/*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package opennlp.tools.namefind;
import java.io.ByteArrayInputStream;
import java.io.File;
import java.io.IOException;
import java.io.InputStream;
import java.io.OutputStream;
import java.net.URL;
import java.util.ArrayList;
import java.util.Collections;
import java.util.List;
import java.util.Map;
import opennlp.model.AbstractModel;
import opennlp.model.MaxentModel;
import opennlp.tools.util.InvalidFormatException;
import opennlp.tools.util.featuregen.AdaptiveFeatureGenerator;
import opennlp.tools.util.featuregen.AggregatedFeatureGenerator;
import opennlp.tools.util.featuregen.FeatureGeneratorResourceProvider;
import opennlp.tools.util.featuregen.GeneratorFactory;
import opennlp.tools.util.model.ArtifactSerializer;
import opennlp.tools.util.model.BaseModel;
import opennlp.tools.util.model.ModelUtil;
/**
* The {@link TokenNameFinderModel} is the model used
* by a learnable {@link TokenNameFinder}.
*
* @see NameFinderME
*/
public class TokenNameFinderModel extends BaseModel {
public static class FeatureGeneratorCreationError extends RuntimeException {
FeatureGeneratorCreationError(Throwable t) {
super(t);
}
}
private static class ByteArraySerializer implements ArtifactSerializer<byte[]> {
public byte[] create(InputStream in) throws IOException,
InvalidFormatException {
return ModelUtil.read(in);
}
public void serialize(byte[] artifact, OutputStream out) throws IOException {
out.write(artifact);
}
}
private static final String COMPONENT_NAME = "NameFinderME";
private static final String MAXENT_MODEL_ENTRY_NAME = "nameFinder.model";
private static final String GENERATOR_DESCRIPTOR_ENTRY_NAME = "generator.featuregen";
public TokenNameFinderModel(String languageCode, AbstractModel nameFinderModel,
byte[] generatorDescriptor, Map<String, Object> resources, Map<String, String> manifestInfoEntries) {
super(COMPONENT_NAME, languageCode, manifestInfoEntries);
if (!isModelValid(nameFinderModel)) {
throw new IllegalArgumentException("Model not compatible with name finder!");
}
artifactMap.put(MAXENT_MODEL_ENTRY_NAME, nameFinderModel);
if (generatorDescriptor != null && generatorDescriptor.length > 0)
artifactMap.put(GENERATOR_DESCRIPTOR_ENTRY_NAME, generatorDescriptor);
if (resources != null) {
// The resource map must not contain key which are already taken
// like the name finder maxent model name
if (resources.containsKey(MAXENT_MODEL_ENTRY_NAME) ||
resources.containsKey(GENERATOR_DESCRIPTOR_ENTRY_NAME)) {
throw new IllegalArgumentException();
}
// TODO: Add checks to not put resources where no serializer exists,
// make that case fail here, should be done in the BaseModel
artifactMap.putAll(resources);
}
checkArtifactMap();
}
public TokenNameFinderModel(String languageCode, AbstractModel nameFinderModel,
Map<String, Object> resources, Map<String, String> manifestInfoEntries) {
this(languageCode, nameFinderModel, null, resources, manifestInfoEntries);
}
public TokenNameFinderModel(InputStream in) throws IOException, InvalidFormatException {
super(COMPONENT_NAME, in);
}
public TokenNameFinderModel(File modelFile) throws IOException, InvalidFormatException {
super(COMPONENT_NAME, modelFile);
}
public TokenNameFinderModel(URL modelURL) throws IOException, InvalidFormatException {
super(COMPONENT_NAME, modelURL);
}
/**
* Retrieves the {@link TokenNameFinder} model.
*
* @return the classification model
*/
public AbstractModel getNameFinderModel() {
return (AbstractModel) artifactMap.get(MAXENT_MODEL_ENTRY_NAME);
}
/**
* Creates the {@link AdaptiveFeatureGenerator}. Usually this
* is a set of generators contained in the {@link AggregatedFeatureGenerator}.
*
* Note:
* The generators are created on every call to this method.
