Line data Source code
1 : /*-------------------------------------------------------------------------
2 : *
3 : * ISO 8859 2-16 <--> UTF8
4 : *
5 : * Portions Copyright (c) 1996-2017, 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_iso8859/utf8_and_iso8859.c
10 : *
11 : *-------------------------------------------------------------------------
12 : */
13 :
14 : #include "postgres.h"
15 : #include "fmgr.h"
16 : #include "mb/pg_wchar.h"
17 : #include "../../Unicode/iso8859_10_to_utf8.map"
18 : #include "../../Unicode/iso8859_13_to_utf8.map"
19 : #include "../../Unicode/iso8859_14_to_utf8.map"
20 : #include "../../Unicode/iso8859_15_to_utf8.map"
21 : #include "../../Unicode/iso8859_2_to_utf8.map"
22 : #include "../../Unicode/iso8859_3_to_utf8.map"
23 : #include "../../Unicode/iso8859_4_to_utf8.map"
24 : #include "../../Unicode/iso8859_5_to_utf8.map"
25 : #include "../../Unicode/iso8859_6_to_utf8.map"
26 : #include "../../Unicode/iso8859_7_to_utf8.map"
27 : #include "../../Unicode/iso8859_8_to_utf8.map"
28 : #include "../../Unicode/iso8859_9_to_utf8.map"
29 : #include "../../Unicode/utf8_to_iso8859_10.map"
30 : #include "../../Unicode/utf8_to_iso8859_13.map"
31 : #include "../../Unicode/utf8_to_iso8859_14.map"
32 : #include "../../Unicode/utf8_to_iso8859_15.map"
33 : #include "../../Unicode/utf8_to_iso8859_16.map"
34 : #include "../../Unicode/utf8_to_iso8859_2.map"
35 : #include "../../Unicode/utf8_to_iso8859_3.map"
36 : #include "../../Unicode/utf8_to_iso8859_4.map"
37 : #include "../../Unicode/utf8_to_iso8859_5.map"
38 : #include "../../Unicode/utf8_to_iso8859_6.map"
39 : #include "../../Unicode/utf8_to_iso8859_7.map"
40 : #include "../../Unicode/utf8_to_iso8859_8.map"
41 : #include "../../Unicode/utf8_to_iso8859_9.map"
42 : #include "../../Unicode/iso8859_16_to_utf8.map"
43 :
44 4 : PG_MODULE_MAGIC;
45 :
46 54 : PG_FUNCTION_INFO_V1(iso8859_to_utf8);
47 54 : PG_FUNCTION_INFO_V1(utf8_to_iso8859);
48 :
49 : /* ----------
50 : * conv_proc(
51 : * INTEGER, -- source encoding id
52 : * INTEGER, -- destination encoding id
53 : * CSTRING, -- source string (null terminated C string)
54 : * CSTRING, -- destination string (null terminated C string)
55 : * INTEGER -- source string length
56 : * ) returns VOID;
57 : * ----------
58 : */
59 :
60 : typedef struct
61 : {
62 : pg_enc encoding;
63 : const pg_mb_radix_tree *map1; /* to UTF8 map name */
64 : const pg_mb_radix_tree *map2; /* from UTF8 map name */
65 : } pg_conv_map;
66 :
67 : static const pg_conv_map maps[] = {
68 : {PG_LATIN2, &iso8859_2_to_unicode_tree,
69 : &iso8859_2_from_unicode_tree}, /* ISO-8859-2 Latin 2 */
70 : {PG_LATIN3, &iso8859_3_to_unicode_tree,
71 : &iso8859_3_from_unicode_tree}, /* ISO-8859-3 Latin 3 */
72 : {PG_LATIN4, &iso8859_4_to_unicode_tree,
