PostgreSQL Source Code  git master
cyrillic_and_mic.c
Go to the documentation of this file.
1 /*-------------------------------------------------------------------------
2  *
3  * Cyrillic and MULE_INTERNAL
4  *
5  * Portions Copyright (c) 1996-2024, PostgreSQL Global Development Group
6  * Portions Copyright (c) 1994, Regents of the University of California
7  *
8  * IDENTIFICATION
9  * src/backend/utils/mb/conversion_procs/cyrillic_and_mic/cyrillic_and_mic.c
10  *
11  *-------------------------------------------------------------------------
12  */
13 
14 #include "postgres.h"
15 #include "fmgr.h"
16 #include "mb/pg_wchar.h"
17 
18 PG_MODULE_MAGIC;
19 
20 PG_FUNCTION_INFO_V1(koi8r_to_mic);
21 PG_FUNCTION_INFO_V1(mic_to_koi8r);
22 PG_FUNCTION_INFO_V1(iso_to_mic);
23 PG_FUNCTION_INFO_V1(mic_to_iso);
24 PG_FUNCTION_INFO_V1(win1251_to_mic);
25 PG_FUNCTION_INFO_V1(mic_to_win1251);
26 PG_FUNCTION_INFO_V1(win866_to_mic);
27 PG_FUNCTION_INFO_V1(mic_to_win866);
28 PG_FUNCTION_INFO_V1(koi8r_to_win1251);
29 PG_FUNCTION_INFO_V1(win1251_to_koi8r);
30 PG_FUNCTION_INFO_V1(koi8r_to_win866);
31 PG_FUNCTION_INFO_V1(win866_to_koi8r);
32 PG_FUNCTION_INFO_V1(win866_to_win1251);
33 PG_FUNCTION_INFO_V1(win1251_to_win866);
34 PG_FUNCTION_INFO_V1(iso_to_koi8r);
35 PG_FUNCTION_INFO_V1(koi8r_to_iso);
36 PG_FUNCTION_INFO_V1(iso_to_win1251);
37 PG_FUNCTION_INFO_V1(win1251_to_iso);
38 PG_FUNCTION_INFO_V1(iso_to_win866);
39 PG_FUNCTION_INFO_V1(win866_to_iso);
40 
41 /* ----------
42  * conv_proc(
43  * INTEGER, -- source encoding id
44  * INTEGER, -- destination encoding id
45  * CSTRING, -- source string (null terminated C string)
46  * CSTRING, -- destination string (null terminated C string)
47  * INTEGER, -- source string length
48  * BOOL -- if true, don't throw an error if conversion fails
49  * ) returns INTEGER;
50  *
51  * Returns the number of bytes successfully converted.
52  * ----------
53  */
54 
55 /*
56  * Cyrillic support
57  * currently supported Cyrillic encodings:
58  *
59  * KOI8-R (this is also the charset for the mule internal code for Cyrillic)
60  * ISO-8859-5
61  * Microsoft's CP1251 (windows-1251)
62  * Alternativny Variant (MS-DOS CP866)
63  */
64 
65 /* ISO-8859-5 to KOI8-R */
66 static const unsigned char iso2koi[] = {
67  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
68  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
69  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
70  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
71  0x00, 0xB3, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
72  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
73  0xE1, 0xE2, 0xF7, 0xE7, 0xE4, 0xE5, 0xF6, 0xFA,
74  0xE9, 0xEA, 0xEB, 0xEC, 0xED, 0xEE, 0xEF, 0xF0,
75  0xF2, 0xF3, 0xF4, 0xF5, 0xE6, 0xE8, 0xE3, 0xFE,
76  0xFB, 0xFD, 0xFF, 0xF9, 0xF8, 0xFC, 0xE0, 0xF1,
77  0xC1, 0xC2, 0xD7, 0xC7, 0xC4, 0xC5, 0xD6, 0xDA,
78  0xC9, 0xCA, 0xCB, 0xCC, 0xCD, 0xCE, 0xCF, 0xD0,
79  0xD2, 0xD3, 0xD4, 0xD5, 0xC6, 0xC8, 0xC3, 0xDE,
80  0xDB, 0xDD, 0xDF, 0xD9, 0xD8, 0xDC, 0xC0, 0xD1,
81  0x00, 0xA3, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
82  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
83 };
84 
85 /* KOI8-R to ISO-8859-5 */
86 static const unsigned char koi2iso[] = {
87  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
88  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
89  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
90  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
91  0x00, 0x00, 0x00, 0xF1, 0x00, 0x00, 0x00, 0x00,
