1 /*
2 * _codecs_hk.c: Codecs collection for encodings from Hong Kong
3 *
4 * Written by Hye-Shik Chang <perky@FreeBSD.org>
5 */
6
7 #define USING_IMPORTED_MAPS
8
9 #include "cjkcodecs.h"
10 #include "mappings_hk.h"
11
12 /*
13 * BIG5HKSCS codec
14 */
15
16 static const encode_map *big5_encmap = NULL;
17 static const decode_map *big5_decmap = NULL;
18
CODEC_INIT(big5hkscs)19 CODEC_INIT(big5hkscs)
20 {
21 static int initialized = 0;
22
23 if (!initialized && IMPORT_MAP(tw, big5, &big5_encmap, &big5_decmap))
24 return -1;
25 initialized = 1;
26 return 0;
27 }
28
29 /*
30 * There are four possible pair unicode -> big5hkscs maps as in HKSCS 2004:
31 * U+00CA U+0304 -> 8862 (U+00CA alone is mapped to 8866)
32 * U+00CA U+030C -> 8864
33 * U+00EA U+0304 -> 88a3 (U+00EA alone is mapped to 88a7)
34 * U+00EA U+030C -> 88a5
35 * These are handled by not mapping tables but a hand-written code.
36 */
37 static const DBCHAR big5hkscs_pairenc_table[4] = {0x8862, 0x8864, 0x88a3, 0x88a5};
38
ENCODER(big5hkscs)39 ENCODER(big5hkscs)
40 {
41 while (inleft > 0) {
42 ucs4_t c = **inbuf;
43 DBCHAR code;
44 Py_ssize_t insize;
45
46 if (c < 0x80) {
47 REQUIRE_OUTBUF(1)
48 **outbuf = (unsigned char)c;
49 NEXT(1, 1)
50 continue;
51 }
52
53 DECODE_SURROGATE(c)
54 insize = GET_INSIZE(c);
55
56 REQUIRE_OUTBUF(2)
57
58 if (c < 0x10000) {
59 TRYMAP_ENC(big5hkscs_bmp, code, c) {
60 if (code == MULTIC) {
61 if (inleft >= 2 &&
62 ((c & 0xffdf) == 0x00ca) &&
63 (((*inbuf)[1] & 0xfff7) == 0x0304)) {
64 code = big5hkscs_pairenc_table[
65 ((c >> 4) |
66 ((*inbuf)[1] >> 3)) & 3];
67 insize = 2;
68 }
69 else if (inleft < 2 &&
70 !(flags & MBENC_FLUSH))
71 return MBERR_TOOFEW;
72 else {
73 if (c == 0xca)
74 code = 0x8866;
75 else /* c == 0xea */
76 code = 0x88a7;
77 }
78 }
79 }
80 else TRYMAP_ENC(big5, code, c);
81 else return 1;
82 }
83 else if (c < 0x20000)
84 return insize;
85 else if (c < 0x30000) {
86 TRYMAP_ENC(big5hkscs_nonbmp, code, c & 0xffff);
87 else return insize;
88 }
89 else
90 return insize;
91
92 OUT1(code >> 8)
93 OUT2(code & 0xFF)
94 NEXT(insize, 2)
95 }
96
97 return 0;
98 }
99
100 #define BH2S(c1, c2) (((c1) - 0x87) * (0xfe - 0x40 + 1) + ((c2) - 0x40))
101
DECODER(big5hkscs)102 DECODER(big5hkscs)
103 {
104 while (inleft > 0) {
105 unsigned char c = IN1;
106 ucs4_t decoded;
107
108 REQUIRE_OUTBUF(1)
109
110 if (c < 0x80) {
111 OUT1(c)
112 NEXT(1, 1)
113 continue;
114 }
115
116 REQUIRE_INBUF(2)
117
118 if (0xc6 > c || c > 0xc8 || (c < 0xc7 && IN2 < 0xa1)) {
119 TRYMAP_DEC(big5, **outbuf, c, IN2) {
120 NEXT(2, 1)
121 continue;
122 }
123 }
124
125 TRYMAP_DEC(big5hkscs, decoded, c, IN2)
126 {
127 int s = BH2S(c, IN2);
128 const unsigned char *hintbase;
129
130 assert(0x87 <= c && c <= 0xfe);
131 assert(0x40 <= IN2 && IN2 <= 0xfe);
132
133 if (BH2S(0x87, 0x40) <= s && s <= BH2S(0xa0, 0xfe)) {
134 hintbase = big5hkscs_phint_0;
135 s -= BH2S(0x87, 0x40);
136 }
137 else if (BH2S(0xc6,0xa1) <= s && s <= BH2S(0xc8,0xfe)){
138 hintbase = big5hkscs_phint_12130;
139 s -= BH2S(0xc6, 0xa1);
140 }
141 else if (BH2S(0xf9,0xd6) <= s && s <= BH2S(0xfe,0xfe)){
142 hintbase = big5hkscs_phint_21924;
143 s -= BH2S(0xf9, 0xd6);
144 }
145 else
146 return MBERR_INTERNAL;
147
148 if (hintbase[s >> 3] & (1 << (s & 7))) {
149 WRITEUCS4(decoded | 0x20000)
150 NEXT_IN(2)
151 }
152 else {
153 OUT1(decoded)
154 NEXT(2, 1)
155 }
156 continue;
157 }
158
159 switch ((c << 8) | IN2) {
160 case 0x8862: WRITE2(0x00ca, 0x0304); break;
161 case 0x8864: WRITE2(0x00ca, 0x030c); break;
162 case 0x88a3: WRITE2(0x00ea, 0x0304); break;
163 case 0x88a5: WRITE2(0x00ea, 0x030c); break;
164 default: return 2;
165 }
166
167 NEXT(2, 2) /* all decoded code points are pairs, above. */
168 }
169
170 return 0;
171 }
172
173
174 BEGIN_MAPPINGS_LIST
175 MAPPING_DECONLY(big5hkscs)
176 MAPPING_ENCONLY(big5hkscs_bmp)
177 MAPPING_ENCONLY(big5hkscs_nonbmp)
178 END_MAPPINGS_LIST
179
180 BEGIN_CODECS_LIST
181 CODEC_STATELESS_WINIT(big5hkscs)
182 END_CODECS_LIST
183
184 I_AM_A_MODULE_FOR(hk)
185