More fixes

[SXSI/XMLTree.git] / XMLTree.cpp

diff --git a/XMLTree.cpp b/XMLTree.cpp
index e17c89d..057a117 100644 (file)
--- a/XMLTree.cpp
+++ b/XMLTree.cpp
@@ -67,6 +67,16 @@ inline void * umalloc(size_t size){
   return dest;\r
 }\r
 \r
+void XMLTree::print_stats() {\r
+       uint total_space = Tags->size()+sizeof(static_sequence*);\r
+       total_space += sizeof(uint*)+sizeof(uint)*(2+uint_len(tags_blen,tags_len));\r
+       cout << "Space usage for XMLTree:" << endl\r
+               << " - tags static_sequence: " << Tags->size()+sizeof(static_sequence*) << endl\r
+               << " - tags access array:    " << sizeof(uint*)+sizeof(uint)*(2+uint_len(tags_blen,tags_len)) << endl\r
+               << " ... add Diego structures ... " << endl\r
+               << " *total* " << total_space << endl;\r
+}\r
+\r
 // Save: saves XML tree data structure to file. \r
 void XMLTree::Save(unsigned char *filename) \r
  {\r
@@ -230,13 +240,14 @@ XMLTree *XMLTree::Load(unsigned char *filename, int sample_rate_text)
 \r
     fclose(fp);\r
 \r
-    std::cerr << "Tree part is " << s_tree/1024 << " Kbytes,\n"\r
+    /*std::cerr << "Tree part is " << s_tree/1024 << " Kbytes,\n"\r
              << "with node->tagid part " << XML_Tree->Tags->size()/1024+(uint_len(XML_Tree->tags_blen,XML_Tree->tags_len)*sizeof(uint))/1024  << "Kbytes \n"\r
              << "size of Tag part : " << XML_Tree->Tags->length () << " elements\n"\r
              << "sizof(unsigned int)* " <<  XML_Tree->Tags->length () << " = " << \r
       sizeof(unsigned int) * XML_Tree->Tags->length () / 1024 << " Kbytes\n"\r
              << "Tag part is " << s_tags/1024 << " Kbytes,\n"\r
-             << "Text collection is " << s_text/1024 << " Kbytes \n";\r
+             << "Text collection is " << s_text/1024 << " Kbytes \n";*/\r
+               XML_Tree->print_stats();\r
     return XML_Tree;\r
  }\r
 \r
@@ -851,18 +862,21 @@ int XMLTree::CloseDocument()
 \r
     // If we found an attribute then "<@>" is present in the tree\r
     // if we didn't then it is not. "<$>" is never present in the tree\r
+               uint max_tag = 0;\r
+               for(uint i=0;i<(uint)npar-1;i++)\r
+                       max_tag = max(max_tag,tags_aux[i]);\r
+               max_tag++;\r
     int ntagsize = found_attributes ? 2*ntagnames-1 : 2*ntagnames - 2;\r
+               tags_aux = (TagType *) urealloc(tags_aux, sizeof(TagType)*(npar + 1));\r
+               tags_aux[npar++] = max_tag;\r
 \r
                static_bitsequence_builder * bmb = new static_bitsequence_builder_brw32(20);\r
                alphabet_mapper *am = new alphabet_mapper_none();\r
-               wt_coder * wc = new wt_coder_huff((uint*)tags_aux,npar-1,am);\r
-               Tags = new static_sequence_wvtree((uint*)tags_aux,npar-1,wc ,bmb, am);\r
+               wt_coder * wc = new wt_coder_huff((uint*)tags_aux,npar,am);\r
+               Tags = new static_sequence_wvtree((uint*)tags_aux,npar,wc ,bmb, am);\r
     //Tags = new static_sequence_gmr((uint *) tags_aux, (uint) npar-1,ntagsize, bmb, ssb);\r
-               uint max_tag = 0;\r
-               for(uint i=0;i<(uint)npar-1;i++)\r
-                       max_tag = max(max_tag,tags_aux[i]);\r
                tags_blen = bits(max_tag);\r
-               tags_len = (uint)npar-1;\r
+               tags_len = (uint)npar;\r
                tags_fix = new uint[uint_len(tags_blen,tags_len)];\r
                for(uint i=0;i<(uint)npar-1;i++)\r
                        set_field(tags_fix,tags_blen,i,tags_aux[i]);\r
@@ -887,6 +901,7 @@ int XMLTree::CloseDocument()
     tags_aux = NULL;\r
 \r
     finished = true;\r
+               print_stats();\r
 \r
     return 1; // indicates success in the inicialization\r
  }\r
@@ -1049,7 +1064,7 @@ TagType XMLTree::GetTagId(unsigned char *tagname)
     // this should be changed for more efficient processing\r
     for (i=0; i<ntagnames; i++)\r
        if (strcmp((const char *)tagname,(const char *)TagName[i])==0) break; \r
-    if (i==ntagnames) return (TagType)NULLT; // tagname does not exists in the table\r
+    if (i==ntagnames) return ntagnames; //(TagType)NULLT; // tagname does not exists in the table\r
     else return i;\r
  }\r
 \r