void ScfgRuleWriter::Write(const ScfgRule &rule)
{
std::ostringstream sourceSS;
std::ostringstream targetSS;
if (m_options.unpairedExtractFormat) {
WriteUnpairedFormat(rule, sourceSS, targetSS);
} else {
WriteStandardFormat(rule, sourceSS, targetSS);
}
// Write the rule to the forward and inverse extract files.
m_fwd << sourceSS.str() << " ||| " << targetSS.str() << " |||";
m_inv << targetSS.str() << " ||| " << sourceSS.str() << " |||";
const Alignment &alignment = rule.GetAlignment();
for (Alignment::const_iterator p = alignment.begin();
p != alignment.end(); ++p) {
m_fwd << " " << p->first << "-" << p->second;
m_inv << " " << p->second << "-" << p->first;
}
// Write a count of 1 and an empty NT length column to the forward extract
// file.
// TODO Add option to write NT length?
m_fwd << " ||| 1 ||| |||";
if (m_options.pcfg) {
// Write the PCFG score.
m_fwd << " " << std::exp(rule.GetPcfgScore());
}
m_fwd << std::endl;
// Write a count of 1 to the inverse extract file.
m_inv << " ||| 1" << std::endl;
}
void ScfgRuleWriter::WriteStandardFormat(const ScfgRule &rule)
{
const std::vector<Symbol> &sourceRHS = rule.GetSourceRHS();
const std::vector<Symbol> &targetRHS = rule.GetTargetRHS();
std::map<int, int> sourceToTargetNTMap;
std::map<int, int> targetToSourceNTMap;
const Alignment &alignment = rule.GetAlignment();
for (Alignment::const_iterator p(alignment.begin());
p != alignment.end(); ++p) {
if (sourceRHS[p->first].GetType() == NonTerminal) {
assert(targetRHS[p->second].GetType() == NonTerminal);
sourceToTargetNTMap[p->first] = p->second;
targetToSourceNTMap[p->second] = p->first;
}
}
std::ostringstream sourceSS;
std::ostringstream targetSS;
// Write the source side of the rule to sourceSS.
int i = 0;
for (std::vector<Symbol>::const_iterator p(sourceRHS.begin());
p != sourceRHS.end(); ++p, ++i) {
WriteSymbol(*p, sourceSS);
if (p->GetType() == NonTerminal) {
int targetIndex = sourceToTargetNTMap[i];
WriteSymbol(targetRHS[targetIndex], sourceSS);
}
sourceSS << " ";
}
WriteSymbol(rule.GetSourceLHS(), sourceSS);
// Write the target side of the rule to targetSS.
i = 0;
for (std::vector<Symbol>::const_iterator p(targetRHS.begin());
p != targetRHS.end(); ++p, ++i) {
if (p->GetType() == NonTerminal) {
int sourceIndex = targetToSourceNTMap[i];
WriteSymbol(sourceRHS[sourceIndex], targetSS);
}
WriteSymbol(*p, targetSS);
targetSS << " ";
}
WriteSymbol(rule.GetTargetLHS(), targetSS);
// Write the rule to the forward and inverse extract files.
m_fwd << sourceSS.str() << " ||| " << targetSS.str() << " |||";
m_inv << targetSS.str() << " ||| " << sourceSS.str() << " |||";
for (Alignment::const_iterator p(alignment.begin());
p != alignment.end(); ++p) {
m_fwd << " " << p->first << "-" << p->second;
m_inv << " " << p->second << "-" << p->first;
}
m_fwd << " ||| 1" << std::endl;
m_inv << " ||| 1" << std::endl;
}
void ScfgRuleWriter::WriteStandardFormat(const ScfgRule &rule,
std::ostream &sourceSS,
std::ostream &targetSS)
{
const std::vector<Symbol> &sourceRHS = rule.GetSourceRHS();
const std::vector<Symbol> &targetRHS = rule.GetTargetRHS();
std::map<int, int> sourceToTargetNTMap;
std::map<int, int> targetToSourceNTMap;
const Alignment &alignment = rule.GetAlignment();
for (Alignment::const_iterator p(alignment.begin());
p != alignment.end(); ++p) {
if (sourceRHS[p->first].GetType() == NonTerminal) {
assert(targetRHS[p->second].GetType() == NonTerminal);
sourceToTargetNTMap[p->first] = p->second;
targetToSourceNTMap[p->second] = p->first;
}
}
// Write the source side of the rule to sourceSS.
int i = 0;
for (std::vector<Symbol>::const_iterator p(sourceRHS.begin());
p != sourceRHS.end(); ++p, ++i) {
WriteSymbol(*p, sourceSS);
if (p->GetType() == NonTerminal) {
int targetIndex = sourceToTargetNTMap[i];
WriteSymbol(targetRHS[targetIndex], sourceSS);
}
sourceSS << " ";
}
if (m_options.conditionOnTargetLhs) {
WriteSymbol(rule.GetTargetLHS(), sourceSS);
} else {
WriteSymbol(rule.GetSourceLHS(), sourceSS);
}
// Write the target side of the rule to targetSS.
