+++ /dev/null
-/*\r
- * _codecs_kr.c: Codecs collection for Korean encodings\r
- *\r
- * Written by Hye-Shik Chang <perky@FreeBSD.org>\r
- */\r
-\r
-#include "cjkcodecs.h"\r
-#include "mappings_kr.h"\r
-\r
-/*\r
- * EUC-KR codec\r
- */\r
-\r
-#define EUCKR_JAMO_FIRSTBYTE 0xA4\r
-#define EUCKR_JAMO_FILLER 0xD4\r
-\r
-static const unsigned char u2cgk_choseong[19] = {\r
- 0xa1, 0xa2, 0xa4, 0xa7, 0xa8, 0xa9, 0xb1, 0xb2,\r
- 0xb3, 0xb5, 0xb6, 0xb7, 0xb8, 0xb9, 0xba, 0xbb,\r
- 0xbc, 0xbd, 0xbe\r
-};\r
-static const unsigned char u2cgk_jungseong[21] = {\r
- 0xbf, 0xc0, 0xc1, 0xc2, 0xc3, 0xc4, 0xc5, 0xc6,\r
- 0xc7, 0xc8, 0xc9, 0xca, 0xcb, 0xcc, 0xcd, 0xce,\r
- 0xcf, 0xd0, 0xd1, 0xd2, 0xd3\r
-};\r
-static const unsigned char u2cgk_jongseong[28] = {\r
- 0xd4, 0xa1, 0xa2, 0xa3, 0xa4, 0xa5, 0xa6, 0xa7,\r
- 0xa9, 0xaa, 0xab, 0xac, 0xad, 0xae, 0xaf, 0xb0,\r
- 0xb1, 0xb2, 0xb4, 0xb5, 0xb6, 0xb7, 0xb8, 0xba,\r
- 0xbb, 0xbc, 0xbd, 0xbe\r
-};\r
-\r
-ENCODER(euc_kr)\r
-{\r
- while (inleft > 0) {\r
- Py_UNICODE c = IN1;\r
- DBCHAR code;\r
-\r
- if (c < 0x80) {\r
- WRITE1((unsigned char)c)\r
- NEXT(1, 1)\r
- continue;\r
- }\r
- UCS4INVALID(c)\r
-\r
- REQUIRE_OUTBUF(2)\r
- TRYMAP_ENC(cp949, code, c);\r
- else return 1;\r
-\r
- if ((code & 0x8000) == 0) {\r
- /* KS X 1001 coded character */\r
- OUT1((code >> 8) | 0x80)\r
- OUT2((code & 0xFF) | 0x80)\r
- NEXT(1, 2)\r
- }\r
- else { /* Mapping is found in CP949 extension,\r
- * but we encode it in KS X 1001:1998 Annex 3,\r
- * make-up sequence for EUC-KR. */\r
-\r
- REQUIRE_OUTBUF(8)\r
-\r
- /* syllable composition precedence */\r
- OUT1(EUCKR_JAMO_FIRSTBYTE)\r
- OUT2(EUCKR_JAMO_FILLER)\r
-\r
- /* All code points in CP949 extension are in unicode\r
- * Hangul Syllable area. */\r
- assert(0xac00 <= c && c <= 0xd7a3);\r
- c -= 0xac00;\r
-\r
- OUT3(EUCKR_JAMO_FIRSTBYTE)\r
- OUT4(u2cgk_choseong[c / 588])\r
- NEXT_OUT(4)\r
-\r
- OUT1(EUCKR_JAMO_FIRSTBYTE)\r
- OUT2(u2cgk_jungseong[(c / 28) % 21])\r
- OUT3(EUCKR_JAMO_FIRSTBYTE)\r
- OUT4(u2cgk_jongseong[c % 28])\r
- NEXT(1, 4)\r
- }\r
- }\r
-\r
- return 0;\r
-}\r
-\r
-#define NONE 127\r
-\r
-static const unsigned char cgk2u_choseong[] = { /* [A1, BE] */\r
- 0, 1, NONE, 2, NONE, NONE, 3, 4,\r
- 5, NONE, NONE, NONE, NONE, NONE, NONE, NONE,\r
- 6, 7, 8, NONE, 9, 10, 11, 12,\r
- 13, 14, 15, 16, 17, 18\r
-};\r
-static const unsigned char cgk2u_jongseong[] = { /* [A1, BE] */\r
- 1, 2, 3, 4, 5, 6, 7, NONE,\r
- 8, 9, 10, 11, 12, 13, 14, 15,\r
- 16, 17, NONE, 18, 19, 20, 21, 22,\r
- NONE, 23, 24, 25, 26, 27\r
-};\r
-\r
-DECODER(euc_kr)\r
-{\r
- while (inleft > 0) {\r
- unsigned char c = IN1;\r
-\r
- REQUIRE_OUTBUF(1)\r
-\r
- if (c < 0x80) {\r
- OUT1(c)\r
- NEXT(1, 1)\r
- continue;\r
- }\r
-\r
- REQUIRE_INBUF(2)\r
-\r
