X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLTreeBuilder.cpp;h=7ff1effb9964ac13ee15a3d5e6d56d5a8e2fd9bc;hb=00d137d79d67d8a79c8ff7ab6623a00a3783df5f;hp=3115e4d1d71ddb38059dc34c1cb626517cea1216;hpb=806bef044cdb34e9bf4cd375d336bf08f65618d2;p=SXSI%2FXMLTree.git

diff --git a/XMLTreeBuilder.cpp b/XMLTreeBuilder.cpp
index 3115e4d..7ff1eff 100644
--- a/XMLTreeBuilder.cpp
+++ b/XMLTreeBuilder.cpp
@@ -1,27 +1,33 @@
-#include "basics.h"
+#include "common.h"
 #include "XMLTreeBuilder.h"
 #include "timings.h"
 using std::string;
 
 XMLTreeBuilder::~XMLTreeBuilder(){
-  
+  //free(par_aux);
+  free(tags_aux);
+  //delete other stuff.
+
 }
 
 // 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 XMLTreeBuilder::OpenDocument(bool empty_texts, int sample_rate_text, bool dtc)
+int XMLTreeBuilder::OpenDocument(bool empty_texts,
+				 int sample_rate_text,
+				 bool dtc,
+				 TextCollectionBuilder::index_type_t index_type)
  {
     npar = 0;
     parArraySize = 1;
     disable_tc = dtc;
-
+    text_index_type = index_type;
     STARTTIMER();
-   
+
     par_aux = (pb *)umalloc(sizeof(pb)*parArraySize);
-    
+
     tags_aux = (TagType *) umalloc(sizeof(TagType));
-    
+
     TagName = new vector<string>();
     tIdMap = new std::unordered_map<string,int>();
 
@@ -38,8 +44,9 @@ int XMLTreeBuilder::OpenDocument(bool empty_texts, int sample_rate_text, bool dt
 
     if (disable_tc)
         TextBuilder = 0;
-    else 
-        TextBuilder = TextCollectionBuilder::create((unsigned)sample_rate_text, TextCollectionBuilder::index_type_default);
+    else
+        TextBuilder = TextCollectionBuilder::create((unsigned)sample_rate_text, index_type);
+
     Text = 0;
     empty_texts_aux = (unsigned int *)ucalloc(sizeof(unsigned int),1);
     eta_size = sizeof(unsigned int);
@@ -47,16 +54,16 @@ int XMLTreeBuilder::OpenDocument(bool empty_texts, int sample_rate_text, bool dt
  }
 
 // CloseDocument(): it finishes the construction of the data structure for the XML
-// document. Tree and tags are represented in the final form, dynamic data 
-// structures are made static, and the flag "finished" is set to true. After that, 
+// document. Tree and tags are represented in the final form, dynamic data
+// structures are made static, and the flag "finished" is set to true. After that,
 // the data structure can be queried.
 XMLTree *XMLTreeBuilder::CloseDocument()
- {    
+ {
     //closing parenthesis for the tree root
     //par_aux = (pb *)urealloc(par_aux, sizeof(pb)*(1+npar/(8*sizeof(pb))));
     //setbit(par_aux, npar, CP);
     //npar++;
-    
+
     // makes the text collection static
    STOPTIMER(Parsing);
    PRINTTIME("Parsing XML Document", Parsing);
@@ -72,16 +79,19 @@ XMLTree *XMLTreeBuilder::CloseDocument()
        PRINTTIME("Building TextCollection", Building);
 
     }
-    
+
     XMLTree *T = new XMLTree(par_aux,
-			     npar, 
+			     npar,
 			     TagName,
 			     tIdMap,
-			     empty_texts_aux,                // freed by the constructor
-			     tags_aux,                       //freed by the constructor
+			     empty_texts_aux, // freed by the constructor
+			     tags_aux,        // freed by the constructor
 			     Text,
-			     disable_tc);
-    return T; 
+			     disable_tc,
+			     text_index_type);
+    tags_aux = 0;
+    empty_texts_aux = 0;
+    return T;
  }
 
 
@@ -91,17 +101,17 @@ XMLTree *XMLTreeBuilder::CloseDocument()
 int XMLTreeBuilder::NewOpenTag(string tagname)
  {
     int i;
-    
+
     // inserts a new opening parentheses in the bit sequence
     if (sizeof(pb)*8*parArraySize == npar) { // no space left for the new parenthesis
        par_aux = (pb *)urealloc(par_aux, sizeof(pb)*2*parArraySize);
        parArraySize *= 2;
     }
-    
-    setbit(par_aux,npar,OP);  // marks a new opening parenthesis
-    
+
+    bp_setbit(par_aux,npar,OP);  // marks a new opening parenthesis
+
     TagIdMapIT tag_id = tIdMap->find(tagname);
-   
+
     if (tag_id == tIdMap->end()){
       REGISTER_TAG(TagName,tIdMap,tagname);
       i = TagName->size() - 1;
@@ -112,14 +122,14 @@ int XMLTreeBuilder::NewOpenTag(string tagname)
     if (tagname.compare(PCDATA_OPEN_TAG) == 0 ||
 	tagname.compare(ATTRIBUTE_DATA_OPEN_TAG) == 0){
     };
-    
+
     tags_aux = (TagType *) urealloc(tags_aux, sizeof(TagType)*(npar + 1));
-    
+
     tags_aux[npar] = i; // inserts the new tag id within the preorder sequence of tags
-    
+
     npar++;
-    
-    return 1; // success    
+
+    return 1; // success
  }
 
 
@@ -129,18 +139,18 @@ int XMLTreeBuilder::NewOpenTag(string tagname)
 int XMLTreeBuilder::NewClosingTag(string tagname)
  {
     int i;
-    
+
     // inserts a new closing parentheses in the bit sequence
     if (sizeof(pb)*8*parArraySize == npar) { // no space left for the new parenthesis
        par_aux = (pb *)urealloc(par_aux, sizeof(pb)*2*parArraySize);
        parArraySize *= 2;
     }
-    
-    setbit(par_aux,npar,CP);  // marks a new closing parenthesis
-    
+
+    bp_setbit(par_aux,npar,CP);  // marks a new closing parenthesis
+
     //tagname.insert(0,"/");
 
-    //TagIdMapIT tag_id = tIdMap->find(tagname);    
+    //TagIdMapIT tag_id = tIdMap->find(tagname);
 
     // if (tag_id == tIdMap->end()){
     //   REGISTER_TAG(TagName,tIdMap,tagname);
@@ -152,7 +162,7 @@ int XMLTreeBuilder::NewClosingTag(string tagname)
     tags_aux = (TagType *)urealloc(tags_aux, sizeof(TagType)*(npar + 1));
 
     tags_aux[npar] = CLOSING_TAG_ID; // inserts the new tag id within the preorder sequence of tags
-    
+
     npar++;
 
     return 1; // success
@@ -173,7 +183,7 @@ int XMLTreeBuilder::NewText(string 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.
-   
+
    empty_texts_aux = (uint *)urecalloc(empty_texts_aux,eta_size,n_eta_size);
    eta_size = n_eta_size;
    bitset(empty_texts_aux, npar-1);  // marks the non-empty text with a 1 in the bit vector