X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=TextStorage.h;h=1f809b8e4878d186aee720a0cdf802dc05685617;hb=a56dda3a95d2df268969bd3c2c994b5d2f020410;hp=fcd66e58f3e0dcf4915c2c870a55392a8e711e3f;hpb=54d7f377217134290545df4c3f7c254478dc1f88;p=SXSI%2FTextCollection.git

diff --git a/TextStorage.h b/TextStorage.h
index fcd66e5..1f809b8 100644
--- a/TextStorage.h
+++ b/TextStorage.h
@@ -23,7 +23,16 @@
 
 #include "TextCollection.h"
 #include "Tools.h"
+
 #include "incbwt/bits/deltavector.h"
+// Re-define word size to ulong:
+#undef W
+#if __WORDSIZE == 64
+#   define W 64
+#else
+#   define W 32
+#endif
+
 #include <cassert>
 #include <stdexcept>
 
@@ -59,20 +68,22 @@ public:
 
     virtual ~TextStorage()
     {
+        delete offit_;
+        offit_ = 0;
         delete offsets_;
-        offsets_ = 0;
+        offsets_ = 0;        
     }
 
     TextCollection::DocId DocIdAtTextPos(TextCollection::TextPosition i) const
     {
         assert(i < n_);
-        return offsets_->rank(i)-1;
+        return offit_->rank(i)-1;
     }
 
     TextCollection::TextPosition TextStartPos(TextCollection::DocId i) const
     {
         assert(i < (TextCollection::DocId)numberOfTexts_);
-        return offsets_->select(i);
+        return offit_->select(i);
     }
 
     bool IsEndmarker(TextCollection::TextPosition i) const
@@ -80,7 +91,7 @@ public:
         assert(i < n_);
         if (i >= n_ - 1)
             return true;
-        return offsets_->isSet(i+1);
+        return offit_->isSet(i+1);
     }
 
 
@@ -91,7 +102,7 @@ protected:
     const static CSA::usint DV_BLOCK_SIZE = 32;
 
     TextStorage(uchar const * text, TextPosition n)
-        : n_(n), offsets_(0), numberOfTexts_(0)
+        : n_(n), offsets_(0), offit_(0), numberOfTexts_(0)
     { 
         // Delta encoded bitvector of text offsets.
         CSA::DeltaEncoder encoder(DV_BLOCK_SIZE);
@@ -104,12 +115,8 @@ protected:
         
 
         offsets_ = new CSA::DeltaVector(encoder, n_);
-
-/*        for (ulong i = 0; i < n_-1; ++i)
-            if ((text[i] == '\0') != IsEndmarker(i))
-                std::cout << "misplaced endmarker at i = " << i << std::endl;
-*/
-        numberOfTexts_ = offsets_->rank(n_ - 1);
+        offit_ = new CSA::DeltaVector::Iterator(*(offsets_));
+        numberOfTexts_ = offit_->rank(n_ - 1);
     }
     
     TextStorage(std::FILE *);
@@ -117,6 +124,7 @@ protected:
 
     TextPosition n_;
     CSA::DeltaVector *offsets_;
+    CSA::DeltaVector::Iterator *offit_;
     TextPosition numberOfTexts_;
 };
 
@@ -157,7 +165,7 @@ public:
     {
         assert(docId < (TextCollection::DocId)numberOfTexts_);
 
-        TextPosition offset = offsets_->select(docId);
+        TextPosition offset = offit_->select(docId);
         return &text_[offset];
     }
 
@@ -166,7 +174,7 @@ public:
         assert(i < (TextCollection::DocId)numberOfTexts_);
         assert(j < (TextCollection::DocId)numberOfTexts_);
 
-        TextPosition offset = offsets_->select(i);
+        TextPosition offset = offit_->select(i);
         return &text_[offset];
     }
 
@@ -182,7 +190,7 @@ private:
 /******************************************************************
  * LZ-index text collection.
  */
-struct LzTriePimpl; // Pimpl, declared in .cpp
+struct LzTriePimpl; // Using Pimpl idiom to hide LzTrie implementation.
 
 class TextStorageLzIndex : public TextStorage
 {
@@ -252,7 +260,7 @@ public:
 
 private:
     TextPosition n_;
-    uchar *text_; // FIXME Replace with a succinct representation.
+    uchar *text_;
     bool freeText;
 }; // class TextStorageBuilder