EDIT

// $Id$

/**

* Copyright (C) 2016 EDIT

* European Distributed Institute of Taxonomy

* http://www.e-taxonomy.eu

*

* The contents of this file are subject to the Mozilla Public License Version 1.1

* See LICENSE.TXT at the top of this package for the full license terms.

*/

package eu.etaxonomy.taxeditor.molecular.io;

import info.bioinfweb.commons.bio.CharacterStateSetType;

import info.bioinfweb.commons.text.StringUtils;

import info.bioinfweb.jphyloio.ReadWriteConstants;

import info.bioinfweb.jphyloio.ReadWriteParameterMap;

import info.bioinfweb.jphyloio.dataadapters.JPhyloIOEventReceiver;

import info.bioinfweb.jphyloio.dataadapters.MatrixDataAdapter;

import info.bioinfweb.jphyloio.dataadapters.ObjectListDataAdapter;

import info.bioinfweb.jphyloio.dataadapters.implementations.NoCharDefsNoSetsMatrixDataAdapter;

import info.bioinfweb.jphyloio.dataadapters.implementations.store.StoreObjectListDataAdapter;

import info.bioinfweb.jphyloio.events.CharacterSetIntervalEvent;

import info.bioinfweb.jphyloio.events.LinkedLabeledIDEvent;

import info.bioinfweb.jphyloio.events.SequenceTokensEvent;

import info.bioinfweb.jphyloio.events.TokenSetDefinitionEvent;

import info.bioinfweb.jphyloio.events.type.EventContentType;

import java.io.IOException;

import java.util.Iterator;

import eu.etaxonomy.cdm.model.molecular.Sequence;

/**

 * In implementation of {@link MatrixDataAdapter} that delegates to a {@link Sequence} object.

 *

 * @author Ben Stöver

 * @date 29.04.2016

 */

public class CDMSequenceMatrixAdapter extends NoCharDefsNoSetsMatrixDataAdapter implements ReadWriteConstants {

    public static final String MATRIX_ID = DEFAULT_MATRIX_ID_PREFIX + "ContigAlignment";

    public static final String SINGLE_READ_SEQUENCE_ID_PREFIX = DEFAULT_SEQUENCE_ID_PREFIX + "SingleRead";

    public static final String CONSENSUS_SEQUENCE_ID= DEFAULT_SEQUENCE_ID_PREFIX + "Consensus";

    private Sequence sequence;

    private String consensusSequenceLabel;

    private ObjectListDataAdapter<TokenSetDefinitionEvent> tokenSetList;

    //TODO Also allow specifying single read labels?

    /**

     * Creates a new instance of this class.

     *

     * @param sequence

     * @param consensusSequenceLabel

     */

    public CDMSequenceMatrixAdapter(Sequence sequence, String consensusSequenceLabel) {

        super();

        this.sequence = sequence;

        this.consensusSequenceLabel = consensusSequenceLabel;

        tokenSetList = createTokenSetList();

    }

    private ObjectListDataAdapter<TokenSetDefinitionEvent> createTokenSetList() {

        StoreObjectListDataAdapter<TokenSetDefinitionEvent> result = new StoreObjectListDataAdapter<TokenSetDefinitionEvent>();

        final String id = ReadWriteConstants.DEFAULT_TOKEN_SET_ID_PREFIX;

        result.setObjectStartEvent(new TokenSetDefinitionEvent(CharacterStateSetType.DNA, id, null));

        result.getObjectContent(id).add(new CharacterSetIntervalEvent(0, getColumnCount(null)));  //TODO Change this expression, if column count should return -1 in the future.

        return result;

    }

    /**

     * @return the sequence

     */

    public Sequence getCDMSequence() {

        return sequence;

    }

    private int extractSingleReadIndexFromID(String sequenceID) {

        if (sequenceID.startsWith(SINGLE_READ_SEQUENCE_ID_PREFIX)) {

            try {

                return Integer.parseInt(sequenceID.substring(SINGLE_READ_SEQUENCE_ID_PREFIX.length()));

            }

            catch (NumberFormatException e) {}  // fall through

        }

        return -1;

    }

    @Override

    public LinkedLabeledIDEvent getStartEvent(ReadWriteParameterMap parameters) {

        return new LinkedLabeledIDEvent(EventContentType.ALIGNMENT, MATRIX_ID, "Contig alignment", null);

                //TODO Use label according to derivate and markers.

