Index: trunk/CHANGELOG =================================================================== --- trunk/CHANGELOG (revision 880) +++ trunk/CHANGELOG (revision 881) @@ -10,7 +10,10 @@ # = Bugfix ! = Update/Change -------------------------------------- 2.7.1 ------------------------------------- +------------------------------------- 2.7.1 ------------------------------------- +22-Nov-2008 Thomas Hornik +! search: replaced most of $string_ul_umlauts (from search_convert.php) through use of preg's u-switch. +! search: replaced strtr() by str_replace() (it's just faster). Changed ' to ' 06-Nov-2008 Thomas Hornik # Module code: curly backets get deleted by pparse() from phplib --> old fix in #569 wasn't that smart. Fixed. # fixed some typos in DE language-file, added missing tab in search.php Index: trunk/wb/framework/functions-utf8.php =================================================================== --- trunk/wb/framework/functions-utf8.php (revision 880) +++ trunk/wb/framework/functions-utf8.php (revision 881) @@ -146,7 +146,7 @@ function utf8_fast_entities_to_umlauts($str) { if(UTF8_MBSTRING) { // we need this for use with mb_convert_encoding - $str = str_replace(array('&','>','<','"',''',' '), array('&amp;','&gt;','&lt;','&quot;','&#39;','&nbsp;'), $str); + $str = str_replace(array('&','>','<','"',''',' '), array('&amp;','&gt;','&lt;','&quot;','&#39;','&nbsp;'), $str); // we need two mb_convert_encoding()-calls - is this a bug? // mb_convert_encoding("öö", 'UTF-8', 'HTML-ENTITIES'); // with string in utf-8-encoding doesn't work. Result: "öö" // Work-around: convert all umlauts to entities first ("öö"->"öö"), then all entities to umlauts ("öö"->"öö") @@ -404,11 +404,11 @@ // missed some? - Many UTF-8-chars can't be romanized // convert to HTML-entities, and replace entites by hex-numbers $str = utf8_fast_umlauts_to_entities($str, false); - $str = str_replace(''', ''', $str); + $str = str_replace(''', ''', $str); $str = preg_replace('/&#([0-9]+);/e', "dechex('$1')", $str); // maybe there are some > < ' " &   left, replace them too - $entities = array('>'=>'','<'=>'','''=>'','\''=>'','"'=>'','&'=>'',' '=>' '); - $str = strtr($str, $entities); + $str = str_replace(array('>', '<', ''', '\'', '"', '&'), '', $str); + $str = str_replace('&', '', $str); return($str); } @@ -417,7 +417,7 @@ * Convert a string from mixed html-entities/umlauts to pure $charset_out-umlauts * * Will replace all numeric and named entities except - * > < ' " '   + * > < ' " '   * @author thorn */ function entities_to_umlauts2($string, $charset_out=DEFAULT_CHARSET) { Index: trunk/wb/framework/charsets_table.php =================================================================== --- trunk/wb/framework/charsets_table.php (revision 880) +++ trunk/wb/framework/charsets_table.php (revision 881) @@ -391,7 +391,7 @@ "\xe0\xb9\x94"=>"\xf4","\xe0\xb9\x95"=>"\xf5","\xe0\xb9\x96"=>"\xf6","\xe0\xb9\x97"=>"\xf7","\xe0\xb9\x98"=>"\xf8","\xe0\xb9\x99"=>"\xf9","\xe0\xb9\x9a"=>"\xfa","\xe0\xb9\x9b"=>"\xfb" ); -// nicht enthalten > < ' ' " &   +// nicht enthalten > < ' ' " &   global $named_entities; $named_entities=array( 'Á','á','Â','â','´','Æ','æ','À','à','ℵ', @@ -421,7 +421,7 @@ '¥','Ÿ','ÿ','Ζ','ζ','‍','‌' ); -// nicht enthalten > < ' ' " &   +// nicht enthalten > < ' ' " &   global $numbered_entities; $numbered_entities=array( 'Á','á','Â','â','´','Æ','æ','À','à','ℵ', Index: trunk/wb/framework/frontend.functions.php =================================================================== --- trunk/wb/framework/frontend.functions.php (revision 880) +++ trunk/wb/framework/frontend.functions.php (revision 881) @@ -70,13 +70,14 @@ if (!function_exists('search_highlight')) { function search_highlight($foo='', $arr_string=array()) { require_once(WB_PATH.'/framework/functions.php'); - static $string_ul_umlauts=array(); - if($string_ul_umlauts == array()) + static $string_ul_umlaut = FALSE; + static $string_ul_regex = FALSE; + if($string_ul_umlaut===FALSE || $string_ul_regex===FALSE) require(WB_PATH.'/search/search_convert.php'); $foo = entities_to_umlauts($foo, 'UTF-8'); array_walk($arr_string, create_function('&$v,$k','$v = preg_quote($v, \'/\');')); $search_string = implode("|", $arr_string); - $string = strtr($search_string, $string_ul_umlauts); + $string = str_replace($string_ul_umlaut, $string_ul_regex, $search_string); // the highlighting // match $string, but not inside , , or HTML-Tags // split $string into pieces - "cut away" styles, scripts, comments, HTML-tags and eMail-addresses @@ -86,9 +87,9 @@ $foo = ""; foreach($matches as $match) { if($match{0}!="<" && !preg_match('/^[A-Z0-9._%+-]+@[A-Z0-9.-]+\.[A-Z]{2,8}$/i', $match)) { - $match = strtr($match, array('<'=>'<', '>'=>'>', '&'=>'&', '"'=>'"', '''=>'\'', ' '=>"\xC2\xA0")); - $match = preg_replace('/('.$string.')/iS', '_span class=_highlight__$1_/span_',$match); - $match = strtr($match, array('<'=>'<', '>'=>'>', '&'=>'&', '"'=>'"', '\''=>''', "\xC2\xA0"=>' ')); + $match = str_replace(array('<', '>', '&', '"', ''', ' '), array('<', '>', '&', '"', '\'', "\xC2\xA0"), $match); + $match = preg_replace('/('.$string.')/ui', '_span class=_highlight__$1_/span_',$match); + $match = str_replace(array('<', '>', '&', '"', '\'', "\xC2\xA0"), array('<', '>', '&', '"', ''', ' '), $match); $match = str_replace(array('_span class=_highlight__', '_/span_'), array('', ''), $match); } $foo .= $match; @@ -215,22 +216,22 @@ if(defined('SEC_ANCHOR') && SEC_ANCHOR!='') { echo ''; } + + // fetch content -- this is where to place possible output-filters (before highlighting) + ob_start(); // fetch original content + require(WB_PATH.'/modules/'.$module.'/view.php'); + $content = ob_get_contents(); + ob_end_clean(); + // highlights searchresults - if (isset($_GET['searchresult']) AND is_numeric($_GET['searchresult']) AND !isset($_GET['nohighlight'])) { - if (isset($_GET['sstring']) AND !empty($_GET['sstring']) ){ - $arr_string = explode(" ", $_GET['sstring']); - if($_GET['searchresult'] == 2) { - // exact match - $arr_string[0] = strtr($arr_string[0], "_"," "); - } - ob_start(); //start output buffer - require(WB_PATH.'/modules/'.$module.'/view.php'); - $foo = ob_get_contents(); // put outputbuffer in $foo - ob_end_clean(); // clear outputbuffer - echo search_highlight($foo, $arr_string); + if(isset($_GET['searchresult']) && is_numeric($_GET['searchresult']) && !isset($_GET['nohighlight']) && isset($_GET['sstring']) && !empty($_GET['sstring'])) { + $arr_string = explode(" ", $_GET['sstring']); + if($_GET['searchresult']==2) { // exact match + $arr_string[0] = str_replace("_", " ", $arr_string[0]); } + echo search_highlight($content, $arr_string); } else { - require(WB_PATH.'/modules/'.$module.'/view.php'); + echo $content; } } } else { Index: trunk/wb/framework/functions.php =================================================================== --- trunk/wb/framework/functions.php (revision 880) +++ trunk/wb/framework/functions.php (revision 881) @@ -350,7 +350,7 @@ } // Convert a string from mixed html-entities/umlauts to pure $charset_out-umlauts -// Will replace all numeric and named entities except > < ' " '   +// Will replace all numeric and named entities except > < ' " '   // In case of error the returned string is unchanged, and a message is emitted. function entities_to_umlauts($string, $charset_out=DEFAULT_CHARSET) { require_once(WB_PATH.'/framework/functions-utf8.php'); Index: trunk/wb/search/search.php =================================================================== --- trunk/wb/search/search.php (revision 880) +++ trunk/wb/search/search.php (revision 881) @@ -238,7 +238,7 @@ require_once(WB_PATH.'/search/search_convert.php'); $search_words = array(); foreach($search_normal_array AS $str) { - $str = strtr($str, $string_ul_umlauts); + $str = str_replace($string_ul_umlaut, $string_ul_regex, $str); $search_words[] = $str; } @@ -650,7 +650,7 @@ $sstring = implode(" ", $search_normal_array); $link = $link."?searchresult=1&sstring=".urlencode($sstring); } else { - $sstring = strtr($search_normal_array[0], " ", "_"); + $sstring = str_replace(" ", "_",$search_normal_array[0]); $link = $link."?searchresult=2&sstring=".urlencode($sstring); } // Set vars to be replaced by values Index: trunk/wb/search/search_convert.php =================================================================== --- trunk/wb/search/search_convert.php (revision 880) +++ trunk/wb/search/search_convert.php (revision 881) @@ -34,1449 +34,65 @@ } if(!isset($search_lang)) $search_lang = LANGUAGE; -//umlauts to '(upper|lower)' for preg_match() -//this is UTF-8-encoded -$string_ul_umlauts = array( - "i" => "(?:i|\xc4\xb0|I)", - "I" => "(?:I|\xc4\xb1|i)", - "k" => "(?:k|\xe2\x84\xaa|K)", - "S" => "(?:S|\xc5\xbf|s)", - "\xc2\xb5" => "(?:\xc2\xb5|\xce\x9c)", - "\xc3\x9f" => "(?:\xc3\x9f|SS|ss)", // german ß - "\xc3\xa0" => "(?:\xc3\xa0|\xc3\x80)", - "\xc3\xa1" => "(?:\xc3\xa1|\xc3\x81)", - "\xc3\xa2" => "(?:\xc3\xa2|\xc3\x82)", - "\xc3\xa3" => "(?:\xc3\xa3|\xc3\x83)", - "\xc3\xa4" => "(?:\xc3\xa4|\xc3\x84)", - "\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)", - "\xc3\xa9" => "(?:\xc3\xa9|\xc3\x89)", - "\xc3\xaa" => "(?:\xc3\xaa|\xc3\x8a)", - "\xc3\xab" => "(?:\xc3\xab|\xc3\x8b)", - "\xc3\xac" => "(?:\xc3\xac|\xc3\x8c)", - "\xc3\xad" => "(?:\xc3\xad|\xc3\x8d)", - "\xc3\xae" => "(?:\xc3\xae|\xc3\x8e)", - "\xc3\xaf" => "(?:\xc3\xaf|\xc3\x8f)", - "\xc3\xb0" => "(?:\xc3\xb0|\xc3\x90)", - "\xc3\xb1" => "(?:\xc3\xb1|\xc3\x91)", - "\xc3\xb2" => "(?:\xc3\xb2|\xc3\x92)", - "\xc3\xb3" => "(?:\xc3\xb3|\xc3\x93)", - "\xc3\xb4" => "(?:\xc3\xb4|\xc3\x94)", - "\xc3\xb5" => "(?:\xc3\xb5|\xc3\x95)", - "\xc3\xb6" => "(?:\xc3\xb6|\xc3\x96)", - "\xc3\xb8" => "(?:\xc3\xb8|\xc3\x98)", - "\xc3\xb9" => "(?:\xc3\xb9|\xc3\x99)", - "\xc3\xba" => "(?:\xc3\xba|\xc3\x9a)", - "\xc3\xbb" => "(?:\xc3\xbb|\xc3\x9b)", - "\xc3\xbc" => "(?:\xc3\xbc|\xc3\x9c)", - "\xc3\xbd" => "(?:\xc3\xbd|\xc3\x9d)", - "\xc3\xbe" => "(?:\xc3\xbe|\xc3\x9e)", - "\xc3\xbf" => "(?:\xc3\xbf|\xc5\xb8)", - "\xc4\x81" => "(?:\xc4\x81|\xc4\x80)", - "\xc4\x83" => "(?:\xc4\x83|\xc4\x82)", - "\xc4\x85" => "(?:\xc4\x85|\xc4\x84)", - "\xc4\x87" => "(?:\xc4\x87|\xc4\x86)", - "\xc4\x89" => "(?:\xc4\x89|\xc4\x88)", - "\xc4\x8b" => "(?:\xc4\x8b|\xc4\x8a)", - "\xc4\x8d" => "(?:\xc4\x8d|\xc4\x8c)", - "\xc4\x8f" => "(?:\xc4\x8f|\xc4\x8e)", - "\xc4\x91" => "(?:\xc4\x91|\xc4\x90)", - "\xc4\x93" => "(?:\xc4\x93|\xc4\x92)", - "\xc4\x95" => "(?:\xc4\x95|\xc4\x94)", - "\xc4\x97" => "(?:\xc4\x97|\xc4\x96)", - "\xc4\x99" => "(?:\xc4\x99|\xc4\x98)", - "\xc4\x9b" => "(?:\xc4\x9b|\xc4\x9a)", - "\xc4\x9d" => "(?:\xc4\x9d|\xc4\x9c)", - "\xc4\x9f" => "(?:\xc4\x9f|\xc4\x9e)", - "\xc4\xa1" => "(?:\xc4\xa1|\xc4\xa0)", - "\xc4\xa3" => "(?:\xc4\xa3|\xc4\xa2)", - "\xc4\xa5" => "(?:\xc4\xa5|\xc4\xa4)", - "\xc4\xa7" => "(?:\xc4\xa7|\xc4\xa6)", - "\xc4\xa9" => "(?:\xc4\xa9|\xc4\xa8)", - "\xc4\xab" => "(?:\xc4\xab|\xc4\xaa)", - "\xc4\xad" => "(?:\xc4\xad|\xc4\xac)", - "\xc4\xaf" => "(?:\xc4\xaf|\xc4\xae)", - "\xc4\xb1" => "(?:\xc4\xb1|I)", - "\xc4\xb3" => "(?:\xc4\xb3|\xc4\xb2)", - "\xc4\xb5" => "(?:\xc4\xb5|\xc4\xb4)", - "\xc4\xb7" => "(?:\xc4\xb7|\xc4\xb6)", - "\xc4\xba" => "(?:\xc4\xba|\xc4\xb9)", - "\xc4\xbc" => "(?:\xc4\xbc|\xc4\xbb)", - "\xc4\xbe" => "(?:\xc4\xbe|\xc4\xbd)", - "\xc5\x80" => "(?:\xc5\x80|\xc4\xbf)", - "\xc5\x82" => "(?:\xc5\x82|\xc5\x81)", - "\xc5\x84" => "(?:\xc5\x84|\xc5\x83)", - "\xc5\x86" => "(?:\xc5\x86|\xc5\x85)", - "\xc5\x88" => "(?:\xc5\x88|\xc5\x87)", - "\xc5\x8b" => "(?:\xc5\x8b|\xc5\x8a)", - "\xc5\x8d" => "(?:\xc5\x8d|\xc5\x8c)", - "\xc5\x8f" => "(?:\xc5\x8f|\xc5\x8e)", - "\xc5\x91" => "(?:\xc5\x91|\xc5\x90)", - "\xc5\x93" => "(?:\xc5\x93|\xc5\x92)", - "\xc5\x95" => "(?:\xc5\x95|\xc5\x94)", - "\xc5\x97" => "(?:\xc5\x97|\xc5\x96)", - "\xc5\x99" => "(?:\xc5\x99|\xc5\x98)", - "\xc5\x9b" => "(?:\xc5\x9b|\xc5\x9a)", - "\xc5\x9d" => "(?:\xc5\x9d|\xc5\x9c)", - "\xc5\x9f" => "(?:\xc5\x9f|\xc5\x9e)", - "\xc5\xa1" => "(?:\xc5\xa1|\xc5\xa0)", - "\xc5\xa3" => "(?:\xc5\xa3|\xc5\xa2)", - "\xc5\xa5" => "(?:\xc5\xa5|\xc5\xa4)", - "\xc5\xa7" => "(?:\xc5\xa7|\xc5\xa6)", - "\xc5\xa9" => "(?:\xc5\xa9|\xc5\xa8)", - "\xc5\xab" => "(?:\xc5\xab|\xc5\xaa)", - "\xc5\xad" => "(?:\xc5\xad|\xc5\xac)", - "\xc5\xaf" => "(?:\xc5\xaf|\xc5\xae)", - "\xc5\xb1" => "(?:\xc5\xb1|\xc5\xb0)", - "\xc5\xb3" => "(?:\xc5\xb3|\xc5\xb2)", - "\xc5\xb5" => "(?:\xc5\xb5|\xc5\xb4)", - "\xc5\xb7" => "(?:\xc5\xb7|\xc5\xb6)", - "\xc5\xba" => "(?:\xc5\xba|\xc5\xb9)", - "\xc5\xbc" => "(?:\xc5\xbc|\xc5\xbb)", - "\xc5\xbe" => "(?:\xc5\xbe|\xc5\xbd)", - "\xc5\xbf" => "(?:\xc5\xbf|S)", - "\xc6\x83" => "(?:\xc6\x83|\xc6\x82)", - "\xc6\x85" => "(?:\xc6\x85|\xc6\x84)", - "\xc6\x88" => "(?:\xc6\x88|\xc6\x87)", - "\xc6\x8c" => "(?:\xc6\x8c|\xc6\x8b)", - "\xc6\x92" => "(?:\xc6\x92|\xc6\x91)", - "\xc6\x95" => "(?:\xc6\x95|\xc7\xb6)", - "\xc6\x99" => "(?:\xc6\x99|\xc6\x98)", - "\xc6\xa1" => "(?:\xc6\xa1|\xc6\xa0)", - "\xc6\xa3" => "(?:\xc6\xa3|\xc6\xa2)", - "\xc6\xa5" => "(?:\xc6\xa5|\xc6\xa4)", - "\xc6\xa8" => "(?:\xc6\xa8|\xc6\xa7)", - "\xc6\xad" => "(?:\xc6\xad|\xc6\xac)", - "\xc6\xb0" => "(?:\xc6\xb0|\xc6\xaf)", - "\xc6\xb4" => "(?:\xc6\xb4|\xc6\xb3)", - "\xc6\xb6" => "(?:\xc6\xb6|\xc6\xb5)", - "\xc6\xb9" => "(?:\xc6\xb9|\xc6\xb8)", - "\xc6\xbd" => "(?:\xc6\xbd|\xc6\xbc)", - "\xc6\xbf" => "(?:\xc6\xbf|\xc7\xb7)", - "\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)", - "\xc7\x94" => "(?:\xc7\x94|\xc7\x93)", - "\xc7\x96" => "(?:\xc7\x96|\xc7\x95)", - "\xc7\x98" => "(?:\xc7\x98|\xc7\x97)", - "\xc7\x9a" => "(?:\xc7\x9a|\xc7\x99)", - "\xc7\x9c" => "(?:\xc7\x9c|\xc7\x9b)", - "\xc7\x9d" => "(?:\xc7\x9d|\xc6\x8e)", - "\xc7\x9f" => "(?:\xc7\x9f|\xc7\x9e)", - "\xc7\xa1" => "(?:\xc7\xa1|\xc7\xa0)", - "\xc7\xa3" => "(?:\xc7\xa3|\xc7\xa2)", - "\xc7\xa5" => "(?:\xc7\xa5|\xc7\xa4)", - "\xc7\xa7" => "(?:\xc7\xa7|\xc7\xa6)", - "\xc7\xa9" => "(?:\xc7\xa9|\xc7\xa8)", - "\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\xb3|\xc7\xb1|\xc7\xb2)", - "\xc7\xb5" => "(?:\xc7\xb5|\xc7\xb4)", - "\xc7\xb9" => "(?:\xc7\xb9|\xc7\xb8)", - "\xc7\xbb" => "(?:\xc7\xbb|\xc7\xba)", - "\xc7\xbd" => "(?:\xc7\xbd|\xc7\xbc)", - "\xc7\xbf" => "(?:\xc7\xbf|\xc7\xbe)", - "\xc8\x81" => "(?:\xc8\x81|\xc8\x80)", - "\xc8\x83" => "(?:\xc8\x83|\xc8\x82)", - "\xc8\x85" => "(?:\xc8\x85|\xc8\x84)", - "\xc8\x87" => "(?:\xc8\x87|\xc8\x86)", - "\xc8\x89" => "(?:\xc8\x89|\xc8\x88)", - "\xc8\x8b" => "(?:\xc8\x8b|\xc8\x8a)", - "\xc8\x8d" => "(?:\xc8\x8d|\xc8\x8c)", - "\xc8\x8f" => "(?:\xc8\x8f|\xc8\x8e)", - "\xc8\x91" => "(?:\xc8\x91|\xc8\x90)", - "\xc8\x93" => "(?:\xc8\x93|\xc8\x92)", - "\xc8\x95" => "(?:\xc8\x95|\xc8\x94)", - "\xc8\x97" => "(?:\xc8\x97|\xc8\x96)", - "\xc8\x99" => "(?:\xc8\x99|\xc8\x98)", - "\xc8\x9b" => "(?:\xc8\x9b|\xc8\x9a)", - "\xc8\x9d" => "(?:\xc8\x9d|\xc8\x9c)", - "\xc8\x9f" => "(?:\xc8\x9f|\xc8\x9e)", - "\xc8\xa3" => "(?:\xc8\xa3|\xc8\xa2)", - "\xc8\xa5" => "(?:\xc8\xa5|\xc8\xa4)", - "\xc8\xa7" => "(?:\xc8\xa7|\xc8\xa6)", - "\xc8\xa9" => "(?:\xc8\xa9|\xc8\xa8)", - "\xc8\xab" => "(?:\xc8\xab|\xc8\xaa)", - "\xc8\xad" => "(?:\xc8\xad|\xc8\xac)", - "\xc8\xaf" => "(?:\xc8\xaf|\xc8\xae)", - "\xc8\xb1" => "(?:\xc8\xb1|\xc8\xb0)", - "\xc8\xb3" => "(?:\xc8\xb3|\xc8\xb2)", - "\xc9\x93" => "(?:\xc9\x93|\xc6\x81)", - "\xc9\x94" => "(?:\xc9\x94|\xc6\x86)", - "\xc9\x96" => "(?:\xc9\x96|\xc6\x89)", - "\xc9\x97" => "(?:\xc9\x97|\xc6\x8a)", - "\xc9\x99" => "(?:\xc9\x99|\xc6\x8f)", - "\xc9\x9b" => "(?:\xc9\x9b|\xc6\x90)", - "\xc9\xa0" => "(?:\xc9\xa0|\xc6\x93)", - "\xc9\xa3" => "(?:\xc9\xa3|\xc6\x94)", - "\xc9\xa8" => "(?:\xc9\xa8|\xc6\x97)", - "\xc9\xa9" => "(?:\xc9\xa9|\xc6\x96)", - "\xc9\xaf" => "(?:\xc9\xaf|\xc6\x9c)", - "\xc9\xb2" => "(?:\xc9\xb2|\xc6\x9d)", - "\xc9\xb5" => "(?:\xc9\xb5|\xc6\x9f)", - "\xca\x80" => "(?:\xca\x80|\xc6\xa6)", - "\xca\x83" => "(?:\xca\x83|\xc6\xa9)", - "\xca\x88" => "(?:\xca\x88|\xc6\xae)", - "\xca\x8a" => "(?:\xca\x8a|\xc6\xb1)", - "\xca\x8b" => "(?:\xca\x8b|\xc6\xb2)", - "\xca\x92" => "(?:\xca\x92|\xc6\xb7)", - "\xcd\x85" => "(?:\xcd\x85|\xce\x99)", - "\xce\xac" => "(?:\xce\xac|\xce\x86)", - "\xce\xad" => "(?:\xce\xad|\xce\x88)", - "\xce\xae" => "(?:\xce\xae|\xce\x89)", - "\xce\xaf" => "(?:\xce\xaf|\xce\x8a)", - "\xce\xb1" => "(?:\xce\xb1|\xce\x91)", - "\xce\xb2" => "(?:\xce\xb2|\xce\x92)", - "\xce\xb3" => "(?:\xce\xb3|\xce\x93)", - "\xce\xb4" => "(?:\xce\xb4|\xce\x94)", - "\xce\xb5" => "(?:\xce\xb5|\xce\x95)", - "\xce\xb6" => "(?:\xce\xb6|\xce\x96)", - "\xce\xb7" => "(?:\xce\xb7|\xce\x97)", - "\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)", - "\xce\xbc" => "(?:\xce\xbc|\xce\x9c)", - "\xce\xbd" => "(?:\xce\xbd|\xce\x9d)", - "\xce\xbe" => "(?:\xce\xbe|\xce\x9e)", - "\xce\xbf" => "(?:\xce\xbf|\xce\x9f)", - "\xcf\x80" => "(?:\xcf\x80|\xce\xa0)", - "\xcf\x81" => "(?:\xcf\x81|\xce\xa1)", - "\xcf\x82" => "(?:\xcf\x82|\xce\xa3)", - "\xcf\x83" => "(?:\xcf\x83|\xce\xa3)", - "\xcf\x84" => "(?:\xcf\x84|\xce\xa4)", - "\xcf\x85" => "(?:\xcf\x85|\xce\xa5)", - "\xcf\x86" => "(?:\xcf\x86|\xce\xa6)", - "\xcf\x87" => "(?:\xcf\x87|\xce\xa7)", - "\xcf\x88" => "(?:\xcf\x88|\xce\xa8)", - "\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)", - "\xcf\x8d" => "(?:\xcf\x8d|\xce\x8e)", - "\xcf\x8e" => "(?:\xcf\x8e|\xce\x8f)", - "\xcf\x90" => "(?:\xcf\x90|\xce\x92)", - "\xcf\x91" => "(?:\xcf\x91|\xce\x98)", - "\xcf\x95" => "(?:\xcf\x95|\xce\xa6)", - "\xcf\x96" => "(?:\xcf\x96|\xce\xa0)", - "\xcf\x9b" => "(?:\xcf\x9b|\xcf\x9a)", - "\xcf\x9d" => "(?:\xcf\x9d|\xcf\x9c)", - "\xcf\x9f" => "(?:\xcf\x9f|\xcf\x9e)", - "\xcf\xa1" => "(?:\xcf\xa1|\xcf\xa0)", - "\xcf\xa3" => "(?:\xcf\xa3|\xcf\xa2)", - "\xcf\xa5" => "(?:\xcf\xa5|\xcf\xa4)", - "\xcf\xa7" => "(?:\xcf\xa7|\xcf\xa6)", - "\xcf\xa9" => "(?:\xcf\xa9|\xcf\xa8)", - "\xcf\xab" => "(?:\xcf\xab|\xcf\xaa)", - "\xcf\xad" => "(?:\xcf\xad|\xcf\xac)", - "\xcf\xaf" => "(?:\xcf\xaf|\xcf\xae)", - "\xcf\xb0" => "(?:\xcf\xb0|\xce\x9a)", - "\xcf\xb1" => "(?:\xcf\xb1|\xce\xa1)", - "\xcf\xb2" => "(?:\xcf\xb2|\xce\xa3)", - "\xcf\xb5" => "(?:\xcf\xb5|\xce\x95)", - "\xd0\xb0" => "(?:\xd0\xb0|\xd0\x90)", - "\xd0\xb1" => "(?:\xd0\xb1|\xd0\x91)", - "\xd0\xb2" => "(?:\xd0\xb2|\xd0\x92)", - "\xd0\xb3" => "(?:\xd0\xb3|\xd0\x93)", - "\xd0\xb4" => "(?:\xd0\xb4|\xd0\x94)", - "\xd0\xb5" => "(?:\xd0\xb5|\xd0\x95)", - "\xd0\xb6" => "(?:\xd0\xb6|\xd0\x96)", - "\xd0\xb7" => "(?:\xd0\xb7|\xd0\x97)", - "\xd0\xb8" => "(?:\xd0\xb8|\xd0\x98)", - "\xd0\xb9" => "(?:\xd0\xb9|\xd0\x99)", - "\xd0\xba" => "(?:\xd0\xba|\xd0\x9a)", - "\xd0\xbb" => "(?:\xd0\xbb|\xd0\x9b)", - "\xd0\xbc" => "(?:\xd0\xbc|\xd0\x9c)", - "\xd0\xbd" => "(?:\xd0\xbd|\xd0\x9d)", - "\xd0\xbe" => "(?:\xd0\xbe|\xd0\x9e)", - "\xd0\xbf" => "(?:\xd0\xbf|\xd0\x9f)", - "\xd1\x80" => "(?:\xd1\x80|\xd0\xa0)", - "\xd1\x81" => "(?:\xd1\x81|\xd0\xa1)", - "\xd1\x82" => "(?:\xd1\x82|\xd0\xa2)", - "\xd1\x83" => "(?:\xd1\x83|\xd0\xa3)", - "\xd1\x84" => "(?:\xd1\x84|\xd0\xa4)", - "\xd1\x85" => "(?:\xd1\x85|\xd0\xa5)", - "\xd1\x86" => "(?:\xd1\x86|\xd0\xa6)", - "\xd1\x87" => "(?:\xd1\x87|\xd0\xa7)", - "\xd1\x88" => "(?:\xd1\x88|\xd0\xa8)", - "\xd1\x89" => "(?:\xd1\x89|\xd0\xa9)", - "\xd1\x8a" => "(?:\xd1\x8a|\xd0\xaa)", - "\xd1\x8b" => "(?:\xd1\x8b|\xd0\xab)", - "\xd1\x8c" => "(?:\xd1\x8c|\xd0\xac)", - "\xd1\x8d" => "(?:\xd1\x8d|\xd0\xad)", - "\xd1\x8e" => "(?:\xd1\x8e|\xd0\xae)", - "\xd1\x8f" => "(?:\xd1\x8f|\xd0\xaf)", - "\xd1\x90" => "(?:\xd1\x90|\xd0\x80)", - "\xd1\x91" => "(?:\xd1\x91|\xd0\x81)", - "\xd1\x92" => "(?:\xd1\x92|\xd0\x82)", - "\xd1\x93" => "(?:\xd1\x93|\xd0\x83)", - "\xd1\x94" => "(?:\xd1\x94|\xd0\x84)", - "\xd1\x95" => "(?:\xd1\x95|\xd0\x85)", - "\xd1\x96" => "(?:\xd1\x96|\xd0\x86)", - "\xd1\x97" => "(?:\xd1\x97|\xd0\x87)", - "\xd1\x98" => "(?:\xd1\x98|\xd0\x88)", - "\xd1\x99" => "(?:\xd1\x99|\xd0\x89)", - "\xd1\x9a" => "(?:\xd1\x9a|\xd0\x8a)", - "\xd1\x9b" => "(?:\xd1\x9b|\xd0\x8b)", - "\xd1\x9c" => "(?:\xd1\x9c|\xd0\x8c)", - "\xd1\x9d" => "(?:\xd1\x9d|\xd0\x8d)", - "\xd1\x9e" => "(?:\xd1\x9e|\xd0\x8e)", - "\xd1\x9f" => "(?:\xd1\x9f|\xd0\x8f)", - "\xd1\xa1" => "(?:\xd1\xa1|\xd1\xa0)", - "\xd1\xa3" => "(?:\xd1\xa3|\xd1\xa2)", - "\xd1\xa5" => "(?:\xd1\xa5|\xd1\xa4)", - "\xd1\xa7" => "(?:\xd1\xa7|\xd1\xa6)", - "\xd1\xa9" => "(?:\xd1\xa9|\xd1\xa8)", - "\xd1\xab" => "(?:\xd1\xab|\xd1\xaa)", - "\xd1\xad" => "(?:\xd1\xad|\xd1\xac)", - "\xd1\xaf" => "(?:\xd1\xaf|\xd1\xae)", - "\xd1