]> git.ipfire.org Git - thirdparty/postgresql.git/blob
a6d8490bdea173b71ee5e9e1d6274d68245f8a00
[thirdparty/postgresql.git] /
1 /*-------------------------------------------------------------------------
2 *
3 * EUC_JIS_2004, SHIFT_JIS_2004
4 *
5 * Copyright (c) 2007-2008, PostgreSQL Global Development Group
6 *
7 * IDENTIFICATION
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 1.3.2.1 2009/01/29 19:23:56 tgl Exp $
9 *
10 *-------------------------------------------------------------------------
11 */
12
13 #include "postgres.h"
14 #include "fmgr.h"
15 #include "mb/pg_wchar.h"
16
17 PG_MODULE_MAGIC;
18
19 PG_FUNCTION_INFO_V1(euc_jis_2004_to_shift_jis_2004);
20 PG_FUNCTION_INFO_V1(shift_jis_2004_to_euc_jis_2004);
21
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);
24
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);
27
28 /* ----------
29 * conv_proc(
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
35 * ) returns VOID;
36 * ----------
37 */
38
39 Datum
40 euc_jis_2004_to_shift_jis_2004(PG_FUNCTION_ARGS)
41 {
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);
45
46 CHECK_ENCODING_CONVERSION_ARGS(PG_EUC_JIS_2004, PG_SHIFT_JIS_2004);
47
48 euc_jis_20042shift_jis_2004(src, dest, len);
49
50 PG_RETURN_VOID();
51 }
52
53 Datum
54 shift_jis_2004_to_euc_jis_2004(PG_FUNCTION_ARGS)
55 {
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);
59
60 CHECK_ENCODING_CONVERSION_ARGS(PG_SHIFT_JIS_2004, PG_EUC_JIS_2004);
61
62 shift_jis_20042euc_jis_2004(src, dest, len);
63
64 PG_RETURN_VOID();
65 }
66
67 /*
68 * EUC_JIS_2004 -> SHIFT_JIS_2004
69 */
70 static void
71 euc_jis_20042shift_jis_2004(const unsigned char *euc, unsigned char *p, int len)
72 {
73 int c1,
74 ku,
75 ten;
76 int l;
77
78 while (len > 0)
79 {
80 c1 = *euc;
81 if (!IS_HIGHBIT_SET(c1))
82 {
83 /* ASCII */
84 if (c1 == 0)
85 report_invalid_encoding(PG_EUC_JIS_2004,
86 (const char *) euc, len);
87 *p++ = c1;
88 euc++;
89 len--;
90 continue;
91 }
92
93 l = pg_encoding_verifymb(PG_EUC_JIS_2004, (const char *) euc, len);
94
95 if (l < 0)
96 report_invalid_encoding(PG_EUC_JIS_2004,
97 (const char *) euc, len);
98
99 if (c1 == SS2 && l == 2) /* JIS X 0201 kana? */
100 {
101 *p++ = euc[1];
102 }
103 else if (c1 == SS3 && l == 3) /* JIS X 0213 plane 2? */
104 {
105 ku = euc[1] - 0xa0;
106 ten = euc[2] - 0xa0;
107
108 switch (ku)
109 {
110 case 1:
111 case 3:
112 case 4:
113 case 5:
114 case 8:
115 case 12:
116 case 13:
117 case 14:
118 case 15:
119 *p++ = ((ku + 0x1df) >> 1) - (ku >> 3) * 3;
120 break;
121 default:
122 if (ku >= 78 && ku <= 94)
123 {
124 *p++ = (ku + 0x19b) >> 1;
125 }
126 else
127 report_invalid_encoding(PG_EUC_JIS_2004,
128 (const char *) euc, len);
129 }
130
131 if (ku % 2)
132 {
133 if (ten >= 1 && ten <= 63)
134 *p++ = ten + 0x3f;
135 else if (ten >= 64 && ten <= 94)
136 *p++ = ten + 0x40;
137 else
138 report_invalid_encoding(PG_EUC_JIS_2004,
139 (const char *) euc, len);
140 }
141 else
142 *p++ = ten + 0x9e;
143 }
144
145 else if (l == 2) /* JIS X 0213 plane 1? */
146 {
147 ku = c1 - 0xa0;
148 ten = euc[1] - 0xa0;
149
150 if (ku >= 1 && ku <= 62)
151 *p++ = (ku + 0x101) >> 1;
152 else if (ku >= 63 && ku <= 94)
153 *p++ = (ku + 0x181) >> 1;
154 else
155 report_invalid_encoding(PG_EUC_JIS_2004,
156 (const char *) euc, len);
157
158 if (ku % 2)
159 {
160 if (ten >= 1 && ten <= 63)
