PostgreSQL Source Code  git master
latin2_and_win1250.c
Go to the documentation of this file.
1 /*-------------------------------------------------------------------------
2  *
3  * LATIN2 and WIN1250
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/latin2_and_win1250/latin2_and_win1250.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(latin2_to_mic);
21 PG_FUNCTION_INFO_V1(mic_to_latin2);
22 PG_FUNCTION_INFO_V1(win1250_to_mic);
23 PG_FUNCTION_INFO_V1(mic_to_win1250);
24 PG_FUNCTION_INFO_V1(latin2_to_win1250);
25 PG_FUNCTION_INFO_V1(win1250_to_latin2);
26 
27 /* ----------
28  * conv_proc(
29  * INTEGER, -- source encoding id
30  * INTEGER, -- destination encoding id
31  * CSTRING, -- source string (null terminated C string)
32  * CSTRING, -- destination string (null terminated C string)
33  * INTEGER, -- source string length
34  * BOOL -- if true, don't throw an error if conversion fails
35  * ) returns INTEGER;
36  *
37  * Returns the number of bytes successfully converted.
38  * ----------
39  */
40 
41 /* WIN1250 to ISO-8859-2 */
42 static const unsigned char win1250_2_iso88592[] = {
43  0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87,
44  0x88, 0x89, 0xA9, 0x8B, 0xA6, 0xAB, 0xAE, 0xAC,
45  0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97,
46  0x98, 0x99, 0xB9, 0x9B, 0xB6, 0xBB, 0xBE, 0xBC,
47  0xA0, 0xB7, 0xA2, 0xA3, 0xA4, 0xA1, 0x00, 0xA7,
48  0xA8, 0x00, 0xAA, 0x00, 0x00, 0xAD, 0x00, 0xAF,
49  0xB0, 0x00, 0xB2, 0xB3, 0xB4, 0x00, 0x00, 0x00,
50  0xB8, 0xB1, 0xBA, 0x00, 0xA5, 0xBD, 0xB5, 0xBF,
51  0xC0, 0xC1, 0xC2, 0xC3, 0xC4, 0xC5, 0xC6, 0xC7,
52  0xC8, 0xC9, 0xCA, 0xCB, 0xCC, 0xCD, 0xCE, 0xCF,
53  0xD0, 0xD1, 0xD2, 0xD3, 0xD4, 0xD5, 0xD6, 0xD7,
54  0xD8, 0xD9, 0xDA, 0xDB, 0xDC, 0xDD, 0xDE, 0xDF,
55  0xE0, 0xE1, 0xE2, 0xE3, 0xE4, 0xE5, 0xE6, 0xE7,
56  0xE8, 0xE9, 0xEA, 0xEB, 0xEC, 0xED, 0xEE, 0xEF,
57  0xF0, 0xF1, 0xF2, 0xF3, 0xF4, 0xF5, 0xF6, 0xF7,
58  0xF8, 0xF9, 0xFA, 0xFB, 0xFC, 0xFD, 0xFE, 0xFF
59 };
60 
61 /* ISO-8859-2 to WIN1250 */
62 static const unsigned char iso88592_2_win1250[] = {
63  0x80, 0x81, 0x82, 0x83, 0x84, 0x85, 0x86, 0x87,
64  0x88, 0x89, 0x00, 0x8B, 0x00, 0x00, 0x00, 0x00,
65  0x90, 0x91, 0x92, 0x93, 0x94, 0x95, 0x96, 0x97,
66  0x98, 0x99, 0x00, 0x9B, 0x00, 0x00, 0x00, 0x00,
67  0xA0, 0xA5, 0xA2, 0xA3, 0xA4, 0xBC, 0x8C, 0xA7,
68  0xA8, 0x8A, 0xAA, 0x8D, 0x8F, 0xAD, 0x8E, 0xAF,
69  0xB0, 0xB9, 0xB2, 0xB3, 0xB4, 0xBE, 0x9C, 0xA1,
70  0xB8, 0x9A, 0xBA, 0x9D, 0x9F, 0xBD, 0x9E, 0xBF,
71  0xC0, 0xC1, 0xC2, 0xC3, 0xC4, 0xC5, 0xC6, 0xC7,
72  0xC8, 0xC9, 0xCA, 0xCB, 0xCC, 0xCD, 0xCE, 0xCF,
73  0xD0, 0xD1, 0xD2, 0xD3, 0xD4, 0xD5, 0xD6, 0xD7,
74  0xD8, 0xD9, 0xDA, 0xDB, 0xDC, 0xDD, 0xDE, 0xDF,
75  0xE0, 0xE1, 0xE2, 0xE3, 0xE4, 0xE5, 0xE6, 0xE7,
76  0xE8, 0xE9, 0xEA, 0xEB, 0xEC, 0xED, 0xEE, 0xEF,
77  0xF0, 0xF1, 0xF2, 0xF3, 0xF4, 0xF5, 0xF6, 0xF7,
78  0xF8, 0xF9, 0xFA, 0xFB, 0xFC, 0xFD, 0xFE, 0xFF
79 };
80 
81 
82 Datum
83 latin2_to_mic(PG_FUNCTION_ARGS)
84 {
85  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
86  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
87  int len = PG_GETARG_INT32(4);
88  bool noError = PG_GETARG_BOOL(5);
89  int converted;
90 
91  CHECK_ENCODING_CONVERSION_ARGS(PG_LATIN2, PG_MULE_INTERNAL);
92 
93  converted = latin2mic(src, dest, len, LC_ISO8859_2, PG_LATIN2, noError);
94 
95  PG_RETURN_INT32(converted);
96 }
97 
98 Datum
99 mic_to_latin2(PG_FUNCTION_ARGS)
100 {
101  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
102  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
103  int len = PG_GETARG_INT32(4);
104  bool noError = PG_GETARG_BOOL(5);
105  int converted;
106 
107  CHECK_ENCODING_CONVERSION_ARGS(PG_MULE_INTERNAL, PG_LATIN2);
