struggle

cxxmph/Makefile.am

@@ -1,5 +1,5 @@
 TESTS = $(check_PROGRAMS)
-check_PROGRAMS = hollow_iterator_test mph_map_test mph_index_test trigraph_test
+check_PROGRAMS = mph_bits_test hollow_iterator_test mph_map_test mph_index_test trigraph_test
 noinst_PROGRAMS = bm_index bm_map
 bin_PROGRAMS = cxxmph
 lib_LTLIBRARIES = libcxxmph.la
@@ -27,4 +27,6 @@ cxxmph_LDADD   = libcxxmph.la
 cxxmph_SOURCES = cxxmph.cc
 
 hollow_iterator_test_SOURCES = hollow_iterator_test.cc
+mph_bits_test_SOURCES = mph_bits_test.cc
+mph_bits_test_LDADD   = libcxxmph.la
 

cxxmph/mph_bits.cc

@@ -1,4 +1,7 @@
 #include "mph_bits.h"
 
 namespace cxxmph {
+
+const uint8_t dynamic_2bitset::vmask[] = { 0xfc, 0xf3, 0xcf, 0x3f};
+
 }

cxxmph/mph_bits.h

@@ -2,13 +2,63 @@
 #define __CXXMPH_MPH_BITS_H__
 
 #include <stdint.h>  // for uint32_t and friends
+#include <cassert>
 #include <climits>
+#include <cstdio>
+#include <cstring>
+#include <limits>
+#include <vector>
 
 namespace cxxmph {
 
-static const uint8_t valuemask[] = { 0xfc, 0xf3, 0xcf, 0x3f};
+class dynamic_2bitset {
+ public:
+  dynamic_2bitset() : data_(NULL), size_(0), one_initialized_(false)  {}
+  dynamic_2bitset(uint32_t size, bool one_initialized = false)
+      : data_(NULL), size_(0), one_initialized_(one_initialized) {
+    resize(size);
+  }
+  ~dynamic_2bitset() { delete [] data_; }
+
+  const uint8_t operator[](uint32_t i) const { return get(i); }
+  uint8_t get(uint32_t i) const { 
+    return (data_[(i >> 2)] >> (((i & 3) << 1)) & 3);
+  }
+  uint8_t set(uint32_t i, uint8_t v) { 
+    uint8_t sf = ((v << ((i & 3) << 1)) | dynamic_2bitset::vmask[i & 3]);
+    fprintf(stderr, "v %d sf %d\n", v, sf);
+    data_[(i >> 2)] &= ((v << ((i & 3) << 1)) | dynamic_2bitset::vmask[i & 3]);
+    assert(get(i) == v);
+  }
+  void resize(uint32_t size) {
+    uint8_t* new_data = new uint8_t[size << 2];
+    assert(one_initialized_);
+    assert(one_initialized_ * ones() == ones());
+    memset(new_data, one_initialized_*ones(), size << 2);
+    assert(new_data[0] == ones());
+    uint8_t* old_data_ = data_;
+    for (int i = 0; i < size_; ++i) { 
+      data_ = old_data_;
+      auto v = get(i);
+      data_ = new_data;
+      set(i, v);
+    }
+    size_ = size;
+    delete [] old_data_;
+    data_ = new_data;
+    assert(data_[0] == ones());
+    assert(get(0) == 3);
+  }
+  static const uint8_t vmask[];
+ private:
+  uint8_t* data_;
+  uint32_t size_;
+  bool one_initialized_;
+  uint8_t ones() { return std::numeric_limits<uint8_t>::max(); }
+};
+
 static void set_2bit_value(uint8_t *d, uint32_t i, uint8_t v) {
-  d[(i >> 2)] &= ((v << ((i & 3) << 1)) | valuemask[i & 3]);
+  d[(i >> 2)] &= ((v << ((i & 3) << 1)) | dynamic_2bitset::vmask[i & 3]);
 }
 static uint32_t get_2bit_value(const uint8_t* d, uint32_t i) {
   return (d[(i >> 2)] >> (((i & 3) << 1)) & 3);

cxxmph/mph_map.h

@@ -126,7 +126,7 @@ class mph_map {
    std::vector<bool> present_;
    std::vector<uint8_t> nests_;
    SimpleMPHIndex<Key, typename seeded_hash<HashFcn>::hash_function> index_;
-   // TODO(davi) optimize slack to no hold a copy of the key
+   // TODO(davi) optimize slack to hold 128 unique bits from hash64 as key
    typedef unordered_map<Key, uint32_t, HashFcn, EqualKey, Alloc> slack_type;
    slack_type slack_;
    size_type size_;