Line data Source code
1 : /*-------------------------------------------------------------------------
2 : *
3 : * WIN <--> UTF8
4 : *
5 : * Portions Copyright (c) 1996-2025, 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/utf8_and_win/utf8_and_win.c
10 : *
11 : *-------------------------------------------------------------------------
12 : */
13 :
14 : #include "postgres.h"
15 : #include "fmgr.h"
16 : #include "mb/pg_wchar.h"
17 : #include "../../Unicode/utf8_to_win1250.map"
18 : #include "../../Unicode/utf8_to_win1251.map"
19 : #include "../../Unicode/utf8_to_win1252.map"
20 : #include "../../Unicode/utf8_to_win1253.map"
21 : #include "../../Unicode/utf8_to_win1254.map"
22 : #include "../../Unicode/utf8_to_win1255.map"
23 : #include "../../Unicode/utf8_to_win1256.map"
24 : #include "../../Unicode/utf8_to_win1257.map"
25 : #include "../../Unicode/utf8_to_win1258.map"
26 : #include "../../Unicode/utf8_to_win866.map"
27 : #include "../../Unicode/utf8_to_win874.map"
28 : #include "../../Unicode/win1250_to_utf8.map"
29 : #include "../../Unicode/win1251_to_utf8.map"
30 : #include "../../Unicode/win1252_to_utf8.map"
31 : #include "../../Unicode/win1253_to_utf8.map"
32 : #include "../../Unicode/win1254_to_utf8.map"
33 : #include "../../Unicode/win1255_to_utf8.map"
34 : #include "../../Unicode/win1256_to_utf8.map"
35 : #include "../../Unicode/win1257_to_utf8.map"
36 : #include "../../Unicode/win866_to_utf8.map"
37 : #include "../../Unicode/win874_to_utf8.map"
38 : #include "../../Unicode/win1258_to_utf8.map"
39 :
40 6 : PG_MODULE_MAGIC_EXT(
41 : .name = "utf8_and_win",
42 : .version = PG_VERSION
43 : );
44 :
45 6 : PG_FUNCTION_INFO_V1(win_to_utf8);
46 6 : PG_FUNCTION_INFO_V1(utf8_to_win);
47 :
48 : /* ----------
49 : * conv_proc(
50 : * INTEGER, -- source encoding id
51 : * INTEGER, -- destination encoding id
52 : * CSTRING, -- source string (null terminated C string)
53 : * CSTRING, -- destination string (null terminated C string)
54 : * INTEGER, -- source string length
55 : * BOOL -- if true, don't throw an error if conversion fails
56 : * ) returns INTEGER;
57 : *
58 : * Returns the number of bytes successfully converted.
59 : * ----------
60 : */
61 :
62 : typedef struct
63 : {
64 : pg_enc encoding;
65 : const pg_mb_radix_tree *map1; /* to UTF8 map name */
66 : const pg_mb_radix_tree *map2; /* from UTF8 map name */
67 : } pg_conv_map;
68 :
69 : static const pg_conv_map maps[] = {
70 : {PG_WIN866, &win866_to_unicode_tree, &win866_from_unicode_tree},
71 : {PG_WIN874, &win874_to_unicode_tree, &win874_from_unicode_tree},
72 : {PG_WIN1250, &win1250_to_unicode_tree, &win1250_from_unicode_tree},
73 : {PG_WIN1251, &win1251_to_unicode_tree, &win1251_from_unicode_tree},
74 : {PG_WIN1252, &win1252_to_unicode_tree, &win1252_from_unicode_tree},
75 : {PG_WIN1253, &win1253_to_unicode_tree, &win1253_from_unicode_tree},
76 : {PG_WIN1254, &win1254_to_unicode_tree, &win1254_from_unicode_tree},
77 : {PG_WIN1255, &win1255_to_unicode_tree, &win1255_from_unicode_tree},
78 : {PG_WIN1256, &win1256_to_unicode_tree, &win1256_from_unicode_tree},
79 : {PG_WIN1257, &win1257_to_unicode_tree, &win1257_from_unicode_tree},
80 : {PG_WIN1258, &win1258_to_unicode_tree, &win1258_from_unicode_tree},
81 : };
82 :
83 : Datum
84 66 : win_to_utf8(PG_FUNCTION_ARGS)
85 : {
86 66 : int encoding = PG_GETARG_INT32(0);
87 66 : unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
88 66 : unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
89 66 : int len = PG_GETARG_INT32(4);
90 66 : bool noError = PG_GETARG_BOOL(5);
91 : int i;
92 :
93 66 : CHECK_ENCODING_CONVERSION_ARGS(-1, PG_UTF8);
94 :
95 396 : for (i = 0; i < lengthof(maps); i++)
96 : {
97 396 : if (encoding == maps[i].encoding)
98 : {
99 : int converted;
100 :
101 66 : converted = LocalToUtf(src, len, dest,
102 : maps[i].map1,
103 : NULL, 0,
104 : NULL,
105 : encoding,
106 : noError);
107 66 : PG_RETURN_INT32(converted);
108 : }
109 : }
110 :
111 0 : ereport(ERROR,
112 : (errcode(ERRCODE_INTERNAL_ERROR),
113 : errmsg("unexpected encoding ID %d for WIN character sets",
114 : encoding)));
115 :
116 : PG_RETURN_INT32(0);
117 : }
118 :
119 : Datum
120 66 : utf8_to_win(PG_FUNCTION_ARGS)
121 : {
122 66 : int encoding = PG_GETARG_INT32(1);
123 66 : unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
124 66 : unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
125 66 : int len = PG_GETARG_INT32(4);
126 66 : bool noError = PG_GETARG_BOOL(5);
127 : int i;
128 :
129 66 : CHECK_ENCODING_CONVERSION_ARGS(PG_UTF8, -1);
130 :
131 396 : for (i = 0; i < lengthof(maps); i++)
132 : {
133 396 : if (encoding == maps[i].encoding)
134 : {
135 : int converted;
136 :
137 66 : converted = UtfToLocal(src, len, dest,
138 : maps[i].map2,
139 : NULL, 0,
140 : NULL,
141 : encoding,
142 : noError);
143 66 : PG_RETURN_INT32(converted);
144 : }
145 : }
146 :
147 0 : ereport(ERROR,
148 : (errcode(ERRCODE_INTERNAL_ERROR),
149 : errmsg("unexpected encoding ID %d for WIN character sets",
150 : encoding)));
151 :
152 : PG_RETURN_INT32(0);
153 : }
|