xref: /freebsd/lib/libc/locale/collate.c (revision 377da8e86794d748bd62696eaed3d8f2b4071075)
1c3d0cca4SAndrey A. Chernov /*-
2c3d0cca4SAndrey A. Chernov  * Copyright (c) 1995 Alex Tatmanjants <alex@elvisti.kiev.ua>
3c3d0cca4SAndrey A. Chernov  *		at Electronni Visti IA, Kiev, Ukraine.
4c3d0cca4SAndrey A. Chernov  *			All rights reserved.
5c3d0cca4SAndrey A. Chernov  *
6c3d0cca4SAndrey A. Chernov  * Redistribution and use in source and binary forms, with or without
7c3d0cca4SAndrey A. Chernov  * modification, are permitted provided that the following conditions
8c3d0cca4SAndrey A. Chernov  * are met:
9c3d0cca4SAndrey A. Chernov  * 1. Redistributions of source code must retain the above copyright
10c3d0cca4SAndrey A. Chernov  *    notice, this list of conditions and the following disclaimer.
11c3d0cca4SAndrey A. Chernov  * 2. Redistributions in binary form must reproduce the above copyright
12c3d0cca4SAndrey A. Chernov  *    notice, this list of conditions and the following disclaimer in the
13c3d0cca4SAndrey A. Chernov  *    documentation and/or other materials provided with the distribution.
14c3d0cca4SAndrey A. Chernov  *
15c3d0cca4SAndrey A. Chernov  * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND
16c3d0cca4SAndrey A. Chernov  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
17c3d0cca4SAndrey A. Chernov  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
18c3d0cca4SAndrey A. Chernov  * ARE DISCLAIMED.  IN NO EVENT SHALL THE AUTHOR BE LIABLE
19c3d0cca4SAndrey A. Chernov  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
20c3d0cca4SAndrey A. Chernov  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
21c3d0cca4SAndrey A. Chernov  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
22c3d0cca4SAndrey A. Chernov  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
23c3d0cca4SAndrey A. Chernov  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
24c3d0cca4SAndrey A. Chernov  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
25c3d0cca4SAndrey A. Chernov  * SUCH DAMAGE.
26c3d0cca4SAndrey A. Chernov  *
27377da8e8SAndrey A. Chernov  * $Id: collate.c,v 1.4 1995/05/30 05:40:40 rgrimes Exp $
28c3d0cca4SAndrey A. Chernov  */
29c3d0cca4SAndrey A. Chernov 
30c3d0cca4SAndrey A. Chernov #include <rune.h>
31c3d0cca4SAndrey A. Chernov #include <stdio.h>
32c3d0cca4SAndrey A. Chernov #include <stdlib.h>
33c3d0cca4SAndrey A. Chernov #include <string.h>
34926f20c9SAndrey A. Chernov #include <errno.h>
35926f20c9SAndrey A. Chernov #include <unistd.h>
36c3d0cca4SAndrey A. Chernov #include <sysexits.h>
37c3d0cca4SAndrey A. Chernov #include "collate.h"
38c3d0cca4SAndrey A. Chernov 
39377da8e8SAndrey A. Chernov char *_PathLocale;
40c3d0cca4SAndrey A. Chernov int __collate_load_error = 1;
41c3d0cca4SAndrey A. Chernov u_char __collate_charmap_table[UCHAR_MAX + 1][STR_LEN];
42c3d0cca4SAndrey A. Chernov u_char __collate_substitute_table[UCHAR_MAX + 1][STR_LEN];
43c3d0cca4SAndrey A. Chernov struct __collate_st_char_pri __collate_char_pri_table[UCHAR_MAX + 1];
