public ParallelTextCorpus(ITextCorpus sourceCorpus, ITextCorpus targetCorpus,
ITextAlignmentCorpus textAlignmentCorpus = null, IComparer <object> segmentRefComparer = null)
{
SourceCorpus = sourceCorpus;
TargetCorpus = targetCorpus;
TextAlignmentCorpus = textAlignmentCorpus;
_segmentRefComparer = segmentRefComparer;
}
private void CloseProject()
{
CurrentText = null;
_texts.Clear();
CurrentText = new TextViewModel(_tokenizer);
_sourceCorpus = null;
_targetCorpus = null;
if (_hybridEngine != null)
{
_hybridEngine.Dispose();
_hybridEngine = null;
}
if (_smtModel != null)
{
_smtModel.Dispose();
_smtModel = null;
}
_saveProjectCommand.UpdateCanExecute();
RebuildTask.UpdateCanExecute();
}
private bool LoadProject(string fileName)
{
XElement projectElem;
try
{
projectElem = XElement.Load(fileName);
}
catch (Exception)
{
return(false);
}
XElement engineElem = projectElem.Element("TranslationEngine");
if (engineElem == null)
{
return(false);
}
var smtConfig = (string)engineElem.Element("SmtConfig");
if (smtConfig == null)
{
return(false);
}
var hcSrcConfig = (string)engineElem.Element("SourceAnalyzerConfig");
var hcTrgConfig = (string)engineElem.Element("TargetGeneratorConfig");
string configDir = Path.GetDirectoryName(fileName);
Debug.Assert(configDir != null);
ITranslationEngine transferEngine = null;
if (hcSrcConfig != null && hcTrgConfig != null)
{
Language srcLang = XmlLanguageLoader.Load(Path.Combine(configDir, hcSrcConfig));
var srcMorpher = new Morpher(_hcTraceManager, srcLang);
Language trgLang = XmlLanguageLoader.Load(Path.Combine(configDir, hcTrgConfig));
var trgMorpher = new Morpher(_hcTraceManager, trgLang);
transferEngine = new TransferEngine(srcMorpher,
new SimpleTransferer(new GlossMorphemeMapper(trgMorpher)), trgMorpher);
}
_smtModel = new ThotSmtModel(Path.Combine(configDir, smtConfig));
IInteractiveSmtEngine smtEngine = _smtModel.CreateInteractiveEngine();
_hybridEngine = new HybridTranslationEngine(smtEngine, transferEngine);
var sourceTexts = new List <IText>();
var targetTexts = new List <IText>();
var alignmentCollections = new List <ITextAlignmentCollection>();
using (_texts.BulkUpdate())
{
foreach (XElement textElem in projectElem.Elements("Texts").Elements("Text"))
{
var name = (string)textElem.Attribute("name");
var metadataFileName = (string)textElem.Element("MetadataFile");
if (metadataFileName == null)
{
return(false);
}
metadataFileName = Path.Combine(configDir, metadataFileName);
var srcTextFileName = (string)textElem.Element("SourceFile");
if (srcTextFileName == null)
{
return(false);
}
srcTextFileName = Path.Combine(configDir, srcTextFileName);
var trgTextFileName = (string)textElem.Element("TargetFile");
if (trgTextFileName == null)
{
return(false);
}
trgTextFileName = Path.Combine(configDir, trgTextFileName);
var alignmentsFileName = (string)textElem.Element("AlignmentsFile");
if (alignmentsFileName != null)
{
alignmentsFileName = Path.Combine(configDir, alignmentsFileName);
}
var text = new TextViewModel(_tokenizer, name, metadataFileName, srcTextFileName, trgTextFileName,
alignmentsFileName)
{
Engine = _hybridEngine
};
text.PropertyChanged += TextPropertyChanged;
_texts.Add(text);
Func <TextSegment, bool> segmentFilter = s => text.IsApproved((TextSegmentRef)s.SegmentRef);
sourceTexts.Add(new FilteredText(new TextFileText(_tokenizer, name, srcTextFileName),
segmentFilter));
targetTexts.Add(new FilteredText(new TextFileText(_tokenizer, name, trgTextFileName),
segmentFilter));
if (alignmentsFileName != null)
{
alignmentCollections.Add(new TextFileTextAlignmentCollection(name, alignmentsFileName));
}
}
}
if (_texts.Count == 0)
{
return(false);
}
_sourceCorpus = new DictionaryTextCorpus(sourceTexts);
_targetCorpus = new DictionaryTextCorpus(targetTexts);
_alignmentCorpus = new DictionaryTextAlignmentCorpus(alignmentCollections);
CurrentText = _texts[0];
AcceptChanges();
RebuildTask.UpdateCanExecute();
return(true);
}
public FilteredTextCorpus(ITextCorpus corpus, Func <IText, bool> filter)
{
_corpus = corpus;
_filter = filter;
}
public ISmtBatchTrainer CreateBatchTrainer(Func <string, string> sourcePreprocessor, ITextCorpus sourceCorpus,
Func <string, string> targetPreprocessor, ITextCorpus targetCorpus,
ITextAlignmentCorpus alignmentCorpus = null)
{
CheckDisposed();
var corpus = new ParallelTextCorpus(sourceCorpus, targetCorpus, alignmentCorpus);
return(string.IsNullOrEmpty(ConfigFileName)
? new BatchTrainer(this, Parameters, sourcePreprocessor, targetPreprocessor, corpus)
: new BatchTrainer(this, ConfigFileName, sourcePreprocessor, targetPreprocessor, corpus));
}