1*16d86563SAlexander Pyhalov /*
2*16d86563SAlexander Pyhalov  * CDDL HEADER START
3*16d86563SAlexander Pyhalov  *
4*16d86563SAlexander Pyhalov  * The contents of this file are subject to the terms of the
5*16d86563SAlexander Pyhalov  * Common Development and Distribution License (the "License").
6*16d86563SAlexander Pyhalov  * You may not use this file except in compliance with the License.
7*16d86563SAlexander Pyhalov  *
8*16d86563SAlexander Pyhalov  * You can obtain a copy of the license at src/OPENSOLARIS.LICENSE
9*16d86563SAlexander Pyhalov  * or http://www.opensolaris.org/os/licensing.
10*16d86563SAlexander Pyhalov  * See the License for the specific language governing permissions
11*16d86563SAlexander Pyhalov  * and limitations under the License.
12*16d86563SAlexander Pyhalov  *
13*16d86563SAlexander Pyhalov  * When distributing Covered Code, include this CDDL HEADER in each
14*16d86563SAlexander Pyhalov  * file and include the License file at src/OPENSOLARIS.LICENSE.
15*16d86563SAlexander Pyhalov  * If applicable, add the following below this CDDL HEADER, with the
16*16d86563SAlexander Pyhalov  * fields enclosed by brackets "[]" replaced with your own identifying
17*16d86563SAlexander Pyhalov  * information: Portions Copyright [yyyy] [name of copyright owner]
18*16d86563SAlexander Pyhalov  *
19*16d86563SAlexander Pyhalov  * CDDL HEADER END
20*16d86563SAlexander Pyhalov  */
21*16d86563SAlexander Pyhalov /*
22*16d86563SAlexander Pyhalov  * Copyright 2004 Sun Microsystems, Inc.  All rights reserved.
23*16d86563SAlexander Pyhalov  * Use is subject to license terms.
24*16d86563SAlexander Pyhalov  */
25*16d86563SAlexander Pyhalov 
26*16d86563SAlexander Pyhalov 
27*16d86563SAlexander Pyhalov 
28*16d86563SAlexander Pyhalov #include <stdio.h>
29*16d86563SAlexander Pyhalov #include <stdlib.h>
30*16d86563SAlexander Pyhalov #include <ctype.h>
31*16d86563SAlexander Pyhalov #include <strings.h>
32*16d86563SAlexander Pyhalov #include "../common_defs.h"
33*16d86563SAlexander Pyhalov 
34*16d86563SAlexander Pyhalov int
main(int ac,char ** av)35*16d86563SAlexander Pyhalov main(int ac, char **av)
36*16d86563SAlexander Pyhalov {
37*16d86563SAlexander Pyhalov 	to_utf8_table_component_t tbl[256];
38*16d86563SAlexander Pyhalov 	register int i, j;
39*16d86563SAlexander Pyhalov 	char buf[BUFSIZ], num[100];
40*16d86563SAlexander Pyhalov 	unsigned int l, k;
41*16d86563SAlexander Pyhalov 	char ascii_only = 0;
42*16d86563SAlexander Pyhalov 
43*16d86563SAlexander Pyhalov 	if (ac > 1 && strcmp(av[1], "-ascii") == 0)
44*16d86563SAlexander Pyhalov 		ascii_only = 1;
45*16d86563SAlexander Pyhalov 
46*16d86563SAlexander Pyhalov 	for (i = 0; i < 256; i++) {
