1/*
2 * Copyright (C) 1999-2002, 2005 Free Software Foundation, Inc.
3 * This file is part of the GNU LIBICONV Library.
4 *
5 * The GNU LIBICONV Library is free software; you can redistribute it
6 * and/or modify it under the terms of the GNU Library General Public
7 * License as published by the Free Software Foundation; either version 2
8 * of the License, or (at your option) any later version.
9 *
10 * The GNU LIBICONV Library is distributed in the hope that it will be
11 * useful, but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
13 * Library General Public License for more details.
14 *
15 * You should have received a copy of the GNU Library General Public
16 * License along with the GNU LIBICONV Library; see the file COPYING.LIB.
17 * If not, write to the Free Software Foundation, Inc., 51 Franklin Street,
18 * Fifth Floor, Boston, MA 02110-1301, USA.
19 */
20
21/*
22 * CP932
23 */
24
25/*
26 * Microsoft CP932 is a slightly extended version of SHIFT_JIS.
27 * The differences between the EASTASIA/JIS/SHIFTJIS.TXT and the
28 * VENDORS/MICSFT/WINDOWS/CP932.TXT tables found on ftp.unicode.org are
29 * as follows:
30 *
31 * 1. CP932 uses ASCII, not JISX0201 Roman.
32 *
33 * 2. Some characters in the JISX0208 range are defined differently:
34 *
35 *     code   SHIFTJIS.TXT                   CP932.TXT
36 *    0x815F  0x005C # REVERSE SOLIDUS       0xFF3C # FULLWIDTH REVERSE SOLIDUS
37 *    0x8160  0x301C # WAVE DASH             0xFF5E # FULLWIDTH TILDE
38 *    0x8161  0x2016 # DOUBLE VERTICAL LINE  0x2225 # PARALLEL TO
39 *    0x817C  0x2212 # MINUS SIGN            0xFF0D # FULLWIDTH HYPHEN-MINUS
40 *    0x8191  0x00A2 # CENT SIGN             0xFFE0 # FULLWIDTH CENT SIGN
41 *    0x8192  0x00A3 # POUND SIGN            0xFFE1 # FULLWIDTH POUND SIGN
42 *    0x81CA  0x00AC # NOT SIGN              0xFFE2 # FULLWIDTH NOT SIGN
43 *
44 *    We don't implement the latter 6 of these changes, only the first one.
45 *    SHIFTJIS.TXT makes more sense. However, as a compromise with user
46 *    expectation, we implement the middle 5 of these changes in the
47 *    Unicode to CP932 direction. We don't implement the last one at all,
48 *    because it would collide with the mapping of 0xFA54.
49 *
50 * 3. A few new rows. See cp932ext.h.
51 *
52 * Many variants of CP932 (in GNU libc, JDK, OSF/1, Windows-2000, ICU) also
53 * add:
54 *
55 * 4. Private area mappings:
56 *
57 *              code                 Unicode
58 *    0x{F0..F9}{40..7E,80..FC}  U+E000..U+E757
59 *
60 * We add them too because, although there are backward compatibility problems
61 * when a character from a private area is moved to an official Unicode code
62 * point, they are useful for some people in practice.
63 */
64
65#include "cp932ext.h"
66
67/*
68   Conversion between SJIS codes (s1,s2) and JISX0208 codes (c1,c2):
69   Example. (s1,s2) = 0x8140, (c1,c2) = 0x2121.
70   0x81 <= s1 <= 0x9F || 0xE0 <= s1 <= 0xEA,
71   0x40 <= s2 <= 0x7E || 0x80 <= s2 <= 0xFC,
72   0x21 <= c1 <= 0x74, 0x21 <= c2 <= 0x7E.
73   Invariant:
74     94*2*(s1 < 0xE0 ? s1-0x81 : s1-0xC1) + (s2 < 0x80 ? s2-0x40 : s2-0x41)
75     = 94*(c1-0x21)+(c2-0x21)
76   Conversion (s1,s2) -> (c1,c2):
77     t1 := (s1 < 0xE0 ? s1-0x81 : s1-0xC1)
78     t2 := (s2 < 0x80 ? s2-0x40 : s2-0x41)
79     c1 := 2*t1 + (t2 < 0x5E ? 0 : 1) + 0x21
80     c2 := (t2 < 0x5E ? t2 : t2-0x5E) + 0x21
81   Conversion (c1,c2) -> (s1,s2):
82     t1 := (c1 - 0x21) >> 1
83     t2 := ((c1 - 0x21) & 1) * 0x5E + (c2 - 0x21)
84     s1 := (t1 < 0x1F ? t1+0x81 : t1+0xC1)
85     s2 := (t2 < 0x3F ? t2+0x40 : t2+0x41)
86 */
87
88static int
89cp932_mbtowc (conv_t conv, ucs4_t *pwc, const unsigned char *s, int n)
90{
91  unsigned char c = *s;
92  if (c < 0x80)
93    return ascii_mbtowc(conv,pwc,s,n);
94  else if (c >= 0xa1 && c <= 0xdf)
95    return jisx0201_mbtowc(conv,pwc,s,n);
96  else {
97    unsigned char s1, s2;
