X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLDocShredder.cpp;h=3daaf45dfbbcdc4a53cef508465a013e42d1d29f;hb=9be0c0e2a5597148fdc3a3cca2bdaf69da0aa27d;hp=6e9c41a3e5f2ff138bedc4a91edd4cec972e7b1f;hpb=3623eefccfb5fc69e19ad975a3669f51a2a8b276;p=SXSI%2Fxpathcomp.git

diff --git a/XMLDocShredder.cpp b/XMLDocShredder.cpp
index 6e9c41a..3daaf45 100644
--- a/XMLDocShredder.cpp
+++ b/XMLDocShredder.cpp
@@ -18,7 +18,7 @@
 
 #include <iostream>
 #include "XMLDocShredder.h"
-#include "OCamlStorageInterface.h"
+#include "SXSIStorageInterface.h"
 #include <libxml++/exceptions/parse_error.h>
 #include "Utils.h"
 
@@ -55,20 +55,19 @@ void XMLDocShredder::setProperties(){
 XMLDocShredder::XMLDocShredder(const unsigned char * data,
 			       TextReader::size_type size) 			
 {
+  last_text = false;
   reader_ = new TextReader(data,size,"");
   setProperties();
-  storageIfc_ = new OCamlStorageInterface();
-  //tagsID_ = new unordered_map<int,string>(107);
-  //idTags_ = new unordered_map<string,int>(107);
+  storageIfc_ = new SXSIStorageInterface();
 }
 
 XMLDocShredder::XMLDocShredder(const string inFileName)
 {
+  last_text = false;
   reader_ = new TextReader(inFileName);
   setProperties();
-  storageIfc_ = new OCamlStorageInterface();
-  //  tagsID_ = new unordered_map<int,string>(107);
-  // idTags_ = new unordered_map<string,int>(107);
+  storageIfc_ = new SXSIStorageInterface();
+
 }
 
 XMLDocShredder::~XMLDocShredder()
@@ -78,23 +77,16 @@ XMLDocShredder::~XMLDocShredder()
 
 }
 
-int XMLDocShredder::tagID(string name)
-{
-  int res = tagsID_[name];
-  return  res;
-}
-string XMLDocShredder::idTag(int id)
-{
-
-  return  idTags_[id];
-}
-
 
 void XMLDocShredder::processStartElement()
 {
 	// fetch element name; this will be the full qualified name
 	ustring name = reader_->get_name();
 	bool empty = false;
+	
+	if (!last_text)
+	  storageIfc_->newText(""); //prevText
+	last_text = false;
 
 	storageIfc_->newChild(name);
 
@@ -113,8 +105,8 @@ void XMLDocShredder::processStartElement()
 
 	
 	if (empty){
-	  DPRINT("Node " << name <<" is empty!\n")
-	    storageIfc_->nodeFinished();
+	    storageIfc_->newText("");  //myText
+	    storageIfc_->nodeFinished(name);	   
 	};
 
 
@@ -125,18 +117,24 @@ void XMLDocShredder::processStartElement()
 
 void XMLDocShredder::processEndElement()
 {
-	// tell the storage interface that the current node has been completely processed
-	storageIfc_->nodeFinished();
+  // tell the storage interface that the current node has been completely processed
+  if (!last_text)
+    storageIfc_->newText(""); //nextText of previous node
+  last_text = false;
+  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())
 	{	  
-	  storageIfc_->newChild("<$>");
 	  storageIfc_->newText(reader_->get_value());
+	  last_text = true;
 	}
+	else 
+	  storageIfc_->newText("");
 }
 
 void XMLDocShredder::processAttributes()
@@ -157,7 +155,7 @@ void XMLDocShredder::processAttributes()
 		if ((name.find(nspaceStr.c_str(), 0, 5)) == 0)
 		{
 			storageIfc_->newChild(":" + value);
-			storageIfc_->nodeFinished();	
+			storageIfc_->nodeFinished(":" + value);	
 		}
 		
 		/* otherwise, this is an ordinary attribute, so we construct a new child node of the 
@@ -167,39 +165,38 @@ void XMLDocShredder::processAttributes()
 		 
 		else
 		{
-			storageIfc_->newChild(name);
-			storageIfc_->newChild("<$>");
-			storageIfc_->newText(value);
-			storageIfc_->nodeFinished();
-			//			storageIfc_->nodeFinished();
+		  storageIfc_->newText(""); //prevText
+		  storageIfc_->newChild(name);
+		  storageIfc_->newText(value);
+		  storageIfc_->nodeFinished(name);
 		}
 	}
 	while (reader_->move_to_next_attribute());
-	storageIfc_->nodeFinished();
+	storageIfc_->newText(""); //nextText
+	storageIfc_->nodeFinished("<@>");
 }
 
 void XMLDocShredder::processSignificantWhitespace()
 {
-	ustring value = reader_->get_value();
-	
+	ustring value = reader_->get_value();	
 	// each significant whitespace sequence constructs a text node
-	storageIfc_->newChild("<$>");
 	storageIfc_->newText(value);
-	//storageIfc_->nodeFinished();
 	
 }
 
 void XMLDocShredder::processStartDocument(const string docName)
 {
   // tell storage interface to construct the document name
-  //  storageIfc_->newChild("");  
+  storageIfc_->newChild("");  
+  
 }
 
 void XMLDocShredder::processEndDocument()
 {
 	/* tell the storage interface that document parsing has finished, and structures
 	 * can now be written to disk. */
-  //  storageIfc_->nodeFinished();
+  storageIfc_->newText("");
+  storageIfc_->nodeFinished("");
   storageIfc_->parsingFinished();	
 }
 
@@ -237,10 +234,10 @@ 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_->newChild("<$>");
 	storageIfc_->newText(value);
-	//	storageIfc_->nodeFinished();
+	last_text = true;
 
 }