Index: trunk/wb/search/search.php =================================================================== --- trunk/wb/search/search.php (revision 475) +++ trunk/wb/search/search.php (revision 476) @@ -62,7 +62,7 @@ // and do some convertion to both require(WB_PATH.'/search/search_convert.php'); $string = strtr($string,$string_conv_all); - $string_entities = strtr($string_entities,$string_entities_conv_all); + $string_entities = strtr($string_entities,$string_conv_all); $search_string = $string_entities; } else { $string = ''; Index: trunk/wb/search/search_convert.php =================================================================== --- trunk/wb/search/search_convert.php (revision 475) +++ trunk/wb/search/search_convert.php (revision 476) @@ -33,19 +33,20 @@ exit(0); } +//htmlspecialchars_decode +$string_htmlspecialchars_decode=array( + "<"=>"<", ">"=>">", "&"=>"&", """=>"\"", "'"=>"\'" +); +//htmlspecialchars_decode +$string_htmlspecialchars_encode=array( + "<"=>"<", ">"=>">", "&"=>"&", "\""=>""", "\'"=>"'" +); + //convert some bad characters $string_conv_all=array( - "<"=>"<",">"=>">","&"=>"&","\""=>""","\'"=>"'","!"=>" ","-"=>" ","="=>" " -); -$string_entities_conv_all=array( "!"=>" ","-"=>" ","="=>" " ); -//htmlspecialchars_decode -$string_htmlspecialchars_decode=array( - "<"=>"<", ">"=>">", "&"=>"&", """=>"\"", "''"=>"\'" -); - //umlauts to '(upper|lower)' for preg_match() //this is UTF-8-encoded $string_ul_umlauts = array( @@ -55,7 +56,7 @@ "\xc3\xa2" => "(\xc3\xa2|\xc3\x82)", "\xc3\xa3" => "(\xc3\xa3|\xc3\x83)", "\xc3\xa4" => "(\xc3\xa4|\xc3\x84)", - "\xc3\xa5" => "(\xc3\xa5|\xc3\x85)", + "\xc3\xa5" => "(\xc3\xa5|\xc3\x85|\xe2\x84\xab)", "\xc3\xa6" => "(\xc3\xa6|\xc3\x86)", "\xc3\xa7" => "(\xc3\xa7|\xc3\x87)", "\xc3\xa8" => "(\xc3\xa8|\xc3\x88)", @@ -162,12 +163,9 @@ "\xc6\xb9" => "(\xc6\xb9|\xc6\xb8)", "\xc6\xbd" => "(\xc6\xbd|\xc6\xbc)", "\xc6\xbf" => "(\xc6\xbf|\xc7\xb7)", - "\xc7\x85" => "(\xc7\x85|\xc7\x84)", - "\xc7\x86" => "(\xc7\x86|\xc7\x84)", - "\xc7\x88" => "(\xc7\x88|\xc7\x87)", - "\xc7\x89" => "(\xc7\x89|\xc7\x87)", - "\xc7\x8b" => "(\xc7\x8b|\xc7\x8a)", - "\xc7\x8c" => "(\xc7\x8c|\xc7\x8a)", + "\xc7\x86" => "(\xc7\x86|\xc7\x84|\xc7\x85)", + "\xc7\x89" => "(\xc7\x89|\xc7\x87|\xc7\x88)", + "\xc7\x8c" => "(\xc7\x8c|\xc7\x8a|\xc7\x8b)", "\xc7\x8e" => "(\xc7\x8e|\xc7\x8d)", "\xc7\x90" => "(\xc7\x90|\xc7\x8f)", "\xc7\x92" => "(\xc7\x92|\xc7\x91)", @@ -186,8 +184,8 @@ "\xc7\xab" => "(\xc7\xab|\xc7\xaa)", "\xc7\xad" => "(\xc7\xad|\xc7\xac)", "\xc7\xaf" => "(\xc7\xaf|\xc7\xae)", - "\xc7\xb2" => "(\xc7\xb2|\xc7\xb1)", - "\xc7\xb3" => "(\xc7\xb3|\xc7\xb1)", + "\xc7\xb2" => "(\xc7\xb2|\xc7\xb1|\xc7\xb3)", + "\xc7\xb3" => "(\xc7\xb3|\xc7\xb1|\xc7\xb2)", "\xc7\xb5" => "(\xc7\xb5|\xc7\xb4)", "\xc7\xb9" => "(\xc7\xb9|\xc7\xb8)", "\xc7\xbb" => "(\xc7\xbb|\xc7\xba)", @@ -249,7 +247,7 @@ "\xce\xb5" => "(\xce\xb5|\xce\x95)", "\xce\xb6" => "(\xce\xb6|\xce\x96)", "\xce\xb7" => "(\xce\xb7|\xce\x97)", - "\xce\xb8" => "(\xce\xb8|\xce\x98)", + "\xce\xb8" => "(\xce\xb8|\xce\x98|\xcf\xb4)", "\xce\xb9" => "(\xce\xb9|\xce\x99)", "\xce\xba" => "(\xce\xba|\xce\x9a)", "\xce\xbb" => "(\xce\xbb|\xce\x9b)", @@ -266,7 +264,7 @@ "\xcf\x86" => "(\xcf\x86|\xce\xa6)", "\xcf\x87" => "(\xcf\x87|\xce\xa7)", "\xcf\x88" => "(\xcf\x88|\xce\xa8)", - "\xcf\x89" => "(\xcf\x89|\xce\xa9)", + "\xcf\x89" => "(\xcf\x89|\xce\xa9|\xe2\x84\xa6)", "\xcf\x8a" => "(\xcf\x8a|\xce\xaa)", "\xcf\x8b" => "(\xcf\x8b|\xce\xab)", "\xcf\x8c" => "(\xcf\x8c|\xce\x8c)", @@ -897,12 +895,12 @@ "\xc6\xb7" => "(\xc6\xb7|\xca\x92)", "\xc6\xb8" => "(\xc6\xb8|\xc6\xb9)", "\xc6\xbc" => "(\xc6\xbc|\xc6\xbd)", - "\xc7\x84" => "(\xc7\x84|\xc7\x86)", - "\xc7\x85" => "(\xc7\x85|\xc7\x86)", - "\xc7\x87" => "(\xc7\x87|\xc7\x89)", - "\xc7\x88" => "(\xc7\x88|\xc7\x89)", - "\xc7\x8a" => "(\xc7\x8a|\xc7\x8c)", - "\xc7\x8b" => "(\xc7\x8b|\xc7\x8c)", + "\xc7\x84" => "(\xc7\x84|\xc7\x86|\xc7\x85)", + "\xc7\x85" => "(\xc7\x85|\xc7\x84|\xc7\x86)", + "\xc7\x87" => "(\xc7\x87|\xc7\x89|\xc7\x88)", + "\xc7\x88" => "(\xc7\x88|\xc7\x87|\xc7\x89)", + "\xc7\x8a" => "(\xc7\x8a|\xc7\x8c|\xc7\x8b)", + "\xc7\x8b" => "(\xc7\x8b|\xc7\x8a|\xc7\x8c)", "\xc7\x8d" => "(\xc7\x8d|\xc7\x8e)", "\xc7\x8f" => "(\xc7\x8f|\xc7\x90)", "\xc7\x91" => "(\xc7\x91|\xc7\x92)", @@ -920,8 +918,7 @@ "\xc7\xaa" => "(\xc7\xaa|\xc7\xab)", "\xc7\xac" => "(\xc7\xac|\xc7\xad)", "\xc7\xae" => "(\xc7\xae|\xc7\xaf)", - "\xc7\xb1" => "(\xc7\xb1|\xc7\xb3)", - "\xc7\xb2" => "(\xc7\xb2|\xc7\xb3)", + "\xc7\xb1" => "(\xc7\xb1|\xc7\xb3|\xc7\xb2)", "\xc7\xb4" => "(\xc7\xb4|\xc7\xb5)", "\xc7\xb6" => "(\xc7\xb6|\xc6\x95)", "\xc7\xb7" => "(\xc7\xb7|\xc6\xbf)", @@ -962,26 +959,26 @@ "\xce\x8e" => "(\xce\x8e|\xcf\x8d)", "\xce\x8f" => "(\xce\x8f|\xcf\x8e)", "\xce\x91" => "(\xce\x91|\xce\xb1)", - "\xce\x92" => "(\xce\x92|\xce\xb2)", + "\xce\x92" => "(\xce\x92|\xce\xb2|\xcf\x90)", "\xce\x93" => "(\xce\x93|\xce\xb3)", "\xce\x94" => "(\xce\x94|\xce\xb4)", - "\xce\x95" => "(\xce\x95|\xce\xb5)", + "\xce\x95" => "(\xce\x95|\xce\xb5|\xcf\xb5)", "\xce\x96" => "(\xce\x96|\xce\xb6)", "\xce\x97" => "(\xce\x97|\xce\xb7)", - "\xce\x98" => "(\xce\x98|\xce\xb8)", - "\xce\x99" => "(\xce\x99|\xce\xb9)", - "\xce\x9a" => "(\xce\x9a|\xce\xba)", + "\xce\x98" => "(\xce\x98|\xce\xb8|\xcf\x91)", + "\xce\x99" => "(\xce\x99|\xce\xb9|\xcd\x85|\xe1\xbe\xbe)", + "\xce\x9a" => "(\xce\x9a|\xce\xba|\xcf\xb0)", "\xce\x9b" => "(\xce\x9b|\xce\xbb)", - "\xce\x9c" => "(\xce\x9c|\xce\xbc)", + "\xce\x9c" => "(\xce\x9c