X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLTree.h;h=9b79ae3eab6040b473c0eb65473aabf4ea4836bd;hb=06201fcce6255906ad6b4d305b8e643c634780a3;hp=a00487ac710eb82d96cdcf94437fa541fff3692b;hpb=dd9992dcd5366f37820c24ed7cddf24ecbc0d549;p=SXSI%2FXMLTree.git

diff --git a/XMLTree.h b/XMLTree.h
index a00487a..9b79ae3 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
@@ -84,9 +85,15 @@ class XMLTree {
 		      
    /** Tag sequence represented with a data structure for rank and select */
    static_sequence *Tags;
+	 uint * tags_fix;
+	 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;
    
    /** Flag indicating whether the whole data structure has been constructed or 
     * not. If the value is true, you cannot add more texts, nodes, etc. */
@@ -118,9 +125,10 @@ class XMLTree {
    bool disable_tc;
    
 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();
@@ -202,9 +210,21 @@ public:
     * is none. */
    treeNode TaggedDesc(treeNode x, TagType tag);
 
-   /** TaggedNext(x,tag): returns the first node tagged tag with larger 
-    * preorder than x. Returns NULT if there is none. */
-   treeNode TaggedNext(treeNode x, TagType tag);
+
+   treeNode TaggedBelow(treeNode x, TagType *childtags, unsigned int ctlen,
+				 TagType *desctags, unsigned int dtlen);
+   
+   treeNode TaggedNext(treeNode x, TagType *childtags, unsigned int ctlen,
+				TagType *folltags, unsigned int flen,treeNode root);
+
+   treeNode TaggedDescOnly(treeNode x, TagType *desctags, unsigned int dtlen);
+   
+   treeNode TaggedDescOrFollOnly(treeNode x, TagType *folltags, unsigned int flen,
+			   treeNode root);
+
+   treeNode TaggedFollOnly(treeNode x, TagType *folltags, unsigned int flen,
+			   treeNode root);
+   
 
    /** TaggedPrec(x,tag): returns the first node tagged tag with smaller 
     * preorder than x and not an ancestor of x. Returns NULLT if there 
@@ -215,7 +235,16 @@ 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);
+
+   /** TaggedAncestor(x, tag): returns the closest ancestor of x tagged 
+     * tag. Return NULLT is there is none. */
+   treeNode TaggedAncestor(treeNode x, TagType tag);
+ 
    /** PrevText(x): returns the document identifier of the text to the left of 
     * node x, or NULLT if x is the root node. */
    DocID PrevText(treeNode x);
@@ -238,6 +267,7 @@ public:
    
    /** ParentNode(d): returns the parent node of document identifier d. */
    treeNode ParentNode(DocID d);
+   treeNode PrevNode(DocID d);
 
    /** OpenDocument(empty_texts,sample_rate_text,dtc): initilizes the construction
     * of the data structure for the XML document. Parameter empty_texts 
@@ -304,7 +334,7 @@ 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) {
@@ -359,6 +389,11 @@ 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) {
@@ -388,7 +423,13 @@ 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) {
+     uchar * str = (uchar*) calloc(sizeof(char),(CachedText.at(d).size() + 1));
+     strcpy((char*) str,(const char*) CachedText.at(d).c_str());
+     return (uchar*) (str);
    }
    
    TextCollection *getTextCollection() {