VCF reader now creates the correct type of header line for each header type

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2423 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/utils/genotype/vcf/VCFFilterHeaderLine.java

@@ -26,6 +26,24 @@ public class VCFFilterHeaderLine extends VCFHeaderLine {
         mDescription = description;
     }
 
+    /**
+     * create a VCF info header line
+     *
+     * @param line   the header line
+     */
+    protected VCFFilterHeaderLine(String line) {
+        super("FILTER", "");
+        String[] pieces = line.split(",");
+        if ( pieces.length < 2 )
+            throw new IllegalArgumentException("There are too few values in the VCF FILTER header line: " + line);
+
+        mName = pieces[0];
+        mDescription = pieces[1];
+        // just in case there were some commas in the description
+        for (int i = 1; i < pieces.length; i++)
+            mDescription += "," + pieces[i];
+    }
+
     protected String makeStringRep() {
         return String.format("FILTER=%s,\"%s\"", mName, mDescription);
     }

java/src/org/broadinstitute/sting/utils/genotype/vcf/VCFFormatHeaderLine.java

@@ -37,6 +37,26 @@ public class VCFFormatHeaderLine extends VCFHeaderLine {
         mDescription = description;
     }
 
+    /**
+     * create a VCF format header line
+     *
+     * @param line   the header line
+     */
+    protected VCFFormatHeaderLine(String line) {
+        super("FORMAT", "");
+        String[] pieces = line.split(",");
+        if ( pieces.length < 4 )
+            throw new IllegalArgumentException("There are too few values in the VCF FORMAT header line: " + line);
+
+        mName = pieces[0];
+        mCount = Integer.valueOf(pieces[1]);
+        mType = INFO_TYPE.valueOf(pieces[2]);
+        mDescription = pieces[3];
+        // just in case there were some commas in the description
+        for (int i = 4; i < pieces.length; i++)
+            mDescription += "," + pieces[i];
+    }
+
     protected String makeStringRep() {
         return String.format("FORMAT=%s,%d,%s,\"%s\"", mName, mCount, mType.toString(), mDescription);
     }

java/src/org/broadinstitute/sting/utils/genotype/vcf/VCFInfoHeaderLine.java

@@ -37,6 +37,26 @@ public class VCFInfoHeaderLine extends VCFHeaderLine {
         mDescription = description;
     }
 
+    /**
+     * create a VCF info header line
+     *
+     * @param line   the header line
+     */
+    protected VCFInfoHeaderLine(String line) {
+        super("INFO", "");
+        String[] pieces = line.split(",");
+        if ( pieces.length < 4 )
+            throw new IllegalArgumentException("There are too few values in the VCF INFO header line: " + line);
+
+        mName = pieces[0];
+        mCount = Integer.valueOf(pieces[1]);
+        mType = INFO_TYPE.valueOf(pieces[2]);
+        mDescription = pieces[3];
+        // just in case there were some commas in the description
+        for (int i = 4; i < pieces.length; i++)
+            mDescription += "," + pieces[i];
+    }
+
     protected String makeStringRep() {
         return String.format("INFO=%s,%d,%s,\"%s\"", mName, mCount, mType.toString(), mDescription);
     }

java/src/org/broadinstitute/sting/utils/genotype/vcf/VCFReader.java

@@ -169,11 +169,19 @@ public class VCFReader implements Iterator<VCFRecord>, Iterable<VCFRecord> {
                     arrayIndex++;
                 }
             } else {
+                if ( str.startsWith("##INFO=") )
+                    metaData.add(new VCFInfoHeaderLine(str.substring(7)));
+                else if ( str.startsWith("##FILTER=") )
+                    metaData.add(new VCFFilterHeaderLine(str.substring(7)));
+                else if ( str.startsWith("##FORMAT=") )
+                    metaData.add(new VCFFormatHeaderLine(str.substring(7)));
+                else {
                     int equals = str.indexOf("=");
                     if ( equals != -1 )
                         metaData.add(new VCFHeaderLine(str.substring(2, equals), str.substring(equals+1)));
                 }
             }
+        }
 
         return new VCFHeader(metaData, auxTags);
     }

java/test/org/broadinstitute/sting/utils/genotype/vcf/VCFReaderTest.java

@@ -10,10 +10,7 @@ import org.junit.Test;
 import org.junit.BeforeClass;
 
 import java.io.*;
-import java.util.ArrayList;
+import java.util.*;
-import java.util.Iterator;
-import java.util.List;
-import java.util.Map;
 
 /** test the VCFReader class test */
 public class VCFReaderTest extends BaseTest {
@@ -206,6 +203,23 @@ public class VCFReaderTest extends BaseTest {
     public void testComplexExample() {
         VCFReader reader = new VCFReader(complexFile);
 
+        // test header
+        Set<VCFHeaderLine> headerLines = reader.getHeader().getMetaData();
+        int formatCount = 0;
+        int infoCount = 0;
+        int filterCount = 0;
+        for ( VCFHeaderLine line : headerLines ) {
+            if ( line instanceof VCFFormatHeaderLine )
+                formatCount++;
+            else if ( line instanceof VCFInfoHeaderLine )
+                infoCount++;
+            else if ( line instanceof VCFFilterHeaderLine )
+                filterCount++;
+        }
+        Assert.assertEquals(3, formatCount);
+        Assert.assertEquals(4, infoCount);
+        Assert.assertEquals(0, filterCount);
+
         // record #1: test dbsnp id
         if (!reader.hasNext()) Assert.fail("The reader should have a record");
         VCFRecord rec = reader.next();