JAL-1793 only decode specific special characters in VCF data

diff --git a/src/jalview/io/vcf/VCFLoader.java b/src/jalview/io/vcf/VCFLoader.java
index d461811..bb2948d 100644 (file)
--- a/src/jalview/io/vcf/VCFLoader.java
+++ b/src/jalview/io/vcf/VCFLoader.java
@@ -52,6 +52,16 @@ import htsjdk.variant.vcf.VCFInfoHeaderLine;
  */
 public class VCFLoader
 {
+  private static final String ENCODED_COMMA = "%2C";
+
+  private static final String ENCODED_PERCENT = "%25";
+
+  private static final String ENCODED_EQUALS = "%3D";
+
+  private static final String ENCODED_SEMICOLON = "%3B";
+
+  private static final String ENCODED_COLON = "%3A";
+
   private static final String UTF_8 = "UTF-8";
 
   private static final String DEFAULT_SPECIES = "homo_sapiens";
@@ -1231,22 +1241,42 @@ public class VCFLoader
       String value = getAttributeValue(variant, key, index);
       if (value != null)
       {
-        /*
-         * VCF spec requires encoding of special characters e.g. '='
-         * so decode them here before storing
-         */
-        try
-        {
-          value = URLDecoder.decode(value, UTF_8);
-        } catch (UnsupportedEncodingException e)
-        {
-        }
+        value = decodeSpecialCharacters(value);
         sf.setValue(key, value);
       }
     }
   }
 
   /**
+   * Decodes colon, semicolon, equals sign, percent sign, comma to their decoded
+   * form. The VCF specification (para 1.2) requires these to be encoded where not
+   * used with their special meaning in the VCF syntax. Note that general URL
+   * decoding should not be applied, since this would incorrectly decode (for
+   * example) a '+' sign.
+   * 
+   * @param value
+   * @return
+   */
+  protected static String decodeSpecialCharacters(String value)
+  {
+    /*
+     * avoid regex compilation if it is not needed!
+     */
+    if (!value.contains(ENCODED_COLON) && !value.contains(ENCODED_SEMICOLON)
+            && !value.contains(ENCODED_EQUALS)
+            && !value.contains(ENCODED_PERCENT)
+            && !value.contains(ENCODED_COMMA))
+    {
+      return value;
+    }
+
+    value = value.replace(ENCODED_COLON, ":")
+            .replace(ENCODED_SEMICOLON, ";").replace(ENCODED_EQUALS, "=")
+            .replace(ENCODED_PERCENT, "%").replace(ENCODED_COMMA, ",");
+    return value;
+  }
+
+  /**
    * Inspects CSQ data blocks (consequences) and adds attributes on the sequence
    * feature.
    * <p>

diff --git a/test/jalview/io/vcf/VCFLoaderTest.java b/test/jalview/io/vcf/VCFLoaderTest.java
index fb7a4e4..999fc6c 100644 (file)
--- a/test/jalview/io/vcf/VCFLoaderTest.java
+++ b/test/jalview/io/vcf/VCFLoaderTest.java
@@ -1,6 +1,7 @@
 package jalview.io.vcf;
 
 import static org.testng.Assert.assertEquals;
+import static org.testng.Assert.assertSame;
 import static org.testng.Assert.assertTrue;
 
 import jalview.bin.Cache;
@@ -733,4 +734,16 @@ public class VCFLoaderTest
     assertEquals(sf.getEnd(), 15);
     assertEquals(sf.getDescription(), "T,C");
   }
+
+  @Test(groups = "Functional")
+  public void testDecodeSpecialCharacters() throws IOException
+  {
+    String encoded = "hello world";
+    String decoded = VCFLoader.decodeSpecialCharacters(encoded);
+    assertSame(encoded, decoded); // no change needed
+
+    encoded = "ab%3Acd%3Bef%3Dgh%25ij%2Ckl%3A";
+    decoded = VCFLoader.decodeSpecialCharacters(encoded);
+    assertEquals(decoded, "ab:cd;ef=gh%ij,kl:");
+  }
 }
\ No newline at end of file