73 : &iso8859_4_from_unicode_tree}, /* ISO-8859-4 Latin 4 */
74 : {PG_LATIN5, &iso8859_9_to_unicode_tree,
75 : &iso8859_9_from_unicode_tree}, /* ISO-8859-9 Latin 5 */
76 : {PG_LATIN6, &iso8859_10_to_unicode_tree,
77 : &iso8859_10_from_unicode_tree}, /* ISO-8859-10 Latin 6 */
78 : {PG_LATIN7, &iso8859_13_to_unicode_tree,
79 : &iso8859_13_from_unicode_tree}, /* ISO-8859-13 Latin 7 */
80 : {PG_LATIN8, &iso8859_14_to_unicode_tree,
81 : &iso8859_14_from_unicode_tree}, /* ISO-8859-14 Latin 8 */
82 : {PG_LATIN9, &iso8859_15_to_unicode_tree,
83 : &iso8859_15_from_unicode_tree}, /* ISO-8859-15 Latin 9 */
84 : {PG_LATIN10, &iso8859_16_to_unicode_tree,
85 : &iso8859_16_from_unicode_tree}, /* ISO-8859-16 Latin 10 */
86 : {PG_ISO_8859_5, &iso8859_5_to_unicode_tree,
87 : &iso8859_5_from_unicode_tree}, /* ISO-8859-5 */
88 : {PG_ISO_8859_6, &iso8859_6_to_unicode_tree,
89 : &iso8859_6_from_unicode_tree}, /* ISO-8859-6 */
90 : {PG_ISO_8859_7, &iso8859_7_to_unicode_tree,
91 : &iso8859_7_from_unicode_tree}, /* ISO-8859-7 */
92 : {PG_ISO_8859_8, &iso8859_8_to_unicode_tree,
93 : &iso8859_8_from_unicode_tree}, /* ISO-8859-8 */
94 : };
95 :
96 : Datum
97 52 : iso8859_to_utf8(PG_FUNCTION_ARGS)
98 : {
99 52 : int encoding = PG_GETARG_INT32(0);
100 52 : unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
101 52 : unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
102 52 : int len = PG_GETARG_INT32(4);
103 : int i;
104 :
105 52 : CHECK_ENCODING_CONVERSION_ARGS(-1, PG_UTF8);
106 :
107 364 : for (i = 0; i < lengthof(maps); i++)
108 : {
109 364 : if (encoding == maps[i].encoding)
110 : {
111 52 : LocalToUtf(src, len, dest,
112 : maps[i].map1,
113 : NULL, 0,
114 : NULL,
115 : encoding);
116 52 : PG_RETURN_VOID();
117 : }
118 : }
119 :
120 0 : ereport(ERROR,
121 : (errcode(ERRCODE_INTERNAL_ERROR),
122 : errmsg("unexpected encoding ID %d for ISO 8859 character sets",
123 : encoding)));
124 :
125 : PG_RETURN_VOID();
126 : }
127 :
128 : Datum
129 52 : utf8_to_iso8859(PG_FUNCTION_ARGS)
130 : {
131 52 : int encoding = PG_GETARG_INT32(1);
132 52 : unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
133 52 : unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
134 52 : int len = PG_GETARG_INT32(4);
135 : int i;
136 :
137 52 : CHECK_ENCODING_CONVERSION_ARGS(PG_UTF8, -1);
138 :
139 364 : for (i = 0; i < lengthof(maps); i++)
140 : {
141 364 : if (encoding == maps[i].encoding)
142 : {
143 52 : UtfToLocal(src, len, dest,
144 : maps[i].map2,
145 : NULL, 0,
146 : NULL,
147 : encoding);
148 52 : PG_RETURN_VOID();
149 : }
150 : }
151 :
152 0 : ereport(ERROR,
153 : (errcode(ERRCODE_INTERNAL_ERROR),
154 : errmsg("unexpected encoding ID %d for ISO 8859 character sets",
155 : encoding)));
156 :
157 : PG_RETURN_VOID();
158 : }
|