1 /*-------------------------------------------------------------------------
3 * EUC_JIS_2004, SHIFT_JIS_2004
5 * Copyright (c) 2007-2009, PostgreSQL Global Development Group
8 * $PostgreSQL: pgsql/src/backend/utils/mb/conversion_procs/euc_jis_2004_and_shift_jis_2004/euc_jis_2004_and_shift_jis_2004.c,v 1.6 2009/01/29 19:23:39 tgl Exp $
10 *-------------------------------------------------------------------------
15 #include "mb/pg_wchar.h"
19 PG_FUNCTION_INFO_V1(euc_jis_2004_to_shift_jis_2004);
20 PG_FUNCTION_INFO_V1(shift_jis_2004_to_euc_jis_2004);
22 extern Datum euc_jis_2004_to_shift_jis_2004(PG_FUNCTION_ARGS);
23 extern Datum shift_jis_2004_to_euc_jis_2004(PG_FUNCTION_ARGS);
25 static void euc_jis_20042shift_jis_2004(const unsigned char *euc, unsigned char *p, int len);
26 static void shift_jis_20042euc_jis_2004(const unsigned char *sjis, unsigned char *p, int len);
30 * INTEGER, -- source encoding id
31 * INTEGER, -- destination encoding id
32 * CSTRING, -- source string (null terminated C string)
33 * CSTRING, -- destination string (null terminated C string)
34 * INTEGER -- source string length
40 euc_jis_2004_to_shift_jis_2004(PG_FUNCTION_ARGS)
42 unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
43 unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
44 int len = PG_GETARG_INT32(4);
46 CHECK_ENCODING_CONVERSION_ARGS(PG_EUC_JIS_2004, PG_SHIFT_JIS_2004);
48 euc_jis_20042shift_jis_2004(src, dest, len);
54 shift_jis_2004_to_euc_jis_2004(PG_FUNCTION_ARGS)
56 unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
57 unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
58 int len = PG_GETARG_INT32(4);
60 CHECK_ENCODING_CONVERSION_ARGS(PG_SHIFT_JIS_2004, PG_EUC_JIS_2004);
62 shift_jis_20042euc_jis_2004(src, dest, len);
68 * EUC_JIS_2004 -> SHIFT_JIS_2004
71 euc_jis_20042shift_jis_2004(const unsigned char *euc, unsigned char *p, int len)
81 if (!IS_HIGHBIT_SET(c1))
85 report_invalid_encoding(PG_EUC_JIS_2004,
86 (const char *) euc, len);
93 l = pg_encoding_verifymb(PG_EUC_JIS_2004, (const char *) euc, len);
96 report_invalid_encoding(PG_EUC_JIS_2004,
97 (const char *) euc, len);
99 if (c1 == SS2 && l == 2) /* JIS X 0201 kana? */
103 else if (c1 == SS3 && l == 3) /* JIS X 0213 plane 2? */
119 *p++ = ((ku + 0x1df) >> 1) - (ku >> 3) * 3;
122 if (ku >= 78 && ku <= 94)
124 *p++ = (ku + 0x19b) >> 1;
127 report_invalid_encoding(PG_EUC_JIS_2004,
128 (const char *) euc, len);
133 if (ten >= 1 && ten <= 63)
135 else if (ten >= 64 && ten <= 94)
138 report_invalid_encoding(PG_EUC_JIS_2004,
139 (const char *) euc, len);
145 else if (l == 2) /* JIS X 0213 plane 1? */
150 if (ku >= 1 && ku <= 62)
151 *p++ = (ku + 0x101) >> 1;
152 else if (ku >= 63 && ku <= 94)
153 *p++ = (ku + 0x181) >> 1;
155 report_invalid_encoding(PG_EUC_JIS_2004,
156 (const char *) euc, len);
160 if (ten >= 1 && ten <= 63)
162 else if (ten >= 64 && ten <= 94)
165 report_invalid_encoding(PG_EUC_JIS_2004,
166 (const char *) euc, len);
172 report_invalid_encoding(PG_EUC_JIS_2004,
173 (const char *) euc, len);
182 * returns SHIFT_JIS_2004 "ku" code indicated by second byte
183 * *ku = 0: "ku" = even
184 * *ku = 1: "ku" = odd
187 get_ten(int b, int *ku)
191 if (b >= 0x40 && b <= 0x7e)
196 else if (b >= 0x80 && b <= 0x9e)
201 else if (b >= 0x9f && b <= 0xfc)
208 ten = -1; /* error */
209 *ku = 0; /* keep compiler quiet */
215 * SHIFT_JIS_2004 ---> EUC_JIS_2004
219 shift_jis_20042euc_jis_2004(const unsigned char *sjis, unsigned char *p, int len)
234 if (!IS_HIGHBIT_SET(c1))
238 report_invalid_encoding(PG_SHIFT_JIS_2004,
239 (const char *) sjis, len);
246 l = pg_encoding_verifymb(PG_SHIFT_JIS_2004, (const char *) sjis, len);
249 report_invalid_encoding(PG_SHIFT_JIS_2004,
250 (const char *) sjis, len);
252 if (c1 >= 0xa1 && c1 <= 0xdf && l == 1)
254 /* JIS X0201 (1 byte kana) */
267 if (c1 >= 0x81 && c1 <= 0x9f) /* plane 1 1ku-62ku */
269 ku = (c1 << 1) - 0x100;
270 ten = get_ten(c2, &kubun);
272 report_invalid_encoding(PG_SHIFT_JIS_2004,
273 (const char *) sjis, len);
276 else if (c1 >= 0xe0 && c1 <= 0xef) /* plane 1 62ku-94ku */
278 ku = (c1 << 1) - 0x180;
279 ten = get_ten(c2, &kubun);
281 report_invalid_encoding(PG_SHIFT_JIS_2004,
283 (const char *) sjis, len);
286 else if (c1 >= 0xf0 && c1 <= 0xf3) /* plane 2
287 * 1,3,4,5,8,12,13,14,15 ku */
290 ten = get_ten(c2, &kubun);
292 report_invalid_encoding(PG_SHIFT_JIS_2004,
293 (const char *) sjis, len);
297 ku = kubun == 0 ? 8 : 1;
300 ku = kubun == 0 ? 4 : 3;
303 ku = kubun == 0 ? 12 : 5;
306 ku = kubun == 0 ? 14 : 13;
310 else if (c1 >= 0xf4 && c1 <= 0xfc) /* plane 2 78-94ku */
313 ten = get_ten(c2, &kubun);
315 report_invalid_encoding(PG_SHIFT_JIS_2004,
316 (const char *) sjis, len);
317 if (c1 == 0xf4 && kubun == 1)
320 ku = (c1 << 1) - 0x19a - kubun;
323 report_invalid_encoding(PG_SHIFT_JIS_2004,
324 (const char *) sjis, len);