JAL-1705 regular expression updates, tests, other refactoring

[jalview.git] / src / jalview / ext / ensembl / EnsemblProtein.java

diff --git a/src/jalview/ext/ensembl/EnsemblProtein.java b/src/jalview/ext/ensembl/EnsemblProtein.java
index 29c7eda..fb79ccf 100644 (file)
--- a/src/jalview/ext/ensembl/EnsemblProtein.java
+++ b/src/jalview/ext/ensembl/EnsemblProtein.java
@@ -10,9 +10,13 @@ import com.stevesoft.pat.Regex;
 
 public class EnsemblProtein extends EnsemblSeqProxy
 {
-  // TODO modify to accept other species e.g. ENSMUSPnnn
+  /*
+   * accepts ENSP 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(
-          "(ENSP|CCDS)[0-9.]{3,}$");
+          "(ENS([A-Z]{3}|)P[0-9]{11}$)" + "|" + "(CCDS[0-9.]{3,}$)");
 
   private static final List<String> CROSSREFS = Arrays.asList(new String[] {
       "PDB", "Uniprot/SPTREMBL", "Uniprot/SWISSPROT" });