X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLDocShredder.h;h=ca555fa7224c14719bffc47f73e2332d3e6665b9;hb=f0557f21878be17ddc75b1bc8f4f86da68c8e604;hp=19ed10f252b19ce917729ad2eb6ddacc56fbb052;hpb=3623eefccfb5fc69e19ad975a3669f51a2a8b276;p=SXSI%2Fxpathcomp.git

diff --git a/XMLDocShredder.h b/XMLDocShredder.h
index 19ed10f..ca555fa 100644
--- a/XMLDocShredder.h
+++ b/XMLDocShredder.h
@@ -15,47 +15,48 @@
 #include <libxml++/libxml++.h>
 #include <libxml++/parsers/textreader.h>
 #include <string>
-#include <unordered_map>
-#include "StorageInterface.h"
+#include "XMLTree.h"
+#include "XMLTreeBuilder.h"
 
 using namespace std;
 using namespace xmlpp;
-/* For Hashmap. Seems fairly well supported */
-using namespace __gnu_cxx; 
 
-typedef pair<int,string> cons_str;
-typedef pair<string,int> cons_int;
 
 class XMLDocShredder
 {
+  void doText();
 public:
-	XMLDocShredder(const string inFileName);
-	XMLDocShredder(const unsigned char * data, TextReader::size_type size);
-	virtual ~XMLDocShredder();
-	virtual void processStartElement();
-	virtual void processEndElement();
-	virtual void processPCDATA();
-	virtual void processAttributes();
-	virtual void processSignificantWhitespace();
-	virtual void processStartDocument(const string docName);
-	virtual void processEndDocument();
-	virtual void processComment();
-	virtual void processProcessingInstruction();
-	virtual void processDocTypeDeclaration();
-	virtual void processUnknownNodeType();
-	virtual void processCDATASection();
-	virtual void parse();
-	virtual int tagID(string);
-	virtual string idTag(int);
-	
-	StorageInterface *storageIfc_;
+  XMLDocShredder(const string inFileName,int sf, bool iet, bool dtc);
+  XMLDocShredder(const unsigned char * data, TextReader::size_type size,int sf, bool iet, bool dtc);
+  virtual ~XMLDocShredder();
+  virtual void processStartElement();
+  virtual void processEndElement();
+  virtual void processPCDATA();
+  virtual void processAttributes();
+  virtual void processSignificantWhitespace();
+  virtual void processStartDocument(const string docName);
+  virtual void processEndDocument();
+  virtual void processComment();
+  virtual void processProcessingInstruction();
+  virtual void processDocTypeDeclaration();
+  virtual void processUnknownNodeType();
+  virtual void processCDATASection();
+  virtual void parse();
+  
+  XMLTree * getXMLTree(){
+    return tree;
+  }
 
 	
-private:
-	TextReader *reader_;
-	void setProperties();
-	unordered_map<int,string> idTags_;
-	unordered_map<string,int> tagsID_;
+ private:
+  XMLTreeBuilder * tb;
+  XMLTree * tree;
+  TextReader *reader_;
+  void setProperties();
+  bool last_text;
+  string buffer; 
+  // used to coalece successive text events
+  // which can occur if we discard pi and comment nodes.
 };
 
 #endif /*XMLDOCSHREDDER_H_*/