print the current location in read walkers (in addition to the number of reads processed), along with some refactoring to support the change.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2006 348d0f76-0448-11de-a6fe-93d51630548a

java/src/org/broadinstitute/sting/gatk/traversals/TraversalEngine.java

@@ -21,6 +21,9 @@ public abstract class TraversalEngine {
     /** our log, which we want to capture anything from this class */
     protected static Logger logger = Logger.getLogger(TraversalEngine.class);
 
+    /** what kind of traversal we're undertaking.  This allows us to format output correctly */
+    public enum TRAVERSAL_TYPE { READ, LOCUS, LOCUS_WINDOW, DUPLICATE };
+
     /**
      * set the max number of iterations
      * @param maximumIterations the number of iterations
@@ -41,10 +44,10 @@ public abstract class TraversalEngine {
     /**
      * Forward request to printProgress
      *
-     * @param type the type of traversal
+     * @param type the TRAVERSAL_TYPE of the traversal
      * @param loc  the location
      */
-    public void printProgress(final String type, GenomeLoc loc) {
+    public void printProgress(final TRAVERSAL_TYPE type, GenomeLoc loc) {
         printProgress(false, type, loc);
     }
 
@@ -56,7 +59,7 @@ public abstract class TraversalEngine {
      * @param type      String to print out describing our atomic traversal type ("read", "locus", etc)
      * @param loc       Current location
      */
-    private void printProgress(boolean mustPrint, final String type, GenomeLoc loc) {
+    private void printProgress(boolean mustPrint, final TRAVERSAL_TYPE type, GenomeLoc loc) {
         final long nRecords = TraversalStatistics.nRecords;
         final long curTime = System.currentTimeMillis();
         final double elapsed = (curTime - startTime) / 1000.0;
@@ -65,12 +68,39 @@ public abstract class TraversalEngine {
         if (mustPrint || nRecords == 1 || nRecords % N_RECORDS_TO_PRINT == 0 || maxElapsedIntervalForPrinting(curTime)) {
             this.lastProgressPrintTime = curTime;
             final double secsPer1MReads = (elapsed * 1000000.0) / nRecords;
-            if (loc != null)
+            switch (type) {
-                logger.info(String.format("[PROGRESS] Traversed to %s, processing %,d %s in %.2f secs (%.2f secs per 1M %s)", loc, nRecords, type, elapsed, secsPer1MReads, type));
+                case LOCUS:
-            else
+                    logger.info(String.format("[PROGRESS] Traversed to %s, processing %,d loci in %.2f secs (%.2f secs per 1M loci)",
-                logger.info(String.format("[PROGRESS] Traversed %,d %s in %.2f secs (%.2f secs per 1M %s)", nRecords, type, elapsed, secsPer1MReads, type));
+                                loc,
-
+                                nRecords,
-        }
+                                elapsed,
+                                secsPer1MReads));break;
+                case READ:
+                    logger.info(String.format("[PROGRESS] Traversed %,d reads in %.2f secs %s(%.2f secs per 1M reads)",
+                                nRecords,
+                                elapsed,
+                                (loc != null) ? String.format("at location %s ",loc) : "",
+                                secsPer1MReads));break;
+                case DUPLICATE:
+                    logger.info(String.format("[PROGRESS] Traversed %,d dups in %.2f secs %s(%.2f secs per 1M dups)",
+                                nRecords,
+                                elapsed,
+                                (loc != null) ? String.format("at location %s ",loc) : "",
+                                secsPer1MReads));break;
+                case LOCUS_WINDOW:
+                    logger.info(String.format("[PROGRESS] Traversed %,d intervals in %.2f secs over interval %s (%.2f secs per 1M intervals)",
+                                nRecords,
+                                elapsed,
+                                loc,
+                                secsPer1MReads));break;
+                default:
+                    logger.info(String.format("[PROGRESS] Traversed %,d records in %.2f secs (%.2f secs per 1M intervals)%s",
+                                nRecords,
+                                elapsed,
+                                secsPer1MReads,
+                                (loc != null) ? String.format(", last location seen was %s",loc) : ""));
+            }
+        }   
     }
 
     /**
@@ -84,11 +114,11 @@ public abstract class TraversalEngine {
     /**
      * Called after a traversal to print out information about the traversal process
      *
-     * @param type String describing this type of traversal ("loci", "read")
+     * @param type TRAVERSAL_TYPE describing this type of traversal
      * @param sum  The reduce result of the traversal
      * @param <T>  ReduceType of the traversal
      */
-    protected <T> void printOnTraversalDone(final String type, T sum) {
+    protected <T> void printOnTraversalDone(final TRAVERSAL_TYPE type, T sum) {
         printProgress(true, type, null);
         logger.info("Traversal reduce result is " + sum);
         final long curTime = System.currentTimeMillis();

java/src/org/broadinstitute/sting/gatk/traversals/TraverseDuplicates.java

@@ -203,7 +203,7 @@ public class TraverseDuplicates extends TraversalEngine {
             if (duplicateReads.size() > 0)
                 sum = mapOne(dupWalker, uniqueReads, duplicateReads, site, refBases, locus, sum);
 
