[Phpxd-commits] CVS: phpXD/include/parser DTDParser.php,1.1,1.2

SourceForge Headquarters 225 Broadway Suite 1600 San Diego, CA 92101 +1 (858) 422-6466

Update of /cvsroot/phpxd/phpXD/include/parser
In directory usw-pr-cvs1:/tmp/cvs-serv24978/include/parser

Modified Files:
	DTDParser.php 
Log Message:
Parsing of Element declarations implemented.
Parameter Entities are supported, but it is just a hack (lines 26-43).
Attlists, Entities and Notations are still missing.


Index: DTDParser.php
===================================================================
RCS file: /cvsroot/phpxd/phpXD/include/parser/DTDParser.php,v
retrieving revision 1.1
retrieving revision 1.2
diff -C2 -r1.1 -r1.2
*** DTDParser.php	2002/01/26 13:20:01	1.1
--- DTDParser.php	2002/01/27 01:23:10	1.2
***************
*** 16,20 ****
--- 16,24 ----
  class DTDParser {
    var $parseDTD = false;
+   var $parseInternalDTD = false;
+   var $DTDTokens;
    var $parameterEntities;
+   var $elements;
+   var $currentElement;
  
    function parse($str) {
***************
*** 22,40 ****
        while (preg_match('=(.*)\%([a-z,A-Z,0-9,\.]*);(.*)$=sU', $str, $ent)) {
  	$str = preg_replace('=<!--.*-->=sU', '', $str);
! 	while (preg_match('=(.*)<!ENTITY[ ,\n,\r,\t]*\%[ ,\n,\r,\t]*([a-z,A-Z,0-9,\.]*)[ ,\n,\r,\t]*PUBLIC[ ,\n,\r,\t]*"([a-z,A-Z,0-9,\-,/, ]*)"[ ,\n,\r,\t]*"([a-z,A-Z,0-9,\.,\-,_,/]*)"[ ,\n,\r,\t]*>(.*)$=sU', $str, $ent)) {
! 	  if ($this->file_exists($this->currentDir."/".$ent[4])) {
! 	    $this->parameterEntities[$ent[2]] = 
! 	       preg_replace('=<!--.*-->=sU', '', 
  			    implode("", file($this->currentDir."/".$ent[4])));
! 	  }
! 	  $str = $ent[1].$ent[5];
! 	}
  	while (preg_match('=(.*)<!ENTITY[ ,\n,\r,\t]*\%[ ,\n,\r,\t]*([a-z,A-Z,0-9,\.]*)[ ,\n,\r,\t]*"(.*)"[ ,\n,\r,\t]*>(.*)$=sU', $str, $ent)) {
  	  $this->parameterEntities[$ent[2]] = $ent[3];
  	  $str = $ent[1].$ent[4];
- 	}
- 	while (preg_match('=(.*)\%([a-z,A-Z,0-9,\.]*);(.*)$=sU', $str, $ent)) {
- 	  $str = $ent[1].$this->parameterEntities[$ent[2]].$ent[3];
  	}
        }
        $parser = xml_parser_create();
--- 26,44 ----
        while (preg_match('=(.*)\%([a-z,A-Z,0-9,\.]*);(.*)$=sU', $str, $ent)) {
  	$str = preg_replace('=<!--.*-->=sU', '', $str);
!  	while (preg_match('=(.*)<!ENTITY[ ,\n,\r,\t]*\%[ ,\n,\r,\t]*([a-z,A-Z,0-9,\.]*)[ ,\n,\r,\t]*PUBLIC[ ,\n,\r,\t]*"([a-z,A-Z,0-9,\-,/, ]*)"[ ,\n,\r,\t]*"([a-z,A-Z,0-9,\.,\-,_,/]*)"[ ,\n,\r,\t]*>(.*)$=sU', $str, $ent)) {
!  	  if ($this->file_exists($this->currentDir."/".$ent[4])) {
!  	    $this->parameterEntities[$ent[2]] = 
!  	       preg_replace('=<!--.*-->=sU', '', 
  			    implode("", file($this->currentDir."/".$ent[4])));
!  	  }
!  	  $str = $ent[1].$ent[5];
!  	}
  	while (preg_match('=(.*)<!ENTITY[ ,\n,\r,\t]*\%[ ,\n,\r,\t]*([a-z,A-Z,0-9,\.]*)[ ,\n,\r,\t]*"(.*)"[ ,\n,\r,\t]*>(.*)$=sU', $str, $ent)) {
  	  $this->parameterEntities[$ent[2]] = $ent[3];
  	  $str = $ent[1].$ent[4];
  	}
+ 	foreach ($this->parameterEntities as $ent => $body) {
+ 	  $str = str_replace("%".$ent.";", $body, $str);
+  	}
        }
        $parser = xml_parser_create();
***************
*** 86,118 ****
    function parseDoctype($tokens) {
      $token = $tokens[0];
!     if ($this->compareToken($token, "<!DOCTYPE")) {
!       $this->parseDocumentElement(array_slice($tokens, 1));
      }
    }
  
