From 4a85628075aa27923821a910758c9b9d69cb1708 Mon Sep 17 00:00:00 2001
From: gmungoc <g.m.carstairs@dundee.ac.uk>
Date: Fri, 23 Oct 2015 10:27:20 +0100
Subject: [PATCH] JAL-1499 slight refactoring of handling interleaved data
 blocks

---
 src/jalview/io/MegaFile.java      |   20 ++++++++++++++------
 test/jalview/io/MegaFileTest.java |   34 +++++++++++++++++++++++++++++++++-
 2 files changed, 47 insertions(+), 7 deletions(-)

diff --git a/src/jalview/io/MegaFile.java b/src/jalview/io/MegaFile.java
index 5a76d07..73cfe23 100644
--- a/src/jalview/io/MegaFile.java
+++ b/src/jalview/io/MegaFile.java
@@ -175,7 +175,7 @@ public class MegaFile extends AlignFile
   private Boolean nucleotide;
 
   // set once we have seen one block of interleaved data
-  private boolean firstDataBlockRead = false;
+  private boolean seenAllSequences = false;
 
   // this can be True, False or null (meaning we don't know yet)
   private Boolean interleaved;
@@ -284,11 +284,13 @@ public class MegaFile extends AlignFile
         if (upperCased.startsWith(BANG + GENE.toUpperCase())
                 || upperCased.startsWith(BANG + DOMAIN.toUpperCase()))
         {
+          endDataBlock();
           parseGeneOrDomain(dataLine);
         }
         else if (upperCased.startsWith(BANG + LABEL.toUpperCase()))
         {
           parseLabel(dataLine);
+          endDataBlock();
         }
         else
         {
@@ -300,7 +302,7 @@ public class MegaFile extends AlignFile
         /*
          * Blank line after processing some data...
          */
-        endOfDataBlock();
+        endDataBlock();
       }
       dataLine = nextNonCommentLine();
     }
@@ -410,10 +412,8 @@ public class MegaFile extends AlignFile
   /**
    * Post-processing after reading one block of interleaved data
    */
-  protected void endOfDataBlock()
+  protected void endDataBlock()
   {
-    this.firstDataBlockRead = true;
-
     padAnnotations(labelAnnotations);
   }
 
@@ -1029,6 +1029,14 @@ public class MegaFile extends AlignFile
       // and a placeholder for any SequenceFeature found
       sequenceFeatures.put(currentId, new ArrayList<SequenceFeature>());
     }
+    else
+    {
+      /*
+       * we are appending to a previously seen sequence; flag that we have seen
+       * all sequences
+       */
+      this.seenAllSequences = true;
+    }
     return sb;
   }
 
@@ -1049,7 +1057,7 @@ public class MegaFile extends AlignFile
     /*
      * New sequence found in second or later data block - error.
      */
-    if (this.firstDataBlockRead && !seqData.containsKey(seqId))
+    if (this.seenAllSequences && !seqData.containsKey(seqId))
     {
       throw new FileFormatException(
               "Parse error: misplaced new sequence starting at " + dataLine);
diff --git a/test/jalview/io/MegaFileTest.java b/test/jalview/io/MegaFileTest.java
index 4a3918e..860a898 100644
--- a/test/jalview/io/MegaFileTest.java
+++ b/test/jalview/io/MegaFileTest.java
@@ -74,7 +74,8 @@ public class MegaFileTest
           + "!TITLE Interleaved sequence data\n\n"
           + "#U455   ABCDEF\n" 
           + "#CPZANT  MNOPQR\n\n"
-          + "#U456   KLMNOP\n";
+          + "#U455   GHIJKL\n" 
+          + "#U456   KLMNOP\n"; // wossis?
 
   // interleaved with description, bases/gaps in triplet groups
   private static final String INTERLEAVED_WITH_DESCRIPTION = 
@@ -1019,4 +1020,35 @@ public class MegaFileTest
     assertEquals("Noncoding", MegaFile.getPropertyFromAnnotation(3, aa));
     assertEquals("Coding", MegaFile.getPropertyFromAnnotation(4, aa));
   }
+
+  //@formatter:on
+  
+  /**
+   * Test parse of interleaved data with no blank lines to separate blocks of
+   * sequence data; to confirm we can handle this correctly
+   * 
+   * @throws IOException
+   */
+  @Test(groups = { "Functional" })
+  public void testParse_interleaved_noBlankLines() throws IOException
+  {
+    String data = INTERLEAVED.replace("\n\n", "\n");
+    MegaFile testee = new MegaFile(data, AppletFormatAdapter.PASTE);
+    assertEquals("Title not as expected", "Interleaved sequence data",
+            testee.getAlignmentProperty(MegaFile.PROP_TITLE));
+    Vector<SequenceI> seqs = testee.getSeqs();
+    // should be 2 sequences
+    assertEquals("Expected two sequences", 2, seqs.size());
+    // check sequence names correct and order preserved
+    assertEquals("First sequence id wrong", "U455", seqs.get(0).getName());
+    assertEquals("Second sequence id wrong", "CPZANT", seqs.get(1)
+            .getName());
+    // check sequence data
+    assertEquals("First sequence data wrong", "ABCDEFKLMNOP", seqs.get(0)
+            .getSequenceAsString());
+    assertEquals("Second sequence data wrong", "MNOPQRWXYZGC", seqs.get(1)
+            .getSequenceAsString());
+    assertTrue("File format is not flagged as interleaved",
+            testee.isInterleaved());
+  }
 }
-- 
1.7.10.2