-            printProgress("dups", site);
+            printProgress(TRAVERSAL_TYPE.DUPLICATE, site);
 
             if (this.maximumIterations > 0 && TraversalStatistics.nRecords > this.maximumIterations) {
                 logger.warn(String.format(("Maximum number of duplicate sets encountered, terminating traversal " + TraversalStatistics.nRecords)));
@@ -309,6 +309,6 @@ public class TraverseDuplicates extends TraversalEngine {
      * @param <T> Type of the result.
      */
     public <T> void printOnTraversalDone(T sum) {
-        printOnTraversalDone("reads", sum);
+        printOnTraversalDone(TRAVERSAL_TYPE.DUPLICATE, sum);
     }
 }

java/src/org/broadinstitute/sting/gatk/traversals/TraverseLoci.java

@@ -86,7 +86,7 @@ public class TraverseLoci extends TraversalEngine {
                     break;
                 }
 
-                printProgress(UNIT_STRING, locus.getLocation());
+                printProgress(TRAVERSAL_TYPE.LOCUS, locus.getLocation());
             }
         }
 
@@ -113,7 +113,7 @@ public class TraverseLoci extends TraversalEngine {
      * @param <T> Type of the result.
      */
     public <T> void printOnTraversalDone( T sum ) {
-        printOnTraversalDone( UNIT_STRING, sum );
+        printOnTraversalDone( TRAVERSAL_TYPE.LOCUS, sum );
     }
 
     /**

java/src/org/broadinstitute/sting/gatk/traversals/TraverseLocusWindows.java

@@ -58,7 +58,7 @@ public class TraverseLocusWindows extends TraversalEngine {
             sum = locusWindowWalker.reduce(x, sum);
         }
 
-        printProgress("intervals", locus.getLocation());
+        printProgress(TRAVERSAL_TYPE.LOCUS_WINDOW, locus.getLocation());
 
         return sum;
     }
@@ -98,7 +98,7 @@ public class TraverseLocusWindows extends TraversalEngine {
      * @param <T> Type of the result.
      */
     public <T> void printOnTraversalDone( T sum ) {
-        printOnTraversalDone( "intervals", sum );
+        printOnTraversalDone( TRAVERSAL_TYPE.LOCUS_WINDOW, sum );
     }
 
 }

java/src/org/broadinstitute/sting/gatk/traversals/TraverseReads.java

@@ -121,7 +121,10 @@ public class TraverseReads extends TraversalEngine {
                 sum = readWalker.reduce(x, sum);
             }
 
-            printProgress("reads", null);
+            printProgress(TRAVERSAL_TYPE.READ,
+                          (read.getReferenceIndex() == SAMRecord.NO_ALIGNMENT_REFERENCE_INDEX) ?
+                                  null :
+                                  GenomeLocParser.createGenomeLoc(read.getReferenceIndex(),read.getAlignmentStart()));
         }
         return sum;
     }
@@ -133,6 +136,6 @@ public class TraverseReads extends TraversalEngine {
      * @param <T> Type of the result.
      */
     public <T> void printOnTraversalDone( T sum ) {
-        printOnTraversalDone( "reads", sum );
+        printOnTraversalDone(TRAVERSAL_TYPE.READ, sum );
     }
 }

java/src/org/broadinstitute/sting/utils/sam/ArtificialReadsTraversal.java

@@ -118,7 +118,7 @@ public class ArtificialReadsTraversal extends TraversalEngine {
                 sum = readWalker.reduce(x, sum);
             }
 
-            if (alignment != null) { printProgress("loci", alignment.getLocation()); }
+            if (alignment != null) { printProgress(TRAVERSAL_TYPE.READ, alignment.getLocation()); }
         }
         return sum;
     }
@@ -131,7 +131,7 @@ public class ArtificialReadsTraversal extends TraversalEngine {
      * @param <T> Type of the result.
      */
     public <T> void printOnTraversalDone( T sum ) {
-        printOnTraversalDone("reads", sum);
+        printOnTraversalDone(TRAVERSAL_TYPE.READ, sum);
     }
 
 

java/test/org/broadinstitute/sting/gatk/traversals/TraverseReadsTest.java

@@ -139,7 +139,7 @@ public class TraverseReadsTest extends BaseTest {
 
         }
 
-        traversalEngine.printOnTraversalDone("loci", accumulator);
+        traversalEngine.printOnTraversalDone(TraversalEngine.TRAVERSAL_TYPE.READ, accumulator);
         countReadWalker.onTraversalDone(accumulator);
 
         if (!(accumulator instanceof Integer)) {
@@ -185,7 +185,7 @@ public class TraverseReadsTest extends BaseTest {
             dataProvider.close();
         }
 
-        traversalEngine.printOnTraversalDone("loci", accumulator);
+        traversalEngine.printOnTraversalDone(TraversalEngine.TRAVERSAL_TYPE.READ, accumulator);
         countReadWalker.onTraversalDone(accumulator);
 
         if (!(accumulator instanceof Integer)) {