\xb1" => "(?:\xd1\xb1|\xd1\xb0)", - "\xd1\xb3" => "(?:\xd1\xb3|\xd1\xb2)", - "\xd1\xb5" => "(?:\xd1\xb5|\xd1\xb4)", - "\xd1\xb7" => "(?:\xd1\xb7|\xd1\xb6)", - "\xd1\xb9" => "(?:\xd1\xb9|\xd1\xb8)", - "\xd1\xbb" => "(?:\xd1\xbb|\xd1\xba)", - "\xd1\xbd" => "(?:\xd1\xbd|\xd1\xbc)", - "\xd1\xbf" => "(?:\xd1\xbf|\xd1\xbe)", - "\xd2\x81" => "(?:\xd2\x81|\xd2\x80)", - "\xd2\x8d" => "(?:\xd2\x8d|\xd2\x8c)", - "\xd2\x8f" => "(?:\xd2\x8f|\xd2\x8e)", - "\xd2\x91" => "(?:\xd2\x91|\xd2\x90)", - "\xd2\x93" => "(?:\xd2\x93|\xd2\x92)", - "\xd2\x95" => "(?:\xd2\x95|\xd2\x94)", - "\xd2\x97" => "(?:\xd2\x97|\xd2\x96)", - "\xd2\x99" => "(?:\xd2\x99|\xd2\x98)", - "\xd2\x9b" => "(?:\xd2\x9b|\xd2\x9a)", - "\xd2\x9d" => "(?:\xd2\x9d|\xd2\x9c)", - "\xd2\x9f" => "(?:\xd2\x9f|\xd2\x9e)", - "\xd2\xa1" => "(?:\xd2\xa1|\xd2\xa0)", - "\xd2\xa3" => "(?:\xd2\xa3|\xd2\xa2)", - "\xd2\xa5" => "(?:\xd2\xa5|\xd2\xa4)", - "\xd2\xa7" => "(?:\xd2\xa7|\xd2\xa6)", - "\xd2\xa9" => "(?:\xd2\xa9|\xd2\xa8)", - "\xd2\xab" => "(?:\xd2\xab|\xd2\xaa)", - "\xd2\xad" => "(?:\xd2\xad|\xd2\xac)", - "\xd2\xaf" => "(?:\xd2\xaf|\xd2\xae)", - "\xd2\xb1" => "(?:\xd2\xb1|\xd2\xb0)", - "\xd2\xb3" => "(?:\xd2\xb3|\xd2\xb2)", - "\xd2\xb5" => "(?:\xd2\xb5|\xd2\xb4)", - "\xd2\xb7" => "(?:\xd2\xb7|\xd2\xb6)", - "\xd2\xb9" => "(?:\xd2\xb9|\xd2\xb8)", - "\xd2\xbb" => "(?:\xd2\xbb|\xd2\xba)", - "\xd2\xbd" => "(?:\xd2\xbd|\xd2\xbc)", - "\xd2\xbf" => "(?:\xd2\xbf|\xd2\xbe)", - "\xd3\x82" => "(?:\xd3\x82|\xd3\x81)", - "\xd3\x84" => "(?:\xd3\x84|\xd3\x83)", - "\xd3\x88" => "(?:\xd3\x88|\xd3\x87)", - "\xd3\x8c" => "(?:\xd3\x8c|\xd3\x8b)", - "\xd3\x91" => "(?:\xd3\x91|\xd3\x90)", - "\xd3\x93" => "(?:\xd3\x93|\xd3\x92)", - "\xd3\x95" => "(?:\xd3\x95|\xd3\x94)", - "\xd3\x97" => "(?:\xd3\x97|\xd3\x96)", - "\xd3\x99" => "(?:\xd3\x99|\xd3\x98)", - "\xd3\x9b" => "(?:\xd3\x9b|\xd3\x9a)", - "\xd3\x9d" => "(?:\xd3\x9d|\xd3\x9c)", - "\xd3\x9f" => "(?:\xd3\x9f|\xd3\x9e)", - "\xd3\xa1" => "(?:\xd3\xa1|\xd3\xa0)", - "\xd3\xa3" => "(?:\xd3\xa3|\xd3\xa2)", - "\xd3\xa5" => "(?:\xd3\xa5|\xd3\xa4)", - "\xd3\xa7" => "(?:\xd3\xa7|\xd3\xa6)", - "\xd3\xa9" => "(?:\xd3\xa9|\xd3\xa8)", - "\xd3\xab" => "(?:\xd3\xab|\xd3\xaa)", - "\xd3\xad" => "(?:\xd3\xad|\xd3\xac)", - "\xd3\xaf" => "(?:\xd3\xaf|\xd3\xae)", - "\xd3\xb1" => "(?:\xd3\xb1|\xd3\xb0)", - "\xd3\xb3" => "(?:\xd3\xb3|\xd3\xb2)", - "\xd3\xb5" => "(?:\xd3\xb5|\xd3\xb4)", - "\xd3\xb9" => "(?:\xd3\xb9|\xd3\xb8)", - "\xd5\xa1" => "(?:\xd5\xa1|\xd4\xb1)", - "\xd5\xa2" => "(?:\xd5\xa2|\xd4\xb2)", - "\xd5\xa3" => "(?:\xd5\xa3|\xd4\xb3)", - "\xd5\xa4" => "(?:\xd5\xa4|\xd4\xb4)", - "\xd5\xa5" => "(?:\xd5\xa5|\xd4\xb5)", - "\xd5\xa6" => "(?:\xd5\xa6|\xd4\xb6)", - "\xd5\xa7" => "(?:\xd5\xa7|\xd4\xb7)", - "\xd5\xa8" => "(?:\xd5\xa8|\xd4\xb8)", - "\xd5\xa9" => "(?:\xd5\xa9|\xd4\xb9)", - "\xd5\xaa" => "(?:\xd5\xaa|\xd4\xba)", - "\xd5\xab" => "(?:\xd5\xab|\xd4\xbb)", - "\xd5\xac" => "(?:\xd5\xac|\xd4\xbc)", - "\xd5\xad" => "(?:\xd5\xad|\xd4\xbd)", - "\xd5\xae" => "(?:\xd5\xae|\xd4\xbe)", - "\xd5\xaf" => "(?:\xd5\xaf|\xd4\xbf)", - "\xd5\xb0" => "(?:\xd5\xb0|\xd5\x80)", - "\xd5\xb1" => "(?:\xd5\xb1|\xd5\x81)", - "\xd5\xb2" => "(?:\xd5\xb2|\xd5\x82)", - "\xd5\xb3" => "(?:\xd5\xb3|\xd5\x83)", - "\xd5\xb4" => "(?:\xd5\xb4|\xd5\x84)", - "\xd5\xb5" => "(?:\xd5\xb5|\xd5\x85)", - "\xd5\xb6" => "(?:\xd5\xb6|\xd5\x86)", - "\xd5\xb7" => "(?:\xd5\xb7|\xd5\x87)", - "\xd5\xb8" => "(?:\xd5\xb8|\xd5\x88)", - "\xd5\xb9" => "(?:\xd5\xb9|\xd5\x89)", - "\xd5\xba" => "(?:\xd5\xba|\xd5\x8a)", - "\xd5\xbb" => "(?:\xd5\xbb|\xd5\x8b)", - "\xd5\xbc" => "(?:\xd5\xbc|\xd5\x8c)", - "\xd5\xbd" => "(?:\xd5\xbd|\xd5\x8d)", - "\xd5\xbe" => "(?:\xd5\xbe|\xd5\x8e)", - "\xd5\xbf" => "(?:\xd5\xbf|\xd5\x8f)", - "\xd6\x80" => "(?:\xd6\x80|\xd5\x90)", - "\xd6\x81" => "(?:\xd6\x81|\xd5\x91)", - "\xd6\x82" => "(?:\xd6\x82|\xd5\x92)", - "\xd6\x83" => "(?:\xd6\x83|\xd5\x93)", - "\xd6\x84" => "(?:\xd6\x84|\xd5\x94)", - "\xd6\x85" => "(?:\xd6\x85|\xd5\x95)", - "\xd6\x86" => "(?:\xd6\x86|\xd5\x96)", - "\xe1\xb8\x81" => "(?:\xe1\xb8\x81|\xe1\xb8\x80)", - "\xe1\xb8\x83" => "(?:\xe1\xb8\x83|\xe1\xb8\x82)", - "\xe1\xb8\x85" => "(?:\xe1\xb8\x85|\xe1\xb8\x84)", - "\xe1\xb8\x87" => "(?:\xe1\xb8\x87|\xe1\xb8\x86)", - "\xe1\xb8\x89" => "(?:\xe1\xb8\x89|\xe1\xb8\x88)", - "\xe1\xb8\x8b" => "(?:\xe1\xb8\x8b|\xe1\xb8\x8a)", - "\xe1\xb8\x8d" => "(?:\xe1\xb8\x8d|\xe1\xb8\x8c)", - "\xe1\xb8\x8f" => "(?:\xe1\xb8\x8f|\xe1\xb8\x8e)", - "\xe1\xb8\x91" => "(?:\xe1\xb8\x91|\xe1\xb8\x90)", - "\xe1\xb8\x93" => "(?:\xe1\xb8\x93|\xe1\xb8\x92)", - "\xe1\xb8\x95" => "(?:\xe1\xb8\x95|\xe1\xb8\x94)", - "\xe1\xb8\x97" => "(?:\xe1\xb8\x97|\xe1\xb8\x96)", - "\xe1\xb8\x99" => "(?:\xe1\xb8\x99|\xe1\xb8\x98)", - "\xe1\xb8\x9b" => "(?:\xe1\xb8\x9b|\xe1\xb8\x9a)", - "\xe1\xb8\x9d" => "(?:\xe1\xb8\x9d|\xe1\xb8\x9c)", - "\xe1\xb8\x9f" => "(?:\xe1\xb8\x9f|\xe1\xb8\x9e)", - "\xe1\xb8\xa1" => "(?:\xe1\xb8\xa1|\xe1\xb8\xa0)", - "\xe1\xb8\xa3" => "(?:\xe1\xb8\xa3|\xe1\xb8\xa2)", - "\xe1\xb8\xa5" => "(?:\xe1\xb8\xa5|\xe1\xb8\xa4)", - "\xe1\xb8\xa7" => "(?:\xe1\xb8\xa7|\xe1\xb8\xa6)", - "\xe1\xb8\xa9" => "(?:\xe1\xb8\xa9|\xe1\xb8\xa8)", - "\xe1\xb8\xab" => "(?:\xe1\xb8\xab|\xe1\xb8\xaa)", - "\xe1\xb8\xad" => "(?:\xe1\xb8\xad|\xe1\xb8\xac)", - "\xe1\xb8\xaf" => "(?:\xe1\xb8\xaf|\xe1\xb8\xae)", - "\xe1\xb8\xb1" => "(?:\xe1\xb8\xb1|\xe1\xb8\xb0)", - "\xe1\xb8\xb3" => "(?:\xe1\xb8\xb3|\xe1\xb8\xb2)", - "\xe1\xb8\xb5" => "(?:\xe1\xb8\xb5|\xe1\xb8\xb4)", - "\xe1\xb8\xb7" => "(?:\xe1\xb8\xb7|\xe1\xb8\xb6)", - "\xe1\xb8\xb9" => "(?:\xe1\xb8\xb9|\xe1\xb8\xb8)", - "\xe1\xb8\xbb" => "(?:\xe1\xb8\xbb|\xe1\xb8\xba)", - "\xe1\xb8\xbd" => "(?:\xe1\xb8\xbd|\xe1\xb8\xbc)", - "\xe1\xb8\xbf" => "(?:\xe1\xb8\xbf|\xe1\xb8\xbe)", - "\xe1\xb9\x81" => "(?:\xe1\xb9\x81|\xe1\xb9\x80)", - "\xe1\xb9\x83" => "(?:\xe1\xb9\x83|\xe1\xb9\x82)", - "\xe1\xb9\x85" => "(?:\xe1\xb9\x85|\xe1\xb9\x84)", - "\xe1\xb9\x87" => "(?:\xe1\xb9\x87|\xe1\xb9\x86)", - "\xe1\xb9\x89" => "(?:\xe1\xb9\x89|\xe1\xb9\x88)", - "\xe1\xb9\x8b" => "(?:\xe1\xb9\x8b|\xe1\xb9\x8a)", - "\xe1\xb9\x8d" => "(?:\xe1\xb9\x8d|\xe1\xb9\x8c)", - "\xe1\xb9\x8f" => "(?:\xe1\xb9\x8f|\xe1\xb9\x8e)", - "\xe1\xb9\x91" => "(?:\xe1\xb9\x91|\xe1\xb9\x90)", - "\xe1\xb9\x93" => "(?:\xe1\xb9\x93|\xe1\xb9\x92)", - "\xe1\xb9\x95" => "(?:\xe1\xb9\x95|\xe1\xb9\x94)", - "\xe1\xb9\x97" => "(?:\xe1\xb9\x97|\xe1\xb9\x96)", - "\xe1\xb9\x99" => "(?:\xe1\xb9\x99|\xe1\xb9\x98)", - "\xe1\xb9\x9b" => "(?:\xe1\xb9\x9b|\xe1\xb9\x9a)", - "\xe1\xb9\x9d" => "(?:\xe1\xb9\x9d|\xe1\xb9\x9c)", - "\xe1\xb9\x9f" => "(?:\xe1\xb9\x9f|\xe1\xb9\x9e)", - "\xe1\xb9\xa1" => "(?:\xe1\xb9\xa1|\xe1\xb9\xa0)", - "\xe1\xb9\xa3" => "(?:\xe1\xb9\xa3|\xe1\xb9\xa2)", - "\xe1\xb9\xa5" => "(?:\xe1\xb9\xa5|\xe1\xb9\xa4)", - "\xe1\xb9\xa7" => "(?:\xe1\xb9\xa7|\xe1\xb9\xa6)", - "\xe1\xb9\xa9" => "(?:\xe1\xb9\xa9|\xe1\xb9\xa8)", - "\xe1\xb9\xab" => "(?:\xe1\xb9\xab|\xe1\xb9\xaa)", - "\xe1\xb9\xad" => "(?:\xe1\xb9\xad|\xe1\xb9\xac)", - "\xe1\xb9\xaf" => "(?:\xe1\xb9\xaf|\xe1\xb9\xae)", - "\xe1\xb9\xb1" => "(?:\xe1\xb9\xb1|\xe1\xb9\xb0)", - "\xe1\xb9\xb3" => "(?:\xe1\xb9\xb3|\xe1\xb9\xb2)", - "\xe1\xb9\xb5" => "(?:\xe1\xb9\xb5|\xe1\xb9\xb4)", - "\xe1\xb9\xb7" => "(?:\xe1\xb9\xb7|\xe1\xb9\xb6)", - "\xe1\xb9\xb9" => "(?:\xe1\xb9\xb9|\xe1\xb9\xb8)", - "\xe1\xb9\xbb" => "(?:\xe1\xb9\xbb|\xe1\xb9\xba)", - "\xe1\xb9\xbd" => "(?:\xe1\xb9\xbd|\xe1\xb9\xbc)", - "\xe1\xb9\xbf" => "(?:\xe1\xb9\xbf|\xe1\xb9\xbe)", - "\xe1\xba\x81" => "(?:\xe1\xba\x81|\xe1\xba\x80)", - "\xe1\xba\x83" => "(?:\xe1\xba\x83|\xe1\xba\x82)", - "\xe1\xba\x85" => "(?:\xe1\xba\x85|\xe1\xba\x84)", - "\xe1\xba\x87" => "(?:\xe1\xba\x87|\xe1\xba\x86)", - "\xe1\xba\x89" => "(?:\xe1\xba\x89|\xe1\xba\x88)", - "\xe1\xba\x8b" => "(?:\xe1\xba\x8b|\xe1\xba\x8a)", - "\xe1\xba\x8d" => "(?:\xe1\xba\x8d|\xe1\xba\x8c)", - "\xe1\xba\x8f" => "(?:\xe1\xba\x8f|\xe1\xba\x8e)", - "\xe1\xba\x91" => "(?:\xe1\xba\x91|\xe1\xba\x90)", - "\xe1\xba\x93" => "(?:\xe1\xba\x93|\xe1\xba\x92)", - "\xe1\xba\x95" => "(?:\xe1\xba\x95|\xe1\xba\x94)", - "\xe1\xba\x9b" => "(?:\xe1\xba\x9b|\xe1\xb9\xa0)", - "\xe1\xba\xa1" => "(?:\xe1\xba\xa1|\xe1\xba\xa0)", - "\xe1\xba\xa3" => "(?:\xe1\xba\xa3|\xe1\xba\xa2)", - "\xe1\xba\xa5" => "(?:\xe1\xba\xa5|\xe1\xba\xa4)", - "\xe1\xba\xa7" => "(?:\xe1\xba\xa7|\xe1\xba\xa6)", - "\xe1\xba\xa9" => "(?:\xe1\xba\xa9|\xe1\xba\xa8)", - "\xe1\xba\xab" => "(?:\xe1\xba\xab|\xe1\xba\xaa)", - "\xe1\xba\xad" => "(?:\xe1\xba\xad|\xe1\xba\xac)", - "\xe1\xba\xaf" => "(?:\xe1\xba\xaf|\xe1\xba\xae)", - "\xe1\xba\xb1" => "(?:\xe1\xba\xb1|\xe1\xba\xb0)", - "\xe1\xba\xb3" => "(?:\xe1\xba\xb3|\xe1\xba\xb2)", - "\xe1\xba\xb5" => "(?:\xe1\xba\xb5|\xe1\xba\xb4)", - "\xe1\xba\xb7" => "(?:\xe1\xba\xb7|\xe1\xba\xb6)", - "\xe1\xba\xb9" => "(?:\xe1\xba\xb9|\xe1\xba\xb8)", - "\xe1\xba\xbb" => "(?:\xe1\xba\xbb|\xe1\xba\xba)", - "\xe1\xba\xbd" => "(?:\xe1\xba\xbd|\xe1\xba\xbc)", - "\xe1\xba\xbf" => "(?:\xe1\xba\xbf|\xe1\xba\xbe)", - "\xe1\xbb\x81" => "(?:\xe1\xbb\x81|\xe1\xbb\x80)", - "\xe1\xbb\x83" => "(?:\xe1\xbb\x83|\xe1\xbb\x82)", - "\xe1\xbb\x85" => "(?:\xe1\xbb\x85|\xe1\xbb\x84)", - "\xe1\xbb\x87" => "(?:\xe1\xbb\x87|\xe1\xbb\x86)", - "\xe1\xbb\x89" => "(?:\xe1\xbb\x89|\xe1\xbb\x88)", - "\xe1\xbb\x8b" => "(?:\xe1\xbb\x8b|\xe1\xbb\x8a)", - "\xe1\xbb\x8d" => "(?:\xe1\xbb\x8d|\xe1\xbb\x8c)", - "\xe1\xbb\x8f" => "(?:\xe1\xbb\x8f|\xe1\xbb\x8e)", - "\xe1\xbb\x91" => "(?:\xe1\xbb\x91|\xe1\xbb\x90)", - "\xe1\xbb\x93" => "(?:\xe1\xbb\x93|\xe1\xbb\x92)", - "\xe1\xbb\x95" => "(?:\xe1\xbb\x95|\xe1\xbb\x94)", - "\xe1\xbb\x97" => "(?:\xe1\xbb\x97|\xe1\xbb\x96)", - "\xe1\xbb\x99" => "(?:\xe1\xbb\x99|\xe1\xbb\x98)", - "\xe1\xbb\x9b" => "(?:\xe1\xbb\x9b|\xe1\xbb\x9a)", - "\xe1\xbb\x9d" => "(?:\xe1\xbb\x9d|\xe1\xbb\x9c)", - "\xe1\xbb\x9f" => "(?:\xe1\xbb\x9f|\xe1\xbb\x9e)", - "\xe1\xbb\xa1" => "(?:\xe1\xbb\xa1|\xe1\xbb\xa0)", - "\xe1\xbb\xa3" => "(?:\xe1\xbb\xa3|\xe1\xbb\xa2)", - "\xe1\xbb\xa5" => "(?:\xe1\xbb\xa5|\xe1\xbb\xa4)", - "\xe1\xbb\xa7" => "(?:\xe1\xbb\xa7|\xe1\xbb\xa6)", - "\xe1\xbb\xa9" => "(?:\xe1\xbb\xa9|\xe1\xbb\xa8)", - "\xe1\xbb\xab" => "(?:\xe1\xbb\xab|\xe1\xbb\xaa)", - "\xe1\xbb\xad" => "(?:\xe1\xbb\xad|\xe1\xbb\xac)", - "\xe1\xbb\xaf" => "(?:\xe1\xbb\xaf|\xe1\xbb\xae)", - "\