if (*ep == ']') {
    292  1.1  glass 				s->str = ep + 1;
    293  1.1  glass 				break;
    294  1.1  glass 			}
    295  1.1  glass 		}
    296  1.1  glass 		err("illegal sequence count");
    297  1.1  glass 		/* NOTREACHED */
    298  1.1  glass 	}
    299  1.1  glass 
    300  1.1  glass 	s->state = s->cnt ? SEQUENCE : INFINITE;
    301  1.1  glass }
    302  1.1  glass 
    303  1.1  glass /* Use the #defines isXXX() here, DON'T use them above. */
    304  1.1  glass #include <ctype.h>
    305  1.1  glass 
    306  1.1  glass /*
    307  1.1  glass  * Translate \??? into a character.  Up to 3 octal digits, if no digits either
    308  1.1  glass  * an escape code or a literal character.
    309  1.1  glass  */
    310  1.1  glass static int
    311  1.1  glass backslash(s)
    312  1.1  glass 	register STR *s;
    313  1.1  glass {
    314  1.1  glass 	register int ch, cnt, val;
    315  1.1  glass 
    316  1.1  glass 	for (cnt = val = 0;;) {
    317  1.1  glass 		ch = *++s->str;
    318  1.1  glass 		if (!isascii(ch) || !isdigit(ch))
    319  1.1  glass 			break;
    320  1.1  glass 		val = val * 8 + ch - '0';
    321  1.1  glass 		if (++cnt == 3) {
    322  1.1  glass 			++s->str;
    323  1.1  glass 			break;
    324  1.1  glass 		}
    325  1.1  glass 	}
    326  1.1  glass 	if (cnt)
    327  1.1  glass 		return (val);
    328  1.1  glass 	if (ch != '\0')
    329  1.1  glass 		++s->str;
    330  1.1  glass 	switch (ch) {
    331  1.1  glass 		case 'a':			/* escape characters */
    332  1.1  glass 			return ('\7');
    333  1.1  glass 		case 'b':
    334  1.1  glass 			return ('\b');
    335  1.1  glass 		case 'f':
    336  1.1  glass 			return ('\f');
    337  1.1  glass 		case 'n':
    338  1.1  glass 			return ('\n');
    339  1.1  glass 		case 'r':
    340  1.1  glass 			return ('\r');
    341  1.1  glass 		case 't':
    342  1.1  glass 			return ('\t');
    343  1.1  glass 		case 'v':
    344  1.1  glass 			return ('\13');
    345  1.1  glass 		case '\0':			/*  \" -> \ */
    346  1.1  glass 			s->state = EOS;
    347  1.1  glass 			return ('\\');
    348  1.1  glass 		default:			/* \x" -> x */
    349  1.1  glass 			return (ch);
    350  1.1  glass 	}
    351  1.1  glass }
    352