From 7fd7bf631f5132771260baa81a957457ceb89694 Mon Sep 17 00:00:00 2001
From: gmungoc <g.m.carstairs@dundee.ac.uk>
Date: Thu, 8 Oct 2015 09:57:17 +0100
Subject: [PATCH] JAL-1499 convert unsupported gap character to '-'

---
 src/jalview/io/MegaFile.java      |   39 ++++++++++++++++++++++++++-----------
 test/jalview/io/MegaFileTest.java |   32 ++++++++++++++++++++++++++++++
 2 files changed, 60 insertions(+), 11 deletions(-)

diff --git a/src/jalview/io/MegaFile.java b/src/jalview/io/MegaFile.java
index 3096b60..97d7775 100644
--- a/src/jalview/io/MegaFile.java
+++ b/src/jalview/io/MegaFile.java
@@ -24,6 +24,7 @@ import jalview.datamodel.Annotation;
 import jalview.datamodel.Sequence;
 import jalview.datamodel.SequenceFeature;
 import jalview.datamodel.SequenceI;
+import jalview.util.Comparison;
 
 import java.io.IOException;
 import java.util.ArrayList;
@@ -143,6 +144,8 @@ public class MegaFile extends AlignFile
 
   private static final String TAB = "\t";
 
+  private static final char DEFAULT_GAP = '-';
+
   /*
    * number of sequence positions output per line
    */
@@ -151,7 +154,7 @@ public class MegaFile extends AlignFile
   private String title;
 
   // gap character may be explicitly declared, default is -
-  private char gapCharacter = '-';
+  private char gapCharacter = DEFAULT_GAP;
 
   // identity character if declared
   private char identityCharacter = 0;
@@ -219,7 +222,7 @@ public class MegaFile extends AlignFile
   @Override
   public void parse() throws IOException
   {
-    gapCharacter = '-';
+    gapCharacter = DEFAULT_GAP;
     sequenceFeatures = new HashMap<String, List<SequenceFeature>>();
     geneStart = new HashMap<String, Integer>();
     domainStart = new HashMap<String, Integer>();
@@ -954,18 +957,22 @@ public class MegaFile extends AlignFile
     for (int i = 0; i < formatted.length(); i++)
     {
       char nextChar = formatted.charAt(i);
-      if (nextChar != gapCharacter)
-      {
-        nonGapped++;
-      }
-      if (nextChar == identityCharacter
-              && len + i < referenceSequence.length())
+      if (nextChar == gapCharacter)
       {
-        sb1.append(referenceSequence.charAt(len + i));
+        sb1.append(Comparison.isGap(nextChar) ? nextChar : DEFAULT_GAP);
       }
       else
       {
-        sb1.append(nextChar);
+        nonGapped++;
+        if (nextChar == identityCharacter
+                && len + i < referenceSequence.length())
+        {
+          sb1.append(referenceSequence.charAt(len + i));
+        }
+        else
+        {
+          sb1.append(nextChar);
+        }
       }
     }
     formatted = sb1.toString();
@@ -1179,6 +1186,11 @@ public class MegaFile extends AlignFile
     else if (keyword.equalsIgnoreCase(INDEL))
     {
       this.gapCharacter = value.charAt(0);
+      if (!Comparison.isGap(gapCharacter))
+      {
+        System.err.println("Jalview doesn't support '" + gapCharacter
+                + "' for gaps, will be converted to '" + DEFAULT_GAP + "'");
+      }
     }
 
     else if (keyword.equalsIgnoreCase(IDENTICAL)
@@ -1604,9 +1616,14 @@ public class MegaFile extends AlignFile
   public void addProperties(AlignmentI al)
   {
     super.addProperties(al);
-    al.setGapCharacter(gapCharacter);
     
     /*
+     * record gap character specified, but convert to '-' if not one we support
+     */
+    al.setGapCharacter(Comparison.isGap(gapCharacter) ? gapCharacter
+            : DEFAULT_GAP);
+
+    /*
      * warn if e.g. DataType=DNA but data is protein (or vice versa)
      */
     if (this.nucleotide != null && this.nucleotide != al.isNucleotide()) {
diff --git a/test/jalview/io/MegaFileTest.java b/test/jalview/io/MegaFileTest.java
index 2b2422f..f7d83c0 100644
--- a/test/jalview/io/MegaFileTest.java
+++ b/test/jalview/io/MegaFileTest.java
@@ -13,6 +13,7 @@ import jalview.datamodel.SequenceFeature;
 import jalview.datamodel.SequenceI;
 
 import java.io.IOException;
+import java.util.List;
 import java.util.Vector;
 
 import org.testng.annotations.Test;
@@ -777,4 +778,35 @@ public class MegaFileTest
       verifySequenceFeature(sfs[2], "gene2", "Gene", 7, 12);
     }
   }
+
+  //@formatter:on
+  
+  /**
+   * Test case where the declared gap character is one Jalview does not support;
+   * it should be converted to a '-'
+   * 
+   * @throws IOException
+   */
+  @Test(groups = { "Functional" })
+  public void testParse_weirdGapCharacter() throws IOException
+  {
+    //@formatter:off
+    String data = "#MEGA\n"+ 
+    "!TITLE Interleaved sequence data;\n" +
+    "!Format Identical=. Indel=%;\n\n" +
+    "#U455 %BC%EF\n" + 
+    "#CPZANT M..P.R\n\n" + 
+    "#U455 KLMNOP\n" +
+    "#CPZANT .%%Z..";
+    AppletFormatAdapter fa = new AppletFormatAdapter();
+    AlignmentI al = fa.readFile(data,
+            AppletFormatAdapter.PASTE, "MEGA");
+    //@formatter:on
+    List<SequenceI> seqs = al.getSequences();
+    assertEquals("First sequence data wrong", "-BC-EFKLMNOP", seqs.get(0)
+            .getSequenceAsString());
+    assertEquals("Second sequence data wrong", "MBCPERK--ZOP", seqs.get(1)
+            .getSequenceAsString());
+    assertEquals('-', al.getGapCharacter());
+  }
 }
-- 
1.7.10.2