Refactoring the code; also, now it prints continuously instead of potentially storing one long string.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1421 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/playground/gatk/walkers/fasta/FastaAlternateReferenceWalker.java

@@ -14,14 +14,13 @@ import java.util.Iterator;
 
 @WalkerName("FastaAlternateReferenceMaker")
 @Requires(value={DataSource.REFERENCE})
-public class FastaAlternateReferenceWalker extends RefWalker<Pair<GenomeLoc, String>, Pair<GenomeLoc, String>> {
+public class FastaAlternateReferenceWalker extends FastaReferenceWalker {
 
     @Argument(fullName="maskSNPs", shortName="mask", doc="print 'N' at SNP sites instead of the alternate allele", required=false)
     private Boolean MASK_SNPS = false;
     @Argument(fullName="outputSequenomFormat", shortName="sequenom", doc="output results in sequenom format (overrides 'maskSNPs' argument)", required=false)
     private Boolean SEQUENOM = false;
 
-	private StringBuffer sb = new StringBuffer();
     int deletionBasesRemaining = 0;
 
     public Pair<GenomeLoc, String> map(RefMetaDataTracker rodData, ReferenceContext ref, AlignmentContext context) {
@@ -54,45 +53,4 @@ public class FastaAlternateReferenceWalker extends RefWalker<Pair<GenomeLoc, Str
         // if we got here then we're just ref
         return new Pair<GenomeLoc, String>(context.getLocation(), refBase);
 	}
-
-    public Pair<GenomeLoc, String> reduceInit() {
-        return new Pair<GenomeLoc, String>(null, "");
-    }
-
-	public Pair<GenomeLoc, String> reduce(Pair<GenomeLoc, String> value, Pair<GenomeLoc, String> sum) {
-        // if there is no interval to the left, then this is the first one
-        if ( sum.first == null ) {
-            sum.first = value.first;
-            sum.second = value.second;
-        }
-        // if the intervals don't overlap, print out the leftmost one and start a new one
-        // (end of contig or new interval)
-        else if ( value.first.getStart() != sum.first.getStop() + 1 ) {
-            printFasta(sum.first, sum.second);
-            sum.first = value.first;
-            sum.second = value.second;
-        }
-        // otherwise, merge them
-        else {
-            sum.first = GenomeLocParser.setStop(sum.first,value.first.getStop());
-            sum.second = sum.second.concat(value.second);
-        }
-		return sum;
-	}
-
-    public void onTraversalDone(Pair<GenomeLoc, String> sum) {
-        if (sum.second != null)
-            printFasta(sum.first, sum.second);
-    }
-
-    private void printFasta(GenomeLoc loc, String s) {
-        out.println(">" + loc);
-        int lines = s.length() / 60;
-        int currentStart = 0;
-        for (int i=0; i < lines; i++) {
-            out.println(s.substring(currentStart, currentStart+60));
-            currentStart += 60;
-        }
-        out.println(s.substring(currentStart));
-    }
 }

java/src/org/broadinstitute/sting/playground/gatk/walkers/fasta/FastaReferenceWalker.java

@@ -12,50 +12,44 @@ import org.broadinstitute.sting.utils.Pair;
 // create a fasta sequence file from a reference and intervals
 
 @WalkerName("FastaReferenceMaker")
-public class FastaReferenceWalker extends RefWalker<Pair<GenomeLoc, Character>, Pair<GenomeLoc, String>> {
+public class FastaReferenceWalker extends RefWalker<Pair<GenomeLoc, String>, GenomeLoc> {
 
-	public Pair<GenomeLoc, Character> map(RefMetaDataTracker rodData, ReferenceContext ref, AlignmentContext context) {
+    protected FastaSequence fasta;
-        return new Pair<GenomeLoc, Character>(context.getLocation(), ref.getBase());
+
+    public void initialize() {
+        fasta = new FastaSequence(out);
     }
 
