src/backend/utils/mb/conversion_procs/utf8_and_win/utf8_and_win.c

   1 /*-------------------------------------------------------------------------
   2  *
   3  *        WIN <--> UTF8
   4  *
   5  * Portions Copyright (c) 1996-2009, PostgreSQL Global Development Group
   6  * Portions Copyright (c) 1994, Regents of the University of California
   7  *
   8  * IDENTIFICATION
   9  *        $PostgreSQL: pgsql/src/backend/utils/mb/conversion_procs/utf8_and_win/utf8_and_win.c,v 1.13 2009/01/01 17:23:53 momjian Exp $
  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 PG_MODULE_MAGIC;
  41
  42 PG_FUNCTION_INFO_V1(win_to_utf8);
  43 PG_FUNCTION_INFO_V1(utf8_to_win);
  44
  45 extern Datum win_to_utf8(PG_FUNCTION_ARGS);
  46 extern Datum utf8_to_win(PG_FUNCTION_ARGS);
  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  * ) returns VOID;
  56  * ----------
  57  */
  58
  59 typedef struct
  60 {
  61         pg_enc          encoding;
  62         pg_local_to_utf *map1;          /* to UTF8 map name */
  63         pg_utf_to_local *map2;          /* from UTF8 map name */
  64         int                     size1;                  /* size of map1 */
  65         int                     size2;                  /* size of map2 */
  66 } pg_conv_map;
  67
  68 static pg_conv_map maps[] = {
  69         {PG_WIN866, LUmapWIN866, ULmapWIN866,
  70                 sizeof(LUmapWIN866) / sizeof(pg_local_to_utf),
  71         sizeof(ULmapWIN866) / sizeof(pg_utf_to_local)},
  72         {PG_WIN874, LUmapWIN874, ULmapWIN874,
  73                 sizeof(LUmapWIN874) / sizeof(pg_local_to_utf),
  74         sizeof(ULmapWIN874) / sizeof(pg_utf_to_local)},
  75         {PG_WIN1250, LUmapWIN1250, ULmapWIN1250,
  76                 sizeof(LUmapWIN1250) / sizeof(pg_local_to_utf),
  77         sizeof(ULmapWIN1250) / sizeof(pg_utf_to_local)},
  78         {PG_WIN1251, LUmapWIN1251, ULmapWIN1251,
  79                 sizeof(LUmapWIN1251) / sizeof(pg_local_to_utf),
  80         sizeof(ULmapWIN1251) / sizeof(pg_utf_to_local)},
  81         {PG_WIN1252, LUmapWIN1252, ULmapWIN1252,
  82                 sizeof(LUmapWIN1252) / sizeof(pg_local_to_utf),
  83         sizeof(ULmapWIN1252) / sizeof(pg_utf_to_local)},
  84         {PG_WIN1253, LUmapWIN1253, ULmapWIN1253,
  85                 sizeof(LUmapWIN1253) / sizeof(pg_local_to_utf),
  86         sizeof(ULmapWIN1253) / sizeof(pg_utf_to_local)},
  87         {PG_WIN1254, LUmapWIN1254, ULmapWIN1254,
  88                 sizeof(LUmapWIN1254) / sizeof(pg_local_to_utf),
  89         sizeof(ULmapWIN1254) / sizeof(pg_utf_to_local)},
  90         {PG_WIN1255, LUmapWIN1255, ULmapWIN1255,
  91                 sizeof(LUmapWIN1255) / sizeof(pg_local_to_utf),
  92         sizeof(ULmapWIN1255) / sizeof(pg_utf_to_local)},
  93         {PG_WIN1256, LUmapWIN1256, ULmapWIN1256,
  94                 sizeof(LUmapWIN1256) / sizeof(pg_local_to_utf),
  95         sizeof(ULmapWIN1256) / sizeof(pg_utf_to_local)},
  96         {PG_WIN1257, LUmapWIN1257, ULmapWIN1257,
  97                 sizeof(LUmapWIN1257) / sizeof(pg_local_to_utf),
  98         sizeof(ULmapWIN1257) / sizeof(pg_utf_to_local)},
  99         {PG_WIN1258, LUmapWIN1258, ULmapWIN1258,
 100                 sizeof(LUmapWIN1258) / sizeof(pg_local_to_utf),
 101         sizeof(ULmapWIN1258) / sizeof(pg_utf_to_local)},
 102 };
 103
 104 Datum
 105 win_to_utf8(PG_FUNCTION_ARGS)
 106 {
 107         int                     encoding = PG_GETARG_INT32(0);
 108         unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
 109         unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
 110         int                     len = PG_GETARG_INT32(4);
 111         int                     i;
 112
 113         Assert(PG_GETARG_INT32(1) == PG_UTF8);
 114         Assert(len >= 0);
 115
 116         for (i = 0; i < sizeof(maps) / sizeof(pg_conv_map); i++)
 117         {
 118                 if (encoding == maps[i].encoding)
 119                 {
 120                         LocalToUtf(src, dest, maps[i].map1, NULL, maps[i].size1, 0, encoding, len);
 121                         PG_RETURN_VOID();
 122                 }
 123         }
 124
 125         ereport(ERROR,
 126                         (errcode(ERRCODE_INTERNAL_ERROR),
 127           errmsg("unexpected encoding ID %d for WIN character sets", encoding)));
 128
 129         PG_RETURN_VOID();
 130 }
 131
 132 Datum
 133 utf8_to_win(PG_FUNCTION_ARGS)
 134 {
 135         int                     encoding = PG_GETARG_INT32(1);
 136         unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
 137         unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
 138         int                     len = PG_GETARG_INT32(4);
 139         int                     i;
 140
 141         Assert(PG_GETARG_INT32(0) == PG_UTF8);
 142         Assert(len >= 0);
 143
 144         for (i = 0; i < sizeof(maps) / sizeof(pg_conv_map); i++)
 145         {
 146                 if (encoding == maps[i].encoding)
 147                 {
 148                         UtfToLocal(src, dest, maps[i].map2, NULL, maps[i].size2, 0, encoding, len);
 149                         PG_RETURN_VOID();
 150                 }
 151         }
 152
 153         ereport(ERROR,
 154                         (errcode(ERRCODE_INTERNAL_ERROR),
 155           errmsg("unexpected encoding ID %d for WIN character sets", encoding)));
 156
 157         PG_RETURN_VOID();
 158 }