Protect against NPE if URI does not parse with a host component

Author: apurtell

hbase-it/src/test/java/org/apache/hadoop/hbase/test/IntegrationTestLoadCommonCrawl.java

@@ -558,6 +558,10 @@ protected void map(LongWritable key, WARCWritable value, Context output)
             byte[] rowKey;
             try {
               rowKey = rowKeyFromTargetURI(warcHeader.getTargetURI());
+            } catch (IllegalArgumentException e) {
+              LOG.debug("URI for record " + warcHeader.getRecordID() +
+                " did not parse with a host component");
+              return;
             } catch (URISyntaxException e) {
               LOG.warn("Could not parse URI \"" + warcHeader.getTargetURI() + "\" for record " +
                 warcHeader.getRecordID());
@@ -612,17 +616,22 @@ protected void map(LongWritable key, WARCWritable value, Context output)
         }
       }
 
-      private byte[] rowKeyFromTargetURI(String targetURI) throws URISyntaxException {
+      private byte[] rowKeyFromTargetURI(String targetURI)
+          throws URISyntaxException, IllegalArgumentException {
         URI uri = new URI(targetURI);
         StringBuffer sb = new StringBuffer();
         // Ignore the scheme
         // Reverse the components of the hostname
-        String[] hostComponents = uri.getHost().split("\\.");
-        for (int i = hostComponents.length - 1; i >= 0; i--) {
-          sb.append(hostComponents[i]);
-          if (i != 0) {
-            sb.append('.');
+        if (uri.getHost() != null) {
+          String[] hostComponents = uri.getHost().split("\\.");
+          for (int i = hostComponents.length - 1; i >= 0; i--) {
+            sb.append(hostComponents[i]);
+            if (i != 0) {
+              sb.append('.');
+            }
           }
+        } else {
+          throw new IllegalArgumentException("URI is missing host component");
         }
         // Port
         if (uri.getPort() != -1) {