xref: /freebsd/lib/libc/locale/collate.c (revision c3d0cca4e90381dd0ec3c78b0720dfcca6682dbe)
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  *
27c3d0cca4SAndrey A. Chernov  * $Id: collate.c,v 1.8 1995/01/27 12:51:06 alex Exp alex $
28c3d0cca4SAndrey A. Chernov  */
29c3d0cca4SAndrey A. Chernov 
30c3d0cca4SAndrey A. Chernov #include <err.h>
31c3d0cca4SAndrey A. Chernov #include <rune.h>
32c3d0cca4SAndrey A. Chernov #include <stdio.h>
33c3d0cca4SAndrey A. Chernov #include <stdlib.h>
34c3d0cca4SAndrey A. Chernov #include <string.h>
35c3d0cca4SAndrey A. Chernov #include <sysexits.h>
36c3d0cca4SAndrey A. Chernov #include "collate.h"
37c3d0cca4SAndrey A. Chernov 
38c3d0cca4SAndrey A. Chernov int __collate_load_error = 1;
39c3d0cca4SAndrey A. Chernov u_char __collate_charmap_table[UCHAR_MAX + 1][STR_LEN];
40c3d0cca4SAndrey A. Chernov u_char __collate_substitute_table[UCHAR_MAX + 1][STR_LEN];
41c3d0cca4SAndrey A. Chernov struct __collate_st_char_pri __collate_char_pri_table[UCHAR_MAX + 1];
42c3d0cca4SAndrey A. Chernov struct __collate_st_name_pri __collate_name_pri_table[TABLE_SIZE];
43c3d0cca4SAndrey A. Chernov struct __collate_st_chain_pri __collate_chain_pri_table[TABLE_SIZE];
44c3d0cca4SAndrey A. Chernov 
45c3d0cca4SAndrey A. Chernov #define FREAD(a, b, c, d) \
46c3d0cca4SAndrey A. Chernov 	do { \
47c3d0cca4SAndrey A. Chernov 		if(fread(a, b, c, d) != c) { \
48c3d0cca4SAndrey A. Chernov 			fclose(d); \
49c3d0cca4SAndrey A. Chernov 			return -1; \
50c3d0cca4SAndrey A. Chernov 		} \
51c3d0cca4SAndrey A. Chernov 	} while(0)
52c3d0cca4SAndrey A. Chernov 
53c3d0cca4SAndrey A. Chernov  int
54c3d0cca4SAndrey A. Chernov __collate_load_tables(encoding)
55c3d0cca4SAndrey A. Chernov 	char *encoding;
56c3d0cca4SAndrey A. Chernov {
57c3d0cca4SAndrey A. Chernov 	char buf[PATH_MAX];
58c3d0cca4SAndrey A. Chernov 	static char *path_locale;
59c3d0cca4SAndrey A. Chernov 	FILE *fp;
60c3d0cca4SAndrey A. Chernov 
61c3d0cca4SAndrey A. Chernov 	__collate_load_error = 1;
62c3d0cca4SAndrey A. Chernov 	if (!encoding)
63c3d0cca4SAndrey A. Chernov 		return -1;
64c3d0cca4SAndrey A. Chernov 	if (!path_locale && !(path_locale = getenv("PATH_LOCALE")))
65c3d0cca4SAndrey A. Chernov 		path_locale = _PATH_LOCALE;
66c3d0cca4SAndrey A. Chernov 	strcpy(buf, path_locale);
67c3d0cca4SAndrey A. Chernov 	strcat(buf, "/");
68c3d0cca4SAndrey A. Chernov 	strcat(buf, encoding);
69c3d0cca4SAndrey A. Chernov 	strcat(buf, "/LC_COLLATE");
70c3d0cca4SAndrey A. Chernov 	if ((fp = fopen(buf, "r")) == NULL)
71c3d0cca4SAndrey A. Chernov 		return -1;
72c3d0cca4SAndrey A. Chernov 	FREAD(__collate_charmap_table, sizeof(__collate_charmap_table), 1, fp);
73c3d0cca4SAndrey A. Chernov 	FREAD(__collate_substitute_table, sizeof(__collate_substitute_table),
74c3d0cca4SAndrey A. Chernov 	      1, fp);
75c3d0cca4SAndrey A. Chernov 	FREAD(__collate_char_pri_table, sizeof(__collate_char_pri_table), 1,
