mirror of
https://github.com/moses-smt/mosesdecoder.git
synced 2024-07-14 23:00:29 +03:00
daily automatic beautifier
This commit is contained in:
parent
a950605795
commit
739165c2b0
@ -47,21 +47,21 @@ ChartParserUnknown::~ChartParserUnknown()
|
||||
// RemoveAllInColl(m_cacheTargetPhraseCollection);
|
||||
}
|
||||
|
||||
AllOptions::ptr const&
|
||||
AllOptions::ptr const&
|
||||
ChartParserUnknown::
|
||||
options() const
|
||||
{
|
||||
return m_ttask.lock()->options();
|
||||
}
|
||||
options() const
|
||||
{
|
||||
return m_ttask.lock()->options();
|
||||
}
|
||||
|
||||
void
|
||||
void
|
||||
ChartParserUnknown::
|
||||
Process(const Word &sourceWord, const Range &range, ChartParserCallback &to)
|
||||
{
|
||||
// unknown word, add as trans opt
|
||||
const StaticData &staticData = StaticData::Instance();
|
||||
const UnknownWordPenaltyProducer &unknownWordPenaltyProducer
|
||||
= UnknownWordPenaltyProducer::Instance();
|
||||
const UnknownWordPenaltyProducer &unknownWordPenaltyProducer
|
||||
= UnknownWordPenaltyProducer::Instance();
|
||||
|
||||
size_t isDigit = 0;
|
||||
if (options()->unk.drop) {
|
||||
@ -101,8 +101,8 @@ Process(const Word &sourceWord, const Range &range, ChartParserCallback &to)
|
||||
//const Word &sourceLHS = staticData.GetInputDefaultNonTerminal();
|
||||
Word *targetLHS = new Word(true);
|
||||
|
||||
targetLHS->CreateFromString(Output, options()->output.factor_order,
|
||||
targetLHSStr, true);
|
||||
targetLHS->CreateFromString(Output, options()->output.factor_order,
|
||||
targetLHSStr, true);
|
||||
UTIL_THROW_IF2(targetLHS->GetFactor(0) == NULL, "Null factor for target LHS");
|
||||
|
||||
// add to dictionary
|
||||
@ -142,8 +142,8 @@ Process(const Word &sourceWord, const Range &range, ChartParserCallback &to)
|
||||
//float prob = iterLHS->second;
|
||||
|
||||
Word *targetLHS = new Word(true);
|
||||
targetLHS->CreateFromString(Output, staticData.options().output.factor_order,
|
||||
targetLHSStr, true);
|
||||
targetLHS->CreateFromString(Output, staticData.options().output.factor_order,
|
||||
targetLHSStr, true);
|
||||
UTIL_THROW_IF2(targetLHS->GetFactor(0) == NULL, "Null factor for target LHS");
|
||||
|
||||
targetPhrase->GetScoreBreakdown().Assign(&unknownWordPenaltyProducer, unknownScore);
|
||||
@ -300,12 +300,12 @@ long ChartParser::GetTranslationId() const
|
||||
}
|
||||
|
||||
|
||||
AllOptions::ptr const&
|
||||
AllOptions::ptr const&
|
||||
ChartParser::
|
||||
options() const
|
||||
{
|
||||
return m_ttask.lock()->options();
|
||||
}
|
||||
options() const
|
||||
{
|
||||
return m_ttask.lock()->options();
|
||||
}
|
||||
|
||||
|
||||
} // namespace Moses
|
||||
|
@ -57,7 +57,7 @@ public:
|
||||
private:
|
||||
std::vector<Phrase*> m_unksrcs;
|
||||
std::list<TargetPhraseCollection::shared_ptr> m_cacheTargetPhraseCollection;
|
||||
AllOptions::ptr const& options() const;
|
||||
AllOptions::ptr const& options() const;
|
||||
};
|
||||
|
||||
