JAL-2682 non-regex sequence id parsing, and tests added

[jalview.git] / src / jalview / datamodel / Sequence.java

diff --git a/src/jalview/datamodel/Sequence.java b/src/jalview/datamodel/Sequence.java
index a01d185..1b91ee5 100755 (executable)
--- a/src/jalview/datamodel/Sequence.java
+++ b/src/jalview/datamodel/Sequence.java
@@ -38,8 +38,6 @@ import java.util.List;
 import java.util.ListIterator;
 import java.util.Vector;
 
-import com.stevesoft.pat.Regex;
-
 import fr.orsay.lri.varna.models.rna.RNA;
 
 /**
@@ -51,11 +49,6 @@ import fr.orsay.lri.varna.models.rna.RNA;
  */
 public class Sequence extends ASequence implements SequenceI
 {
-  private static final Regex limitrx = new Regex(
-          "[/][0-9]{1,}[-][0-9]{1,}$");
-
-  private static final Regex endrx = new Regex("[0-9]{1,}$");
-
   SequenceI datasetSequence;
 
   String name;
@@ -151,25 +144,49 @@ public class Sequence extends ASequence implements SequenceI
     checkValidRange();
   }
 
+  /**
+   * If 'name' ends in /i-j, where i >= j > 0 are integers, extracts i and j as
+   * start and end respectively and removes the suffix from the name
+   */
   void parseId()
   {
     if (name == null)
     {
-      System.err
-              .println("POSSIBLE IMPLEMENTATION ERROR: null sequence name passed to constructor.");
+      System.err.println(
+              "POSSIBLE IMPLEMENTATION ERROR: null sequence name passed to constructor.");
       name = "";
     }
-    // Does sequence have the /start-end signature?
-    if (limitrx.search(name))
+    int slashPos = name.lastIndexOf('/');
+    if (slashPos > -1 && slashPos < name.length() - 1)
     {
-      name = limitrx.left();
-      endrx.search(limitrx.stringMatched());
-      setStart(Integer.parseInt(limitrx.stringMatched().substring(1,
-              endrx.matchedFrom() - 1)));
-      setEnd(Integer.parseInt(endrx.stringMatched()));
+      String suffix = name.substring(slashPos + 1);
+      String[] range = suffix.split("-");
+      if (range.length == 2)
+      {
+        try
+        {
+          int from = Integer.valueOf(range[0]);
+          int to = Integer.valueOf(range[1]);
+          if (from > 0 && to >= from)
+          {
+            name = name.substring(0, slashPos);
+            setStart(from);
+            setEnd(to);
+            checkValidRange();
+          }
+        } catch (NumberFormatException e)
+        {
+          // leave name unchanged if suffix is invalid
+        }
+      }
     }
   }
 
+  /**
+   * Ensures that 'end' is not before the end of the sequence, that is,
+   * (end-start+1) is at least as long as the count of ungapped positions. Note
+   * that end is permitted to be beyond the end of the sequence data.
+   */
   void checkValidRange()
   {
     // Note: JAL-774 :
@@ -178,7 +195,7 @@ public class Sequence extends ASequence implements SequenceI
       int endRes = 0;
       for (int j = 0; j < sequence.length; j++)
       {
-        if (!jalview.util.Comparison.isGap(sequence[j]))
+        if (!Comparison.isGap(sequence[j]))
         {
           endRes++;
         }
@@ -453,15 +470,15 @@ public class Sequence extends ASequence implements SequenceI
   }
 
   /**
-   * DOCUMENT ME!
+   * Sets the sequence name. If the name ends in /start-end, then the start-end
+   * values are parsed out and set, and the suffix is removed from the name.
    * 
-   * @param name
-   *          DOCUMENT ME!
+   * @param theName
    */
   @Override
-  public void setName(String name)
+  public void setName(String theName)
   {
-    this.name = name;
+    this.name = theName;
     this.parseId();
   }
 
@@ -1364,8 +1381,9 @@ public class Sequence extends ASequence implements SequenceI
   @Override
   public AlignmentAnnotation[] getAnnotation()
   {
-    return annotation == null ? null : annotation
-            .toArray(new AlignmentAnnotation[annotation.size()]);
+    return annotation == null ? null
+            : annotation
+                    .toArray(new AlignmentAnnotation[annotation.size()]);
   }
 
   @Override
@@ -1477,8 +1495,9 @@ public class Sequence extends ASequence implements SequenceI
   {
     if (datasetSequence == null)
     {
-      Sequence dsseq = new Sequence(getName(), AlignSeq.extractGaps(
-              jalview.util.Comparison.GapChars, getSequenceAsString()),
+      Sequence dsseq = new Sequence(getName(),
+              AlignSeq.extractGaps(jalview.util.Comparison.GapChars,
+                      getSequenceAsString()),
               getStart(), getEnd());
 
       datasetSequence = dsseq;
@@ -1620,7 +1639,7 @@ public class Sequence extends ASequence implements SequenceI
       List<SequenceFeature> sfs = entry.getSequenceFeatures();
       for (SequenceFeature feature : sfs)
       {
-        SequenceFeature sf[] = (mp != null) ? mp.locateFeature(feature)
+       SequenceFeature sf[] = (mp != null) ? mp.locateFeature(feature)
                 : new SequenceFeature[] { new SequenceFeature(feature) };
         if (sf != null)
         {
@@ -1777,8 +1796,8 @@ public class Sequence extends ASequence implements SequenceI
           }
         }
         // whilst it looks like it is a primary ref, we also sanity check type
-        if (DBRefUtils.getCanonicalName(DBRefSource.PDB).equals(
-                DBRefUtils.getCanonicalName(ref.getSource())))
+        if (DBRefUtils.getCanonicalName(DBRefSource.PDB)
+                .equals(DBRefUtils.getCanonicalName(ref.getSource())))
         {
           // PDB dbrefs imply there should be a PDBEntry associated
           // TODO: tighten PDB dbrefs