2015-10-23 20:33:12 +03:00
|
|
|
/*
|
2015-10-26 00:20:55 +03:00
|
|
|
* System.cpp
|
2015-10-23 20:33:12 +03:00
|
|
|
*
|
|
|
|
* Created on: 23 Oct 2015
|
|
|
|
* Author: hieu
|
|
|
|
*/
|
2015-10-27 18:46:37 +03:00
|
|
|
#include <string>
|
|
|
|
#include <iostream>
|
|
|
|
#include <boost/foreach.hpp>
|
2015-11-10 15:46:26 +03:00
|
|
|
#include <boost/thread.hpp>
|
|
|
|
#include <boost/thread/mutex.hpp>
|
2015-10-26 00:20:55 +03:00
|
|
|
#include "System.h"
|
2015-11-03 16:24:39 +03:00
|
|
|
#include "FF/FeatureFunction.h"
|
2015-11-04 16:09:53 +03:00
|
|
|
#include "TranslationModel/UnknownWordPenalty.h"
|
2015-11-11 19:23:49 +03:00
|
|
|
#include "legacy/Util2.h"
|
2015-10-27 18:46:37 +03:00
|
|
|
#include "util/exception.hh"
|
|
|
|
|
|
|
|
using namespace std;
|
2015-10-23 20:33:12 +03:00
|
|
|
|
2015-12-10 23:49:30 +03:00
|
|
|
namespace Moses2
|
|
|
|
{
|
|
|
|
|
2016-03-31 23:00:16 +03:00
|
|
|
System::System(const Parameter ¶msArg) :
|
|
|
|
params(paramsArg), featureFunctions(*this)
|
2015-10-27 18:46:37 +03:00
|
|
|
{
|
2016-04-07 16:51:43 +03:00
|
|
|
options.init(paramsArg);
|
|
|
|
|
2016-03-31 23:00:16 +03:00
|
|
|
bestCollector.reset(new OutputCollector());
|
|
|
|
|
|
|
|
ini_performance_options();
|
|
|
|
params.SetParameter(cpuAffinityOffset, "cpu-affinity-offset", 0);
|
|
|
|
params.SetParameter(cpuAffinityOffsetIncr, "cpu-affinity-increment", 1);
|
|
|
|
|
|
|
|
reportSegmentation = (params.GetParam("report-segmentation-enriched") ? 2 :
|
|
|
|
params.GetParam("report-segmentation") ? 1 : 0);
|
|
|
|
|
|
|
|
params.SetParameter(outputHypoScore, "output-hypo-score", false);
|
|
|
|
|
|
|
|
const PARAM_VEC *section;
|
|
|
|
|
|
|
|
section = params.GetParam("n-best-list");
|
|
|
|
if (section) {
|
|
|
|
if (section->size() >= 2) {
|
|
|
|
nBestPath = section->at(0);
|
|
|
|
nbestSize = Scan<size_t>(section->at(1));
|
|
|
|
distinctNBest = (section->size() > 2 && section->at(2) == "distinct");
|
2015-12-01 02:03:33 +03:00
|
|
|
}
|
2016-03-31 23:00:16 +03:00
|
|
|
else {
|
|
|
|
throw "wrong format for switch -n-best-list file size [disinct]";
|
2016-03-17 20:54:25 +03:00
|
|
|
}
|
2016-03-31 23:00:16 +03:00
|
|
|
}
|
|
|
|
else {
|
|
|
|
nbestSize = 0;
|
|
|
|
}
|
|
|
|
|
|
|
|
// output collectors
|
|
|
|
if (nbestSize) {
|
|
|
|
nbestCollector.reset(new OutputCollector(nBestPath));
|
|
|
|
}
|
2016-03-17 20:54:25 +03:00
|
|
|
|
2016-03-31 23:00:16 +03:00
|
|
|
featureFunctions.Create();
|
|
|
|
LoadWeights();
|
|
|
|
cerr << "START featureFunctions.Load()" << endl;
|
|
|
|
featureFunctions.Load();
|
|
|
|
cerr << "START LoadMappings()" << endl;
|
|
|
|
LoadMappings();
|
|
|
|
cerr << "END LoadMappings()" << endl;
|
2015-10-27 18:46:37 +03:00
|
|
|
}
|
|
|
|
|
2016-03-20 14:40:39 +03:00
|
|
|
System::~System()
|
|
|
|
{
|
2015-10-27 18:46:37 +03:00
|
|
|
}
|
|
|
|
|
2015-10-27 19:54:15 +03:00
|
|
|
void System::LoadWeights()
|
2015-10-24 14:39:15 +03:00
|
|
|
{
|
2015-11-11 20:31:05 +03:00
|
|
|
const PARAM_VEC *vec = params.GetParam("weight");
|
2015-11-05 14:19:37 +03:00
|
|
|
UTIL_THROW_IF2(vec == NULL, "Must have [weight] section");
|
2015-10-27 18:46:37 +03:00
|
|
|
|
2015-11-05 14:19:37 +03:00
|
|
|
weights.Init(featureFunctions);
|
2016-03-31 23:00:16 +03:00
|
|
|
BOOST_FOREACH(const std::string &line, *vec){
|
|
|
|
cerr << "line=" << line << endl;
|
|
|
|
weights.CreateFromString(featureFunctions, line);
|
|
|
|
}
|
2015-10-27 19:54:15 +03:00
|
|
|
}
|
2015-10-27 18:46:37 +03:00
|
|
|
|
2015-10-31 05:45:01 +03:00
|
|
|
void System::LoadMappings()
|
|
|
|
{
|
2015-11-13 13:40:55 +03:00
