Merge pull request iipc#12 from ukwa/master

Check WARC/GZ via file extension

src/main/java/org/archive/io/warc/WARCReaderFactory.java

@@ -100,12 +100,20 @@ public static ArchiveReader get(final String s, final InputStream is,
             atFirstRecord);
     }
 
+	/*
+	 * Note that the ARC companion does this differently, with quite a lot of duplication.
+	 * 
+	 * @see org.archive.io.arc.ARCReaderFactory.getArchiveReader(String, InputStream, boolean)
+	 */
     protected ArchiveReader getArchiveReader(final String f,
 			final InputStream is, final boolean atFirstRecord)
 			throws IOException {
-		// For now, assume stream is compressed. Later add test of input
-		// stream or handle exception thrown when figure not compressed stream.
-		return new CompressedWARCReader(f, is, atFirstRecord);
+    	// Check if it's compressed, based on file extension.
+    	if( f.endsWith(".gz") ) {
+    		return new CompressedWARCReader(f, is, atFirstRecord);
+    	} else {
+    		return new UncompressedWARCReader(f, is);
+    	}
 	}
 
     public static WARCReader get(final URL arcUrl, final long offset)

src/test/java/org/archive/io/warc/WARCReaderFactoryTest.java

@@ -0,0 +1,34 @@
+package org.archive.io.warc;
+
+import java.io.FileInputStream;
+import java.io.IOException;
+
+import org.archive.format.warc.WARCConstants;
+import org.archive.format.warc.WARCConstants.WARCRecordType;
+import org.archive.io.ArchiveReader;
+import org.archive.io.ArchiveRecord;
+
+import junit.framework.TestCase;
+
+public class WARCReaderFactoryTest extends TestCase {
+
+	// Test files:
+	String[] files = new String[] {
+			"src/test/resources/org/archive/format/gzip/IAH-urls-wget.warc.gz",
+			"src/test/resources/org/archive/format/warc/IAH-urls-wget.warc"
+	};
+
+	public void testGetStringInputstreamBoolean() throws IOException {
+		// Check the test files can be opened:
+		for( String file : files ) {
+			FileInputStream is = new FileInputStream(file);
+			ArchiveReader ar = WARCReaderFactory.get(file, is, true);
+			ArchiveRecord r = ar.get();
+			String type = (String) r.getHeader().getHeaderValue(WARCConstants.HEADER_KEY_TYPE);
+			// Check the first record comes out as a 'warcinfo' record.
+			assertEquals(WARCRecordType.warcinfo.name(), type);
+		}
+	}
+
+
+}