New bit code works, need to cleanup logging.

cxxmph/mph_bits.h

@@ -17,16 +17,23 @@ namespace cxxmph {
 
 class dynamic_2bitset {
  public:
-  dynamic_2bitset() : fill_(false)  {}
+  dynamic_2bitset() : size_(0), fill_(false)  {}
   dynamic_2bitset(uint32_t size, bool fill = false)
       : size_(size), fill_(fill), data_(ceil(size / 4.0), ones()*fill) {
+   if (data_.size()) fprintf(stderr, "creating %p size %d\n", &data_[0], data_.size());
+  }
+  ~dynamic_2bitset() {
+    if (data_.size()) fprintf(stderr, "Deleting %p size %d\n", &data_[0], data_.size());
   }
 
   const uint8_t operator[](uint32_t i) const { return get(i); }
-  uint8_t get(uint32_t i) const { 
+  const uint8_t get(uint32_t i) const { 
+    assert(i < size());
+    assert((i >> 2) < data_.size());
     return (data_[(i >> 2)] >> (((i & 3) << 1)) & 3);
   }
   uint8_t set(uint32_t i, uint8_t v) { 
+    assert((i >> 2) < data_.size());
     data_[(i >> 2)] |= ones() ^ dynamic_2bitset::vmask[i & 3];
     data_[(i >> 2)] &= ((v << ((i & 3) << 1)) | dynamic_2bitset::vmask[i & 3]);
     assert(v <= 3);
@@ -39,17 +46,18 @@ class dynamic_2bitset {
   void swap(dynamic_2bitset& other) {
     std::swap(other.size_, size_);
     std::swap(other.fill_, fill_);
-    std::swap(other.data_, data_);
+    other.data_.swap(data_);
   }
-  void clear() { data_.clear(); }
+  void clear() { data_.clear(); size_ = 0; }
     
   uint32_t size() const { return size_; }
   static const uint8_t vmask[];
- private:
+  const std::vector<uint8_t>& data() const { return data_; }
+// private:
   uint32_t size_;
   bool fill_;
   std::vector<uint8_t> data_;
-  uint8_t ones() { return std::numeric_limits<uint8_t>::max(); }
+  const uint8_t ones() { return std::numeric_limits<uint8_t>::max(); }
 };
 
 static void set_2bit_value(uint8_t *d, uint32_t i, uint8_t v) {
@@ -67,6 +75,8 @@ static uint32_t nextpoweroftwo(uint32_t k) {
 
 // Interesting bit tricks that might end up here:
 // http://graphics.stanford.edu/~seander/bithacks.html#ZeroInWord
+// Fast a % (k*2^t)
+// http://www.azillionmonkeys.com/qed/adiv.html
   
 }  // namespace cxxmph
 

cxxmph/mph_bits_test.cc

@@ -40,10 +40,18 @@ int main(int argc, char** argv) {
   if (size_corner1.size() != 1) exit(-1);
   dynamic_2bitset size_corner2(2);
   if (size_corner2.size() != 2) exit(-1);
-  (dynamic_2bitset(4)).swap(size_corner2);
+  (dynamic_2bitset(4, true)).swap(size_corner2);
   if (size_corner2.size() != 4) exit(-1);
+  for (int i = 0; i < size_corner2.size(); ++i) {
+    if (size_corner2[i] != 3) exit(-1);
+  }
+  size_corner2.clear();
+  if (size_corner2.size() != 0) exit(-1);
 
-
+  dynamic_2bitset empty;
+  empty.clear();
+  dynamic_2bitset large(1000, true);
+  empty.swap(large);
 }
   
   

cxxmph/mph_index.cc

@@ -44,9 +44,6 @@ MPHIndex::~MPHIndex() {
 }
 
 void MPHIndex::clear() {
-  if (!deserialized_) delete [] g_;
-  g_ = NULL;
-  g_size_ = 0;
   if (!deserialized_) delete [] ranktable_;
   ranktable_ = NULL;
   ranktable_size_ = 0;
@@ -113,13 +110,9 @@ void MPHIndex::Assigning(
     const vector<TriGraph::Edge>& edges, const vector<uint32_t>& queue) {
   uint32_t current_edge = 0;
   vector<bool> marked_vertices(n_ + 1);
+  dynamic_2bitset().swap(g_);
   // Initialize vector of half nibbles with all bits set.
-  g_size_ = static_cast<uint32_t>(ceil(n_/4.0));
+  dynamic_2bitset g(n_, true /* set bits to 1 */);
-  if (!deserialized_) delete [] g_;
-  g_ = NULL;
-  uint8_t* g = new uint8_t[g_size_];
-  memset(g, std::numeric_limits<uint8_t>::max(), g_size_);
-  assert(g[g_size_ - 1] == 255);
 
