JAL-2154 belt-and-braces patch:

* when dataset XML doc is read in, all vamsasSet sequences should be processed in sync with the ‘view’ JSeq entries containing start/end metadata
* features, dbrefs and pdbentrys are now added direct to dataset sequences, but old behaviour should be preserved for reading alignment views containing dbrefs that get propagated to dataset.

diff --git a/src/jalview/gui/Jalview2XML.java b/src/jalview/gui/Jalview2XML.java
index 9e312db..3125cba 100644 (file)
--- a/src/jalview/gui/Jalview2XML.java
+++ b/src/jalview/gui/Jalview2XML.java
@@ -793,37 +793,42 @@ public class Jalview2XML
 
     JSeq jseq;
     Set<String> calcIdSet = new HashSet<String>();
-
+    // record the set of vamsas sequence XML POJO we create.
+    HashMap<String,Sequence> vamsasSetIds = new HashMap<String,Sequence>(); 
     // SAVE SEQUENCES
     for (final SequenceI jds : rjal.getSequences())
     {
       final SequenceI jdatasq = jds.getDatasetSequence() == null ? jds
               : jds.getDatasetSequence();
       String id = seqHash(jds);
-
-      if (seqRefIds.get(id) != null)
-      {
-        // This happens for two reasons: 1. multiple views are being serialised.
-        // 2. the hashCode has collided with another sequence's code. This DOES
-        // HAPPEN! (PF00072.15.stk does this)
-        // JBPNote: Uncomment to debug writing out of files that do not read
-        // back in due to ArrayOutOfBoundExceptions.
-        // System.err.println("vamsasSeq backref: "+id+"");
-        // System.err.println(jds.getName()+"
-        // "+jds.getStart()+"-"+jds.getEnd()+" "+jds.getSequenceAsString());
-        // System.err.println("Hashcode: "+seqHash(jds));
-        // SequenceI rsq = (SequenceI) seqRefIds.get(id + "");
-        // System.err.println(rsq.getName()+"
-        // "+rsq.getStart()+"-"+rsq.getEnd()+" "+rsq.getSequenceAsString());
-        // System.err.println("Hashcode: "+seqHash(rsq));
-      }
-      else
-      {
-        vamsasSeq = createVamsasSequence(id, jds);
-        vamsasSet.addSequence(vamsasSeq);
-        seqRefIds.put(id, jds);
+      if (vamsasSetIds.get(id) == null)
+      {
+        if (seqRefIds.get(id) != null && !storeDS)
+        {
+          // This happens for two reasons: 1. multiple views are being
+          // serialised.
+          // 2. the hashCode has collided with another sequence's code. This
+          // DOES
+          // HAPPEN! (PF00072.15.stk does this)
+          // JBPNote: Uncomment to debug writing out of files that do not read
+          // back in due to ArrayOutOfBoundExceptions.
+          // System.err.println("vamsasSeq backref: "+id+"");
+          // System.err.println(jds.getName()+"
+          // "+jds.getStart()+"-"+jds.getEnd()+" "+jds.getSequenceAsString());
+          // System.err.println("Hashcode: "+seqHash(jds));
+          // SequenceI rsq = (SequenceI) seqRefIds.get(id + "");
+          // System.err.println(rsq.getName()+"
+          // "+rsq.getStart()+"-"+rsq.getEnd()+" "+rsq.getSequenceAsString());
+          // System.err.println("Hashcode: "+seqHash(rsq));
+        }
+        else
+        {
+          vamsasSeq = createVamsasSequence(id, jds);
+          vamsasSet.addSequence(vamsasSeq);
+          vamsasSetIds.put(id, vamsasSeq);
+          seqRefIds.put(id, jds);
+        }
       }
-
       jseq = new JSeq();
       jseq.setStart(jds.getStart());
       jseq.setEnd(jds.getEnd());
@@ -2814,15 +2819,28 @@ public class Jalview2XML
           { 
             System.err
                     .println("Warning JAL-2154 regression: updating start/end for sequence "
-                    + tmpSeq.toString());
+                            + tmpSeq.toString() + " to " + jseqs[i]);
           }
         } else {
           incompleteSeqs.remove(seqId);
         }
+        if (vamsasSeq[vi].getId().equals(seqId))
+        {
+          // most likely we are reading a dataset XML document so
+          // update from vamsasSeq section of XML for this sequence
+          tmpSeq.setName(vamsasSeq[vi].getName());
+          tmpSeq.setDescription(vamsasSeq[vi].getDescription());
+          tmpSeq.setSequence(vamsasSeq[vi].getSequence());
+          vi++;
+        }
+        else
+        {
+          // reading multiple views, so vamsasSeq set is a subset of JSeq
+          multipleView = true;
+        }
         tmpSeq.setStart(jseqs[i].getStart());
         tmpSeq.setEnd(jseqs[i].getEnd());
         tmpseqs.add(tmpSeq);
-        multipleView = true;
       }
       else
       {
@@ -2911,6 +2929,12 @@ public class Jalview2XML
     {
       // load sequence features, database references and any associated PDB
       // structures for the alignment
+      //
+      // prior to 2.10, this part would only be executed the first time a
+      // sequence was encountered, but not afterwards.
+      // now, for 2.10 projects, this is also done if the xml doc includes
+      // dataset sequences not actually present in any particular view.
+      //
       for (int i = 0; i < vamsasSeq.length; i++)
       {
         if (jseqs[i].getFeaturesCount() > 0)
@@ -2937,13 +2961,17 @@ public class Jalview2XML
               }
 
