libraries/iconv_wrapper.lib.php

   1 <?php
   2 /* vim: set expandtab sw=4 ts=4 sts=4: */
   3 /**
   4  *
   5  * @version $Id$
   6  */
   7
   8 /**
   9  *
  10  */
  11 # GNU iconv code set to IBM AIX libiconv code set table
  12 # Keys of this table should be in lowercase, and searches should be performed using lowercase!
  13 $gnu_iconv_to_aix_iconv_codepage_map = array (
  14     // "iso-8859-[1-9]" --> "ISO8859-[1-9]" according to http://publibn.boulder.ibm.com/doc_link/en_US/a_doc_lib/libs/basetrf2/setlocale.htm
  15     'iso-8859-1' => 'ISO8859-1',
  16     'iso-8859-2' => 'ISO8859-2',
  17     'iso-8859-3' => 'ISO8859-3',
  18     'iso-8859-4' => 'ISO8859-4',
  19     'iso-8859-5' => 'ISO8859-5',
  20     'iso-8859-6' => 'ISO8859-6',
  21     'iso-8859-7' => 'ISO8859-7',
  22     'iso-8859-8' => 'ISO8859-8',
  23     'iso-8859-9' => 'ISO8859-9',
  24
  25     // "big5" --> "IBM-eucTW" according to http://kadesh.cepba.upc.es/mancpp/classref/ref/ITranscoder_DSC.htm
  26     'big5' => 'IBM-eucTW',
  27
  28     // Other mappings corresponding to the phpMyAdmin dropdown box when using the AllowAnywhereRecoding feature
  29     'euc-jp' => 'IBM-eucJP',
  30     'koi8-r' => 'IBM-eucKR',
  31     'ks_c_5601-1987' => 'KSC5601.1987-0',
  32     'tis-620' => 'TIS-620',
  33     'utf-8' => 'UTF-8'
  34 );
  35
  36 /**
  37  * Wrapper around IBM AIX iconv(), whose character set naming differs
  38  * from the GNU version of iconv().
  39  *
  40  * @param   string   input character set
  41  * @param   string   output character set
  42  * @param   string   the string to convert
  43  *
  44  * @return  mixed    converted string or FALSE on failure
  45  *
  46  * @access  public
  47  *
  48  * @author  bwiberg  Björn Wiberg <Bjorn.Wiberg@its.uu.se>
  49  */
  50 function PMA_aix_iconv_wrapper($in_charset, $out_charset, $str) {
  51
  52     global $gnu_iconv_to_aix_iconv_codepage_map;
  53
  54     // Check for transliteration argument at the end of output character set name
  55     $translit_search = strpos(strtolower($out_charset), '//translit');
  56     $using_translit = (!($translit_search === FALSE));
  57
  58     // Extract "plain" output character set name (without any transliteration argument)
  59     $out_charset_plain = ($using_translit ? substr($out_charset, 0, $translit_search) : $out_charset);
  60
  61     // Transform name of input character set (if found)
  62     if (array_key_exists(strtolower($in_charset), $gnu_iconv_to_aix_iconv_codepage_map)) {
  63         $in_charset = $gnu_iconv_to_aix_iconv_codepage_map[strtolower($in_charset)];
  64     }
  65
  66     // Transform name of "plain" output character set (if found)
  67     if (array_key_exists(strtolower($out_charset_plain), $gnu_iconv_to_aix_iconv_codepage_map)) {
  68         $out_charset_plain = $gnu_iconv_to_aix_iconv_codepage_map[strtolower($out_charset_plain)];
  69     }
  70
  71     // Add transliteration argument again (exactly as specified by user) if used
  72     // Build the output character set name that we will use
  73     $out_charset = ($using_translit ? $out_charset_plain . substr($out_charset, $translit_search) : $out_charset_plain);
  74
  75     // NOTE: Transliteration not supported; we will use the "plain" output character set name
  76     $out_charset = $out_charset_plain;
  77
  78     // Call iconv() with the possibly modified parameters
  79     $result = iconv($in_charset, $out_charset, $str);
  80     return $result;
  81 } //  end of the "PMA_aix_iconv_wrapper()" function
  82
  83 ?>