92  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
93  0x00, 0x00, 0x00, 0xA1, 0x00, 0x00, 0x00, 0x00,
94  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
95  0xEE, 0xD0, 0xD1, 0xE6, 0xD4, 0xD5, 0xE4, 0xD3,
96  0xE5, 0xD8, 0xD9, 0xDA, 0xDB, 0xDC, 0xDD, 0xDE,
97  0xDF, 0xEF, 0xE0, 0xE1, 0xE2, 0xE3, 0xD6, 0xD2,
98  0xEC, 0xEB, 0xD7, 0xE8, 0xED, 0xE9, 0xE7, 0xEA,
99  0xCE, 0xB0, 0xB1, 0xC6, 0xB4, 0xB5, 0xC4, 0xB3,
100  0xC5, 0xB8, 0xB9, 0xBA, 0xBB, 0xBC, 0xBD, 0xBE,
101  0xBF, 0xCF, 0xC0, 0xC1, 0xC2, 0xC3, 0xB6, 0xB2,
102  0xCC, 0xCB, 0xB7, 0xC8, 0xCD, 0xC9, 0xC7, 0xCA
103 };
104 
105 /* WIN1251 to KOI8-R */
106 static const unsigned char win12512koi[] = {
107  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
108  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
109  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
110  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
111  0x00, 0x00, 0x00, 0x00, 0x00, 0xBD, 0x00, 0x00,
112  0xB3, 0x00, 0xB4, 0x00, 0x00, 0x00, 0x00, 0xB7,
113  0x00, 0x00, 0xB6, 0xA6, 0xAD, 0x00, 0x00, 0x00,
114  0xA3, 0x00, 0xA4, 0x00, 0x00, 0x00, 0x00, 0xA7,
115  0xE1, 0xE2, 0xF7, 0xE7, 0xE4, 0xE5, 0xF6, 0xFA,
116  0xE9, 0xEA, 0xEB, 0xEC, 0xED, 0xEE, 0xEF, 0xF0,
117  0xF2, 0xF3, 0xF4, 0xF5, 0xE6, 0xE8, 0xE3, 0xFE,
118  0xFB, 0xFD, 0xFF, 0xF9, 0xF8, 0xFC, 0xE0, 0xF1,
119  0xC1, 0xC2, 0xD7, 0xC7, 0xC4, 0xC5, 0xD6, 0xDA,
120  0xC9, 0xCA, 0xCB, 0xCC, 0xCD, 0xCE, 0xCF, 0xD0,
121  0xD2, 0xD3, 0xD4, 0xD5, 0xC6, 0xC8, 0xC3, 0xDE,
122  0xDB, 0xDD, 0xDF, 0xD9, 0xD8, 0xDC, 0xC0, 0xD1
123 };
124 
125 /* KOI8-R to WIN1251 */
126 static const unsigned char koi2win1251[] = {
127  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
128  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
129  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
130  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
131  0x00, 0x00, 0x00, 0xB8, 0xBA, 0x00, 0xB3, 0xBF,
132  0x00, 0x00, 0x00, 0x00, 0x00, 0xB4, 0x00, 0x00,
133  0x00, 0x00, 0x00, 0xA8, 0xAA, 0x00, 0xB2, 0xAF,
134  0x00, 0x00, 0x00, 0x00, 0x00, 0xA5, 0x00, 0x00,
135  0xFE, 0xE0, 0xE1, 0xF6, 0xE4, 0xE5, 0xF4, 0xE3,
136  0xF5, 0xE8, 0xE9, 0xEA, 0xEB, 0xEC, 0xED, 0xEE,
137  0xEF, 0xFF, 0xF0, 0xF1, 0xF2, 0xF3, 0xE6, 0xE2,
138  0xFC, 0xFB, 0xE7, 0xF8, 0xFD, 0xF9, 0xF7, 0xFA,
139  0xDE, 0xC0, 0xC1, 0xD6, 0xC4, 0xC5, 0xD4, 0xC3,
140  0xD5, 0xC8, 0xC9, 0xCA, 0xCB, 0xCC, 0xCD, 0xCE,
141  0xCF, 0xDF, 0xD0, 0xD1, 0xD2, 0xD3, 0xC6, 0xC2,
142  0xDC, 0xDB, 0xC7, 0xD8, 0xDD, 0xD9, 0xD7, 0xDA
143 };
144 
145 /* WIN866 to KOI8-R */
146 static const unsigned char win8662koi[] = {
147  0xE1, 0xE2, 0xF7, 0xE7, 0xE4, 0xE5, 0xF6, 0xFA,
148  0xE9, 0xEA, 0xEB, 0xEC, 0xED, 0xEE, 0xEF, 0xF0,
149  0xF2, 0xF3, 0xF4, 0xF5, 0xE6, 0xE8, 0xE3, 0xFE,
150  0xFB, 0xFD, 0xFF, 0xF9, 0xF8, 0xFC, 0xE0, 0xF1,
151  0xC1, 0xC2, 0xD7, 0xC7, 0xC4, 0xC5, 0xD6, 0xDA,
152  0xC9, 0xCA, 0xCB, 0xCC, 0xCD, 0xCE, 0xCF, 0xD0,
153  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
154  0x00, 0x00, 0x00, 0x00, 0x00, 0xBD, 0x00, 0x00,
155  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
156  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
157  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
158  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
159  0xD2, 0xD3, 0xD4, 0xD5, 0xC6, 0xC8, 0xC3, 0xDE,
160  0xDB, 0xDD, 0xDF, 0xD9, 0xD8, 0xDC, 0xC0, 0xD1,
161  0xB3, 0xA3, 0xB4, 0xA4, 0xB7, 0xA7, 0x00, 0x00,
162  0xB6, 0xA6, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
163 };
164 
165 /* KOI8-R to WIN866 */
166 static const unsigned char koi2win866[] = {
