X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLTree.h;h=1268cb6b015c369e0bfdc87f7425879d8b88cc47;hb=2134520212940b32a0d0e430df02e71591cc4550;hp=42e02b736abc8f3a55ed0b408e664b301eca387d;hpb=3b0e5683fd88917737d9adb681a13bb0b2160ce5;p=SXSI%2FXMLTree.git

diff --git a/XMLTree.h b/XMLTree.h
index 42e02b7..1268cb6 100644
--- a/XMLTree.h
+++ b/XMLTree.h
@@ -21,7 +21,7 @@
 
 #ifndef XMLTREE_H_
 #define XMLTREE_H_
-#include "TextCollection/TextCollection.h"
+#include "TextCollection/TextCollectionBuilder.h"
 #include <stdio.h>
 #include <stdlib.h>
 #include <cstring>
@@ -37,6 +37,7 @@
 #include <alphabet_mapper.h>
 #include <static_sequence.h>
 using SXSI::TextCollection;
+using SXSI::TextCollectionBuilder;
 
 
 // this constant is used to efficiently compute the child operation in the tree
@@ -88,7 +89,9 @@ class XMLTree {
 	 uint tags_blen, tags_len;
 
    /** The texts in the XML document */
+   TextCollectionBuilder *TextBuilder;
    TextCollection *Text;
+
    /** The texts in the XML document (cached for faster display) */
    vector<string> CachedText;
    
@@ -125,7 +128,7 @@ public:
 	 void print_stats();
 
    /** Data structure constructor */
-   XMLTree() {finished = false; initialized = false;}; 
+   XMLTree() {finished = false; initialized = false; Text = 0; TextBuilder = 0; }; 
  
    /** Data structure destructor */
    ~XMLTree();
@@ -150,6 +153,9 @@ public:
   
    /** IsChild(x,y): returns whether node x is parent of node y. */
    bool IsChild(treeNode x, treeNode y);
+
+   /** IsChild(x,y): returns whether node x is the first child of its parent */
+   bool IsFirstChild(treeNode x);
    
    /** NumChildren(x): number of children of node x. Constant time with the 
     * data structure of Sadakane. */
@@ -188,8 +194,13 @@ public:
     * Very fast in BP. */
    treeNode FirstChild(treeNode x);
    
+   /** LastChild(x): returns the last child of node x. 
+    * Implemented by Kim naively. */
+   treeNode LastChild(treeNode x);
+
    /** NextSibling(x): returns the next sibling of node x, assuming it 
     * exists. */
+
    treeNode NextSibling(treeNode x);
    
    /** PrevSibling(x): returns the previous sibling of node x, assuming it 
@@ -232,7 +243,8 @@ public:
     * preorder than x and not in the subtree of x. Returns NULLT if there 
     * is none. */
    treeNode TaggedFoll(treeNode x, TagType tag);
-     
+
+   treeNode TaggedFollBelow(treeNode x, TagType tag,treeNode root);     
   
    /** TaggedFollowingSibling(x,tag) */
    treeNode TaggedFollowingSibling(treeNode x, TagType tag);
@@ -330,7 +342,7 @@ public:
    TagType RegisterTag(unsigned char *tagname);
 
    bool EmptyText(DocID i) {
-      return Text->EmptyText(i);
+       return Text->EmptyText(i);
    }
    /** Prefix(s): search for texts prefixed by string s. */
    TextCollection::document_result Prefix(uchar const *s) {
@@ -385,6 +397,11 @@ public:
    bool IsLessThan(uchar const *s) {
       return Text->IsLessThan(s);
    }
+   
+   /** Count(s): Global counting  */
+   unsigned Count(uchar const *s) {
+      return Text->Count(s);
+   }
 
    /** CountPrefix(s): counting version of Prefix(s). */
    unsigned CountPrefix(uchar const *s) {
@@ -414,7 +431,7 @@ public:
    /** GetText(d): returns the text corresponding to document with
     * id d. */
    uchar* GetText(DocID d) {
-      return Text->GetText(d);
+       return Text->GetText(d);
    }
 
    uchar* GetCachedText(DocID d) {