big5_emacs.h revision 1ab64890 
1/* $XFree86$ */
2
3/*
4 * BIG5-0 and BIG5-1
5 */
6
7/*
8   BIG5 with its 13494 characters doesn't fit in a single 94x94 or 96x96
9   block. Therefore Emacs/Mule developers, in a typically Japanese way of
10   thinking, have developed an alternative encoding of BIG5 in two 94x94
11   planes, very similar to the SHIFT_JIS encoding for JISX0208.
12
13   Conversion between BIG5 codes (s1,s2) and BIG5-0 codes (c1,c2):
14   Example. (s1,s2) = 0xA140, (c1,c2) = 0x2121.
15   0xA1 <= s1 <= 0xC7, 0x40 <= s2 <= 0x7E || 0xA1 <= s2 <= 0xFE,
16   0x21 <= c1 <= 0x62, 0x21 <= c2 <= 0x7E.
17   Invariant:
18     157*(s1-0xA1) + (s2 < 0x80 ? s2-0x40 : s2-0x62)
19     = 94*(c1-0x21)+(c2-0x21)
20   Conversion (s1,s2) -> (c1,c2):
21     t := 157*(s1-0xA1) + (s2 < 0x80 ? s2-0x40 : s2-0x62)
22     c1 := (t div 94) + 0x21
23     c2 := (t mod 94) + 0x21
24   Conversion (c1,c2) -> (s1,s2):
25     t := 94*(c1-0x21)+(c2-0x21)
26     t2 := t mod 157
27     s1 := (t div 157) + 0xA1
28     s2 := (t2 < 0x3F ? t2+0x40 : t2+0x62)
29
30   Conversion between BIG5 codes (s1,s2) and BIG5-1 codes (c1,c2):
31   Example. (s1,s2) = 0xC940, (c1,c2) = 0x2121.
32   0xC9 <= s1 <= 0xF9, 0x40 <= s2 <= 0x7E || 0xA1 <= s2 <= 0xFE,
33   0x21 <= c1 <= 0x72, 0x21 <= c2 <= 0x7E.
34   Invariant:
35     157*(s1-0xC9) + (s2 < 0x80 ? s2-0x40 : s2-0x62)
36     = 94*(c1-0x21)+(c2-0x21)
37   Conversion (s1,s2) -> (c1,c2):
38     t := 157*(s1-0xC9) + (s2 < 0x80 ? s2-0x40 : s2-0x62)
39     c1 := (t div 94) + 0x21
40     c2 := (t mod 94) + 0x21
41   Conversion (c1,c2) -> (s1,s2):
42     t := 94*(c1-0x21)+(c2-0x21)
43     t2 := t mod 157
44     s1 := (t div 157) + 0xC9
45     s2 := (t2 < 0x3F ? t2+0x40 : t2+0x62)
46 */
47
48static int
49big5_0_mbtowc (conv_t conv, ucs4_t *pwc, const unsigned char *s, int n)
50{
51  unsigned char c1 = s[0];
52  if (c1 >= 0x21 && c1 <= 0x62) {
53    if (n >= 2) {
54      unsigned char c2 = s[1];
55      if (c2 >= 0x21 && c2 <= 0x7e) {
56        unsigned int i = 94 * (c1 - 0x21) + (c2 - 0x21);
57        if (0) {
58          /* Unoptimized. */
59          unsigned char buf[2];
60          buf[0] = (i / 157) + 0xa1;
61          i = i % 157;
62          buf[1] = i + (i < 0x3f ? 0x40 : 0x62);
63          return big5_mbtowc(conv,pwc,buf,2);
64        } else {
65          /* Inline the implementation of big5_mbtowc. */
66          if (i < 6121) {
67            unsigned short wc = big5_2uni_pagea1[i];
68            if (wc != 0xfffd) {
69              *pwc = (ucs4_t) wc;
70              return 2;
71            }
72          }
73        }
74      }
75      return RET_ILSEQ;
76    }
77    return RET_TOOFEW(0);
78  }
79  return RET_ILSEQ;
80}
81
82static int
83big5_1_mbtowc (conv_t conv, ucs4_t *pwc, const unsigned char *s, int n)
84{
85  unsigned char c1 = s[0];
86  if (c1 >= 0x21 && c1 <= 0x72) {
87    if (n >= 2) {
88      unsigned char c2 = s[1];
89      if (c2 >= 0x21 && c2 <= 0x7e) {
90        unsigned int i = 94 * (c1 - 0x21) + (c2 - 0x21);
91        if (0) {
92          /* Unoptimized. */
93          unsigned char buf[2];
94          buf[0] = (i / 157) + 0xc9;
95          i = i % 157;
96          buf[1] = i + (i < 0x3f ? 0x40 : 0x62);
97          return big5_mbtowc(conv,pwc,buf,2);
98        } else {
99          /* Inline the implementation of big5_mbtowc. */
100          if (i < 7652) {
101            unsigned short wc = big5_2uni_pagec9[i];
102            if (wc != 0xfffd) {
103              *pwc = (ucs4_t) wc;
104              return 2;
105            }
106          }
107        }
108      }
109      return RET_ILSEQ;
110    }
111    return RET_TOOFEW(0);
112  }
113  return RET_ILSEQ;
114}
115
116static int
117big5_0_wctomb (conv_t conv, unsigned char *r, ucs4_t wc, int n)
118{
119  if (n >= 2) {
120    unsigned char buf[2];
121    int ret = big5_wctomb(conv,buf,wc,2);
122    if (ret != RET_ILSEQ) {
123      unsigned char s1, s2;
124      if (ret != 2) abort();
125      s1 = buf[0];
126      s2 = buf[1];
127      if (!(s1 >= 0xa1)) abort();
128      if (!((s2 >= 0x40 && s2 <= 0x7e) || (s2 >= 0xa1 && s2 <= 0xfe))) abort();
129      if (s1 < 0xc9) {
130        unsigned int t = 157 * (s1 - 0xa1) + s2 - (s2 < 0x80 ? 0x40 : 0x62);
131        r[0] = (t / 94) + 0x21;
132        r[1] = (t % 94) + 0x21;
133        return 2;
134      }
135    }
136    return RET_ILSEQ;
137  }
138  return RET_TOOSMALL;
139}
140
141static int
142big5_1_wctomb (conv_t conv, unsigned char *r, ucs4_t wc, int n)
143{
144  if (n >= 2) {
145    unsigned char buf[2];
146    int ret = big5_wctomb(conv,buf,wc,2);
147    if (ret != RET_ILSEQ) {
148      unsigned char s1, s2;
149      if (ret != 2) abort();
150      s1 = buf[0];
151      s2 = buf[1];
152      if (!(s1 <= 0xf9)) abort();
153      if (!((s2 >= 0x40 && s2 <= 0x7e) || (s2 >= 0xa1 && s2 <= 0xfe))) abort();
154      if (s1 >= 0xc9) {
155        unsigned int t = 157 * (s1 - 0xc9) + s2 - (s2 < 0x80 ? 0x40 : 0x62);
156        r[0] = (t / 94) + 0x21;
157        r[1] = (t % 94) + 0x21;
158        return 2;
159      }
160    }
161    return RET_ILSEQ;
162  }
163  return RET_TOOSMALL;
164}
165