package jalview.datamodel;

import static org.junit.Assert.assertEquals;
import static org.junit.Assert.assertFalse;
import static org.junit.Assert.assertTrue;
import jalview.io.AppletFormatAdapter;
import jalview.io.FormatAdapter;
import jalview.util.MapList;

import java.io.IOException;
import java.util.Iterator;

import org.junit.Before;
import org.junit.Test;

/**
 * Unit tests for Alignment datamodel.
 * 
 * @author gmcarstairs
 *
 */
public class AlignmentTest
{
  // @formatter:off
  private static final String TEST_DATA = 
          "# STOCKHOLM 1.0\n" +
          "#=GS D.melanogaster.1 AC AY119185.1/838-902\n" +
          "#=GS D.melanogaster.2 AC AC092237.1/57223-57161\n" +
          "#=GS D.melanogaster.3 AC AY060611.1/560-627\n" +
          "D.melanogaster.1          G.AGCC.CU...AUGAUCGA\n" +
          "#=GR D.melanogaster.1 SS  ................((((\n" +
          "D.melanogaster.2          C.AUUCAACU.UAUGAGGAU\n" +
          "#=GR D.melanogaster.2 SS  ................((((\n" +
          "D.melanogaster.3          G.UGGCGCU..UAUGACGCA\n" +
          "#=GR D.melanogaster.3 SS  (.(((...(....(((((((\n" +
          "//";

  private static final String AA_SEQS_1 = 
          ">Seq1Name\n" +
          "K-QY--L\n" +
          ">Seq2Name\n" +
          "-R-FP-W-\n";

  private static final String CDNA_SEQS_1 = 
          ">Seq1Name\n" +
          "AC-GG--CUC-CAA-CT\n" +
          ">Seq2Name\n" +
          "-CG-TTA--ACG---AAGT\n";

  private static final String CDNA_SEQS_2 = 
          ">Seq1Name\n" +
          "GCTCGUCGTACT\n" +
          ">Seq2Name\n" +
          "GGGTCAGGCAGT\n";
  // @formatter:on

  private AlignmentI al;

  /**
   * Helper method to load an alignment and ensure dataset sequences are set up.
   * 
   * @param data
   * @param format TODO
   * @return
   * @throws IOException
   */
  protected AlignmentI loadAlignment(final String data, String format) throws IOException
  {
    Alignment a = new FormatAdapter().readFile(data,
            AppletFormatAdapter.PASTE, format);
    a.setDataset(null);
    return a;
  }

  /*
   * Read in Stockholm format test data including secondary structure
   * annotations.
   */
  @Before
  public void setUp() throws IOException
  {
    al = loadAlignment(TEST_DATA, "STH");
    int i = 0;
    for (AlignmentAnnotation ann : al.getAlignmentAnnotation())
    {
      ann.setCalcId("CalcIdFor"
              + al.getSequenceAt(i).getName());
      i++;
    }
  }

  /**
   * Test method that returns annotations that match on calcId.
   */
  @Test
  public void testFindAnnotation_byCalcId()
  {
    Iterable<AlignmentAnnotation> anns = al
            .findAnnotation("CalcIdForD.melanogaster.2");
    Iterator<AlignmentAnnotation> iter = anns.iterator();
    assertTrue(iter.hasNext());
    AlignmentAnnotation ann = iter.next();
    assertEquals("D.melanogaster.2", ann.sequenceRef.getName());
    assertFalse(iter.hasNext());
  }

  /**
   * Tests for realigning as per a supplied alignment: Dna as Dna.
   * 
   * Note: AlignedCodonFrame's state variables are named for protein-to-cDNA
   * mapping, but can be exploited for a general 'sequence-to-sequence' mapping
   * as here.
   * 
   * @throws IOException
   */
  @Test
  public void testAlignAs_dnaAsDna() throws IOException
  {
    // aligned cDNA:
    AlignmentI al1 = loadAlignment(CDNA_SEQS_1, "FASTA");
    // unaligned cDNA:
    AlignmentI al2 = loadAlignment(CDNA_SEQS_2, "FASTA");

    /*
     * Make mappings between sequences. The 'aligned cDNA' is playing the role
     * of what would normally be protein here.
     */
    AlignedCodonFrame acf = new AlignedCodonFrame();
    MapList ml = new MapList(new int[]
    { 1, 12 }, new int[]
    { 1, 12 }, 1, 1);
    acf.addMap(al2.getSequenceAt(0), al1.getSequenceAt(0), ml);
    acf.addMap(al2.getSequenceAt(1), al1.getSequenceAt(1), ml);
    al1.addCodonFrame(acf);

    al2.alignAs(al1);
    assertEquals("GC-TC--GUC-GTA-CT", al2.getSequenceAt(0)
            .getSequenceAsString());
    assertEquals("-GG-GTC--AGG---CAGT", al2.getSequenceAt(1)
            .getSequenceAsString());
  }

  /**
   * Aligning protein from cDNA yet to be implemented, does nothing.
   * 
   * @throws IOException
   */
  @Test
  public void testAlignAs_proteinAsCdna() throws IOException
  {
    AlignmentI al1 = loadAlignment(CDNA_SEQS_1, "FASTA");
    AlignmentI al2 = loadAlignment(AA_SEQS_1, "FASTA");
    String before0 = al2.getSequenceAt(0).getSequenceAsString();
    String before1 = al2.getSequenceAt(1).getSequenceAsString();

    al2.alignAs(al1);
    assertEquals(before0, al2.getSequenceAt(0).getSequenceAsString());
    assertEquals(before1, al2.getSequenceAt(1).getSequenceAsString());
  }

  /**
   * Test aligning cdna as per protein alignment.
   * 
   * @throws IOException
   */
  @Test
  public void testAlignAs_cdnaAsProtein() throws IOException
  {
    /*
     * Load alignments and add mappings for cDNA to protein
     */
    AlignmentI al1 = loadAlignment(CDNA_SEQS_1, "FASTA");
    AlignmentI al2 = loadAlignment(AA_SEQS_1, "FASTA");
    AlignedCodonFrame acf = new AlignedCodonFrame();
    MapList ml = new MapList(new int[]
    { 1, 12 }, new int[]
    { 1, 4 }, 3, 1);
    acf.addMap(al1.getSequenceAt(0), al2.getSequenceAt(0), ml);
    acf.addMap(al1.getSequenceAt(1), al2.getSequenceAt(1), ml);
    al2.addCodonFrame(acf);

    al1.alignAs(al2);
    assertEquals("ACG---GCUCCA------ACT", al1.getSequenceAt(0)
            .getSequenceAsString());
    assertEquals("---CGT---TAACGA---AGT---", al1.getSequenceAt(1)
            .getSequenceAsString());
  }
}