<?php
/*
Usage: 1. include 'bijoy2unicode.php'
2. Call convertBijoyToUnicode($stringToFormat) function, Unicode formatted string will be returned.
*/
$preConversionMap = array(
' +' => ' ',
'yy' => 'y', //Double Hrosh-u-Kar
'vv' => 'v', //Double Aa-Kar
'' => '', //Double Jukto-L - L+Double-L = Triple L
'y&' => 'y', //Hoshonto+Hrosh-u
'„&' => '„', //Hoshonto+Ri-Kar
'‡u' => 'u‡', //ChondroBindu Error /Typing Mistake
'wu' => 'uw', //ChondroBindu Error /Typing Mistake
' ,' => ',',
' \\|' => '\\|',
'\\\\ ' => '',
' \\\\' => '',
'\\\\' => '',
'\n +' => '\n',
' +\n' => '\n',
'\n\n\n\n\n' => '\n\n',
'\n\n\n\n' => '\n\n',
'\n\n\n' => '\n\n'
);
$conversionMap = array(
// Vowels Start
'Av' => 'আ',
'A' => 'অ',
'B' => 'ই',
'C' => 'ঈ',
'D' => 'উ',
'E' => 'ঊ',
'F' => 'ঋ',
'G' => 'এ',
'H' => 'ঐ',
'I' => 'ও',
'J' => 'ঔ',
// Constants
'K' => 'ক',
'L' => 'খ',
'M' => 'গ',
'N' => 'ঘ',
'O' => 'ঙ',
'P' => 'চ',
'Q' => 'ছ',
'R' => 'জ',
'S' => 'ঝ',
'T' => 'ঞ',
'U' => 'ট',
'V' => 'ঠ',
'W' => 'ড',
'X' => 'ঢ',
'Y' => 'ণ',
'Z' => 'ত',
'_' => 'থ',
'`' => 'দ',
'a' => 'ধ',
'b' => 'ন',
'c' => 'প',
'd' => 'ফ',
'e' => 'ব',
'f' => 'ভ',
'g' => 'ম',
'h' => 'য',
'i' => 'র',
'j' => 'ল',
'k' => 'শ',
'l' => 'ষ',
'm' => 'স',
'n' => 'হ',
'o' => 'ড়',
'p' => 'ঢ়',
'q' => 'য়',
'r' => 'ৎ',
's' => 'ং',
't' => 'ঃ',
'u' => 'ঁ',
// Numbers
'0' => '০',
'1' => '১',
'2' => '২',
'3' => '৩',
'4' => '৪',
'5' => '৫',
'6' => '৬',
'7' => '৭',
'8' => '৮',
'9' => '৯',
// Kars
'•' => 'ঙ্',
'v' => 'া', // Aa-Kar
'w' => 'ি', // i-Kar
'x' => 'ী', // I-Kar
'y' => 'ু', // u-Kar
'z' => 'ু', // u-Kar
'“' => 'ু', // u-kar
'–' => 'ু', // u-kar
'~' => 'ূ', // U-kar
'ƒ' => 'ূ', // U-kaar
'‚' => 'ূ', // U-kaar
'„„' => 'ৃ', //Double Rri-kar Bug
'„' => 'ৃ', // Ri-Kar
'…' => 'ৃ', // Ri-Kar
'†' => 'ে', // E-Kar
'‡' => 'ে', // E-Kar
'ˆ' => 'ৈ', // Oi-Kar
'‰' => 'ৈ', // Oi-Kar
'Š' => 'ৗ', // Ou-Kar
'\\|' => '।', // Full-Stop
'\\&' => '্', // Ho-shonto
// Jukto Okkhor
'\\^' => '্ব',
'‘' => '্তু',
'’' => '্থ',
'‹' => '্ক',
'Œ' => '্ক্র',
'”' => 'চ্',
'—' => '্ত',
'˜' => 'দ্',
'™' => 'দ্',
'š' => 'ন্',
'›' => 'ন্',
'œ' => '্ন',
'Ÿ' => '্ব',
'¡' => '্ব',
'¢' => '্ভ',
'£' => '্ভ্র',
'¤' => 'ম্',
'¥' => '্ম',
'¦' => '্ব',
'§' => '্ম',
'¨' => '্য',
'©' => 'র্',
'ª' => '্র',
'«' => '্র',
'¬' => '্ল',
'' => '্ল',
'®' => 'ষ্',
'¯' => 'স্',
'°' => 'ক্ক',
'±' => 'ক্ট',
'²' => 'ক্ষ্ণ', //shu(kkhno)
'³' => 'ক্ত',
'´' => 'ক্ম',
'µ' => 'ক্র',
'¶' => 'ক্ষ',