class ChartParser
|
||||
@ -79,7 +79,7 @@ public:
|
||||
return m_unknown.GetUnknownSources();
|
||||
}
|
||||
|
||||
AllOptions::ptr const& options() const;
|
||||
AllOptions::ptr const& options() const;
|
||||
|
||||
private:
|
||||
ChartParserUnknown m_unknown;
|
||||
|
@ -46,8 +46,8 @@ public:
|
||||
|
||||
const Column& GetColumn(size_t i) const {
|
||||
UTIL_THROW_IF2(i >= data.size(),
|
||||
"Out of bounds. Trying to access " << i
|
||||
<< " when vector only contains " << data.size());
|
||||
"Out of bounds. Trying to access " << i
|
||||
<< " when vector only contains " << data.size());
|
||||
return data[i];
|
||||
}
|
||||
const Column& operator[](size_t i) const {
|
||||
|
@ -569,8 +569,8 @@ void SoftSourceSyntacticConstraintsFeature::EvaluateWithSourceContext(const Inpu
|
||||
}
|
||||
if ( treeInputLabelsLHS.size() == 0 ) {
|
||||
scoreBreakdown.PlusEquals(this,
|
||||
"LHSPAIR_" + targetLHS->GetString().as_string() + "_"
|
||||
+ m_output_default_nonterminal[0]->GetString().as_string(),
|
||||
"LHSPAIR_" + targetLHS->GetString().as_string() + "_"
|
||||
+ m_output_default_nonterminal[0]->GetString().as_string(),
|
||||
1);
|
||||
if (!m_targetSourceLHSJointCountFile.empty()) {
|
||||
t2sLabelsScore = TransformScore(m_floor);
|
||||
|
@ -48,10 +48,10 @@ void SourceGHKMTreeInputMatchFeature::EvaluateWithSourceContext(const InputType
|
||||
|
||||
const StaticData& staticData = StaticData::Instance();
|
||||
|
||||
std::vector<float> newScores(m_numScoreComponents,0.0);
|
||||
std::vector<float> newScores(m_numScoreComponents,0.0);
|
||||
// m_numScoreComponents == 2 // first fires for matches, second for mismatches
|
||||
|
||||
if ( (treeInputLabels.find(lhsLabel) != treeInputLabels.end())
|
||||
if ( (treeInputLabels.find(lhsLabel) != treeInputLabels.end())
|
||||
&& (lhsLabel != m_output_default_nonterminal) ) {
|
||||
// match
|
||||
newScores[0] = 1.0;
|
||||
|
@ -68,7 +68,7 @@ Read(std::istream &in)
|
||||
assert(topVertices.size() >= 1);
|
||||
}
|
||||
|
||||
|
||||
|
||||
const std::vector<FactorType>& factorOrder = m_options->input.factor_order;
|
||||
|
||||
// Add <s> vertex.
|
||||
@ -175,7 +175,7 @@ void ForestInput::FindTopVertices(Forest &forest,
|
||||
std::back_inserter(topVertices));
|
||||
}
|
||||
|
||||
void
|
||||
void
|
||||
ForestInput::
|
||||
ParseHyperedgeLine(const std::string &line)
|
||||
{
|
||||
|
@ -74,13 +74,12 @@ pair<HypothesisStackNormal::iterator, bool> HypothesisStackNormal::Add(Hypothesi
|
||||
// prune only if stack is twice as big as needed (lazy pruning)
|
||||
size_t toleratedSize = 2*m_maxHypoStackSize-1;
|
||||
// add in room for stack diversity
|
||||
if (m_minHypoStackDiversity)
|
||||
{
|
||||
// so what happens if maxdistortion is negative?
|
||||
toleratedSize += m_minHypoStackDiversity
|
||||
<< m_manager.options()->reordering.max_distortion;
|
||||
}
|
||||
|
||||
if (m_minHypoStackDiversity) {
|
||||
// so what happens if maxdistortion is negative?
