From add1a8a0b8a7ca05bb099ec38a60f3c52b9dfd71 Mon Sep 17 00:00:00 2001
From: kim <kim@3cdefd35-fc62-479d-8e8d-bae585ffb9ca>
Date: Wed, 28 Jan 2009 00:06:34 +0000
Subject: [PATCH] Hardcode "<@>" and "<$>" at position 0 and 1 in the TagName
 table. These are used to denote attributes and PCDATA nodes.

git-svn-id: svn+ssh://idea.nguyen.vg/svn/sxsi/trunk/XMLTree@81 3cdefd35-fc62-479d-8e8d-bae585ffb9ca
---
 XMLTree.cpp | 40 ++++++++++++++++++++++++++++++++++++----
 XMLTree.h   |  1 +
 2 files changed, 37 insertions(+), 4 deletions(-)

diff --git a/XMLTree.cpp b/XMLTree.cpp
index 22ba4bb..fe01896 100644
--- a/XMLTree.cpp
+++ b/XMLTree.cpp
@@ -574,9 +574,10 @@ int XMLTree::OpenDocument(bool empty_texts, int sample_rate_text)
  {
     initialized = true;
     finished = false;
+    found_attributes = false;
     npar = 0;
     parArraySize = 1;
-    ntagnames = 0;    
+    ntagnames = 2;    
  
     indexing_empty_texts = empty_texts;
     
@@ -592,7 +593,28 @@ 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));
@@ -633,7 +655,12 @@ 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;
@@ -679,8 +706,13 @@ 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));
        
diff --git a/XMLTree.h b/XMLTree.h
index 88aae9a..cc5f089 100644
--- a/XMLTree.h
+++ b/XMLTree.h
@@ -90,6 +90,7 @@ class XMLTree {
    int parArraySize;
    int ntagnames;
    unsigned int *empty_texts_aux;
+   bool found_attributes;
    
 public:
 
-- 
2.17.1