[Summary view] [Print] [Text view]

   1  <?php
   2  
   3  define ("CONVERT_TABLES_DIR", PHPWCMS_ROOT.'/include/inc_ext/ConvertCharset/ConvertTables/');
   4  define ("DEBUG_MODE", 1);
   5  
   6  
   7  class ConvertCharset{
   8       var $RecognizedEncoding; // (boolean) This value keeps information if string contains multibyte chars.
   9       var $Entities; // (boolean) This value keeps information if output should be with numeric entities.
  10       var $FromCharset; // (string) This value keeps information about source (from) encoding
  11       var $ToCharset; // (string) This value keeps information about destination (to) encoding
  12       var $CharsetTable; // (array) This property keeps convert Table inside
  13  
  14  
  15  
  16       function ConvertCharset ($FromCharset, $ToCharset, $TurnOnEntities = false)
  17      {
  18  
  19           $this -> FromCharset = strtolower($FromCharset);
  20           $this -> ToCharset = strtolower($ToCharset);
  21           $this -> Entities = $TurnOnEntities;
  22  
  23  
  24           if ($this -> FromCharset == $this -> ToCharset)
  25          {
  26               print $this -> DebugOutput(1, 0, $this -> FromCharset);
  27               }
  28           if (($this -> FromCharset == $this -> ToCharset) AND ($this -> FromCharset == "utf-8"))
  29              {
  30               print $this -> DebugOutput(0, 4, $this -> FromCharset);
  31               exit;
  32               }
  33  
  34  
  35           if ($this -> FromCharset == "utf-8")
  36          {
  37               $this -> CharsetTable = $this -> MakeConvertTable ($this -> ToCharset);
  38               }
  39          else if ($this -> ToCharset == "utf-8")
  40          {
  41               $this -> CharsetTable = $this -> MakeConvertTable ($this -> FromCharset);
  42               }
  43          else
  44              {
  45               $this -> CharsetTable = $this -> MakeConvertTable ($this -> FromCharset, $this -> ToCharset);
  46               }
  47  
  48           }
  49  
  50  
  51       function UnicodeEntity ($UnicodeString)
  52      {
  53           $OutString = "";
  54           $StringLenght = strlen ($UnicodeString);
  55           for ($CharPosition = 0; $CharPosition < $StringLenght; $CharPosition++)
  56          {
  57               $Char = $UnicodeString [$CharPosition];
  58               $AsciiChar = ord ($Char);
  59  
  60               if ($AsciiChar < 128){
  61                   $OutString .= $Char;
  62                   }
  63              else if ($AsciiChar >> 5 == 6){
  64                   $FirstByte = ($AsciiChar & 31);
  65                   $CharPosition++;
  66                   $Char = $UnicodeString [$CharPosition];
  67                   $AsciiChar = ord ($Char);
  68                   $SecondByte = ($AsciiChar & 63);
  69                   $AsciiChar = ($FirstByte * 64) + $SecondByte;
  70                   $Entity = sprintf ("&#%d;", $AsciiChar);
  71                   $OutString .= $Entity;
  72                   }
  73              else if ($AsciiChar >> 4 == 14){
  74                   $FirstByte = ($AsciiChar & 31);
  75                   $CharPosition++;
  76                   $Char = $UnicodeString [$CharPosition];
  77                   $AsciiChar = ord ($Char);
  78                   $SecondByte = ($AsciiChar & 63);
  79                   $CharPosition++;
  80                   $Char = $UnicodeString [$CharPosition];
  81                   $AsciiChar = ord ($Char);
  82                   $ThidrByte = ($AsciiChar & 63);
  83                   $AsciiChar = ((($FirstByte * 64) + $SecondByte) * 64) + $ThidrByte;
  84  
  85                   $Entity = sprintf ("&#%d;", $AsciiChar);
  86                   $OutString .= $Entity;
  87                   }
  88              else if ($AsciiChar >> 3 == 30){
  89                   $FirstByte = ($AsciiChar & 31);
  90                   $CharPosition++;
  91                   $Char = $UnicodeString [$CharPosition];
  92                   $AsciiChar = ord ($Char);
  93                   $SecondByte = ($AsciiChar & 63);
  94                   $CharPosition++;
  95                   $Char = $UnicodeString [$CharPosition];
  96                   $AsciiChar = ord ($Char);
  97                   $ThidrByte = ($AsciiChar & 63);
  98                   $CharPosition++;
  99                   $Char = $UnicodeString [$CharPosition];
 100                   $AsciiChar = ord ($Char);
 101                   $FourthByte = ($AsciiChar & 63);
 102                   $AsciiChar = ((((($FirstByte * 64) + $SecondByte) * 64) + $ThidrByte) * 64) + $FourthByte;
 103  
 104                   $Entity = sprintf ("&#%d;", $AsciiChar);
 105                   $OutString .= $Entity;
 106                   }
 107               }
 108           return $OutString;
 109           }
 110  
 111  
 112       function HexToUtf ($UtfCharInHex)
 113      {
 114           $OutputChar = "";
 115           $UtfCharInDec = hexdec($UtfCharInHex);
 116           if($UtfCharInDec < 128) $OutputChar .= chr($UtfCharInDec);
 117           else if($UtfCharInDec < 2048)$OutputChar .= chr(($UtfCharInDec >> 6) + 192) . chr(($UtfCharInDec & 63) + 128);