|
|
|
const PARAM_VEC *vec = params.GetParam("mapping");
|
2015-11-05 14:19:37 +03:00
|
|
|
UTIL_THROW_IF2(vec == NULL, "Must have [mapping] section");
|
2015-10-31 05:45:01 +03:00
|
|
|
|
2016-03-31 23:00:16 +03:00
|
|
|
BOOST_FOREACH(const std::string &line, *vec){
|
|
|
|
vector<string> toks = Tokenize(line);
|
|
|
|
assert( (toks.size() == 2 && toks[0] == "T") || (toks.size() == 3 && toks[1] == "T") );
|
|
|
|
|
|
|
|
size_t ptInd;
|
|
|
|
if (toks.size() == 2) {
|
|
|
|
ptInd = Scan<size_t>(toks[1]);
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
ptInd = Scan<size_t>(toks[2]);
|
2015-10-31 05:45:01 +03:00
|
|
|
}
|
2016-03-31 23:00:16 +03:00
|
|
|
const PhraseTable *pt = featureFunctions.GetPhraseTablesExcludeUnknownWordPenalty(ptInd);
|
|
|
|
mappings.push_back(pt);
|
|
|
|
}
|
2015-10-27 18:46:37 +03:00
|
|
|
|
2016-03-31 23:00:16 +03:00
|
|
|
// unk pt
|
|
|
|
const UnknownWordPenalty *unkWP =
|
|
|
|
dynamic_cast<const UnknownWordPenalty*>(featureFunctions.FindFeatureFunction(
|
|
|
|
"UnknownWordPenalty0"));
|
2015-12-07 21:16:51 +03:00
|
|
|
if (unkWP) {
|
2016-03-31 23:00:16 +03:00
|
|
|
mappings.push_back(unkWP);
|
2015-12-07 21:16:51 +03:00
|
|
|
}
|
2015-10-31 05:45:01 +03:00
|
|
|
}
|
2015-10-24 15:19:42 +03:00
|
|
|
|
2016-01-03 00:44:44 +03:00
|
|
|
MemPool &System::GetSystemPool() const
|
|
|
|
{
|
|
|
|
MemPool &ret = GetThreadSpecificObj(m_systemPool);
|
|
|
|
return ret;
|
|
|
|
}
|
|
|
|
|
2015-11-10 14:28:17 +03:00
|
|
|
MemPool &System::GetManagerPool() const
|
|
|
|
{
|
2015-11-24 14:24:09 +03:00
|
|
|
MemPool &ret = GetThreadSpecificObj(m_managerPool);
|
|
|
|
return ret;
|
2015-11-10 14:28:17 +03:00
|
|
|
}
|
|
|
|
|
2016-03-31 23:00:16 +03:00
|
|
|
void System::ini_performance_options()
|
2015-11-10 15:46:26 +03:00
|
|
|
{
|
2016-03-31 23:00:16 +03:00
|
|
|
const PARAM_VEC *paramsVec;
|
|
|
|
// m_parameter->SetParameter<size_t>(m_timeout_threshold, "time-out", -1);
|
|
|
|
// m_timeout = (GetTimeoutThreshold() == (size_t)-1) ? false : true;
|
|
|
|
|
|
|
|
numThreads = 1;
|
|
|
|
paramsVec = params.GetParam("threads");
|
|
|
|
if (paramsVec && paramsVec->size()) {
|
|
|
|
if (paramsVec->at(0) == "all") {
|
|
|
|
#ifdef WITH_THREADS
|
|
|
|
numThreads = boost::thread::hardware_concurrency();
|
|
|
|
if (!numThreads) {
|
|
|
|
std::cerr
|
|
|
|
<< "-threads all specified but Boost doesn't know how many cores there are";
|
|
|
|
throw;
|
|
|
|
}
|
|
|
|
#else
|
|
|
|
std::cerr << "-threads all specified but moses not built with thread support";
|
|
|
|
return false;
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
else {
|
|
|
|
numThreads = Scan<int>(paramsVec->at(0));
|
|
|
|
if (numThreads < 1) {
|
|
|
|
std::cerr << "Specify at least one thread.";
|
|
|
|
throw;
|
|
|
|
}
|
|
|
|
#ifndef WITH_THREADS
|
|
|
|
if (numThreads > 1) {
|
|
|
|
std::cerr << "Error: Thread count of " << params->at(0)
|
|
|
|
<< " but moses not built with thread support";
|
|
|
|
throw
|
|
|
|
}
|
|
|
|
#endif
|
|
|
|
}
|
|
|
|
}
|
|
|
|
return;
|
2015-11-10 15:46:26 +03:00
|
|
|
}
|
2015-11-18 18:33:42 +03:00
|
|
|
|
|
|
|
FactorCollection &System::GetVocab() const
|
|
|
|
{
|
2016-03-31 23:00:16 +03:00
|
|
|
return m_vocab;
|
2015-11-18 18:33:42 +03:00
|
|
|
}
|
2015-11-20 03:49:48 +03:00
|
|
|
|
2016-02-29 13:55:24 +03:00
|
|
|
Recycler<HypothesisBase*> &System::GetHypoRecycler() const
|
2016-02-26 16:15:27 +03:00
|
|
|
{
|
2016-03-31 23:00:16 +03:00
|
|
|
return GetThreadSpecificObj(m_hypoRecycler);
|
2016-02-26 16:15:27 +03:00
|
|
|
}
|
|
|
|
|
2015-12-10 23:49:30 +03:00
|
|
|
}
|
|
|
|
|