|\xce\xbc|\xc2\xb5)", "\xce\x9d" => "(\xce\x9d|\xce\xbd)", "\xce\x9e" => "(\xce\x9e|\xce\xbe)", "\xce\x9f" => "(\xce\x9f|\xce\xbf)", - "\xce\xa0" => "(\xce\xa0|\xcf\x80)", - "\xce\xa1" => "(\xce\xa1|\xcf\x81)", - "\xce\xa3" => "(\xce\xa3|\xcf\x83)", + "\xce\xa0" => "(\xce\xa0|\xcf\x80|\xcf\x96)", + "\xce\xa1" => "(\xce\xa1|\xcf\x81|\xcf\xb1)", + "\xce\xa3" => "(\xce\xa3|\xcf\x83|\xcf\x82|\xcf\xb2)", "\xce\xa4" => "(\xce\xa4|\xcf\x84)", "\xce\xa5" => "(\xce\xa5|\xcf\x85)", - "\xce\xa6" => "(\xce\xa6|\xcf\x86)", + "\xce\xa6" => "(\xce\xa6|\xcf\x86|\xcf\x95)", "\xce\xa7" => "(\xce\xa7|\xcf\x87)", "\xce\xa8" => "(\xce\xa8|\xcf\x88)", "\xce\xa9" => "(\xce\xa9|\xcf\x89)", @@ -1200,7 +1197,7 @@ "\xe1\xb9\x9a" => "(\xe1\xb9\x9a|\xe1\xb9\x9b)", "\xe1\xb9\x9c" => "(\xe1\xb9\x9c|\xe1\xb9\x9d)", "\xe1\xb9\x9e" => "(\xe1\xb9\x9e|\xe1\xb9\x9f)", - "\xe1\xb9\xa0" => "(\xe1\xb9\xa0|\xe1\xb9\xa1)", + "\xe1\xb9\xa0" => "(\xe1\xb9\xa0|\xe1\xb9\xa1|\xe1\xba\x9b)", "\xe1\xb9\xa2" => "(\xe1\xb9\xa2|\xe1\xb9\xa3)", "\xe1\xb9\xa4" => "(\xe1\xb9\xa4|\xe1\xb9\xa5)", "\xe1\xb9\xa6" => "(\xe1\xb9\xa6|\xe1\xb9\xa7)", @@ -1476,7 +1473,11 @@ "\xf0\x90\x90\xa2" => "(\xf0\x90\x90\xa2|\xf0\x90\x91\x8a)", "\xf0\x90\x90\xa3" => "(\xf0\x90\x90\xa3|\xf0\x90\x91\x8b)", "\xf0\x90\x90\xa4" => "(\xf0\x90\x90\xa4|\xf0\x90\x91\x8c)", - "\xf0\x90\x90\xa5" => "(\xf0\x90\x90\xa5|\xf0\x90\x91\x8d)" + "\xf0\x90\x90\xa5" => "(\xf0\x90\x90\xa5|\xf0\x90\x91\x8d)", + "i" => "(i|\xc4\xb0|I)", + "I" => "(I|\xc4\xb1|i)", + "k" => "(k|\xe2\x84\xaa|K)", + "S" => "(S|\xc5\xbf|s)" ); ?> \ No newline at end of file Index: trunk/wb/framework/frontend.functions.php =================================================================== --- trunk/wb/framework/frontend.functions.php (revision 475) +++ trunk/wb/framework/frontend.functions.php (revision 476) @@ -71,16 +71,14 @@ require_once(WB_PATH.'/framework/functions.php'); require(WB_PATH.'/search/search_convert.php'); $foo = entities_to_umlauts($foo, 'UTF-8'); - foreach($arr_string as $string) { - $string = strtr($string, $string_htmlspecialchars_decode); - $string = entities_to_umlauts($string, 'UTF-8'); - $string = preg_quote($string, '/'); - $string = strtr($string, $string_ul_umlauts); - $foo = preg_replace('/('.$string.')(?=[^>]*<)/iUS', '$1',$foo); - $pos = strpos($foo, '<'); - if ($pos === false) { // "===" means identicaly - $foo = preg_replace('/('.$string.')/i', '$1',$foo); - } + array_walk($arr_string, create_function('&$v,$k','$v = preg_quote($v, \'/\');')); + $search_string = implode("|", $arr_string); + $string = entities_to_umlauts($search_string, 'UTF-8'); + $string = strtr($string, $string_ul_umlauts); + $foo = preg_replace('/('.$string.')(?