76c3d0cca4SAndrey A. Chernov 	      fp);
77c3d0cca4SAndrey A. Chernov 	FREAD(__collate_chain_pri_table, sizeof(__collate_chain_pri_table), 1,
78c3d0cca4SAndrey A. Chernov 	      fp);
79c3d0cca4SAndrey A. Chernov 	FREAD(__collate_name_pri_table, sizeof(__collate_name_pri_table), 1,
80c3d0cca4SAndrey A. Chernov 	      fp);
81c3d0cca4SAndrey A. Chernov 	fclose(fp);
82c3d0cca4SAndrey A. Chernov 	__collate_load_error = 0;
83c3d0cca4SAndrey A. Chernov 	return 0;
84c3d0cca4SAndrey A. Chernov }
85c3d0cca4SAndrey A. Chernov 
86c3d0cca4SAndrey A. Chernov u_char *
87c3d0cca4SAndrey A. Chernov __collate_substitute(s)
88c3d0cca4SAndrey A. Chernov 	const u_char *s;
89c3d0cca4SAndrey A. Chernov {
90c3d0cca4SAndrey A. Chernov 	int dest_len = 0, len = 0;
91c3d0cca4SAndrey A. Chernov 	int delta = strlen(s);
92c3d0cca4SAndrey A. Chernov 	u_char *dest_str = 0;
93c3d0cca4SAndrey A. Chernov 
94c3d0cca4SAndrey A. Chernov 	if(s == 0 || *s == '\0')
95c3d0cca4SAndrey A. Chernov 		return __collate_strdup("");
96c3d0cca4SAndrey A. Chernov 	while(*s) {
97c3d0cca4SAndrey A. Chernov 		len += strlen(__collate_substitute_table[*s]);
98c3d0cca4SAndrey A. Chernov 		while(dest_len <= len) {
99c3d0cca4SAndrey A. Chernov 			if(!dest_str)
100c3d0cca4SAndrey A. Chernov 				dest_str = calloc(dest_len = delta, 1);
101c3d0cca4SAndrey A. Chernov 			else
102c3d0cca4SAndrey A. Chernov 				dest_str = realloc(dest_str, dest_len += delta);
103c3d0cca4SAndrey A. Chernov 			if(!dest_str)
104c3d0cca4SAndrey A. Chernov 				err(EX_OSERR, "%s(\"%s\")", __FUNCTION__, s);
105c3d0cca4SAndrey A. Chernov 		}
106c3d0cca4SAndrey A. Chernov 		strcat(dest_str, __collate_substitute_table[*s++]);
107c3d0cca4SAndrey A. Chernov 	}
108c3d0cca4SAndrey A. Chernov 	return dest_str;
109c3d0cca4SAndrey A. Chernov }
110c3d0cca4SAndrey A. Chernov 
111c3d0cca4SAndrey A. Chernov void
112c3d0cca4SAndrey A. Chernov __collate_lookup(t, len, prim, sec)
113c3d0cca4SAndrey A. Chernov 	u_char *t;
114c3d0cca4SAndrey A. Chernov 	int *len, *prim, *sec;
115c3d0cca4SAndrey A. Chernov {
116c3d0cca4SAndrey A. Chernov 	struct __collate_st_name_pri *p;
117c3d0cca4SAndrey A. Chernov 	struct __collate_st_chain_pri *p2;
118c3d0cca4SAndrey A. Chernov 
119c3d0cca4SAndrey A. Chernov 	*len = 1;
120c3d0cca4SAndrey A. Chernov 	*prim = *sec = 0;
121c3d0cca4SAndrey A. Chernov 	if(__collate_charmap_table[*t][0]) {
122c3d0cca4SAndrey A. Chernov 		for(p = __collate_name_pri_table; p->str[0]; p++) {
123c3d0cca4SAndrey A. Chernov 			if(strncmp(__collate_charmap_table[*t], p->str, strlen(p->str))
124c3d0cca4SAndrey A. Chernov 			   == 0) {
125c3d0cca4SAndrey A. Chernov 				*prim = p->prim;
126c3d0cca4SAndrey A. Chernov 				*sec = p->sec;
127c3d0cca4SAndrey A. Chernov 				return;
128c3d0cca4SAndrey A. Chernov 			}
129c3d0cca4SAndrey A. Chernov 		}
130c3d0cca4SAndrey A. Chernov 		return;
131c3d0cca4SAndrey A. Chernov 	}
132c3d0cca4SAndrey A. Chernov 	for(p2 = __collate_chain_pri_table; p2->str[0]; p2++) {
