Added rankLessThan

[SXSI/XMLTree.git] / libcds / src / static_sequence / wt_node_internal.cpp

diff --git a/libcds/src/static_sequence/wt_node_internal.cpp b/libcds/src/static_sequence/wt_node_internal.cpp
index 9dc87a2..50bb35f 100644 (file)
--- a/libcds/src/static_sequence/wt_node_internal.cpp
+++ b/libcds/src/static_sequence/wt_node_internal.cpp
@@ -70,6 +70,7 @@ wt_node_internal::wt_node_internal(uint * symbols, uint n, uint l, wt_coder * c,
        delete [] right;
 }
 
+// Deletes symbols array!
 wt_node_internal::wt_node_internal(uchar * symbols, uint n, uint l, wt_coder * c, static_bitsequence_builder * bmb) {
        uint * ibitmap = new uint[n/W+1];
        for(uint i=0;i<n/W+1;i++)
@@ -99,24 +100,42 @@ wt_node_internal::wt_node_internal(uchar * symbols, uint n, uint l, wt_coder * c
                                        match_left = false;
                }
        }
+
+        delete [] symbols; 
+        symbols = 0;
+
        if(count_left>0) {
                if(match_left/* && c->done(left[0],l+1)*/)
-                       left_child = new wt_node_leaf((uint)left[0], count_left);
+                {
+                    left_child = new wt_node_leaf((uint)left[0], count_left);
+                    delete [] left;
+                    left = 0;
+                }
                else
-                       left_child = new wt_node_internal(left, count_left, l+1, c, bmb);
+                {
+                    left_child = new wt_node_internal(left, count_left, l+1, c, bmb);
+                    left = 0; // Already deleted
+                }
        } else {
                left_child = NULL;
        }
        if(count_right>0) {
                if(match_right/* && c->done(right[0],l+1)*/)
-                       right_child = new wt_node_leaf((uint)right[0], count_right);
-               else
-                       right_child = new wt_node_internal(right, count_right, l+1, c, bmb);
+                {
+                    right_child = new wt_node_leaf((uint)right[0], count_right);
+                    delete [] right;
+                    right = 0;
+                }
+               else 
+                {
+                    right_child = new wt_node_internal(right, count_right, l+1, c, bmb);
+                    right = 0; // Already deleted
+                }
        } else {
                right_child = NULL;
        }
-       delete [] left;
-       delete [] right;
+//     delete [] left; // already deleted
+//     delete [] right;
 }
 
 
@@ -140,6 +159,52 @@ uint wt_node_internal::rank(uint symbol, uint pos, uint l, wt_coder * c) {
        }
 }
 
+// return value is rank of symbol (less or equal to the given symbol) that has rank > 0, 
+// the parameter symbol is updated accordinly
+uint wt_node_internal::rankLessThan(uint &symbol, uint pos, uint l, wt_coder * c) 
+{
+    bool is_set = c->is_set(symbol,l);
+    using std::cout;
+    using std::endl;
+//    cout << "l = " << l << ", symbol = " << (uchar)symbol << ", rank0 = " << bitmap->rank0(pos) << ", rank1 = " << bitmap->rank1(pos) << endl;
+
+    uint result = -1;
+    if(!is_set) {
+        if(left_child==NULL) return -1;
+        uint rank = bitmap->rank0(pos);
+        if(rank != 0)
+            result = left_child->rankLessThan(symbol,rank-1,l+1,c);
+        return result;
+    }
+
+    uint rank = bitmap->rank1(pos);
+    if (rank != 0 && right_child != NULL)
+        result = right_child->rankLessThan(symbol, rank-1,l+1,c);
+
+//    cout << "recursion to leftchild at l = " << l << ", symbol = " << (uchar)symbol << ", rank0 = " << bitmap->rank0(pos) << ", rank1 = " << bitmap->rank1(pos) << endl;
+    // check left child for symbols <= givenSymbol
+    if (result != -1 || left_child == NULL)
+        return result;
+    return left_child->rankLessThan(symbol, bitmap->rank0(pos)-1);
+}
+
+uint wt_node_internal::rankLessThan(uint &symbol, uint pos) 
+{
+    uint result = -1;
+    using std::cout;
+    using std::endl;
+//    cout << "pos = " << pos << ", symbol = " << (uchar)symbol << endl;
+    
+    if (pos == -1)
+        return -1;
+    if(right_child!=NULL)
+        result = right_child->rankLessThan(symbol, bitmap->rank1(pos)-1);
+    if(result == -1 && left_child!=NULL)
+        return left_child->rankLessThan(symbol, bitmap->rank0(pos)-1);
+    return result;
+}
+
+
 uint wt_node_internal::select(uint symbol, uint pos, uint l, wt_coder * c) {
        bool is_set = c->is_set(symbol, l);
        uint ret = 0;
@@ -171,6 +236,45 @@ uint wt_node_internal::access(uint pos) {
        }
 }
 
+// Returns the value at given position and its rank
+uint wt_node_internal::access(uint pos, uint &rank) 
+{
+    // p is the internal node we are pointing our finger at each step
+    wt_node_internal *p = this;
+
+    while(1)
+    {
+        bool is_set = p->bitmap->access(pos);
+//        cout << "is_set = " << is_set << ", pos = " << pos << ", rank0 = " << bitmap->rank0(pos) << ", rank1 = " << bitmap->rank1(pos) << endl;
+        if(!is_set)
+        {
+            // recurse left
+            pos = p->bitmap->rank0(pos)-1;
+            wt_node_internal *tmp = dynamic_cast<wt_node_internal *>(p->left_child);
+            if (tmp == NULL)
+            {
+                // it's a leaf
+                rank = pos+1;
+                return p->left_child->access(0);
+            }
+            p = tmp; // new internal node
+        } 
+        else 
+        {
+            // recurse right
+            pos = p->bitmap->rank1(pos)-1;
+            wt_node_internal *tmp = dynamic_cast<wt_node_internal *>(p->right_child);
+            if (tmp == NULL)
+            {
+                // it's a leaf
+                rank = pos+1;
+                return p->right_child->access(0);
+            }
+            p = tmp; // new internal node
+        }
+    }
+}
+
 uint wt_node_internal::size() {
        uint s = bitmap->size()+sizeof(wt_node_internal);
        if(left_child!=NULL)