1
Fork 0
turbonss/cxxmph/cmph_hash_map.h

176 lines
5.4 KiB
C
Raw Normal View History

2010-11-08 22:19:44 +02:00
#include <algorithm>
#include <unordered_map>
2010-06-28 22:01:18 +03:00
#include <vector>
#include <utility> // for std::pair
2010-10-29 09:26:37 +03:00
#include "MurmurHash2.h"
#include "mphtable.h"
namespace cxxmph {
2010-06-28 22:01:18 +03:00
// Save on repetitive typing.
#define CMPH_TMPL_SPEC template <class Key, class Data, class HashFcn, class EqualKey, class Alloc>
2010-06-28 22:01:18 +03:00
#define CMPH_CLASS_SPEC cmph_hash_map<Key, Data, HashFcn, EqualKey, Alloc>
#define CMPH_METHOD_DECL(r, m) CMPH_TMPL_SPEC typename CMPH_CLASS_SPEC::r CMPH_CLASS_SPEC::m
2010-11-08 22:19:44 +02:00
template <class Key, class Data, class HashFcn = std::hash<Key>, class EqualKey = std::equal_to<Key>, class Alloc = std::allocator<Data> >
2010-06-28 22:01:18 +03:00
class cmph_hash_map {
public:
typedef Key key_type;
typedef Data data_type;
typedef std::pair<Key, Data> value_type;
typedef HashFcn hasher;
typedef EqualKey key_equal;
typedef typename std::vector<value_type>::pointer pointer;
typedef typename std::vector<value_type>::reference reference;
typedef typename std::vector<value_type>::const_reference const_reference;
typedef typename std::vector<value_type>::size_type size_type;
typedef typename std::vector<value_type>::difference_type difference_type;
typedef typename std::vector<value_type>::iterator iterator;
typedef typename std::vector<value_type>::const_iterator const_iterator;
// For making macros simpler.
typedef void void_type;
typedef bool bool_type;
typedef std::pair<iterator, bool> insert_return_type;
cmph_hash_map();
~cmph_hash_map();
iterator begin();
iterator end();
const_iterator begin() const;
const_iterator end() const;
size_type size() const;
bool empty() const;
void clear();
void erase(iterator pos);
void erase(const key_type& k);
std::pair<iterator, bool> insert(const value_type& x);
iterator find(const key_type& k);
const_iterator find(const key_type& k) const;
data_type& operator[](const key_type &k);
void pack() { rehash(); }
private:
2010-11-05 08:40:15 +02:00
template <typename iterator>
struct iterator_first : public iterator {
iterator_first(iterator it) : iterator(it) { }
const typename iterator::value_type::first_type& operator*() const {
return this->iterator::operator*().first;
}
};
template <typename iterator>
iterator_first<iterator> make_iterator_first(iterator it) {
return iterator_first<iterator>(it);
}
void rehash();
std::vector<value_type> values_;
2011-02-14 00:40:26 +02:00
SimpleMPHTable<Key, typename cxxmph_hash<HashFcn>::hash_function> table_;
2010-11-08 22:19:44 +02:00
// TODO(davi) optimize slack to no hold a copy of the key
2011-02-14 00:40:26 +02:00
typedef typename std::unordered_map<Key, uint32_t, HashFcn, EqualKey, Alloc> slack_type;
2010-11-05 08:40:15 +02:00
slack_type slack_;
2010-06-28 22:01:18 +03:00
};
CMPH_TMPL_SPEC
bool operator==(const CMPH_CLASS_SPEC& lhs, const CMPH_CLASS_SPEC& rhs) {
return lhs.values_ == rhs.values_;
}
2010-10-29 09:26:37 +03:00
CMPH_TMPL_SPEC CMPH_CLASS_SPEC::cmph_hash_map() {
2010-06-28 22:01:18 +03:00
rehash();
}
CMPH_TMPL_SPEC CMPH_CLASS_SPEC::~cmph_hash_map() {
}
CMPH_METHOD_DECL(insert_return_type, insert)(const value_type& x) {
iterator it = find(x.first);
if (it != end()) return std::make_pair(it, false);
values_.push_back(x);
2010-11-08 22:19:44 +02:00
slack_.insert(std::make_pair(x.first, values_.size() - 1));
if (slack_.size() == table_.size() ||
(slack_.size() >= 256 && table_.size() == 0)) {
2010-10-29 09:26:37 +03:00
rehash();
}
2010-06-28 22:01:18 +03:00
it = find(x.first);
return std::make_pair(it, true);
}
CMPH_METHOD_DECL(void_type, rehash)() {
if (values_.empty()) return;
2010-11-08 22:19:44 +02:00
std::cerr << "Calling Reset with "
<< table_.size() << " keys in table "
<< slack_.size() << " keys in slack "
<< values_.size() << " key in total" << std::endl;
2010-06-28 22:01:18 +03:00
slack_type().swap(slack_);
2010-11-09 02:02:18 +02:00
bool success = table_.Reset(
make_iterator_first(values_.begin()),
make_iterator_first(values_.end()));
assert(success);
2010-06-28 22:01:18 +03:00
std::vector<value_type> new_values(values_.size());
2010-11-06 13:14:07 +02:00
for (const_iterator it = values_.begin(), end = values_.end();
it != end; ++it) {
size_type id = table_.index(it->first);
new_values[id] = *it;
2010-06-28 22:01:18 +03:00
}
values_.swap(new_values);
}
CMPH_METHOD_DECL(iterator, begin)() { return values_.begin(); }
CMPH_METHOD_DECL(iterator, end)() { return values_.end(); }
CMPH_METHOD_DECL(const_iterator, begin)() const { return values_.begin(); }
CMPH_METHOD_DECL(const_iterator, end)() const { return values_.end(); }
CMPH_METHOD_DECL(bool_type, empty)() const { return values_.empty(); }
CMPH_METHOD_DECL(void_type, clear)() {
2010-06-28 22:01:18 +03:00
values_.clear();
slack_.clear();
table_.clear();
2010-06-28 22:01:18 +03:00
}
CMPH_METHOD_DECL(void_type, erase)(iterator pos) {
values_.erase(pos);
rehash();
}
CMPH_METHOD_DECL(void_type, erase)(const key_type& k) {
iterator it = find(k);
if (it == end()) return;
erase(it);
}
CMPH_METHOD_DECL(const_iterator, find)(const key_type& k) const {
if (!slack_.empty()) {
2010-11-08 22:19:44 +02:00
typename slack_type::const_iterator it = slack_.find(k);
if (it != slack_.end()) return values_.begin() + it->second;
2010-06-28 22:01:18 +03:00
}
2010-10-29 09:26:37 +03:00
if (table_.size() == 0) return end();
size_type id = table_.index(k);
2010-06-28 22:01:18 +03:00
if (key_equal()(values_[id].first, k)) {
return values_.begin() + id;
}
return end();
}
CMPH_METHOD_DECL(iterator, find)(const key_type& k) {
if (!slack_.empty()) {
typename slack_type::const_iterator it = slack_.find(k);
if (it != slack_.end()) return values_.begin() + it->second;
}
2010-10-29 09:26:37 +03:00
if (table_.size() == 0) return end();
size_type id = table_.index(k);
2010-06-28 22:01:18 +03:00
if (key_equal()(values_[id].first, k)) {
return values_.begin() + id;
}
return end();
}
CMPH_METHOD_DECL(data_type&, operator[])(const key_type& k) {
return insert(std::make_pair(k, data_type())).first->second;
}
2010-10-29 09:26:37 +03:00
} // namespace cxxmph