Update copyright for 2022
[pgsql.git] / src / backend / utils / mb / conversion_procs / euc2004_sjis2004 / euc2004_sjis2004.c
blob6891600754c10e93550dcf01c9c6e7cb74a5383c
1 /*-------------------------------------------------------------------------
3 * EUC_JIS_2004, SHIFT_JIS_2004
5 * Copyright (c) 2007-2022, PostgreSQL Global Development Group
7 * IDENTIFICATION
8 * src/backend/utils/mb/conversion_procs/euc2004_sjis2004/euc2004_sjis2004.c
10 *-------------------------------------------------------------------------
13 #include "postgres.h"
14 #include "fmgr.h"
15 #include "mb/pg_wchar.h"
17 PG_MODULE_MAGIC;
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 static int euc_jis_20042shift_jis_2004(const unsigned char *euc, unsigned char *p, int len, bool noError);
23 static int shift_jis_20042euc_jis_2004(const unsigned char *sjis, unsigned char *p, int len, bool noError);
25 /* ----------
26 * conv_proc(
27 * INTEGER, -- source encoding id
28 * INTEGER, -- destination encoding id
29 * CSTRING, -- source string (null terminated C string)
30 * CSTRING, -- destination string (null terminated C string)
31 * INTEGER, -- source string length
32 * BOOL -- if true, don't throw an error if conversion fails
33 * ) returns INTEGER;
35 * Returns the number of bytes successfully converted.
36 * ----------
39 Datum
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);
45 bool noError = PG_GETARG_BOOL(5);
46 int converted;
48 CHECK_ENCODING_CONVERSION_ARGS(PG_EUC_JIS_2004, PG_SHIFT_JIS_2004);
50 converted = euc_jis_20042shift_jis_2004(src, dest, len, noError);
52 PG_RETURN_INT32(converted);
55 Datum
56 shift_jis_2004_to_euc_jis_2004(PG_FUNCTION_ARGS)
58 unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
59 unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
60 int len = PG_GETARG_INT32(4);
61 bool noError = PG_GETARG_BOOL(5);
62 int converted;
64 CHECK_ENCODING_CONVERSION_ARGS(PG_SHIFT_JIS_2004, PG_EUC_JIS_2004);
66 converted = shift_jis_20042euc_jis_2004(src, dest, len, noError);
68 PG_RETURN_INT32(converted);
72 * EUC_JIS_2004 -> SHIFT_JIS_2004
74 static int
75 euc_jis_20042shift_jis_2004(const unsigned char *euc, unsigned char *p, int len, bool noError)
77 const unsigned char *start = euc;
78 int c1,
79 ku,
80 ten;
81 int l;
83 while (len > 0)
85 c1 = *euc;
86 if (!IS_HIGHBIT_SET(c1))
88 /* ASCII */
89 if (c1 == 0)
91 if (noError)
92 break;
93 report_invalid_encoding(PG_EUC_JIS_2004,
94 (const char *) euc, len);
96 *p++ = c1;
97 euc++;
98 len--;
99 continue;
102 l = pg_encoding_verifymbchar(PG_EUC_JIS_2004, (const char *) euc, len);
104 if (l < 0)
106 if (noError)
107 break;
108 report_invalid_encoding(PG_EUC_JIS_2004,
109 (const char *) euc, len);
112 if (c1 == SS2 && l == 2) /* JIS X 0201 kana? */
114 *p++ = euc[1];
116 else if (c1 == SS3 && l == 3) /* JIS X 0213 plane 2? */
118 ku = euc[1] - 0xa0;
119 ten = euc[2] - 0xa0;
121 switch (ku)
123 case 1:
124 case 3:
125 case 4:
126 case 5:
127 case 8:
128 case 12:
129 case 13:
130 case 14:
131 case 15:
132 *p++ = ((ku + 0x1df) >> 1) - (ku >> 3) * 3;
133 break;
134 default:
135 if (ku >= 78 && ku <= 94)
137 *p++ = (ku + 0x19b) >> 1;
139 else
141 if (noError)
142 break;
143 report_invalid_encoding(PG_EUC_JIS_2004,
144 (const char *) euc, len);
148 if (ku % 2)
150 if (ten >= 1 && ten <= 63)
151 *p++ = ten + 0x3f;
152 else if (ten >= 64 && ten <= 94)
153 *p++ = ten + 0x40;
154 else
156 if (noError)
157 break;
158 report_invalid_encoding(PG_EUC_JIS_2004,
159 (const char *) euc, len);
162 else
163 *p++ = ten + 0x9e;
166 else if (l == 2) /* JIS X 0213 plane 1? */
168 ku = c1 - 0xa0;
169 ten = euc[1] - 0xa0;
171 if (ku >= 1 && ku <= 62)
