1: <?php
  2:     class QLexer extends QBaseClass {
  3:         private $objRegexMode = array();
  4:         private $objModeHandlers = array();
  5:         private $objTokenModeMap = array();
  6:         private $objTokens = array();
  7:         private $objModeStack;
  8:     
  9:         const UNMATCHED = "__UNMATCHED__";
 10:         const DefaultMode = "default_mode";
 11:     
 12:         public function __construct($strStartMode = QLexer::DefaultMode) {
 13:             $this->objModeStack = new QStack();
 14:             $this->objModeStack->Push($strStartMode);
 15:         }
 16:     
 17:         public function addPattern($strPattern, $strTokenName, $strMode = QLexer::DefaultMode) {
 18:             if(!isset($this->objRegexMode[$strMode])) {
 19:                 $this->objRegexMode[$strMode] = new QRegex();
 20:             }
 21:             $this->objRegexMode[$strMode]->addPattern($strPattern,$strTokenName);
 22:         }
 23:     
 24:         public function addEntryPattern($strPattern, $strTokenName, $strMode = QLexer::DefaultMode, $strNewMode) {
 25:             if(!isset($this->objRegexMode[$strMode])) {
 26:                 $this->objRegexMode[$strMode] = new QRegex();
 27:             }
 28:             $this->objRegexMode[$strMode]->addPattern($strPattern,$strTokenName);
 29:             $this->objTokenModeMap[$strTokenName] = $strNewMode;
 30:         }
 31:     
 32:         public function addExitPattern($strPattern, $strTokenName, $strMode, $strNewMode = QLexer::DefaultMode) {
 33:             if(!isset($this->objRegexMode[$strMode])) {
 34:                 $this->objRegexMode[$strMode] = new QRegex();
 35:             }
 36:             $this->objRegexMode[$strMode]->addPattern($strPattern,$strTokenName);
 37:             $this->objTokenModeMap[$strTokenName] = "__exit";
 38:         }
 39:     
 40:         public function Tokenize(&$strRaw) {
 41:             $objTokens = array();
 42:             $intLength = strlen($strRaw);
 43:             while (is_array($objParsed = $this->Reduce($strRaw))) {
 44:                 list($strUnmatched,$strMatched,$strToken) = $objParsed;
 45:                 if($strUnmatched != "") {
 46:                     array_push($objTokens,array("token"=>QLexer::UNMATCHED,'raw'=>$strUnmatched));
 47:                 }
 48:     
 49:                 if(array_key_exists($strToken,$this->objTokenModeMap)) {
 50:                     if($this->objTokenModeMap[$strToken] == "__exit") {
 51:                         $this->objModeStack->Pop();
 52:                         array_push($objTokens,array("token"=>$strToken,'raw'=>$strMatched));
 53:                         return $objTokens;
 54:                     }
 55:                     else {
 56:                         $this->objModeStack->Push($this->objTokenModeMap[$strToken]);
 57:                         array_push($objTokens,array("token"=>$strToken,'raw'=>$this->Tokenize($strRaw)));
 58:                     }
 59:                 }
 60:                 else {
 61:                     array_push($objTokens,array("token"=>$strToken,'raw'=>$strMatched));
 62:                 }
 63:             }
 64:     
 65:             if($objParsed) {
 66:                 array_push($objTokens,array("token"=>QLexer::UNMATCHED,'raw'=>$strRaw));
 67:             }
 68:     
 69:             /**
 70:              * If we get here, we've parsed everything possible.  Pop one 
 71:              * off the stack and see if we can continue.
 72:              */
 73:             if($this->objModeStack->Size() > 1) // Don't short the stack
 74:                 $this->objModeStack->Pop();
 75:     
 76:             /**
 77:              * Try a little data cleanup
 78:              */
 79:             if(is_array($objParsed = $this->Reduce($strRaw))) {
 80:                 list($strUnmatched,$strMatched,$strToken) = $objParsed;
 81:                 if($strMatched != "") {
 82:                     // We got a match when we pop the stack ...
 83:                     // Replace things the way they should be.
 84:                     $objLastToken = array_pop($objTokens);
 85:                     $objLastToken["raw"] = $strUnmatched;
 86:                     array_push($objTokens,$objLastToken);
 87:                     $strRaw = $strMatched . $strRaw;
 88:                 }
 89:             }
 90:     
 91:             return $objTokens;
 92:         }
 93:     
 94:         private function Reduce(&$strRaw) {
 95:             // Are we in a valid mode?
 96:             if (!isset($this->objRegexMode[$this->objModeStack->PeekLast()])) {
 97:                 return FALSE;
 98:             }
 99:     
100:             // Empty String?
101:             if ($strRaw === "") {
102:                 return TRUE;
103:             }
104:     
105:             $strToken = $this->objRegexMode[$this->objModeStack->PeekLast()]->match($strRaw, $strMatch);
106:             if ($strToken) {
107:                 // Where in the string did we match?
108:                 $intMatchPosition = strpos($strRaw, $strMatch);
109:                 $strUnparsed = substr($strRaw, 0, $intMatchPosition);
110:                 $strRaw = substr($strRaw, $intMatchPosition + strlen($strMatch));
111:                 return array($strUnparsed, $strMatch, $strToken);
112:             }
113:     
114:             return TRUE;
115:         }
116:     }