xe1\xbb\xb1" => "(?:\xe1\xbb\xb1|\xe1\xbb\xb0)", - "\xe1\xbb\xb3" => "(?:\xe1\xbb\xb3|\xe1\xbb\xb2)", - "\xe1\xbb\xb5" => "(?:\xe1\xbb\xb5|\xe1\xbb\xb4)", - "\xe1\xbb\xb7" => "(?:\xe1\xbb\xb7|\xe1\xbb\xb6)", - "\xe1\xbb\xb9" => "(?:\xe1\xbb\xb9|\xe1\xbb\xb8)", - "\xe1\xbc\x80" => "(?:\xe1\xbc\x80|\xe1\xbc\x88)", - "\xe1\xbc\x81" => "(?:\xe1\xbc\x81|\xe1\xbc\x89)", - "\xe1\xbc\x82" => "(?:\xe1\xbc\x82|\xe1\xbc\x8a)", - "\xe1\xbc\x83" => "(?:\xe1\xbc\x83|\xe1\xbc\x8b)", - "\xe1\xbc\x84" => "(?:\xe1\xbc\x84|\xe1\xbc\x8c)", - "\xe1\xbc\x85" => "(?:\xe1\xbc\x85|\xe1\xbc\x8d)", - "\xe1\xbc\x86" => "(?:\xe1\xbc\x86|\xe1\xbc\x8e)", - "\xe1\xbc\x87" => "(?:\xe1\xbc\x87|\xe1\xbc\x8f)", - "\xe1\xbc\x90" => "(?:\xe1\xbc\x90|\xe1\xbc\x98)", - "\xe1\xbc\x91" => "(?:\xe1\xbc\x91|\xe1\xbc\x99)", - "\xe1\xbc\x92" => "(?:\xe1\xbc\x92|\xe1\xbc\x9a)", - "\xe1\xbc\x93" => "(?:\xe1\xbc\x93|\xe1\xbc\x9b)", - "\xe1\xbc\x94" => "(?:\xe1\xbc\x94|\xe1\xbc\x9c)", - "\xe1\xbc\x95" => "(?:\xe1\xbc\x95|\xe1\xbc\x9d)", - "\xe1\xbc\xa0" => "(?:\xe1\xbc\xa0|\xe1\xbc\xa8)", - "\xe1\xbc\xa1" => "(?:\xe1\xbc\xa1|\xe1\xbc\xa9)", - "\xe1\xbc\xa2" => "(?:\xe1\xbc\xa2|\xe1\xbc\xaa)", - "\xe1\xbc\xa3" => "(?:\xe1\xbc\xa3|\xe1\xbc\xab)", - "\xe1\xbc\xa4" => "(?:\xe1\xbc\xa4|\xe1\xbc\xac)", - "\xe1\xbc\xa5" => "(?:\xe1\xbc\xa5|\xe1\xbc\xad)", - "\xe1\xbc\xa6" => "(?:\xe1\xbc\xa6|\xe1\xbc\xae)", - "\xe1\xbc\xa7" => "(?:\xe1\xbc\xa7|\xe1\xbc\xaf)", - "\xe1\xbc\xb0" => "(?:\xe1\xbc\xb0|\xe1\xbc\xb8)", - "\xe1\xbc\xb1" => "(?:\xe1\xbc\xb1|\xe1\xbc\xb9)", - "\xe1\xbc\xb2" => "(?:\xe1\xbc\xb2|\xe1\xbc\xba)", - "\xe1\xbc\xb3" => "(?:\xe1\xbc\xb3|\xe1\xbc\xbb)", - "\xe1\xbc\xb4" => "(?:\xe1\xbc\xb4|\xe1\xbc\xbc)", - "\xe1\xbc\xb5" => "(?:\xe1\xbc\xb5|\xe1\xbc\xbd)", - "\xe1\xbc\xb6" => "(?:\xe1\xbc\xb6|\xe1\xbc\xbe)", - "\xe1\xbc\xb7" => "(?:\xe1\xbc\xb7|\xe1\xbc\xbf)", - "\xe1\xbd\x80" => "(?:\xe1\xbd\x80|\xe1\xbd\x88)", - "\xe1\xbd\x81" => "(?:\xe1\xbd\x81|\xe1\xbd\x89)", - "\xe1\xbd\x82" => "(?:\xe1\xbd\x82|\xe1\xbd\x8a)", - "\xe1\xbd\x83" => "(?:\xe1\xbd\x83|\xe1\xbd\x8b)", - "\xe1\xbd\x84" => "(?:\xe1\xbd\x84|\xe1\xbd\x8c)", - "\xe1\xbd\x85" => "(?:\xe1\xbd\x85|\xe1\xbd\x8d)", - "\xe1\xbd\x91" => "(?:\xe1\xbd\x91|\xe1\xbd\x99)", - "\xe1\xbd\x93" => "(?:\xe1\xbd\x93|\xe1\xbd\x9b)", - "\xe1\xbd\x95" => "(?:\xe1\xbd\x95|\xe1\xbd\x9d)", - "\xe1\xbd\x97" => "(?:\xe1\xbd\x97|\xe1\xbd\x9f)", - "\xe1\xbd\xa0" => "(?:\xe1\xbd\xa0|\xe1\xbd\xa8)", - "\xe1\xbd\xa1" => "(?:\xe1\xbd\xa1|\xe1\xbd\xa9)", - "\xe1\xbd\xa2" => "(?:\xe1\xbd\xa2|\xe1\xbd\xaa)", - "\xe1\xbd\xa3" => "(?:\xe1\xbd\xa3|\xe1\xbd\xab)", - "\xe1\xbd\xa4" => "(?:\xe1\xbd\xa4|\xe1\xbd\xac)", - "\xe1\xbd\xa5" => "(?:\xe1\xbd\xa5|\xe1\xbd\xad)", - "\xe1\xbd\xa6" => "(?:\xe1\xbd\xa6|\xe1\xbd\xae)", - "\xe1\xbd\xa7" => "(?:\xe1\xbd\xa7|\xe1\xbd\xaf)", - "\xe1\xbd\xb0" => "(?:\xe1\xbd\xb0|\xe1\xbe\xba)", - "\xe1\xbd\xb1" => "(?:\xe1\xbd\xb1|\xe1\xbe\xbb)", - "\xe1\xbd\xb2" => "(?:\xe1\xbd\xb2|\xe1\xbf\x88)", - "\xe1\xbd\xb3" => "(?:\xe1\xbd\xb3|\xe1\xbf\x89)", - "\xe1\xbd\xb4" => "(?:\xe1\xbd\xb4|\xe1\xbf\x8a)", - "\xe1\xbd\xb5" => "(?:\xe1\xbd\xb5|\xe1\xbf\x8b)", - "\xe1\xbd\xb6" => "(?:\xe1\xbd\xb6|\xe1\xbf\x9a)", - "\xe1\xbd\xb7" => "(?:\xe1\xbd\xb7|\xe1\xbf\x9b)", - "\xe1\xbd\xb8" => "(?:\xe1\xbd\xb8|\xe1\xbf\xb8)", - "\xe1\xbd\xb9" => "(?:\xe1\xbd\xb9|\xe1\xbf\xb9)", - "\xe1\xbd\xba" => "(?:\xe1\xbd\xba|\xe1\xbf\xaa)", - "\xe1\xbd\xbb" => "(?:\xe1\xbd\xbb|\xe1\xbf\xab)", - "\xe1\xbd\xbc" => "(?:\xe1\xbd\xbc|\xe1\xbf\xba)", - "\xe1\xbd\xbd" => "(?:\xe1\xbd\xbd|\xe1\xbf\xbb)", - "\xe1\xbe\x80" => "(?:\xe1\xbe\x80|\xe1\xbe\x88)", - "\xe1\xbe\x81" => "(?:\xe1\xbe\x81|\xe1\xbe\x89)", - "\xe1\xbe\x82" => "(?:\xe1\xbe\x82|\xe1\xbe\x8a)", - "\xe1\xbe\x83" => "(?:\xe1\xbe\x83|\xe1\xbe\x8b)", - "\xe1\xbe\x84" => "(?:\xe1\xbe\x84|\xe1\xbe\x8c)", - "\xe1\xbe\x85" => "(?:\xe1\xbe\x85|\xe1\xbe\x8d)", - "\xe1\xbe\x86" => "(?:\xe1\xbe\x86|\xe1\xbe\x8e)", - "\xe1\xbe\x87" => "(?:\xe1\xbe\x87|\xe1\xbe\x8f)", - "\xe1\xbe\x90" => "(?:\xe1\xbe\x90|\xe1\xbe\x98)", - "\xe1\xbe\x91" => "(?:\xe1\xbe\x91|\xe1\xbe\x99)", - "\xe1\xbe\x92" => "(?:\xe1\xbe\x92|\xe1\xbe\x9a)", - "\xe1\xbe\x93" => "(?:\xe1\xbe\x93|\xe1\xbe\x9b)", - "\xe1\xbe\x94" => "(?:\xe1\xbe\x94|\xe1\xbe\x9c)", - "\xe1\xbe\x95" => "(?:\xe1\xbe\x95|\xe1\xbe\x9d)", - "\xe1\xbe\x96" => "(?:\xe1\xbe\x96|\xe1\xbe\x9e)", - "\xe1\xbe\x97" => "(?:\xe1\xbe\x97|\xe1\xbe\x9f)", - "\xe1\xbe\xa0" => "(?:\xe1\xbe\xa0|\xe1\xbe\xa8)", - "\xe1\xbe\xa1" => "(?:\xe1\xbe\xa1|\xe1\xbe\xa9)", - "\xe1\xbe\xa2" => "(?:\xe1\xbe\xa2|\xe1\xbe\xaa)", - "\xe1\xbe\xa3" => "(?:\xe1\xbe\xa3|\xe1\xbe\xab)", - "\xe1\xbe\xa4" => "(?:\xe1\xbe\xa4|\xe1\xbe\xac)", - "\xe1\xbe\xa5" => "(?:\xe1\xbe\xa5|\xe1\xbe\xad)", - "\xe1\xbe\xa6" => "(?:\xe1\xbe\xa6|\xe1\xbe\xae)", - "\xe1\xbe\xa7" => "(?:\xe1\xbe\xa7|\xe1\xbe\xaf)", - "\xe1\xbe\xb0" => "(?:\xe1\xbe\xb0|\xe1\xbe\xb8)", - "\xe1\xbe\xb1" => "(?:\xe1\xbe\xb1|\xe1\xbe\xb9)", - "\xe1\xbe\xb3" => "(?:\xe1\xbe\xb3|\xe1\xbe\xbc)", - "\xe1\xbe\xbe" => "(?:\xe1\xbe\xbe|\xce\x99)", - "\xe1\xbf\x83" => "(?:\xe1\xbf\x83|\xe1\xbf\x8c)", - "\xe1\xbf\x90" => "(?:\xe1\xbf\x90|\xe1\xbf\x98)", - "\xe1\xbf\x91" => "(?:\xe1\xbf\x91|\xe1\xbf\x99)", - "\xe1\xbf\xa0" => "(?:\xe1\xbf\xa0|\xe1\xbf\xa8)", - "\xe1\xbf\xa1" => "(?:\xe1\xbf\xa1|\xe1\xbf\xa9)", - "\xe1\xbf\xa5" => "(?:\xe1\xbf\xa5|\xe1\xbf\xac)", - "\xe1\xbf\xb3" => "(?:\xe1\xbf\xb3|\xe1\xbf\xbc)", - "\xe2\x85\xb0" => "(?:\xe2\x85\xb0|\xe2\x85\xa0)", - "\xe2\x85\xb1" => "(?:\xe2\x85\xb1|\xe2\x85\xa1)", - "\xe2\x85\xb2" => "(?:\xe2\x85\xb2|\xe2\x85\xa2)", - "\xe2\x85\xb3" => "(?:\xe2\x85\xb3|\xe2\x85\xa3)", - "\xe2\x85\xb4" => "(?:\xe2\x85\xb4|\xe2\x85\xa4)", - "\xe2\x85\xb5" => "(?:\xe2\x85\xb5|\xe2\x85\xa5)", - "\xe2\x85\xb6" => "(?:\xe2\x85\xb6|\xe2\x85\xa6)", - "\xe2\x85\xb7" => "(?:\xe2\x85\xb7|\xe2\x85\xa7)", - "\xe2\x85\xb8" => "(?:\xe2\x85\xb8|\xe2\x85\xa8)", - "\xe2\x85\xb9" => "(?:\xe2\x85\xb9|\xe2\x85\xa9)", - "\xe2\x85\xba" => "(?:\xe2\x85\xba|\xe2\x85\xaa)", - "\xe2\x85\xbb" => "(?:\xe2\x85\xbb|\xe2\x85\xab)", - "\xe2\x85\xbc" => "(?:\xe2\x85\xbc|\xe2\x85\xac)", - "\xe2\x85\xbd" => "(?:\xe2\x85\xbd|\xe2\x85\xad)", - "\xe2\x85\xbe" => "(?:\xe2\x85\xbe|\xe2\x85\xae)", - "\xe2\x85\xbf" => "(?:\xe2\x85\xbf|\xe2\x85\xaf)", - "\xe2\x93\x90" => "(?:\xe2\x93\x90|\xe2\x92\xb6)", - "\xe2\x93\x91" => "(?:\xe2\x93\x91|\xe2\x92\xb7)", - "\xe2\x93\x92" => "(?:\xe2\x93\x92|\xe2\x92\xb8)", - "\xe2\x93\x93" => "(?:\xe2\x93\x93|\xe2\x92\xb9)", - "\xe2\x93\x94" => "(?:\xe2\x93\x94|\xe2\x92\xba)", - "\xe2\x93\x95" => "(?:\xe2\x93\x95|\xe2\x92\xbb)", - "\xe2\x93\x96" => "(?:\xe2\x93\x96|\xe2\x92\xbc)", - "\xe2\x93\x97" => "(?:\xe2\x93\x97|\xe2\x92\xbd)", - "\xe2\x93\x98" => "(?:\xe2\x93\x98|\xe2\x92\xbe)", - "\xe2\x93\x99" => "(?:\xe2\x93\x99|\xe2\x92\xbf)", - "\xe2\x93\x9a" => "(?:\xe2\x93\x9a|\xe2\x93\x80)", - "\xe2\x93\x9b" => "(?:\xe2\x93\x9b|\xe2\x93\x81)", - "\xe2\x93\x9c" => "(?:\xe2\x93\x9c|\xe2\x93\x82)", - "\xe2\x93\x9d" => "(?:\xe2\x93\x9d|\xe2\x93\x83)", - "\xe2\x93\x9e" => "(?:\xe2\x93\x9e|\xe2\x93\x84)", - "\xe2\x93\x9f" => "(?:\xe2\x93\x9f|\xe2\x93\x85)", - "\xe2\x93\xa0" => "(?:\xe2\x93\xa0|\xe2\x93\x86)", - "\xe2\x93\xa1" => "(?:\xe2\x93\xa1|\xe2\x93\x87)", - "\xe2\x93\xa2" => "(?:\xe2\x93\xa2|\xe2\x93\x88)", - "\xe2\x93\xa3" => "(?:\xe2\x93\xa3|\xe2\x93\x89)", - "\xe2\x93\xa4" => "(?:\xe2\x93\xa4|\xe2\x93\x8a)", - "\xe2\x93\xa5" => "(?:\xe2\x93\xa5|\xe2\x93\x8b)", - "\xe2\x93\xa6" => "(?:\xe2\x93\xa6|\xe2\x93\x8c)", - "\xe2\x93\xa7" => "(?:\xe2\x93\xa7|\xe2\x93\x8d)", - "\xe2\x93\xa8" => "(?:\xe2\x93\xa8|\xe2\x93\x8e)", - "\xe2\x93\xa9" => "(?:\xe2\x93\xa9|\xe2\x93\x8f)", - "\xef\xbd\x81" => "(?:\xef\xbd\x81|\xef\xbc\xa1)", - "\xef\xbd\x82" => "(?:\xef\xbd\x82|\xef\xbc\xa2)", - "\xef\xbd\x83" => "(?:\xef\xbd\x83|\xef\xbc\xa3)", - "\xef\xbd\x84" => "(?:\xef\xbd\x84|\xef\xbc\xa4)", - "\xef\xbd\x85" => "(?:\xef\xbd\x85|\xef\xbc\xa5)", - "\xef\xbd\x86" => "(?:\xef\xbd\x86|\xef\xbc\xa6)", - "\xef\xbd\x87" => "(?:\xef\xbd\x87|\xef\xbc\xa7)", - "\xef\xbd\x88" => "(?:\xef\xbd\x88|\xef\xbc\xa8)", - "\xef\xbd\x89" => "(?:\xef\xbd\x89|\xef\xbc\xa9)", - "\xef\xbd\x8a" => "(?:\xef\xbd\x8a|\xef\xbc\xaa)", - "\xef\xbd\x8b" => "(?:\xef\xbd\x8b|\xef\xbc\xab)", - "\xef\xbd\x8c" => "(?:\xef\xbd\x8c|\xef\xbc\xac)", - "\xef\xbd\x8d" => "(?:\xef\xbd\x8d|\xef\xbc\xad)", - "\xef\xbd\x8e" => "(?:\xef\xbd\x8e|\xef\xbc\xae)", - "\xef\xbd\x8f" => "(?:\xef\xbd\x8f|\xef\xbc\xaf)", - "\xef\xbd\x90" => "(?:\xef\xbd\x90|\xef\xbc\xb0)", - "\xef\xbd\x91" => "(?:\xef\xbd\x91|\xef\xbc\xb1)", - "\xef\xbd\x92" => "(?:\xef\xbd\x92|\xef\xbc\xb2)", - "\xef\xbd\x93" => "(?:\xef\xbd\x93|\xef\xbc\xb3)", - "\xef\xbd\x94" => "(?:\xef\xbd\x94|\xef\xbc\xb4)", - "\xef\xbd\x95" => "(?:\xef\xbd\x95|\xef\xbc\xb5)", - "\xef\xbd\x96" => "(?:\xef\xbd\x96|\xef\xbc\xb6)", - "\xef\xbd\x97" => "(?:\xef\xbd\x97|\xef\xbc\xb7)", - "\xef\xbd\x98" => "(?:\xef\xbd\x98|\xef\xbc\xb8)", - "\xef\xbd\x99" => "(?:\xef\xbd\x99|\xef\xbc\xb9)", - "\xef\xbd\x9a" => "(?:\xef\xbd\x9a|\xef\xbc\xba)", - "\xf0\x90\x90\xa8" => "(?:\xf0\x90\x90\xa8|\xf0\x90\x90\x80)", - "\xf0\x90\x90\xa9" => "(?:\xf0\x90\x90\xa9|\xf0\x90\x90\x81)", - "\xf0\x90\x90\xaa" => "(?:\xf0\x90\x90\xaa|\xf0\x90\x90\x82)", - "\xf0\x90\x90\xab" => "(?:\xf0\x90\x90\xab|\xf0\x90\x90\x83)", - "\xf0\x90\x90\xac" => "(?:\xf0\x90\x90\xac|\xf0\x90\x90\x84)", - "\xf0\x90\x90\xad" => "(?:\xf0\x90\x90\xad|\xf0\x90\x90\x85)", - "\xf0\x90\x90\xae" => "(?:\xf0\x90\x90\xae|\xf0\x90\x90\x86)", - "\xf0\x90\x90\xaf" => "(?:\xf0\x90\x90\xaf|\xf0\x90\x90\x87)", - "\xf0\x90\x90\xb0" => "(?:\xf0\x90\x90\xb0|\xf0\x90\x90\x88)", - "\xf0\x90\x90\xb1" => "(?:\xf0\x90\x90\xb1|\xf0\x90\x90\x89)", - "\xf0\x90\x90\xb2" => "(?:\xf0\x90\x90\xb2|\xf0\x90\x90\x8a)", - "\xf0\x90\x90\xb3" => "(?:\xf0\x90\x90\xb3|\xf0\x90\x90\x8b)", - "\xf0\x90\x90\xb4" => "(?:\xf0\x90\x90\xb4|\xf0\x90\x90\x8c)", - "\xf0\x90\x90\xb5" => "(?:\xf0\x90\x90\xb5|\xf0\x90\x90\x8d)", - "\xf0\x90\x90\xb6" => "(?:\xf0\x90\x90\xb6|\xf0\x90\x90\x8e)", - "\xf0\x90\x90\xb7" => "(?:\xf0\x90\x90\xb7|\xf0\x90\x90\x8f)", - "\xf0\x90\x90\xb8" => "(?:\xf0\x90\x90\xb8|\xf0\x90\x90\x90)", - "\xf0\x90\x90\