JAL-2738 query unindexed file, unit test for VCFLoader

[jalview.git] / src / jalview / ext / htsjdk / VCFReader.java

diff --git a/src/jalview/ext/htsjdk/VCFReader.java b/src/jalview/ext/htsjdk/VCFReader.java
index c5e09e0..14c057f 100644 (file)
--- a/src/jalview/ext/htsjdk/VCFReader.java
+++ b/src/jalview/ext/htsjdk/VCFReader.java
@@ -72,9 +72,10 @@ public class VCFReader implements Closeable, Iterable<VariantContext>
 
   /**
    * Queries for records overlapping the region specified. Note that this method
-   * requires a VCF file with an associated index. If no index exists a
-   * TribbleException will be thrown. Client code should call close() on the
-   * iterator when finished with it.
+   * is performant if the VCF file is indexed, and may be very slow if it is
+   * not.
+   * <p>
+   * Client code should call close() on the iterator when finished with it.
    * 
    * @param chrom
    *          the chromosome to query
@@ -87,7 +88,107 @@ public class VCFReader implements Closeable, Iterable<VariantContext>
   public CloseableIterator<VariantContext> query(final String chrom,
           final int start, final int end)
   {
-    return reader == null ? null : reader.query(chrom, start, end);
+   if (reader == null) {
+     return null;
+   }
+    if (indexed)
+    {
+      return reader.query(chrom, start, end);
+    }
+    else
+    {
+      return queryUnindexed(chrom, start, end);
+    }
+  }
+
+  /**
+   * Returns an iterator over variant records read from a flat file which
+   * overlap the specified chromosomal positions. Call close() on the iterator
+   * when finished with it!
+   * 
+   * @param chrom
+   * @param start
+   * @param end
+   * @return
+   */
+  protected CloseableIterator<VariantContext> queryUnindexed(
+          final String chrom, final int start, final int end)
+  {
+    final CloseableIterator<VariantContext> it = reader.iterator();
+    
+    return new CloseableIterator<VariantContext>()
+    {
+      boolean atEnd = false;
+
+      // prime look-ahead buffer with next matching record
+      private VariantContext next = findNext();
+
+      private VariantContext findNext()
+      {
+        if (atEnd)
+        {
+          return null;
+        }
+        VariantContext variant = null;
+        while (it.hasNext())
+        {
+          variant = it.next();
+          int vstart = variant.getStart();
+
+          if (vstart > end)
+          {
+            atEnd = true;
+            close();
+            return null;
+          }
+
+          int vend = variant.getEnd();
+          // todo what is the undeprecated way to get
+          // the chromosome for the variant?
+          if (chrom.equals(variant.getChr()) && (vstart <= end)
+                  && (vend >= start))
+          {
+            return variant;
+          }
+        }
+        return null;
+      }
+
+      @Override
+      public boolean hasNext()
+      {
+        boolean hasNext = !atEnd && (next != null);
+        if (!hasNext)
+        {
+          close();
+        }
+        return hasNext;
+      }
+
+      @Override
+      public VariantContext next()
+      {
+        /*
+         * return the next match, and then re-prime
+         * it with the following one (if any)
+         */
+        VariantContext temp = next;
+        next = findNext();
+        return temp;
+      }
+
+      @Override
+      public void remove()
+      {
+        // not implemented
+      }
+
+      @Override
+      public void close()
+      {
+        it.close();
+      }
+    };
   }
 
   /**
@@ -99,4 +200,15 @@ public class VCFReader implements Closeable, Iterable<VariantContext>
   {
     return reader == null ? null : reader.getFileHeader();
   }
+
+  /**
+   * Answers true if we are processing a tab-indexed VCF file, false if it is a
+   * plain text (uncompressed) file.
+   * 
+   * @return
+   */
+  public boolean isIndex()
+  {
+    return indexed;
+  }
 }