=[^>]*<)/iUS', '$1',$foo); + $pos = strpos($foo, '<'); + if ($pos === false) { // "===" means identicaly + $foo = preg_replace('/('.$string.')/i', '$1',$foo); } $foo = umlauts_to_entities($foo, 'UTF-8', 0); return $foo; Index: trunk/wb/framework/functions.php =================================================================== --- trunk/wb/framework/functions.php (revision 475) +++ trunk/wb/framework/functions.php (revision 476) @@ -340,7 +340,9 @@ // Function as replecement for php's htmlspecialchars() function my_htmlspecialchars($string) { - $string = strtr($string, array("<"=>"<", ">"=>">", "\""=>""", "\'"=>"'")); + $string = preg_replace("/&(?=[#a-z0-9]+;)/i", "_x_", $string); + $string = strtr($string, array("<"=>"<", ">"=>">", "&"=>"&", "\""=>""", "\'"=>"'")); + $string = preg_replace("/_x_(?=[#a-z0-9]+;)/i", "&", $string); return($string); } @@ -369,11 +371,18 @@ if ($charset_out == $charset_in) { return $string; } + $use_iconv = true; + $use_mbstring = true; + if(version_compare(PHP_VERSION, "5.1.0", "<")) { + $use_mbstring = false; // don't rely on mb_convert_encoding if php<5.1.0 + $use_iconv = false; // don't rely on iconv neither + } + // try mb_convert_encoding(). This can handle to or from HTML-ENTITIES, too - if (function_exists('mb_convert_encoding')) { + if ($use_mbstring && function_exists('mb_convert_encoding')) { // there's no GB2312 or ISO-8859-11 encoding in php's mb_* functions if ($charset_in=='ISO-8859-11' || $charset_in=='GB2312') { - if (function_exists('iconv')) { + if ($use_iconv && function_exists('iconv')) { $string = iconv($charset_in, 'UTF-8', $string); } else { @@ -390,7 +399,7 @@ } if ($charset_out=='ISO-8859-11' || $charset_out=='GB2312') { $string=mb_convert_encoding($string, 'UTF-8', $charset_in); - if (function_exists('iconv')) { + if ($use_iconv && function_exists('iconv')) { $string = iconv('UTF-8', $charset_out, $string); } else { @@ -409,75 +418,14 @@ } // try iconv(). This can't handle to or from HTML-ENTITIES. - if (function_exists('iconv') && $charset_out!='HTML-ENTITIES' && $charset_in!='HTML-ENTITIES' ) { + if ($use_iconv && function_exists('iconv') && $charset_out!='HTML-ENTITIES' && $charset_in!='HTML-ENTITIES' ) { $string = iconv($charset_in, $charset_out, $string); return $string; } - // do the UTF-8->HTML-ENTITIES or HTML-ENTITIES->UTF-8 translation + // do the UTF-8->HTML-ENTITIES or HTML-ENTITIES->UTF-8 translation if mb_convert_encoding isn't available if (($charset_in=='HTML-ENTITIES' && $charset_out=='UTF-8') || ($charset_in=='UTF-8' && $charset_out=='HTML-ENTITIES')) { - $named_to_numbered_entities=array( - ' '=>' ','¡'=>'¡','¢'=>'¢','£'=>'£','¤'=>'¤', - '¥'=>'¥','¦'=>'¦','§'=>'§','¨'=>'¨','ª'=>'ª', - '«'=>'«','¬'=>'¬','­'=>'­','®'=>'®','¯'=>'¯', - '°'=>'°','±'=>'±','²'=>'²','³'=>'³','´'=>'´', - 