167  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
168  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
169  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
170  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
171  0x00, 0x00, 0x00, 0xF1, 0xF3, 0x00, 0xF9, 0xF5,
172  0x00, 0x00, 0x00, 0x00, 0x00, 0xAD, 0x00, 0x00,
173  0x00, 0x00, 0x00, 0xF0, 0xF2, 0x00, 0xF8, 0xF4,
174  0x00, 0x00, 0x00, 0x00, 0x00, 0xBD, 0x00, 0x00,
175  0xEE, 0xA0, 0xA1, 0xE6, 0xA4, 0xA5, 0xE4, 0xA3,
176  0xE5, 0xA8, 0xA9, 0xAA, 0xAB, 0xAC, 0xAD, 0xAE,
177  0xAF, 0xEF, 0xE0, 0xE1, 0xE2, 0xE3, 0xA6, 0xA2,
178  0xEC, 0xEB, 0xA7, 0xE8, 0xED, 0xE9, 0xE7, 0xEA,
179  0x9E, 0x80, 0x81, 0x96, 0x84, 0x85, 0x94, 0x83,
180  0x95, 0x88, 0x89, 0x8A, 0x8B, 0x8C, 0x8D, 0x8E,
181  0x8F, 0x9F, 0x90, 0x91, 0x92, 0x93, 0x86, 0x82,
182  0x9C, 0x9B, 0x87, 0x98, 0x9D, 0x99, 0x97, 0x9A
183 };
184 
185 /* WIN866 to WIN1251 */
186 static const unsigned char win8662win1251[] = {
187  0xC0, 0xC1, 0xC2, 0xC3, 0xC4, 0xC5, 0xC6, 0xC7,
188  0xC8, 0xC9, 0xCA, 0xCB, 0xCC, 0xCD, 0xCE, 0xCF,
189  0xD0, 0xD1, 0xD2, 0xD3, 0xD4, 0xD5, 0xD6, 0xD7,
190  0xD8, 0xD9, 0xDA, 0xDB, 0xDC, 0xDD, 0xDE, 0xDF,
191  0xE0, 0xE1, 0xE2, 0xE3, 0xE4, 0xE5, 0xE6, 0xE7,
192  0xE8, 0xE9, 0xEA, 0xEB, 0xEC, 0xED, 0xEE, 0xEF,
193  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
194  0x00, 0x00, 0x00, 0x00, 0x00, 0xA5, 0x00, 0x00,
195  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
196  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
197  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
198  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
199  0xF0, 0xF1, 0xF2, 0xF3, 0xF4, 0xF5, 0xF6, 0xF7,
200  0xF8, 0xF9, 0xFA, 0xFB, 0xFC, 0xFD, 0xFE, 0xFF,
201  0xA8, 0xB8, 0xAA, 0xBA, 0xAF, 0xBF, 0x00, 0x00,
202  0xB2, 0xB3, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
203 };
204 
205 /* WIN1251 to WIN866 */
206 static const unsigned char win12512win866[] = {
207  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
208  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
209  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
210  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
211  0x00, 0x00, 0x00, 0x00, 0x00, 0xBD, 0x00, 0x00,
212  0xF0, 0x00, 0xF2, 0x00, 0x00, 0x00, 0x00, 0xF4,
213  0x00, 0x00, 0xF8, 0xF9, 0xAD, 0x00, 0x00, 0x00,
214  0xF1, 0x00, 0xF3, 0x00, 0x00, 0x00, 0x00, 0xF5,
215  0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87,
216  0x88, 0x89, 0x8A, 0x8B, 0x8C, 0x8D, 0x8E, 0x8F,
217  0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97,
218  0x98, 0x99, 0x9A, 0x9B, 0x9C, 0x9D, 0x9E, 0x9F,
219  0xA0, 0xA1, 0xA2, 0xA3, 0xA4, 0xA5, 0xA6, 0xA7,
220  0xA8, 0xA9, 0xAA, 0xAB, 0xAC, 0xAD, 0xAE, 0xAF,
221  0xE0, 0xE1, 0xE2, 0xE3, 0xE4, 0xE5, 0xE6, 0xE7,
222  0xE8, 0xE9, 0xEA, 0xEB, 0xEC, 0xED, 0xEE, 0xEF
223 };
224 
225 /* ISO-8859-5 to WIN1251 */
226 static const unsigned char iso2win1251[] = {
227  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
228  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
229  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
230  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
231  0x00, 0xA8, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
232  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
233  0xC0, 0xC1, 0xC2, 0xC3, 0xC4, 0xC5, 0xC6, 0xC7,
234  0xC8, 0xC9, 0xCA, 0xCB, 0xCC, 0xCD, 0xCE, 0xCF,
235  0xD0, 0xD1, 0xD2, 0xD3, 0xD4, 0xD5, 0xD6, 0xD7,
236  0xD8, 0xD9, 0xDA, 0xDB, 0xDC, 0xDD, 0xDE, 0xDF,
237  0xE0, 0xE1, 0xE2, 0xE3, 0xE4, 0xE5, 0xE6, 0xE7,
