X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=TCImplementation.h;h=0f0cab49391db87a9082c3e3128151964bc13eec;hb=65e613f046f3316769cc66eb27bc891d90ec3dc6;hp=5e488db94da8a623ace9f6dad8df4ead78643732;hpb=40ddf9aca842bdc081b6350a4ebfe36b066c94c9;p=SXSI%2FTextCollection.git

diff --git a/TCImplementation.h b/TCImplementation.h
index 5e488db..0f0cab4 100644
--- a/TCImplementation.h
+++ b/TCImplementation.h
@@ -30,6 +30,7 @@
 #include <static_bitsequence.h>
 #include <alphabet_mapper.h>
 #include <static_sequence.h>
+#include <static_sequence_wvtree_noptrs.h>
 
 // Re-define word size to ulong:
 #undef W
@@ -43,8 +44,6 @@
 namespace SXSI 
 {
 
-// Un-comment to compare BWT against a BWT generated from class dynFMI:
-//#define CSA_TEST_BWT
 
 /**
  * Implementation of the TextCollection interface
@@ -125,18 +124,11 @@ private:
 
     // Total number of texts in the collection
     unsigned numberOfTexts;
-    // Total number of texts including empty texts
-    unsigned numberOfAllTexts;
     // Length of the longest text
     ulong maxTextLength;
 
     // Array of document id's in the order of end-markers in BWT
-    // Access by endmarkerDocId[rank_$(L, p) - 1].
-    BlockArray *endmarkerDocId;
-
-    // FIXME Replace with a more succinct data structure
-    // Note: Empty texts are already handled inside XMLTree class.
-    BSGAP *emptyTextRank; // FIXME Remove
+    static_sequence *Doc;
 
     // Following are not part of the public API
     uchar * BWT(uchar *);
@@ -161,7 +153,50 @@ private:
     
         return alphabetrank->rank(0, ep) - ranksp;
     }
-    unsigned CountEndmarkers(TextPosition, TextPosition, DocId, DocId) const;
+
+    /**
+     * Count end-markers in given interval and
+     * within docIds [min,max]
+     */
+    inline unsigned CountEndmarkers(TextPosition sp, TextPosition ep, DocId min, DocId max) const
+    {
+        if (sp != 0)
+            sp = alphabetrank->rank(0, sp - 1);
+        ep = alphabetrank->rank(0, ep);
+        if (ep == 0)
+            return 0;
+        
+        return Doc->count(sp, ep-1, min, max);
+    }
+    
+    /**
+     * Enumerate all end-markers in given interval
+     */
+    inline document_result EnumerateEndmarkers(TextPosition sp, TextPosition ep) const
+    {
+        if (sp != 0)
+            sp = alphabetrank->rank(0, sp - 1);
+        ep = alphabetrank->rank(0, ep);
+        if (ep == 0)
+            return document_result();
+        
+        return Doc->accessAll(sp, ep-1);
+    }
+
+    /**
+     * Enumerate end-markers in given interval and
+     * within docIds [min,max]
+     */
+    inline document_result EnumerateEndmarkers(TextPosition sp, TextPosition ep, DocId min, DocId max) const
+    {
+        if (sp != 0)
+            sp = alphabetrank->rank(0, sp - 1);
+        ep = alphabetrank->rank(0, ep);
+        if (ep == 0)
+            return document_result();
+        
+        return Doc->access(sp, ep-1, min, max);
+    }
 }; // class TCImplementation
 
 } // namespace SXSI