X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLTree.cpp;h=ecc9a2786ee545c97deb9e4a76b465c2c0d466ce;hb=46dd7565a6c8d813459b81cf2b19446db1b51010;hp=21658e99c323c4aaaff8b72e385d8580542bddf4;hpb=22ee86b6a6088bf49747da202d619439fd8880e2;p=SXSI%2FXMLTree.git

diff --git a/XMLTree.cpp b/XMLTree.cpp
index 21658e9..ecc9a27 100644
--- a/XMLTree.cpp
+++ b/XMLTree.cpp
@@ -570,14 +570,16 @@ treeNode XMLTree::ParentNode(DocID d)
 // OpenDocument(empty_texts): it starts the construction of the data structure for
 // the XML document. Parameter empty_texts indicates whether we index empty texts
 // in document or not. Returns a non-zero value upon success, NULLT in case of error.
-int XMLTree::OpenDocument(bool empty_texts, int sample_rate_text)
+int XMLTree::OpenDocument(bool empty_texts, int sample_rate_text,bool dtc)
  {
     initialized = true;
     finished = false;
+    found_attributes = false;
     npar = 0;
     parArraySize = 1;
-    ntagnames = 0;    
- 
+    ntagnames = 2;    
+    disable_tc = dtc;
+    
     indexing_empty_texts = empty_texts;
     
     par_aux = (pb *)malloc(sizeof(pb)*parArraySize);
@@ -592,7 +594,28 @@ 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));
@@ -633,17 +656,29 @@ 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;
     delete ssb;
     // makes the text collection static
-    Text->MakeStatic();
+    if (!disable_tc)
+      Text->MakeStatic();
     
     // creates the data structure marking the non-empty texts (just in the case it is necessary)
-    if (!indexing_empty_texts) 
+    if (!indexing_empty_texts)  {
        EBVector = new static_bitsequence_rrr02((uint *)empty_texts_aux,(ulong)npar,(uint)32);
+       free (empty_texts_aux);
+       empty_texts_aux = NULL;
+    }
+   
+    free(tags_aux);
+    tags_aux = NULL;
 
     finished = true;
 
@@ -679,8 +714,13 @@ 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));
        
@@ -777,6 +817,11 @@ int XMLTree::NewText(unsigned char *s)
        return NULLT;
     }
 
+    if (disable_tc) {
+      XMLTree::NewEmptyText();
+      return 1;
+    };
+
     if (!indexing_empty_texts) {
        empty_texts_aux = (unsigned int *)realloc(empty_texts_aux, sizeof(pb)*(1+(npar-1)/(8*sizeof(pb))));
        if (!empty_texts_aux) {
@@ -846,4 +891,19 @@ 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;
+}