108 
109  converted = mic2latin(src, dest, len, LC_ISO8859_2, PG_LATIN2, noError);
110 
111  PG_RETURN_INT32(converted);
112 }
113 
114 Datum
115 win1250_to_mic(PG_FUNCTION_ARGS)
116 {
117  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
118  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
119  int len = PG_GETARG_INT32(4);
120  bool noError = PG_GETARG_BOOL(5);
121  int converted;
122 
123  CHECK_ENCODING_CONVERSION_ARGS(PG_WIN1250, PG_MULE_INTERNAL);
124 
125  converted = latin2mic_with_table(src, dest, len, LC_ISO8859_2, PG_WIN1250,
126  win1250_2_iso88592, noError);
127 
128  PG_RETURN_INT32(converted);
129 }
130 
131 Datum
132 mic_to_win1250(PG_FUNCTION_ARGS)
133 {
134  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
135  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
136  int len = PG_GETARG_INT32(4);
137  bool noError = PG_GETARG_BOOL(5);
138  int converted;
139 
140  CHECK_ENCODING_CONVERSION_ARGS(PG_MULE_INTERNAL, PG_WIN1250);
141 
142  converted = mic2latin_with_table(src, dest, len, LC_ISO8859_2, PG_WIN1250,
143  iso88592_2_win1250, noError);
144 
145  PG_RETURN_INT32(converted);
146 }
147 
148 Datum
149 latin2_to_win1250(PG_FUNCTION_ARGS)
150 {
151  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
152  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
153  int len = PG_GETARG_INT32(4);
154  bool noError = PG_GETARG_BOOL(5);
155  int converted;
156 
157  CHECK_ENCODING_CONVERSION_ARGS(PG_LATIN2, PG_WIN1250);
158 
159  converted = local2local(src, dest, len, PG_LATIN2, PG_WIN1250,
160  iso88592_2_win1250, noError);
161 
162  PG_RETURN_INT32(converted);
163 }
164 
165 Datum
166 win1250_to_latin2(PG_FUNCTION_ARGS)
167 {
168  unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
169  unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
170  int len = PG_GETARG_INT32(4);
171  bool noError = PG_GETARG_BOOL(5);
172  int converted;
173 
174  CHECK_ENCODING_CONVERSION_ARGS(PG_WIN1250, PG_LATIN2);
175 
176  converted = local2local(src, dest, len, PG_WIN1250, PG_LATIN2,
177  win1250_2_iso88592, noError);
178 
179  PG_RETURN_INT32(converted);
180 }
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
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
mic_to_latin2
Datum mic_to_latin2(PG_FUNCTION_ARGS)
Definition: latin2_and_win1250.c:99
win1250_2_iso88592
static const unsigned char win1250_2_iso88592[]
Definition: latin2_and_win1250.c:42
iso88592_2_win1250
static const unsigned char iso88592_2_win1250[]
Definition: latin2_and_win1250.c:62
latin2_to_win1250
Datum latin2_to_win1250(PG_FUNCTION_ARGS)
Definition: latin2_and_win1250.c:149
PG_MODULE_MAGIC
PG_MODULE_MAGIC
Definition: latin2_and_win1250.c:18
mic_to_win1250
Datum mic_to_win1250(PG_FUNCTION_ARGS)
Definition: latin2_and_win1250.c:132
latin2_to_mic
Datum latin2_to_mic(PG_FUNCTION_ARGS)
Definition: latin2_and_win1250.c:83
win1250_to_mic
Datum win1250_to_mic(PG_FUNCTION_ARGS)
Definition: latin2_and_win1250.c:115
win1250_to_latin2
Datum win1250_to_latin2(PG_FUNCTION_ARGS)
Definition: latin2_and_win1250.c:166
PG_FUNCTION_INFO_V1
PG_FUNCTION_INFO_V1(latin2_to_mic)
generate_unaccent_rules.dest
dest
Definition: generate_unaccent_rules.py:285
len
const void size_t len
Definition: pg_crc32c_sse42.c:24
pg_wchar.h
PG_MULE_INTERNAL
@ PG_MULE_INTERNAL
Definition: pg_wchar.h:233
PG_LATIN2
@ PG_LATIN2
Definition: pg_wchar.h:235
PG_WIN1250
@ PG_WIN1250
Definition: pg_wchar.h:255
LC_ISO8859_2
#define LC_ISO8859_2
Definition: pg_wchar.h:106
CHECK_ENCODING_CONVERSION_ARGS
#define CHECK_ENCODING_CONVERSION_ARGS(srcencoding, destencoding)
Definition: pg_wchar.h:507
Datum
uintptr_t Datum
Definition: postgres.h:64