98    s1 = c;
99    if ((s1 >= 0x81 && s1 <= 0x9f && s1 != 0x87) || (s1 >= 0xe0 && s1 <= 0xea)) {
100      if (n < 2)
101        return RET_TOOFEW(0);
102      s2 = s[1];
103      if ((s2 >= 0x40 && s2 <= 0x7e) || (s2 >= 0x80 && s2 <= 0xfc)) {
104        unsigned char t1 = (s1 < 0xe0 ? s1-0x81 : s1-0xc1);
105        unsigned char t2 = (s2 < 0x80 ? s2-0x40 : s2-0x41);
106        unsigned char buf[2];
107        buf[0] = 2*t1 + (t2 < 0x5e ? 0 : 1) + 0x21;
108        buf[1] = (t2 < 0x5e ? t2 : t2-0x5e) + 0x21;
109        return jisx0208_mbtowc(conv,pwc,buf,2);
110      }
111    } else if ((s1 == 0x87) || (s1 >= 0xed && s1 <= 0xee) || (s1 >= 0xfa)) {
112      if (n < 2)
113        return RET_TOOFEW(0);
114      return cp932ext_mbtowc(conv,pwc,s,2);
115    } else if (s1 >= 0xf0 && s1 <= 0xf9) {
116      /* User-defined range. See
117       * Ken Lunde's "CJKV Information Processing", table 4-66, p. 206. */
118      if (n < 2)
119        return RET_TOOFEW(0);
120      s2 = s[1];
121      if ((s2 >= 0x40 && s2 <= 0x7e) || (s2 >= 0x80 && s2 <= 0xfc)) {
122        *pwc = 0xe000 + 188*(s1 - 0xf0) + (s2 < 0x80 ? s2-0x40 : s2-0x41);
123        return 2;
124      }
125    }
126    return RET_ILSEQ;
127  }
128}
129
130static int
131cp932_wctomb (conv_t conv, unsigned char *r, ucs4_t wc, int n)
132{
133  unsigned char buf[2];
134  int ret;
135
136  /* Try ASCII. */
137  ret = ascii_wctomb(conv,buf,wc,1);
138  if (ret != RET_ILUNI) {
139    unsigned char c;
140    if (ret != 1) abort();
141    c = buf[0];
142    if (c < 0x80) {
143      r[0] = c;
144      return 1;
145    }
146  }
147
148  /* Try JIS X 0201-1976 Katakana. */
149  ret = jisx0201_wctomb(conv,buf,wc,1);
150  if (ret != RET_ILUNI) {
151    unsigned char c;
152    if (ret != 1) abort();
153    c = buf[0];
154    if (c >= 0xa1 && c <= 0xdf) {
155      r[0] = c;
156      return 1;
157    }
158  }
159
160  /* Try JIS X 0208-1990. */
161  ret = jisx0208_wctomb(conv,buf,wc,2);
162  if (ret != RET_ILUNI) {
163    unsigned char c1, c2;
164    if (ret != 2) abort();
165    if (n < 2)
166      return RET_TOOSMALL;
167    c1 = buf[0];
168    c2 = buf[1];
169    if ((c1 >= 0x21 && c1 <= 0x74) && (c2 >= 0x21 && c2 <= 0x7e)) {
170      unsigned char t1 = (c1 - 0x21) >> 1;
171      unsigned char t2 = (((c1 - 0x21) & 1) ? 0x5e : 0) + (c2 - 0x21);
172      r[0] = (t1 < 0x1f ? t1+0x81 : t1+0xc1);
173      r[1] = (t2 < 0x3f ? t2+0x40 : t2+0x41);
174      return 2;
175    }
176  }
177
178  /* Try CP932 extensions. */
179  ret = cp932ext_wctomb(conv,buf,wc,2);
180  if (ret != RET_ILUNI) {
181    if (ret != 2) abort();
182    if (n < 2)
183      return RET_TOOSMALL;
184    r[0] = buf[0];
185    r[1] = buf[1];
186    return 2;
187  }
188
189  /* User-defined range. See
190   * Ken Lunde's "CJKV Information Processing", table 4-66, p. 206. */
191  if (wc >= 0xe000 && wc < 0xe758) {
192    unsigned char c1, c2;
193    if (n < 2)
194      return RET_TOOSMALL;
195    c1 = (unsigned int) (wc - 0xe000) / 188;
196    c2 = (unsigned int) (wc - 0xe000) % 188;
197    r[0] = c1+0xf0;
198    r[1] = (c2 < 0x3f ? c2+0x40 : c2+0x41);
199    return 2;
200  }
201
202  /* Irreversible mappings.  */
203  if (wc == 0xff5e) {
204    if (n < 2)
205      return RET_TOOSMALL;
206    r[0] = 0x81;
207    r[1] = 0x60;
208    return 2;
209  }
210  if (wc == 0x2225) {
211    if (n < 2)
212      return RET_TOOSMALL;
213    r[0] = 0x81;
214    r[1] = 0x61;
215    return 2;
216  }
217  if (wc == 0xff0d) {
218    if (n < 2)
219      return RET_TOOSMALL;
220    r[0] = 0x81;
221    r[1] = 0x7c;
222    return 2;
223  }
224  if (wc == 0xffe0) {
225    if (n < 2)
226      return RET_TOOSMALL;
227    r[0] = 0x81;
228    r[1] = 0x91;
229    return 2;
230  }
231  if (wc == 0xffe1) {
232    if (n < 2)
233      return RET_TOOSMALL;
234    r[0] = 0x81;
235    r[1] = 0x92;
236    return 2;
237  }
238
239  return RET_ILUNI;
240}
241