commit 10bb3d158eb6bed6e584ce79e1aa661369488d7e
author Mike Miller <mmiller@apache.org> Fri Jun 07 14:11:03 2019 -0400
committer Mike Miller <mmiller@apache.org> Fri Jun 07 14:11:03 2019 -0400
tree 6ad70c7033f17a3d6c336b4d3b634dbc366970ab
parent 134b78df77f45184f2193e6f5d822f1d99071f59

Minor fixes

ingest/pom.xml

diff --git a/ingest/pom.xml b/ingest/pom.xml
index 1f6bc99..a8a0a6c 100644
--- a/ingest/pom.xml
+++ b/ingest/pom.xml
@@ -84,8 +84,8 @@
             <phase>prepare-package</phase>
             <configuration>
               <outputDirectory>lib</outputDirectory>
-              &lt;!&ndash; just grab the non-provided runtime dependencies &ndash;&gt;
-              &lt;!&ndash; XXX we include guava at the same version as hadoop 2 provides so that we have it on hadoop 1 &ndash;&gt;
+	      <!-- just grab the non-provided runtime dependencies -->
+	      <!-- XXX we include guava at the same version as hadoop 2 provides so that we have it on hadoop 1 -->
               <includeArtifactIds>commons-lang,guava,lucene-core,lucene-analyzers,lucene-wikipedia,protobuf-java,accumulo-core,hadoop-core,libthrift,zookeeper,commons-codec,accumulo-fate,accumulo-trace</includeArtifactIds>
               <excludeTransitive>false</excludeTransitive>
             </configuration>

ingest/src/main/java/org/apache/accumulo/examples/wikisearch/ingest/WikipediaMapper.java

diff --git a/ingest/src/main/java/org/apache/accumulo/examples/wikisearch/ingest/WikipediaMapper.java b/ingest/src/main/java/org/apache/accumulo/examples/wikisearch/ingest/WikipediaMapper.java
index c2fed03..63f1d42 100644
--- a/ingest/src/main/java/org/apache/accumulo/examples/wikisearch/ingest/WikipediaMapper.java
+++ b/ingest/src/main/java/org/apache/accumulo/examples/wikisearch/ingest/WikipediaMapper.java
@@ -226,13 +226,7 @@
   static Set<String> getTokens(Article article) {
     Set<String> tokenList = new HashSet<>();
     try (WikipediaTokenizer tok = new WikipediaTokenizer(new StringReader(article.getText()))) {
-      Attribute term = tok.addAttribute(Attribute.class);
-      while (tok.incrementToken()) {
-        String token = term.toString();
-        if (!StringUtils.isEmpty(token)) {
-          tokenList.add(token);
-        }
-      }
+      tokenList.add(tok.toString());
     } catch (IOException e) {
       log.error("Error tokenizing text", e);
     }

pom.xml

diff --git a/pom.xml b/pom.xml
index ba62cf0..e997c9f 100644
--- a/pom.xml
+++ b/pom.xml
@@ -375,7 +375,7 @@
             <phase>prepare-package</phase>
             <configuration>
               <outputDirectory>../../lib</outputDirectory>
-                &lt;!&ndash; just grab the non-provided runtime dependencies &ndash;&gt;
+                <!-- just grab the non-provided runtime dependencies -->
               <includeArtifactIds>commons-collections,commons-configuration,commons-io,commons-lang,jline,log4j,libthrift,commons-jci-core,commons-jci-fam,commons-logging,commons-logging-api</includeArtifactIds>
               <excludeGroupIds>accumulo</excludeGroupIds>
               <excludeTransitive>true</excludeTransitive>