JAL-3171 record mapping between unmarshalled sequence’s dataset sequence ID and mater...

diff --git a/src/jalview/project/Jalview2XML.java b/src/jalview/project/Jalview2XML.java
index 4cc43d3..90c09d7 100644 (file)
--- a/src/jalview/project/Jalview2XML.java
+++ b/src/jalview/project/Jalview2XML.java
@@ -5220,6 +5220,20 @@ public class Jalview2XML
   {
     jalview.datamodel.AlignmentI ds = getDatasetFor(
             vamsasSet.getDatasetId());
+    AlignmentI xtant_ds = ds;
+    if (xtant_ds == null)
+    {
+      // good chance we are about to create a new dataset, but check if we've
+      // seen some of the dataset sequence IDs before.
+      // TODO: skip this check if we are working with project generated by
+      // version 2.11 or later
+      xtant_ds = checkIfHasDataset(vamsasSet.getSequence());
+      if (xtant_ds != null)
+      {
+        ds = xtant_ds;
+        addDatasetRef(vamsasSet.getDatasetId(), ds);
+      }
+    }
     Vector dseqs = null;
     if (!ignoreUnrefed)
     {
@@ -5231,6 +5245,13 @@ public class Jalview2XML
         {
           warn("JAL-3171 regression: Overwriting a dataset reference for an alignment"
                   + " - CDS/Protein crossreference data may be lost");
+          if (xtant_ds != null)
+          {
+            // This can only happen if the unique sequence set ID was bound to a
+            // dataset that did not contain any of the sequences in the view
+            // currently being restored.
+            warn("JAL-3171 SERIOUS!  TOTAL CONFUSION - please consider contacting the Jalview Development team so they can investigate why your project caused this message to be displayed.");
+          }
         }
         ds = seqSetDS;
         addDatasetRef(vamsasSet.getDatasetId(), ds);
@@ -5238,6 +5259,8 @@ public class Jalview2XML
     }
     if (ds == null)
     {
+      // try even harder to restore dataset
+      AlignmentI xtantDS = checkIfHasDataset(vamsasSet.getSequence());
       // create a list of new dataset sequences
       dseqs = new Vector();
     }
@@ -5263,9 +5286,55 @@ public class Jalview2XML
       // register dataset for the alignment's uniqueSeqSetId for legacy projects
       addDatasetRef(UNIQSEQSETID + uniqueSeqSetId, ds);
     }
+    updateSeqDatasetBinding(vamsasSet.getSequence(), ds);
+  }
+
+  /**
+   * XML dataset sequence ID to materialised dataset reference
+   */
+  HashMap<String, AlignmentI> seqToDataset = new HashMap<>();
+
+  /**
+   * @return the first materialised dataset reference containing a dataset
+   *         sequence referenced in the given view
+   * @param list
+   *          - sequences from the view
+   */
+  AlignmentI checkIfHasDataset(List<Sequence> list)
+  {
+    for (Sequence restoredSeq : list)
+    {
+      AlignmentI datasetFor = seqToDataset.get(restoredSeq.getDsseqid());
+      if (datasetFor != null)
+      {
+        return datasetFor;
+      }
+    }
+    return null;
   }
 
   /**
+   * Register ds as the containing dataset for the dataset sequences referenced
+   * by sequences in list
+   * 
+   * @param list
+   *          - sequences in a view
+   * @param ds
+   */
+  void updateSeqDatasetBinding(List<Sequence> list, AlignmentI ds)
+  {
+    for (Sequence restoredSeq : list)
+    {
+      AlignmentI prevDS = seqToDataset.put(restoredSeq.getDsseqid(), ds);
+      if (prevDS != null && prevDS != ds)
+      {
+        warn("Dataset sequence appears in many datasets: "
+                + restoredSeq.getDsseqid());
+        // TODO: try to merge!
+      }
+    }
+  }
+  /**
    * 
    * @param vamsasSeq
    *          sequence definition to create/merge dataset sequence for