From deeb5434b7e061741700eff08588539d34b717a6 Mon Sep 17 00:00:00 2001
From: nvalimak <nvalimak@3cdefd35-fc62-479d-8e8d-bae585ffb9ca>
Date: Mon, 23 Mar 2009 15:33:35 +0000
Subject: [PATCH] Jouni's Incremental BWT integrated into TextCollection

git-svn-id: svn+ssh://idea.nguyen.vg/svn/sxsi/trunk/XMLTree@272 3cdefd35-fc62-479d-8e8d-bae585ffb9ca
---
 XMLTree.cpp | 31 ++++++++++++++++++++-----------
 XMLTree.h   | 11 +++++++----
 makefile    |  2 +-
 3 files changed, 28 insertions(+), 16 deletions(-)

diff --git a/XMLTree.cpp b/XMLTree.cpp
index 8f8739d..70cf4b7 100644
--- a/XMLTree.cpp
+++ b/XMLTree.cpp
@@ -223,8 +223,7 @@ XMLTree *XMLTree::Load(unsigned char *filename, int sample_rate_text)
 
     // 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;
@@ -1013,9 +1014,11 @@ 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
  }
@@ -1073,10 +1076,16 @@ 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 +1223,7 @@ 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 +1248,7 @@ 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    
  }
diff --git a/XMLTree.h b/XMLTree.h
index 80173a4..9b79ae3 100644
--- a/XMLTree.h
+++ b/XMLTree.h
@@ -21,7 +21,7 @@
 
 #ifndef XMLTREE_H_
 #define XMLTREE_H_
-#include "TextCollection/TextCollection.h"
+#include "TextCollection/TextCollectionBuilder.h"
 #include <stdio.h>
 #include <stdlib.h>
 #include <cstring>
@@ -37,6 +37,7 @@
 #include <alphabet_mapper.h>
 #include <static_sequence.h>
 using SXSI::TextCollection;
+using SXSI::TextCollectionBuilder;
 
 
 // this constant is used to efficiently compute the child operation in the tree
@@ -88,7 +89,9 @@ class XMLTree {
 	 uint tags_blen, tags_len;
 
    /** The texts in the XML document */
+   TextCollectionBuilder *TextBuilder;
    TextCollection *Text;
+
    /** The texts in the XML document (cached for faster display) */
    vector<string> CachedText;
    
@@ -125,7 +128,7 @@ public:
 	 void print_stats();
 
    /** Data structure constructor */
-   XMLTree() {finished = false; initialized = false;}; 
+   XMLTree() {finished = false; initialized = false; Text = 0; TextBuilder = 0; }; 
  
    /** Data structure destructor */
    ~XMLTree();
@@ -331,7 +334,7 @@ public:
    TagType RegisterTag(unsigned char *tagname);
 
    bool EmptyText(DocID i) {
-      return Text->EmptyText(i);
+       return Text->EmptyText(i);
    }
    /** Prefix(s): search for texts prefixed by string s. */
    TextCollection::document_result Prefix(uchar const *s) {
@@ -420,7 +423,7 @@ public:
    /** GetText(d): returns the text corresponding to document with
     * id d. */
    uchar* GetText(DocID d) {
-      return Text->GetText(d);
+       return Text->GetText(d);
    }
 
    uchar* GetCachedText(DocID d) {
diff --git a/makefile b/makefile
index 4f06f3b..e1705ef 100644
--- a/makefile
+++ b/makefile
@@ -1,7 +1,7 @@
 FLAGS =-std=c++0x -O3 -I./libcds/includes/
 
 OBJECTS=libcds/lib/libcds.a
-OBJECTS_TC= TextCollection/TextCollection.o TextCollection/CSA.o TextCollection/Tools.o TextCollection/BitRank.o TextCollection/bittree.o TextCollection/rbtree.o TextCollection/dynFMI.o TextCollection/RLWaveletTree.o TextCollection/GapEncode.o TextCollection/BSGAP.o
+OBJECTS_TC= TextCollection/TextCollection.o TextCollection/TextCollectionBuilder.o TextCollection/TCImplementation.o TextCollection/Tools.o TextCollection/BitRank.o TextCollection/BSGAP.o TextCollection/incbwt/rlcsa.a
 
 all: libcds text_collection XMLTree
 
-- 
2.17.1