44c3d0cca4SAndrey A. Chernov struct __collate_st_name_pri __collate_name_pri_table[TABLE_SIZE];
45c3d0cca4SAndrey A. Chernov struct __collate_st_chain_pri __collate_chain_pri_table[TABLE_SIZE];
46c3d0cca4SAndrey A. Chernov 
47c3d0cca4SAndrey A. Chernov #define FREAD(a, b, c, d) \
48c3d0cca4SAndrey A. Chernov 	do { \
49c3d0cca4SAndrey A. Chernov 		if(fread(a, b, c, d) != c) { \
50c3d0cca4SAndrey A. Chernov 			fclose(d); \
51c3d0cca4SAndrey A. Chernov 			return -1; \
52c3d0cca4SAndrey A. Chernov 		} \
53c3d0cca4SAndrey A. Chernov 	} while(0)
54c3d0cca4SAndrey A. Chernov 
55926f20c9SAndrey A. Chernov __dead void __collate_err(int ex, const char *f) __dead2;
56926f20c9SAndrey A. Chernov 
57c3d0cca4SAndrey A. Chernov int
58c3d0cca4SAndrey A. Chernov __collate_load_tables(encoding)
59c3d0cca4SAndrey A. Chernov 	char *encoding;
60c3d0cca4SAndrey A. Chernov {
61c3d0cca4SAndrey A. Chernov 	char buf[PATH_MAX];
62c3d0cca4SAndrey A. Chernov 	FILE *fp;
63377da8e8SAndrey A. Chernov 	int save_load_error;
64c3d0cca4SAndrey A. Chernov 
65377da8e8SAndrey A. Chernov 	save_load_error = __collate_load_error;
66c3d0cca4SAndrey A. Chernov 	__collate_load_error = 1;
67377da8e8SAndrey A. Chernov 	if (!encoding) {
68377da8e8SAndrey A. Chernov 		__collate_load_error = save_load_error;
69c3d0cca4SAndrey A. Chernov 		return -1;
70377da8e8SAndrey A. Chernov 	}
71377da8e8SAndrey A. Chernov 	if (!*encoding || !strcmp(encoding, "C") || !strcmp(encoding, "POSIX"))
72377da8e8SAndrey A. Chernov 		return 0;
73377da8e8SAndrey A. Chernov 	if (!_PathLocale && !(_PathLocale = getenv("PATH_LOCALE")))
74377da8e8SAndrey A. Chernov 		_PathLocale = _PATH_LOCALE;
75377da8e8SAndrey A. Chernov 	strcpy(buf, _PathLocale);
76c3d0cca4SAndrey A. Chernov 	strcat(buf, "/");
77c3d0cca4SAndrey A. Chernov 	strcat(buf, encoding);
78c3d0cca4SAndrey A. Chernov 	strcat(buf, "/LC_COLLATE");
79377da8e8SAndrey A. Chernov 	if ((fp = fopen(buf, "r")) == NULL) {
80377da8e8SAndrey A. Chernov 		__collate_load_error = save_load_error;
81c3d0cca4SAndrey A. Chernov 		return -1;
82377da8e8SAndrey A. Chernov 	}
83c3d0cca4SAndrey A. Chernov 	FREAD(__collate_charmap_table, sizeof(__collate_charmap_table), 1, fp);
84c3d0cca4SAndrey A. Chernov 	FREAD(__collate_substitute_table, sizeof(__collate_substitute_table),
85c3d0cca4SAndrey A. Chernov 	      1, fp);
86c3d0cca4SAndrey A. Chernov 	FREAD(__collate_char_pri_table, sizeof(__collate_char_pri_table), 1,
87c3d0cca4SAndrey A. Chernov 	      fp);
88c3d0cca4SAndrey A. Chernov 	FREAD(__collate_chain_pri_table, sizeof(__collate_chain_pri_table), 1,
89c3d0cca4SAndrey A. Chernov 	      fp);
90c3d0cca4SAndrey A. Chernov 	FREAD(__collate_name_pri_table, sizeof(__collate_name_pri_table), 1,
91c3d0cca4SAndrey A. Chernov 	      fp);
92c3d0cca4SAndrey A. Chernov 	fclose(fp);
93c3d0cca4SAndrey A. Chernov 	__collate_load_error = 0;
94c3d0cca4SAndrey A. Chernov 	return 0;
95c3d0cca4SAndrey A. Chernov }
96c3d0cca4SAndrey A. Chernov 
97c3d0cca4SAndrey A. Chernov u_char *
98c3d0cca4SAndrey A. Chernov __collate_substitute(s)
99c3d0cca4SAndrey A. Chernov 	const u_char *s;
100c3d0cca4SAndrey A. Chernov {
