X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLTree.cpp;h=057a1178245fcd98fb18cc7179e20c1f4fc328bd;hb=4e6e8f179979f24c8f77e11eb8f6cc3d609a52d4;hp=e17c89d90358731f2cbad205b9c5a4534f96af76;hpb=d6ecf291b7dd99ec191daa090a2ff3dac08f8150;p=SXSI%2FXMLTree.git

diff --git a/XMLTree.cpp b/XMLTree.cpp
index e17c89d..057a117 100644
--- a/XMLTree.cpp
+++ b/XMLTree.cpp
@@ -67,6 +67,16 @@ inline void * umalloc(size_t size){
   return dest;
 }
 
+void XMLTree::print_stats() {
+	uint total_space = Tags->size()+sizeof(static_sequence*);
+	total_space += sizeof(uint*)+sizeof(uint)*(2+uint_len(tags_blen,tags_len));
+	cout << "Space usage for XMLTree:" << endl
+		<< " - tags static_sequence: " << Tags->size()+sizeof(static_sequence*) << endl
+		<< " - tags access array:    " << sizeof(uint*)+sizeof(uint)*(2+uint_len(tags_blen,tags_len)) << endl
+		<< " ... add Diego structures ... " << endl
+		<< " *total* " << total_space << endl;
+}
+
 // Save: saves XML tree data structure to file. 
 void XMLTree::Save(unsigned char *filename) 
  {
@@ -230,13 +240,14 @@ XMLTree *XMLTree::Load(unsigned char *filename, int sample_rate_text)
 
     fclose(fp);
 
-    std::cerr << "Tree part is " << s_tree/1024 << " Kbytes,\n"
+    /*std::cerr << "Tree part is " << s_tree/1024 << " Kbytes,\n"
 	      << "with node->tagid part " << XML_Tree->Tags->size()/1024+(uint_len(XML_Tree->tags_blen,XML_Tree->tags_len)*sizeof(uint))/1024  << "Kbytes \n"
 	      << "size of Tag part : " << XML_Tree->Tags->length () << " elements\n"
 	      << "sizof(unsigned int)* " <<  XML_Tree->Tags->length () << " = " << 
       sizeof(unsigned int) * XML_Tree->Tags->length () / 1024 << " Kbytes\n"
 	      << "Tag part is " << s_tags/1024 << " Kbytes,\n"
-	      << "Text collection is " << s_text/1024 << " Kbytes \n";
+	      << "Text collection is " << s_text/1024 << " Kbytes \n";*/
+		XML_Tree->print_stats();
     return XML_Tree;
  }
 
@@ -851,18 +862,21 @@ 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]);
+		max_tag++;
     int ntagsize = found_attributes ? 2*ntagnames-1 : 2*ntagnames - 2;
+		tags_aux = (TagType *) urealloc(tags_aux, sizeof(TagType)*(npar + 1));
+		tags_aux[npar++] = max_tag;
 
 		static_bitsequence_builder * bmb = new static_bitsequence_builder_brw32(20);
 		alphabet_mapper *am = new alphabet_mapper_none();
-		wt_coder * wc = new wt_coder_huff((uint*)tags_aux,npar-1,am);
-		Tags = new static_sequence_wvtree((uint*)tags_aux,npar-1,wc ,bmb, am);
+		wt_coder * wc = new wt_coder_huff((uint*)tags_aux,npar,am);
+		Tags = new static_sequence_wvtree((uint*)tags_aux,npar,wc ,bmb, am);
     //Tags = new static_sequence_gmr((uint *) tags_aux, (uint) npar-1,ntagsize, bmb, ssb);
-		uint max_tag = 0;
-		for(uint i=0;i<(uint)npar-1;i++)
-			max_tag = max(max_tag,tags_aux[i]);
 		tags_blen = bits(max_tag);
-		tags_len = (uint)npar-1;
+		tags_len = (uint)npar;
 		tags_fix = new uint[uint_len(tags_blen,tags_len)];
 		for(uint i=0;i<(uint)npar-1;i++)
 			set_field(tags_fix,tags_blen,i,tags_aux[i]);
@@ -887,6 +901,7 @@ int XMLTree::CloseDocument()
     tags_aux = NULL;
 
     finished = true;
+		print_stats();
 
     return 1; // indicates success in the inicialization
  }
@@ -1049,7 +1064,7 @@ TagType XMLTree::GetTagId(unsigned char *tagname)
     // this should be changed for more efficient processing
     for (i=0; i<ntagnames; i++)
        if (strcmp((const char *)tagname,(const char *)TagName[i])==0) break; 
-    if (i==ntagnames) return (TagType)NULLT; // tagname does not exists in the table
+    if (i==ntagnames) return ntagnames; //(TagType)NULLT; // tagname does not exists in the table
     else return i;
  }