HashFunctions.cpp

#include "HashFunctions.h"
#include "HashSequence.h"
 
size_t Cogs::murmur2(const void* bytes, const size_t length) noexcept
{
#if defined(_WIN64) || defined(__x86_64__) || defined(__x86_64) || defined(_M_X64) || defined(__LP64__)
  const size_t m = 0xc6a4a7935bd1e995ull;
  const int r = 47;
#else
  const size_t m = 0x5bd1e995;
  const int r = 24;
#endif
 
  const size_t seed = 0;
 
  size_t h = seed ^ (length * m);
 
  const size_t * data = reinterpret_cast<const size_t *>(bytes);
  const size_t * end = data + (length / sizeof(size_t));
 
  while (data != end) {
    size_t k = *data++;
 
    k *= m;
    k ^= k >> r;
    k *= m;
 
    h ^= k;
    h *= m;
  }
 
  const uint8_t* trailing = reinterpret_cast<const uint8_t*>(data);
 
#if defined(_WIN64) || defined(__x86_64__) || defined(__x86_64) || defined(_M_X64) || defined(__LP64__)
  switch (length & 7) {
  case 7: h ^= size_t(trailing[6]) << 48; [[fallthrough]];
  case 6: h ^= size_t(trailing[5]) << 40; [[fallthrough]];
  case 5: h ^= size_t(trailing[4]) << 32; [[fallthrough]];
  case 4: h ^= size_t(trailing[3]) << 24; [[fallthrough]];
#else
  switch (length & 3) {
#endif
  case 3: h ^= size_t(trailing[2]) << 16; [[fallthrough]];
  case 2: h ^= size_t(trailing[1]) << 8;  [[fallthrough]];
  case 1: h ^= size_t(trailing[0]);
    h *= m;
    break;
  }
 
  h ^= h >> r;
  h *= m;
  h ^= h >> r;
 
  return h;
}
 
size_t Cogs::hashLowercase(std::string_view str, size_t hashValue) noexcept
{
  size_t strSize = str.size();
  for (size_t i = 0; i < strSize; ++i) {
    uint8_t chr = str[i];
 
    // std::tolower uses C locale, avoid surprises and do it explicitly.
    if (('A' <= chr) && (chr <= 'Z')) {
      chr += 'a' - 'A';
    }
    hashValue = Cogs::hash(chr, hashValue);
  }
  return hashValue;
}