   uint32_t nedges = m_;  // for legibility
   for (int i = nedges - 1; i + 1 >= 1; --i) {
@@ -133,35 +126,35 @@ void MPHIndex::Assigning(
     */
     if (!marked_vertices[e[0]]) {
       if (!marked_vertices[e[1]]) {
-        set_2bit_value(g, e[1], kUnassigned);
+        g.set(e[1], kUnassigned);
         marked_vertices[e[1]] = true;
       }
       if (!marked_vertices[e[2]]) {
-        set_2bit_value(g, e[2], kUnassigned);
+        g.set(e[2], kUnassigned);
 	assert(marked_vertices.size() > e[2]);
         marked_vertices[e[2]] = true;
       }
-      set_2bit_value(g, e[0], (6 - (get_2bit_value(g, e[1]) + get_2bit_value(g, e[2]))) % 3);
+      g.set(e[0], (6 - (g[e[1]] + g[e[2]])) % 3);
       marked_vertices[e[0]] = true;
     } else if (!marked_vertices[e[1]]) {
       if (!marked_vertices[e[2]]) {
-        set_2bit_value(g, e[2], kUnassigned);
+        g.set(e[2], kUnassigned);
         marked_vertices[e[2]] = true;
       }
-      set_2bit_value(g, e[1], (7 - (get_2bit_value(g, e[0]) + get_2bit_value(g, e[2]))) % 3);
+      g.set(e[1], (7 - (g[e[0]] + g[e[2]])) % 3);
       marked_vertices[e[1]] = true;
     } else {
-      set_2bit_value(g, e[2], (8 - (get_2bit_value(g, e[0]) + get_2bit_value(g, e[1]))) % 3);
+      g.set(e[2], (8 - (g[e[0]] + g[e[1]])) % 3);
       marked_vertices[e[2]] = true;
     }
     /*
     cerr << "A: " << e[0] << " " << e[1] << " " << e[2] << " -> "
-        << get_2bit_value(g, e[0]) << " "
+        << static_cast<uint32_t>(g[e[0]]) << " "
-        << get_2bit_value(g, e[1]) << " "
+        << static_cast<uint32_t>(g[e[1]]) << " "
-        << get_2bit_value(g, e[2]) << " " << endl;
+        << static_cast<uint32_t>(g[e[2]]) << " " << endl;
     */
   }
-  g_ = g;
+  g_.swap(g);
 }
 
 void MPHIndex::Ranking() {
@@ -194,19 +187,17 @@ uint32_t MPHIndex::Rank(uint32_t vertex) const {
   uint32_t beg_idx_v = index << b_;
   uint32_t beg_idx_b = beg_idx_v >> 2;
   uint32_t end_idx_b = vertex >> 2;
-  while (beg_idx_b < end_idx_b) base_rank += kBdzLookupIndex[g_[beg_idx_b++]];
+  while (beg_idx_b < end_idx_b) base_rank += kBdzLookupIndex[g_.data()[beg_idx_b++]];
   beg_idx_v = beg_idx_b << 2;
   // cerr << "beg_idx_v: " << beg_idx_v << endl;
   // cerr << "base rank: " << base_rank << endl;
-  /*
   cerr << "G: ";
   for (unsigned int i = 0; i < n_; ++i) {
-    cerr << get_2bit_value(g_, i) << " ";
+    cerr << static_cast<uint32_t>(g_[i]) << " ";
   }
   cerr << endl;
-  */
   while (beg_idx_v < vertex) {
-    if (get_2bit_value(g_, beg_idx_v) != kUnassigned) ++base_rank;
+    if (g_[beg_idx_v] != kUnassigned) ++base_rank;
     ++beg_idx_v;
   }
   // cerr << "Base rank: " << base_rank << endl;
@@ -214,21 +205,12 @@ uint32_t MPHIndex::Rank(uint32_t vertex) const {
 }
 