*
* @return the feature generator or null if there is no descriptor in the model
*/
public AdaptiveFeatureGenerator createFeatureGenerators() {
byte descriptorBytes[] = (byte[]) artifactMap.get(GENERATOR_DESCRIPTOR_ENTRY_NAME);
if (descriptorBytes != null) {
InputStream descriptorIn = new ByteArrayInputStream(descriptorBytes);
AdaptiveFeatureGenerator generator = null;
try {
generator = GeneratorFactory.create(descriptorIn, new FeatureGeneratorResourceProvider() {
public Object getResource(String key) {
return artifactMap.get(key);
}
});
} catch (InvalidFormatException e) {
// It is assumed that the creation of the feature generation does not
// fail after it succeeded once during model loading.
// But it might still be possible that such an exception is thrown,
// in this case the caller should not be forced to handle the exception
// and a Runtime Exception is thrown instead.
// If the re-creation of the feature generation fails it is assumed
// that this can only be caused by a programming mistake and therefore
// throwing a Runtime Exception is reasonable
throw new FeatureGeneratorCreationError(e);
} catch (IOException e) {
throw new IllegalStateException("Reading from mem cannot result in an I/O error", e);
}
return generator;
}
else {
return null;
}
}
public TokenNameFinderModel updateFeatureGenerator(byte descriptor[]) {
TokenNameFinderModel model = new TokenNameFinderModel(getLanguage(), getNameFinderModel(),
descriptor, Collections.<String, Object>emptyMap(), Collections.<String, String>emptyMap());
// TODO: Not so nice!
model.artifactMap.clear();
model.artifactMap.putAll(artifactMap);
model.artifactMap.put(GENERATOR_DESCRIPTOR_ENTRY_NAME, descriptor);
return model;
}
@Override
protected void createArtifactSerializers(Map<String, ArtifactSerializer> serializers) {
super.createArtifactSerializers(serializers);
serializers.put("featuregen", new ByteArraySerializer());
}
public static Map<String, ArtifactSerializer> createArtifactSerializers() {
// TODO: Not so nice, because code cannot really be reused by the other create serializer method
// Has to be redesigned, we need static access to default serializers
// and these should be able to extend during runtime ?!
Map<String, ArtifactSerializer> serializers = BaseModel.createArtifactSerializers();
serializers.put("featuregen", new ByteArraySerializer());
return serializers;
}
// TODO: Write test for this method
public static boolean isModelValid(MaxentModel model) {
// We should have *optionally* one outcome named "other", some named xyz-start and sometimes
// they have a pair xyz-cont. We should not have any other outcome
// To validate the model we check if we have one outcome named "other", at least
// one outcome with suffix start. After that we check if all outcomes that ends with
// "cont" have a pair that ends with "start".
List<String> start = new ArrayList<String>();
List<String> cont = new ArrayList<String>();
for (int i = 0; i < model.getNumOutcomes(); i++) {
String outcome = model.getOutcome(i);
if (outcome.endsWith(NameFinderME.START)) {
start.add(outcome.substring(0, outcome.length()
- NameFinderME.START.length()));
} else if (outcome.endsWith(NameFinderME.CONTINUE)) {
cont.add(outcome.substring(0, outcome.length()
- NameFinderME.CONTINUE.length()));
} else if (outcome.equals(NameFinderME.OTHER)) {
// don't fail anymore if couldn't find outcome named OTHER
} else {
// got unexpected outcome
return false;
}
}
if (start.size() == 0) {
return false;
} else {
for (String contPreffix : cont) {
if (!start.contains(contPreffix)) {
return false;
}
}
}
return true;
}
@Override
protected void validateArtifactMap() throws InvalidFormatException {
super.validateArtifactMap();
if (artifactMap.get(MAXENT_MODEL_ENTRY_NAME) instanceof AbstractModel) {
AbstractModel model = (AbstractModel) artifactMap.get(MAXENT_MODEL_ENTRY_NAME);
isModelValid(model);
}
else {
throw new InvalidFormatException("Token Name Finder model is incomplete!");
}
}
}