47*16d86563SAlexander Pyhalov 		if (i <= 0x1f || i == 0x7f || (ascii_only && i <= 0x7f)) {
48*16d86563SAlexander Pyhalov 			tbl[i].size = 1;
49*16d86563SAlexander Pyhalov 			tbl[i].u8 = (unsigned int)i;
50*16d86563SAlexander Pyhalov 		} else if (!ascii_only && (i >= 0x80 && i <= 0x9f)) {
51*16d86563SAlexander Pyhalov 			tbl[i].size = 2;
52*16d86563SAlexander Pyhalov 			tbl[i].u8 = (unsigned int)i;
53*16d86563SAlexander Pyhalov 		} else {
54*16d86563SAlexander Pyhalov 			tbl[i].size = ICV_TYPE_ILLEGAL_CHAR;
55*16d86563SAlexander Pyhalov 			tbl[i].u8 = 0;
56*16d86563SAlexander Pyhalov 		}
57*16d86563SAlexander Pyhalov 	}
58*16d86563SAlexander Pyhalov 
59*16d86563SAlexander Pyhalov 
60*16d86563SAlexander Pyhalov 	while (fgets(buf, BUFSIZ, stdin)) {
61*16d86563SAlexander Pyhalov 		i = 0;
62*16d86563SAlexander Pyhalov 		while (buf[i] && isspace(buf[i]))
63*16d86563SAlexander Pyhalov 			i++;
64*16d86563SAlexander Pyhalov 		if (buf[i] == '#' || buf[i] == '\0')
65*16d86563SAlexander Pyhalov 			continue;
66*16d86563SAlexander Pyhalov 
67*16d86563SAlexander Pyhalov 		for (j = 0; !isspace(buf[i]); i++, j++)
68*16d86563SAlexander Pyhalov 			num[j] = buf[i];
69*16d86563SAlexander Pyhalov 		num[j] = '\0';
70*16d86563SAlexander Pyhalov 
71*16d86563SAlexander Pyhalov 		k = strtol(num, (char **)NULL, 0);
72*16d86563SAlexander Pyhalov 
73*16d86563SAlexander Pyhalov 		while (isspace(buf[i]))
74*16d86563SAlexander Pyhalov 			i++;
75*16d86563SAlexander Pyhalov 
76*16d86563SAlexander Pyhalov 		/* Take care of UNDEFINED cases. */
77*16d86563SAlexander Pyhalov 		if (buf[i] == '#' || buf[i] == '\0') {
78*16d86563SAlexander Pyhalov 			tbl[k].size = ICV_TYPE_ILLEGAL_CHAR;
79*16d86563SAlexander Pyhalov 			tbl[k].u8 = 0;
80*16d86563SAlexander Pyhalov 			continue;
81*16d86563SAlexander Pyhalov 		}
82*16d86563SAlexander Pyhalov 
83*16d86563SAlexander Pyhalov 		for (j = 0; !isspace(buf[i]); i++, j++)
84*16d86563SAlexander Pyhalov 			num[j] = buf[i];
85*16d86563SAlexander Pyhalov 		num[j] = '\0';
86*16d86563SAlexander Pyhalov 
87*16d86563SAlexander Pyhalov 		l = strtol(num, (char **)NULL, 0);
88*16d86563SAlexander Pyhalov 
89*16d86563SAlexander Pyhalov 		tbl[k].u8 = l;
90*16d86563SAlexander Pyhalov 		if (l < 0x80)
91*16d86563SAlexander Pyhalov 			tbl[k].size = 1;
92*16d86563SAlexander Pyhalov 		else if (l < 0x800)
93*16d86563SAlexander Pyhalov 			tbl[k].size = 2;
94*16d86563SAlexander Pyhalov 		else if (l < 0x10000)
95*16d86563SAlexander Pyhalov 			tbl[k].size = 3;
96*16d86563SAlexander Pyhalov 		else if (l < 0x200000)
97*16d86563SAlexander Pyhalov 			tbl[k].size = 4;
98*16d86563SAlexander Pyhalov 		else if (l < 0x4000000)