    }

    @Override

    public boolean containsLongTokens(ReadWriteParameterMap parameters) {

        return false;

    }

    @Override

    public long getColumnCount(ReadWriteParameterMap parameters) {

        return getCDMSequence().getSequenceString().length();  //TODO Consider that parts of single reads may lie outside of the current consensus sequence. Possibly return -1 in the future.

    }

    @Override

    public long getSequenceCount(ReadWriteParameterMap parameters) {

        return /*sequence.getSingleReadAlignments().size() +*/ 1;

    }

    @Override

    public Iterator<String> getSequenceIDIterator(ReadWriteParameterMap parameters) {

        return new SequenceIDIterator(getCDMSequence().getSingleReadAlignments().size());

    }

    @Override

    public long getSequenceLength(ReadWriteParameterMap parameters, String sequenceID) throws IllegalArgumentException {

        int singleReadIndex = extractSingleReadIndexFromID(sequenceID);

        if (singleReadIndex >= 0) {

            //TODO Determine single read length. (Current problem: Reads are not ordered in the set of the Sequence object.)

            return 0;

        }

        else if (CONSENSUS_SEQUENCE_ID.equals(sequenceID)) {

            return getCDMSequence().getSequenceString().length();

        }

        else {

            throw new IllegalArgumentException("No sequence with the ID \"" + sequenceID + "\" could be found.");

        }

    }

    @Override

    public LinkedLabeledIDEvent getSequenceStartEvent(ReadWriteParameterMap parameters, String sequenceID) {

        int singleReadIndex = extractSingleReadIndexFromID(sequenceID);

        if (singleReadIndex >= 0) {

            return new LinkedLabeledIDEvent(EventContentType.SEQUENCE, sequenceID, "Single read " + singleReadIndex, null);

                    //TODO Use name displayed in derivate hierarchy or specified name as label instead?

        }

        else if (CONSENSUS_SEQUENCE_ID.equals(sequenceID)) {

            return new LinkedLabeledIDEvent(EventContentType.SEQUENCE, sequenceID, consensusSequenceLabel, null);

        }

        else {

            throw new IllegalArgumentException("No sequence with the ID \"" + sequenceID + "\" could be found.");

        }

    }

    @Override

    public void writeSequencePartContentData(ReadWriteParameterMap parameters, JPhyloIOEventReceiver receiver, String sequenceID, long startColumn,

            long endColumn) throws IOException, IllegalArgumentException {

        int singleReadIndex = extractSingleReadIndexFromID(sequenceID);

        if (singleReadIndex >= 0) {

            //TODO Implement single read output.

        }

        else if (CONSENSUS_SEQUENCE_ID.equals(sequenceID)) {

            if (startColumn == 0) {

                //TODO Add consensus sequence metadata.

                //TODO Possibly export additional properties of sequence (e.g. isBarcode(), getDdbjId(), ...) as metadata.

            }

            receiver.add(new SequenceTokensEvent(StringUtils.charSequenceToStringList(

                    getCDMSequence().getSequenceString().substring((int)startColumn, (int)endColumn))));

        }

        else {

            throw new IllegalArgumentException("No sequence with the ID \"" + sequenceID + "\" could be found.");

        }

    }

    /* (non-Javadoc)

     * @see info.bioinfweb.jphyloio.dataadapters.implementations.NoSetsMatrixDataAdapter#getTokenSets()

     */

    @Override

    public ObjectListDataAdapter<TokenSetDefinitionEvent> getTokenSets(ReadWriteParameterMap parameters) {

        return tokenSetList;

    }

}