Modules/cjkcodecs/_codecs_hk.c

   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
  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
  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
 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                         goto hkscsdec;
 120
 121                 TRYMAP_DEC(big5, **outbuf, c, IN2) {
 122                         NEXT(2, 1)
 123                 }
 124                 else
 125 hkscsdec:       TRYMAP_DEC(big5hkscs, decoded, c, IN2) {
 126                         int s = BH2S(c, IN2);
 127                         const unsigned char *hintbase;
 128
 129                         assert(0x87 <= c && c <= 0xfe);
 130                         assert(0x40 <= IN2 && IN2 <= 0xfe);
 131
 132                         if (BH2S(0x87, 0x40) <= s && s <= BH2S(0xa0, 0xfe)) {
 133                                 hintbase = big5hkscs_phint_0;
 134                                 s -= BH2S(0x87, 0x40);
 135                         }
 136                         else if (BH2S(0xc6,0xa1) <= s && s <= BH2S(0xc8,0xfe)){
 137                                 hintbase = big5hkscs_phint_12130;
 138                                 s -= BH2S(0xc6, 0xa1);
 139                         }
 140                         else if (BH2S(0xf9,0xd6) <= s && s <= BH2S(0xfe,0xfe)){
 141                                 hintbase = big5hkscs_phint_21924;
 142                                 s -= BH2S(0xf9, 0xd6);
 143                         }
 144                         else
 145                                 return MBERR_INTERNAL;
 146
 147                         if (hintbase[s >> 3] & (1 << (s & 7))) {
 148                                 WRITEUCS4(decoded | 0x20000)
 149                                 NEXT_IN(2)
 150                         }
 151                         else {
 152                                 OUT1(decoded)
 153                                 NEXT(2, 1)
 154                         }
 155                 }
 156                 else {
 157                         switch ((c << 8) | IN2) {
 158                         case 0x8862: WRITE2(0x00ca, 0x0304); break;
 159                         case 0x8864: WRITE2(0x00ca, 0x030c); break;
 160                         case 0x88a3: WRITE2(0x00ea, 0x0304); break;
 161                         case 0x88a5: WRITE2(0x00ea, 0x030c); break;
 162                         default: return 2;
 163                         }
 164
 165                         NEXT(2, 2) /* all decoded codepoints are pairs, above. */
 166                 }
 167         }
 168
 169         return 0;
 170 }
 171
 172
 173 BEGIN_MAPPINGS_LIST
 174   MAPPING_DECONLY(big5hkscs)
 175   MAPPING_ENCONLY(big5hkscs_bmp)
 176   MAPPING_ENCONLY(big5hkscs_nonbmp)
 177 END_MAPPINGS_LIST
 178
 179 BEGIN_CODECS_LIST
 180   CODEC_STATELESS_WINIT(big5hkscs)
 181 END_CODECS_LIST
 182
 183 I_AM_A_MODULE_FOR(hk)