238  0xE8, 0xE9, 0xEA, 0xEB, 0xEC, 0xED, 0xEE, 0xEF,
239  0xF0, 0xF1, 0xF2, 0xF3, 0xF4, 0xF5, 0xF6, 0xF7,
240  0xF8, 0xF9, 0xFA, 0xFB, 0xFC, 0xFD, 0xFE, 0xFF,
241  0x00, 0xB8, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
242  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
243 };
244 
245 /* WIN1251 to ISO-8859-5 */
246 static const unsigned char win12512iso[] = {
247  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
248  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
249  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
250  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
251  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
252  0xA1, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
253  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
254  0xF1, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
255  0xB0, 0xB1, 0xB2, 0xB3, 0xB4, 0xB5, 0xB6, 0xB7,
256  0xB8, 0xB9, 0xBA, 0xBB, 0xBC, 0xBD, 0xBE, 0xBF,
257  0xC0, 0xC1, 0xC2, 0xC3, 0xC4, 0xC5, 0xC6, 0xC7,
258  0xC8, 0xC9, 0xCA, 0xCB, 0xCC, 0xCD, 0xCE, 0xCF,
259  0xD0, 0xD1, 0xD2, 0xD3, 0xD4, 0xD5, 0xD6, 0xD7,
260  0xD8, 0xD9, 0xDA, 0xDB, 0xDC, 0xDD, 0xDE, 0xDF,
261  0xE0, 0xE1, 0xE2, 0xE3, 0xE4, 0xE5, 0xE6, 0xE7,
262  0xE8, 0xE9, 0xEA, 0xEB, 0xEC, 0xED, 0xEE, 0xEF
263 };
264 
265 /* ISO-8859-5 to WIN866 */
266 static const unsigned char iso2win866[] = {
267  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
268  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
269  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
270  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
271  0x00, 0xF0, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
272  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
273  0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87,
274  0x88, 0x89, 0x8A, 0x8B, 0x8C, 0x8D, 0x8E, 0x8F,
275  0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97,
276  0x98, 0x99, 0x9A, 0x9B, 0x9C, 0x9D, 0x9E, 0x9F,
277  0xA0, 0xA1, 0xA2, 0xA3, 0xA4, 0xA5, 0xA6, 0xA7,
278  0xA8, 0xA9, 0xAA, 0xAB, 0xAC, 0xAD, 0xAE, 0xAF,
279  0xE0, 0xE1, 0xE2, 0xE3, 0xE4, 0xE5, 0xE6, 0xE7,
280  0xE8, 0xE9, 0xEA, 0xEB, 0xEC, 0xED, 0xEE, 0xEF,
281  0x00, 0xF1, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
282  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
283 };
284 
285 /* WIN866 to ISO-8859-5 */
286 static const unsigned char win8662iso[] = {
287  0xB0, 0xB1, 0xB2, 0xB3, 0xB4, 0xB5, 0xB6, 0xB7,
288  0xB8, 0xB9, 0xBA, 0xBB, 0xBC, 0xBD, 0xBE, 0xBF,
289  0xC0, 0xC1, 0xC2, 0xC3, 0xC4, 0xC5, 0xC6, 0xC7,
290  0xC8, 0xC9, 0xCA, 0xCB, 0xCC, 0xCD, 0xCE, 0xCF,
291  0xD0, 0xD1, 0xD2, 0xD3, 0xD4, 0xD5, 0xD6, 0xD7,
292  0xD8, 0xD9, 0xDA, 0xDB, 0xDC, 0xDD, 0xDE, 0xDF,
293  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
294  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
295  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
296  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
297  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
298  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
299  0xE0, 0xE1, 0xE2, 0xE3, 0xE4, 0xE5, 0xE6, 0xE7,
300  0xE8, 0xE9, 0xEA, 0xEB, 0xEC, 0xED, 0xEE, 0xEF,
301  0xA1, 0xF1, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00,
302  0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
303 };
304 
305 
306 Datum
307 koi8r_to_mic(PG_FUNCTION_ARGS)
308 {
309  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
310  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
311  int len = PG_GETARG_INT32(4);
