X-Git-Url: http://git.nguyen.vg/gitweb/?a=blobdiff_plain;f=incbwt%2Fsasamples.h;fp=incbwt%2Fsasamples.h;h=da154d0cb738aa7f6064e436a81399089865d4f5;hb=13e254b7c0ee22dffbc7c3125cee0408f9b375da;hp=bb4eee828c6218179ccfa3aea62315d20a7edec6;hpb=e4b6bdc7cc2a1372e4d4dae50acac55cebcc7e9b;p=SXSI%2FTextCollection.git

diff --git a/incbwt/sasamples.h b/incbwt/sasamples.h
index bb4eee8..da154d0 100644
--- a/incbwt/sasamples.h
+++ b/incbwt/sasamples.h
@@ -15,37 +15,52 @@ namespace CSA
 class SASamples
 {
   public:
-    const static usint BLOCK_SIZE = 16;
+    const static usint INDEX_BLOCK_SIZE = 16;
 
     SASamples(std::ifstream& sample_file, usint sample_rate);
-    SASamples(usint* array, usint data_size, usint sample_rate);
+
+    // These assume < 2 GB data. Use the second one when there are multiple sequences.
+    SASamples(uint* array, usint data_size, usint sample_rate);
+    SASamples(uint* inverse, DeltaVector* end_points, usint data_size, usint sample_rate);
+
     ~SASamples();
 
     // Destroys contents of index and increment.
     // We assume index and increment have same sample rate.
-    SASamples(SASamples& index, SASamples& increment, usint* positions, usint number_of_positions);
+    // positions must not containt the positions of end of sequence markers.
+    // number_of_sequences is subtracted from each position before the value is used.
+    SASamples(SASamples& index, SASamples& increment, usint* positions, usint number_of_positions, usint number_of_sequences);
 
-    void writeTo(std::ofstream& sample_file);
+    void writeTo(std::ofstream& sample_file) const;
 
     // Returns i such that SA[i] = value.
     // If SA[i] is not sampled, returns the next sampled value. (Don't try!)
     // Value is actual 0-based suffix array value.
     // Returns size if value is too large.
-    usint inverseSA(usint value);
+    inline usint inverseSA(usint value) const
+    {
+      if(value >= this->size) { return this->size; }
+      DeltaVector::Iterator iter(*(this->indexes));
+      return iter.select(this->inverse_samples->readItemConst(value / this->rate));
+    }
 
     // Returns the value of ith sample in suffix array order.
-    inline usint getSample(usint i)
+    inline usint getSample(usint i) const
     {
-      return std::min(this->samples->readItem(i) * this->rate, this->size - 1);
+      return std::min(this->samples->readItemConst(i) * this->rate, this->size - 1);
     }
 
     // Returns (ind, sample number) where ind >= index or (size, ???).
-    pair_type getFirstSampleAfter(usint index);
+    inline pair_type getFirstSampleAfter(usint index) const
+    {
+      DeltaVector::Iterator iter(*(this->indexes));
+      return iter.valueAfter(index);
+    }
 
-    inline usint getSampleRate() { return this->rate; }
-    inline usint getNumberOfSamples() { return this->items; }
+    inline usint getSampleRate() const { return this->rate; }
+    inline usint getNumberOfSamples() const { return this->items; }
 
-    usint reportSize();
+    usint reportSize() const;
 
   private:
     usint integer_bits;
@@ -53,13 +68,18 @@ class SASamples
 
     DeltaVector* indexes;
 
-    FastBitBuffer* samples;
-    FastBitBuffer* inverse_samples;
+    ReadBuffer* samples;
+    ReadBuffer* inverse_samples;
 
     void buildInverseSamples();
 
     // Note: contents of original samples are deleted.
-    void mergeSamples(SASamples& index, SASamples& increment, usint* positions, usint n);
+    void mergeSamples(SASamples& index, SASamples& increment, usint* positions, usint n, usint skip);
+
+    // These are not allowed.
+    SASamples();
+    SASamples(const SASamples&);
+    SASamples& operator = (const SASamples&);
 };