X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLTree.cpp;h=ae80fc2a489680d270c6fba2d59939a8c9bddc00;hb=b53633fb64f387edb5cebefbb3308b6347b2389c;hp=8dcdaec6904f8e017e064e912fa129ced60a8b57;hpb=c1865639eb1a804ff9363035a2a656b64fd98b9e;p=SXSI%2FXMLTree.git

diff --git a/XMLTree.cpp b/XMLTree.cpp
index 8dcdaec..ae80fc2 100644
--- a/XMLTree.cpp
+++ b/XMLTree.cpp
@@ -211,6 +211,14 @@ XMLTree *XMLTree::Load(unsigned char *filename, int sample_rate_text)
 		s_tree+=2*sizeof(uint)+sizeof(uint)*uint_len(XML_Tree->tags_blen,XML_Tree->tags_len);
     s_tree+= XML_Tree->Tags->size();
 
+		/// FIXME:UGLY tests!
+		/*uint * seq = new uint[XML_Tree->tags_len];
+		for(uint i=0;i<XML_Tree->tags_len;i++)
+			seq[i] = get_field(XML_Tree->tags_fix,XML_Tree->tags_blen,i);
+		cout << "Tags test: " << XML_Tree->Tags->test(seq,XML_Tree->tags_len) << endl;
+		delete [] seq;*/
+		/// End ugly tests
+
     s_text = ftell(fp);
 
     // loads the texts
@@ -1030,36 +1038,42 @@ int XMLTree::CloseDocument()
     Par = (bp *)umalloc(sizeof(bp));
     bp_construct(Par, npar, par_aux, OPT_DEGREE|0);    
     // creates structure for tags
-    //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);
-
 
     // 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);
+		//uint max_tag = 0;
+		//for(uint i=0;i<(uint)npar-1;i++)
+		//	max_tag = max(max_tag,tags_aux[i]);
+		//max_tag++;
+		//tags_aux = (TagType *) urealloc(tags_aux, sizeof(TagType)*(npar + 1));
+		//tags_aux[npar++] = max_tag;
+    //int ntagsize = found_attributes ? 2*ntagnames-1 : 2*ntagnames - 2;
+    int ntagsize = 2*ntagnames + 2;
+
+    //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);
+		static_bitsequence_builder * bmb = new static_bitsequence_builder_sdarray();
 		alphabet_mapper *am = new alphabet_mapper_none();
-		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);
+		//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,ntagsize, bmb, ssb);
+		Tags = new static_sequence_bs((uint*)tags_aux,npar,am,bmb);
+		
+		cout << "Tags test: " << Tags->test((uint*)tags_aux,npar) << endl;
+
 		tags_blen = bits(max_tag);
 		tags_len = (uint)npar;
 		tags_fix = new uint[uint_len(tags_blen,tags_len)];
-		for(uint i=0;i<(uint)npar-1;i++)
+		for(uint i=0;i<(uint)npar;i++)
 			set_field(tags_fix,tags_blen,i,tags_aux[i]);
     
     delete bmb;
     //delete pmb;
     //delete ssb;
-    // makes the text collection static
+
+    
+		// makes the text collection static
     if (!disable_tc)
       Text->MakeStatic();
     
@@ -1239,7 +1253,7 @@ 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 ntagnames; //(TagType)NULLT; // tagname does not exists in the table
+    if (i==ntagnames) return (TagType)-1; //ntagnames; //(TagType)NULLT; // tagname does not exists in the table
     else return i;
  }
 
@@ -1249,7 +1263,7 @@ TagType XMLTree::GetTagId(unsigned char *tagname)
 unsigned char *XMLTree::GetTagName(TagType tagid)
  {
     unsigned char *s;
-
+		if(tagid==(uint)-1) return NULL;
     if (tagid >= ntagnames) return NULL; // invalid tag identifier
     s = (unsigned char *)umalloc((strlen((const char *)TagName[tagid])+1)*sizeof(unsigned char));
     strcpy((char *)s, (const char *)TagName[tagid]);
@@ -1261,6 +1275,7 @@ unsigned char *XMLTree::GetTagName(TagType tagid)
 
 const unsigned char *XMLTree::GetTagNameByRef(TagType tagid)
  {
+		if(tagid==(uint)-1) return NULL;
     if (tagid >= ntagnames) return NULL; // invalid tag identifier
     return ((const unsigned char*)  TagName[tagid]);
  }