 118           else if($UtfCharInDec < 65536)$OutputChar .= chr(($UtfCharInDec >> 12) + 224) . chr((($UtfCharInDec >> 6) & 63) + 128) . chr(($UtfCharInDec & 63) + 128);
 119           else if($UtfCharInDec < 2097152)$OutputChar .= chr($UtfCharInDec >> 18 + 240) . chr((($UtfCharInDec >> 12) & 63) + 128) . chr(($UtfCharInDec >> 6) & 63 + 128) . chr($UtfCharInDec & 63 + 128);
 120           return $OutputChar;
 121           }
 122  
 123  
 124  
 125       function MakeConvertTable ($FromCharset, $ToCharset = '')
 126      {
 127           $ConvertTable = array();
 128           for($i = 0; $i < func_num_args(); $i++)
 129          {
 130  
 131               $FileName = func_get_arg($i);
 132               if (!is_file(CONVERT_TABLES_DIR . $FileName))
 133                  {
 134                   print $this -> DebugOutput(0, 0, CONVERT_TABLES_DIR . $FileName); //Print an error message
 135                   exit;
 136                   }
 137               $FileWithEncTabe = fopen(CONVERT_TABLES_DIR . $FileName, "r") or die(); //This die(); is just to make sure...
 138               while(!feof($FileWithEncTabe))
 139              {
 140  
 141                   if($OneLine = trim(fgets($FileWithEncTabe, 1024)))
 142                      {
 143  
 144                       if (substr($OneLine, 0, 1) != "#")
 145                          {
 146  
 147                           $HexValue = preg_split ("/[\s,]+/", $OneLine, 3); //We need only first 2 values
 148  
 149                           if (substr($HexValue[1], 0, 1) != "#")
 150                              {
 151                               $ArrayKey = strtoupper(str_replace(strtolower("0x"), "", $HexValue[1]));
 152                               $ArrayValue = strtoupper(str_replace(strtolower("0x"), "", $HexValue[0]));
 153                               $ConvertTable[func_get_arg($i)][$ArrayKey] = $ArrayValue;
 154                               }
 155                           } //if (substr($OneLine,...
 156                       } //if($OneLine=trim(f...
 157                   } //while(!feof($FirstFileWi...
 158               } //for($i = 0; $i < func_...
 159  
 160           if(!is_array($ConvertTable[$FromCharset])) $ConvertTable[$FromCharset] = array();
 161  
 162           if ((func_num_args() > 1) && (count($ConvertTable[$FromCharset]) == count($ConvertTable[$ToCharset])) && (count(array_diff_assoc($ConvertTable[$FromCharset], $ConvertTable[$ToCharset])) == 0))
 163              {
 164               print $this -> DebugOutput(1, 1, "$FromCharset, $ToCharset");
 165               }
 166           return $ConvertTable;
 167           }
 168  
 169  
 170       function Convert ($StringToChange)
 171      {
 172           if(!strlen($StringToChange)) return '';
 173           $StringToChange = (string)($StringToChange);
 174  
 175           if($this -> FromCharset == $this -> ToCharset) return $StringToChange;
 176  
 177           $NewString = "";
 178  
 179  
 180           if ($this -> FromCharset != "utf-8")
 181          {
 182  
 183               for ($i = 0; $i < strlen($StringToChange); $i++)
 184              {
 185                   $HexChar = "";
 186                   $UnicodeHexChar = "";
 187                   $HexChar = strtoupper(dechex(ord($StringToChange[$i])));
 188                   if (strlen($HexChar) == 1) $HexChar = "0" . $HexChar;
 189                   if (($this -> FromCharset == "gsm0338") && ($HexChar == '1B')){
 190                       $i++;
 191                       $HexChar .= strtoupper(dechex(ord($StringToChange[$i])));
 192                       }
 193                   if ($this -> ToCharset != "utf-8")
 194                  {
 195                       if (in_array($HexChar, $this -> CharsetTable[$this -> FromCharset]))
 196                          {
 197                           $UnicodeHexChar = array_search($HexChar, $this -> CharsetTable[$this -> FromCharset]);
 198                           $UnicodeHexChars = explode("+", $UnicodeHexChar);
 199                           for($UnicodeHexCharElement = 0; $UnicodeHexCharElement < count($UnicodeHexChars); $UnicodeHexCharElement++)
 200                          {
 201                               if (array_key_exists($UnicodeHexChars[$UnicodeHexCharElement], $this -> CharsetTable[$this -> ToCharset]))
 202                                  {
 203                                   if ($this -> Entities == true)
 204                                  {
 205                                       $NewString .= $this -> UnicodeEntity($this -> HexToUtf($UnicodeHexChars[$UnicodeHexCharElement]));
 206                                       }
 207                                  else
 208                                      {
 209                                       $NewString .= chr(hexdec($this -> CharsetTable[$this -> ToCharset][$UnicodeHexChars[$UnicodeHexCharElement]]));
 210                                       }
 211                                   }
 212                              else
 213                                  {
 214                                   print $this -> DebugOutput(0, 1, $StringToChange[$i]);
 215                                   }
 216                               } //for($UnicodeH...
