X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLTree.cpp;h=2a1da8446a816ab17117963aea34bc785b4945ca;hb=dc02a833a150dbef202bc14aca74c51360d4a631;hp=ae80fc2a489680d270c6fba2d59939a8c9bddc00;hpb=b53633fb64f387edb5cebefbb3308b6347b2389c;p=SXSI%2FXMLTree.git

diff --git a/XMLTree.cpp b/XMLTree.cpp
index ae80fc2..2a1da84 100644
--- a/XMLTree.cpp
+++ b/XMLTree.cpp
@@ -212,19 +212,18 @@ XMLTree *XMLTree::Load(unsigned char *filename, int sample_rate_text)
     s_tree+= XML_Tree->Tags->size();
 
 		/// FIXME:UGLY tests!
-		/*uint * seq = new uint[XML_Tree->tags_len];
+		uint * seq = new uint[XML_Tree->tags_len];
 		for(uint i=0;i<XML_Tree->tags_len;i++)
 			seq[i] = get_field(XML_Tree->tags_fix,XML_Tree->tags_blen,i);
 		cout << "Tags test: " << XML_Tree->Tags->test(seq,XML_Tree->tags_len) << endl;
-		delete [] seq;*/
+		delete [] seq;
 		/// End ugly tests
 
     s_text = ftell(fp);
 
     // loads the texts
     if (!XML_Tree->disable_tc){
-      XML_Tree->Text = TextCollection::InitTextCollection(sample_rate_text);
-      XML_Tree->Text->Load(fp,sample_rate_text);
+      XML_Tree->Text = TextCollection::Load(fp,sample_rate_text);
       int sst;
       int st;
       ufread(&sst, sizeof(int),1,fp);
@@ -284,7 +283,9 @@ XMLTree::~XMLTree()
     Tags = NULL;
 
     //Text->~TextCollection();
-    delete Text;
+    delete TextBuilder; 
+    TextBuilder = NULL;
+    delete Text; 
     Text = NULL;
 
     initialized = false;
@@ -356,7 +357,9 @@ bool XMLTree::IsChild(treeNode x, treeNode y)
     if (!is_ancestor(Par, x, y)) return false;
     return depth(Par, x) == (depth(Par, y) + 1);
  }
-
+bool XMLTree::IsFirstChild(treeNode x){
+  return ((x != NULLT)&&(x==Root() || prev_sibling(Par,x) == NULLT));
+}
 // NumChildren(x): number of children of node x. Constant time with the data structure
 // of Sadakane.
 int XMLTree::NumChildren(treeNode x) 
@@ -498,6 +501,14 @@ treeNode XMLTree::FirstChild(treeNode x)
     return first_child(Par, x);
  }
 
+treeNode XMLTree::LastChild(treeNode x) 
+{
+  if (x == Root() || isleaf(Par,x) || x == NULLT)
+    return x;
+  else
+  return find_open(Par,find_close(Par,parent(Par,x))-1);
+}
+
 // NextSibling(x): returns the next sibling of node x, assuming it exists.
 treeNode XMLTree::NextSibling(treeNode x) 
  {
@@ -555,13 +566,14 @@ treeNode XMLTree::TaggedDesc(treeNode x, TagType tag)
        exit(1);
     }
 
-    int r, s;
+    //int r, s;
     treeNode y;
     if (isleaf(Par,x))
       return NULLT;
 
-    r = (int) Tags->rank(tag, node2tagpos(x));
-    s = (int) Tags->select(tag, r+1);
+		int s = (int) Tags->select_next(tag,node2tagpos(x));
+    /*r = (int) Tags->rank(tag, node2tagpos(x));
+    s = (int) Tags->select(tag, r+1);*/
     if (s == -1) return NULLT; // there is no such node
     y = tagpos2node(s); // transforms the tag position into a node position
     if (!is_ancestor(Par, x, y)) return NULLT; // the next node tagged tag (in preorder) is not within the subtree of x.
@@ -648,14 +660,15 @@ treeNode XMLTree::TaggedDescOrFollOnly(treeNode x,TagType *folltags, unsigned in
 {
 
   treeNode res,y,lim;
-  int r,s;
+  //int r,s;
   lim = find_close(Par,root);   
   res=NULLT;
   for (unsigned int i = 0; i < ftlen; i ++ )
     {
 
-      r = (int) Tags->rank(folltags[i], node2tagpos(x));
-      s = (int) Tags->select(folltags[i], r+1);
+			int s = (int) Tags->select_next(folltags[i],node2tagpos(x));
+      /*r = (int) Tags->rank(folltags[i], node2tagpos(x));
+      s = (int) Tags->select(folltags[i], r+1);*/
       if (s == -1) 
 	y = NULLT; // there is no such node
       else {
@@ -751,12 +764,13 @@ treeNode XMLTree::TaggedFoll(treeNode x, TagType tag)
        exit(1);
     }
 
