Imporve property handling in GFFRec

Author: lvaruzza

copy-test-data-to-hadoop.sh

@@ -14,10 +14,12 @@
 public class GFFParser {
 
 	public MapWritable readProperties(byte[] buff) {
+		System.out.println("|" + (new String(buff)) + "|");
 		MapWritable map = new MapWritable();
-		for(byte[] x: ByteArray.splitIterable(buff, (byte) ';', 0, buff.length)) {
-			byte[][] pair = ByteArray.split2(x, (byte) ' '); 
-			map.put(new Text(pair[0]), new Text(pair[1]));
+		for(byte[] x: ByteArray.splitIterable(buff, (byte) ';', 0, buff.length)) {			
+			byte[][] pair = ByteArray.split2(ByteArray.trim(x), (byte) ' '); 
+			System.out.println(new String(pair[0]) +  ":" + new String(pair[1]));
+			map.put(new Text(pair[0]), new Text(ByteArray.unquote(pair[1])));
 		}
 
 		return map;
@@ -45,7 +47,7 @@ public GFFRec parse(byte [] buff,int buffStart,int buffLen) {
 
 		ByteWritable strand = new ByteWritable(it.next()[0]);
 		ByteWritable frame = new ByteWritable(it.next()[0]);
-		MapWritable properties = readProperties(it.next());
+		MapWritable properties = readProperties(ByteArray.trim(it.next()));
 
 		return new GFFRec(seqname,source, feature,start,
 				end, score, strand,

src/main/java/com/lifetech/hadoop/alignment/GFFParser.java

@@ -85,9 +85,9 @@ public String toString() {
 
 		for( Entry<Writable,Writable> pair: properties.entrySet()) {
 			b.append(pair.getKey().toString());
-			b.append(' ');
+			b.append(" \"");
 			b.append(pair.getValue().toString());
-			b.append("; ");
+			b.append("\"; ");
 		}
 		return b.toString();
 	}
@@ -127,4 +127,8 @@ public ByteWritable getFrame() {
 	public MapWritable getProperties() {
 		return properties;
 	}
+	
+	public Text getProperty(Text name) {
+		return (Text)properties.get(name);
+	}
 }

src/main/java/com/lifetech/hadoop/alignment/GFFRec.java

@@ -6,6 +6,10 @@
 
 public class ByteArray {
 
+	/*
+	 * Split Iterator and Iterable 
+	 */
+	
 	static class SplitIterator implements Iterator<byte[]> {
 		private byte[] b; 
 		private byte c;
@@ -90,7 +94,11 @@ public static Iterator<byte[]> splitIterator(byte [] b,byte c,int start,int len)
 		return new SplitIterator(b,c,start,len);
 	}
 
-	public static void printBytes(byte[] x) {
+	/*
+	 *  =======================================================================================
+	 */
+	
+	static public void printBytes(byte[] x) {
 		ByteArray.printBytes(x,x.length);
 	}
 
@@ -99,4 +107,32 @@ static public void printBytes(byte[] x,int size) {
 			System.out.printf("\\x%x",x[i]);
 		}
 	}
+	
+	/*
+	 *  =======================================================================================
+	 */
+	
+	static public byte[] trim(byte[] b) {
+		if(b.length == 0) return b;
+		
+		int i=0;
+		int j=b.length-1;
+		for(;i<b.length & (b[i] == (byte)' ');i++) {}
+		for(;j>i & (b[j] == (byte)' ');j--) {  };
+		return Arrays.copyOfRange(b, i, j+1);
+	}
+	
+	static public byte[] unquote(byte[] b) {
+		if (b.length < 2) return b;
+		if (b[0] == (byte) '"' || b[0] == (byte) '\'') {
+			if (b[0] == b[b.length-1]) {
+				return Arrays.copyOfRange(b, 1, b.length-1);
+			} else {
+				return b;
+			}
+		} else {
+			return b;
+		}
+			
+	}
 }

src/main/java/com/lifetech/utils/ByteArray.java

@@ -42,7 +42,13 @@ public void testGFFParser() {
 		assertEquals('0',(char)gff.getFrame().get());
 		out.println((char)gff.getFrame().get());
 
+		out.println(gff.getProperty(new Text("gene_id")));
+		assertEquals(new Text("001"),gff.getProperty(new Text("gene_id")));
+
+		out.println(gff.getProperty(new Text("transcript_id")));
+		assertEquals(new Text("001"),gff.getProperty(new Text("transcript_id")));
+		
 		out.println(gff.toString());
-		assertEquals("AB000381\tTwinscan\tCDS\t380\t401\t.\t+\t0\t transcript_id \"001.1\"; gene_id \"001\"; ",gff.toString());
+		assertEquals("AB000381\tTwinscan\tCDS\t380\t401\t.\t+\t0\tgene_id \"001\"; transcript_id \"001.1\"; ",gff.toString());
 	}
 }

src/test/java/com/lifetech/hadoop/alignment/TestGFFRec.java

@@ -37,4 +37,29 @@ public void testSplitIterable() {
 		testSplitIterable0("1024","[1024]");
 		testSplitIterable0("1024 13","[1024, 13]");
 	}
+	
+	private void testTrim0(String target,String expected) {
+		byte[] b = target.getBytes();
+		byte[] r = ByteArray.trim(b);
+		assertEquals(expected,new String(r));
+	}
+	
+	@Test
+	public void testTrim() {
+		testTrim0("  ABC","ABC");
+		testTrim0("  ABC  ","ABC");
+	}
+	
+	private void testUnquote0(String target,String expected) {
+		byte[] b = target.getBytes();
+		byte[] r = ByteArray.unquote(b);
+		assertEquals(expected,new String(r));
+	}	
+
+	@Test
+	public void testUnquote() {
+		testUnquote0("\"abc\"","abc");
+		testUnquote0("\"\"","");
+		testUnquote0("\"abc","\"abc");
+	}
 }