mosesdecoder/contrib/other-builds/moses2/Word.cpp

80 lines
1.6 KiB
C++
Raw Normal View History

2015-10-23 18:46:35 +03:00
/*
* Word.cpp
*
* Created on: 23 Oct 2015
* Author: hieu
*/
2015-10-26 17:58:59 +03:00
#include <vector>
2015-10-23 18:46:35 +03:00
#include "Word.h"
2015-11-18 14:08:32 +03:00
#include "System.h"
2015-11-11 19:23:49 +03:00
#include "legacy/Util2.h"
2015-10-24 14:39:15 +03:00
#include "util/murmur_hash.hh"
2015-10-23 18:46:35 +03:00
2015-10-26 17:58:59 +03:00
using namespace std;
2015-10-23 18:46:35 +03:00
Word::Word() {
2015-11-13 01:51:13 +03:00
Init<const Factor*>(m_factors, MAX_NUM_FACTORS, NULL);
2015-10-23 18:46:35 +03:00
}
2015-12-07 19:49:02 +03:00
Word::Word(const Word &copy)
{
memcpy(m_factors, copy.m_factors, sizeof(const Factor *) * MAX_NUM_FACTORS);
}
2015-10-23 18:46:35 +03:00
Word::~Word() {
// TODO Auto-generated destructor stub
}
2015-11-18 14:08:32 +03:00
void Word::CreateFromString(FactorCollection &vocab, const System &system, const std::string &str)
2015-10-26 17:58:59 +03:00
{
2015-11-12 02:28:18 +03:00
vector<string> toks = Tokenize(str, "|");
2015-10-26 17:58:59 +03:00
for (size_t i = 0; i < toks.size(); ++i) {
2015-10-26 19:02:07 +03:00
const string &tok = toks[i];
//cerr << "tok=" << tok << endl;
2015-11-18 16:07:16 +03:00
const Factor *factor = vocab.AddFactor(tok, system, false);
2015-10-26 17:58:59 +03:00
m_factors[i] = factor;
}
}
2015-10-24 14:39:15 +03:00
size_t Word::hash() const
{
uint64_t seed = 0;
2015-11-13 01:51:13 +03:00
size_t ret = util::MurmurHashNative(m_factors, sizeof(Factor*) * MAX_NUM_FACTORS, seed);
2015-10-24 14:39:15 +03:00
return ret;
}
2015-11-26 03:53:12 +03:00
int Word::Compare(const Word &compare) const
2015-10-24 14:39:15 +03:00
{
2015-11-26 03:53:12 +03:00
2015-11-13 01:51:13 +03:00
int cmp = memcmp(m_factors, compare.m_factors, sizeof(Factor*) * MAX_NUM_FACTORS);
2015-11-26 03:53:12 +03:00
return cmp;
/*
int ret = m_factors[0]->GetString().compare(compare.m_factors[0]->GetString());
return ret;
*/
2015-10-24 14:39:15 +03:00
}
2015-10-26 19:32:47 +03:00
2015-11-25 20:35:22 +03:00
bool Word::operator<(const Word &compare) const
{
2015-11-26 03:53:12 +03:00
int cmp = Compare(compare);
2015-11-25 20:35:22 +03:00
return (cmp < 0);
}
2015-10-26 19:32:47 +03:00
std::ostream& operator<<(std::ostream &out, const Word &obj)
{
2015-10-26 21:42:42 +03:00
bool outputAlready = false;
2015-10-26 19:32:47 +03:00
for (size_t i = 0; i < MAX_NUM_FACTORS; ++i) {
2015-11-13 01:51:13 +03:00
const Factor *factor = obj.m_factors[i];
2015-10-26 21:42:42 +03:00
if (factor) {
if (outputAlready) {
out << "|";
}
out << *factor;
outputAlready = true;
}
2015-10-26 19:32:47 +03:00
}
return out;
}