!   function parseDocumentElement($tokens) {
!     $token = $tokens[0];
      $this->documentElement = $token;
!     $token = $tokens[1];
!     if ($this->compareToken($token, "[")) {
!       $this->parseInternalDTD(array_slice($tokens, 2));
      }
!     if ($this->compareToken($token, "SYSTEM")) {
!       $this->parseExternalSystemDTD(array_slice($tokens, 2));
      }
!     if ($this->compareToken($token, "PUBLIC")) {
!       $this->parseExternalPublicDTD(array_slice($tokens, 2));
      }
    }
  
!   function parseInternalDTD($tokens) {
    }
  
!   function parseExternalSystemDTD($tokens) {
    }
  
!   function parseExternalPublicDTD($tokens) {
!     $tokenPublicID = $this->removeQuotes($tokens[0]);
!     $tokenURI = $this->removeQuotes($tokens[1]);
      switch ($tokenPublicID) {
      case "-//W3C//DTD XHTML 1.0 Strict//EN": {
--- 90,174 ----
    function parseDoctype($tokens) {
      $token = $tokens[0];
!     if ($token == "<!DOCTYPE") {
!       $this->parseDocumentElement($tokens, 1);
      }
    }
  
!   function parseDocumentElement($tokens, $offset) {
!     $token = $tokens[$offset];
      $this->documentElement = $token;
!     $token = $tokens[$offset + 1];
!     if ($token == "[") {
!       $this->parseInternalDTD($tokens, $offset + 2);
! //        foreach ($this->elements as $element) {
! // 	 echo $element->tagName." -- ";
! // 	 if ($element->children == null) {
! // 	   echo "EMPTY";
! // 	 }
! // 	 else {
! // 	   echo $element->children->toString();
! // 	 }
! // 	 echo "<br>";
! //        }
!       exit;
!     }
!     if ($token == "SYSTEM") {
!       $this->parseExternalSystemDTD($tokens, $offset + 2);
!     }
!     if ($token == "PUBLIC") {
!       $this->parseExternalPublicDTD($tokens, $offset + 2);
!     }
!   }
! 
!   function parseInternalDTD($tokens, $offset) {
!     $token = $tokens[$offset];
!     switch ($token) {
!     case "<!ELEMENT": {
!       $this->parseElement($tokens, $offset + 1);
!        $currentElement = $this->currentElement;
!        $currentElement->children =& $this->currentElement->children;
!        echo $this->currentElement->tagName." -- ";
!        unset($this->currentElement);
!        if ($currentElement->children != null) {
!  	echo $currentElement->children->toString()."<br>";
!        }
!        $this->elements[] =& $currentElement;
!       $offset = $this->skipToGt($tokens, $offset + 1);
!       break;
!     }
!     case "<!ATTLIST": {
!       $offset = $this->skipToGt($tokens, $offset + 1);
!       break;
!     }
!     case "<!ENTITY": {
!       $offset = $this->skipToGt($tokens, $offset + 1);
!       break;
      }
!     case "<!NOTATION": {
!       $offset = $this->skipToGt($tokens, $offset + 1);
!       break;
!     }
      }
!     if (($tokens[$offset] != "]>") && ($tokens[$offset] != "]")) {
!       $this->parseInternalDTD($tokens, $offset);
      }
    }
  
!   function skipToGt($tokens, $offset) {
!     while ($tokens[$offset] != ">") {
!       $offset++;
!     }
!     return $offset + 1;;
    }
  
!   function parseExternalSystemDTD($tokens, $offset) {
!     $token = $this->removeQuotes($tokens[$offset]);
!     $DTDParser = new DTDParser();
!     $DTDParser->parseFile($token);
    }
  
