X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLTree.cpp;h=249143cfbe456ffc8d78d10262f82167baa773dd;hb=69d191f9549b65e692b38bfceecc86fbcde474ad;hp=2a1ca71b035d4b37c3492f8b583e8dd204f137f6;hpb=4b998a9c4982a8971b2997adbad9fd654de4b634;p=SXSI%2FXMLTree.git

diff --git a/XMLTree.cpp b/XMLTree.cpp
index 2a1ca71..249143c 100644
--- a/XMLTree.cpp
+++ b/XMLTree.cpp
@@ -212,19 +212,18 @@ XMLTree *XMLTree::Load(unsigned char *filename, int sample_rate_text)
     s_tree+= XML_Tree->Tags->size();
 
 		/// FIXME:UGLY tests!
-		/*uint * seq = new uint[XML_Tree->tags_len];
+		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;*/
+		delete [] seq;
 		/// End ugly tests
 
     s_text = ftell(fp);
 
     // loads the texts
     if (!XML_Tree->disable_tc){
-      XML_Tree->Text = TextCollection::InitTextCollection(sample_rate_text);
-      XML_Tree->Text->Load(fp,sample_rate_text);
+      XML_Tree->Text = TextCollection::Load(fp,sample_rate_text);
       int sst;
       int st;
       ufread(&sst, sizeof(int),1,fp);
@@ -284,7 +283,9 @@ XMLTree::~XMLTree()
     Tags = NULL;
 
     //Text->~TextCollection();
-    delete Text;
+    delete TextBuilder; 
+    TextBuilder = NULL;
+    delete Text; 
     Text = NULL;
 
     initialized = false;
@@ -356,7 +357,9 @@ bool XMLTree::IsChild(treeNode x, treeNode y)
     if (!is_ancestor(Par, x, y)) return false;
     return depth(Par, x) == (depth(Par, y) + 1);
  }
-
+bool XMLTree::IsFirstChild(treeNode x){
+  return ((x != NULLT)&&(x==Root() || prev_sibling(Par,x) == NULLT));
+}
 // NumChildren(x): number of children of node x. Constant time with the data structure
 // of Sadakane.
 int XMLTree::NumChildren(treeNode x) 
@@ -498,6 +501,14 @@ treeNode XMLTree::FirstChild(treeNode x)
     return first_child(Par, x);
  }
 
+treeNode XMLTree::LastChild(treeNode x) 
+{
+  if (x == Root() || isleaf(Par,x) || x == NULLT)
+    return x;
+  else
+  return find_open(Par,find_close(Par,parent(Par,x))-1);
+}
+
 // NextSibling(x): returns the next sibling of node x, assuming it exists.
 treeNode XMLTree::NextSibling(treeNode x) 
  {
@@ -1013,9 +1024,11 @@ int XMLTree::OpenDocument(bool empty_texts, int sample_rate_text,bool dtc)
     if (!indexing_empty_texts) 
       empty_texts_aux = (unsigned int *)umalloc(sizeof(unsigned int));
        
-    
-    
-    Text = TextCollection::InitTextCollection((unsigned)sample_rate_text);
+    if (disable_tc)
+        TextBuilder = 0;
+    else 
+        TextBuilder = new TextCollectionBuilder((unsigned)sample_rate_text);
+    Text = 0;
     
     return 1;  // indicates success in the initialization of the data structure
  }
@@ -1044,9 +1057,9 @@ int XMLTree::CloseDocument()
 		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;
+		//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;
 
@@ -1073,10 +1086,16 @@ int XMLTree::CloseDocument()
     //delete ssb;
 
     
-		// makes the text collection static
+    // makes the text collection static
     if (!disable_tc)
-      Text->MakeStatic();
-    
+    {
+        assert(Text == 0);
+        assert(TextBuilder != 0);
+        Text = TextBuilder->InitTextCollection();
+        delete TextBuilder;
+        TextBuilder = 0;
+    }
+
     // creates the data structure marking the non-empty texts (just in the case it is necessary)
     if (!indexing_empty_texts)  {
        EBVector = new static_bitsequence_rrr02((uint *)empty_texts_aux,(ulong)npar,(uint)32);
@@ -1214,7 +1233,7 @@ int XMLTree::NewText(unsigned char *s)
               bitset(empty_texts_aux, npar-1);  // marks the non-empty text with a 1 in the bit vector
     }
     
-    Text->InsertText(s);
+    TextBuilder->InsertText(s);
     string cpps = (char*) s;
     CachedText.push_back(cpps); 
     
@@ -1239,7 +1258,7 @@ int XMLTree::NewEmptyText()
        
        bitclean(empty_texts_aux, npar-1);  // marks the empty text with a 0 in the bit vector
     }
-    else Text->InsertText(&c); // we insert the empty text just in case we index all the texts
+    else TextBuilder->InsertText(&c); // we insert the empty text just in case we index all the texts
     
     return 1; // success    
  }
@@ -1253,7 +1272,7 @@ 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;
  }
 
@@ -1263,7 +1282,7 @@ 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]);
@@ -1275,6 +1294,7 @@ 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]);
  }