From cb79cecb722e9e3b17eb3013ed24f61c5fc6c800 Mon Sep 17 00:00:00 2001
From: gmungoc <g.m.carstairs@dundee.ac.uk>
Date: Tue, 18 Aug 2020 14:59:12 +0100
Subject: [PATCH] JAL-3692 parse multiline CDS location correctly

---
 src/jalview/io/EmblFlatFile.java      |   17 +++++++++++------
 test/jalview/io/EmblFlatFileTest.java |   10 +++++++++-
 test/jalview/io/J03321.embl.txt       |    4 +++-
 3 files changed, 23 insertions(+), 8 deletions(-)

diff --git a/src/jalview/io/EmblFlatFile.java b/src/jalview/io/EmblFlatFile.java
index 900aef8..bfae4ed 100644
--- a/src/jalview/io/EmblFlatFile.java
+++ b/src/jalview/io/EmblFlatFile.java
@@ -118,6 +118,7 @@ public class EmblFlatFile extends AlignFile // FileParse
    * 
    * @throws IOException
    */
+  @Override
   public void parse() throws IOException
   {
     String line = nextLine();
@@ -322,11 +323,14 @@ public class EmblFlatFile extends AlignFile // FileParse
       return nextLine();
     }
 
+    /*
+     * parse location - which may be over more than one line e.g. EAW51554
+     */
     CdsData data = new CdsData();
-    data.cdsLocation = tokens[2];
-    // TODO location can be over >1 line e.g. EAW51554
+    StringBuilder sb = new StringBuilder().append(tokens[2]);
+    line = parseFeatureQualifier(sb, "CDS");
+    data.cdsLocation = sb.toString();
 
-    line = nextLine();
     while (line != null)
     {
       if (!line.startsWith("FT    ")) // 4 spaces
@@ -359,7 +363,7 @@ public class EmblFlatFile extends AlignFile // FileParse
       String qualifier = line.substring(slashPos + 1, eqPos);
       String value = line.substring(eqPos + 1);
       value = removeQuotes(value);
-      StringBuilder sb = new StringBuilder().append(value);
+      sb = new StringBuilder().append(value);
       line = parseFeatureQualifier(sb, qualifier);
       String featureValue = sb.toString();
 
@@ -427,7 +431,7 @@ public class EmblFlatFile extends AlignFile // FileParse
    */
   static String removeQuotes(String value)
   {
-    if (value == null) 
+    if (value == null)
     {
       return null;
     }
@@ -493,7 +497,8 @@ public class EmblFlatFile extends AlignFile // FileParse
        * heuristic rule: most multi-line value (e.g. /product) are text,
        * so add a space for word boundary at a new line; not for translation
        */
-      if (!"translation".equals(qualifierName))
+      if (!"translation".equals(qualifierName)
+              && !"CDS".equals(qualifierName))
       {
         sb.append(" ");
       }
diff --git a/test/jalview/io/EmblFlatFileTest.java b/test/jalview/io/EmblFlatFileTest.java
index 2898a06..5d8ef21 100644
--- a/test/jalview/io/EmblFlatFileTest.java
+++ b/test/jalview/io/EmblFlatFileTest.java
@@ -3,9 +3,9 @@ package jalview.io;
 import static org.testng.Assert.assertEquals;
 import static org.testng.Assert.assertTrue;
 import static org.testng.AssertJUnit.assertNotNull;
+import static org.testng.AssertJUnit.assertNull;
 import static org.testng.AssertJUnit.assertSame;
 import static org.testng.AssertJUnit.fail;
-import static org.testng.AssertJUnit.assertNull;
 
 import java.io.File;
 import java.io.IOException;
@@ -14,8 +14,10 @@ import java.util.Arrays;
 import java.util.List;
 import java.util.Set;
 
+import org.testng.annotations.BeforeClass;
 import org.testng.annotations.Test;
 
+import jalview.bin.Cache;
 import jalview.datamodel.DBRefEntry;
 import jalview.datamodel.Mapping;
 import jalview.datamodel.SequenceFeature;
@@ -25,6 +27,12 @@ import jalview.util.MapList;
 
 public class EmblFlatFileTest
 {
+  @BeforeClass(alwaysRun = true)
+  public void setUp()
+  {
+    Cache.initLogger();
+  }
+
   /**
    * A fairly tough test, using J03321 (circular DNA), which has 8 CDS features,
    * one of them reverse strand
diff --git a/test/jalview/io/J03321.embl.txt b/test/jalview/io/J03321.embl.txt
index 92065b9..b0bc512 100644
--- a/test/jalview/io/J03321.embl.txt
+++ b/test/jalview/io/J03321.embl.txt
@@ -38,6 +38,7 @@ XX
 CC   Draft entry and computer-readable sequence kindly submitted by
 CC   G.Ratti, 28-MAR-1990.
 XX
+XX   ! first CDS location below split across two lines for test purposes !
 FH   Key             Location/Qualifiers
 FH
 FT   source          1..7502
@@ -48,7 +49,8 @@ FT                   /serotype="D"
 FT                   /mol_type="genomic DNA"
 FT                   /isolation_source="trachoma"
 FT                   /db_xref="taxon:813"
-FT   CDS             join(7022..7502,1..437)
+FT   CDS             join(7022..7502,
+FT                   1..437)
 FT                   /codon_start=1
 FT                   /transl_table=11
 FT                   /product="hypothetical protein"
-- 
1.7.10.2