!   function parseExternalPublicDTD($tokens, $offset) {
!     $tokenPublicID = $this->removeQuotes($tokens[$offset]);
!     $tokenURI = $this->removeQuotes($tokens[$offset + 1]);
      switch ($tokenPublicID) {
      case "-//W3C//DTD XHTML 1.0 Strict//EN": {
***************
*** 123,135 ****
      }
      }
-     //    echo $tokenURI;
    }
  
!   function compareToken($tok1, $tok2) {
!     if ($tok1 == $tok2) {
!       return true;
!     } 
      else {
!       return false;
      }
    }
--- 179,300 ----
      }
      }
    }
  
!   function parseElement($tokens, $offset) {
!     $name = $tokens[$offset];
!     $this->currentElement = new DTDElement($name);
!     $this->parseElementChilds($tokens, $offset + 1);
!   }
! 
!   function parseElementChilds($tokens, $offset) {
!     $token = $tokens[$offset];
!     if ($token == "(") {
!       $nestlevel = 1; $cursor = 1;
!       while (($nestlevel > 0) && 
! 	     (($tokens[$offset + $cursor] != ",") || ($nestlevel > 1))) {
! 	if ($tokens[$offset + $cursor] == "(") {
! 	  $nestlevel++;
! 	}
! 	if (substr($tokens[$offset + $cursor], 0, 1) == ")") {
! 	  $nestlevel--;
! 	}
! 	$cursor++;
!       }
!       if ($nestlevel == 1) {
! 	if ($this->currentElement->children == null) {
! 	  $this->currentElement->children = new DTDElementSequence();
! 	} 
! 	else {
! 	  $newChild = new DTDElementSequence();
! 	  $newChild->parent =& $this->currentElement->children;
! 	  $newChild->parent->appendChild($newChild);
! 	  unset($this->currentElement->children);
! 	  $this->currentElement->children =& $newChild;
! 	}
!       } 
!       else {
! 	if ($this->currentElement->children == null) {
! 	  $this->currentElement->children = new DTDElementChoice();
! 	}
! 	else {
! 	  $newChild = new DTDElementChoice();
! 	  $newChild->parent =& $this->currentElement->children;
! 	  $newChild->parent->appendChild($newChild);
! 	  unset($this->currentElement->children);
! 	  $this->currentElement->children =& $newChild;
! 	}
!       }
!       $this->parseElementChilds($tokens, $offset + 1);
!       return;
!     }
!     if ($token[0] == ")") {
!       if (($token[strlen($token) - 1] == "?") || 
! 	  ($token[strlen($token) - 1] == "+") || 
! 	  ($token[strlen($token) - 1] == "*")) {
! 	$this->currentElement->children->setNumber($token[strlen($token) - 1]);
!       }
!       if ($this->currentElement->children->parent != null) {
! 	$parent =& $this->currentElement->children->parent;
!   	$this->currentElement->children =& $parent;
! 	$this->parseElementChilds($tokens, $offset + 1);
! 	return;
!       }
!       else {
! 	// assert: the dtd is correct, so it is the last return.
! 	return;
!       }
!     }
!     if (($token == ",") || ($token == "|") ||
! 	($token == "?") || ($token == "+") || ($token == "*")) {
!       $this->parseElementChilds($tokens, $offset + 1);
!       return;
!     }
!     if ($token == ">") {
!       return;
!     }
!     $this->parseElementChild($tokens, $offset);
!     $this->parseElementChilds($tokens, $offset + 1);
!   }
! 
!   function parseElementChild($tokens, $offset) {
!     $token = $tokens[$offset];
!     if ($token == "ANY") {
!       $this->currentElement->children = 
! 	 new DTDElementChild("", "", true);
!       return;
!     }
!     if ($token == "EMPTY") {
!       $this->currentElement->children = null;
!       return;
!     }
!     if ($token == "#PCDATA") {
!       if ($this->currentElement->children == null) {
! 	$this->currentElement->children = 
! 	   new DTDElementChild("", "", true, true);
! 	return;
!       }
!       else {
! 	$this->currentElement->children->
! 	   appendChild(new DTDElementChild("", "", true, true));
! 	return;
!       }
!     }
!     if (($token[strlen($token) - 1] == "?") || 
! 	($token[strlen($token) - 1] == "+") || 
! 	($token[strlen($token) - 1] == "*")) {
!       $number = $token[strlen($token) - 1];
!       $token = substr($token, 0, strlen($token) - 1);
!     }
      else {
!       $number = "";
!     }
!     $child = new DTDElementChild($token, $number);
!     if ($this->currentElement->children == null) {
!       $this->currentElement->children =& $child;
!       return;
!     }
!     else {
!       $this->currentElement->children->appendChild($child);
!       return;
      }
    }
***************
*** 146,151 ****
      $data = trim($data);
  
