From 1413ae2197d87e87571c9d8d6fc9f20f691fcea3 Mon Sep 17 00:00:00 2001
From: kim <kim@3cdefd35-fc62-479d-8e8d-bae585ffb9ca>
Date: Mon, 11 May 2009 05:37:25 +0000
Subject: [PATCH] Removed the naive text interface

git-svn-id: svn+ssh://idea.nguyen.vg/svn/sxsi/trunk/XMLTree@383 3cdefd35-fc62-479d-8e8d-bae585ffb9ca
---
 XMLTree.cpp        | 36 ++++++------------------------------
 XMLTree.h          | 19 +++++--------------
 XMLTreeBuilder.cpp |  3 ---
 darray.c           |  2 +-
 4 files changed, 12 insertions(+), 48 deletions(-)

diff --git a/XMLTree.cpp b/XMLTree.cpp
index 61f6a4d..f03360e 100644
--- a/XMLTree.cpp
+++ b/XMLTree.cpp
@@ -26,7 +26,7 @@ inline int node2tagpos(treeNode x)
 
 
 XMLTree::XMLTree( pb * const par, uint npar,  vector<string> * const TN,  TagIdMap * const tim, uint *empty_texts_bmp, TagType *tags,
-            TextCollection * const TC, vector<string> * const CT, bool dis_tc)
+            TextCollection * const TC, bool dis_tc)
 
  {
     // creates the data structure for the tree topology
@@ -58,7 +58,7 @@ XMLTree::XMLTree( pb * const par, uint npar,  vector<string> * const TN,  TagIdM
     
     Text = (TextCollection*) TC;
 
-    CachedText = (vector<string>*) CT;
+
     EBVector = new static_bitsequence_rrr02(empty_texts_bmp,npar,32);
     free(empty_texts_bmp);
     empty_texts_bmp = NULL;
@@ -139,18 +139,6 @@ void XMLTree::Save(int fd)
     
     // stores the texts   
     if (!disable_tc) {
-      int st = CachedText->size();
-
-      ufwrite(&st, sizeof(int),1,fp);
-      for (int i = 0; i< CachedText->size(); i++){
-	st = CachedText->at(i).size();
-
-	ufwrite(&st, sizeof(int),1,fp);
-
-	ufwrite(CachedText->at(i).c_str(),sizeof(char),1+CachedText->at(i).size(),fp);
-
-      };
-      
       Text->Save(fp);
     };
 
@@ -228,20 +216,6 @@ XMLTree *XMLTree::Load(int fd)
     int sample_rate_text = 64;
     // loads the texts
     if (!XML_Tree->disable_tc){
-      XML_Tree->CachedText = new vector<string>;
-      int sst;
-      int st;
-      ufread(&sst, sizeof(int),1,fp);
-      
-      for (int i=0;i<sst;i++){
-	ufread(&st, sizeof(int),1,fp);
-	char* str = (char*) malloc(sizeof(char)*st+1);
-	ufread(str,sizeof(char),st+1,fp);
-	string cppstr = str;
-	XML_Tree->CachedText->push_back(cppstr);
-	free(str);
-
-      };
       XML_Tree->Text = TextCollection::Load(fp,sample_rate_text);
     }
     else XML_Tree->Text = NULL;
@@ -640,9 +614,10 @@ DocID XMLTree::MyText(treeNode x)
 // all tree nodes and all text nodes. Assumes that the tree root has preorder 1.
 int XMLTree::TextXMLId(DocID d) 
  {
-   int s = EBVector->select1(d+1);
+   NULLT_IF(d == NULLT);
+     int s = EBVector->select1(d+1);
    return rank_open(Par, s) + d + 1; // +1 because root has preorder 1
- 
+   
  }
 
 // NodeXMLId(x): returns the preorder of node x in the tree consisting 
@@ -650,6 +625,7 @@ int XMLTree::TextXMLId(DocID d)
 // preorder 0;
 int XMLTree::NodeXMLId(treeNode x) 
  {
+   NULLT_IF(x == NULLT);
    if (x == Root()) return 1; // root node has preorder 1
    return rank_open(Par, x) + EBVector->rank1(x-1);
  }
diff --git a/XMLTree.h b/XMLTree.h
index 6a722f9..275b9bc 100644
--- a/XMLTree.h
+++ b/XMLTree.h
@@ -118,9 +118,6 @@ class XMLTree {
    /** The texts in the XML document */
    TextCollection *Text;
 
-   /** The texts in the XML document (cached for faster display) */
-   vector<string> *CachedText;
-
    // Allows to disable the TextCollection for benchmarkin purposes
    bool disable_tc;
    
@@ -130,7 +127,7 @@ class XMLTree {
 
    // non const pointer are freed by this method.
   XMLTree( pb * const par, uint npar,  vector<string> * const TN,  TagIdMap * const tim, uint *empty_texts_bmp, TagType *tags,
-	   TextCollection * const TC, vector<string> * const CT, bool dis_tc);
+	   TextCollection * const TC, bool dis_tc);
 
 public: 
    /** Data structure destructor */
@@ -396,17 +393,11 @@ public:
 
    /** GetText(i, j): returns the texts corresponding to documents with
     * ids i, i+1, ..., j. Texts are separated by '\0' character.  */
-   uchar* GetText(DocID i, DocID j) {
-     uchar * s = Text->GetText(i, j);
-     return (s[0] == 1 ? (uchar*)"" : s);
-   }
+   //   uchar* GetText(DocID i, DocID j) {
+   //  uchar * s = Text->GetText(i, j);
+   // return (s[0] == 1 ? (uchar*)"" : s);
+   //}
 
-   uchar* GetCachedText(DocID d) {
-     uchar * str = (uchar*) calloc(sizeof(char),(CachedText->at(d).size() + 1));
-     strcpy((char*) str,(const char*) CachedText->at(d).c_str());
-     return (uchar*) (str);
-   }
-   
    TextCollection *getTextCollection() {
       return Text;
    }
diff --git a/XMLTreeBuilder.cpp b/XMLTreeBuilder.cpp
index 138db9b..9c0fc6a 100644
--- a/XMLTreeBuilder.cpp
+++ b/XMLTreeBuilder.cpp
@@ -33,7 +33,6 @@ int XMLTreeBuilder::OpenDocument(bool empty_texts, int sample_rate_text, bool dt
     REGISTER_TAG(TagName,tIdMap,ATTRIBUTE_DATA_CLOSE_TAG);
 
 
-    CachedText = new vector<string>;
     if (disable_tc)
         TextBuilder = 0;
     else 
@@ -71,7 +70,6 @@ XMLTree *XMLTreeBuilder::CloseDocument()
 			     empty_texts_aux,                // freed by the constructor
 			     tags_aux,                       //freed by the constructor
 			     Text,
-			     CachedText,
 			     disable_tc);
     return T; 
  }
@@ -163,7 +161,6 @@ int XMLTreeBuilder::NewText(string text)
        TextBuilder->InsertText((uchar *) text.c_str());
    };
 
-   CachedText->push_back(text); 
    int n_eta_size = sizeof(uint)*(1+(npar-1)/(8*sizeof(uint)));
    //see basics.h, recalloc resizes and sets the new area to 0.
    
diff --git a/darray.c b/darray.c
index 7d302c4..3359ed3 100644
--- a/darray.c
+++ b/darray.c
@@ -7,7 +7,7 @@
 //typedef unsigned int dword;
 
 //typedef dword pb;
-#define logD 5
+//#define logD 5
 
 #define PBS (sizeof(pb)*8)
 #define D (1<<logD)
-- 
2.17.1