|
||||
toleratedSize += m_minHypoStackDiversity
|
||||
<< m_manager.options()->reordering.max_distortion;
|
||||
}
|
||||
|
||||
if (m_hypos.size() > toleratedSize) {
|
||||
PruneToSize(m_maxHypoStackSize);
|
||||
} else {
|
||||
|
@ -29,7 +29,7 @@ Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
|
||||
namespace Moses
|
||||
{
|
||||
|
||||
InputType::InputType(AllOptions::ptr const& opts, long translationId)
|
||||
InputType::InputType(AllOptions::ptr const& opts, long translationId)
|
||||
: m_options(opts)
|
||||
, m_translationId(translationId)
|
||||
{
|
||||
|
@ -190,10 +190,10 @@ public:
|
||||
|
||||
//! populate this InputType with data from in stream
|
||||
virtual int
|
||||
Read(std::istream& in) = 0;
|
||||
// ,
|
||||
// std::vector<FactorType> const& factorOrder,
|
||||
// AllOptions const& opts) =0;
|
||||
Read(std::istream& in) = 0;
|
||||
// ,
|
||||
// std::vector<FactorType> const& factorOrder,
|
||||
// AllOptions const& opts) =0;
|
||||
|
||||
//! Output debugging info to stream out
|
||||
virtual void Print(std::ostream&) const =0;
|
||||
|
@ -64,8 +64,8 @@ MockHypothesisGuard
|
||||
m_targetPhrases.back().CreateFromString(Input, factors, *ti, NULL);
|
||||
m_toptions.push_back(new TranslationOption
|
||||
(range,m_targetPhrases.back()));
|
||||
m_hypothesis = new Hypothesis(*prevHypo, *m_toptions.back(), newBitmap,
|
||||
m_manager->GetNextHypoId());
|
||||
m_hypothesis = new Hypothesis(*prevHypo, *m_toptions.back(), newBitmap,
|
||||
m_manager->GetNextHypoId());
|
||||
}
|
||||
|
||||
|
||||
|
@ -151,11 +151,11 @@ aux_interpret_xml(std::string& line, std::vector<size_t> & xmlWalls,
|
||||
// parse XML markup in translation line
|
||||
using namespace std;
|
||||
if (m_options->input.xml_policy != XmlPassThrough) {
|
||||
bool OK = ProcessAndStripXMLTags(*m_options, line,
|
||||
m_xmlOptions,
|
||||
bool OK = ProcessAndStripXMLTags(*m_options, line,
|
||||
m_xmlOptions,
|
||||
m_reorderingConstraint,
|
||||
xmlWalls, placeholders);
|
||||
UTIL_THROW_IF2(!OK, "Unable to parse XML in line: " << line);
|
||||
UTIL_THROW_IF2(!OK, "Unable to parse XML in line: " << line);
|
||||
}
|
||||
}
|
||||
|
||||
@ -170,7 +170,7 @@ init(string line)
|
||||
|
||||
if (m_options->input.continue_partial_translation)
|
||||
aux_init_partial_translation(line);
|
||||
|
||||
|
||||
line = Trim(line);
|
||||
aux_interpret_sgml_markup(line); // for "<seg id=..." markup
|
||||
aux_interpret_dlt(line); // some poorly documented cache-based stuff
|
||||
@ -365,7 +365,7 @@ CreateFromString(vector<FactorType> const& FOrder, string const& phraseString)
|
||||
}
|
||||
|
||||
Sentence::
|
||||
Sentence(AllOptions::ptr const& opts, size_t const transId, string stext)
|
||||
Sentence(AllOptions::ptr const& opts, size_t const transId, string stext)
|
||||
: InputType(opts, transId)
|
||||
{
|
||||
init(stext);
|
||||
|
@ -115,7 +115,7 @@ public:
|
||||
}
|
||||
|
||||
|
||||
void init(std::string line);
|
||||
void init(std::string line);
|
||||
|
||||
std::vector<std::map<std::string,std::string> > const&
|
||||
GetDltMeta() const {
|
||||
|
@ -9,7 +9,7 @@
|
||||
|
||||
namespace Moses
|
||||
{
|
||||
class AllOptions;
|
||||
class AllOptions;
|
||||
namespace Syntax
|
||||
{
|
||||
namespace F2S
|
||||
@ -20,8 +20,8 @@ class GlueRuleSynthesizer : public HyperTreeCreator
|
||||
Word m_input_default_nonterminal;
|
||||
Word m_output_default_nonterminal;
|
||||
public:
|
||||
GlueRuleSynthesizer(Moses::AllOptions const& opts, HyperTree &);
|
||||
|
||||
GlueRuleSynthesizer(Moses::AllOptions const& opts, HyperTree &);
|
||||
|
||||
// Synthesize the minimal, monotone rule that can be applied to the given
|
||||
// hyperedge and add it to the rule trie.
