Commit before changing Tree.ml interface

[SXSI/xpathcomp.git] / XMLDocShredder.cpp

diff --git a/XMLDocShredder.cpp b/XMLDocShredder.cpp
index f81251e..1516c9f 100644 (file)
--- a/XMLDocShredder.cpp
+++ b/XMLDocShredder.cpp
@@ -53,21 +53,25 @@ 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 = "";
 }
 
 XMLDocShredder::~XMLDocShredder()
@@ -84,10 +88,6 @@ void XMLDocShredder::processStartElement()
        ustring name = reader_->get_name();
        bool empty = false;
        
-       if (!last_text)
-         storageIfc_->newText(""); //prevText
-       last_text = false;
-
        storageIfc_->newChild(name);
 
        /* We must be really carefull here. calling process attributes moves
@@ -105,46 +105,34 @@ void XMLDocShredder::processStartElement()
 
        
        if (empty){
-         DPRINT("Node " << name <<" is empty!\n")
-           storageIfc_->newText("");  //myText
-           storageIfc_->nodeFinished(name);
-           storageIfc_->newText("");  //nextText
+         storageIfc_->nodeFinished(name);         
        };
 
 
-
-
-
 }
 
 void XMLDocShredder::processEndElement()
 {
-  // tell the storage interface that the current node has been completely processed
-  if (!last_text)
-    storageIfc_->newText(""); //nextText of previous node
-  last_text = false;
+  // 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())
-       {         
-         storageIfc_->newChild("<$>");
-         storageIfc_->newText(reader_->get_value());
-         last_text = true;
-       }
-       else 
-         storageIfc_->newText("");
+  // send the content of this PCDATA node to the storage interface as a text node
+  
+  if (reader_->has_value()){
+    storageIfc_->newChild("<$>");
+    storageIfc_->newText(reader_->get_value());
+    storageIfc_->nodeFinished("<$>");
+  };
 }
 
 void XMLDocShredder::processAttributes()
 {
        reader_->move_to_first_attribute();
                
-       string nspaceStr = "xmlns";     
+       string nspaceStr = "xmlns";
        storageIfc_->newChild("<@>");
        do
        {
@@ -168,10 +156,12 @@ void XMLDocShredder::processAttributes()
                 
                else
                {
-                       storageIfc_->newChild(name);
-                       storageIfc_->newChild("<$>");
-                       storageIfc_->newText(value);
-                       storageIfc_->nodeFinished("<$>");
+                 string attname = "<@>"+name;
+                 storageIfc_->newChild(attname);
+                 storageIfc_->newChild("<@$>");
+                 storageIfc_->newText(value);
+                 storageIfc_->nodeFinished("<@$>");
+                 storageIfc_->nodeFinished(attname);
                }
        }
        while (reader_->move_to_next_attribute());
@@ -180,19 +170,18 @@ void XMLDocShredder::processAttributes()
 
 void XMLDocShredder::processSignificantWhitespace()
 {
-       ustring value = reader_->get_value();
-       
-       // each significant whitespace sequence constructs a text node
-       storageIfc_->newChild("<$>");
-       storageIfc_->newText(value);
-
-       
+  
+  if (reader_->has_value()){
+    storageIfc_->newChild("<$>");
+    storageIfc_->newText(reader_->get_value());
+    storageIfc_->nodeFinished("<$>");
+  };  
 }
 
 void XMLDocShredder::processStartDocument(const string docName)
 {
   // tell storage interface to construct the document name
-  storageIfc_->newChild("ROOT");  
+  storageIfc_->newChild("");  
   
 }
 
@@ -200,7 +189,7 @@ void XMLDocShredder::processEndDocument()
 {
        /* tell the storage interface that document parsing has finished, and structures
         * can now be written to disk. */
-  storageIfc_->nodeFinished("ROOT");
+  storageIfc_->nodeFinished("");
   storageIfc_->parsingFinished();      
 }
 
@@ -238,12 +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_->newChild("<$>");
-       storageIfc_->newText(value);
-       last_text = true;
-       //      storageIfc_->nodeFinished();
+  if (reader_->has_value()){
+    storageIfc_->newChild("<$>");
+    storageIfc_->newText(reader_->get_value());
+    storageIfc_->nodeFinished("<$>");
+  };
 
 }