'µ'=>'µ','¶'=>'¶','·'=>'·','¸'=>'¸','¹'=>'¹', - 'º'=>'º','»'=>'»','¼'=>'¼','½'=>'½','¾'=>'¾', - '¿'=>'¿','÷'=>'÷','∅'=>'∅','€'=>'€', - 'Á'=>'Á','á'=>'á','Â'=>'Â', - 'â'=>'â','Æ'=>'Æ','æ'=>'æ','À'=>'À','à'=>'à', - 'Å'=>'Å','å'=>'å','Ã'=>'Ã','ã'=>'ã','Ä'=>'Ä', - 'ä'=>'ä','Ç'=>'Ç','ç'=>'ç','É'=>'É','é'=>'é', - 'Ê'=>'Ê','ê'=>'ê','È'=>'È','è'=>'è','Ë'=>'Ë', - 'ë'=>'ë','Í'=>'Í','í'=>'í','Î'=>'Î','î'=>'î', - 'Ì'=>'Ì','ì'=>'ì','Ï'=>'Ï','ï'=>'ï','Ñ'=>'Ñ', - 'ñ'=>'ñ','Ó'=>'Ó','ó'=>'ó','Ô'=>'Ô','ô'=>'ô', - 'Œ'=>'Œ','œ'=>'œ','Ò'=>'Ò','ò'=>'ò','Õ'=>'Õ', - 'õ'=>'õ','Ö'=>'Ö','ö'=>'ö','Š'=>'Š','š'=>'š', - 'ß'=>'ß','Ú'=>'Ú','ú'=>'ú','Û'=>'Û','û'=>'û', - 'Ù'=>'Ù','ù'=>'ù','Ü'=>'Ü','ü'=>'ü','Ý'=>'Ý', - 'ý'=>'ý','Ÿ'=>'Ÿ','ÿ'=>'ÿ','©'=>'©','®'=>'®', - 'Ð'=>'Ð','×'=>'×','Ø'=>'Ø','Þ'=>'Þ','ð'=>'ð', - 'ø'=>'ø','þ'=>'þ'); - $numbered_to_named_entities=array('Á'=>'Á','á'=>'á','Â'=>'Â', - ' '=>' ','¡'=>'¡','¢'=>'¢','£'=>'£','¤'=>'¤', - '¥'=>'¥','¦'=>'¦','§'=>'§','¨'=>'¨','ª'=>'ª', - '«'=>'«','¬'=>'¬','­'=>'­','®'=>'®','¯'=>'¯', - '°'=>'°','±'=>'±','²'=>'²','³'=>'³','´'=>'´', - 'µ'=>'µ','¶'=>'¶','·'=>'·','¸'=>'¸','¹'=>'¹', - 'º'=>'º','»'=>'»','¼'=>'¼','½'=>'½','¾'=>'¾', - '¿'=>'¿','÷'=>'÷','∅'=>'∅','€'=>'€', - 'â'=>'â','Æ'=>'Æ','æ'=>'æ','À'=>'À','à'=>'à', - 'Å'=>'Å','å'=>'å','Ã'=>'Ã','ã'=>'ã','Ä'=>'Ä', - 'ä'=>'ä','Ç'=>'Ç','ç'=>'ç','É'=>'É','é'=>'é', - 'Ê'=>'Ê','ê'=>'ê','È'=>'È','è'=>'è','Ë'=>'Ë', - 'ë'=>'ë','Í'=>'Í','í'=>'í','Î'=>'Î','î'=>'î', - 'Ì'=>'Ì','ì'=>'ì','Ï'=>'Ï','ï'=>'ï','Ñ'=>'Ñ', - 'ñ'=>'ñ','Ó'=>'Ó','ó'=>'ó','Ô'=>'Ô','ô'=>'ô', - 'Œ'=>'Œ','œ'=>'œ','Ò'=>'Ò','ò'=>'ò','Õ'=>'Õ', - 'õ'=>'õ','Ö'=>'Ö','ö'=>'ö','Š'=>'Š','š'=>'š', - 'ß'=>'ß','Ú'=>'Ú','ú'=>'ú','Û'=>'Û','û'=>'û', - 'Ù'=>'Ù','ù'=>'ù','Ü'=>'Ü','ü'=>'ü','Ý'=>'Ý', - 'ý'=>'ý','Ÿ'=>'Ÿ','ÿ'=>'ÿ','©'=>'©','®'=>'®', - 'Ð'=>'Ð','×'=>'×','Ø'=>'Ø','Þ'=>'Þ','ð'=>'ð', - 'ø'=>'ø','þ'=>'þ'); - if ($charset_in == 'HTML-ENTITIES') { - $string = strtr($string, $named_to_numbered_entities); - $string = preg_replace("/&#([0-9]+);/e", "code_to_utf8($1)", $string); - } - elseif ($charset_out == 'HTML-ENTITIES') { - $string = preg_replace("/&#([0-9]+);/e", "code_to_utf8($1)", $string); - $char = ""; - while (strlen($string) > 0) { - preg_match("/^(.)(.*)$/su", $string, $match); - if (strlen($match[1]) > 1) { - $char .= "&#".uniord($match[1]).";"; - } else $char .= $match[1]; - $string = $match[2]; - } - $string = $char; - $string = strtr($string, $numbered_to_named_entities); - } + $string = string_decode_encode_entities($string, $charset_out, $charset_in); return $string; } @@ -530,7 +478,140 @@ // and mbstring _and_ iconv aren't available. return $string; } -// support-function for mb_convert_encoding_wrapper() + +// Decodes or encodes html-entities. Works for utf-8 only! +function string_decode_encode_entities($string, $out='HTML-ENTITIES', $in='UTF-8') { + if(!