312  bool noError = PG_GETARG_BOOL(5);
313  int converted;
314 
315  CHECK_ENCODING_CONVERSION_ARGS(PG_KOI8R, PG_MULE_INTERNAL);
316 
317  converted = latin2mic(src, dest, len, LC_KOI8_R, PG_KOI8R, noError);
318 
319  PG_RETURN_INT32(converted);
320 }
321 
322 Datum
323 mic_to_koi8r(PG_FUNCTION_ARGS)
324 {
325  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
326  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
327  int len = PG_GETARG_INT32(4);
328  bool noError = PG_GETARG_BOOL(5);
329  int converted;
330 
331  CHECK_ENCODING_CONVERSION_ARGS(PG_MULE_INTERNAL, PG_KOI8R);
332 
333  converted = mic2latin(src, dest, len, LC_KOI8_R, PG_KOI8R, noError);
334 
335  PG_RETURN_INT32(converted);
336 }
337 
338 Datum
339 iso_to_mic(PG_FUNCTION_ARGS)
340 {
341  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
342  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
343  int len = PG_GETARG_INT32(4);
344  bool noError = PG_GETARG_BOOL(5);
345  int converted;
346 
347  CHECK_ENCODING_CONVERSION_ARGS(PG_ISO_8859_5, PG_MULE_INTERNAL);
348 
349  converted = latin2mic_with_table(src, dest, len, LC_KOI8_R, PG_ISO_8859_5, iso2koi, noError);
350 
351  PG_RETURN_INT32(converted);
352 }
353 
354 Datum
355 mic_to_iso(PG_FUNCTION_ARGS)
356 {
357  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
358  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
359  int len = PG_GETARG_INT32(4);
360  bool noError = PG_GETARG_BOOL(5);
361  int converted;
362 
363  CHECK_ENCODING_CONVERSION_ARGS(PG_MULE_INTERNAL, PG_ISO_8859_5);
364 
365  converted = mic2latin_with_table(src, dest, len, LC_KOI8_R, PG_ISO_8859_5, koi2iso, noError);
366 
367  PG_RETURN_INT32(converted);
368 }
369 
370 Datum
371 win1251_to_mic(PG_FUNCTION_ARGS)
372 {
373  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
374  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
375  int len = PG_GETARG_INT32(4);
376  bool noError = PG_GETARG_BOOL(5);
377  int converted;
378 
379  CHECK_ENCODING_CONVERSION_ARGS(PG_WIN1251, PG_MULE_INTERNAL);
380 
381  converted = latin2mic_with_table(src, dest, len, LC_KOI8_R, PG_WIN1251, win12512koi, noError);
382 
383  PG_RETURN_INT32(converted);
384 }
385 
386 Datum
387 mic_to_win1251(PG_FUNCTION_ARGS)
388 {
389  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
390  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
391  int len = PG_GETARG_INT32(4);
392  bool noError = PG_GETARG_BOOL(5);
393  int converted;
394 
395  CHECK_ENCODING_CONVERSION_ARGS(PG_MULE_INTERNAL, PG_WIN1251);
396 
397  converted = mic2latin_with_table(src, dest, len, LC_KOI8_R, PG_WIN1251, koi2win1251, noError);
398 
399  PG_RETURN_INT32(converted);
400 }
401 
402 Datum
403 win866_to_mic(PG_FUNCTION_ARGS)
404 {
405  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
406  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
407  int len = PG_GETARG_INT32(4);
408  bool noError = PG_GETARG_BOOL(5);
409  int converted;
410 
411  CHECK_ENCODING_CONVERSION_ARGS(PG_WIN866, PG_MULE_INTERNAL);
412 
413  converted = latin2mic_with_table(src, dest, len, LC_KOI8_R, PG_WIN866, win8662koi, noError);
414 
415  PG_RETURN_INT32(converted);
416 }
417 
418 Datum
419 mic_to_win866(PG_FUNCTION_ARGS)
420 {
421  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
422  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
423  int len = PG_GETARG_INT32(4);
424  bool noError = PG_GETARG_BOOL(5);
425  int converted;
426 
427  CHECK_ENCODING_CONVERSION_ARGS(PG_MULE_INTERNAL, PG_WIN866);
428 