             }
-
-            al.getSequenceAt(i).getDatasetSequence().addSequenceFeature(sf);
+            // adds feature to datasequence's feature set (since Jalview 2.10)
+            al.getSequenceAt(i).addSequenceFeature(sf);
           }
         }
         if (vamsasSeq[i].getDBRefCount() > 0)
         {
-          addDBRefs(al.getSequenceAt(i).getDatasetSequence(), vamsasSeq[i]);
+          // adds dbrefs to datasequence's set (since Jalview 2.10)
+          addDBRefs(
+                  al.getSequenceAt(i).getDatasetSequence() == null ? al.getSequenceAt(i)
+                          : al.getSequenceAt(i).getDatasetSequence(),
+                  vamsasSeq[i]);
         }
         if (jseqs[i].getPdbidsCount() > 0)
         {
@@ -2976,7 +3004,15 @@ public class Jalview2XML
             }
             StructureSelectionManager.getStructureSelectionManager(
                     Desktop.instance).registerPDBEntry(entry);
-            al.getSequenceAt(i).getDatasetSequence().addPDBId(entry);
+            // adds PDBEntry to datasequence's set (since Jalview 2.10)
+            if (al.getSequenceAt(i).getDatasetSequence() != null)
+            {
+              al.getSequenceAt(i).getDatasetSequence().addPDBId(entry);
+            }
+            else
+            {
+              al.getSequenceAt(i).addPDBId(entry);
+            }
           }
         }
       }
@@ -3005,16 +3041,16 @@ public class Jalview2XML
             if (maps[m].getMapping() != null)
             {
               mapping = addMapping(maps[m].getMapping());
-            }
-            if (dnaseq != null && mapping.getTo() != null)
-            {
-              cf.addMap(dnaseq, mapping.getTo(), mapping.getMap());
-            }
-            else
-            {
-              // defer to later
-              frefedSequence.add(newAlcodMapRef(maps[m].getDnasq(), cf,
-                      mapping));
+              if (dnaseq != null && mapping.getTo() != null)
+              {
+                cf.addMap(dnaseq, mapping.getTo(), mapping.getMap());
+              }
+              else
+              {
+                // defer to later
+                frefedSequence.add(newAlcodMapRef(maps[m].getDnasq(), cf,
+                        mapping));
+              }
             }
           }
           al.addCodonFrame(cf);