X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLDocShredder.cpp;h=1516c9fc1cd724c346c66877b250af4d94d843f3;hb=451e60ad59e35344dff62da5ca27fcd5eec1bff9;hp=1c25d611d336558886d096132ef871f87bce481a;hpb=496df5f7d3e6f8271763314f2067719cc2904c71;p=SXSI%2Fxpathcomp.git

diff --git a/XMLDocShredder.cpp b/XMLDocShredder.cpp
index 1c25d61..1516c9f 100644
--- a/XMLDocShredder.cpp
+++ b/XMLDocShredder.cpp
@@ -53,21 +53,24 @@ void XMLDocShredder::setProperties(){
   
 }
 XMLDocShredder::XMLDocShredder(const unsigned char * data,
-			       TextReader::size_type size) 			
+			       TextReader::size_type size,
+			       int sf, 
+			       bool iet, 
+			       bool dtc) 			
 {
   last_text = false;
   reader_ = new TextReader(data,size,"");
   setProperties();
-  storageIfc_ = new SXSIStorageInterface();
+  storageIfc_ = new SXSIStorageInterface(sf,iet,dtc);
   buffer = "";
 }
 
-XMLDocShredder::XMLDocShredder(const string inFileName)
+XMLDocShredder::XMLDocShredder(const string inFileName,int sf, bool iet, bool dtc)
 {
   last_text = false;
   reader_ = new TextReader(inFileName);
   setProperties();
-  storageIfc_ = new SXSIStorageInterface();
+  storageIfc_ = new SXSIStorageInterface(sf,iet,dtc);
   buffer = "";
 }
 
@@ -85,9 +88,6 @@ void XMLDocShredder::processStartElement()
 	ustring name = reader_->get_name();
 	bool empty = false;
 	
-	storageIfc_->newText(buffer); //prevText
-	buffer.erase();
-	
 	storageIfc_->newChild(name);
 
 	/* We must be really carefull here. calling process attributes moves
@@ -105,8 +105,7 @@ void XMLDocShredder::processStartElement()
 
 	
 	if (empty){
-	    storageIfc_->newText("");  //myText
-	    storageIfc_->nodeFinished(name);	   
+	  storageIfc_->nodeFinished(name);	   
 	};
 
 
@@ -114,22 +113,19 @@ void XMLDocShredder::processStartElement()
 
 void XMLDocShredder::processEndElement()
 {
-  // tell the storage interface that the current node has been completely processed
-  storageIfc_->newText(buffer); //prevText
-  buffer.erase();
+  // tell the storage interface that the current node has been completely processed  
   storageIfc_->nodeFinished(reader_->get_name());
 }
 
 void XMLDocShredder::processPCDATA()
 {
-	// send the content of this PCDATA node to the storage interface as a text node
-         
-	if (reader_->has_value())
-	{
-	  buffer += reader_->get_value();
-
-	};
-
+  // send the content of this PCDATA node to the storage interface as a text node
+  
+  if (reader_->has_value()){
+    storageIfc_->newChild("<$>");
+    storageIfc_->newText(reader_->get_value());
+    storageIfc_->nodeFinished("<$>");
+  };
 }
 
 void XMLDocShredder::processAttributes()
@@ -137,7 +133,6 @@ void XMLDocShredder::processAttributes()
 	reader_->move_to_first_attribute();
 		
 	string nspaceStr = "xmlns";
-	storageIfc_->newText(""); //prevText
 	storageIfc_->newChild("<@>");
 	do
 	{
@@ -161,22 +156,26 @@ void XMLDocShredder::processAttributes()
 		 
 		else
 		{
-		  storageIfc_->newText(""); //prevText
-		  storageIfc_->newChild(name);
+		  string attname = "<@>"+name;
+		  storageIfc_->newChild(attname);
+		  storageIfc_->newChild("<@$>");
 		  storageIfc_->newText(value);
-		  storageIfc_->nodeFinished(name);
+		  storageIfc_->nodeFinished("<@$>");
+		  storageIfc_->nodeFinished(attname);
 		}
 	}
 	while (reader_->move_to_next_attribute());
-	storageIfc_->newText(""); //nextText
 	storageIfc_->nodeFinished("<@>");
 }
 
 void XMLDocShredder::processSignificantWhitespace()
 {
-  // each significant whitespace sequence constructs a text node
-  buffer += reader_->get_value();	
-	
+  
+  if (reader_->has_value()){
+    storageIfc_->newChild("<$>");
+    storageIfc_->newText(reader_->get_value());
+    storageIfc_->nodeFinished("<$>");
+  };  
 }
 
 void XMLDocShredder::processStartDocument(const string docName)
@@ -190,7 +189,6 @@ void XMLDocShredder::processEndDocument()
 {
 	/* tell the storage interface that document parsing has finished, and structures
 	 * can now be written to disk. */
-  storageIfc_->newText("");
   storageIfc_->nodeFinished("");
   storageIfc_->parsingFinished();	
 }
@@ -229,10 +227,11 @@ void XMLDocShredder::processCDATASection()
 	 * model.  Instead, we simply pass the converted text value to the storage interface as 
 	 * a text node attached to the current context node.
 	 */
-  
-	ustring value = reader_->get_value();
-	storageIfc_->newText(value);
-	last_text = true;
+  if (reader_->has_value()){
+    storageIfc_->newChild("<$>");
+    storageIfc_->newText(reader_->get_value());
+    storageIfc_->nodeFinished("<$>");
+  };
 
 }