de.julielab.geneexpbase.configuration.Configuration Maven / Gradle / Ivy

Go to download
package de.julielab.geneexpbase.configuration;

import de.julielab.java.utilities.FileUtilities;

import java.io.File;
import java.io.IOException;
import java.util.*;

/**
 * 
 * A simple configuration class used to specify global static configuration values.
 * 
 * Those are mostly file system paths to resources.
 * Algorithmical parameters are represented by the {@link Parameters} class.
 */
public class Configuration extends Properties {
    public static final String EHCACHE_CONFIG = "cache.ehcache.config";
    /**
     * Short for "machine learning". A special path element specifying the training and usage of a machine learning model.
     * The idea is that the {@link #ML} element indicates the existence of a machine learning configuration. The directly
     * succeeding path element should than represent the root of the configuration parameters for the model.
     * E.g. candidate_retrieval.ml.approxmatches
     * would be the root for the re-ranking ML model for approximate candidate matches for a gene mention.
     */
    public static final String ML = "ml";
    public static final String PARAM_ALGORITHM = "algorithm";
    /**
     * Configuration path element to denote the training instances creates for a machine learning approach.
     * Should be located directly at the root of a machine learning configuration.
     *
     * @see {@link #ML}
     */
    public static final String TRAINING_INSTANCES = "training_instances";
    /**
     * Instance field: The concrete classifier to employ for disambiguation.
     */
    public static final String KEY_CLASSIFIER = "classifier";
    /**
     * Instance field: The concrete ranger to employ for disambiguation.
     */
    public static final String KEY_RANKER = "ranker";
    public static final String PARAM_STANDARDIZE_FEATURES = "standardize_features";
    public static final String PARAM_MINMAX_SCALE_FEATURES = "minmax_scale_features";
    public static final String KEY_STANDARDIZATION_VALUES = "standardization_values";
    public static final String KEY_MINMAX_SCALING_VALUES = "minmax_scaling_values";
    public static final String KEY_DATA_ALPHABET = "data_alphabet";
    public static final String KEY_TARGET_ALPHABET = "target_alphabet";
    public static final String VALUE_LTR = "ltr";
    public static final String VALUE_LUCENE = "lucene";
    public static final String VALUE_MAXENT = "maxent";
    public static final String VALUE_SVM = "svm";
    public static final String VALUE_TRANSFORMER = "transformer";
    public static final String PARAM_SVM_TYPE = "svm.type";
    public static final String PARAM_SVM_C = "svm.c";
    public static final String PARAM_SVM_KERNEL_TYPE = "svm.kerneltype";
    public static final String PARAM_SVM_GAMMA = "svm.gamma";
    public static final String PARAM_SVM_COEF0 = "svm.coef0";
    public static final String PARAM_SVM_DEGREE = "svm.degree";
    public static final String PARAM_LTR_METRIC = "ltr.metric";
    public static final String PARAM_LTR_K = "ltr.k";
    public static final String PARAM_LTR_ALGORITHM = "ltr.algorithm";
    public static final String PARAM_PYTHON_EXECUTABLE = "python_executable";
    public static final String PYTHON_PROCESS_LIMIT = "python_process_limit";
    public static final String CONCURRENCY_LEVEL = "concurrency_level";
    /**
     * For SMAC optimization with ML approaches: Whether to use the training data from all corpora or only the
     * corpus the current instance is from.
     */
    public static final String PARAM_ML_USE_ALL_CORPUS_TRAINSPLITS = "use_all_corpus_trainsplits";
    // Mention and semantic score scaling to [0,1]. This is supposed to help with finding good rejection thresholds
    // when the score is not a probability by nature.
    public static final String PARAM_SCALE_RESULT_SCORE = "scale_result_score";
    public static final String KEY_MAX_RESULT_SCORE = "max_result_score";
    public static final String KEY_MIN_RESULT_SCORE = "min_result_score";
    // Model paths
    public static final String PARAM_MODEL = "model";
    public static final String GENE_ORTHOLOGS_PATH = "gene_orthologs";

    public Configuration() {
    }

    public Configuration(String... parameters) {
        if (parameters != null) {
            if (parameters.length % 2 == 1)
                throw new IllegalArgumentException("Uneven number parameter-value items.");
            for (int i = 0; i < parameters.length - 1; i++) {
                String parameter = parameters[i];
                String value = parameters[i + 1];
                setProperty(parameter, value);
            }
        }
    }

    public Configuration(File configurationFile) throws IOException {
        this.load(FileUtilities.getInputStreamFromFile(configurationFile));
    }

    /**
     * Adds all properties in baseConfiguration to this configuration.
     * @param baseConfiguration The configuration to copy.
     */
    public Configuration(de.julielab.geneexpbase.configuration.Configuration baseConfiguration) {
        for (String key : baseConfiguration.stringPropertyNames())
            setProperty(key, baseConfiguration.getProperty(key));
    }

    public static String dot(String... inputs) {
        return String.join(".", inputs);
    }

    public Optional getBoolean(String parameter) {
        String property = getProperty(parameter.trim());
        if (property != null)
            return Optional.of(Boolean.parseBoolean(property.trim()));
        return Optional.empty();
    }

    public OptionalInt getInteger(String parameter) {
        String property = getProperty(parameter.trim());
        if (property != null)
            return OptionalInt.of(Integer.parseInt(property));
        return OptionalInt.empty();
    }

    public OptionalDouble getDouble(String parameter) {
        String property = getProperty(parameter.trim());
        if (property != null)
            return OptionalDouble.of(Float.parseFloat(property));
        return OptionalDouble.empty();
    }

    public OptionalLong getLong(String parameter) {
        String property = getProperty(parameter.trim());
        if (property != null)
            return OptionalLong.of(Long.parseLong(property));
        return OptionalLong.empty();
    }
}