- if (c == EUCKR_JAMO_FIRSTBYTE &&\r
- IN2 == EUCKR_JAMO_FILLER) {\r
- /* KS X 1001:1998 Annex 3 make-up sequence */\r
- DBCHAR cho, jung, jong;\r
-\r
- REQUIRE_INBUF(8)\r
- if ((*inbuf)[2] != EUCKR_JAMO_FIRSTBYTE ||\r
- (*inbuf)[4] != EUCKR_JAMO_FIRSTBYTE ||\r
- (*inbuf)[6] != EUCKR_JAMO_FIRSTBYTE)\r
- return 8;\r
-\r
- c = (*inbuf)[3];\r
- if (0xa1 <= c && c <= 0xbe)\r
- cho = cgk2u_choseong[c - 0xa1];\r
- else\r
- cho = NONE;\r
-\r
- c = (*inbuf)[5];\r
- jung = (0xbf <= c && c <= 0xd3) ? c - 0xbf : NONE;\r
-\r
- c = (*inbuf)[7];\r
- if (c == EUCKR_JAMO_FILLER)\r
- jong = 0;\r
- else if (0xa1 <= c && c <= 0xbe)\r
- jong = cgk2u_jongseong[c - 0xa1];\r
- else\r
- jong = NONE;\r
-\r
- if (cho == NONE || jung == NONE || jong == NONE)\r
- return 8;\r
-\r
- OUT1(0xac00 + cho*588 + jung*28 + jong);\r
- NEXT(8, 1)\r
- }\r
- else TRYMAP_DEC(ksx1001, **outbuf, c ^ 0x80, IN2 ^ 0x80) {\r
- NEXT(2, 1)\r
- }\r
- else\r
- return 2;\r
- }\r
-\r
- return 0;\r
-}\r
-#undef NONE\r
-\r
-\r
-/*\r
- * CP949 codec\r
- */\r
-\r
-ENCODER(cp949)\r
-{\r
- while (inleft > 0) {\r
- Py_UNICODE c = IN1;\r
- DBCHAR code;\r
-\r
- if (c < 0x80) {\r
- WRITE1((unsigned char)c)\r
- NEXT(1, 1)\r
- continue;\r
- }\r
- UCS4INVALID(c)\r
-\r
- REQUIRE_OUTBUF(2)\r
- TRYMAP_ENC(cp949, code, c);\r
- else return 1;\r
-\r
- OUT1((code >> 8) | 0x80)\r
- if (code & 0x8000)\r
- OUT2(code & 0xFF) /* MSB set: CP949 */\r
- else\r
- OUT2((code & 0xFF) | 0x80) /* MSB unset: ks x 1001 */\r
- NEXT(1, 2)\r
- }\r
-\r
- return 0;\r
-}\r
-\r
-DECODER(cp949)\r
-{\r
- while (inleft > 0) {\r
- unsigned char c = IN1;\r
-\r
- REQUIRE_OUTBUF(1)\r
-\r
- if (c < 0x80) {\r
- OUT1(c)\r
- NEXT(1, 1)\r
- continue;\r
- }\r
-\r
- REQUIRE_INBUF(2)\r
- TRYMAP_DEC(ksx1001, **outbuf, c ^ 0x80, IN2 ^ 0x80);\r
- else TRYMAP_DEC(cp949ext, **outbuf, c, IN2);\r
- else return 2;\r
-\r
- NEXT(2, 1)\r
- }\r
-\r
- return 0;\r
-}\r
-\r
-\r
-/*\r
- * JOHAB codec\r
- */\r
-\r
-static const unsigned char u2johabidx_choseong[32] = {\r
- 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,\r
- 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,\r
- 0x10, 0x11, 0x12, 0x13, 0x14,\r
-};\r
-static const unsigned char u2johabidx_jungseong[32] = {\r
- 0x03, 0x04, 0x05, 0x06, 0x07,\r
- 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,\r
- 0x12, 0x13, 0x14, 0x15, 0x16, 0x17,\r
- 0x1a, 0x1b, 0x1c, 0x1d,\r
-};\r
-static const unsigned char u2johabidx_jongseong[32] = {\r
- 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07,\r
- 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f,\r
- 0x10, 0x11, 0x13, 0x14, 0x15, 0x16, 0x17,\r
- 0x18, 0x19, 0x1a, 0x1b, 0x1c, 0x1d,\r
-};\r
-static const DBCHAR u2johabjamo[] = {\r
- 0x8841, 0x8c41, 0x8444, 0x9041, 0x8446, 0x8447, 0x9441,\r