(($in=='UTF-8' || $in=='HTML-ENTITIES') && ($out=='UTF-8' || $out=='HTML-ENTITIES'))) { + return $string; + } + $named_to_numbered_entities=array( + 'Á'=>'Á','á'=>'á', + 'Â'=>'Â','â'=>'â','´'=>'´','Æ'=>'Æ','æ'=>'æ', + 'À'=>'À','à'=>'à','ℵ'=>'ℵ','Α'=>'Α','α'=>'α', + '&'=>'&','∧'=>'∧','∠'=>'∠','''=>''','Å'=>'Å','å'=>'å', + '≈'=>'≈','Ã'=>'Ã','ã'=>'ã','Ä'=>'Ä','ä'=>'ä', + '„'=>'„','Β'=>'Β','β'=>'β','¦'=>'¦','•'=>'•', + '∩'=>'∩','Ç'=>'Ç','ç'=>'ç','¸'=>'¸','¢'=>'¢', + 'Χ'=>'Χ','χ'=>'χ','ˆ'=>'ˆ','♣'=>'♣','≅'=>'≅', + '©'=>'©','↵'=>'↵','∪'=>'∪','¤'=>'¤','‡'=>'‡', + '†'=>'†','⇓'=>'⇓','↓'=>'↓','°'=>'°','Δ'=>'Δ', + 'δ'=>'δ','♦'=>'&v#9830;','÷'=>'÷','É'=>'É','é'=>'é', + 'Ê'=>'Ê','ê'=>'ê','È'=>'È','è'=>'è','∅'=>'∅', + ' '=>' ',' '=>' ','Ε'=>'Ε','ε'=>'ε','≡'=>'≡', + 'Η'=>'Η','η'=>'η','Ð'=>'Ð','ð'=>'ð','Ë'=>'Ë','ë'=>'ë', + '€'=>'€','∃'=>'∃','ƒ'=>'ƒ','∀'=>'∀','½'=>'½', + '¼'=>'¼','¾'=>'¾','⁄'=>'⁄','Γ'=>'Γ','γ'=>'γ', + '≥'=>'≥','>'=>'>','⇔'=>'⇔','↔'=>'↔','♥'=>'♥', + '…'=>'…','Í'=>'Í','í'=>'í','Î'=>'Î','î'=>'î', + '¡'=>'¡','Ì'=>'Ì','ì'=>'ì','ℑ'=>'ℑ','∞'=>'∞', + '∫'=>'∫','Ι'=>'Ι','ι'=>'ι','¿'=>'¿','∈'=>'∈', + 'Ï'=>'Ï','ï'=>'ï','Κ'=>'Κ','κ'=>'κ','Λ'=>'Λ', + 'λ'=>'λ','⟨'=>'〈','«'=>'«','⇐'=>'⇐','←'=>'←', + '⌈'=>'⌈','“'=>'“','≤'=>'≤','⌊'=>'⌊','∗'=>'∗', + '◊'=>'◊','‎'=>'‎','‹'=>'‹','‘'=>'‘','<'=>'<', + '¯'=>'¯','—'=>'—','µ'=>'µ','·'=>'·','−'=>'−', + 'Μ'=>'Μ','μ'=>'μ','∇'=>'∇',' '=>' ','–'=>'–', + '≠'=>'≠','∋'=>'∋','¬'=>'¬','∉'=>'∉','⊄'=>'⊄', + 'Ñ'=>'Ñ','ñ'=>'ñ','Ν'=>'Ν','ν'=>'ν','Ó'=>'Ó', + 'ó'=>'ó','Ô'=>'Ô','ô'=>'ô','Œ'=>'Œ','œ'=>'œ', + 'Ò'=>'Ò','ò'=>'ò','‾'=>'‾','Ω'=>'Ω','ω'=>'ω', + 'Ο'=>'Ο','ο'=>'ο','⊕'=>'⊕','∨'=>'∨','ª'=>'ª', + 'º'=>'º','Ø'=>'Ø','ø'=>'ø','Õ'=>'Õ','õ'=>'õ', + '⊗'=>'⊗','Ö'=>'Ö','ö'=>'ö','¶'=>'¶','∂'=>'∂', + '‰'=>'‰','⊥'=>'⊥','Φ'=>'Φ','φ'=>'φ','Π'=>'Π', + 'π'=>'π','ϖ'=>'ϖ','±'=>'±','£'=>'£','″'=>'″', + '′'=>'′','∏'=>'∏','∝'=>'∝','Ψ'=>'Ψ','ψ'=>'ψ', + '"'=>'"','√'=>'√','⟩'=>'〉','»'=>'»','⇒'=>'⇒', + '→'=>'→','⌉'=>'⌉','”'=>'”','ℜ'=>'ℜ','®'=>'®', + '⌋'=>'⌋','Ρ'=>'Ρ','ρ'=>'ρ','‏'=>'‏','›'=>'›', + '’'=>'’','‚'=>'‚','Š'=>'Š','š'=>'š','⋅'=>'⋅', + '§'=>'§','­'=>'­','Σ'=>'Σ','σ'=>'σ','ς'=>'ς', + '∼'=>'∼','♠'=>'♠','⊂'=>'⊂','⊆'=>'⊆','∑'=>'∑', + '⊃'=>'⊃','¹'=>'¹','²'=>'²','³'=>'³','⊇'=>'⊇', + 'ß'=>'ß','Τ'=>'Τ','τ'=>'τ','∴'=>'∴','Θ'=>'Θ', + 'θ'=>'θ','ϑ'=>'ϑ',' '=>' ','Þ'=>'Þ','þ'=>'þ', + '˜'=>'˜','×'=>'×','™'=>'™','Ú'=>'Ú','ú'=>'ú', + '⇑'=>'⇑','↑'=>'↑','Û'=>'Û','û'=>'û','Ù'=>'Ù', + 'ù'=>'ù','¨'=>'¨','ϒ'=>'ϒ','Υ'=>'Υ','υ'=>'υ', + 'Ü'=>'Ü','ü'=>'ü','℘'=>'℘','Ξ'=>'Ξ','ξ'=>'ξ', + 'Ý'=>'Ý','ý'=>'ý','¥'=>'¥','Ÿ'=>'Ÿ','ÿ'=>'ÿ', + 