i = 0;
for (std::vector<Symbol>::const_iterator p(targetRHS.begin());
p != targetRHS.end(); ++p, ++i) {
if (p->GetType() == NonTerminal) {
int sourceIndex = targetToSourceNTMap[i];
WriteSymbol(sourceRHS[sourceIndex], targetSS);
}
WriteSymbol(*p, targetSS);
targetSS << " ";
}
WriteSymbol(rule.GetTargetLHS(), targetSS);
}
void ScfgRuleWriter::WriteUnpairedFormat(const ScfgRule &rule)
{
const std::vector<Symbol> &sourceRHS = rule.GetSourceRHS();
const std::vector<Symbol> &targetRHS = rule.GetTargetRHS();
const Alignment &alignment = rule.GetAlignment();
std::ostringstream sourceSS;
std::ostringstream targetSS;
// Write the source side of the rule to sourceSS.
int i = 0;
for (std::vector<Symbol>::const_iterator p(sourceRHS.begin());
p != sourceRHS.end(); ++p, ++i) {
WriteSymbol(*p, sourceSS);
sourceSS << " ";
}
WriteSymbol(rule.GetSourceLHS(), sourceSS);
// Write the target side of the rule to targetSS.
i = 0;
for (std::vector<Symbol>::const_iterator p(targetRHS.begin());
p != targetRHS.end(); ++p, ++i) {
WriteSymbol(*p, targetSS);
targetSS << " ";
}
WriteSymbol(rule.GetTargetLHS(), targetSS);
// Write the rule to the forward and inverse extract files.
m_fwd << sourceSS.str() << " ||| " << targetSS.str() << " |||";
m_inv << targetSS.str() << " ||| " << sourceSS.str() << " |||";
for (Alignment::const_iterator p(alignment.begin());
p != alignment.end(); ++p) {
m_fwd << " " << p->first << "-" << p->second;
m_inv << " " << p->second << "-" << p->first;
}
m_fwd << " ||| 1" << std::endl;
m_inv << " ||| 1" << std::endl;
}
void ScfgRuleWriter::Write(const ScfgRule &rule, bool printEndl)
{
std::ostringstream sourceSS;
std::ostringstream targetSS;
if (m_options.unpairedExtractFormat) {
WriteUnpairedFormat(rule, sourceSS, targetSS);
} else {
WriteStandardFormat(rule, sourceSS, targetSS);
}
// Write the rule to the forward and inverse extract files.
m_fwd << sourceSS.str() << " ||| " << targetSS.str() << " |||";
m_inv << targetSS.str() << " ||| " << sourceSS.str() << " |||";
const Alignment &alignment = rule.GetAlignment();
for (Alignment::const_iterator p = alignment.begin();
p != alignment.end(); ++p) {
m_fwd << " " << p->first << "-" << p->second;
m_inv << " " << p->second << "-" << p->first;
}
// Write a count of 1.
m_fwd << " ||| 1";
m_inv << " ||| 1";
// Write the PCFG score (if requested).
if (m_options.pcfg) {
m_fwd << " ||| " << std::exp(rule.GetPcfgScore());
}
if (printEndl) {
m_fwd << std::endl;
m_inv << std::endl;
}
}
void ScfgRuleWriter::Write(const ScfgRule &rule, size_t lineNum, bool printEndl)
{
std::ostringstream sourceSS;
std::ostringstream targetSS;
if (m_options.unpairedExtractFormat) {
WriteUnpairedFormat(rule, sourceSS, targetSS);
} else {
WriteStandardFormat(rule, sourceSS, targetSS);
}
// Write the rule to the forward and inverse extract files.
if (m_options.t2s) {
// If model is tree-to-string then flip the source and target.
m_fwd << targetSS.str() << " ||| " << sourceSS.str() << " |||";
m_inv << sourceSS.str() << " ||| " << targetSS.str() << " |||";
} else {
m_fwd << sourceSS.str() << " ||| " << targetSS.str() << " |||";
m_inv << targetSS.str() << " ||| " << sourceSS.str() << " |||";
}
const Alignment &alignment = rule.GetAlignment();
for (Alignment::const_iterator p = alignment.begin();
p != alignment.end(); ++p) {
if (m_options.t2s) {
// If model is tree-to-string then flip the source and target.
m_fwd << " " << p->second << "-" << p->first;
m_inv << " " << p->first << "-" << p->second;
} else {
m_fwd << " " << p->first << "-" << p->second;
m_inv << " " << p->second << "-" << p->first;
}
}
if (m_options.includeSentenceId) {
if (m_options.t2s) {
m_inv << " ||| " << lineNum;
} else {
m_fwd << " ||| " << lineNum;
}
}
// Write a count of 1.
m_fwd << " ||| 1";
m_inv << " ||| 1";
// Write the PCFG score (if requested).
if (m_options.pcfg) {
m_fwd << " ||| " << std::exp(rule.GetPcfgScore());
}
m_fwd << " |||";
if (m_options.sourceLabels && rule.HasSourceLabels()) {
m_fwd << " {{SourceLabels";
rule.PrintSourceLabels(m_fwd);
m_fwd << "}}";
}
if (printEndl) {
m_fwd << std::endl;
m_inv << std::endl;
}
}