xref: /freebsd/lib/libc/locale/collate.c (revision 63407d3487909d41524a1ae94e60b2fcec7e5c27)
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  *
271130b656SJordan K. Hubbard  * $FreeBSD$
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"
3863407d34SAndrey A. Chernov #include "setlocale.h"
39c3d0cca4SAndrey A. Chernov 
40c3d0cca4SAndrey A. Chernov int __collate_load_error = 1;
411642f84dSAndrey A. Chernov char __collate_version[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_chain_pri __collate_chain_pri_table[TABLE_SIZE];
45c3d0cca4SAndrey A. Chernov 
46c3d0cca4SAndrey A. Chernov #define FREAD(a, b, c, d) \
47c3d0cca4SAndrey A. Chernov 	do { \
48c3d0cca4SAndrey A. Chernov 		if(fread(a, b, c, d) != c) { \
49c3d0cca4SAndrey A. Chernov 			fclose(d); \
50c3d0cca4SAndrey A. Chernov 			return -1; \
51c3d0cca4SAndrey A. Chernov 		} \
52c3d0cca4SAndrey A. Chernov 	} while(0)
53c3d0cca4SAndrey A. Chernov 
54eaa86f9dSBruce Evans void __collate_err(int ex, const char *f) __dead2;
55926f20c9SAndrey A. Chernov 
56c3d0cca4SAndrey A. Chernov int
57c3d0cca4SAndrey A. Chernov __collate_load_tables(encoding)
58c3d0cca4SAndrey A. Chernov 	char *encoding;
59c3d0cca4SAndrey A. Chernov {
60c3d0cca4SAndrey A. Chernov 	char buf[PATH_MAX];
61c3d0cca4SAndrey A. Chernov 	FILE *fp;
62377da8e8SAndrey A. Chernov 	int save_load_error;
63c3d0cca4SAndrey A. Chernov 
64377da8e8SAndrey A. Chernov 	save_load_error = __collate_load_error;
65c3d0cca4SAndrey A. Chernov 	__collate_load_error = 1;
66377da8e8SAndrey A. Chernov 	if (!encoding) {
67377da8e8SAndrey A. Chernov 		__collate_load_error = save_load_error;
68c3d0cca4SAndrey A. Chernov 		return -1;
69377da8e8SAndrey A. Chernov 	}
70c8f931a8SAndrey A. Chernov 	if (!strcmp(encoding, "C") || !strcmp(encoding, "POSIX"))
71377da8e8SAndrey A. Chernov 		return 0;
72c8f931a8SAndrey A. Chernov 	if (!_PathLocale) {
73c8f931a8SAndrey A. Chernov 		__collate_load_error = save_load_error;
74c8f931a8SAndrey A. Chernov 		return -1;
75c8f931a8SAndrey A. Chernov 	}
76d81a0916SAndrey A. Chernov 	/* Range checking not needed, encoding has fixed size */
77af155bdfSAndrey A. Chernov 	(void) strcpy(buf, _PathLocale);
78af155bdfSAndrey A. Chernov 	(void) strcat(buf, "/");
79af155bdfSAndrey A. Chernov 	(void) strcat(buf, encoding);
80af155bdfSAndrey A. Chernov 	(void) strcat(buf, "/LC_COLLATE");
81377da8e8SAndrey A. Chernov 	if ((fp = fopen(buf, "r")) == NULL) {
82377da8e8SAndrey A. Chernov 		__collate_load_error = save_load_error;
83c3d0cca4SAndrey A. Chernov 		return -1;
84377da8e8SAndrey A. Chernov 	}
851642f84dSAndrey A. Chernov 	FREAD(__collate_version, sizeof(__collate_version), 1, fp);
861642f84dSAndrey A. Chernov 	if (strcmp(__collate_version, COLLATE_VERSION) != 0) {
871642f84dSAndrey A. Chernov 		fclose(fp);
881642f84dSAndrey A. Chernov 		return -1;
891642f84dSAndrey A. Chernov 	}
