turbonss/deps/cmph/cxxmph/seeded_hash.h

148 lines
4.8 KiB
C
Raw Normal View History

2011-05-16 02:47:42 +03:00
#ifndef __CXXMPH_SEEDED_HASH_H__
#define __CXXMPH_SEEDED_HASH_H__
2011-02-14 00:40:26 +02:00
#include <stdint.h> // for uint32_t and friends
2010-11-05 08:40:15 +02:00
#include <cstdlib>
2012-03-21 15:20:30 +02:00
#include <cstring>
2011-11-10 20:44:37 +02:00
#include <unordered_map> // for std::hash
2010-11-05 08:40:15 +02:00
2012-03-14 00:34:24 +02:00
#include "MurmurHash3.h"
2010-11-05 08:40:15 +02:00
#include "stringpiece.h"
namespace cxxmph {
2012-03-20 16:47:55 +02:00
struct h128 {
2012-03-21 15:20:30 +02:00
const uint32_t& operator[](uint8_t i) const { return uint32[i]; }
2012-03-20 16:47:55 +02:00
uint32_t& operator[](uint8_t i) { return uint32[i]; }
2014-03-24 03:51:31 +02:00
uint64_t get64(bool second) const { return (static_cast<uint64_t>(uint32[second << 1]) << 32) | uint32[1 + (second << 1)]; }
2012-03-21 15:20:30 +02:00
void set64(uint64_t v, bool second) { uint32[second << 1] = v >> 32; uint32[1+(second<<1)] = ((v << 32) >> 32); }
bool operator==(const h128 rhs) const { return memcmp(uint32, rhs.uint32, sizeof(uint32)) == 0; }
2012-03-20 16:47:55 +02:00
uint32_t uint32[4];
2012-03-21 15:20:30 +02:00
struct hash32 { uint32_t operator()(const cxxmph::h128& h) const { return h[3]; } };
2012-03-20 16:47:55 +02:00
};
2010-11-05 08:40:15 +02:00
template <class HashFcn>
struct seeded_hash_function {
template <class Key>
2011-02-14 00:40:26 +02:00
uint32_t operator()(const Key& k, uint32_t seed) const {
2012-06-09 08:27:39 +03:00
uint32_t h;
uint32_t h0 = HashFcn()(k);
MurmurHash3_x86_32(reinterpret_cast<const void*>(&h0), 4, seed, &h);
return h;
2010-11-05 08:40:15 +02:00
}
2012-03-14 00:34:24 +02:00
template <class Key>
2012-03-20 16:47:55 +02:00
h128 hash128(const Key& k, uint32_t seed) const {
h128 h;
2012-06-09 08:27:39 +03:00
uint32_t h0 = HashFcn()(k);
MurmurHash3_x64_128(reinterpret_cast<const void*>(&h0), 4, seed, &h);
2012-03-20 16:47:55 +02:00
return h;
2012-03-14 00:34:24 +02:00
}
2010-11-05 08:40:15 +02:00
};
2012-03-14 00:34:24 +02:00
struct Murmur3 {
2010-11-05 08:40:15 +02:00
template<class Key>
2011-02-14 00:40:26 +02:00
uint32_t operator()(const Key& k) const {
2012-03-14 00:34:24 +02:00
uint32_t out;
MurmurHash3_x86_32(reinterpret_cast<const void*>(&k), sizeof(Key), 1 /* seed */, &out);
return out;
}
template <class Key>
2012-03-20 16:47:55 +02:00
h128 hash128(const Key& k) const {
h128 h;
MurmurHash3_x64_128(reinterpret_cast<const void*>(&k), sizeof(Key), 1 /* seed */, &h);
return h;
2010-11-05 08:40:15 +02:00
}
};
2012-03-14 00:34:24 +02:00
struct Murmur3StringPiece {
2010-11-05 08:40:15 +02:00
template <class Key>
2011-02-14 00:40:26 +02:00
uint32_t operator()(const Key& k) const {
2010-11-05 08:40:15 +02:00
StringPiece s(k);
2012-03-14 00:34:24 +02:00
uint32_t out;
MurmurHash3_x86_32(s.data(), s.length(), 1 /* seed */, &out);
return out;
}
template <class Key>
2012-03-20 16:47:55 +02:00
h128 hash128(const Key& k) const {
h128 h;
2012-03-14 00:34:24 +02:00
StringPiece s(k);
2012-03-20 16:47:55 +02:00
MurmurHash3_x64_128(s.data(), s.length(), 1 /* seed */, &h);
return h;
2010-11-05 08:40:15 +02:00
}
};
template <>
2012-03-14 00:34:24 +02:00
struct seeded_hash_function<Murmur3> {
2010-11-05 08:40:15 +02:00
template <class Key>
2011-02-14 00:40:26 +02:00
uint32_t operator()(const Key& k, uint32_t seed) const {
2012-03-14 00:34:24 +02:00
uint32_t out;
MurmurHash3_x86_32(reinterpret_cast<const void*>(&k), sizeof(Key), seed, &out);
return out;
}
template <class Key>
2012-03-20 16:47:55 +02:00
h128 hash128(const Key& k, uint32_t seed) const {