101c3d0cca4SAndrey A. Chernov 	int dest_len = 0, len = 0;
102c3d0cca4SAndrey A. Chernov 	int delta = strlen(s);
103926f20c9SAndrey A. Chernov 	u_char *dest_str = NULL;
104c3d0cca4SAndrey A. Chernov 
105926f20c9SAndrey A. Chernov 	if(s == NULL || *s == '\0')
106c3d0cca4SAndrey A. Chernov 		return __collate_strdup("");
107c3d0cca4SAndrey A. Chernov 	while(*s) {
108c3d0cca4SAndrey A. Chernov 		len += strlen(__collate_substitute_table[*s]);
109c3d0cca4SAndrey A. Chernov 		while(dest_len <= len) {
110c3d0cca4SAndrey A. Chernov 			if(!dest_str)
111c3d0cca4SAndrey A. Chernov 				dest_str = calloc(dest_len = delta, 1);
112c3d0cca4SAndrey A. Chernov 			else
113c3d0cca4SAndrey A. Chernov 				dest_str = realloc(dest_str, dest_len += delta);
114926f20c9SAndrey A. Chernov 			if(dest_str == NULL)
115926f20c9SAndrey A. Chernov 				__collate_err(EX_OSERR, __FUNCTION__);
116c3d0cca4SAndrey A. Chernov 		}
117c3d0cca4SAndrey A. Chernov 		strcat(dest_str, __collate_substitute_table[*s++]);
118c3d0cca4SAndrey A. Chernov 	}
119c3d0cca4SAndrey A. Chernov 	return dest_str;
120c3d0cca4SAndrey A. Chernov }
121c3d0cca4SAndrey A. Chernov 
122c3d0cca4SAndrey A. Chernov void
123c3d0cca4SAndrey A. Chernov __collate_lookup(t, len, prim, sec)
124c3d0cca4SAndrey A. Chernov 	u_char *t;
125c3d0cca4SAndrey A. Chernov 	int *len, *prim, *sec;
126c3d0cca4SAndrey A. Chernov {
127c3d0cca4SAndrey A. Chernov 	struct __collate_st_name_pri *p;
128c3d0cca4SAndrey A. Chernov 	struct __collate_st_chain_pri *p2;
129c3d0cca4SAndrey A. Chernov 
130c3d0cca4SAndrey A. Chernov 	*len = 1;
131c3d0cca4SAndrey A. Chernov 	*prim = *sec = 0;
132c3d0cca4SAndrey A. Chernov 	if(__collate_charmap_table[*t][0]) {
133c3d0cca4SAndrey A. Chernov 		for(p = __collate_name_pri_table; p->str[0]; p++) {
134c3d0cca4SAndrey A. Chernov 			if(strncmp(__collate_charmap_table[*t], p->str, strlen(p->str))
135c3d0cca4SAndrey A. Chernov 			   == 0) {
136c3d0cca4SAndrey A. Chernov 				*prim = p->prim;
137c3d0cca4SAndrey A. Chernov 				*sec = p->sec;
138c3d0cca4SAndrey A. Chernov 				return;
139c3d0cca4SAndrey A. Chernov 			}
140c3d0cca4SAndrey A. Chernov 		}
141c3d0cca4SAndrey A. Chernov 		return;
142c3d0cca4SAndrey A. Chernov 	}
143c3d0cca4SAndrey A. Chernov 	for(p2 = __collate_chain_pri_table; p2->str[0]; p2++) {
144c3d0cca4SAndrey A. Chernov 		if(strncmp(t, p2->str, strlen(p2->str)) == 0) {
145c3d0cca4SAndrey A. Chernov 			*len = strlen(p2->str);
146c3d0cca4SAndrey A. Chernov 			*prim = p2->prim;
147c3d0cca4SAndrey A. Chernov 			*sec = p2->sec;
148c3d0cca4SAndrey A. Chernov 			return;
149c3d0cca4SAndrey A. Chernov 		}
150c3d0cca4SAndrey A. Chernov 	}
151c3d0cca4SAndrey A. Chernov 	*prim = __collate_char_pri_table[*t].prim;
152c3d0cca4SAndrey A. Chernov 	*sec = __collate_char_pri_table[*t].sec;
153c3d0cca4SAndrey A. Chernov }
154c3d0cca4SAndrey A. Chernov 
155c3d0cca4SAndrey A. Chernov u_char *
156c3d0cca4SAndrey A. Chernov __collate_strdup(s)
157c3d0cca4SAndrey A. Chernov 	u_char *s;
158c3d0cca4SAndrey A. Chernov {
159926f20c9SAndrey A. Chernov 	u_char *t = strdup(s);
160c3d0cca4SAndrey A. Chernov 
