X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=XMLTreeBuilder.h;h=336ca2dfff33fe86765143f8fe028ff912422397;hb=235e3214904e390d2f101c5d5bf7def98745b132;hp=38d81baf7a3849f477249203845ed6cec2f5b4b1;hpb=aa6692a9fd2badf8e8e686b92075f041dc03bbef;p=SXSI%2FXMLTree.git

diff --git a/XMLTreeBuilder.h b/XMLTreeBuilder.h
index 38d81ba..336ca2d 100644
--- a/XMLTreeBuilder.h
+++ b/XMLTreeBuilder.h
@@ -21,6 +21,7 @@
 
 #ifndef XMLTREEBUILDER_H_
 #define XMLTREEBUILDER_H_
+#include <unordered_map>
 #include "TextCollection/TextCollectionBuilder.h"
 #include <stdio.h>
 #include <stdlib.h>
@@ -31,8 +32,9 @@
 #undef WW
 #undef Wminusone
 
-#include "bp.h"
+
 #include "XMLTree.h"
+#include "bp.h"
 #include <static_bitsequence.h>
 #include <alphabet_mapper.h>
 #include <static_sequence.h>
@@ -47,6 +49,7 @@ using SXSI::TextCollectionBuilder;
 #define bitclean(e,p) ((e)[(p)/W] &= ~(1<<((p)%W)))
 
 
+
 class XMLTreeBuilder {
   
    /** Array containing the balanced parentheses sequence */
@@ -55,9 +58,8 @@ class XMLTreeBuilder {
    int npar;
 
    /** Mapping from tag identifer to tag name */  
-   unsigned char **TagName;
-   int ntagnames;
-
+   vector<string> *TagName;
+   TagIdMap * tIdMap;
    /** Array containing the sequence of tags */
    TagType *tags_aux;
    
@@ -66,19 +68,11 @@ class XMLTreeBuilder {
    TextCollection *Text;
    
    /** The texts in the XML document (cached for faster display) */
-   vector<string> CachedText;
 
-   /** boolean flag indicating whether we are indexing empty texts or not */
-   bool indexing_empty_texts; 
-   unsigned int *empty_texts_aux;
-
-   // The TagName array should always contains two special tags
-   // <@> for attributes and <$> for PCDATA.
-   // <$> can never be in a document (since we handle the text differently)
-   // but <@> can be returned by the parser. This boolean is needed for the construction
-   // of the Tag bitmap to know if <@> must be taken into account or not
-   bool found_attributes;
+   vector<string> *CachedText;
 
+   unsigned int *empty_texts_aux;
+   int eta_size;
    // Allows to disable the TextCollection for benchmarkin purposes
    bool disable_tc;
 
@@ -88,9 +82,8 @@ public:
 
    ~XMLTreeBuilder();
    
-   /** OpenDocument(empty_texts,sample_rate_text,dtc): initilizes the construction
-    * of the data structure for the XML document. Parameter empty_texts 
-    * indicates whether we index empty texts in document or not. Parameter 
+   /** OpenDocument(sample_rate_text,dtc): initilizes the construction
+    * of the data structure for the XML document.  Parameter 
     * sample_rate_text indicates the sampling rate for the text searching data
     * structures (small values get faster searching but a bigger space 
     * requirement). dtc disable the use of the TextCollection
@@ -108,24 +101,23 @@ public:
    /** NewOpenTag(tagname): indicates the event of finding a new opening tag 
     * in the document. Tag name is given. Returns a non-zero value upon 
     * success, and returns NULLT in case of error. */
-   int NewOpenTag(unsigned char *tagname);
+   int NewOpenTag(string tagname);
    
    /** NewClosingTag(tagname): indicates the event of finding a new closing tag
     *  in the document. Tag name is given. Returns a non-zero value upon 
     *  success, and returns NULLT in case of error. */
-   int NewClosingTag(unsigned char *tagname);
+   int NewClosingTag(string tagname);
  
-   /** NewText(s): indicates the event of finding a new (non-empty) text s in 
+   /** NewText(s): indicates the event of finding a new text s in 
     * the document. The new text is inserted within the text collection. 
-    * Returns a non-zero value upon success, NULLT in case of error. */
-   int NewText(unsigned char *s);
-
-   /** NewEmptyText(): indicates the event of finding a new empty text in the 
-    * document. In case of indexing empty and non-empty texts, we insert the 
-    * empty texts into the text collection. In case of indexing only non-empty
-    * texts, it just indicates an empty text in the bit vector of empty texts. 
-    * Returns a non-zero value upon success, NULLT in case of error. */
-   int NewEmptyText();
+    * Returns a non-zero value upon success, NULLT in case of error. 
+    * If the string is empty, which is legal in attributes, then
+    * the string the sequence '\0x01\0x00' is inserted in the TextCollection
+    * It is ok to do so since a non printable character cannot occur in an XML document
+    */
+   int NewText(string text);
+
+
 };
 #endif