|
||||
void SynthesizeRule(const Forest::Hyperedge &);
|
||||
|
@ -12,7 +12,7 @@ namespace Syntax
|
||||
namespace T2S
|
||||
{
|
||||
|
||||
void
|
||||
void
|
||||
GlueRuleSynthesizer::
|
||||
SynthesizeRule(const InputTree::Node &node)
|
||||
{
|
||||
|
@ -18,11 +18,11 @@ class GlueRuleSynthesizer : public RuleTrieCreator
|
||||
{
|
||||
Word m_output_default_nonterminal;
|
||||
public:
|
||||
GlueRuleSynthesizer(RuleTrie &trie, Word dflt_nonterm)
|
||||
GlueRuleSynthesizer(RuleTrie &trie, Word dflt_nonterm)
|
||||
: m_ruleTrie(trie)
|
||||
, m_output_default_nonterminal(dflt_nonterm)
|
||||
{}
|
||||
|
||||
|
||||
// Synthesize the minimal, montone rule that can be applied to the given node
|
||||
// and add it to the rule trie.
|
||||
void SynthesizeRule(const InputTree::Node &);
|
||||
|
@ -77,7 +77,7 @@ TranslationTask
|
||||
boost::shared_ptr<IOWrapper> const& ioWrapper)
|
||||
: m_source(source) , m_ioWrapper(ioWrapper)
|
||||
{
|
||||
m_options = source->options();
|
||||
m_options = source->options();
|
||||
}
|
||||
|
||||
TranslationTask::~TranslationTask()
|
||||
@ -104,7 +104,7 @@ TranslationTask
|
||||
|
||||
else if (algo == SyntaxS2T) {
|
||||
// new-style string-to-tree decoding (ask Phil Williams)
|
||||
S2TParsingAlgorithm algorithm = m_options->syntax.s2t_parsing_algo;
|
||||
S2TParsingAlgorithm algorithm = m_options->syntax.s2t_parsing_algo;
|
||||
if (algorithm == RecursiveCYKPlus) {
|
||||
typedef Syntax::S2T::EagerParserCallback Callback;
|
||||
typedef Syntax::S2T::RecursiveCYKPlusParser<Callback> Parser;
|
||||
|
@ -245,7 +245,7 @@ Read(std::istream& in)
|
||||
string line;
|
||||
if (getline(in, line, '\n').eof())
|
||||
return 0;
|
||||
|
||||
|
||||
m_labelledSpans.clear();
|
||||
ProcessAndStripXMLTags(*m_options, line, m_labelledSpans, m_xmlOptions);
|
||||
|
||||
@ -253,7 +253,7 @@ Read(std::istream& in)
|
||||
stringstream strme;
|
||||
strme << line << endl;
|
||||
|
||||
Sentence::Read(strme);
|
||||
Sentence::Read(strme);
|
||||
|
||||
// size input chart
|
||||
size_t sourceSize = GetSize();
|
||||
@ -265,7 +265,7 @@ Read(std::istream& in)
|
||||
|
||||
// do source labels
|
||||
vector<XMLParseOutput>::const_iterator iterLabel;
|
||||
for (iterLabel = m_labelledSpans.begin();
|
||||
for (iterLabel = m_labelledSpans.begin();
|
||||
iterLabel != m_labelledSpans.end(); ++iterLabel) {
|
||||
const XMLParseOutput &labelItem = *iterLabel;
|
||||
const Range &range = labelItem.m_range;
|
||||
@ -279,7 +279,7 @@ Read(std::istream& in)
|
||||
for (size_t endPos = startPos; endPos < sourceSize; ++endPos) {
|
||||
NonTerminalSet &list = GetLabelSet(startPos, endPos);
|
||||
if (list.size() == 0 || ! only4empty ) {
|
||||
AddChartLabel(startPos, endPos, m_options->syntax.input_default_non_terminal);
|
||||
AddChartLabel(startPos, endPos, m_options->syntax.input_default_non_terminal);
|
||||
}
|
||||
}
|
||||
}
|
||||