161926f20c9SAndrey A. Chernov 	if (t == NULL)
162926f20c9SAndrey A. Chernov 		__collate_err(EX_OSERR, __FUNCTION__);
163c3d0cca4SAndrey A. Chernov 	return t;
164c3d0cca4SAndrey A. Chernov }
165c3d0cca4SAndrey A. Chernov 
166926f20c9SAndrey A. Chernov __dead void __collate_err(int ex, const char *f)
167926f20c9SAndrey A. Chernov {
168926f20c9SAndrey A. Chernov 	extern char *__progname;                /* Program name, from crt0. */
169926f20c9SAndrey A. Chernov 	const char *s;
170926f20c9SAndrey A. Chernov 	int serrno = errno;
171926f20c9SAndrey A. Chernov 
172926f20c9SAndrey A. Chernov 	s = __progname;
173926f20c9SAndrey A. Chernov 	write(STDERR_FILENO, s, strlen(s));
1743d4c9a4bSAndrey A. Chernov 	write(STDERR_FILENO, ": ", 2);
175926f20c9SAndrey A. Chernov 	s = f;
176926f20c9SAndrey A. Chernov 	write(STDERR_FILENO, s, strlen(s));
1773d4c9a4bSAndrey A. Chernov 	write(STDERR_FILENO, ": ", 2);
178926f20c9SAndrey A. Chernov 	s = strerror(serrno);
179926f20c9SAndrey A. Chernov 	write(STDERR_FILENO, s, strlen(s));
1803d4c9a4bSAndrey A. Chernov 	write(STDERR_FILENO, "\n", 1);
181926f20c9SAndrey A. Chernov 	exit(ex);
182926f20c9SAndrey A. Chernov }
183926f20c9SAndrey A. Chernov 
184c3d0cca4SAndrey A. Chernov #ifdef COLLATE_DEBUG
185c3d0cca4SAndrey A. Chernov void
186c3d0cca4SAndrey A. Chernov __collate_print_tables()
187c3d0cca4SAndrey A. Chernov {
188c3d0cca4SAndrey A. Chernov 	int i;
189c3d0cca4SAndrey A. Chernov 	struct __collate_st_name_pri *p;
190c3d0cca4SAndrey A. Chernov 	struct __collate_st_chain_pri *p2;
191c3d0cca4SAndrey A. Chernov 
192c3d0cca4SAndrey A. Chernov 	printf("Substitute table:\n");
193c3d0cca4SAndrey A. Chernov 	for (i = 0; i < UCHAR_MAX + 1; i++)
194c3d0cca4SAndrey A. Chernov 		printf("\t'%c' --> \"%s\"\n", i,
195c3d0cca4SAndrey A. Chernov 		       __collate_substitute_table[i]);
196c3d0cca4SAndrey A. Chernov 	printf("Charmap table:\n");
197c3d0cca4SAndrey A. Chernov 	for (i = 0; i < UCHAR_MAX + 1; i++) {
198c3d0cca4SAndrey A. Chernov 		if (__collate_charmap_table[i][0])
199c3d0cca4SAndrey A. Chernov 			printf("\t\\x%02x --> \"%s\"\n", i,
200c3d0cca4SAndrey A. Chernov 			       __collate_charmap_table[i]);
201c3d0cca4SAndrey A. Chernov 	}
202c3d0cca4SAndrey A. Chernov 	printf("Name priority table:\n");
203c3d0cca4SAndrey A. Chernov 	for (p = __collate_name_pri_table; p->str[0]; p++)
204c3d0cca4SAndrey A. Chernov 		printf("\t\"%s\" : %d %d\n\n", p->str, p->prim, p->sec);
205c3d0cca4SAndrey A. Chernov 	printf("Chain priority table:\n");
206c3d0cca4SAndrey A. Chernov 	for (p2 = __collate_chain_pri_table; p2->str[0]; p2++)
207c3d0cca4SAndrey A. Chernov 		printf("\t\"%s\" : %d %d\n\n", p2->str, p2->prim, p2->sec);
208c3d0cca4SAndrey A. Chernov 	printf("Char priority table:\n");
209c3d0cca4SAndrey A. Chernov 	for (i = 0; i < UCHAR_MAX + 1; i++)
210c3d0cca4SAndrey A. Chernov 		printf("\t'%c' : %d %d\n", i, __collate_char_pri_table[i].prim,
211c3d0cca4SAndrey A. Chernov 		       __collate_char_pri_table[i].sec);
212c3d0cca4SAndrey A. Chernov }
213c3d0cca4SAndrey A. Chernov #endif
214