429  converted = mic2latin_with_table(src, dest, len, LC_KOI8_R, PG_WIN866, koi2win866, noError);
430 
431  PG_RETURN_INT32(converted);
432 }
433 
434 Datum
435 koi8r_to_win1251(PG_FUNCTION_ARGS)
436 {
437  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
438  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
439  int len = PG_GETARG_INT32(4);
440  bool noError = PG_GETARG_BOOL(5);
441  int converted;
442 
443  CHECK_ENCODING_CONVERSION_ARGS(PG_KOI8R, PG_WIN1251);
444 
445  converted = local2local(src, dest, len, PG_KOI8R, PG_WIN1251, koi2win1251, noError);
446 
447  PG_RETURN_INT32(converted);
448 }
449 
450 Datum
451 win1251_to_koi8r(PG_FUNCTION_ARGS)
452 {
453  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
454  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
455  int len = PG_GETARG_INT32(4);
456  bool noError = PG_GETARG_BOOL(5);
457  int converted;
458 
459  CHECK_ENCODING_CONVERSION_ARGS(PG_WIN1251, PG_KOI8R);
460 
461  converted = local2local(src, dest, len, PG_WIN1251, PG_KOI8R, win12512koi, noError);
462 
463  PG_RETURN_INT32(converted);
464 }
465 
466 Datum
467 koi8r_to_win866(PG_FUNCTION_ARGS)
468 {
469  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
470  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
471  int len = PG_GETARG_INT32(4);
472  bool noError = PG_GETARG_BOOL(5);
473  int converted;
474 
475  CHECK_ENCODING_CONVERSION_ARGS(PG_KOI8R, PG_WIN866);
476 
477  converted = local2local(src, dest, len, PG_KOI8R, PG_WIN866, koi2win866, noError);
478 
479  PG_RETURN_INT32(converted);
480 }
481 
482 Datum
483 win866_to_koi8r(PG_FUNCTION_ARGS)
484 {
485  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
486  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
487  int len = PG_GETARG_INT32(4);
488  bool noError = PG_GETARG_BOOL(5);
489  int converted;
490 
491  CHECK_ENCODING_CONVERSION_ARGS(PG_WIN866, PG_KOI8R);
492 
493  converted = local2local(src, dest, len, PG_WIN866, PG_KOI8R, win8662koi, noError);
494 
495  PG_RETURN_INT32(converted);
496 }
497 
498 Datum
499 win866_to_win1251(PG_FUNCTION_ARGS)
500 {
501  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
502  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
503  int len = PG_GETARG_INT32(4);
504  bool noError = PG_GETARG_BOOL(5);
505  int converted;
506 
507  CHECK_ENCODING_CONVERSION_ARGS(PG_WIN866, PG_WIN1251);
508 
509  converted = local2local(src, dest, len, PG_WIN866, PG_WIN1251, win8662win1251, noError);
510 
511  PG_RETURN_INT32(converted);
512 }
513 
514 Datum
515 win1251_to_win866(PG_FUNCTION_ARGS)
516 {
517  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
518  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
519  int len = PG_GETARG_INT32(4);
520  bool noError = PG_GETARG_BOOL(5);
521  int converted;
522 
523  CHECK_ENCODING_CONVERSION_ARGS(PG_WIN1251, PG_WIN866);
524 
525  converted = local2local(src, dest, len, PG_WIN1251, PG_WIN866, win12512win866, noError);
526 
527  PG_RETURN_INT32(converted);
528 }
529 
530 Datum
531 iso_to_koi8r(PG_FUNCTION_ARGS)
532 {
533  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
534  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
535  int len = PG_GETARG_INT32(4);
536  bool noError = PG_GETARG_BOOL(5);
537  int converted;
538 
539  CHECK_ENCODING_CONVERSION_ARGS(PG_ISO_8859_5, PG_KOI8R);
540 
541  converted = local2local(src, dest, len, PG_ISO_8859_5, PG_KOI8R, iso2koi, noError);
542 
543  PG_RETURN_INT32(converted);
544 }
545 
546 Datum
547 koi8r_to_iso(PG_FUNCTION_ARGS)
548 {
549  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