133c3d0cca4SAndrey A. Chernov 		if(strncmp(t, p2->str, strlen(p2->str)) == 0) {
134c3d0cca4SAndrey A. Chernov 			*len = strlen(p2->str);
135c3d0cca4SAndrey A. Chernov 			*prim = p2->prim;
136c3d0cca4SAndrey A. Chernov 			*sec = p2->sec;
137c3d0cca4SAndrey A. Chernov 			return;
138c3d0cca4SAndrey A. Chernov 		}
139c3d0cca4SAndrey A. Chernov 	}
140c3d0cca4SAndrey A. Chernov 	*prim = __collate_char_pri_table[*t].prim;
141c3d0cca4SAndrey A. Chernov 	*sec = __collate_char_pri_table[*t].sec;
142c3d0cca4SAndrey A. Chernov }
143c3d0cca4SAndrey A. Chernov 
144c3d0cca4SAndrey A. Chernov u_char *
145c3d0cca4SAndrey A. Chernov __collate_strdup(s)
146c3d0cca4SAndrey A. Chernov 	u_char *s;
147c3d0cca4SAndrey A. Chernov {
148c3d0cca4SAndrey A. Chernov 	void *t = strdup(s);
149c3d0cca4SAndrey A. Chernov 
150c3d0cca4SAndrey A. Chernov 	if (!t)
151c3d0cca4SAndrey A. Chernov 		err(EX_OSERR, "__collate_strdup(\"%s\")", s);
152c3d0cca4SAndrey A. Chernov 	return t;
153c3d0cca4SAndrey A. Chernov }
154c3d0cca4SAndrey A. Chernov 
155c3d0cca4SAndrey A. Chernov #ifdef COLLATE_DEBUG
156c3d0cca4SAndrey A. Chernov void
157c3d0cca4SAndrey A. Chernov __collate_print_tables()
158c3d0cca4SAndrey A. Chernov {
159c3d0cca4SAndrey A. Chernov 	int i;
160c3d0cca4SAndrey A. Chernov 	struct __collate_st_name_pri *p;
161c3d0cca4SAndrey A. Chernov 	struct __collate_st_chain_pri *p2;
162c3d0cca4SAndrey A. Chernov 
163c3d0cca4SAndrey A. Chernov 	printf("Substitute table:\n");
164c3d0cca4SAndrey A. Chernov 	for (i = 0; i < UCHAR_MAX + 1; i++)
165c3d0cca4SAndrey A. Chernov 		printf("\t'%c' --> \"%s\"\n", i,
166c3d0cca4SAndrey A. Chernov 		       __collate_substitute_table[i]);
167c3d0cca4SAndrey A. Chernov 	printf("Charmap table:\n");
168c3d0cca4SAndrey A. Chernov 	for (i = 0; i < UCHAR_MAX + 1; i++) {
169c3d0cca4SAndrey A. Chernov 		if (__collate_charmap_table[i][0])
170c3d0cca4SAndrey A. Chernov 			printf("\t\\x%02x --> \"%s\"\n", i,
171c3d0cca4SAndrey A. Chernov 			       __collate_charmap_table[i]);
172c3d0cca4SAndrey A. Chernov 	}
173c3d0cca4SAndrey A. Chernov 	printf("Name priority table:\n");
174c3d0cca4SAndrey A. Chernov 	for (p = __collate_name_pri_table; p->str[0]; p++)
175c3d0cca4SAndrey A. Chernov 		printf("\t\"%s\" : %d %d\n\n", p->str, p->prim, p->sec);
176c3d0cca4SAndrey A. Chernov 	printf("Chain priority table:\n");
177c3d0cca4SAndrey A. Chernov 	for (p2 = __collate_chain_pri_table; p2->str[0]; p2++)
178c3d0cca4SAndrey A. Chernov 		printf("\t\"%s\" : %d %d\n\n", p2->str, p2->prim, p2->sec);
179c3d0cca4SAndrey A. Chernov 	printf("Char priority table:\n");
180c3d0cca4SAndrey A. Chernov 	for (i = 0; i < UCHAR_MAX + 1; i++)
181c3d0cca4SAndrey A. Chernov 		printf("\t'%c' : %d %d\n", i, __collate_char_pri_table[i].prim,
182c3d0cca4SAndrey A. Chernov 		       __collate_char_pri_table[i].sec);
183c3d0cca4SAndrey A. Chernov }
184c3d0cca4SAndrey A. Chernov #endif
185