Merge branch 'develop' into features/JAL-2094_colourInterface

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

diff --git a/src/jalview/datamodel/Alignment.java b/src/jalview/datamodel/Alignment.java
index 32bb761..2f64759 100755 (executable)
--- a/src/jalview/datamodel/Alignment.java
+++ b/src/jalview/datamodel/Alignment.java
@@ -21,6 +21,7 @@
 package jalview.datamodel;
 
 import jalview.analysis.AlignmentUtils;
+import jalview.datamodel.AlignedCodonFrame.SequenceToSequenceMapping;
 import jalview.io.FastaFile;
 import jalview.util.Comparison;
 import jalview.util.MessageManager;
@@ -225,18 +226,21 @@ public class Alignment implements AlignmentI
   {
     if (dataset != null)
     {
+
       // maintain dataset integrity
-      if (snew.getDatasetSequence() != null)
-      {
-        getDataset().addSequence(snew.getDatasetSequence());
-      }
-      else
+      SequenceI dsseq = snew.getDatasetSequence();
+      if (dsseq == null)
       {
         // derive new sequence
         SequenceI adding = snew.deriveSequence();
-        getDataset().addSequence(adding.getDatasetSequence());
         snew = adding;
+        dsseq = snew.getDatasetSequence();
       }
+      if (getDataset().findIndex(dsseq) == -1)
+      {
+        getDataset().addSequence(dsseq);
+      }
+
     }
     if (sequences == null)
     {
@@ -255,18 +259,22 @@ public class Alignment implements AlignmentI
     }
   }
 
-  /**
-   * Adds a sequence to the alignment. Recalculates maxLength and size.
-   * 
-   * @param snew
-   */
   @Override
-  public void setSequenceAt(int i, SequenceI snew)
+  public SequenceI replaceSequenceAt(int i, SequenceI snew)
   {
     synchronized (sequences)
     {
-      deleteSequence(i);
-      sequences.set(i, snew);
+      if (sequences.size() > i)
+      {
+        return sequences.set(i, snew);
+
+      }
+      else
+      {
+        sequences.add(snew);
+        hiddenSequences.adjustHeightSequenceAdded();
+      }
+      return null;
     }
   }
 
@@ -1029,6 +1037,62 @@ public class Alignment implements AlignmentI
   }
 
   /**
+   * add dataset sequences to seq for currentSeq and any sequences it references
+   */
+  private void resolveAndAddDatasetSeq(SequenceI currentSeq,
+          Set<SequenceI> seqs, boolean createDatasetSequence)
+  {
+    if (currentSeq.getDatasetSequence() != null)
+    {
+      currentSeq = currentSeq.getDatasetSequence();
+    }
+    else
+    {
+      if (createDatasetSequence)
+      {
+        currentSeq = currentSeq.createDatasetSequence();
+      }
+    }
+    if (seqs.contains(currentSeq))
+    {
+      return;
+    }
+    List<SequenceI> toProcess = new ArrayList<SequenceI>();
+    toProcess.add(currentSeq);
+    while (toProcess.size() > 0)
+    {
+      // use a queue ?
+      SequenceI curDs = toProcess.remove(0);
+      if (seqs.contains(curDs))
+      {
+        continue;
+      }
+      seqs.add(curDs);
+      // iterate over database references, making sure we add forward referenced
+      // sequences
+      if (curDs.getDBRefs() != null)
+      {
+        for (DBRefEntry dbr : curDs.getDBRefs())
+        {
+          if (dbr.getMap() != null && dbr.getMap().getTo() != null)
+          {
+            if (dbr.getMap().getTo().getDatasetSequence() != null)
+            {
+              throw new Error("Implementation error: Map.getTo() for dbref"
+                      + dbr + " is not a dataset sequence.");
+              // TODO: if this happens, could also rewrite the reference to
+              // point to new dataset sequence
+            }
+            // we recurse to add all forward references to dataset sequences via
+            // DBRefs/etc
+            toProcess.add(dbr.getMap().getTo());
+          }
+        }
+      }
+    }
+  }
+
+  /**
    * Creates a new dataset for this alignment. Can only be done once - if
    * dataset is not null this will not be performed.
    */
@@ -1038,22 +1102,32 @@ public class Alignment implements AlignmentI
     {
       return;
     }
-    SequenceI[] seqs = new SequenceI[getHeight()];
-    SequenceI currentSeq;
+    // try to avoid using SequenceI.equals at this stage, it will be expensive
+    Set<SequenceI> seqs = new jalview.util.LinkedIdentityHashSet<SequenceI>();
+
     for (int i = 0; i < getHeight(); i++)
     {
-      currentSeq = getSequenceAt(i);
-      if (currentSeq.getDatasetSequence() != null)
-      {
-        seqs[i] = currentSeq.getDatasetSequence();
-      }
-      else
+      SequenceI currentSeq = getSequenceAt(i);
+      resolveAndAddDatasetSeq(currentSeq, seqs, true);
+    }
+
+    // verify all mappings are in dataset
+    for (AlignedCodonFrame cf : codonFrameList)
+    {
+      for (SequenceToSequenceMapping ssm : cf.getMappings())
       {
-        seqs[i] = currentSeq.createDatasetSequence();
+        if (!seqs.contains(ssm.getFromSeq()))
+        {
+          resolveAndAddDatasetSeq(ssm.getFromSeq(), seqs, false);
+        }
+        if (!seqs.contains(ssm.getMapping().getTo()))
+        {
+          resolveAndAddDatasetSeq(ssm.getMapping().getTo(), seqs, false);
+        }
       }
     }
-
-    dataset = new Alignment(seqs);
+    // finally construct dataset
+    dataset = new Alignment(seqs.toArray(new SequenceI[seqs.size()]));
     // move mappings to the dataset alignment
     dataset.codonFrameList = this.codonFrameList;
     this.codonFrameList = null;