src/backend/utils/mb/conversion_procs/utf8_and_iso8859/utf8_and_iso8859.c

   1 /*-------------------------------------------------------------------------
   2  *
   3  *        ISO 8859 2-16 <--> UTF8
   4  *
   5  * Portions Copyright (c) 1996-2022, 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 PG_MODULE_MAGIC;
  45
  46 PG_FUNCTION_INFO_V1(iso8859_to_utf8);
  47 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  *              BOOL            -- if true, don't throw an error if conversion fails
  57  * ) returns INTEGER;
  58  *
  59  * Returns the number of bytes successfully converted.
  60  * ----------
  61  */
  62
  63 typedef struct
  64 {
  65         pg_enc          encoding;
  66         const pg_mb_radix_tree *map1;   /* to UTF8 map name */
  67         const pg_mb_radix_tree *map2;   /* from UTF8 map name */
  68 } pg_conv_map;
  69
  70 static const pg_conv_map maps[] = {
  71         {PG_LATIN2, &iso8859_2_to_unicode_tree,
  72         &iso8859_2_from_unicode_tree},  /* ISO-8859-2 Latin 2 */
  73         {PG_LATIN3, &iso8859_3_to_unicode_tree,
  74         &iso8859_3_from_unicode_tree},  /* ISO-8859-3 Latin 3 */
  75         {PG_LATIN4, &iso8859_4_to_unicode_tree,
  76         &iso8859_4_from_unicode_tree},  /* ISO-8859-4 Latin 4 */
  77         {PG_LATIN5, &iso8859_9_to_unicode_tree,
  78         &iso8859_9_from_unicode_tree},  /* ISO-8859-9 Latin 5 */
  79         {PG_LATIN6, &iso8859_10_to_unicode_tree,
  80         &iso8859_10_from_unicode_tree}, /* ISO-8859-10 Latin 6 */
  81         {PG_LATIN7, &iso8859_13_to_unicode_tree,
  82         &iso8859_13_from_unicode_tree}, /* ISO-8859-13 Latin 7 */
  83         {PG_LATIN8, &iso8859_14_to_unicode_tree,
  84         &iso8859_14_from_unicode_tree}, /* ISO-8859-14 Latin 8 */
  85         {PG_LATIN9, &iso8859_15_to_unicode_tree,
  86         &iso8859_15_from_unicode_tree}, /* ISO-8859-15 Latin 9 */
  87         {PG_LATIN10, &iso8859_16_to_unicode_tree,
  88         &iso8859_16_from_unicode_tree}, /* ISO-8859-16 Latin 10 */
  89         {PG_ISO_8859_5, &iso8859_5_to_unicode_tree,
  90         &iso8859_5_from_unicode_tree},  /* ISO-8859-5 */
  91         {PG_ISO_8859_6, &iso8859_6_to_unicode_tree,
  92         &iso8859_6_from_unicode_tree},  /* ISO-8859-6 */
  93         {PG_ISO_8859_7, &iso8859_7_to_unicode_tree,
  94         &iso8859_7_from_unicode_tree},  /* ISO-8859-7 */
  95         {PG_ISO_8859_8, &iso8859_8_to_unicode_tree,
  96         &iso8859_8_from_unicode_tree},  /* ISO-8859-8 */
  97 };
  98
  99 Datum
 100 iso8859_to_utf8(PG_FUNCTION_ARGS)
 101 {
 102         int                     encoding = PG_GETARG_INT32(0);
 103         unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
 104         unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
 105         int                     len = PG_GETARG_INT32(4);
 106         bool            noError = PG_GETARG_BOOL(5);
 107         int                     i;
 108
 109         CHECK_ENCODING_CONVERSION_ARGS(-1, PG_UTF8);
 110
 111         for (i = 0; i < lengthof(maps); i++)
 112         {
 113                 if (encoding == maps[i].encoding)
 114                 {
 115                         int                     converted;
 116
 117                         converted = LocalToUtf(src, len, dest,
 118                                                                    maps[i].map1,
 119                                                                    NULL, 0,
 120                                                                    NULL,
 121                                                                    encoding,
 122                                                                    noError);
 123                         PG_RETURN_INT32(converted);
 124                 }
 125         }
 126
 127         ereport(ERROR,
 128                         (errcode(ERRCODE_INTERNAL_ERROR),
 129                          errmsg("unexpected encoding ID %d for ISO 8859 character sets",
 130                                         encoding)));
 131
 132         PG_RETURN_INT32(0);
 133 }
 134
 135 Datum
 136 utf8_to_iso8859(PG_FUNCTION_ARGS)
 137 {
 138         int                     encoding = PG_GETARG_INT32(1);
 139         unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
 140         unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
 141         int                     len = PG_GETARG_INT32(4);
 142         bool            noError = PG_GETARG_BOOL(5);
 143         int                     i;
 144
 145         CHECK_ENCODING_CONVERSION_ARGS(PG_UTF8, -1);
 146
 147         for (i = 0; i < lengthof(maps); i++)
 148         {
 149                 if (encoding == maps[i].encoding)
 150                 {
 151                         int                     converted;
 152
 153                         converted = UtfToLocal(src, len, dest,
 154                                                                    maps[i].map2,
 155                                                                    NULL, 0,
 156                                                                    NULL,
 157                                                                    encoding,
 158                                                                    noError);
 159                         PG_RETURN_INT32(converted);
 160                 }
 161         }
 162
 163         ereport(ERROR,
 164                         (errcode(ERRCODE_INTERNAL_ERROR),
 165                          errmsg("unexpected encoding ID %d for ISO 8859 character sets",
 166                                         encoding)));
 167
 168         PG_RETURN_INT32(0);
 169 }