172 *p++ = (ku + 0x101) >> 1;
173 else if (ku >= 63 && ku <= 94)
174 *p++ = (ku + 0x181) >> 1;
175 else
177 if (noError)
178 break;
179 report_invalid_encoding(PG_EUC_JIS_2004,
180 (const char *) euc, len);
183 if (ku % 2)
185 if (ten >= 1 && ten <= 63)
186 *p++ = ten + 0x3f;
187 else if (ten >= 64 && ten <= 94)
188 *p++ = ten + 0x40;
189 else
191 if (noError)
192 break;
193 report_invalid_encoding(PG_EUC_JIS_2004,
194 (const char *) euc, len);
197 else
198 *p++ = ten + 0x9e;
200 else
202 if (noError)
203 break;
204 report_invalid_encoding(PG_EUC_JIS_2004,
205 (const char *) euc, len);
208 euc += l;
209 len -= l;
211 *p = '\0';
213 return euc - start;
217 * returns SHIFT_JIS_2004 "ku" code indicated by second byte
218 * *ku = 0: "ku" = even
219 * *ku = 1: "ku" = odd
221 static int
222 get_ten(int b, int *ku)
224 int ten;
226 if (b >= 0x40 && b <= 0x7e)
228 ten = b - 0x3f;
229 *ku = 1;
231 else if (b >= 0x80 && b <= 0x9e)
233 ten = b - 0x40;
234 *ku = 1;
236 else if (b >= 0x9f && b <= 0xfc)
238 ten = b - 0x9e;
239 *ku = 0;
241 else
243 ten = -1; /* error */
244 *ku = 0; /* keep compiler quiet */
246 return ten;
250 * SHIFT_JIS_2004 ---> EUC_JIS_2004
253 static int
254 shift_jis_20042euc_jis_2004(const unsigned char *sjis, unsigned char *p, int len, bool noError)
256 const unsigned char *start = sjis;
257 int c1;
258 int ku,
259 ten,
260 kubun;
261 int plane;
262 int l;
264 while (len > 0)
266 c1 = *sjis;
268 if (!IS_HIGHBIT_SET(c1))
270 /* ASCII */
271 if (c1 == 0)
273 if (noError)
274 break;
275 report_invalid_encoding(PG_SHIFT_JIS_2004,
276 (const char *) sjis, len);
278 *p++ = c1;
279 sjis++;
280 len--;
281 continue;
284 l = pg_encoding_verifymbchar(PG_SHIFT_JIS_2004, (const char *) sjis, len);
286 if (l < 0 || l > len)
288 if (noError)
289 break;
290 report_invalid_encoding(PG_SHIFT_JIS_2004,
291 (const char *) sjis, len);
294 if (c1 >= 0xa1 && c1 <= 0xdf && l == 1)
296 /* JIS X0201 (1 byte kana) */
297 *p++ = SS2;
298 *p++ = c1;
300 else if (l == 2)
302 int c2 = sjis[1];
304 plane = 1;
305 ku = 1;
306 ten = 1;
309 * JIS X 0213
311 if (c1 >= 0x81 && c1 <= 0x9f) /* plane 1 1ku-62ku */
313 ku = (c1 << 1) - 0x100;
314 ten = get_ten(c2, &kubun);
315 if (ten < 0)
317 if (noError)
318 break;
319 report_invalid_encoding(PG_SHIFT_JIS_2004,
320 (const char *) sjis, len);
322 ku -= kubun;
324 else if (c1 >= 0xe0 && c1 <= 0xef) /* plane 1 62ku-94ku */
326 ku = (c1 << 1) - 0x180;
327 ten = get_ten(c2, &kubun);
328 if (ten < 0)
330 if (noError)
331 break;
332 report_invalid_encoding(PG_SHIFT_JIS_2004,
333 (const char *) sjis, len);
335 ku -= kubun;
337 else if (c1 >= 0xf0 && c1 <= 0xf3) /* plane 2
338 * 1,3,4,5,8,12,13,14,15 ku */
340 plane = 2;
341 ten = get_ten(c2, &kubun);
342 if (ten < 0)
344 if (noError)
345 break;
346 report_invalid_encoding(PG_SHIFT_JIS_2004,
347 (const char *) sjis, len);
349 switch (c1)
351 case 0xf0:
352 ku = kubun == 0 ? 8 : 1;
353 break;
354 case 0xf1:
355 ku = kubun == 0 ? 4 : 3;
356 break;
357 case 0xf2:
358 ku = kubun == 0 ? 12 : 5;
359 break;
360 default:
361 ku = kubun == 0 ? 14 : 13;
362 break;
365 else if (c1 >= 0xf4 && c1 <= 0xfc) /* plane 2 78-94ku */
367 plane = 2;
368 ten = get_ten(c2, &kubun);
369 if (ten < 0)
371 if (noError)
372 break;
373 report_invalid_encoding(PG_SHIFT_JIS_2004,
374 (const char *) sjis, len);
376 if (c1 == 0xf4 && kubun == 1)
377 ku = 15;
378 else
379 ku = (c1 << 1) - 0x19a - kubun;
381 else
383 if (noError)
384 break;
385 report_invalid_encoding(PG_SHIFT_JIS_2004,
386 (const char *) sjis, len);
389 if (plane == 2)
390 *p++ = SS3;
392 *p++ = ku + 0xa0;
393 *p++ = ten + 0xa0;
395 sjis += l;
396 len -= l;
398 *p = '\0';
400 return sjis - start;