- 0x9841, 0x9c41, 0x844a, 0x844b, 0x844c, 0x844d, 0x844e, 0x844f,\r
- 0x8450, 0xa041, 0xa441, 0xa841, 0x8454, 0xac41, 0xb041, 0xb441,\r
- 0xb841, 0xbc41, 0xc041, 0xc441, 0xc841, 0xcc41, 0xd041, 0x8461,\r
- 0x8481, 0x84a1, 0x84c1, 0x84e1, 0x8541, 0x8561, 0x8581, 0x85a1,\r
- 0x85c1, 0x85e1, 0x8641, 0x8661, 0x8681, 0x86a1, 0x86c1, 0x86e1,\r
- 0x8741, 0x8761, 0x8781, 0x87a1,\r
-};\r
-\r
-ENCODER(johab)\r
-{\r
- while (inleft > 0) {\r
- Py_UNICODE c = IN1;\r
- DBCHAR code;\r
-\r
- if (c < 0x80) {\r
- WRITE1((unsigned char)c)\r
- NEXT(1, 1)\r
- continue;\r
- }\r
- UCS4INVALID(c)\r
-\r
- REQUIRE_OUTBUF(2)\r
-\r
- if (c >= 0xac00 && c <= 0xd7a3) {\r
- c -= 0xac00;\r
- code = 0x8000 |\r
- (u2johabidx_choseong[c / 588] << 10) |\r
- (u2johabidx_jungseong[(c / 28) % 21] << 5) |\r
- u2johabidx_jongseong[c % 28];\r
- }\r
- else if (c >= 0x3131 && c <= 0x3163)\r
- code = u2johabjamo[c - 0x3131];\r
- else TRYMAP_ENC(cp949, code, c) {\r
- unsigned char c1, c2, t2;\r
- unsigned short t1;\r
-\r
- assert((code & 0x8000) == 0);\r
- c1 = code >> 8;\r
- c2 = code & 0xff;\r
- if (((c1 >= 0x21 && c1 <= 0x2c) ||\r
- (c1 >= 0x4a && c1 <= 0x7d)) &&\r
- (c2 >= 0x21 && c2 <= 0x7e)) {\r
- t1 = (c1 < 0x4a ? (c1 - 0x21 + 0x1b2) :\r
- (c1 - 0x21 + 0x197));\r
- t2 = ((t1 & 1) ? 0x5e : 0) + (c2 - 0x21);\r
- OUT1(t1 >> 1)\r
- OUT2(t2 < 0x4e ? t2 + 0x31 : t2 + 0x43)\r
- NEXT(1, 2)\r
- continue;\r
- }\r
- else\r
- return 1;\r
- }\r
- else\r
- return 1;\r
-\r
- OUT1(code >> 8)\r
- OUT2(code & 0xff)\r
- NEXT(1, 2)\r
- }\r
-\r
- return 0;\r
-}\r
-\r
-#define FILL 0xfd\r
-#define NONE 0xff\r
-\r
-static const unsigned char johabidx_choseong[32] = {\r
- NONE, FILL, 0x00, 0x01, 0x02, 0x03, 0x04, 0x05,\r
- 0x06, 0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d,\r
- 0x0e, 0x0f, 0x10, 0x11, 0x12, NONE, NONE, NONE,\r
- NONE, NONE, NONE, NONE, NONE, NONE, NONE, NONE,\r
-};\r
-static const unsigned char johabidx_jungseong[32] = {\r
- NONE, NONE, FILL, 0x00, 0x01, 0x02, 0x03, 0x04,\r
- NONE, NONE, 0x05, 0x06, 0x07, 0x08, 0x09, 0x0a,\r
- NONE, NONE, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x10,\r
- NONE, NONE, 0x11, 0x12, 0x13, 0x14, NONE, NONE,\r
-};\r
-static const unsigned char johabidx_jongseong[32] = {\r
- NONE, FILL, 0x01, 0x02, 0x03, 0x04, 0x05, 0x06,\r
- 0x07, 0x08, 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e,\r
- 0x0f, 0x10, NONE, 0x11, 0x12, 0x13, 0x14, 0x15,\r
- 0x16, 0x17, 0x18, 0x19, 0x1a, 0x1b, NONE, NONE,\r
-};\r
-\r
-static const unsigned char johabjamo_choseong[32] = {\r
- NONE, FILL, 0x31, 0x32, 0x34, 0x37, 0x38, 0x39,\r
- 0x41, 0x42, 0x43, 0x45, 0x46, 0x47, 0x48, 0x49,\r
- 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, NONE, NONE, NONE,\r
- NONE, NONE, NONE, NONE, NONE, NONE, NONE, NONE,\r