90c3d0cca4SAndrey A. Chernov 	FREAD(__collate_substitute_table, sizeof(__collate_substitute_table),
91c3d0cca4SAndrey A. Chernov 	      1, fp);
92c3d0cca4SAndrey A. Chernov 	FREAD(__collate_char_pri_table, sizeof(__collate_char_pri_table), 1,
93c3d0cca4SAndrey A. Chernov 	      fp);
94c3d0cca4SAndrey A. Chernov 	FREAD(__collate_chain_pri_table, sizeof(__collate_chain_pri_table), 1,
95c3d0cca4SAndrey A. Chernov 	      fp);
96c3d0cca4SAndrey A. Chernov 	fclose(fp);
97c3d0cca4SAndrey A. Chernov 	__collate_load_error = 0;
98c3d0cca4SAndrey A. Chernov 	return 0;
99c3d0cca4SAndrey A. Chernov }
100c3d0cca4SAndrey A. Chernov 
101c3d0cca4SAndrey A. Chernov u_char *
102c3d0cca4SAndrey A. Chernov __collate_substitute(s)
103c3d0cca4SAndrey A. Chernov 	const u_char *s;
104c3d0cca4SAndrey A. Chernov {
105c3d0cca4SAndrey A. Chernov 	int dest_len = 0, len = 0;
106c3d0cca4SAndrey A. Chernov 	int delta = strlen(s);
107926f20c9SAndrey A. Chernov 	u_char *dest_str = NULL;
108c3d0cca4SAndrey A. Chernov 
109926f20c9SAndrey A. Chernov 	if(s == NULL || *s == '\0')
110c3d0cca4SAndrey A. Chernov 		return __collate_strdup("");
111c3d0cca4SAndrey A. Chernov 	while(*s) {
112c3d0cca4SAndrey A. Chernov 		len += strlen(__collate_substitute_table[*s]);
113c3d0cca4SAndrey A. Chernov 		while(dest_len <= len) {
114c3d0cca4SAndrey A. Chernov 			if(!dest_str)
115c3d0cca4SAndrey A. Chernov 				dest_str = calloc(dest_len = delta, 1);
116c3d0cca4SAndrey A. Chernov 			else
117c3d0cca4SAndrey A. Chernov 				dest_str = realloc(dest_str, dest_len += delta);
118926f20c9SAndrey A. Chernov 			if(dest_str == NULL)
119926f20c9SAndrey A. Chernov 				__collate_err(EX_OSERR, __FUNCTION__);
120c3d0cca4SAndrey A. Chernov 		}
121c3d0cca4SAndrey A. Chernov 		strcat(dest_str, __collate_substitute_table[*s++]);
122c3d0cca4SAndrey A. Chernov 	}
123c3d0cca4SAndrey A. Chernov 	return dest_str;
124c3d0cca4SAndrey A. Chernov }
125c3d0cca4SAndrey A. Chernov 
126c3d0cca4SAndrey A. Chernov void
127c3d0cca4SAndrey A. Chernov __collate_lookup(t, len, prim, sec)
128c3d0cca4SAndrey A. Chernov 	u_char *t;
129c3d0cca4SAndrey A. Chernov 	int *len, *prim, *sec;
130c3d0cca4SAndrey A. Chernov {
131c3d0cca4SAndrey A. Chernov 	struct __collate_st_chain_pri *p2;
132c3d0cca4SAndrey A. Chernov 
133c3d0cca4SAndrey A. Chernov 	*len = 1;
134c3d0cca4SAndrey A. Chernov 	*prim = *sec = 0;
135c3d0cca4SAndrey A. Chernov 	for(p2 = __collate_chain_pri_table; p2->str[0]; p2++) {
136c3d0cca4SAndrey A. Chernov 		if(strncmp(t, p2->str, strlen(p2->str)) == 0) {
137c3d0cca4SAndrey A. Chernov 			*len = strlen(p2->str);
138c3d0cca4SAndrey A. Chernov 			*prim = p2->prim;
139c3d0cca4SAndrey A. Chernov 			*sec = p2->sec;
140c3d0cca4SAndrey A. Chernov 			return;
141c3d0cca4SAndrey A. Chernov 		}
142c3d0cca4SAndrey A. Chernov 	}
143c3d0cca4SAndrey A. Chernov 	*prim = __collate_char_pri_table[*t].prim;