-    public Pair<GenomeLoc, String> reduceInit() {
+	public Pair<GenomeLoc, String> map(RefMetaDataTracker rodData, ReferenceContext ref, AlignmentContext context) {
-        return new Pair<GenomeLoc, String>(null, "");
+        return new Pair<GenomeLoc, String>(context.getLocation(), String.valueOf(ref.getBase()));
 	}
 
-	public Pair<GenomeLoc, String> reduce(Pair<GenomeLoc, Character> value, Pair<GenomeLoc, String> sum) {
+    public GenomeLoc reduceInit() {
+        return null;
+    }
+
+	public GenomeLoc reduce(Pair<GenomeLoc, String> value, GenomeLoc sum) {
         // if there is no interval to the left, then this is the first one
-        if ( sum.first == null ) {
+        if ( sum == null ) {
-            sum.first = value.first;
+            sum = value.first;
-            sum.second = value.second.toString();
+            fasta.append(value.second.toString());
         }
         // if the intervals don't overlap, print out the leftmost one and start a new one
         // (end of contig or new interval)
-        else if ( value.first.getStart() != sum.first.getStop() + 1 ) {
+        else if ( value.first.getStart() != sum.getStop() + 1 ) {
-            printFasta(sum.first, sum.second);
+            fasta.flush();
-            sum.first = value.first;
+            sum = value.first;
-            sum.second = value.second.toString();
+            fasta.append(value.second.toString());
         }
         // otherwise, merge them
         else {
-            sum.first = GenomeLocParser.setStop(sum.first,value.first.getStop());
+            sum = GenomeLocParser.setStop(sum, value.first.getStop());
-            sum.second = new String(sum.second + value.second);
+            fasta.append(value.second.toString());
         }
 		return sum;
 	}
 
-    public void onTraversalDone(Pair<GenomeLoc, String> sum) {
+    public void onTraversalDone(GenomeLoc sum) {
-        if (sum.second != null)
+        fasta.flush();
-            printFasta(sum.first, sum.second);
-    }
-
-    private void printFasta(GenomeLoc loc, String s) {
-        out.println(">" + loc);
-        int lines = s.length() / 60;
-        int currentStart = 0;
-        for (int i=0; i < lines; i++) {
-            out.println(s.substring(currentStart, currentStart+60));
-            currentStart += 60;
-        }
-        out.println(s.substring(currentStart));
     }
 }

java/src/org/broadinstitute/sting/playground/gatk/walkers/fasta/FastaSequence.java

@@ -0,0 +1,50 @@
+package org.broadinstitute.sting.playground.gatk.walkers.fasta;
+
+import org.broadinstitute.sting.utils.GenomeLoc;
+
+import java.io.PrintStream;
+
+// fasta sequence holder class
+
+public class FastaSequence {
+
+    private PrintStream out;
+    private StringBuffer sb = new StringBuffer();
+    private long sequenceCounter = 1;
+    private boolean printedHeader = false;
+
+	public FastaSequence(PrintStream out) {
+        this.out = out;
+    }
+
+    public void append(String s) {
+        sb.append(s);
+        printFasta(false);
+    }
+
+    public void flush() {
+        printFasta(true);
+        printedHeader = false;
+    }
+
+    private void printFasta(boolean printAll) {
+        if ( sb.length() == 0 || (!printAll && sb.length() < 60) )
+            return;
+        if ( !printedHeader ) {
+            out.println(">" + sequenceCounter++);
+            printedHeader = true;
+        }
+        int lines = sb.length() / 60;
+        int currentStart = 0;
+        for (int i=0; i < lines; i++) {
+            out.println(sb.substring(currentStart, currentStart+60));
+            currentStart += 60;
+        }
+        if ( printAll ) {
+            out.println(sb.substring(currentStart));
+            sb.setLength(0);
+        } else {
+            sb.delete(0, currentStart);
+        }
+    }
+}