| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
|
|
| #include "SentenceAlignmentWithSyntax.h" |
|
|
| #include <map> |
| #include <set> |
| #include <string> |
|
|
| #include "tables-core.h" |
| #include "XmlException.h" |
| #include "XmlTree.h" |
| #include "util/tokenize.hh" |
|
|
| using namespace std; |
|
|
| namespace MosesTraining |
| { |
|
|
| bool SentenceAlignmentWithSyntax::processTargetSentence(const char * targetString, int sentenceID, bool boundaryRules) |
| { |
| if (!m_targetSyntax) { |
| return SentenceAlignment::processTargetSentence(targetString, sentenceID, boundaryRules); |
| } |
|
|
| string targetStringCPP(targetString); |
| try { |
| ProcessAndStripXMLTags(targetStringCPP, targetTree, |
| m_targetLabelCollection, |
| m_targetTopLabelCollection, |
| false); |
| } catch (const XmlException & e) { |
| std::cerr << "WARNING: failed to process target sentence at line " |
| << sentenceID << ": " << e.getMsg() << std::endl; |
| return false; |
| } |
| target = util::tokenize(targetStringCPP); |
| return true; |
| } |
|
|
| bool SentenceAlignmentWithSyntax::processSourceSentence(const char * sourceString, int sentenceID, bool boundaryRules) |
| { |
| if (!m_sourceSyntax) { |
| return SentenceAlignment::processSourceSentence(sourceString, sentenceID, boundaryRules); |
| } |
|
|
| string sourceStringCPP(sourceString); |
| try { |
| ProcessAndStripXMLTags(sourceStringCPP, sourceTree, |
| m_sourceLabelCollection , |
| m_sourceTopLabelCollection, |
| false); |
| } catch (const XmlException & e) { |
| std::cerr << "WARNING: failed to process source sentence at line " |
| << sentenceID << ": " << e.getMsg() << std::endl; |
| return false; |
| } |
| source = util::tokenize(sourceStringCPP); |
| return true; |
| } |
|
|
| } |
|
|