xb9" => "(?:\xf0\x90\x90\xb9|\xf0\x90\x90\x91)", - "\xf0\x90\x90\xba" => "(?:\xf0\x90\x90\xba|\xf0\x90\x90\x92)", - "\xf0\x90\x90\xbb" => "(?:\xf0\x90\x90\xbb|\xf0\x90\x90\x93)", - "\xf0\x90\x90\xbc" => "(?:\xf0\x90\x90\xbc|\xf0\x90\x90\x94)", - "\xf0\x90\x90\xbd" => "(?:\xf0\x90\x90\xbd|\xf0\x90\x90\x95)", - "\xf0\x90\x90\xbe" => "(?:\xf0\x90\x90\xbe|\xf0\x90\x90\x96)", - "\xf0\x90\x90\xbf" => "(?:\xf0\x90\x90\xbf|\xf0\x90\x90\x97)", - "\xf0\x90\x91\x80" => "(?:\xf0\x90\x91\x80|\xf0\x90\x90\x98)", - "\xf0\x90\x91\x81" => "(?:\xf0\x90\x91\x81|\xf0\x90\x90\x99)", - "\xf0\x90\x91\x82" => "(?:\xf0\x90\x91\x82|\xf0\x90\x90\x9a)", - "\xf0\x90\x91\x83" => "(?:\xf0\x90\x91\x83|\xf0\x90\x90\x9b)", - "\xf0\x90\x91\x84" => "(?:\xf0\x90\x91\x84|\xf0\x90\x90\x9c)", - "\xf0\x90\x91\x85" => "(?:\xf0\x90\x91\x85|\xf0\x90\x90\x9d)", - "\xf0\x90\x91\x86" => "(?:\xf0\x90\x91\x86|\xf0\x90\x90\x9e)", - "\xf0\x90\x91\x87" => "(?:\xf0\x90\x91\x87|\xf0\x90\x90\x9f)", - "\xf0\x90\x91\x88" => "(?:\xf0\x90\x91\x88|\xf0\x90\x90\xa0)", - "\xf0\x90\x91\x89" => "(?:\xf0\x90\x91\x89|\xf0\x90\x90\xa1)", - "\xf0\x90\x91\x8a" => "(?:\xf0\x90\x91\x8a|\xf0\x90\x90\xa2)", - "\xf0\x90\x91\x8b" => "(?:\xf0\x90\x91\x8b|\xf0\x90\x90\xa3)", - "\xf0\x90\x91\x8c" => "(?:\xf0\x90\x91\x8c|\xf0\x90\x90\xa4)", - "\xf0\x90\x91\x8d" => "(?:\xf0\x90\x91\x8d|\xf0\x90\x90\xa5)", - "\xc3\x80" => "(?:\xc3\x80|\xc3\xa0)", - "\xc3\x81" => "(?:\xc3\x81|\xc3\xa1)", - "\xc3\x82" => "(?:\xc3\x82|\xc3\xa2)", - "\xc3\x83" => "(?:\xc3\x83|\xc3\xa3)", - "\xc3\x84" => "(?:\xc3\x84|\xc3\xa4)", - "\xc3\x85" => "(?:\xc3\x85|\xc3\xa5)", - "\xc3\x86" => "(?:\xc3\x86|\xc3\xa6)", - "\xc3\x87" => "(?:\xc3\x87|\xc3\xa7)", - "\xc3\x88" => "(?:\xc3\x88|\xc3\xa8)", - "\xc3\x89" => "(?:\xc3\x89|\xc3\xa9)", - "\xc3\x8a" => "(?:\xc3\x8a|\xc3\xaa)", - "\xc3\x8b" => "(?:\xc3\x8b|\xc3\xab)", - "\xc3\x8c" => "(?:\xc3\x8c|\xc3\xac)", - "\xc3\x8d" => "(?:\xc3\x8d|\xc3\xad)", - "\xc3\x8e" => "(?:\xc3\x8e|\xc3\xae)", - "\xc3\x8f" => "(?:\xc3\x8f|\xc3\xaf)", - "\xc3\x90" => "(?:\xc3\x90|\xc3\xb0)", - "\xc3\x91" => "(?:\xc3\x91|\xc3\xb1)", - "\xc3\x92" => "(?:\xc3\x92|\xc3\xb2)", - "\xc3\x93" => "(?:\xc3\x93|\xc3\xb3)", - "\xc3\x94" => "(?:\xc3\x94|\xc3\xb4)", - "\xc3\x95" => "(?:\xc3\x95|\xc3\xb5)", - "\xc3\x96" => "(?:\xc3\x96|\xc3\xb6)", - "\xc3\x98" => "(?:\xc3\x98|\xc3\xb8)", - "\xc3\x99" => "(?:\xc3\x99|\xc3\xb9)", - "\xc3\x9a" => "(?:\xc3\x9a|\xc3\xba)", - "\xc3\x9b" => "(?:\xc3\x9b|\xc3\xbb)", - "\xc3\x9c" => "(?:\xc3\x9c|\xc3\xbc)", - "\xc3\x9d" => "(?:\xc3\x9d|\xc3\xbd)", - "\xc3\x9e" => "(?:\xc3\x9e|\xc3\xbe)", - "\xc4\x80" => "(?:\xc4\x80|\xc4\x81)", - "\xc4\x82" => "(?:\xc4\x82|\xc4\x83)", - "\xc4\x84" => "(?:\xc4\x84|\xc4\x85)", - "\xc4\x86" => "(?:\xc4\x86|\xc4\x87)", - "\xc4\x88" => "(?:\xc4\x88|\xc4\x89)", - "\xc4\x8a" => "(?:\xc4\x8a|\xc4\x8b)", - "\xc4\x8c" => "(?:\xc4\x8c|\xc4\x8d)", - "\xc4\x8e" => "(?:\xc4\x8e|\xc4\x8f)", - "\xc4\x90" => "(?:\xc4\x90|\xc4\x91)", - "\xc4\x92" => "(?:\xc4\x92|\xc4\x93)", - "\xc4\x94" => "(?:\xc4\x94|\xc4\x95)", - "\xc4\x96" => "(?:\xc4\x96|\xc4\x97)", - "\xc4\x98" => "(?:\xc4\x98|\xc4\x99)", - "\xc4\x9a" => "(?:\xc4\x9a|\xc4\x9b)", - "\xc4\x9c" => "(?:\xc4\x9c|\xc4\x9d)", - "\xc4\x9e" => "(?:\xc4\x9e|\xc4\x9f)", - "\xc4\xa0" => "(?:\xc4\xa0|\xc4\xa1)", - "\xc4\xa2" => "(?:\xc4\xa2|\xc4\xa3)", - "\xc4\xa4" => "(?:\xc4\xa4|\xc4\xa5)", - "\xc4\xa6" => "(?:\xc4\xa6|\xc4\xa7)", - "\xc4\xa8" => "(?:\xc4\xa8|\xc4\xa9)", - "\xc4\xaa" => "(?:\xc4\xaa|\xc4\xab)", - "\xc4\xac" => "(?:\xc4\xac|\xc4\xad)", - "\xc4\xae" => "(?:\xc4\xae|\xc4\xaf)", - "\xc4\xb0" => "(?:\xc4\xb0|i)", - "\xc4\xb2" => "(?:\xc4\xb2|\xc4\xb3)", - "\xc4\xb4" => "(?:\xc4\xb4|\xc4\xb5)", - "\xc4\xb6" => "(?:\xc4\xb6|\xc4\xb7)", - "\xc4\xb9" => "(?:\xc4\xb9|\xc4\xba)", - "\xc4\xbb" => "(?:\xc4\xbb|\xc4\xbc)", - "\xc4\xbd" => "(?:\xc4\xbd|\xc4\xbe)", - "\xc4\xbf" => "(?:\xc4\xbf|\xc5\x80)", - "\xc5\x81" => "(?:\xc5\x81|\xc5\x82)", - "\xc5\x83" => "(?:\xc5\x83|\xc5\x84)", - "\xc5\x85" => "(?:\xc5\x85|\xc5\x86)", - "\xc5\x87" => "(?:\xc5\x87|\xc5\x88)", - "\xc5\x8a" => "(?:\xc5\x8a|\xc5\x8b)", - "\xc5\x8c" => "(?:\xc5\x8c|\xc5\x8d)", - "\xc5\x8e" => "(?:\xc5\x8e|\xc5\x8f)", - "\xc5\x90" => "(?:\xc5\x90|\xc5\x91)", - "\xc5\x92" => "(?:\xc5\x92|\xc5\x93)", - "\xc5\x94" => "(?:\xc5\x94|\xc5\x95)", - "\xc5\x96" => "(?:\xc5\x96|\xc5\x97)", - "\xc5\x98" => "(?:\xc5\x98|\xc5\x99)", - "\xc5\x9a" => "(?:\xc5\x9a|\xc5\x9b)", - "\xc5\x9c" => "(?:\xc5\x9c|\xc5\x9d)", - "\xc5\x9e" => "(?:\xc5\x9e|\xc5\x9f)", - "\xc5\xa0" => "(?:\xc5\xa0|\xc5\xa1)", - "\xc5\xa2" => "(?:\xc5\xa2|\xc5\xa3)", - "\xc5\xa4" => "(?:\xc5\xa4|\xc5\xa5)", - "\xc5\xa6" => "(?:\xc5\xa6|\xc5\xa7)", - "\xc5\xa8" => "(?:\xc5\xa8|\xc5\xa9)", - "\xc5\xaa" => "(?:\xc5\xaa|\xc5\xab)", - "\xc5\xac" => "(?:\xc5\xac|\xc5\xad)", - "\xc5\xae" => "(?:\xc5\xae|\xc5\xaf)", - "\xc5\xb0" => "(?:\xc5\xb0|\xc5\xb1)", - "\xc5\xb2" => "(?:\xc5\xb2|\xc5\xb3)", - "\xc5\xb4" => "(?:\xc5\xb4|\xc5\xb5)", - "\xc5\xb6" => "(?:\xc5\xb6|\xc5\xb7)", - "\xc5\xb8" => "(?:\xc5\xb8|\xc3\xbf)", - "\xc5\xb9" => "(?:\xc5\xb9|\xc5\xba)", - "\xc5\xbb" => "(?:\xc5\xbb|\xc5\xbc)", - "\xc5\xbd" => "(?:\xc5\xbd|\xc5\xbe)", - "\xc6\x81" => "(?:\xc6\x81|\xc9\x93)", - "\xc6\x82" => "(?:\xc6\x82|\xc6\x83)", - "\xc6\x84" => "(?:\xc6\x84|\xc6\x85)", - "\xc6\x86" => "(?:\xc6\x86|\xc9\x94)", - "\xc6\x87" => "(?:\xc6\x87|\xc6\x88)", - "\xc6\x89" => "(?:\xc6\x89|\xc9\x96)", - "\xc6\x8a" => "(?:\xc6\x8a|\xc9\x97)", - "\xc6\x8b" => "(?:\xc6\x8b|\xc6\x8c)", - "\xc6\x8e" => "(?:\xc6\x8e|\xc7\x9d)", - "\xc6\x8f" => "(?:\xc6\x8f|\xc9\x99)", - "\xc6\x90" => "(?:\xc6\x90|\xc9\x9b)", - "\xc6\x91" => "(?:\xc6\x91|\xc6\x92)", - "\xc6\x93" => "(?:\xc6\x93|\xc9\xa0)", - "\xc6\x94" => "(?:\xc6\x94|\xc9\xa3)", - "\xc6\x96" => "(?:\xc6\x96|\xc9\xa9)", - "\xc6\x97" => "(?:\xc6\x97|\xc9\xa8)", - "\xc6\x98" => "(?:\xc6\x98|\xc6\x99)", - "\xc6\x9c" => "(?:\xc6\x9c|\xc9\xaf)", - "\xc6\x9d" => "(?:\xc6\x9d|\xc9\xb2)", - "\xc6\x9f" => "(?:\xc6\x9f|\xc9\xb5)", - "\xc6\xa0" => "(?:\xc6\xa0|\xc6\xa1)", - "\xc6\xa2" => "(?:\xc6\xa2|\xc6\xa3)", - "\xc6\xa4" => "(?:\xc6\xa4|\xc6\xa5)", - "\xc6\xa6" => "(?:\xc6\xa6|\xca\x80)", - "\xc6\xa7" => "(?:\xc6\xa7|\xc6\xa8)", - "\xc6\xa9" => "(?:\xc6\xa9|\xca\x83)", - "\xc6\xac" => "(?:\xc6\xac|\xc6\xad)", - "\xc6\xae" => "(?:\xc6\xae|\xca\x88)", - "\xc6\xaf" => "(?:\xc6\xaf|\xc6\xb0)", - "\xc6\xb1" => "(?:\xc6\xb1|\xca\x8a)", - "\xc6\xb2" => "(?:\xc6\xb2|\xca\x8b)", - "\xc6\xb3" => "(?:\xc6\xb3|\xc6\xb4)", - "\xc6\xb5" => "(?:\xc6\xb5|\xc6\xb6)", - "\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\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)", - "\xc7\x93" => "(?:\xc7\x93|\xc7\x94)", - "\xc7\x95" => "(?:\xc7\x95|\xc7\x96)", - "\xc7\x97" => "(?:\xc7\x97|\xc7\x98)", - "\xc7\x99" => "(?:\xc7\x99|\xc7\x9a)", - "\xc7\x9b" => "(?:\xc7\x9b|\xc7\x9c)", - "\xc7\x9e" => "(?:\xc7\x9e|\xc7\x9f)", - "\xc7\xa0" => "(?:\xc7\xa0|\xc7\xa1)", - "\xc7\xa2" => "(?:\xc7\xa2|\xc7\xa3)", - "\xc7\xa4" => "(?:\xc7\xa4|\xc7\xa5)", - "\xc7\xa6" => "(?:\xc7\xa6|\xc7\xa7)", - "\xc7\xa8" => "(?:\xc7\xa8|\xc7\xa9)", - "\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\xb4" => "(?:\xc7\xb4|\xc7\xb5)", - "\xc7\xb6" => "(?:\xc7\xb6|\xc6\x95)", - "\xc7\xb7" => "(?:\xc7\xb7|\xc6\xbf)", - "\xc7\xb8" => "(?:\xc7\xb8|\xc7\xb9)", - "\xc7\xba" => "(?:\xc7\xba|\xc7\xbb)", - "\xc7\xbc" => "(?:\xc7\xbc|\xc7\xbd)", - "\xc7\xbe" => "(?:\xc7\xbe|\xc7\xbf)", - "\xc8\x80" => "(?:\xc8\x80|\xc8\x81)", - "\xc8\x82" => "(?:\xc8\x82|\xc8\x83)", - "\xc8\x84" => "(?:\xc8\x84|\xc8\x85)", - "\xc8\x86" => "(?:\xc8\x86|\xc8\x87)", - "\xc8\x88" => "(?:\xc8\x88|\xc8\x89)", - "\xc8\x8a" => "(?:\xc8\x8a|\xc8\x8b)", - "\xc8\x8c" => "(?:\xc8\x8c|\xc8\x8d)", - "\xc8\x8e" => "(?:\xc8\x8e|\xc8\x8f)", - "\xc8\x90" => "(?:\xc8\x90|\xc8\x91)", - "\xc8\x92" => "(?:\xc8\x92|\xc8\x93)", - "\xc8\x94" => "(?:\xc8\x94|\xc8\x95)", - "\xc8\x96" => "(?:\xc8\x96|\xc8\x97)", - "\xc8\x98" => "(?:\xc8\x98|\xc8\x99)", - "\xc8\x9a" => "(?:\xc8\x9a|\xc8\x9b)", - "\xc8\x9c" => "(?:\xc8\x9c|\xc8\x9d)", - "\xc8\x9e" => "(?:\xc8\x9e|\xc8\x9f)", - "\xc8\xa2" => "(?:\xc8\xa2|\xc8\xa3)", - "\xc8\xa4" => "(?:\xc8\xa4|\xc8\xa5)", - "\xc8\xa6" => "(?:\xc8\xa6|\xc8\xa7)", - "\xc8\xa8" => "(?:\xc8\xa8|\xc8\xa9)", - "\xc8\xaa" => "(?:\xc8\xaa|\xc8\xab)", - "\xc8\xac" => "(?:\xc8\xac|\xc8\xad)", - "\xc8\xae" => "(?:\xc8\xae|\xc8\xaf)", - "\xc8\xb0" => "(?:\xc8\xb0|\xc8\xb1)", - "\xc8\xb2" => "(?:\xc8\xb2|\xc8\xb3)", - "\xce\x86" => "(?:\xce\x86|\xce\xac)", - "\xce\x88" => "(?:\xce\x88|\xce\xad)", - "\xce\x89" => "(?:\xce\x89|\xce\xae)", - "\xce\x8a" => "(?:\xce\x8a|\xce\xaf)", - "\xce\x8c" => "(?:\xce\x8c|\xcf\x8c)", - "\xce\x8e" => "(?:\xce\x8e|\xcf\x8d)", - "\xce\x8f" => "(?:\xce\x8f|\xcf\x8e)", - "\xce\x91" => "(?:\xce\x91|\xce\xb1)", - "\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|\xcf\xb5)", - "\xce\x96" => "(?:\xce\x96|\xce\xb6)", - "\xce\x97" => "(?:\xce\x97|\xce\xb7)", - "\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|\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|\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|\xcf\x95)", - "\xce\xa7" => "(?:\xce\xa7|\xcf\x87)", - "\xce\xa8" => "(?:\xce\xa8|\xcf\x88)", - "\xce\xa9" => "(?:\xce\xa9|\xcf\x89)", - "\xce\xaa" => "(?:\xce\xaa|\xcf\x8a)", - "\xce\xab" => "(?:\xce\xab|\xcf\x8b)", - "\xcf\x9a" => "(?:\xcf\x9a|\xcf\x9b)", - "\xcf\x9c" => "(?:\xcf\x9c|\xcf\x9d)", - "\xcf\x9e" => "(?:\xcf\x9e|\xcf\x9f)", - "\xcf\xa0" => "(?:\xcf\xa0|\xcf\xa1)", - "\xcf\xa2" => "(?:\xcf\xa2|\xcf\xa3)", - "\xcf\xa4" => "(?:\xcf\xa4|\xcf\xa5)", - "\xcf\xa6" => "(?:\xcf\xa6|\xcf\xa7)", - "\xcf\xa8" => "(?:\xcf\xa8|\xcf\xa9)", - "\xcf\xaa" => "(?:\xcf\xaa|\xcf\xab)", - "\xcf\xac" => "(?:\xcf\xac|\xcf\xad)", - "\xcf\xae" => "(?:\xcf\xae|\xcf\xaf)", - "\xcf\xb4" => "(?:\xcf\xb4|\xce\xb8)", - "\xd0\x80" => "(?:\xd0\x80|\xd1\x90)", - "\xd0\x81" => "(?:\xd0\x81|\xd1\x91)", - "\xd0\x82" => "(?:\xd0\x82|\xd1\x92)", - "\xd0\x83" => "(?:\xd0\x83|\xd1\x93)", - "\xd0\x84" => "(?:\xd0\x84|\xd1\x94)", - "\xd0\x85" => "(?:\xd0\x85|\xd1\x95)", - "\xd0\x86" => "(?:\xd0\x86|\xd1\x96)", - "\xd0\x87" => "(?:\xd0\x87|\xd1\x97)", - "\xd0\x88" => "(?:\xd0\x88|\xd1\x98)", - "\xd0\x89" => "(?:\xd0\x89|\xd1\x99)", - "\xd0\x8a" => "(?:\xd0\x8a|\xd1\x9a)", - "\xd0\x8b" => "(?:\xd0\x8b|\xd1\x9b)", - "\xd0\x8c" => "(?:\xd0\x8c|\xd1\x9c)", - "\xd0\x8d" => "(?:\xd0\x8d|\xd1\x9d)", - "\xd0\x8e" => "(?:\xd0\x8e|\xd1\x9e)", - "\xd0\x8f" => "(?:\xd0\x8f|\xd1\x9f)", - "\xd0\x90" => "(?:\xd0\x90|\xd0\xb0)", - "\xd0\x91" => "(?:\xd0\x91|\xd0\xb1)", - "\xd0\x92" => "(?:\xd0\x92|\xd0\xb2)", - "\xd0\x93" => "(?:\xd0\x93|\xd0\xb3)", - "\xd0\x94" => "(?:\xd0\x94|\xd0\xb4)", - "\xd0\x95" => "(?:\xd0\x95|\xd0\xb5)", - "\xd0\x96" => "(?:\xd0\x96|\xd0\xb6)", - "\xd0\x97" => "(?:\xd0\x97|\xd0\xb7)", - "\xd0\x98" => "(?:\xd0\x98|\xd0\xb8)", - "\xd0\x99" => "(?:\xd0\x99|\xd0\xb9)", - "\xd0\x9a" => "(?:\xd0\x9a|\xd0\xba)", - "\xd0\x9b" => "(?:\xd0\x9b|\xd0\xbb)", - "\xd0\x9c" => "(?:\xd0\x9c|\xd0\xbc)", - "\xd0\x9d" => "(?:\xd0\x9d|\xd0\xbd)", - "\xd0\x9e" => "(?:\xd0\x9e|\xd0\xbe)", - "\xd0\x9f" => "(?:\xd0\x9f|\xd0\xbf)", - "\xd0\xa0" => "(?:\xd0\xa0|\xd1\x80)", - "\xd0\xa1" => "(?:\xd0\xa1|\xd1\x81)", - "\xd0\xa2" => "(?:\xd0\xa2|\xd1\x82)", - "\xd0\xa3" => "(?:\xd0\xa3|\xd1\x83)", - "\xd0\xa4" => "(?:\xd0\xa4|\xd1\x84)", - "\xd0\xa5" => "(?:\xd0\xa5|\xd1\x85)", - "\xd0\xa6" => "(?:\xd0\xa6|\xd1\x86)", - "\xd0\xa7" => "(?:\xd0\xa7|\xd1\x87)", - "\xd0\xa8" => "(?:\xd0\xa8|\xd1\x88)", - "\xd0\xa9" => "(?:\xd0\xa9|\xd1\x89)", - "\xd0\xaa" => "(?:\xd0\xaa|\xd1\x8a)", - "\xd0\xab" => "(?:\xd0\xab|\xd1\x8b)", - "\xd0\xac" => "(?:\xd0\xac|\xd1\x8c)", - "\xd0\xad" => "(?:\xd0\xad|\xd1\x8d)", - "\xd0\xae" => "(?:\xd0\xae|\xd1\x8e)", - "\xd0\xaf" => "(?:\xd0\xaf|\xd1\x8f)", - "\xd1\xa0" => "(?:\xd1\xa0|\xd1\xa1)", - "\xd1\xa2" => "(?:\xd1\xa2|\xd1\xa3)", - "\xd1\xa4" => "(?:\xd1\xa4|\xd1\xa5)", - "\xd1\xa6" => "(?:\xd1\xa6|\xd1\xa7)", - "\xd1\xa8" => "(?:\xd1\xa8|\xd1\xa9)", - "\xd1\xaa" => "(?:\xd1\xaa|\xd1\xab)", - "\xd1\xac" => "(?:\xd1\xac|\xd1\xad)", - "\xd1\xae" => "(?:\xd1\xae|\xd1\xaf)", - "\xd1\xb0" => "(?:\xd1\xb0|\xd1\xb1)", - "\xd1\xb2" => "(?:\xd1\xb2|\xd1\xb3)", - "\xd1\xb4" => "(?:\xd1\xb4|\xd1\xb5)", - "\xd1\xb6" => "(?:\xd1\xb6|\xd1\xb7)", - "\xd1\xb8" => "(?:\xd1\xb8|\xd1\xb9)", - "\xd1\xba" => "(?:\xd1\xba|\xd1\xbb)", - "\xd1\xbc" => "(?:\xd1\xbc|\xd1\xbd)", - "\xd1\xbe" => "(?:\xd1\xbe|\xd1\xbf)", - "\xd2\x80" => "(?:\xd2\x80|\xd2\x81)", - "\xd2\x8c" => "(?:\xd2\x8c|\xd2\x8d)", - "\xd2\x8e" => "(?:\xd2\x8e|\xd2\x8f)", - "\xd2\x90" => "(?:\xd2\x90|\xd2\x91)", - "\xd2\x92" => "(?:\xd2\x92|\xd2\x93)", - "\xd2\x94" => "(?:\xd2\x94|\xd2\x95)", - "\xd2\x96" => "(?:\xd2\x96|\xd2\x97)", - "\xd2\x98" => "(?:\xd2\x98|\xd2\x99)", - "\xd2\x9a" => "(?:\xd2\x9a|\xd2\x9b)", - "\xd2\x9c" => "(?:\xd2\x9c|\xd2\x9d)", - "\xd2\x9e" => "(?:\xd2\x9e|\xd2\x9f)", - "\xd2\xa0" => "(?:\xd2\xa0|\xd2\xa1)", - "\xd2\xa2" => "(?:\xd2\xa2|\xd2\xa3)", - "\xd2\xa4" => "(?:\xd2\xa4|\xd2\xa5)", - "\xd2\xa6" => "(?:\xd2\xa6|\xd2\xa7)", - "\xd2\xa8" => "(?:\xd2\xa8|\xd2\xa9)", - "\xd2\xaa" => "(?:\xd2\xaa|\xd2\xab)", - "\xd2\xac" => "(?:\xd2\xac|\xd2\xad)", - "\xd2\xae" => "(?:\xd2\xae|\xd2\xaf)", - "\xd2\xb0" => "(?:\xd2\xb0|\xd2\xb1)", - "\xd2\xb2" => "(?:\xd2\xb2|\xd2\xb3)", - "\xd2\xb4" => "(?:\xd2\xb4|\xd2\xb5)", - "\xd2\xb6" => "(?:\xd2\xb6|\xd2\xb7)", - "\xd2\xb8" => "(?:\xd2\xb8|\xd2\xb9)", - "\xd2\xba" => "(?:\xd2\xba|\xd2\xbb)", - "\xd2\xbc" => "(?:\xd2\xbc|\xd2\xbd)", - "\xd2\xbe" => "(?:\xd2\xbe|\xd2\xbf)", - "\xd3\x81" => "(?:\xd3\x81|\xd3\x82)", - "\xd3\x83" => "(?:\xd3\x83|\xd3\x84)", - "\xd3\x87" => "(?:\xd3\x87|\xd3\x88)", - "\xd3\x8b" => "(?:\xd3\x8b|\xd3\x8c)", - "\xd3\x90" => "(?:\xd3\x90|\xd3\x91)", - "\xd3\x92" => "(?:\xd3\x92|\xd3\x93)", - "\xd3\x94" => "(?:\xd3\x94|\xd3\x95)", - "\xd3\x96" => "(?:\xd3\x96|\xd3\x97)", - "\xd3\x98" => "(?:\xd3\x98|\xd3\x99)", - "\xd3\x9a" => "(?:\xd3\x9a|\xd3\x9b)", - "\xd3\x9c" => "(?:\xd3\x9c|\xd3\x9d)", - "\xd3\x9e" => "(?:\xd3\x9e|\xd3\x9f)", - "\xd3\xa0" => "(?:\xd3\xa0|\xd3\xa1)", - "\xd3\xa2" => "(?:\xd3\xa2|\xd3\xa3)", - "\xd3\xa4" => "(?:\xd3\xa4|\xd3\xa5)", - "\xd3\xa6" => "(?:\xd3\xa6|\xd3\xa7)", - "\xd3\xa8" => "(?:\xd3\xa8|\xd3\xa9)", - "\xd3\xaa" => "(?:\xd3\xaa|\xd3\xab)", - "\xd3\xac" => "(?:\xd3\xac|\xd3\xad)", - "\xd3\xae" => "(?:\xd3\xae|\xd3\xaf)", - "\xd3\xb0" => "(?:\xd3\xb0|\xd3\xb1)", - "\xd3\xb2" => "(?:\xd3\xb2|\xd3\xb3)", - "\xd3\xb4" => "(?:\xd3\xb4|\xd3\xb5)", - "\xd3\xb8" => "(?:\xd3\xb8|\xd3\xb9)", - "\xd4\xb1" => "(?:\xd4\xb1|\xd5\xa1)", - "\xd4\xb2" => "(?:\xd4\xb2|\xd5\xa2)", - "\xd4\xb3" => "(?:\xd4\xb3|\xd5\xa3)", - "\xd4\xb4" => "(?:\xd4\xb4|\xd5\xa4)", - "\xd4\xb5" => "(?:\xd4\xb5|\xd5\xa5)", - "\xd4\xb6" => "(?:\xd4\xb6|\xd5\xa6)", - "\xd4\xb7" => "(?:\xd4\xb7|\xd5\xa7)", - "\xd4\xb8" => "(?:\xd4\xb8|\xd5\xa8)", - "\xd4\xb9" => "(?:\xd4\xb9|\xd5\xa9)", - "\xd4\xba" => "(?:\xd4\xba|\xd5\xaa)", - "\xd4\xbb" => "(?:\xd4\xbb|\xd5\xab)", - "\xd4\xbc" => "(?:\xd4\xbc|\xd5\xac)", - "\xd4\xbd" => "(?:\xd4\xbd|\xd5\xad)", - "\xd4\xbe" => "(?:\xd4\xbe|\xd5\xae)", - "\xd4\xbf" => "(?:\xd4\xbf|\xd5\xaf)", - "\xd5\x80" => "(?:\xd5\x80|\xd5\xb0)", - "\xd5\x81" => "(?:\xd5\x81|\xd5\xb1)", - "\xd5\x82" => "(?:\xd5\x82|\xd5\xb2)", - "\xd5\x83" => "(?:\xd5\x83|\xd5\xb3)", - "\xd5\x84" => "(?:\xd5\x84|\xd5\xb4)", - "\xd5\x85" => "(?:\xd5\x85|\xd5\xb5)", - "\xd5\x86" => "(?:\xd5\x86|\xd5\xb6)", - "\xd5\x87" => "(?:\xd5\x87|\xd5\xb7)", - "\xd5\x88" => "(?:\xd5\x88|\xd5\xb8)", - "\xd5\x89" => "(?:\xd5\x89|\xd5\xb9)", - "\xd5\x8a" => "(?:\xd5\x8a|\xd5\xba)", - "\xd5\x8b" => "(?:\xd5\x8b|\xd5\xbb)", - "\xd5\x8c" => "(?:\xd5\x8c|\xd5\xbc)", - "\xd5\x8d" => "(?:\xd5\x8d|\xd5\xbd)", - "\xd5\x8e" => "(?:\xd5\x8e|\xd5\xbe)", - "\xd5\x8f" => "(?:\xd5\x8f|\xd5\xbf)", - "\xd5\x90" => "(?:\xd5\x90|\xd6\x80)", - "\xd5\x91" => "(?:\xd5\x91|\xd6\x81)", - "\xd5\x92" => "(?:\xd5\x92|\xd6\x82)", - "\xd5\x93" => "(?:\xd5\x93|\xd6\x83)", - "\xd5\x94" => "(?:\xd5\x94|\xd6\x84)", - "\xd5\x95" => "(?:\xd5\x95|\xd6\x85)", - "\xd5\x96" => "(?:\xd5\x96|\xd6\x86)", - "\xe1\xb8\x80" => "(?:\xe1\xb8\x80|\xe1\xb8\x81)", - "\xe1\xb8\x82" => "(?:\xe1\xb8\x82|\xe1\xb8\x83)", - "\xe1\xb8\x84" => "(?:\xe1\xb8\x84|\xe1\xb8\x85)", - "\xe1\xb8\x86" => "(?:\xe1\xb8\x86|\xe1\xb8\x87)", - "\xe1\xb8\x88" => "(?:\xe1\xb8\x88|\xe1\xb8\x89)", - "\xe1\xb8\x8a" => "(?:\xe1\xb8\x8a|\xe1\xb8\x8b)", - "\xe1\xb8\x8c" => "(?:\xe1\xb8\x8c|\xe1\xb8\x8d)", - "\xe1\xb8\x8e" => "(?:\xe1\xb8\x8e|\xe1\xb8\x8f)", - "\xe1\xb8\x90" => "(?:\xe1\xb8\x90|\xe1\xb8\x91)", - "\xe1\xb8\x92" => "(?:\xe1\xb8\x92|\xe1\xb8\x93)", - "\xe1\xb8\x94" => "(?:\xe1\xb8\x94|\xe1\xb8\x95)", - "\xe1\xb8\x96" => "(?:\xe1\xb8\x96|\xe1\xb8\x97)", - "\xe1\xb8\x98" => "(?:\xe1\xb8\x98|\xe1\xb8\x99)", - "\xe1\xb8\x9a" => "(?:\xe1\xb8\x9a|\xe1\xb8\x9b)", - "\xe1\xb8\x9c" => "(?:\xe1\xb8\x9c|\xe1\xb8\x9d)", - "\xe1\xb8\x9e" => "(?:\xe1\xb8\x9e|\xe1\xb8\x9f)", - "\xe1\xb8\xa0" => "(?:\xe1\xb8\xa0|\xe1\xb8\xa1)", - "\xe1\xb8\xa2" => "(?:\xe1\xb8\xa2|\xe1\xb8\xa3)", - "\xe1\xb8\xa4" => "(?:\xe1\xb8\xa4|\xe1\xb8\xa5)", - "\xe1\xb8\xa6" => "(?:\xe1\xb8\xa6|\xe1\xb8\xa7)", - "\xe1\xb8\xa8" => "(?:\xe1\xb8\xa8|\xe1\xb8\xa9)", - "\xe1\xb8\xaa" => "(?:\xe1\xb8\xaa|\xe1\xb8\xab)", - "\xe1\xb8\xac" => "(?:\xe1\xb8\xac|\xe1\xb8\xad)", - "\xe1\xb8\xae" => "(?:\xe1\xb8\xae|\xe1\xb8\xaf)", - "\xe1\xb8\xb0" => "(?:\xe1\xb8\xb0|\xe1\xb8\xb1)", - "\xe1\xb8\xb2" => "(?:\xe1\xb8\xb2|\xe1\xb8\xb3)", - "\xe1\xb8\xb4" => "(?:\xe1\xb8\xb4|\xe1\xb8\xb5)", - "\xe1\xb8\xb6" => "(?:\xe1\xb8\xb6|\xe1\xb8\xb7)", - "\xe1\xb8\xb8" => "(?:\xe1\xb8\xb8|\xe1\xb8\xb9)", - "\xe1\xb8\xba" => "(?:\xe1\xb8\xba|\xe1\xb8\xbb)", - "\xe1\xb8\xbc" => "(?:\xe1\xb8\xbc|\xe1\xb8\xbd)", - "\xe1\xb8\xbe" => "(?:\xe1\xb8\xbe|\xe1\xb8\xbf)", - "\xe1\xb9\x80" => "(?:\xe1\xb9\x80|\xe1\xb9\x81)", - "\xe1\xb9\x82" => "(?:\xe1\xb9\x82|\xe1\xb9\x83)", - "\xe1\xb9\x84" => "(?:\xe1\xb9\x84|\xe1\xb9\x85)", - "\xe1\xb9\x86" => "(?:\xe1\xb9\x86|\xe1\xb9\x87)", - "\xe1\xb9\x88" => "(?:\xe1\xb9\x88|\xe1\xb9\x89)", - "\xe1\xb9\x8a" => "(?:\xe1\xb9\x8a|\xe1\xb9\x8b)", - "\xe1\xb9\x8c" => "(?:\xe1\xb9\x8c|\xe1\xb9\x8d)", - "\xe1\xb9\x8e" => "(?:\xe1\xb9\x8e|\xe1\xb9\x8f)", - "\xe1\xb9\x90" => "(?:\xe1\xb9\x90|\xe1\xb9\x91)", - "\xe1\xb9\x92" => "(?:\xe1\xb9\x92|\xe1\xb9\x93)", - "\xe1\xb9\x94" => "(?:\xe1\xb9\x94|\xe1\xb9\x95)", - "\xe1\xb9\x96" => "(?:\xe1\xb9\x96|\xe1\xb9\x97)", - "\xe1\xb9\x98" => "(?:\xe1\xb9\x98|\xe1\xb9\x99)", - "\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\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)", - "\xe1\xb9\xa8" => "(?:\xe1\xb9\xa8|\xe1\xb9\xa9)", - "\xe1\xb9\xaa" => "(?:\xe1\xb9\xaa|\xe1\xb9\xab)", - "\xe1\xb9\xac" => "(?:\xe1\xb9\xac|\xe1\xb9\xad)", - "\xe1\xb9\xae" => "(?:\xe1\xb9\xae|\xe1\xb9\xaf)", - "\xe1\xb9\xb0" => "(?:\xe1\xb9\xb0|\xe1\xb9\xb1)", - "\xe1\xb9\xb2" => "(?:\xe1\xb9\xb2|\xe1\xb9\xb3)", - "\xe1\xb9\xb4" => "(?:\xe1\xb9\xb4|\xe1\xb9\xb5)", - "\xe1\xb9\xb6" => "(?:\xe1\xb9\xb6|\xe1\xb9\xb7)", - "\xe1\xb9\xb8" => "(?:\xe1\xb9\xb8|\xe1\xb9\xb9)", - "\xe1\xb9\xba" => "(?:\xe1\xb9\xba|\xe1\xb9\xbb)", - "\xe1\xb9\xbc" => "(?:\xe1\xb9\xbc|\xe1\xb9\xbd)", - "\xe1\xb9\xbe" => "(?:\xe1\xb9\xbe|\xe1\xb9\xbf)", - "\xe1\xba\x80" => "(?:\xe1\xba\x80|\xe1\xba\x81)", - "\xe1\xba\x82" => "(?:\xe1\xba\x82|\xe1\xba\x83)", - "\xe1\xba\x84" => "(?:\xe1\xba\x84|\xe1\xba\x85)", - "\xe1\xba\x86" => "(?:\xe1\xba\x86|\xe1\xba\x87)", - "\xe1\xba\x88" => "(?:\xe1\xba\x88|\xe1\xba\x89)", - "\xe1\xba\x8a" => "(?:\xe1\xba\x8a|\xe1\xba\x8b)", - "\xe1\xba\x8c" => "(?:\xe1\xba\x8c|\xe1\xba\x8d)", - "\xe1\xba\x8e" => "(?:\xe1\xba\x8e|\xe1\xba\x8f)", - "\xe1\xba\x90" => "(?:\xe1\xba\x90|\xe1\xba\x91)", - "\xe1\xba\x92" => "(?:\xe1\xba\x92|\xe1\xba\x93)", - "\xe1\xba\x94" => "(?:\xe1\xba\x94|\xe1\xba\x95)", - "\xe1\xba\xa0" => "(?:\xe1\xba\xa0|\xe1\xba\xa1)", - "\xe1\xba\xa2" => "(?:\xe1\xba\xa2|\xe1\xba\xa3)", - "\xe1\xba\xa4" => "(?:\xe1\xba\xa4|\xe1\xba\xa5)", - "\xe1\xba\xa6" => "(?:\xe1\xba\xa6|\xe1\xba\xa7)", - "\xe1\xba\xa8" => "(?:\xe1\xba\xa8|\xe1\xba\xa9)", - "\xe1\xba\xaa" => "(?:\xe1\xba\xaa|\xe1\xba\xab)", - "\xe1\xba\xac" => "(?:\xe1\xba\xac|\xe1\xba\xad)", - "\xe1\xba\xae" => "(?:\xe1\xba\xae|\xe1\xba\xaf)", - "\xe1\xba\xb0" => "(?:\xe1\xba\xb