-};\r
-static const unsigned char johabjamo_jungseong[32] = {\r
- NONE, NONE, FILL, 0x4f, 0x50, 0x51, 0x52, 0x53,\r
- NONE, NONE, 0x54, 0x55, 0x56, 0x57, 0x58, 0x59,\r
- NONE, NONE, 0x5a, 0x5b, 0x5c, 0x5d, 0x5e, 0x5f,\r
- NONE, NONE, 0x60, 0x61, 0x62, 0x63, NONE, NONE,\r
-};\r
-static const unsigned char johabjamo_jongseong[32] = {\r
- NONE, FILL, 0x31, 0x32, 0x33, 0x34, 0x35, 0x36,\r
- 0x37, 0x39, 0x3a, 0x3b, 0x3c, 0x3d, 0x3e, 0x3f,\r
- 0x40, 0x41, NONE, 0x42, 0x44, 0x45, 0x46, 0x47,\r
- 0x48, 0x4a, 0x4b, 0x4c, 0x4d, 0x4e, NONE, NONE,\r
-};\r
-\r
-DECODER(johab)\r
-{\r
- while (inleft > 0) {\r
- unsigned char c = IN1, c2;\r
-\r
- REQUIRE_OUTBUF(1)\r
-\r
- if (c < 0x80) {\r
- OUT1(c)\r
- NEXT(1, 1)\r
- continue;\r
- }\r
-\r
- REQUIRE_INBUF(2)\r
- c2 = IN2;\r
-\r
- if (c < 0xd8) {\r
- /* johab hangul */\r
- unsigned char c_cho, c_jung, c_jong;\r
- unsigned char i_cho, i_jung, i_jong;\r
-\r
- c_cho = (c >> 2) & 0x1f;\r
- c_jung = ((c << 3) | c2 >> 5) & 0x1f;\r
- c_jong = c2 & 0x1f;\r
-\r
- i_cho = johabidx_choseong[c_cho];\r
- i_jung = johabidx_jungseong[c_jung];\r
- i_jong = johabidx_jongseong[c_jong];\r
-\r
- if (i_cho == NONE || i_jung == NONE || i_jong == NONE)\r
- return 2;\r
-\r
- /* we don't use U+1100 hangul jamo yet. */\r
- if (i_cho == FILL) {\r
- if (i_jung == FILL) {\r
- if (i_jong == FILL)\r
- OUT1(0x3000)\r
- else\r
- OUT1(0x3100 |\r
- johabjamo_jongseong[c_jong])\r
- }\r
- else {\r
- if (i_jong == FILL)\r
- OUT1(0x3100 |\r
- johabjamo_jungseong[c_jung])\r
- else\r
- return 2;\r
- }\r
- } else {\r
- if (i_jung == FILL) {\r
- if (i_jong == FILL)\r
- OUT1(0x3100 |\r
- johabjamo_choseong[c_cho])\r
- else\r
- return 2;\r
- }\r
- else\r
- OUT1(0xac00 +\r
- i_cho * 588 +\r
- i_jung * 28 +\r
- (i_jong == FILL ? 0 : i_jong))\r
- }\r
- NEXT(2, 1)\r
- } else {\r
- /* KS X 1001 except hangul jamos and syllables */\r
- if (c == 0xdf || c > 0xf9 ||\r
- c2 < 0x31 || (c2 >= 0x80 && c2 < 0x91) ||\r
- (c2 & 0x7f) == 0x7f ||\r
- (c == 0xda && (c2 >= 0xa1 && c2 <= 0xd3)))\r
- return 2;\r
- else {\r
- unsigned char t1, t2;\r
-\r
- t1 = (c < 0xe0 ? 2 * (c - 0xd9) :\r
- 2 * c - 0x197);\r
- t2 = (c2 < 0x91 ? c2 - 0x31 : c2 - 0x43);\r
- t1 = t1 + (t2 < 0x5e ? 0 : 1) + 0x21;\r
- t2 = (t2 < 0x5e ? t2 : t2 - 0x5e) + 0x21;\r
-\r
- TRYMAP_DEC(ksx1001, **outbuf, t1, t2);\r
- else return 2;\r
- NEXT(2, 1)\r
- }\r
- }\r
- }\r
-\r
- return 0;\r
-}\r
-#undef NONE\r
-#undef FILL\r
-\r
-\r
-BEGIN_MAPPINGS_LIST\r
- MAPPING_DECONLY(ksx1001)\r
- MAPPING_ENCONLY(cp949)\r
- MAPPING_DECONLY(cp949ext)\r
-END_MAPPINGS_LIST\r
-\r
-BEGIN_CODECS_LIST\r
- CODEC_STATELESS(euc_kr)\r
- CODEC_STATELESS(cp949)\r
- CODEC_STATELESS(johab)\r
-END_CODECS_LIST\r
-\r
-I_AM_A_MODULE_FOR(kr)\r
projects / mirror_edk2.git / blobdiff