fixed the text codec to enable empty strings TA:Z:"" as per latest h…

…ts-spec

* Added a test to show that htsJDK can handle empty strings as tag values.

* - fixed the text codec to enable empty strings
- testing that we can code and decode into various formats

- Enables the reading and writing of sam/bam/cram records with empty string tags as required by samtools/hts-specs#135

src/main/java/htsjdk/samtools/SAMRecord.java

@@ -1397,7 +1397,7 @@ protected void setAttribute(final short tag, final Object value) {
 
      * @deprecated
      * The attribute type and value checks have been moved directly into
-     * {@code SAMBinaryTagAndValue}.
+     * {@link SAMBinaryTagAndValue}.
      */
     @Deprecated
     protected static boolean isAllowedAttributeValue(final Object value) {

src/main/java/htsjdk/samtools/TextTagCodec.java

@@ -41,6 +41,7 @@
  * instance is used in multiple threads.
  */
 public class TextTagCodec {
+    // 3 fields for non-empty strings 2 fields if the string is empty.
     private static final int NUM_TAG_FIELDS = 3;
 
     /**
@@ -149,12 +150,12 @@ public String encodeUntypedTag(final String tagName, final Object value) {
      */
     public Map.Entry<String, Object> decode(final String tag) {
         final int numFields = StringUtil.splitConcatenateExcessTokens(tag, fields, ':');
-        if (numFields != TextTagCodec.NUM_TAG_FIELDS) {
+        if (numFields != TextTagCodec.NUM_TAG_FIELDS && numFields != TextTagCodec.NUM_TAG_FIELDS - 1) {
             throw new SAMFormatException("Not enough fields in tag '" + tag + "'");
         }
         final String key = fields[0];
         final String type = fields[1];
-        final String stringVal = fields[2];
+        final String stringVal = numFields == TextTagCodec.NUM_TAG_FIELDS ? fields[2] : "";
         final Object val = convertStringToObject(type, stringVal);
         return new Map.Entry<String, Object>() {
             public String getKey() {

src/test/java/htsjdk/samtools/SAMIntegerTagTest.java

@@ -98,6 +98,52 @@ public void testGetTypedAttributeMethods() throws Exception {
         Assert.assertEquals(rec.getIntegerAttribute(INTEGER_TAG).intValue(), 1);
     }
 
+
+    @DataProvider
+    public Object[][] formatsAndValues(){
+        return new Object[][]{
+                new Object[]{"sam","Hello World!"},
+                new Object[]{"bam","Hello World!"},
+                new Object[]{"cram","Hello World!"},
+                new Object[]{"cram",""},
+                new Object[]{"bam",""},
+                new Object[]{"sam",""},
+        };
+    }
+    /**
+     * Should be able to write empty and non-empty strings
+     */
+    @Test(dataProvider = "formatsAndValues")
+    public void testWriteAndReadStrings(final String format,final String value) throws Exception {
+        final SAMRecord rec = createSamRecord();
+        rec.setAttribute(STRING_TAG, value);
+        writeAndReadSamRecord(format, rec);
+        Assert.assertEquals(rec.getStringAttribute(STRING_TAG),value);
+    }
+
+
+    @DataProvider
+    public Object[][] formatsAndValues2(){
+        return new Object[][]{
+                new Object[]{"sam",'a'},
+                new Object[]{"bam",'a'},
+                new Object[]{"cram",'a'},
+                new Object[]{"cram",null},
+                new Object[]{"bam",null},
+                new Object[]{"sam",null},
+        };
+    }
+    /**
+     * Should be able to write empty and non-empty strings
+     */
+    @Test(dataProvider = "formatsAndValues2")
+    public void testWriteAndReadCharacters(final String format,final Character value) throws Exception {
+        final SAMRecord rec = createSamRecord();
+        rec.setAttribute(STRING_TAG, value);
+        writeAndReadSamRecord(format, rec);
+        Assert.assertEquals(rec.getCharacterAttribute(STRING_TAG),value);
+    }
+
     /**
      * Should be an exception if a typed attribute call is made for the wrong type.
      */

src/test/java/htsjdk/samtools/SAMRecordUnitTest.java

@@ -417,6 +417,19 @@ public void test_isAllowedAttributeDataType() {
         Assert.assertFalse(SAMRecord.isAllowedAttributeValue(new Long(Integer.MIN_VALUE - 1L)));
     }
 
+    @Test()
+    public void test_setAttribute_empty_string() {
+        final SAMFileHeader header = new SAMFileHeader();
+        final SAMRecord record = new SAMRecord(header);
+        Assert.assertNull(record.getStringAttribute(SAMTag.MD.name()));
+        record.setAttribute(SAMTag.MD.name(), "");
+        Assert.assertNotNull(record.getStringAttribute(SAMTag.MD.name()));
+        Assert.assertEquals(record.getStringAttribute(SAMTag.MD.name()),"");
+        record.setAttribute(SAMTag.MD.name(), null);
+        Assert.assertNull(record.getStringAttribute(SAMTag.MD.name()));
+    }
+
+
     @Test(expectedExceptions = IllegalArgumentException.class)
     public void test_setAttribute_unsigned_int_negative() {
         SAMFileHeader header = new SAMFileHeader();