'·' => 'ক্স',
'¸' => 'গু',
'¹' => 'জ্ঞ',
'º' => 'গ্দ',
'»' => 'গ্ধ',
'¼' => 'ঙ্ক',
'½' => 'ঙ্গ',
'¾' => 'জ্জ',
'¿' => '্ত্র',
'À' => 'জ্ঝ',
'Á' => 'জ্ঞ',
'Â' => 'ঞ্চ',
'Ã' => 'ঞ্ছ',
'Ä' => 'ঞ্জ',
'Å' => 'ঞ্ঝ',
'Æ' => 'ট্ট',
'Ç' => 'ড্ড',
'È' => 'ণ্ট',
'É' => 'ণ্ঠ',
'Ê' => 'ণ্ড',
'Ë' => 'ত্ত',
'Ì' => 'ত্থ',
'Í' => 'ত্ম',
'Î' => 'ত্র',
'Ï' => 'দ্দ',
'Ð' => '-',
'Ñ' => '-',
'Ò' => '"',
'Ó' => '"',
'Ô' => "'",
'Õ' => "'",
'Ö' => '্র',
'×' => 'দ্ধ',
'Ø' => 'দ্ব',
'Ù' => 'দ্ম',
'Ú' => 'ন্ঠ',
'Û' => 'ন্ড',
'Ü' => 'ন্ধ',
'Ý' => 'ন্স',
'Þ' => 'প্ট',
'ß' => 'প্ত',
'à' => 'প্প',
'á' => 'প্স',
'â' => 'ব্জ',
'ã' => 'ব্দ',
'ä' => 'ব্ধ',
'å' => 'ভ্র',
'æ' => 'ম্ন',
'ç' => 'ম্ফ',
'è' => '্ন',
'é' => 'ল্ক',
'ê' => 'ল্গ',
'ë' => 'ল্ট',
'ì' => 'ল্ড',
'í' => 'ল্প',
'î' => 'ল্ফ',
'ï' => 'শু',
'ð' => 'শ্চ',
'ñ' => 'শ্ছ',
'ò' => 'ষ্ণ',
'ó' => 'ষ্ট',
'ô' => 'ষ্ঠ',
'õ' => 'ষ্ফ',
'ö' => 'স্খ',
'÷' => 'স্ট',
'ø' => 'স্ন', //(sn)eho //†ønØ
'ù' => 'স্ফ',
'ú' => '্প',
'û' => 'হু',
'ü' => 'হৃ',
'ý' => 'হ্ন',
'þ' => 'হ্ম'
);
$proConversionMap = array('্্' => '্');
$postConversionMap = array(
//Colon with Number/Space
'০ঃ' => '০:',
'১ঃ' => '১:',
'২ঃ' => '২:',
'৩ঃ' => '৩:',
'৪ঃ' => '৪:',
'৫ঃ' => '৫:',
'৬ঃ' => '৬:',
'৭ঃ' => '৭:',
'৮ঃ' => '৮:',
'৯ঃ' => '৯:',
' ঃ' => ' :',
'\nঃ' => '\n:',
']ঃ' => ']:',
'\\[ঃ' => '\\[:',
' ' => ' ',
'অা' => 'আ',
'্্' => '্'
);
function IsBanglaDigit($c) {
if ($c >= '০' && $c <= '৯')
return true;
return false;
}
function IsBanglaPreKar($c) {
if ($c == 'ি' || $c == 'ৈ' || $c == 'ে')
return true;
return false;
}
function IsBanglaPostKar($c) {
if ($c == 'া' || $c == 'ো' || $c == 'ৌ' || $c == 'ৗ' || $c == 'ু' || $c == 'ূ' || $c == 'ী' || $c == 'ৃ')
return true;
return false;
}
function IsBanglaKar($c) {
if (IsBanglaPreKar($c) || IsBanglaPostKar($c))
return true;
return false;
}
function IsBanglaBanjonborno($c) {
if ($c == 'ক' || $c == 'খ' || $c == 'গ' || $c == 'ঘ' || $c == 'ঙ' || $c == 'চ' || $c == 'ছ' || $c == 'জ' || $c == 'ঝ' || $c == 'ঞ' || $c == 'ট' || $c == 'ঠ' || $c == 'ড' || $c == 'ঢ' || $c == 'ণ' || $c == 'ত' || $c == 'থ' || $c == 'দ' || $c == 'ধ' || $c == 'ন' || $c == 'প' || $c == 'ফ' || $c == 'ব' || $c == 'ভ' || $c == 'ম' || $c == 'য' || $c == 'র' || $c == 'ল' || $c == 'শ' || $c == 'ষ' || $c == 'স' || $c == 'হ' || $c == 'ড়' || $c == 'ঢ়' || $c == 'য়' || $c == 'ৎ' || $c == 'ং' || $c == 'ঃ' || $c == 'ঁ')
return true;
return false;
}
function IsBanglaSoroborno($c) {