 uint32_t MPHIndex::serialize_bytes_needed() const {
-  return sizeof(MPHIndex) + g_size_ + ranktable_size_*sizeof(uint32_t);
+  return 0;
 }
 void MPHIndex::serialize(char* memory) const {
-  memcpy(memory, this, sizeof(MPHIndex));
-  memcpy(memory + sizeof(MPHIndex), g_, g_size_);
-  memcpy(memory + sizeof(MPHIndex) + g_size_,
-      ranktable_, ranktable_size_*sizeof(uint32_t));
 }
 
 bool MPHIndex::deserialize(const char* serialized_memory) {
-  memcpy(this, serialized_memory, sizeof(MPHIndex)); 
-  g_ = reinterpret_cast<const uint8_t*>(serialized_memory + sizeof(MPHIndex));
-  ranktable_ = reinterpret_cast<const uint32_t*>(
-      serialized_memory + sizeof(MPHIndex) + g_size_);
-  deserialized_ = true;
   return true;
 }
 

cxxmph/mph_index.h

@@ -45,7 +45,6 @@ class MPHIndex {
  public:
   MPHIndex(double c = 1.23, uint8_t b = 7) :
       c_(c), b_(b), m_(0), n_(0), k_(0), r_(1),
-      g_(NULL), g_size_(0),
       ranktable_(NULL), ranktable_size_(0),
       deserialized_(false) { }
   ~MPHIndex();
@@ -103,10 +102,8 @@ class MPHIndex {
   uint32_t r_;
   uint32_t nest_displacement_[3];  // derived from r_
 
-  // The array containing the minimal perfect hash function graph. Do not use
+  // The array containing the minimal perfect hash function graph.
-  // c++ vector to make mmap based backing easier.
+  dynamic_2bitset g_;
-  const uint8_t* g_;
-  uint32_t g_size_;
   uint8_t threebit_mod3[10];  // speed up mod3 calculation for 3bit ints
   // The table used for the rank step of the minimal perfect hash function
   const uint32_t* ranktable_;
@@ -156,6 +153,7 @@ bool MPHIndex::Reset(
   }
   if (iterations == 0) return false;
   Assigning(edges, queue);
+  fprintf(stderr, "Assignment finished\n");
   std::vector<TriGraph::Edge>().swap(edges);
   Ranking();
   deserialized_ = false;
@@ -192,7 +190,7 @@ void MPHIndex::hash_vector(const Key& key, uint32_t* h) const {
 template <class SeededHashFcn, class Key>
 uint32_t MPHIndex::perfect_hash(const Key& key) const {
   uint32_t h[4];
-  if (!g_size_) return 0;
+  if (!g_.size()) return 0;
   SeededHashFcn().hash64(key, hash_seed_[0], h);
   h[0] = (h[0] % r_) + nest_displacement_[0];
   h[1] = (h[1] % r_) + nest_displacement_[1];
@@ -200,12 +198,11 @@ uint32_t MPHIndex::perfect_hash(const Key& key) const {
   // h[0] = (h[0] & (r_-1)) + nest_displacement_[0];
   // h[1] = (h[1] & (r_-1)) + nest_displacement_[1];
   // h[2] = (h[2] & (r_-1)) + nest_displacement_[2];
-  // cerr << "g_.size() " << g_size_ << " h0 >> 2 " << (h[0] >> 2) << endl;
+  assert((h[0]) < g_.size());
-  assert((h[0] >> 2) <g_size_);
+  assert((h[1]) < g_.size());
-  assert((h[1] >> 2) <g_size_);
+  assert((h[2]) < g_.size());
-  assert((h[2] >> 2) <g_size_);
   uint8_t nest = threebit_mod3[
-      get_2bit_value(g_, h[0]) + get_2bit_value(g_, h[1]) + get_2bit_value(g_, h[2])];
+      g_[h[0]] + g_[h[1]] + g_[h[2]]];
   uint32_t vertex = h[nest];
   return vertex;
 }

cxxmph/mph_index_test.cc

@@ -33,8 +33,10 @@ int main(int argc, char** argv) {
   cerr << endl;
   sort(ids.begin(), ids.end());
   for (vector<int>::size_type i = 0; i < ids.size(); ++i) assert(ids[i] == static_cast<vector<int>::value_type>(i));
+   /*
   char* serialized = new char[mph_index.serialize_bytes_needed()];
   mph_index.serialize(serialized);
   SimpleMPHIndex<string> other_mph_index;
   other_mph_index.deserialize(serialized);
+  */
 }