package jalview.ext.ensembl;

import jalview.datamodel.DBRefSource;
import jalview.ws.seqfetcher.DbSourceProxyImpl;

import com.stevesoft.pat.Regex;

/**
 * A base class for Ensembl sequence fetchers
 * 
 * @author gmcarstairs
 */
abstract class EnsemblSequenceFetcher extends DbSourceProxyImpl
{
  /*
   * accepts ENSG/T/E/P with 11 digits
   * or ENSMUSP or similar for other species
   * or CCDSnnnnn.nn with at least 3 digits
   */
  private static final Regex ACCESSION_REGEX = new Regex(
          "(ENS([A-Z]{3}|)[GTEP]{1}[0-9]{11}$)" + "|" + "(CCDS[0-9.]{3,}$)");

  /*
   * possible values for the 'feature' parameter of the /overlap REST service
   * @see http://rest.ensembl.org/documentation/info/overlap_id
   */
  protected enum EnsemblFeatureType
  {
    gene, transcript, cds, exon, repeat, simple, misc, variation,
    somatic_variation, structural_variation, somatic_structural_variation,
    constrained, regulatory
  }

  @Override
  public String getDbSource()
  {
    // NB ensure Uniprot xrefs are canonicalised from "Ensembl" to "ENSEMBL"
    return DBRefSource.ENSEMBL; // "ENSEMBL"
  }

  @Override
  public String getDbVersion()
  {
    return "0";
  }

  @Override
  public String getAccessionSeparator()
  {
    return " ";
  }

  /**
   * Ensembl accession are ENST + 11 digits for human transcript, ENSG for human
   * gene. Other species insert 3 letters e.g. ENSMUST..., ENSMUSG...
   * 
   * @see http://www.ensembl.org/Help/View?id=151
   */
  @Override
  public Regex getAccessionValidator()
  {
    return ACCESSION_REGEX;
  }

  @Override
  public boolean isValidReference(String accession)
  {
    return getAccessionValidator().search(accession);
  }

  @Override
  public int getTier()
  {
    return 0;
  }

  /**
   * Default test query is a transcript
   */
  @Override
  public String getTestQuery()
  {
    // has CDS on reverse strand:
    return "ENST00000288602";
    // ENST00000461457 // forward strand
  }

  @Override
  public boolean isDnaCoding()
  {
    return true;
  }
}