-    int r, s;
+    //int r, s;
     if (x ==NULLT || x == Root())
        return NULLT;
-                   
-    r = (int) Tags->rank(tag, find_close(Par, x));
-    s = (int) Tags->select(tag, r+1);  // select returns -1 in case that there is no r+1-th tag.
+     
+		int s = (int) Tags->select_next(tag,find_close(Par,x));
+    /*r = (int) Tags->rank(tag, find_close(Par, x));
+    s = (int) Tags->select(tag, r+1);  // select returns -1 in case that there is no r+1-th tag. */
     if (s==-1) return NULLT;
     else return tagpos2node(s);
  } 
@@ -766,13 +780,14 @@ treeNode XMLTree::TaggedFoll(treeNode x, TagType tag)
 treeNode XMLTree::TaggedFollBelow(treeNode x, TagType tag, treeNode root)
  {
 
-    int r, s;
+    //int r, s;
     int lim = node2tagpos(find_close(Par,root));
     if (x ==NULLT || x == Root())
        return NULLT;
                    
-    r = (int) Tags->rank(tag,find_close(Par,x));
-    s = (int) Tags->select(tag, r+1);  // select returns -1 in case that there is no r+1-th tag.
+		int s = (int) Tags->select_next(tag,find_close(Par,x));
+    /*r = (int) Tags->rank(tag,find_close(Par,x));
+    s = (int) Tags->select(tag, r+1);  // select returns -1 in case that there is no r+1-th tag.*/
     if (s==-1 || s >= lim) 
       return NULLT;
     else 
@@ -789,14 +804,15 @@ treeNode XMLTree::TaggedFollowingSibling(treeNode x, TagType tag)
        exit(1);
     }
 
-    int r, s;
+    //int r, s;
     treeNode ns = next_sibling(Par,x);
 
     if (x == NULLT || x == Root() || ns == -1)
       return NULLT;
 
-    r = (int) Tags->rank(tag, node2tagpos(ns)-1);
-    s = (int) Tags->select(tag, r+1);  // select returns -1 in case that there is no r+1-th tag.
+		int s = (int) Tags->select_next(tag,node2tagpos(ns)-1);
+    /*r = (int) Tags->rank(tag, node2tagpos(ns)-1);
+    s = (int) Tags->select(tag, r+1);  // select returns -1 in case that there is no r+1-th tag.*/
     if (s==-1) return NULLT;
     else return tagpos2node(s);
  }
@@ -1013,9 +1029,11 @@ int XMLTree::OpenDocument(bool empty_texts, int sample_rate_text,bool dtc)
     if (!indexing_empty_texts) 
       empty_texts_aux = (unsigned int *)umalloc(sizeof(unsigned int));
        
-    
-    
-    Text = TextCollection::InitTextCollection((unsigned)sample_rate_text);
+    if (disable_tc)
+        TextBuilder = 0;
+    else 
+        TextBuilder = new TextCollectionBuilder((unsigned)sample_rate_text);
+    Text = 0;
     
     return 1;  // indicates success in the initialization of the data structure
  }
@@ -1041,9 +1059,9 @@ int XMLTree::CloseDocument()
 
     // If we found an attribute then "<@>" is present in the tree
     // if we didn't then it is not. "<$>" is never present in the tree
-		//uint max_tag = 0;
-		//for(uint i=0;i<(uint)npar-1;i++)
-		//	max_tag = max(max_tag,tags_aux[i]);
+		uint max_tag = 0;
+		for(uint i=0;i<(uint)npar-1;i++)
+			max_tag = max(max_tag,tags_aux[i]);
 		//max_tag++;
 		//tags_aux = (TagType *) urealloc(tags_aux, sizeof(TagType)*(npar + 1));
 		//tags_aux[npar++] = max_tag;
@@ -1073,10 +1091,16 @@ int XMLTree::CloseDocument()
     //delete ssb;
 
     
-		// makes the text collection static
+    // makes the text collection static
     if (!disable_tc)
-      Text->MakeStatic();
-    
+    {
+        assert(Text == 0);
+        assert(TextBuilder != 0);
+        Text = TextBuilder->InitTextCollection();
+        delete TextBuilder;
+        TextBuilder = 0;
+    }
+
     // creates the data structure marking the non-empty texts (just in the case it is necessary)
     if (!indexing_empty_texts)  {
        EBVector = new static_bitsequence_rrr02((uint *)empty_texts_aux,(ulong)npar,(uint)32);
@@ -1214,7 +1238,7 @@ int XMLTree::NewText(unsigned char *s)
               bitset(empty_texts_aux, npar-1);  // marks the non-empty text with a 1 in the bit vector
     }
     
-    Text->InsertText(s);
+    TextBuilder->InsertText(s);
     string cpps = (char*) s;
     CachedText.push_back(cpps); 
     
@@ -1239,7 +1263,7 @@ int XMLTree::NewEmptyText()
        
        bitclean(empty_texts_aux, npar-1);  // marks the empty text with a 0 in the bit vector
     }
-    else Text->InsertText(&c); // we insert the empty text just in case we index all the texts
+    else TextBuilder->InsertText(&c); // we insert the empty text just in case we index all the texts
     
     return 1; // success    
  }