550  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
551  int len = PG_GETARG_INT32(4);
552  bool noError = PG_GETARG_BOOL(5);
553  int converted;
554 
555  CHECK_ENCODING_CONVERSION_ARGS(PG_KOI8R, PG_ISO_8859_5);
556 
557  converted = local2local(src, dest, len, PG_KOI8R, PG_ISO_8859_5, koi2iso, noError);
558 
559  PG_RETURN_INT32(converted);
560 }
561 
562 Datum
563 iso_to_win1251(PG_FUNCTION_ARGS)
564 {
565  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
566  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
567  int len = PG_GETARG_INT32(4);
568  bool noError = PG_GETARG_BOOL(5);
569  int converted;
570 
571  CHECK_ENCODING_CONVERSION_ARGS(PG_ISO_8859_5, PG_WIN1251);
572 
573  converted = local2local(src, dest, len, PG_ISO_8859_5, PG_WIN1251, iso2win1251, noError);
574 
575  PG_RETURN_INT32(converted);
576 }
577 
578 Datum
579 win1251_to_iso(PG_FUNCTION_ARGS)
580 {
581  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
582  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
583  int len = PG_GETARG_INT32(4);
584  bool noError = PG_GETARG_BOOL(5);
585  int converted;
586 
587  CHECK_ENCODING_CONVERSION_ARGS(PG_WIN1251, PG_ISO_8859_5);
588 
589  converted = local2local(src, dest, len, PG_WIN1251, PG_ISO_8859_5, win12512iso, noError);
590 
591  PG_RETURN_INT32(converted);
592 }
593 
594 Datum
595 iso_to_win866(PG_FUNCTION_ARGS)
596 {
597  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
598  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
599  int len = PG_GETARG_INT32(4);
600  bool noError = PG_GETARG_BOOL(5);
601  int converted;
602 
603  CHECK_ENCODING_CONVERSION_ARGS(PG_ISO_8859_5, PG_WIN866);
604 
605  converted = local2local(src, dest, len, PG_ISO_8859_5, PG_WIN866, iso2win866, noError);
606 
607  PG_RETURN_INT32(converted);
608 }
609 
610 Datum
611 win866_to_iso(PG_FUNCTION_ARGS)
612 {
613  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
614  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
615  int len = PG_GETARG_INT32(4);
616  bool noError = PG_GETARG_BOOL(5);
617  int converted;
618 
619  CHECK_ENCODING_CONVERSION_ARGS(PG_WIN866, PG_ISO_8859_5);
620 
621  converted = local2local(src, dest, len, PG_WIN866, PG_ISO_8859_5, win8662iso, noError);
622 
623  PG_RETURN_INT32(converted);
624 }
mic2latin_with_table
int mic2latin_with_table(const unsigned char *mic, unsigned char *p, int len, int lc, int encoding, const unsigned char *tab, bool noError)
Definition: conv.c:257
latin2mic_with_table
int latin2mic_with_table(const unsigned char *l, unsigned char *p, int len, int lc, int encoding, const unsigned char *tab, bool noError)
Definition: conv.c:194
mic2latin
int mic2latin(const unsigned char *mic, unsigned char *p, int len, int lc, int encoding, bool noError)
Definition: conv.c:127
local2local
int local2local(const unsigned char *l, unsigned char *p, int len, int src_encoding, int dest_encoding, const unsigned char *tab, bool noError)
Definition: conv.c:33
latin2mic
int latin2mic(const unsigned char *l, unsigned char *p, int len, int lc, int encoding, bool noError)
Definition: conv.c:89
win8662koi
static const unsigned char win8662koi[]
Definition: cyrillic_and_mic.c:146
iso2win866
static const unsigned char iso2win866[]
Definition: cyrillic_and_mic.c:266
win866_to_win1251
Datum win866_to_win1251(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:499
win1251_to_win866
Datum win1251_to_win866(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:515
win12512win866
static const unsigned char win12512win866[]