'Ζ'=>'Ζ','ζ'=>'ζ','‍'=>'‍','‌'=>'‌' + ); + $numbered_to_named_entities=array( + 'Á'=>'Á','á'=>'á','Â'=>'Â','â'=>'â','´'=>'´', + 'Æ'=>'Æ','æ'=>'æ','À'=>'À','à'=>'à','ℵ'=>'ℵ', + 'Α'=>'Α','α'=>'α','&'=>'&','∧'=>'∧','∠'=>'∠', + '''=>''','Å'=>'Å','å'=>'å','≈'=>'≈','Ã'=>'Ã', + 'ã'=>'ã','Ä'=>'Ä','ä'=>'ä','„'=>'„','Β'=>'Β', + 'β'=>'β','¦'=>'¦','•'=>'•','∩'=>'∩','Ç'=>'Ç', + 'ç'=>'ç','¸'=>'¸','¢'=>'¢','Χ'=>'Χ','χ'=>'χ', + 'ˆ'=>'ˆ','♣'=>'♣','≅'=>'≅','©'=>'©','↵'=>'↵', + '∪'=>'∪','¤'=>'¤','‡'=>'‡','†'=>'†','⇓'=>'⇓', + '↓'=>'↓','°'=>'°','Δ'=>'Δ','δ'=>'δ','&v#9830;'=>'♦', + '÷'=>'÷','É'=>'É','é'=>'é','Ê'=>'Ê','ê'=>'ê', + 'È'=>'È','è'=>'è','∅'=>'∅',' '=>' ',' '=>' ', + 'Ε'=>'Ε','ε'=>'ε','≡'=>'≡','Η'=>'Η','η'=>'η', + 'Ð'=>'Ð','ð'=>'ð','Ë'=>'Ë','ë'=>'ë','€'=>'€', + '∃'=>'∃','ƒ'=>'ƒ','∀'=>'∀','½'=>'½','¼'=>'¼', + '¾'=>'¾','⁄'=>'⁄','Γ'=>'Γ','γ'=>'γ','≥'=>'≥', + '>'=>'>','⇔'=>'⇔','↔'=>'↔','♥'=>'♥','…'=>'…', + 'Í'=>'Í','í'=>'í','Î'=>'Î','î'=>'î','¡'=>'¡', + 'Ì'=>'Ì','ì'=>'ì','ℑ'=>'ℑ','∞'=>'∞','∫'=>'∫', + 'Ι'=>'Ι','ι'=>'ι','¿'=>'¿','∈'=>'∈','Ï'=>'Ï', + 'ï'=>'ï','Κ'=>'Κ','κ'=>'κ','Λ'=>'Λ','λ'=>'λ', + '〈'=>'⟨','«'=>'«','⇐'=>'⇐','←'=>'←','⌈'=>'⌈', + '“'=>'“','≤'=>'≤','⌊'=>'⌊','∗'=>'∗','◊'=>'◊', + '‎'=>'‎','‹'=>'‹','‘'=>'‘','<'=>'<','¯'=>'¯', + '—'=>'—','µ'=>'µ','·'=>'·','−'=>'−','Μ'=>'Μ', + 'μ'=>'μ','∇'=>'∇',' '=>' ','–'=>'–','≠'=>'≠', + '∋'=>'∋','¬'=>'¬','∉'=>'∉','⊄'=>'⊄','Ñ'=>'Ñ', + 'ñ'=>'ñ','Ν'=>'Ν','ν'=>'ν','Ó'=>'Ó','ó'=>'ó', + 'Ô'=>'Ô','ô'=>'ô','Œ'=>'Œ','œ'=>'œ','Ò'=>'Ò', + 'ò'=>'ò','‾'=>'‾','Ω'=>'Ω','ω'=>'ω','Ο'=>'Ο', + 'ο'=>'ο','⊕'=>'⊕','∨'=>'∨','ª'=>'ª','º'=>'º', + 'Ø'=>'Ø','ø'=>'ø','Õ'=>'Õ','õ'=>'õ','⊗'=>'⊗', + 'Ö'=>'Ö','ö'=>'ö','¶'=>'¶','∂'=>'∂','‰'=>'‰', + '⊥'=>'⊥','Φ'=>'Φ','φ'=>'φ','Π'=>'Π','π'=>'π','ϖ'=>'ϖ', + '±'=>'±','£'=>'£','″'=>'″','′'=>'′','∏'=>'∏', + '∝'=>'∝','Ψ'=>'Ψ','ψ'=>'ψ','"'=>'"','√'=>'√', + '〉'=>'⟩','»'=>'»','⇒'=>'⇒','→'=>'→','⌉'=>'⌉', + '”'=>'”','ℜ'=>'ℜ','®'=>'®','⌋'=>'⌋','Ρ'=>'Ρ', + 'ρ'=>'ρ','‏'=>'‏','›'=>'›','’'=>'’','‚'=>'‚', + 'Š'=>'Š','š'=>'š','⋅'=>'⋅','§'=>'§','­'=>'­', + 'Σ'=>'Σ','σ'=>'σ','ς'=>'ς','∼'=>'∼','♠'=>'♠', + '⊂'=>'⊂','⊆'=>'⊆','∑'=>'∑','⊃'=>'⊃','¹'=>'¹', + '²'=>'²','³'=>'³','⊇'=>'⊇','ß'=>'ß','Τ'=>'Τ', + 'τ'=>'τ','∴'=>'∴','Θ'=>'Θ','θ'=>'θ','ϑ'=>'ϑ', + ' '=>' ','Þ'=>'Þ','þ'=>'þ','˜'=>'˜','×'=>'×', + '™'=>'™','Ú'=>'Ú','ú'=>'ú','⇑'=>'⇑','↑'=>'↑', + 'Û'=>'Û','û'=>'û','Ù'=>'Ù','ù'=>'ù','¨'=>'¨', + 'ϒ'=>'ϒ','Υ'=>'Υ','υ'=>'υ','Ü'=>'Ü','ü'=>'ü', + '℘'=>'℘','Ξ'=>'Ξ','ξ'=>'ξ','Ý'=>'Ý','ý'=>'ý', + '¥'=>'¥','Ÿ'=>'Ÿ','ÿ'=>'ÿ','Ζ'=>'Ζ','ζ'=>'ζ','‍'=>'‍', + '‌'=>'‌' + ); + + if ($in == 'HTML-ENTITIES') { + $string = strtr($string, $named_to_numbered_entities); + $string = preg_replace("/&#([0-9]+);/e", "code_to_utf8($1)", $string); + } + elseif ($out == 'HTML-ENTITIES') { + //$string = preg_replace("/&#([0-9]+);/e", "code_to_utf8($1)", $string); + $char = ""; + while (strlen($string) > 0) { + preg_match("/^(.)