h128 h;
MurmurHash3_x64_128(reinterpret_cast<const void*>(&k), sizeof(Key), seed, &h);
return h;
2010-11-05 08:40:15 +02:00
}
};
template <>
2012-03-14 00:34:24 +02:00
struct seeded_hash_function<Murmur3StringPiece> {
2010-11-05 08:40:15 +02:00
template <class Key>
2011-02-14 00:40:26 +02:00
uint32_t operator()(const Key& k, uint32_t seed) const {
2010-11-05 08:40:15 +02:00
StringPiece s(k);
2012-03-14 00:34:24 +02:00
uint32_t out;
MurmurHash3_x86_32(s.data(), s.length(), seed, &out);
return out;
}
template <class Key>
2012-03-20 16:47:55 +02:00
h128 hash128(const Key& k, uint32_t seed) const {
h128 h;
2012-03-14 00:34:24 +02:00
StringPiece s(k);
2012-03-20 16:47:55 +02:00
MurmurHash3_x64_128(s.data(), s.length(), seed, &h);
return h;
2010-11-05 08:40:15 +02:00
}
};
2011-05-16 02:47:42 +03:00
template <class HashFcn> struct seeded_hash
2010-11-05 08:40:15 +02:00
{ typedef seeded_hash_function<HashFcn> hash_function; };
2012-03-14 00:34:24 +02:00
// Use Murmur3 instead for all types defined in std::hash, plus
2010-11-05 08:40:15 +02:00
// std::string which is commonly extended.
2011-11-10 20:44:37 +02:00
template <> struct seeded_hash<std::hash<char*> >
2012-03-14 00:34:24 +02:00
{ typedef seeded_hash_function<Murmur3StringPiece> hash_function; };
2011-11-10 20:44:37 +02:00
template <> struct seeded_hash<std::hash<const char*> >
2012-03-14 00:34:24 +02:00
{ typedef seeded_hash_function<Murmur3StringPiece> hash_function; };
2011-11-10 20:44:37 +02:00
template <> struct seeded_hash<std::hash<std::string> >
2012-03-14 00:34:24 +02:00
{ typedef seeded_hash_function<Murmur3StringPiece> hash_function; };
2011-11-10 20:44:37 +02:00
template <> struct seeded_hash<std::hash<cxxmph::StringPiece> >
2012-03-14 00:34:24 +02:00
{ typedef seeded_hash_function<Murmur3StringPiece> hash_function; };
2010-11-05 08:40:15 +02:00
2011-11-10 20:44:37 +02:00
template <> struct seeded_hash<std::hash<char> >
2012-03-14 00:34:24 +02:00
{ typedef seeded_hash_function<Murmur3> hash_function; };
2011-11-10 20:44:37 +02:00
template <> struct seeded_hash<std::hash<unsigned char> >
2012-03-14 00:34:24 +02:00
{ typedef seeded_hash_function<Murmur3> hash_function; };
2011-11-10 20:44:37 +02:00
template <> struct seeded_hash<std::hash<short> >
2012-03-14 00:34:24 +02:00
{ typedef seeded_hash_function<Murmur3> hash_function; };
2011-11-10 20:44:37 +02:00
template <> struct seeded_hash<std::hash<unsigned short> >
2012-03-14 00:34:24 +02:00
{ typedef seeded_hash_function<Murmur3> hash_function; };
2011-11-10 20:44:37 +02:00
template <> struct seeded_hash<std::hash<int> >
2012-03-14 00:34:24 +02:00
{ typedef seeded_hash_function<Murmur3> hash_function; };
2011-11-10 20:44:37 +02:00
template <> struct seeded_hash<std::hash<unsigned int> >
2012-03-14 00:34:24 +02:00
{ typedef seeded_hash_function<Murmur3> hash_function; };
2011-11-10 20:44:37 +02:00
template <> struct seeded_hash<std::hash<long> >
2012-03-14 00:34:24 +02:00
{ typedef seeded_hash_function<Murmur3> hash_function; };
2011-11-10 20:44:37 +02:00
template <> struct seeded_hash<std::hash<unsigned long> >
2012-03-14 00:34:24 +02:00
{ typedef seeded_hash_function<Murmur3> hash_function; };
2011-11-10 20:44:37 +02:00
template <> struct seeded_hash<std::hash<long long> >
2012-03-14 00:34:24 +02:00
{ typedef seeded_hash_function<Murmur3> hash_function; };
2011-11-10 20:44:37 +02:00
template <> struct seeded_hash<std::hash<unsigned long long> >
2012-03-14 00:34:24 +02:00
{ typedef seeded_hash_function<Murmur3> hash_function; };
2010-11-05 08:40:15 +02:00
} // namespace cxxmph
2011-05-16 02:47:42 +03:00
#endif // __CXXMPH_SEEDED_HASH_H__