if ($c == 'অ' || $c == 'আ' || $c == 'ই' || $c == 'ঈ' || $c == 'উ' || $c == 'ঊ' || $c == 'ঋ' || $c == 'ঌ' || $c == 'এ' || $c == 'ঐ' || $c == 'ও' || $c == 'ঔ')
return true;
return false;
}
function IsBanglaNukta($c) {
if ($c == 'ঁ')
return true;
return false;
}
function IsBanglaHalant($c) {
if ($c == '্')
return true;
return false;
}
function IsSpace($c) {
if ($c == ' ' || $c == '\t' || $c == '\n' || $c == '\r')
return true;
return false;
}
function reArrangeUnicodeConvertedText($str) {
mb_internal_encoding("UTF-8"); //force multi-byte UTF-8 encoding
global $proConversionMap;
for ($i = 0; $i < mb_strlen($str); ++$i) {
// Change refs
if ($i < (mb_strlen($str) - 1) && mbCharAt($str, $i) == 'র' && IsBanglaHalant(mbCharAt($str, $i + 1)) && !IsBanglaHalant(mbCharAt($str, $i - 1))) {
$j = 1;
while (true) {
if ($i - $j < 0) {
break;
}
if (IsBanglaBanjonborno(mbCharAt($str, $i - $j)) && IsBanglaHalant(mbCharAt($str, $i - $j - 1))) {
$j += 2;
} else if ($j == 1 && IsBanglaKar(mbCharAt($str, $i - $j))) {
$j++;
} else {
break;
}
}
$temp = subString($str, 0, $i - $j);
$temp .= mbCharAt($str, $i);
$temp .= mbCharAt($str, $i + 1);
$temp .= subString($str, $i - $j, $i);
$temp .= subString($str, $i + 2, mb_strlen($str));
$str = $temp;
$i += 1;
continue;
}
}
$str = doCharMap($str, $proConversionMap);
for ($i = 0; $i < mb_strlen($str); ++$i) {
if ($i < mb_strlen($str) - 1 && mbCharAt($str, $i) == 'র' && IsBanglaHalant(mbCharAt($str, $i + 1)) && !IsBanglaHalant(mbCharAt($str, $i - 1)) && IsBanglaHalant(mbCharAt($str, $i + 2))) {
$j = 1;
while (true) {
if ($i - $j < 0) {
break;
}
if (IsBanglaBanjonborno(mbCharAt($str, $i - $j)) && IsBanglaHalant(mbCharAt($str, $i - $j - 1))) {
$j += 2;
} else if ($j == 1 && IsBanglaKar(mbCharAt($str, $i - $j))) {
$j++;
} else {
break;
}
}
$temp = subString($str, 0, $i - $j);
$temp .= mbCharAt($str, $i);
$temp .= mbCharAt($str, $i + 1);
$temp .= subString($str, $i - $j, $i);
$temp .= subString($str, $i + 2, mb_strlen($str));
$str = $temp;
$i += 1;
continue;
}
// for 'Vowel + HALANT + Consonant' it should be 'HALANT + Consonant + Vowel'
if ($i > 0 && mbCharAt($str, $i) == '\u09CD' && (IsBanglaKar(mbCharAt($str, $i - 1)) || IsBanglaNukta(mbCharAt($str, $i - 1))) && $i < mb_strlen($str) - 1) {
$temp = subString($str, 0, $i - 1);
$temp .= mbCharAt($str, $i);
$temp .= mbCharAt($str, $i + 1);
$temp .= mbCharAt($str, $i - 1);
$temp .= subString($str, $i + 2, mb_strlen($str));