(.*)$/su", $string, $match); + if (strlen($match[1]) > 1) { + $char .= "&#".uniord($match[1]).";"; + } else $char .= $match[1]; + $string = $match[2]; + } + $string = $char; + $string = strtr($string, $numbered_to_named_entities); + } + return $string; +} + +// support-function for string_decode_encode_entities() function uniord($c) { $ud = 0; if (ord($c{0}) >= 0 && ord($c{0}) <= 127) $ud = ord($c{0}); @@ -604,22 +685,54 @@ function entities_to_umlauts($string, $charset_out=DEFAULT_CHARSET, $convert_htmlspecialchars=0) { $charset_out = strtoupper($charset_out); if ($charset_out == '') { $charset_out = 'ISO-8859-1'; } - $string = string_to_utf8($string); - if($charset_out!='UTF-8' && is_UTF8($string)) { + $charset_in = strtoupper(DEFAULT_CHARSET); + + // string to utf-8 + if ($charset_in == 'ISO-8859-1' || $charset_in == 'UTF-8') { + if ($charset_in == 'ISO-8859-1') { + $string=utf8_encode($string); + } + // decode html-entities + if(preg_match("/&[#a-zA-Z0-9]+;/", $string)) { + $string=string_decode_encode_entities($string, 'UTF-8', 'HTML-ENTITIES'); + } + } + else { + $string = string_to_utf8($string); // will decode html-entities, too. + } + // string to $charset_out + if($charset_out == 'ISO-8859-1') { + $string=utf8_decode($string); + } + elseif($charset_out != 'UTF-8' && is_UTF8($string)) { $string=mb_convert_encoding_wrapper($string, $charset_out, 'UTF-8'); } - return($string); -} + return $string; +} // Function to convert a string from mixed html-entitites/$charset_in-umlauts to pure html-entities function umlauts_to_entities($string, $charset_in=DEFAULT_CHARSET, $convert_htmlspecialchars=0) { $charset_in = strtoupper($charset_in); if ($charset_in == "") { $charset_in = 'ISO-8859-1'; } - $string = string_to_utf8($string, $charset_in); - if (is_UTF8($string)) { - $string=mb_convert_encoding_wrapper($string,'HTML-ENTITIES','UTF-8'); + + // string to utf-8 + if ($charset_in == 'ISO-8859-1' || $charset_in == 'UTF-8') { + if ($charset_in == 'ISO-8859-1') { + $string=utf8_encode($string); + } + // encode html-entities + $string=string_decode_encode_entities($string, 'HTML-ENTITIES', 'UTF-8'); + //$string=mb_convert_encoding_wrapper($string, 'HTML-ENTITIES', 'UTF-8'); } - return($string); + else { + $string = string_to_utf8($string, $charset_in); + // encode html-entities + if (is_UTF8($string)) { + $string=string_decode_encode_entities($string, 'HTML-ENTITIES', 'UTF-8'); + //$string=mb_convert_encoding_wrapper($string, 'HTML-ENTITIES', 'UTF-8'); + } + } + return $string; } // translate any latin/greek/cyrillic html-entities to their plain 7bit equivalents