144c3d0cca4SAndrey A. Chernov 	*sec = __collate_char_pri_table[*t].sec;
145c3d0cca4SAndrey A. Chernov }
146c3d0cca4SAndrey A. Chernov 
147c3d0cca4SAndrey A. Chernov u_char *
148c3d0cca4SAndrey A. Chernov __collate_strdup(s)
149c3d0cca4SAndrey A. Chernov 	u_char *s;
150c3d0cca4SAndrey A. Chernov {
151926f20c9SAndrey A. Chernov 	u_char *t = strdup(s);
152c3d0cca4SAndrey A. Chernov 
153926f20c9SAndrey A. Chernov 	if (t == NULL)
154926f20c9SAndrey A. Chernov 		__collate_err(EX_OSERR, __FUNCTION__);
155c3d0cca4SAndrey A. Chernov 	return t;
156c3d0cca4SAndrey A. Chernov }
157c3d0cca4SAndrey A. Chernov 
158eaa86f9dSBruce Evans void
159eaa86f9dSBruce Evans __collate_err(int ex, const char *f)
160926f20c9SAndrey A. Chernov {
161926f20c9SAndrey A. Chernov 	extern char *__progname;                /* Program name, from crt0. */
162926f20c9SAndrey A. Chernov 	const char *s;
163926f20c9SAndrey A. Chernov 	int serrno = errno;
164926f20c9SAndrey A. Chernov 
165926f20c9SAndrey A. Chernov 	s = __progname;
166926f20c9SAndrey A. Chernov 	write(STDERR_FILENO, s, strlen(s));
1673d4c9a4bSAndrey A. Chernov 	write(STDERR_FILENO, ": ", 2);
168926f20c9SAndrey A. Chernov 	s = f;
169926f20c9SAndrey A. Chernov 	write(STDERR_FILENO, s, strlen(s));
1703d4c9a4bSAndrey A. Chernov 	write(STDERR_FILENO, ": ", 2);
171926f20c9SAndrey A. Chernov 	s = strerror(serrno);
172926f20c9SAndrey A. Chernov 	write(STDERR_FILENO, s, strlen(s));
1733d4c9a4bSAndrey A. Chernov 	write(STDERR_FILENO, "\n", 1);
174926f20c9SAndrey A. Chernov 	exit(ex);
175926f20c9SAndrey A. Chernov }
176926f20c9SAndrey A. Chernov 
177c3d0cca4SAndrey A. Chernov #ifdef COLLATE_DEBUG
178c3d0cca4SAndrey A. Chernov void
179c3d0cca4SAndrey A. Chernov __collate_print_tables()
180c3d0cca4SAndrey A. Chernov {
181c3d0cca4SAndrey A. Chernov 	int i;
182c3d0cca4SAndrey A. Chernov 	struct __collate_st_chain_pri *p2;
183c3d0cca4SAndrey A. Chernov 
184c3d0cca4SAndrey A. Chernov 	printf("Substitute table:\n");
185c3d0cca4SAndrey A. Chernov 	for (i = 0; i < UCHAR_MAX + 1; i++)
186967a5cb1SAndrey A. Chernov 	    if (i != *__collate_substitute_table[i])
187c3d0cca4SAndrey A. Chernov 		printf("\t'%c' --> \"%s\"\n", i,
188c3d0cca4SAndrey A. Chernov 		       __collate_substitute_table[i]);
189c3d0cca4SAndrey A. Chernov 	printf("Chain priority table:\n");
190c3d0cca4SAndrey A. Chernov 	for (p2 = __collate_chain_pri_table; p2->str[0]; p2++)
191c3d0cca4SAndrey A. Chernov 		printf("\t\"%s\" : %d %d\n\n", p2->str, p2->prim, p2->sec);
192c3d0cca4SAndrey A. Chernov 	printf("Char priority table:\n");
193c3d0cca4SAndrey A. Chernov 	for (i = 0; i < UCHAR_MAX + 1; i++)
194c3d0cca4SAndrey A. Chernov 		printf("\t'%c' : %d %d\n", i, __collate_char_pri_table[i].prim,
195c3d0cca4SAndrey A. Chernov 		       __collate_char_pri_table[i].sec);
196c3d0cca4SAndrey A. Chernov }
197c3d0cca4SAndrey A. Chernov #endif
198