X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLTree.cpp;h=21658e99c323c4aaaff8b72e385d8580542bddf4;hb=52cb7bbcda67f4676335cdd4eb96d4d87ad1445d;hp=ea12fad9080f4a09c3f864d694e40a149f6eb3e2;hpb=e421727c5a512154cf2c3ff4cf8390eb88d527d7;p=SXSI%2FXMLTree.git

diff --git a/XMLTree.cpp b/XMLTree.cpp
index ea12fad..21658e9 100644
--- a/XMLTree.cpp
+++ b/XMLTree.cpp
@@ -1,6 +1,5 @@
-
 #include "XMLTree.h"
-
+#include <cstring>
 // functions to convert tag positions to the corresponding tree node and viceversa. 
 // These are implemented in order to be able to change the tree and Tags representations, 
 // without affecting the code so much.
@@ -102,7 +101,7 @@ XMLTree *XMLTree::Load(unsigned char *filename, int sample_rate_text)
     if (!(XML_Tree->indexing_empty_texts)) XML_Tree->EBVector = static_bitsequence_rrr02::load(fp);
 
     // loads the tags
-    XML_Tree->Tags = static_sequence_wvtree::load(fp);
+    XML_Tree->Tags = static_sequence::load(fp);
 
     // loads the texts   
     XML_Tree->Text->Load(fp,sample_rate_text);
@@ -127,16 +126,16 @@ XMLTree::~XMLTree()
     free(TagName);
 
     if (!indexing_empty_texts) {
-       EBVector->~static_bitsequence_rrr02();
+       //EBVector->~static_bitsequence_rrr02();
        delete EBVector;
        EBVector = NULL;
     }
 
-    Tags->~static_sequence_wvtree();
+    //Tags->~static_sequence_wvtree();
     delete Tags;
     Tags = NULL;
 
-    Text->~TextCollection();
+    //Text->~TextCollection();
     delete Text;
     Text = NULL;
 
@@ -312,8 +311,10 @@ treeNode XMLTree::Parent(treeNode x)
        fprintf(stderr, "Error: data structure has not been constructed properly\n");
        exit(1);
     }
-
-    return parent(Par, x);
+    if (x == Root())
+      return NULLT;
+    else
+      return parent(Par, x);
  }
 
 // Child(x,i): returns the i-th child of node x, assuming it exists.
@@ -346,7 +347,9 @@ treeNode XMLTree::NextSibling(treeNode x)
        fprintf(stderr, "Error: data structure has not been constructed properly\n");
        exit(1);
     }
-
+    if (x == Root())
+      return NULLT;
+ 
     return next_sibling(Par, x);
  }
 
@@ -572,19 +575,18 @@ int XMLTree::OpenDocument(bool empty_texts, int sample_rate_text)
     initialized = true;
     finished = false;
     npar = 0;
-    ntagnames = 0;
-    
+    parArraySize = 1;
+    ntagnames = 0;    
+ 
     indexing_empty_texts = empty_texts;
     
-    par_aux = (pb *)malloc(sizeof(pb));
+    par_aux = (pb *)malloc(sizeof(pb)*parArraySize);
     if (!par_aux) {
        fprintf(stderr, "Error: not enough memory\n");
        return NULLT;
     }
-    setbit(par_aux,npar,OP);  // marks a new opening parenthesis for the tree root
-    npar++;
     
-    tags_aux = (TagType *)malloc(sizeof(TagType));
+    tags_aux = (TagType *) malloc(sizeof(TagType));
     if (!tags_aux) {
        fprintf(stderr, "Error: not enough memory\n");
        return NULLT;
@@ -622,18 +624,20 @@ int XMLTree::CloseDocument()
        fprintf(stderr, "Error: not enough memory\n");
        return NULLT;    
     }
-    setbit(par_aux,npar,CP); 
-    npar++;
     
     // creates the data structure for the tree topology
-    Par = (bp *)malloc(sizeof(bp));      
+    Par = (bp *)malloc(sizeof(bp));
     bp_construct(Par, npar, par_aux, OPT_DEGREE|0);    
     // creates structure for tags
-    alphabet_mapper * am = new alphabet_mapper_none();
-    static_bitsequence_builder * bmb = new static_bitsequence_builder_rrr02(32); 
-    wt_coder * wtc = new wt_coder_huff((uint *)tags_aux,npar-1,am);
-    Tags = new static_sequence_wvtree((uint *) tags_aux, (uint) npar-1, wtc, bmb, am);
+    static_bitsequence_builder * bmb = new static_bitsequence_builder_brw32(20);
+    static_permutation_builder * pmb = new static_permutation_builder_mrrr(PERM_SAMPLE, bmb);
+    static_sequence_builder * ssb = new static_sequence_builder_gmr_chunk(bmb, pmb);
 
+    Tags = new static_sequence_gmr((uint *) tags_aux, (uint) npar-1,2*ntagnames, bmb, ssb);
+    
+    delete bmb;
+    delete pmb;
+    delete ssb;
     // makes the text collection static
     Text->MakeStatic();
     
@@ -660,11 +664,16 @@ int XMLTree::NewOpenTag(unsigned char *tagname)
     }
     
     // inserts a new opening parentheses in the bit sequence
-    par_aux = (pb *)realloc(par_aux, sizeof(pb)*(1+npar/(8*sizeof(pb))));
+    if (sizeof(pb)*8*parArraySize == npar) { // no space left for the new parenthesis
+       par_aux = (pb *)realloc(par_aux, sizeof(pb)*2*parArraySize);
+       parArraySize *= 2;
+    }
+    
     if (!par_aux) {
        fprintf(stderr, "Error: not enough memory\n");
        return NULLT;    
     }
+
     setbit(par_aux,npar,OP);  // marks a new opening parenthesis
 
     // transforms the tagname into a tag identifier. If the tag is new, we insert
@@ -684,9 +693,7 @@ int XMLTree::NewOpenTag(unsigned char *tagname)
        TagName[i] = (unsigned char *)malloc(sizeof(unsigned char)*(strlen((const char *)tagname)+1));
        strcpy((char *)TagName[i], (const char *)tagname);
     } 
-    
-    tags_aux = (TagType *)realloc(tags_aux, sizeof(TagType)*(npar + 1));
-
+    tags_aux = (TagType *) realloc(tags_aux, sizeof(TagType)*(npar + 1));
     if (!tags_aux) {
        fprintf(stderr, "Error: not enough memory\n");
        return NULLT;
@@ -714,7 +721,11 @@ int XMLTree::NewClosingTag(unsigned char *tagname)
     }
     
     // inserts a new closing parentheses in the bit sequence
-    par_aux = (pb *)realloc(par_aux, sizeof(pb)*(1+npar/(8*sizeof(pb))));
+    if (sizeof(pb)*8*parArraySize == npar) { // no space left for the new parenthesis
+       par_aux = (pb *)realloc(par_aux, sizeof(pb)*2*parArraySize);
+       parArraySize *= 2;
+    }
+    
     if (!par_aux) {
        fprintf(stderr, "Error: not enough memory\n");
        return NULLT;