$str = $temp;
}
// for 'RA (\u09B0) + HALANT + Vowel' it should be 'Vowel + RA (\u09B0) + HALANT'
if ($i > 0 && $i < mb_strlen($str) - 1 && mbCharAt($str, $i) == '\u09CD' && mbCharAt($str, $i - 1) == '\u09B0' && mbCharAt($str, $i - 2) != '\u09CD' && IsBanglaKar(mbCharAt($str, $i + 1))) {
$temp = subString($str, 0, $i - 1);
$temp .= mbCharAt($str, $i + 1);
$temp .= mbCharAt($str, $i - 1);
$temp .= mbCharAt($str, $i);
$temp .= subString($str, $i + 2, mb_strlen($str));
$str = $temp;
}
// Change pre-kar to post format suitable for unicode
if ($i < mb_strlen($str) - 1 && IsBanglaPreKar(mbCharAt($str, $i)) && IsSpace(mbCharAt($str, $i + 1)) == false) {
$temp = subString($str, 0, $i);
$j = 1;
while (($i + $j) < mb_strlen($str) - 1 && IsBanglaBanjonborno(mbCharAt($str, $i + $j))) {
if (($i + $j) < mb_strlen($str) && IsBanglaHalant(mbCharAt($str, $i + $j + 1))) {
$j += 2;
} else {
break;
}
}
$temp .= subString($str, $i + 1, $i + $j + 1);
$l = 0;
if (mbCharAt($str, $i) == 'ে' && mbCharAt($str, $i + $j + 1) == 'া') {
$temp .= "ো";
$l = 1;
} else if (mbCharAt($str, $i) == 'ে' && mbCharAt($str, $i + $j + 1) == "ৗ") {
$temp .= "ৌ";
$l = 1;
} else {
$temp .= mbCharAt($str, $i);
}
$temp .= subString($str, $i + $j + $l + 1, mb_strlen($str));
$str = $temp;
$i += $j;
}
// nukta should be placed after kars
if ($i < mb_strlen($str) - 1 && IsBanglaNukta(mbCharAt($str, $i)) && IsBanglaPostKar(mbCharAt($str, $i + 1))) {
$temp = subString($str, 0, $i);
$temp .= mbCharAt($str, $i + 1);
$temp .= mbCharAt($str, $i);
$temp .= subString($str, $i + 2, mb_strlen($str));
$str = $temp;
}
}
return $str;
}
//main conversion function
function convertBijoyToUnicode($srcString) {
global $preConversionMap, $conversionMap, $postConversionMap;
$srcString = doCharMap($srcString, $preConversionMap);
$srcString = doCharMap($srcString, $conversionMap);
$srcString = reArrangeUnicodeConvertedText($srcString);
$srcString = doCharMap($srcString, $postConversionMap);
return $srcString;
}
function doCharMap($text, $charMap) {
foreach ($charMap as $srcKey => $keyVal) {
$format = "@$srcKey@";
$text = preg_replace($format, $keyVal, $text);
}
return $text;
}
//returns the $i-th byte of the multi-byte string $str
function mbCharAt($str, $i) {
return mb_substr($str, $i, 1);
}
//returns the javascript 'substring' method equivalent
function subString($string, $from, $to) {
return mb_substr($string, $from, $to - $from);
}
?>
