Line data    Source code

       1             : /*-------------------------------------------------------------------------
       2             :  *
       3             :  *    UTF8 and Cyrillic
       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_cyrillic/utf8_and_cyrillic.c
      10             :  *
      11             :  *-------------------------------------------------------------------------
      12             :  */
      13             : 
      14             : #include "postgres.h"
      15             : #include "fmgr.h"
      16             : #include "mb/pg_wchar.h"
      17             : #include "../../Unicode/utf8_to_koi8r.map"
      18             : #include "../../Unicode/koi8r_to_utf8.map"
      19             : #include "../../Unicode/utf8_to_koi8u.map"
      20             : #include "../../Unicode/koi8u_to_utf8.map"
      21             : 
      22          12 : PG_MODULE_MAGIC_EXT(
      23             :                     .name = "utf8_and_cyrillic",
      24             :                     .version = PG_VERSION
      25             : );
      26             : 
      27          12 : PG_FUNCTION_INFO_V1(utf8_to_koi8r);
      28           6 : PG_FUNCTION_INFO_V1(koi8r_to_utf8);
      29             : 
      30           6 : PG_FUNCTION_INFO_V1(utf8_to_koi8u);
      31           6 : PG_FUNCTION_INFO_V1(koi8u_to_utf8);
      32             : 
      33             : /* ----------
      34             :  * conv_proc(
      35             :  *      INTEGER,    -- source encoding id
      36             :  *      INTEGER,    -- destination encoding id
      37             :  *      CSTRING,    -- source string (null terminated C string)
      38             :  *      CSTRING,    -- destination string (null terminated C string)
      39             :  *      INTEGER,    -- source string length
      40             :  *      BOOL        -- if true, don't throw an error if conversion fails
      41             :  * ) returns INTEGER;
      42             :  *
      43             :  * Returns the number of bytes successfully converted.
      44             :  * ----------
      45             :  */
      46             : 
      47             : Datum
      48         438 : utf8_to_koi8r(PG_FUNCTION_ARGS)
      49             : {
      50         438 :     unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
      51         438 :     unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
      52         438 :     int         len = PG_GETARG_INT32(4);
      53         438 :     bool        noError = PG_GETARG_BOOL(5);
      54             :     int         converted;
      55             : 
      56         438 :     CHECK_ENCODING_CONVERSION_ARGS(PG_UTF8, PG_KOI8R);
      57             : 
      58         438 :     converted = UtfToLocal(src, len, dest,
      59             :                            &koi8r_from_unicode_tree,
      60             :                            NULL, 0,
      61             :                            NULL,
      62             :                            PG_KOI8R,
      63             :                            noError);
      64             : 
      65         240 :     PG_RETURN_INT32(converted);
      66             : }
      67             : 
      68             : Datum
      69           6 : koi8r_to_utf8(PG_FUNCTION_ARGS)
      70             : {
      71           6 :     unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
      72           6 :     unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
      73           6 :     int         len = PG_GETARG_INT32(4);
      74           6 :     bool        noError = PG_GETARG_BOOL(5);
      75             :     int         converted;
      76             : 
      77           6 :     CHECK_ENCODING_CONVERSION_ARGS(PG_KOI8R, PG_UTF8);
      78             : 
      79           6 :     converted = LocalToUtf(src, len, dest,
      80             :                            &koi8r_to_unicode_tree,
      81             :                            NULL, 0,
      82             :                            NULL,
      83             :                            PG_KOI8R,
      84             :                            noError);
      85             : 
      86           6 :     PG_RETURN_INT32(converted);
      87             : }
      88             : 
      89             : Datum
      90           6 : utf8_to_koi8u(PG_FUNCTION_ARGS)
      91             : {
      92           6 :     unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
      93           6 :     unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
      94           6 :     int         len = PG_GETARG_INT32(4);
      95           6 :     bool        noError = PG_GETARG_BOOL(5);
      96             :     int         converted;
      97             : 
      98           6 :     CHECK_ENCODING_CONVERSION_ARGS(PG_UTF8, PG_KOI8U);
      99             : 
     100           6 :     converted = UtfToLocal(src, len, dest,
     101             :                            &koi8u_from_unicode_tree,
     102             :                            NULL, 0,
     103             :                            NULL,
     104             :                            PG_KOI8U,
     105             :                            noError);
     106             : 
     107           6 :     PG_RETURN_INT32(converted);
     108             : }
     109             : 
     110             : Datum
     111           6 : koi8u_to_utf8(PG_FUNCTION_ARGS)
     112             : {
     113           6 :     unsigned char *src = (unsigned char *) PG_GETARG_CSTRING(2);
     114           6 :     unsigned char *dest = (unsigned char *) PG_GETARG_CSTRING(3);
     115           6 :     int         len = PG_GETARG_INT32(4);
     116           6 :     bool        noError = PG_GETARG_BOOL(5);
     117             :     int         converted;
     118             : 
     119           6 :     CHECK_ENCODING_CONVERSION_ARGS(PG_KOI8U, PG_UTF8);
     120             : 
     121           6 :     converted = LocalToUtf(src, len, dest,
     122             :                            &koi8u_to_unicode_tree,
     123             :                            NULL, 0,
     124             :                            NULL,
     125             :                            PG_KOI8U,
     126             :                            noError);
     127             : 
     128           6 :     PG_RETURN_INT32(converted);
     129             : }