99*16d86563SAlexander Pyhalov 			tbl[k].size = 5;
100*16d86563SAlexander Pyhalov 		else
101*16d86563SAlexander Pyhalov 			tbl[k].size = 6;
102*16d86563SAlexander Pyhalov 	}
103*16d86563SAlexander Pyhalov 
104*16d86563SAlexander Pyhalov 	for (i = 0; i < 256; i++) {
105*16d86563SAlexander Pyhalov 		if (tbl[i].u8 < 0x80)
106*16d86563SAlexander Pyhalov 			l = tbl[i].u8;
107*16d86563SAlexander Pyhalov 		else if (tbl[i].u8 < 0x800) {
108*16d86563SAlexander Pyhalov 			l = 0xc080 |
109*16d86563SAlexander Pyhalov 				(((tbl[i].u8 >> 6) & 0x1f) << 8) |
110*16d86563SAlexander Pyhalov 				(tbl[i].u8 & 0x3f);
111*16d86563SAlexander Pyhalov 		} else if (tbl[i].u8 < 0x10000) {
112*16d86563SAlexander Pyhalov 			l = 0xe08080 |
113*16d86563SAlexander Pyhalov 				(((tbl[i].u8 >> 12) & 0x0f) << 16) |
114*16d86563SAlexander Pyhalov 				(((tbl[i].u8 >> 6) & 0x3f) << 8) |
115*16d86563SAlexander Pyhalov 				(tbl[i].u8 & 0x3f);
116*16d86563SAlexander Pyhalov 		} else if (tbl[i].u8 < 0x200000) {
117*16d86563SAlexander Pyhalov 			l = 0xf0808080 |
118*16d86563SAlexander Pyhalov 				(((tbl[i].u8 >> 18) & 0x07) << 24) |
119*16d86563SAlexander Pyhalov 				(((tbl[i].u8 >> 12) & 0x3f) << 16) |
120*16d86563SAlexander Pyhalov 				(((tbl[i].u8 >> 6) & 0x3f) << 8) |
121*16d86563SAlexander Pyhalov 				(tbl[i].u8 & 0x3f);
122*16d86563SAlexander Pyhalov 		} /* We only support characters in range of UTF-16
123*16d86563SAlexander Pyhalov 		else if (tbl[i].u8 < 0x4000000) {
124*16d86563SAlexander Pyhalov 			l = 0xf880808080 |
125*16d86563SAlexander Pyhalov 				(((tbl[i].u8 >> 24) & 0x03) << 32) |
126*16d86563SAlexander Pyhalov 				(((tbl[i].u8 >> 18) & 0x3f) << 24) |
127*16d86563SAlexander Pyhalov 				(((tbl[i].u8 >> 12) & 0x3f) << 16) |
128*16d86563SAlexander Pyhalov 				(((tbl[i].u8 >> 6) & 0x3f) << 8) |
129*16d86563SAlexander Pyhalov 				(tbl[i].u8 & 0x3f);
130*16d86563SAlexander Pyhalov 		} else {
131*16d86563SAlexander Pyhalov 			l = 0xfc8080808080 |
132*16d86563SAlexander Pyhalov 				(((tbl[i].u8 >> 30) & 0x01) << 40) |
133*16d86563SAlexander Pyhalov 				(((tbl[i].u8 >> 24) & 0x3f) << 32) |
134*16d86563SAlexander Pyhalov 				(((tbl[i].u8 >> 18) & 0x3f) << 24) |
135*16d86563SAlexander Pyhalov 				(((tbl[i].u8 >> 12) & 0x3f) << 16) |
136*16d86563SAlexander Pyhalov 				(((tbl[i].u8 >> 6) & 0x3f) << 8) |
137*16d86563SAlexander Pyhalov 				(tbl[i].u8 & 0x3f);
138*16d86563SAlexander Pyhalov 		}
139*16d86563SAlexander Pyhalov 		*/
140*16d86563SAlexander Pyhalov 
141*16d86563SAlexander Pyhalov 		printf("/* 0x%02X */  {  0x%08X, %-3d},\n", i, l, tbl[i].size);
142*16d86563SAlexander Pyhalov 	}
143*16d86563SAlexander Pyhalov }
144