 217                           }
 218                      else
 219                          {
 220                           print $this -> DebugOutput(0, 2, $StringToChange[$i]);
 221                           }
 222                       }
 223                  else
 224                      {
 225                       if (in_array("$HexChar", $this -> CharsetTable[$this -> FromCharset]))
 226                          {
 227                           $UnicodeHexChar = array_search($HexChar, $this -> CharsetTable[$this -> FromCharset]);
 228  
 229                           $UnicodeHexChars = explode("+", $UnicodeHexChar);
 230                           for($UnicodeHexCharElement = 0; $UnicodeHexCharElement < count($UnicodeHexChars); $UnicodeHexCharElement++)
 231                          {
 232                               if ($this -> Entities == true)
 233                              {
 234                                   $NewString .= $this -> UnicodeEntity($this -> HexToUtf($UnicodeHexChars[$UnicodeHexCharElement]));
 235                                   }
 236                              else
 237                                  {
 238                                   $NewString .= $this -> HexToUtf($UnicodeHexChars[$UnicodeHexCharElement]);
 239                                   }
 240                               } // for
 241                           }
 242                      else
 243                          {
 244                           print $this -> DebugOutput(0, 2, $StringToChange[$i]);
 245                           }
 246                       }
 247                   }
 248               }
 249  
 250           else if($this -> FromCharset == "utf-8")
 251          {
 252               $HexChar = "";
 253               $UnicodeHexChar = "";
 254               $this -> CharsetTable = $this -> MakeConvertTable ($this -> ToCharset);
 255               foreach ($this -> CharsetTable[$this -> ToCharset] as $UnicodeHexChar => $HexChar)
 256              {
 257                   if ($this -> Entities == true){
 258                       $EntitieOrChar = $this -> UnicodeEntity($this -> HexToUtf($UnicodeHexChar));
 259                       }
 260                  else
 261                      {
 262                       $EntitieOrChar = chr(hexdec($HexChar));
 263                       }
 264                   $StringToChange = str_replace($this -> HexToUtf($UnicodeHexChar), $EntitieOrChar, $StringToChange);
 265                   }
 266               $NewString = $StringToChange;
 267               }
 268  
 269           return $NewString;
 270           }
 271  
 272  
 273       function ConvertArray(& $array)
 274      {
 275           if (!is_array($array))
 276              {
 277               $array = $this -> Convert($array);
 278               return;
 279               }
 280           while(list($k, $v) = each($array))
 281          {
 282               $this -> ConvertArray($v);
 283               $array[$k] = $v;
 284               }
 285           }
 286  
 287  
 288       function DebugOutput ($Group, $Number, $Value = false)
 289      {
 290           $Debug[0][0] = "Error, can NOT read file: " . $Value . "<br>";
 291           $Debug[0][1] = "Error, can't find maching char \"" . $Value . "\" in destination encoding table!" . "<br>";
 292           $Debug[0][2] = "Error, can't find maching char \"" . $Value . "\" in source encoding table!" . "<br>";
 293           $Debug[0][3] = "Error, you did NOT set variable " . $Value . " in Convert() function." . "<br>";
 294           $Debug[0][4] = "You can NOT convert string from " . $Value . " to " . $Value . "!" . "<BR>";
 295           $Debug[1][0] = "Notice, you are trying to convert string from " . $Value . " to " . $Value . ", don't you feel it's strange? ;-)" . "<br>";
 296           $Debug[1][1] = "Notice, both charsets " . $Value . " are identical! Check encoding tables files." . "<br>";
 297           $Debug[1][2] = "Notice, there is no unicode char in the string you are trying to convert." . "<br>";
 298  
 299           if (DEBUG_MODE >= $Group)
 300          {
 301               return $Debug[$Group][$Number];
 302               }
 303           } // function DebugOutput
 304  
 305      } //class ends here
 306  ?>