-     //  echo "<pre>".$data."</pre>";
-     
      if ($data == "<!DOCTYPE") {
        $this->DTDTokens[] = $data;
--- 311,314 ----
***************
*** 155,159 ****
  
      if ($this->parseDTD) {
!       $this->DTDTokens[] = $data;
        if ($data == "[") {
  	$this->parseInternalDTD = true;
--- 318,324 ----
  
      if ($this->parseDTD) {
!       if ($data != "") {
! 	$this->DTDTokens[] = $data;
!       }
        if ($data == "[") {
  	$this->parseInternalDTD = true;
***************
*** 171,180 ****
  
    function handleExternalEntityRef($parser, $openEntityNames, $base, $systemId, $publicId) {
-     echo "hier";
      return true;
    }
  
    function handleUnparsedEntityDecl($parser, $entityName, $base, $systemId, $publicId, $notationName) {
-     echo "hier";
      return true;
    }
--- 336,343 ----
***************
*** 211,214 ****
--- 374,503 ----
      echo "<strong>phpXD error:</strong> ".$message;
      exit;
+   }
+ }
+     
+ class DTDNode {
+ }
+ 
+ class DTDElement extends DTDNode {
+   var $tagName;
+   var $children;
+ 
+   function DTDElement($tagName) {
+     $this->tagName = $tagName;
+     $this->children = null;
+   }
+ }
+ 
+ class DTDElementChild extends DTDNode {
+   // Number of that child
+   // 1 = one time
+   // 2 = ? - zero or one time
+   // 3 = * - zero or more times
+   // 4 = + - one or more times
+   var $number = 1;
+ 
+   var $any = false;
+   var $pcdata = false;
+   var $tagName;
+   var $id;
+ 
+   function DTDElementChild($tagName, $number, $any = false, $pcdata = false) {
+     if ($any) {
+       $this->any = true;
+       return;
+     }
+ 
+     if ($pcdata) {
+       $this->pcdata = true;
+       return;
+     }
+ 
+     $this->tagName = $tagName;
+     $this->setNumber($number);
+   }
+ 
+   function setNumber($number) {
+     switch ($number) {
+     case "?": {
+       $this->number = 2;
+       break;
+     }
+     case "+": {
+       $this->number = 3;
+       break;
+     }
+     case "*": {
+       $this->number = 4;
+       break;
+     }
+     }
+   }
+ 
+   function toString() {
+     if ($this->any) {
+       return "ANY";
+     }
+     if ($this->pcdata) {
+       return "#PCDAZA";
+     }
+     return $this->tagName." / ".$this->number;
+   }
+ }
+ 
+ class DTDElementSequence extends DTDElementChild {
+   var $sequence;
+   var $length = 0;
+   var $parent = null;
+ 
+   function DTDElementSequence() {
+   }
+ 
+   function appendChild(&$child) {
+     $this->sequence[$this->length++] =& $child;
+   }
+ 
+   function toString() {
+     $str = "(";
+     $count = 0;
+     for ($count = 0; $count < $this->length; ++$count) {
+       $child =& $this->sequence[$count];
+       if ($count != 0) {
+ 	$str .= " , ";
+       }
+       $str .= $child->toString();
+     }
+     $str .= ") / ".$this->number;
+     return $str;
+   }
+ }
+ 
+ class DTDElementChoice extends DTDElementChild  {
+   var $choices;
+   var $length = 0;
+   var $parent = null;
+ 
+   function DTDElementChoice() {
+   }
+ 
+   function DTDElementChoice() {
+   }
+ 
+   function appendChild(&$child) {
+     $this->choices[$this->length++] =& $child;
+   }
+ 
+   function toString() {
+     $str = "(";
+     $count = 0;
+     for ($count = 0; $count < $this->length; ++$count) {
+       $child =& $this->choices[$count];
+       if ($count != 0) {
+ 	$str .= " | ";
+       }
+       $str .= $child->toString();
+     }
+     $str .= ") / ".$this->number;
+     return $str;
    }
  }