0|\xe1\xba\xb1)", - "\xe1\xba\xb2" => "(?:\xe1\xba\xb2|\xe1\xba\xb3)", - "\xe1\xba\xb4" => "(?:\xe1\xba\xb4|\xe1\xba\xb5)", - "\xe1\xba\xb6" => "(?:\xe1\xba\xb6|\xe1\xba\xb7)", - "\xe1\xba\xb8" => "(?:\xe1\xba\xb8|\xe1\xba\xb9)", - "\xe1\xba\xba" => "(?:\xe1\xba\xba|\xe1\xba\xbb)", - "\xe1\xba\xbc" => "(?:\xe1\xba\xbc|\xe1\xba\xbd)", - "\xe1\xba\xbe" => "(?:\xe1\xba\xbe|\xe1\xba\xbf)", - "\xe1\xbb\x80" => "(?:\xe1\xbb\x80|\xe1\xbb\x81)", - "\xe1\xbb\x82" => "(?:\xe1\xbb\x82|\xe1\xbb\x83)", - "\xe1\xbb\x84" => "(?:\xe1\xbb\x84|\xe1\xbb\x85)", - "\xe1\xbb\x86" => "(?:\xe1\xbb\x86|\xe1\xbb\x87)", - "\xe1\xbb\x88" => "(?:\xe1\xbb\x88|\xe1\xbb\x89)", - "\xe1\xbb\x8a" => "(?:\xe1\xbb\x8a|\xe1\xbb\x8b)", - "\xe1\xbb\x8c" => "(?:\xe1\xbb\x8c|\xe1\xbb\x8d)", - "\xe1\xbb\x8e" => "(?:\xe1\xbb\x8e|\xe1\xbb\x8f)", - "\xe1\xbb\x90" => "(?:\xe1\xbb\x90|\xe1\xbb\x91)", - "\xe1\xbb\x92" => "(?:\xe1\xbb\x92|\xe1\xbb\x93)", - "\xe1\xbb\x94" => "(?:\xe1\xbb\x94|\xe1\xbb\x95)", - "\xe1\xbb\x96" => "(?:\xe1\xbb\x96|\xe1\xbb\x97)", - "\xe1\xbb\x98" => "(?:\xe1\xbb\x98|\xe1\xbb\x99)", - "\xe1\xbb\x9a" => "(?:\xe1\xbb\x9a|\xe1\xbb\x9b)", - "\xe1\xbb\x9c" => "(?:\xe1\xbb\x9c|\xe1\xbb\x9d)", - "\xe1\xbb\x9e" => "(?:\xe1\xbb\x9e|\xe1\xbb\x9f)", - "\xe1\xbb\xa0" => "(?:\xe1\xbb\xa0|\xe1\xbb\xa1)", - "\xe1\xbb\xa2" => "(?:\xe1\xbb\xa2|\xe1\xbb\xa3)", - "\xe1\xbb\xa4" => "(?:\xe1\xbb\xa4|\xe1\xbb\xa5)", - "\xe1\xbb\xa6" => "(?:\xe1\xbb\xa6|\xe1\xbb\xa7)", - "\xe1\xbb\xa8" => "(?:\xe1\xbb\xa8|\xe1\xbb\xa9)", - "\xe1\xbb\xaa" => "(?:\xe1\xbb\xaa|\xe1\xbb\xab)", - "\xe1\xbb\xac" => "(?:\xe1\xbb\xac|\xe1\xbb\xad)", - "\xe1\xbb\xae" => "(?:\xe1\xbb\xae|\xe1\xbb\xaf)", - "\xe1\xbb\xb0" => "(?:\xe1\xbb\xb0|\xe1\xbb\xb1)", - "\xe1\xbb\xb2" => "(?:\xe1\xbb\xb2|\xe1\xbb\xb3)", - "\xe1\xbb\xb4" => "(?:\xe1\xbb\xb4|\xe1\xbb\xb5)", - "\xe1\xbb\xb6" => "(?:\xe1\xbb\xb6|\xe1\xbb\xb7)", - "\xe1\xbb\xb8" => "(?:\xe1\xbb\xb8|\xe1\xbb\xb9)", - "\xe1\xbc\x88" => "(?:\xe1\xbc\x88|\xe1\xbc\x80)", - "\xe1\xbc\x89" => "(?:\xe1\xbc\x89|\xe1\xbc\x81)", - "\xe1\xbc\x8a" => "(?:\xe1\xbc\x8a|\xe1\xbc\x82)", - "\xe1\xbc\x8b" => "(?:\xe1\xbc\x8b|\xe1\xbc\x83)", - "\xe1\xbc\x8c" => "(?:\xe1\xbc\x8c|\xe1\xbc\x84)", - "\xe1\xbc\x8d" => "(?:\xe1\xbc\x8d|\xe1\xbc\x85)", - "\xe1\xbc\x8e" => "(?:\xe1\xbc\x8e|\xe1\xbc\x86)", - "\xe1\xbc\x8f" => "(?:\xe1\xbc\x8f|\xe1\xbc\x87)", - "\xe1\xbc\x98" => "(?:\xe1\xbc\x98|\xe1\xbc\x90)", - "\xe1\xbc\x99" => "(?:\xe1\xbc\x99|\xe1\xbc\x91)", - "\xe1\xbc\x9a" => "(?:\xe1\xbc\x9a|\xe1\xbc\x92)", - "\xe1\xbc\x9b" => "(?:\xe1\xbc\x9b|\xe1\xbc\x93)", - "\xe1\xbc\x9c" => "(?:\xe1\xbc\x9c|\xe1\xbc\x94)", - "\xe1\xbc\x9d" => "(?:\xe1\xbc\x9d|\xe1\xbc\x95)", - "\xe1\xbc\xa8" => "(?:\xe1\xbc\xa8|\xe1\xbc\xa0)", - "\xe1\xbc\xa9" => "(?:\xe1\xbc\xa9|\xe1\xbc\xa1)", - "\xe1\xbc\xaa" => "(?:\xe1\xbc\xaa|\xe1\xbc\xa2)", - "\xe1\xbc\xab" => "(?:\xe1\xbc\xab|\xe1\xbc\xa3)", - "\xe1\xbc\xac" => "(?:\xe1\xbc\xac|\xe1\xbc\xa4)", - "\xe1\xbc\xad" => "(?:\xe1\xbc\xad|\xe1\xbc\xa5)", - "\xe1\xbc\xae" => "(?:\xe1\xbc\xae|\xe1\xbc\xa6)", - "\xe1\xbc\xaf" => "(?:\xe1\xbc\xaf|\xe1\xbc\xa7)", - "\xe1\xbc\xb8" => "(?:\xe1\xbc\xb8|\xe1\xbc\xb0)", - "\xe1\xbc\xb9" => "(?:\xe1\xbc\xb9|\xe1\xbc\xb1)", - "\xe1\xbc\xba" => "(?:\xe1\xbc\xba|\xe1\xbc\xb2)", - "\xe1\xbc\xbb" => "(?:\xe1\xbc\xbb|\xe1\xbc\xb3)", - "\xe1\xbc\xbc" => "(?:\xe1\xbc\xbc|\xe1\xbc\xb4)", - "\xe1\xbc\xbd" => "(?:\xe1\xbc\xbd|\xe1\xbc\xb5)", - "\xe1\xbc\xbe" => "(?:\xe1\xbc\xbe|\xe1\xbc\xb6)", - "\xe1\xbc\xbf" => "(?:\xe1\xbc\xbf|\xe1\xbc\xb7)", - "\xe1\xbd\x88" => "(?:\xe1\xbd\x88|\xe1\xbd\x80)", - "\xe1\xbd\x89" => "(?:\xe1\xbd\x89|\xe1\xbd\x81)", - "\xe1\xbd\x8a" => "(?:\xe1\xbd\x8a|\xe1\xbd\x82)", - "\xe1\xbd\x8b" => "(?:\xe1\xbd\x8b|\xe1\xbd\x83)", - "\xe1\xbd\x8c" => "(?:\xe1\xbd\x8c|\xe1\xbd\x84)", - "\xe1\xbd\x8d" => "(?:\xe1\xbd\x8d|\xe1\xbd\x85)", - "\xe1\xbd\x99" => "(?:\xe1\xbd\x99|\xe1\xbd\x91)", - "\xe1\xbd\x9b" => "(?:\xe1\xbd\x9b|\xe1\xbd\x93)", - "\xe1\xbd\x9d" => "(?:\xe1\xbd\x9d|\xe1\xbd\x95)", - "\xe1\xbd\x9f" => "(?:\xe1\xbd\x9f|\xe1\xbd\x97)", - "\xe1\xbd\xa8" => "(?:\xe1\xbd\xa8|\xe1\xbd\xa0)", - "\xe1\xbd\xa9" => "(?:\xe1\xbd\xa9|\xe1\xbd\xa1)", - "\xe1\xbd\xaa" => "(?:\xe1\xbd\xaa|\xe1\xbd\xa2)", - "\xe1\xbd\xab" => "(?:\xe1\xbd\xab|\xe1\xbd\xa3)", - "\xe1\xbd\xac" => "(?:\xe1\xbd\xac|\xe1\xbd\xa4)", - "\xe1\xbd\xad" => "(?:\xe1\xbd\xad|\xe1\xbd\xa5)", - "\xe1\xbd\xae" => "(?:\xe1\xbd\xae|\xe1\xbd\xa6)", - "\xe1\xbd\xaf" => "(?:\xe1\xbd\xaf|\xe1\xbd\xa7)", - "\xe1\xbe\x88" => "(?:\xe1\xbe\x88|\xe1\xbe\x80)", - "\xe1\xbe\x89" => "(?:\xe1\xbe\x89|\xe1\xbe\x81)", - "\xe1\xbe\x8a" => "(?:\xe1\xbe\x8a|\xe1\xbe\x82)", - "\xe1\xbe\x8b" => "(?:\xe1\xbe\x8b|\xe1\xbe\x83)", - "\xe1\xbe\x8c" => "(?:\xe1\xbe\x8c|\xe1\xbe\x84)", - "\xe1\xbe\x8d" => "(?:\xe1\xbe\x8d|\xe1\xbe\x85)", - "\xe1\xbe\x8e" => "(?:\xe1\xbe\x8e|\xe1\xbe\x86)", - "\xe1\xbe\x8f" => "(?:\xe1\xbe\x8f|\xe1\xbe\x87)", - "\xe1\xbe\x98" => "(?:\xe1\xbe\x98|\xe1\xbe\x90)", - "\xe1\xbe\x99" => "(?:\xe1\xbe\x99|\xe1\xbe\x91)", - "\xe1\xbe\x9a" => "(?:\xe1\xbe\x9a|\xe1\xbe\x92)", - "\xe1\xbe\x9b" => "(?:\xe1\xbe\x9b|\xe1\xbe\x93)", - "\xe1\xbe\x9c" => "(?:\xe1\xbe\x9c|\xe1\xbe\x94)", - "\xe1\xbe\x9d" => "(?:\xe1\xbe\x9d|\xe1\xbe\x95)", - "\xe1\xbe\x9e" => "(?:\xe1\xbe\x9e|\xe1\xbe\x96)", - "\xe1\xbe\x9f" => "(?:\xe1\xbe\x9f|\xe1\xbe\x97)", - "\xe1\xbe\xa8" => "(?:\xe1\xbe\xa8|\xe1\xbe\xa0)", - "\xe1\xbe\xa9" => "(?:\xe1\xbe\xa9|\xe1\xbe\xa1)", - "\xe1\xbe\xaa" => "(?:\xe1\xbe\xaa|\xe1\xbe\xa2)", - "\xe1\xbe\xab" => "(?:\xe1\xbe\xab|\xe1\xbe\xa3)", - "\xe1\xbe\xac" => "(?:\xe1\xbe\xac|\xe1\xbe\xa4)", - "\xe1\xbe\xad" => "(?:\xe1\xbe\xad|\xe1\xbe\xa5)", - "\xe1\xbe\xae" => "(?:\xe1\xbe\xae|\xe1\xbe\xa6)", - "\xe1\xbe\xaf" => "(?:\xe1\xbe\xaf|\xe1\xbe\xa7)", - "\xe1\xbe\xb8" => "(?:\xe1\xbe\xb8|\xe1\xbe\xb0)", - "\xe1\xbe\xb9" => "(?:\xe1\xbe\xb9|\xe1\xbe\xb1)", - "\xe1\xbe\xba" => "(?:\xe1\xbe\xba|\xe1\xbd\xb0)", - "\xe1\xbe\xbb" => "(?:\xe1\xbe\xbb|\xe1\xbd\xb1)", - "\xe1\xbe\xbc" => "(?:\xe1\xbe\xbc|\xe1\xbe\xb3)", - "\xe1\xbf\x88" => "(?:\xe1\xbf\x88|\xe1\xbd\xb2)", - "\xe1\xbf\x89" => "(?:\xe1\xbf\x89|\xe1\xbd\xb3)", - "\xe1\xbf\x8a" => "(?:\xe1\xbf\x8a|\xe1\xbd\xb4)", - "\xe1\xbf\x8b" => "(?:\xe1\xbf\x8b|\xe1\xbd\xb5)", - "\xe1\xbf\x8c" => "(?:\xe1\xbf\x8c|\xe1\xbf\x83)", - "\xe1\xbf\x98" => "(?:\xe1\xbf\x98|\xe1\xbf\x90)", - "\xe1\xbf\x99" => "(?:\xe1\xbf\x99|\xe1\xbf\x91)", - "\xe1\xbf\x9a" => "(?:\xe1\xbf\x9a|\xe1\xbd\xb6)", - "\xe1\xbf\x9b" => "(?:\xe1\xbf\x9b|\xe1\xbd\xb7)", - "\xe1\xbf\xa8" => "(?:\xe1\xbf\xa8|\xe1\xbf\xa0)", - "\xe1\xbf\xa9" => "(?:\xe1\xbf\xa9|\xe1\xbf\xa1)", - "\xe1\xbf\xaa" => "(?:\xe1\xbf\xaa|\xe1\xbd\xba)", - "\xe1\xbf\xab" => "(?:\xe1\xbf\xab|\xe1\xbd\xbb)", - "\xe1\xbf\xac" => "(?:\xe1\xbf\xac|\xe1\xbf\xa5)", - "\xe1\xbf\xb8" => "(?:\xe1\xbf\xb8|\xe1\xbd\xb8)", - "\xe1\xbf\xb9" => "(?:\xe1\xbf\xb9|\xe1\xbd\xb9)", - "\xe1\xbf\xba" => "(?:\xe1\xbf\xba|\xe1\xbd\xbc)", - "\xe1\xbf\xbb" => "(?:\xe1\xbf\xbb|\xe1\xbd\xbd)", - "\xe1\xbf\xbc" => "(?:\xe1\xbf\xbc|\xe1\xbf\xb3)", - "\xe2\x84\xa6" => "(?:\xe2\x84\xa6|\xcf\x89)", - "\xe2\x84\xaa" => "(?:\xe2\x84\xaa|k)", - "\xe2\x84\xab" => "(?:\xe2\x84\xab|\xc3\xa5)", - "\xe2\x85\xa0" => "(?:\xe2\x85\xa0|\xe2\x85\xb0)", - "\xe2\x85\xa1" => "(?:\xe2\x85\xa1|\xe2\x85\xb1)", - "\xe2\x85\xa2" => "(?:\xe2\x85\xa2|\xe2\x85\xb2)", - "\xe2\x85\xa3" => "(?:\xe2\x85\xa3|\xe2\x85\xb3)", - "\xe2\x85\xa4" => "(?:\xe2\x85\xa4|\xe2\x85\xb4)", - "\xe2\x85\xa5" => "(?:\xe2\x85\xa5|\xe2\x85\xb5)", - "\xe2\x85\xa6" => "(?:\xe2\x85\xa6|\xe2\x85\xb6)", - "\xe2\x85\xa7" => "(?:\xe2\x85\xa7|\xe2\x85\xb7)", - "\xe2\x85\xa8" => "(?:\xe2\x85\xa8|\xe2\x85\xb8)", - "\xe2\x85\xa9" => "(?:\xe2\x85\xa9|\xe2\x85\xb9)", - "\xe2\x85\xaa" => "(?:\xe2\x85\xaa|\xe2\x85\xba)", - "\xe2\x85\xab" => "(?:\xe2\x85\xab|\xe2\x85\xbb)", - "\xe2\x85\xac" => "(?:\xe2\x85\xac|\xe2\x85\xbc)", - "\xe2\x85\xad" => "(?:\xe2\x85\xad|\xe2\x85\xbd)", - "\xe2\x85\xae" => "(?:\xe2\x85\xae|\xe2\x85\xbe)", - "\xe2\x85\xaf" => "(?:\xe2\x85\xaf|\xe2\x85\xbf)", - "\xe2\x92\xb6" => "(?:\xe2\x92\xb6|\xe2\x93\x90)", - "\xe2\x92\xb7" => "(?:\xe2\x92\xb7|\xe2\x93\x91)", - "\xe2\x92\xb8" => "(?:\xe2\x92\xb8|\xe2\x93\x92)", - "\xe2\x92\xb9" => "(?:\xe2\x92\xb9|\xe2\x93\x93)", - "\xe2\x92\xba" => "(?:\xe2\x92\xba|\xe2\x93\x94)", - "\xe2\x92\xbb" => "(?:\xe2\x92\xbb|\xe2\x93\x95)", - "\xe2\x92\xbc" => "(?:\xe2\x92\xbc|\xe2\x93\x96)", - "\xe2\x92\xbd" => "(?:\xe2\x92\xbd|\xe2\x93\x97)", - "\xe2\x92\xbe" => "(?:\xe2\x92\xbe|\xe2\x93\x98)", - "\xe2\x92\xbf" => "(?:\xe2\x92\xbf|\xe2\x93\x99)", - "\xe2\x93\x80" => "(?:\xe2\x93\x80|\xe2\x93\x9a)", - "\xe2\x93\x81" => "(?:\xe2\x93\x81|\xe2\x93\x9b)", - "\xe2\x93\x82" => "(?:\xe2\x93\x82|\xe2\x93\x9c)", - "\xe2\x93\x83" => "(?:\xe2\x93\x83|\xe2\x93\x9d)", - "\xe2\x93\x84" => "(?:\xe2\x93\x84|\xe2\x93\x9e)", - "\xe2\x93\x85" => "(?:\xe2\x93\x85|\xe2\x93\x9f)", - "\xe2\x93\x86" => "(?:\xe2\x93\x86|\xe2\x93\xa0)", - "\xe2\x93\x87" => "(?:\xe2\x93\x87|\xe2\x93\xa1)", - "\xe2\x93\x88" => "(?:\xe2\x93\x88|\xe2\x93\xa2)", - "\xe2\x93\x89" => "(?:\xe2\x93\x89|\xe2\x93\xa3)", - "\xe2\x93\x8a" => "(?:\xe2\x93\x8a|\xe2\x93\xa4)", - "\xe2\x93\x8b" => "(?:\xe2\x93\x8b|\xe2\x93\xa5)", - "\xe2\x93\x8c" => "(?:\xe2\x93\x8c|\xe2\x93\xa6)", - "\xe2\x93\x8d" => "(?:\xe2\x93\x8d|\xe2\x93\xa7)", - "\xe2\x93\x8e" => "(?:\xe2\x93\x8e|\xe2\x93\xa8)", - "\xe2\x93\x8f" => "(?:\xe2\x93\x8f|\xe2\x93\xa9)", - "\xef\xbc\xa1" => "(?:\xef\xbc\xa1|\xef\xbd\x81)", - "\xef\xbc\xa2" => "(?:\xef\xbc\xa2|\xef\xbd\x82)", - "\xef\xbc\xa3" => "(?:\xef\xbc\xa3|\xef\xbd\x83)", - "\xef\xbc\xa4" => "(?:\xef\xbc\xa4|\xef\xbd\x84)", - "\xef\xbc\xa5" => "(?:\xef\xbc\xa5|\xef\xbd\x85)", - "\xef\xbc\xa6" => "(?:\xef\xbc\xa6|\xef\xbd\x86)", - "\xef\xbc\xa7" => "(?:\xef\xbc\xa7|\xef\xbd\x87)", - "\xef\xbc\xa8" => "(?:\xef\xbc\xa8|\xef\xbd\x88)", - "\xef\xbc\xa9" => "(?:\xef\xbc\xa9|\xef\xbd\x89)", - "\xef\xbc\xaa" => "(?:\xef\xbc\xaa|\xef\xbd\x8a)", - "\xef\xbc\xab" => "(?:\xef\xbc\xab|\xef\xbd\x8b)", - "\xef\xbc\xac" => "(?:\xef\xbc\xac|\xef\xbd\x8c)", - "\xef\xbc\xad" => "(?:\xef\xbc\xad|\xef\xbd\x8d)", - "\xef\xbc\xae" => "(?:\xef\xbc\xae|\xef\xbd\x8e)", - "\xef\xbc\xaf" => "(?:\xef\xbc\xaf|\xef\xbd\x8f)", - "\xef\xbc\xb0" => "(?:\xef\xbc\xb0|\xef\xbd\x90)", - "\xef\xbc\xb1" => "(?:\xef\xbc\xb1|\xef\xbd\x91)", - "\xef\xbc\xb2" => "(?:\xef\xbc\xb2|\xef\xbd\x92)", - "\xef\xbc\xb3" => "(?:\xef\xbc\xb3|\xef\xbd\x93)", - "\xef\xbc\xb4" => "(?:\xef\xbc\xb4|\xef\xbd\x94)", - "\xef\xbc\xb5" => "(?:\xef\xbc\xb5|\xef\xbd\x95)", - "\xef\xbc\xb6" => "(?:\xef\xbc\xb6|\xef\xbd\x96)", - "\xef\xbc\xb7" => "(?:\xef\xbc\xb7|\xef\xbd\x97)", - "\xef\xbc\xb8" => "(?:\xef\xbc\xb8|\xef\xbd\x98)", - "\xef\xbc\xb9" => "(?:\xef\xbc\xb9|\xef\xbd\x99)", - "\xef\xbc\xba" => "(?:\xef\xbc\xba|\xef\xbd\x9a)", - "\xf0\x90\x90\x80" => "(?:\xf0\x90\x90\x80|\xf0\x90\x90\xa8)", - "\xf0\x90\x90\x81" => "(?:\xf0\x90\x90\x81|\xf0\x90\x90\xa9)", - "\xf0\x90\x90\x82" => "(?:\xf0\x90\x90\x82|\xf0\x90\x90\xaa)", - "\xf0\x90\x90\x83" => "(?:\xf0\x90\x90\x83|\xf0\x90\x90\xab)", - "\xf0\x90\x90\x84" => "(?:\xf0\x90\x90\x84|\xf0\x90\x90\xac)", - "\xf0\x90\x90\x85" => "(?:\xf0\x90\x90\x85|\xf0\x90\x90\xad)", - "\xf0\x90\x90\x86" => "(?:\xf0\x90\x90\x86|\xf0\x90\x90\xae)", - "\xf0\x90\x90\x87" => "(?:\xf0\x90\x90\x87|\xf0\x90\x90\xaf)", - "\xf0\x90\x90\x88" => "(?:\xf0\x90\x90\x88|\xf0\x90\x90\xb0)", - "\xf0\x90\x90\x89" => "(?:\xf0\x90\x90\x89|\xf0\x90\x90\xb1)", - "\xf0\x90\x90\x8a" => "(?:\xf0\x90\x90\x8a|\xf0\x90\x90\xb2)", - "\xf0\x90\x90\x8b" => "(?:\xf0\x90\x90\x8b|\xf0\x90\x90\xb3)", - "\xf0\x90\x90\x8c" => "(?:\xf0\x90\x90\x8c|\xf0\x90\x90\xb4)", - "\xf0\x90\x90\x8d" => "(?:\xf0\x90\x90\x8d|\xf0\x90\x90\xb5)", - "\xf0\x90\x90\x8e" => "(?:\xf0\x90\x90\x8e|\xf0\x90\x90\xb6)", - "\xf0\x90\x90\x8f" => "(?:\xf0\x90\x90\x8f|\xf0\x90\x90\xb7)", - "\xf0\x90\x90\x90" => "(?:\xf0\x90\x90\x90|\xf0\x90\x90\xb8)", - "\xf0\x90\x90\x91" => "(?:\xf0\x90\x90\x91|\xf0\x90\x90\xb9)", - "\xf0\x90\x90\x92" => "(?:\xf0\x90\x90\x92|\xf0\x90\x90\xba)", - "\xf0\x90\x90\x93" => "(?:\xf0\x90\x90\x93|\xf0\x90\x90\xbb)", - "\xf0\x90\x90\x94" => "(?:\xf0\x90\x90\x94|\xf0\x90\x90\xbc)", - "\xf0\x90\x90\x95" => "(?:\xf0\x90\x90\x95|\xf0\x90\x90\xbd)", - "\xf0\x90\x90\x96" => "(?:\xf0\x90\x90\x96|\xf0\x90\x90\xbe)", - "\xf0\x90\x90\x97" => "(?:\xf0\x90\x90\x97|\xf0\x90\x90\xbf)", - "\xf0\x90\x90\x98" => "(?:\xf0\x90\x90\x98|\xf0\x90\x91\x80)", - "\xf0\x90\x90\x99" => "(?:\xf0\x90\x90\x99|\xf0\x90\x91\x81)", - "\xf0\x90\x90\x9a" => "(?:\xf0\x90\x90\x9a|\xf0\x90\x91\x82)", - "\xf0\x90\x90\x9b" => "(?:\xf0\x90\x90\x9b|\xf0\x90\x91\x83)", - "\xf0\x90\x90\x9c" => "(?:\xf0\x90\x90\x9c|\xf0\x90\x91\x84)", - "\xf0\x90\x90\x9d" => "(?:\xf0\x90\x90\x9d|\xf0\x90\x91\x85)", - "\xf0\x90\x90\x9e" => "(?:\xf0\x90\x90\x9e|\xf0\x90\x91\x86)", - "\xf0\x90\x90\x9f" => "(?:\xf0\x90\x90\x9f|\xf0\x90\x91\x87)", - "\xf0\x90\x90\xa0" => "(?:\xf0\x90\x90\xa0|\xf0\x90\x91\x88)", - "\xf0\x90\x90\xa1" => "(?:\xf0\x90\x90\xa1|\xf0\x90\x91\x89)", - "\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)" -); +// umlaut to '(upper|lower)' for preg_match() +// this is UTF-8-encoded +// there is no need for a translation-table anymore since we use u-switch (utf-8) for preg-functions +// remember that we use the i-switch, too. [No need for (ae|Ae)] +$string_ul_umlaut = array(); +$string_ul_regex = array(); -if($search_lang=='DE') { - $string_ul_umlauts["\xc3\x9f"] = "(?:\xc3\x9f|SS|ss)"; // german ß - $string_ul_umlauts["\xc3\xa4"] = "(?:\xc3\xa4|\xc3\x84|ae|Ae)"; // german ä - $string_ul_umlauts["\xc3\xb6"] = "(?:\xc3\xb6|\xc3\x96|oe|Oe)"; // german ö - $string_ul_umlauts["\xc3\xbc"] = "(?:\xc3\xbc|\xc3\x9c|ue|Ue)"; // german ü - $string_ul_umlauts["\xc3\x84"] = "(?:\xc3\x84|\xc3\xa4|Ae|ae)"; // german Ä - $string_ul_umlauts["\xc3\x96"] = "(?:\xc3\x96|\xc3\xb6|Oe|oe)"; // german Ö - $string_ul_umlauts["\xc3\x9c"] = "(?:\xc3\x9c|\xc3\xbc|Ue|ue)"; // german Ü +// but add some national stuff +if($search_lang=='DE') { // add special handling for german umlauts (ä==ae, ...) + $string_ul_umlaut_add = array( + "\xc3\x9f", // german SZ-Ligatur + "\xc3\xa4", // german ae + "\xc3\xb6", // german oe + "\xc3\xbc", // german ue + "\xc3\x84", // german Ae + "\xc3\x96", // german Oe + "\xc3\x9c", // german Ue + // these are not that usual + "\xEF\xAC\x84", // german ffl-ligatur + "ffl", // german ffl-ligatur + "\xEF\xAC\x83", // german ffi-ligatur + "ffi", // german ffi-ligatur + "0xEF\xAC\x80", // german ff-Ligatur + "ff", // german ff-Ligatur + "\xEF\xAC\x81", // german fi-ligatur + "fi", // german fi-ligatur + "\xEF\xAC\x82", // german fl-ligatur + "fl", // german fl-ligatur + "\xEF\xAC\x85", // german st-Ligatur (long s) + "st", // german st-Ligatur + "\xEF\xAC\x86" // german st-ligatur (round-s) + ); + $string_ul_regex_add = array( + "(\xc3\x9f|ss)", // german SZ.Ligatur + "(\xc3\xa4|ae)", // german ae + "(\xc3\xb6|oe)", // german oe + "(\xc3\xbc|ue)", // german ue + "(\xc3\x84|Ae)", // german Ae + "(\xc3\x96|Oe)", // german Oe + "(\xc3\x9c|Ue)", // german Ue + // these are not that usual + "(\xEF\xAC\x84|ffl)", // german ffl-ligatur + "(\xEF\xAC\x84|ffl)", // german ffl-ligatur + "(\xEF\xAC\x83|ffi)", // german ffi-ligatur + "(\xEF\xAC\x83|ffi)", // german ffi-ligatur + "(\xEF\xAC\x80|ff)", // german ff-Ligatur + "(\xEF\xAC\x80|ff)", // german ff-Ligatur + "(\xEF\xAC\x81|fi)", // german fi-Ligatur + "(\xEF\xAC\x81|fi)", // german fi-Ligatur + "(\xEF\xAC\x82|fl)", // german fl-ligatur + "(\xEF\xAC\x82|fl)", // german fl-ligatur + "(\xEF\xAC\x85|st)", // german st-Ligatur (long s) + "(\xEF\xAC\x85|st|\xEF\xAC\x86)", // german st-Ligaturs + "(\xEF\xAC\x86|st)" // german st-ligatur (round-s) + ); + $string_ul_umlaut = array_merge($string_ul_umlaut_add, $string_ul_umlaut); + $string_ul_regex = array_merge($string_ul_regex_add, $string_ul_regex); } + ?> \ No newline at end of file Index: trunk/wb/search/search_modext.php =================================================================== --- trunk/wb/search/search_modext.php (revision 880) +++ trunk/wb/search/search_modext.php (revision 881) @@ -66,7 +66,7 @@ function is_all_matched($text, $search_words) { $all_matched = true; foreach ($search_words AS $word) { - if(!preg_match('/'.$word.'/i', $text)) { + if(!preg_match('/'.$word.'/ui', $text)) { $all_matched = false; break; } @@ -78,7 +78,7 @@ function is_any_matched($text, $search_words) { $any_matched = false; $word = '('.implode('|', $search_words).')'; - if(preg_match('/'.$word.'/i', $text)) { + if(preg_match('/'.$word.'/ui', $text)) { $any_matched = true; } return $any_matched; @@ -99,13 +99,13 @@ // stop-sign: .!?; + DOUBLE EXCLAMATION MARK - INTERROBANG - EXCLAMATION QUESTION MARK - QUESTION EXCLAMATION MARK - DOUBLE QUESTION MARK - HALFWIDTH IDEOGRAPHIC FULL STOP - IDEOGRAPHIC FULL STOP - IDEOGRAPHIC COMMA $str2=".!?;"."\xE2\x80\xBC"."\xE2\x80\xBD"."\xE2\x81\x89"."\xE2\x81\x88"."\xE2\x81\x87"."\xEF\xBD\xA1"."\xE3\x80\x82"."\xE3\x80\x81"; } - $regex='/(?:^|\b|['.$str1.'])([^'.$str1.']{0,200}?'.$word.'[^'.$str2.']{0,200}(?:['.$str2.']|\b|$))/isu'; + $regex='/(?:^|\b|['.$str1.'])([^'.$str1.']{0,200}?'.$word.'[^'.$str2.']{0,200}(?:['.$str2.']|\b|$))/uis'; if(version_compare(PHP_VERSION, '4.3.3', '>=') && strpos(strtoupper(PHP_OS), 'WIN')!==0 ) { // this may crash windows server, so skip if on windows // jump from match to match, get excerpt, stop if $max_excerpt_num is reached $last_end = 0; $offset = 0; - while(preg_match('/'.$word.'/Sisu', $text, $match_array, PREG_OFFSET_CAPTURE, $last_end)) { + while(preg_match('/'.$word.'/uis', $text, $match_array, PREG_OFFSET_CAPTURE, $last_end)) { $offset = ($match_array[0][1]-206 < $last_end)?$last_end:$match_array[0][1]-206; if(preg_match($regex, $text, $matches, PREG_OFFSET_CAPTURE, $offset)) { $last_end = $matches[1][1]+strlen($matches[1][0])-1; @@ -152,7 +152,7 @@ foreach($excerpt_array as $str) { $excerpt .= '#,,#'.preg_replace("/($string)/iu","#,,,,#$1#,,,,,#",$str).'#,,,#'; } - $excerpt = str_replace(array('&','<','>','"','\'',"\xC2\xA0"), array('&','<','>','"',''',' '), $excerpt); + $excerpt = str_replace(array('&','<','>','"','\'',"\xC2\xA0"), array('&','<','>','"',''',' '), $excerpt); $excerpt = str_replace(array('#,,,,#','#,,,,,#'), array($EXCERPT_MARKUP_START,$EXCERPT_MARKUP_END), $excerpt); $excerpt = str_replace(array('#,,#','#,,,#'), array($EXCERPT_BEFORE,$EXCERPT_AFTER), $excerpt); // prepare to write out @@ -170,7 +170,7 @@ // 4. $page_link_target=="" - do nothing if(version_compare(PHP_VERSION, '4.3.3', ">=") && substr($page_link_target,0,12)=='#wb_section_') { $word = '('.implode('|', $search_words).')'; - preg_match('/'.$word.'/i', $text, $match, PREG_OFFSET_CAPTURE); + preg_match('/'.$word.'/ui', $text, $match, PREG_OFFSET_CAPTURE); if($match && is_array($match[0])) { $x=$match[0][1]; // position of first match // is there an anchor nearby? @@ -255,10 +255,10 @@ if($mod_no_highlight) // no highlighting { $mod_page_link_target = "&nohighlight=1".$mod_page_link_target; } // clean the text: - $mod_text = preg_replace('#<(!--.*--|style.*#SiU', ' ', $mod_text); - $mod_text = preg_replace('#<(br( /)?|dt|/dd|/?(h[1-6]|tr|table|p|li|ul|pre|code|div|hr))[^>]*>#Si', '.', $mod_text); - $mod_text = preg_replace('/\s+/', ' ', $mod_text); - $mod_text = preg_replace('/ \./', '.', $mod_text); + $mod_text = preg_replace('#<(!--.*--|style.*#iU', ' ', $mod_text); + $mod_text = preg_replace('#<(br( /)?|dt|/dd|/?(h[1-6]|tr|table|p|li|ul|pre|code|div|hr))[^>]*>#i', '.', $mod_text); + $mod_text = preg_replace('/(\v\s?|\s\s)+/', ' ', $mod_text); + $mod_text = preg_replace('/\s\./', '.', $mod_text); if($mod_ext_charset!='') { // data from external database may have a different charset require_once(WB_PATH.'/framework/functions-utf8.php'); switch($mod_ext_charset) { @@ -293,7 +293,7 @@ } $anchor_text = $mod_text; // make an copy containing html-tags $mod_text = strip_tags($mod_text); - $mod_text = str_replace(array('>','<','&','"',''',''',' '), array('>','<','&','"','\'','\'',"\xC2\xA0"), $mod_text); + $mod_text = str_replace(array('>','<','&','"',''',''',' '), array('>','<','&','"','\'','\'',"\xC2\xA0"), $mod_text); $mod_text = '.'.trim($mod_text).'.'; // Do a fast scan over $mod_text first. This will speedup things a lot. if($func_search_match == 'all') {