From 3aaa2e6af84ae474b22de8a97a04cdd44e946b6c Mon Sep 17 00:00:00 2001
From: gmungoc <g.m.carstairs@dundee.ac.uk>
Date: Thu, 1 Oct 2015 12:11:48 +0100
Subject: [PATCH] JAL-1499 tests with gaps; NSites = width

---
 src/jalview/io/MegaFile.java      |   38 ++++++++++++++++++-------------------
 test/jalview/io/MegaFileTest.java |   27 ++++++++++++++------------
 2 files changed, 34 insertions(+), 31 deletions(-)

diff --git a/src/jalview/io/MegaFile.java b/src/jalview/io/MegaFile.java
index 1126cff..88200f5 100644
--- a/src/jalview/io/MegaFile.java
+++ b/src/jalview/io/MegaFile.java
@@ -859,19 +859,23 @@ public class MegaFile extends AlignFile
           throws IOException
   {
     StringBuilder desc = new StringBuilder(256);
-    String line = getValue(firstDescriptionLine);
-    while (line != null)
+    desc.append(getValue(firstDescriptionLine));
+    if (!firstDescriptionLine.endsWith(SEMICOLON))
     {
-      if (line.endsWith(SEMICOLON))
+      String line = nextNonCommentLine();
+      while (line != null)
       {
-        desc.append(line.substring(0, line.length() - 1));
-        break;
-      }
-      else if (line.length() > 0)
-      {
-        desc.append(line).append(newline);
+        if (line.endsWith(SEMICOLON))
+        {
+          desc.append(line.substring(0, line.length() - 1));
+          break;
+        }
+        else if (line.length() > 0)
+        {
+          desc.append(line).append(newline);
+        }
+        line = nextNonCommentLine();
       }
-      line = nextNonCommentLine();
     }
     setAlignmentProperty(PROP_DESCRIPTION, desc.toString());
   }
@@ -988,10 +992,8 @@ public class MegaFile extends AlignFile
     String propertyValue = (String) al.getProperty(PROP_TITLE);
     if (propertyValue != null)
     {
-      sb.append(BANG).append(TITLE).append(SPACE)
-.append(propertyValue)
-              .append(SEMICOLON)
-              .append(newline);
+      sb.append(BANG).append(TITLE).append(SPACE).append(propertyValue)
+              .append(SEMICOLON).append(newline);
     }
     propertyValue = (String) al.getProperty(PROP_DESCRIPTION);
     if (propertyValue != null)
@@ -1017,14 +1019,12 @@ public class MegaFile extends AlignFile
             .append(newline);
     
     /*
-     * !Format NSeqs NSites
-     * NSites the length of any sequence (they should all be the same), excluding
-     * gaps?!?
+     * !Format NSeqs NSites (the length of sequences - they should all be the
+     * same - including gaps)
      */
     sb.append(INDENT).append(N_SEQS).append(EQUALS).append(al.getHeight());
-    SequenceI seq = al.getSequenceAt(0);
     sb.append(SPACE).append(N_SITES).append(EQUALS)
-            .append(seq.getEnd() - seq.getStart() + 1);
+            .append(String.valueOf(al.getWidth()));
     sb.append(newline);
 
     /*
diff --git a/test/jalview/io/MegaFileTest.java b/test/jalview/io/MegaFileTest.java
index 477f982..d868dcb 100644
--- a/test/jalview/io/MegaFileTest.java
+++ b/test/jalview/io/MegaFileTest.java
@@ -72,7 +72,7 @@ public class MegaFileTest
           + "#CPZANT  MNOPQR\n\n"
           + "#U456   KLMNOP\n";
 
-  // the 'fancy' format, different header format, bases in triplet groups
+  // interleaved with description, bases/gaps in triplet groups
   private static final String INTERLEAVED_WITH_DESCRIPTION = 
           "#MEGA\n"
           + "!Title Data with description;\n"
@@ -80,10 +80,10 @@ public class MegaFileTest
           + "!Description\n" 
           + "    Line one of description\n"
           + "    Line two of description;\n\n"
-          + "#U455   CGC GTA\n" 
-          + "#CPZANT ATC GGG\n\n"
-          + "#U455   CGA TTT\n" 
-          + "#CPZANT CAA TGC\n";
+          + "#U455   C-- GTA\n" 
+          + "#CPZANT ATC -G-\n\n"
+          + "#U455   CGA --T\n" 
+          + "#CPZANT CA- -GC\n";
 
   //@formatter:on
 
@@ -358,9 +358,9 @@ public class MegaFileTest
     assertEquals("Second sequence id wrong", "CPZANT", seqs.get(1)
             .getName());
     // check sequence data
-    assertEquals("First sequence data wrong", "CGCGTACGATTT", seqs.get(0)
+    assertEquals("First sequence data wrong", "C--GTACGA--T", seqs.get(0)
             .getSequenceAsString());
-    assertEquals("Second sequence data wrong", "ATCGGGCAATGC", seqs.get(1)
+    assertEquals("Second sequence data wrong", "ATC-G-CA--GC", seqs.get(1)
             .getSequenceAsString());
     assertTrue("File format is not flagged as interleaved",
             testee.isInterleaved());
@@ -431,12 +431,12 @@ public class MegaFileTest
          "    Line two of description;\n" +
          "!Format\n" +
          "    DataType=DNA CodeTable=Standard\n" +
-         "    NSeqs=2 NSites=12\n" +
+         "    NSeqs=2 NSites=12\n" + // NSites includes gaps
          "    Indel=- Identical=. Missing=?;\n\n" +
-         "#U455   CGC GTA [6]\n" +
-         "#CPZANT ATC GGG [6]\n\n" +
-         "#U455   CGA TTT [12]\n" +
-         "#CPZANT CAA TGC [12]\n";
+         "#U455   C-- GTA [6]\n" +
+         "#CPZANT ATC -G- [6]\n\n" +
+         "#U455   CGA --T [12]\n" +
+         "#CPZANT CA- -GC [12]\n";
     //@formatter:on
     assertEquals("Roundtrip didn't match", expected,
             formatted);
@@ -603,4 +603,7 @@ public class MegaFileTest
     assertTrue("File format is not flagged as interleaved",
             testee.isInterleaved());
   }
+
+  //@formatter:on
+
 }
-- 
1.7.10.2