From: gmungoc <g.m.carstairs@dundee.ac.uk>
Date: Thu, 1 Oct 2015 10:46:33 +0000 (+0100)
Subject: JAL-1499 parse files with position number comments like [24]
X-Git-Url: http://source.jalview.org/gitweb/?a=commitdiff_plain;h=5413e3be3afeac58aa213af08b5c355981afc491;p=jalview.git

JAL-1499 parse files with position number comments like [24]
---

diff --git a/src/jalview/io/MegaFile.java b/src/jalview/io/MegaFile.java
index 2f5e35a..1126cff 100644
--- a/src/jalview/io/MegaFile.java
+++ b/src/jalview/io/MegaFile.java
@@ -252,8 +252,10 @@ public class MegaFile extends AlignFile
   }
 
   /**
-   * Returns the next line that is not a comment, or null at end of file.
-   * Comments in MEGA are within [ ] brackets, and may be nested.
+   * Returns the next non-comment line (or part line), or null at end of file.
+   * Comments in MEGA are within [ ] brackets, and may be nested. They may occur
+   * anywhere within a line (for example at the end with position numbers); this
+   * method returns the line with any comments removed.
    * 
    * @param depth
    *          current depth of nesting of comments while parsing
@@ -272,16 +274,6 @@ public class MegaFile extends AlignFile
       }
       return data;
     }
-    int leftBracket = data.indexOf(COMMENT_START);
-
-    /*
-     * reject unnested comment following data on the same line
-     */
-    if (depth == 0 && leftBracket > 0)
-    {
-      throw new FileFormatException(
-              "Can't parse comment following data at " + data);
-    }
 
     /*
      * If we are in a (possibly nested) comment after parsing this line, keep
@@ -295,15 +287,10 @@ public class MegaFile extends AlignFile
     else
     {
       /*
-       * not in a comment by end of this line; return what is left (or the next
-       * line if that is empty)
+       * not in a comment by end of this line; return what is left
        */
       String nonCommentPart = getNonCommentContent(data, depth);
-      // if (nonCommentPart.length() > 0)
-      // {
-        return nonCommentPart;
-      // }
-      // return nextNonCommentLine(0);
+      return nonCommentPart;
     }
   }
 
@@ -1215,10 +1202,19 @@ public class MegaFile extends AlignFile
   public String print(AlignmentI al)
   {
     this.nucleotide = al.isNucleotide();
+
     String lineLength = (String) al.getProperty(PROP_LINELENGTH);
     this.positionsPerLine = lineLength == null ? DEFAULT_LINE_LENGTH : Integer
             .parseInt(lineLength);
-    return printHeaders(al) + print(al.getSequencesArray());
+
+    String interleave = (String) al.getProperty(PROP_INTERLEAVED);
+    if (interleave != null)
+    {
+      this.interleaved = Boolean.valueOf(interleave);
+    }
+
+    String headers = printHeaders(al);
+    return headers + print(al.getSequencesArray());
   }
 
   /**
diff --git a/test/jalview/io/MegaFileTest.java b/test/jalview/io/MegaFileTest.java
index 309ffee..477f982 100644
--- a/test/jalview/io/MegaFileTest.java
+++ b/test/jalview/io/MegaFileTest.java
@@ -88,7 +88,7 @@ public class MegaFileTest
   //@formatter:on
 
   /**
-   * Test paste of interleaved mega format data.
+   * Test parse of interleaved mega format data.
    * 
    * @throws IOException
    */
@@ -115,7 +115,7 @@ public class MegaFileTest
   }
 
   /**
-   * Test paste of noninterleaved mega format data.
+   * Test parse of noninterleaved mega format data.
    * 
    * @throws IOException
    */
@@ -495,7 +495,7 @@ public class MegaFileTest
   //@formatter:on
   
   /**
-   * Test paste of interleaved mega format data where the identity character is
+   * Test parse of interleaved mega format data where the identity character is
    * used in sequences after the first
    * 
    * @throws IOException
@@ -531,7 +531,7 @@ public class MegaFileTest
   }
 
   /**
-   * Test paste of noninterleaved format data including identity symbol
+   * Test parse of noninterleaved format data including identity symbol
    * 
    * @throws IOException
    */
@@ -567,4 +567,40 @@ public class MegaFileTest
     assertFalse("File format is not flagged as noninterleaved",
             testee.isInterleaved());
   }
+
+  //@formatter:on
+  
+  /**
+   * Test parse of interleaved format data including position number comments.
+   * 
+   * @throws IOException
+   */
+  @Test(groups = { "Functional" })
+  public void testParse_interleavedWithPositionNumber() throws IOException
+  {
+    //@formatter:off
+    MegaFile testee = new MegaFile("#MEGA\n"+ 
+    "TITLE: Interleaved sequence data\n\n" + 
+    "#U455   ABCDEF [6]\n" + 
+    "#CPZANT  MNOPQR [6]\n\n" + 
+    "#U455   KLMNOP [12]\n" + 
+    "#CPZANT WXYZGC [12]\n", AppletFormatAdapter.PASTE);
+    //@formatter:on
+    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());
+  }
 }