/// <summary>
/// Initializes a new instance of the <see cref="SentenceDetectorME" /> using the given sentence model.
/// </summary>
/// <param name="sentenceModel">The sentence model.</param>
/// <exception cref="System.ArgumentNullException">
/// The <paramref name="sentenceModel"/> is <c>null</c>.
/// </exception>
public SentenceDetectorME(SentenceModel sentenceModel)
{
if (sentenceModel == null)
{
throw new ArgumentNullException(nameof(sentenceModel));
}
model = sentenceModel.MaxentModel;
cgen = sentenceModel.Factory.GetContextGenerator();
scanner = sentenceModel.Factory.GetEndOfSentenceScanner();
useTokenEnd = sentenceModel.UseTokenEnd;
if (sentenceModel.Abbreviations == null)
{
return;
}
stringComparison = sentenceModel.Abbreviations.IsCaseSensitive
? StringComparison.Ordinal
: StringComparison.OrdinalIgnoreCase;
abbreviationTokens = new Dictionary <string, int>();
foreach (var abbreviation in sentenceModel.Abbreviations)
{
foreach (var token in abbreviation.Tokens)
{
abbreviationTokens.Add(token, token.Length);
}
}
}
public SentenceDetectorME(SentenceModel sentenceModel)
{
model = sentenceModel.MaxentModel;
cgen = sentenceModel.Factory.GetContextGenerator();
scanner = sentenceModel.Factory.GetEndOfSentenceScanner();
useTokenEnd = sentenceModel.UseTokenEnd;
}
public void TestDefault() {
var dic = LoadAbbDictionary();
Assert.NotNull(dic);
char[] eos = {'.', '?'};
var sdModel = Train(new SentenceDetectorFactory("en", true, dic, eos));
Assert.NotNull(sdModel);
SentenceDetectorFactory factory = sdModel.Factory;
Assert.NotNull(factory.AbbreviationDictionary);
Assert.True(factory.GetContextGenerator() is DefaultSentenceContextGenerator);
Assert.True(factory.GetEndOfSentenceScanner() is DefaultEndOfSentenceScanner);
Assert.True(eos.SequenceEqual(factory.EOSCharacters));
var o = new MemoryStream();
sdModel.Serialize(new UnclosableStream(o));
o.Seek(0, SeekOrigin.Begin);
var fromSerialized = new SentenceModel(o);
factory = fromSerialized.Factory;
Assert.NotNull(factory.AbbreviationDictionary);
Assert.True(factory.GetContextGenerator() is DefaultSentenceContextGenerator);
Assert.True(factory.GetEndOfSentenceScanner() is DefaultEndOfSentenceScanner);
Assert.True(eos.SequenceEqual(factory.EOSCharacters));
}
/// <summary>
/// Initializes a new instance of the <see cref="SentenceDetectorME" /> using the given sentence model.
/// </summary>
/// <param name="sentenceModel">The sentence model.</param>
/// <exception cref="System.ArgumentNullException">
/// The <paramref name="sentenceModel"/> is <c>null</c>.
/// </exception>
public SentenceDetectorME(SentenceModel sentenceModel) {
if (sentenceModel == null)
throw new ArgumentNullException("sentenceModel");
model = sentenceModel.MaxentModel;
cgen = sentenceModel.Factory.GetContextGenerator();
scanner = sentenceModel.Factory.GetEndOfSentenceScanner();
useTokenEnd = sentenceModel.UseTokenEnd;
if (sentenceModel.Abbreviations == null)
return;
stringComparison = sentenceModel.Abbreviations.IsCaseSensitive
? StringComparison.Ordinal
: StringComparison.OrdinalIgnoreCase;
abbreviationTokens = new Dictionary<string, int>();
foreach (var abbreviation in sentenceModel.Abbreviations)
foreach (var token in abbreviation.Tokens)
abbreviationTokens.Add(token, token.Length);
}
public void TestCrossCompatibility()
{
var jModel = OpenJavaModel();
var sModel = OpenSharpModel();
var jFile = Path.GetTempFileName();
var sFile = Path.GetTempFileName();
var jFileStream = OpenNLP.CreateOutputStream(jFile);
jModel.serialize(jFileStream);
jFileStream.close();
sModel.Serialize(new FileStream(sFile, FileMode.Create));
// now java opens the csharp model and vice versa :)
var jModel2 = new JavaModel(OpenNLP.CreateInputStream(sFile));
var sModel2 = new SharpModel(jFile);
Assert.Null(jModel2.getAbbreviations());
Assert.Null(sModel2.Abbreviations);
Assert.Null(jModel2.getEosCharacters());
Assert.Null(sModel2.EosCharacters);
Assert.AreEqual(jModel2.useTokenEnd(), sModel2.UseTokenEnd);
var jFactory2 = jModel2.getFactory();
var sFactory2 = sModel2.Factory;
Assert.AreEqual(jFactory2.isUseTokenEnd(), sFactory2.UseTokenEnd);
Assert.AreEqual(jFactory2.getLanguageCode(), sFactory2.LanguageCode);
Assert.True(true);
}
public void TestWithOpenNLPModel() {
using (var file = Tests.OpenFile("/opennlp/models/en-sent.bin")) {
var model = new SentenceModel(file);
EvalSentences(new SentenceDetectorME(model));
}
}
public void TestCrossCompatibility() {
var jModel = OpenJavaModel();
var sModel = OpenSharpModel();
var jFile = Path.GetTempFileName();
var sFile = Path.GetTempFileName();
var jFileStream = OpenNLP.CreateOutputStream(jFile);
jModel.serialize(jFileStream);
jFileStream.close();
sModel.Serialize(new FileStream(sFile, FileMode.Create));
// now java opens the csharp model and vice versa :)
var jModel2 = new JavaModel(OpenNLP.CreateInputStream(sFile));
var sModel2 = new SharpModel(jFile);
Assert.Null(jModel2.getAbbreviations());
Assert.Null(sModel2.Abbreviations);
Assert.Null(jModel2.getEosCharacters());
Assert.Null(sModel2.EosCharacters);
Assert.AreEqual(jModel2.useTokenEnd(), sModel2.UseTokenEnd);
var jFactory2 = jModel2.getFactory();
var sFactory2 = sModel2.Factory;
Assert.AreEqual(jFactory2.isUseTokenEnd(), sFactory2.UseTokenEnd);
Assert.AreEqual(jFactory2.getLanguageCode(), sFactory2.LanguageCode);
Assert.True(true);
}