161 *p++ = ten + 0x3f;
162 else if (ten >= 64 && ten <= 94)
163 *p++ = ten + 0x40;
164 else
165 report_invalid_encoding(PG_EUC_JIS_2004,
166 (const char *) euc, len);
167 }
168 else
169 *p++ = ten + 0x9e;
170 }
171 else
172 report_invalid_encoding(PG_EUC_JIS_2004,
173 (const char *) euc, len);
174
175 euc += l;
176 len -= l;
177 }
178 *p = '\0';
179 }
180
181 /*
182 * returns SHIFT_JIS_2004 "ku" code indicated by second byte
183 * *ku = 0: "ku" = even
184 * *ku = 1: "ku" = odd
185 */
186 static int
187 get_ten(int b, int *ku)
188 {
189 int ten;
190
191 if (b >= 0x40 && b <= 0x7e)
192 {
193 ten = b - 0x3f;
194 *ku = 1;
195 }
196 else if (b >= 0x80 && b <= 0x9e)
197 {
198 ten = b - 0x40;
199 *ku = 1;
200 }
201 else if (b >= 0x9f && b <= 0xfc)
202 {
203 ten = b - 0x9e;
204 *ku = 0;
205 }
206 else
207 {
208 ten = -1; /* error */
209 }
210 return ten;
211 }
212
213 /*
214 * SHIFT_JIS_2004 ---> EUC_JIS_2004
215 */
216
217 static void
218 shift_jis_20042euc_jis_2004(const unsigned char *sjis, unsigned char *p, int len)
219 {
220 int c1,
221 c2;
222 int ku,
223 ten,
224 kubun;
225 int plane;
226 int l;
227
228 while (len > 0)
229 {
230 c1 = *sjis;
231 c2 = sjis[1];
232
233 if (!IS_HIGHBIT_SET(c1))
234 {
235 /* ASCII */
236 if (c1 == 0)
237 report_invalid_encoding(PG_SHIFT_JIS_2004,
238 (const char *) sjis, len);
239 *p++ = c1;
240 sjis++;
241 len--;
242 continue;
243 }
244
245 l = pg_encoding_verifymb(PG_SHIFT_JIS_2004, (const char *) sjis, len);
246
247 if (l < 0)
248 report_invalid_encoding(PG_SHIFT_JIS_2004,
249 (const char *) sjis, len);
250
251 if (c1 >= 0xa1 && c1 <= 0xdf && l == 1)
252 {
253 /* JIS X0201 (1 byte kana) */
254 *p++ = SS2;
255 *p++ = c1;
256 }
257 else if (l == 2)
258 {
259 plane = 1;
260 ku = 1;
261 ten = 1;
262
263 /*
264 * JIS X 0213
265 */
266 if (c1 >= 0x81 && c1 <= 0x9f) /* plane 1 1ku-62ku */
267 {
268 ku = (c1 << 1) - 0x100;
269 ten = get_ten(c2, &kubun);
270 if (ten < 0)
271 report_invalid_encoding(PG_SHIFT_JIS_2004,
272 (const char *) sjis, len);
273 ku -= kubun;
274 }
275 else if (c1 >= 0xe0 && c1 <= 0xef) /* plane 1 62ku-94ku */
276 {
277 ku = (c1 << 1) - 0x180;
278 ten = get_ten(c2, &kubun);
279 if (ten < 0)
280 report_invalid_encoding(PG_SHIFT_JIS_2004,
281
282 (const char *) sjis, len);
283 ku -= kubun;
284 }
285 else if (c1 >= 0xf0 && c1 <= 0xf3) /* plane 2
286 * 1,3,4,5,8,12,13,14,15 ku */
287 {
288 plane = 2;
289 ten = get_ten(c2, &kubun);
290 if (ten < 0)
291 report_invalid_encoding(PG_SHIFT_JIS_2004,
292 (const char *) sjis, len);
293 switch (c1)
294 {
295 case 0xf0:
296 ku = kubun == 0 ? 8 : 1;
297 break;
298 case 0xf1:
299 ku = kubun == 0 ? 4 : 3;
300 break;
301 case 0xf2:
302 ku = kubun == 0 ? 12 : 5;
303 break;
304 default:
305 ku = kubun == 0 ? 14 : 13;
306 break;
307 }
308 }
309 else if (c1 >= 0xf4 && c1 <= 0xfc) /* plane 2 78-94ku */
310 {
311 plane = 2;
312 ten = get_ten(c2, &kubun);
313 if (ten < 0)
314 report_invalid_encoding(PG_SHIFT_JIS_2004,
315 (const char *) sjis, len);
316 if (c1 == 0xf4 && kubun == 1)
317 ku = 15;
318 else
319 ku = (c1 << 1) - 0x19a - kubun;
320 }
321 else
322 report_invalid_encoding(PG_SHIFT_JIS_2004,
323 (const char *) sjis, len);
324
325 if (plane == 2)
326 *p++ = SS3;
327
328 *p++ = ku + 0xa0;
329 *p++ = ten + 0xa0;
330 }
331 sjis += l;
332 len -= l;
333 }
334 *p = '\0';
335 }