protected override void CreateArtifactSerializers()
{
base.CreateArtifactSerializers();
// note from OpenNLP (for future adaptations)
// In 1.6.x the headrules artifact is serialized with the new API
// which uses the Serializable interface
// This change is not backward compatible with the 1.5.x models.
// In order to load 1.5.x model the English headrules serializer must be
// put on the serializer map.
RegisterArtifactType(".headrules",
(artifact, stream) => HeadRulesManager.Serialize(artifact as AbstractHeadRules, stream),
stream => HeadRulesManager.Deserialize(Language, stream));
RegisterArtifactType(".postagger", (artifact, stream) => {
var model = artifact as POSModel;
if (model == null)
{
throw new InvalidOperationException();
}
model.Serialize(stream);
}, stream => new POSModel(stream));
RegisterArtifactType(".chunker", (artifact, stream) => {
var model = artifact as ChunkerModel;
if (model == null)
{
throw new InvalidOperationException();
}
model.Serialize(stream);
}, stream => new ChunkerModel(stream));
}
/// <summary>
/// Registers all serializers for their artifact file name extensions. Override this method to register custom file extensions.
/// </summary>
/// <seealso href="https://msdn.microsoft.com/en-us/library/ms182331.aspx" />
/// <remarks>The subclasses should invoke the <see cref="ArtifactProvider.RegisterArtifactType" /> to register
/// the proper serialization/deserialization methods for an new extension.
/// Warning: This method is called in constructor of the base class!! Be aware that this method is ONLY designed to register serializers.</remarks>
protected override void CreateArtifactSerializers()
{
base.CreateArtifactSerializers();
// note from OpenNLP (for future adaptations)
// In 1.6.x the headrules artifact is serialized with the new API
// which uses the Serializable interface
// This change is not backward compatible with the 1.5.x models.
// In order to load 1.5.x model the English headrules serializer must be
// put on the serializer map.
RegisterArtifactType(".headrules",
(artifact, stream) => HeadRulesManager.Serialize(artifact as AbstractHeadRules, stream),
stream => HeadRulesManager.Deserialize(Language, stream));
RegisterArtifactType(".postagger", (artifact, stream) => {
var model = artifact as POSModel;
if (model == null)
{
throw new InvalidOperationException();
}
model.Serialize(stream);
}, stream => {
var model = new POSModel(stream);
// The 1.6.x models write the non-default beam size into the model itself.
// In 1.5.x the parser configured the beam size when the model was loaded,
// this is not possible anymore with the new APIs
if (model.Version.Major == 1 && model.Version.Minor == 5 && !model.Manifest.Contains(Parameters.BeamSize))
{
return(new POSModel(model.Language, model.MaxentModel, 10, null, model.Factory));
}
return(model);
});
RegisterArtifactType(".chunker", (artifact, stream) => {
var model = artifact as ChunkerModel;
if (model == null)
{
throw new InvalidOperationException();
}
model.Serialize(stream);
}, stream => {
var model = new ChunkerModel(stream);
if (model.Version.Major == 1 && model.Version.Minor == 5)
{
return(new ChunkerModel(model.Language, model.MaxentModel, new ParserChunkerFactory()));
}
return(model);
});
}
internal static AbstractHeadRules CreateTestHeadRules()
{
AbstractHeadRules rules;
using (var file = Tests.OpenFile("/opennlp/tools/parser/en_head_rules")) {
Assert.NotNull(file);
rules = HeadRulesManager.Deserialize("en", file);
Assert.NotNull(rules);
}
return(rules);
}