Definition: cyrillic_and_mic.c:206
koi8r_to_win1251
Datum koi8r_to_win1251(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:435
win1251_to_mic
Datum win1251_to_mic(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:371
win8662iso
static const unsigned char win8662iso[]
Definition: cyrillic_and_mic.c:286
iso_to_win866
Datum iso_to_win866(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:595
win12512iso
static const unsigned char win12512iso[]
Definition: cyrillic_and_mic.c:246
koi2iso
static const unsigned char koi2iso[]
Definition: cyrillic_and_mic.c:86
PG_MODULE_MAGIC
PG_MODULE_MAGIC
Definition: cyrillic_and_mic.c:18
win866_to_iso
Datum win866_to_iso(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:611
win12512koi
static const unsigned char win12512koi[]
Definition: cyrillic_and_mic.c:106
iso2win1251
static const unsigned char iso2win1251[]
Definition: cyrillic_and_mic.c:226
win866_to_mic
Datum win866_to_mic(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:403
mic_to_win1251
Datum mic_to_win1251(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:387
iso_to_mic
Datum iso_to_mic(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:339
koi8r_to_iso
Datum koi8r_to_iso(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:547
win8662win1251
static const unsigned char win8662win1251[]
Definition: cyrillic_and_mic.c:186
koi2win1251
static const unsigned char koi2win1251[]
Definition: cyrillic_and_mic.c:126
koi2win866
static const unsigned char koi2win866[]
Definition: cyrillic_and_mic.c:166
win866_to_koi8r
Datum win866_to_koi8r(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:483
win1251_to_iso
Datum win1251_to_iso(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:579
PG_FUNCTION_INFO_V1
PG_FUNCTION_INFO_V1(koi8r_to_mic)
mic_to_iso
Datum mic_to_iso(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:355
iso_to_koi8r
Datum iso_to_koi8r(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:531
koi8r_to_mic
Datum koi8r_to_mic(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:307
mic_to_win866
Datum mic_to_win866(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:419
iso2koi
static const unsigned char iso2koi[]
Definition: cyrillic_and_mic.c:66
win1251_to_koi8r
Datum win1251_to_koi8r(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:451
iso_to_win1251
Datum iso_to_win1251(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:563
mic_to_koi8r
Datum mic_to_koi8r(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:323
koi8r_to_win866
Datum koi8r_to_win866(PG_FUNCTION_ARGS)
Definition: cyrillic_and_mic.c:467
PG_GETARG_CSTRING
#define PG_GETARG_CSTRING(n)
Definition: fmgr.h:277
PG_RETURN_INT32
#define PG_RETURN_INT32(x)
Definition: fmgr.h:354
PG_GETARG_INT32
#define PG_GETARG_INT32(n)
Definition: fmgr.h:269
PG_GETARG_BOOL
#define PG_GETARG_BOOL(n)
Definition: fmgr.h:274
PG_FUNCTION_ARGS
#define PG_FUNCTION_ARGS
Definition: fmgr.h:193
generate_unaccent_rules.dest
dest
Definition: generate_unaccent_rules.py:280
len
const void size_t len
Definition: pg_crc32c_sse42.c:24
pg_wchar.h
LC_KOI8_R
#define LC_KOI8_R
Definition: pg_wchar.h:119
PG_KOI8R
@ PG_KOI8R
Definition: pg_wchar.h:248
PG_MULE_INTERNAL
@ PG_MULE_INTERNAL
Definition: pg_wchar.h:233
PG_ISO_8859_5
@ PG_ISO_8859_5
Definition: pg_wchar.h:251
PG_WIN1251
@ PG_WIN1251
Definition: pg_wchar.h:249
PG_WIN866
@ PG_WIN866
Definition: pg_wchar.h:246
CHECK_ENCODING_CONVERSION_ARGS
#define CHECK_ENCODING_CONVERSION_ARGS(srcencoding, destencoding)
Definition: pg_wchar.h:507
Datum
uintptr_t Datum
Definition: postgres.h:64