@ -300,7 +300,7 @@ TranslationOptionCollection* TreeInput::CreateTranslationOptionCollection() cons
|
||||
return NULL;
|
||||
}
|
||||
|
||||
void
|
||||
void
|
||||
TreeInput::
|
||||
AddChartLabel(size_t startPos, size_t endPos, const Word &label)
|
||||
{
|
||||
@ -324,14 +324,14 @@ AddChartLabel(size_t startPos, size_t endPos, const Word &label)
|
||||
}
|
||||
}
|
||||
|
||||
void
|
||||
void
|
||||
TreeInput::
|
||||
AddChartLabel(size_t startPos, size_t endPos, const string &label)
|
||||
{
|
||||
const std::vector<FactorType>& fOrder = m_options->input.factor_order;
|
||||
Word word(true);
|
||||
const Factor *factor
|
||||
= FactorCollection::Instance().AddFactor(Input, fOrder[0], label, true);
|
||||
const Factor *factor
|
||||
= FactorCollection::Instance().AddFactor(Input, fOrder[0], label, true);
|
||||
// TODO - no factors
|
||||
word.SetFactor(0, factor);
|
||||
AddChartLabel(startPos, endPos, word);
|
||||
|
@ -159,7 +159,7 @@ vector<string> TokenizeXml(const string& str, const std::string& lbrackStr, cons
|
||||
* \param rbrackStr xml tag's right bracket string, typically ">"
|
||||
*/
|
||||
bool
|
||||
ProcessAndStripXMLTags(AllOptions const& opts, string &line,
|
||||
ProcessAndStripXMLTags(AllOptions const& opts, string &line,
|
||||
vector<XmlOption const*> &res,
|
||||
ReorderingConstraint &reorderingConstraint,
|
||||
vector< size_t > &walls,
|
||||
|
@ -32,8 +32,8 @@ std::vector<std::string> TokenizeXml(const std::string& str, const std::string&
|
||||
|
||||
bool ProcessAndStripXMLTags(AllOptions const& opts,
|
||||
std::string &line, std::vector<XmlOption const*> &res,
|
||||
ReorderingConstraint &reorderingConstraint,
|
||||
std::vector< size_t > &walls,
|
||||
ReorderingConstraint &reorderingConstraint,
|
||||
std::vector< size_t > &walls,
|
||||
std::vector< std::pair<size_t, std::string> > &placeholders);
|
||||
|
||||
|
||||
|
@ -70,7 +70,7 @@ public:
|
||||
, startS(sS)
|
||||
, endS(eS)
|
||||
, count(0)
|
||||
, pcfgScore(0.0)
|
||||
, pcfgScore(0.0)
|
||||
, l2rOrientation(PhraseOrientation::REO_CLASS_UNKNOWN)
|
||||
, r2lOrientation(PhraseOrientation::REO_CLASS_UNKNOWN)
|
||||
{ }
|
||||
|
@ -1210,7 +1210,8 @@ void collectWordLabelCounts( SentenceAlignmentWithSyntax &sentence )
|
||||
}
|
||||
}
|
||||
|
||||
void writeUnknownWordLabel(const string & fileName) {
|
||||
void writeUnknownWordLabel(const string & fileName)
|
||||
{
|
||||
ofstream outFile;
|
||||
outFile.open(fileName.c_str());
|
||||
typedef map<string,int>::const_iterator I;
|
||||
@ -1234,7 +1235,7 @@ void writeUnknownWordLabel(const string & fileName) {
|
||||
outFile.close();
|
||||
}
|
||||
|
||||
void writePhraseOrientationPriors(const string &fileName)
|
||||
void writePhraseOrientationPriors(const string &fileName)
|
||||
{
|
||||
ofstream outFile;
|
||||
outFile.open(fileName.c_str());
|
||||
|
Loading…
Reference in New Issue
Block a user