X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLTree.cpp;h=fe01896d4ca75c93ffdf83121b9d0ba1a42efca0;hb=add1a8a0b8a7ca05bb099ec38a60f3c52b9dfd71;hp=92360a432535065f46769245eadee9f98515c005;hpb=a9846746dc7a55764591fcc273fd48c6049df962;p=SXSI%2FXMLTree.git

diff --git a/XMLTree.cpp b/XMLTree.cpp
index 92360a4..fe01896 100644
--- a/XMLTree.cpp
+++ b/XMLTree.cpp
@@ -311,8 +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.
@@ -345,7 +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);
  }
 
@@ -570,9 +574,10 @@ int XMLTree::OpenDocument(bool empty_texts, int sample_rate_text)
  {
     initialized = true;
     finished = false;
+    found_attributes = false;
     npar = 0;
     parArraySize = 1;
-    ntagnames = 0;    
+    ntagnames = 2;    
  
     indexing_empty_texts = empty_texts;
     
@@ -588,7 +593,28 @@ int XMLTree::OpenDocument(bool empty_texts, int sample_rate_text)
        return NULLT;
     }
     
-    TagName = NULL;
+    TagName = (unsigned char **) malloc(2*sizeof(unsigned char*));
+    if (!TagName){
+       fprintf(stderr, "Error: not enough memory\n");
+       return NULLT;
+    }
+
+    TagName[0] = (unsigned char *) malloc(4*sizeof(unsigned char));
+    strcpy((char *) TagName[0], "<@>");
+
+    if (!TagName[0]){
+      fprintf(stderr, "Error: not enough memory\n");
+      return NULLT;
+    }
+
+    TagName[1] = (unsigned char *) malloc(4*sizeof(unsigned char));
+    if (!TagName[1]){
+      fprintf(stderr, "Error: not enough memory\n");
+      return NULLT;
+    }
+
+    strcpy((char *) TagName[1], "<$>");
+
 
     if (!indexing_empty_texts) {
        empty_texts_aux = (unsigned int *)malloc(sizeof(unsigned int));
@@ -629,7 +655,12 @@ int XMLTree::CloseDocument()
     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);
+
+    // 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
+    int ntagsize = found_attributes ? 2*ntagnames-1 : 2*ntagnames - 2;
+
+    Tags = new static_sequence_gmr((uint *) tags_aux, (uint) npar-1,ntagsize, bmb, ssb);
     
     delete bmb;
     delete pmb;
@@ -675,8 +706,13 @@ int XMLTree::NewOpenTag(unsigned char *tagname)
     // transforms the tagname into a tag identifier. If the tag is new, we insert
     // it in the table.
     for (i=0; i<ntagnames; i++)
-       if (strcmp((const char *)tagname,(const char *)TagName[i])==0) break;
+      if (strcmp((const char *)tagname,(const char *)TagName[i])==0) break;
  
+
+    // NewOpenTag("<@>") was called
+    if (i==0) 
+      found_attributes=true;
+
     if (i==ntagnames) { // the tag is a new one, then we insert it
        TagName = (unsigned char **)realloc(TagName, sizeof(char *)*(ntagnames+1));
        
@@ -842,4 +878,19 @@ unsigned char *XMLTree::GetTagName(TagType tagid)
  }
 
 
-
+TagType XMLTree::RegisterTag(unsigned char *tagname)
+{
+  if (!finished)
+    return NULLT;
+  
+
+  TagType id = XMLTree::GetTagId(tagname);
+  if (id == NULLT){
+    id = ntagnames;
+    ntagnames = ntagnames + 1;    
+    TagName = (unsigned char **) realloc(TagName,ntagnames*(sizeof(unsigned char*)));
+    strcpy((char*)TagName[id], (const char *)tagname);  
+  };
+
+  return id;
+}