blob: 9162bfa41e4f6bbb35d1fae6c34141cb1bc4b8c4 [file] [log] [blame]
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html lang="en">
<head>
<title>Source code</title>
<link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style">
</head>
<body>
<div class="sourceContainer">
<pre><span class="sourceLineNo">001</span>/**<a name="line.1"></a>
<span class="sourceLineNo">002</span> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.2"></a>
<span class="sourceLineNo">003</span> * or more contributor license agreements. See the NOTICE file<a name="line.3"></a>
<span class="sourceLineNo">004</span> * distributed with this work for additional information<a name="line.4"></a>
<span class="sourceLineNo">005</span> * regarding copyright ownership. The ASF licenses this file<a name="line.5"></a>
<span class="sourceLineNo">006</span> * to you under the Apache License, Version 2.0 (the<a name="line.6"></a>
<span class="sourceLineNo">007</span> * "License"); you may not use this file except in compliance<a name="line.7"></a>
<span class="sourceLineNo">008</span> * with the License. You may obtain a copy of the License at<a name="line.8"></a>
<span class="sourceLineNo">009</span> *<a name="line.9"></a>
<span class="sourceLineNo">010</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.10"></a>
<span class="sourceLineNo">011</span> *<a name="line.11"></a>
<span class="sourceLineNo">012</span> * Unless required by applicable law or agreed to in writing, software<a name="line.12"></a>
<span class="sourceLineNo">013</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.13"></a>
<span class="sourceLineNo">014</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.14"></a>
<span class="sourceLineNo">015</span> * See the License for the specific language governing permissions and<a name="line.15"></a>
<span class="sourceLineNo">016</span> * limitations under the License.<a name="line.16"></a>
<span class="sourceLineNo">017</span> */<a name="line.17"></a>
<span class="sourceLineNo">018</span>package org.apache.hadoop.hbase.mapred;<a name="line.18"></a>
<span class="sourceLineNo">019</span><a name="line.19"></a>
<span class="sourceLineNo">020</span>import static org.junit.Assert.assertEquals;<a name="line.20"></a>
<span class="sourceLineNo">021</span>import static org.junit.Assert.assertTrue;<a name="line.21"></a>
<span class="sourceLineNo">022</span><a name="line.22"></a>
<span class="sourceLineNo">023</span>import java.io.File;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import java.io.IOException;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import java.util.ArrayList;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import java.util.Iterator;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import java.util.List;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import java.util.Set;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.apache.hadoop.conf.Configuration;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.apache.hadoop.fs.FileUtil;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import org.apache.hadoop.hbase.HBaseClassTestRule;<a name="line.31"></a>
<span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.HBaseTestingUtility;<a name="line.32"></a>
<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.TableName;<a name="line.33"></a>
<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.client.Put;<a name="line.34"></a>
<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.client.Result;<a name="line.35"></a>
<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.client.Table;<a name="line.36"></a>
<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.37"></a>
<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.testclassification.LargeTests;<a name="line.38"></a>
<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.testclassification.MapReduceTests;<a name="line.39"></a>
<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.40"></a>
<span class="sourceLineNo">041</span>import org.apache.hadoop.mapred.JobClient;<a name="line.41"></a>
<span class="sourceLineNo">042</span>import org.apache.hadoop.mapred.JobConf;<a name="line.42"></a>
<span class="sourceLineNo">043</span>import org.apache.hadoop.mapred.MapReduceBase;<a name="line.43"></a>
<span class="sourceLineNo">044</span>import org.apache.hadoop.mapred.OutputCollector;<a name="line.44"></a>
<span class="sourceLineNo">045</span>import org.apache.hadoop.mapred.Reporter;<a name="line.45"></a>
<span class="sourceLineNo">046</span>import org.apache.hadoop.mapred.RunningJob;<a name="line.46"></a>
<span class="sourceLineNo">047</span>import org.junit.AfterClass;<a name="line.47"></a>
<span class="sourceLineNo">048</span>import org.junit.Assert;<a name="line.48"></a>
<span class="sourceLineNo">049</span>import org.junit.Before;<a name="line.49"></a>
<span class="sourceLineNo">050</span>import org.junit.BeforeClass;<a name="line.50"></a>
<span class="sourceLineNo">051</span>import org.junit.ClassRule;<a name="line.51"></a>
<span class="sourceLineNo">052</span>import org.junit.Test;<a name="line.52"></a>
<span class="sourceLineNo">053</span>import org.junit.experimental.categories.Category;<a name="line.53"></a>
<span class="sourceLineNo">054</span>import org.slf4j.Logger;<a name="line.54"></a>
<span class="sourceLineNo">055</span>import org.slf4j.LoggerFactory;<a name="line.55"></a>
<span class="sourceLineNo">056</span><a name="line.56"></a>
<span class="sourceLineNo">057</span>import org.apache.hbase.thirdparty.com.google.common.collect.ImmutableMap;<a name="line.57"></a>
<span class="sourceLineNo">058</span>import org.apache.hbase.thirdparty.com.google.common.collect.ImmutableSet;<a name="line.58"></a>
<span class="sourceLineNo">059</span><a name="line.59"></a>
<span class="sourceLineNo">060</span>@Category({MapReduceTests.class, LargeTests.class})<a name="line.60"></a>
<span class="sourceLineNo">061</span>public class TestTableMapReduceUtil {<a name="line.61"></a>
<span class="sourceLineNo">062</span><a name="line.62"></a>
<span class="sourceLineNo">063</span> @ClassRule<a name="line.63"></a>
<span class="sourceLineNo">064</span> public static final HBaseClassTestRule CLASS_RULE =<a name="line.64"></a>
<span class="sourceLineNo">065</span> HBaseClassTestRule.forClass(TestTableMapReduceUtil.class);<a name="line.65"></a>
<span class="sourceLineNo">066</span><a name="line.66"></a>
<span class="sourceLineNo">067</span> private static final Logger LOG = LoggerFactory<a name="line.67"></a>
<span class="sourceLineNo">068</span> .getLogger(TestTableMapReduceUtil.class);<a name="line.68"></a>
<span class="sourceLineNo">069</span><a name="line.69"></a>
<span class="sourceLineNo">070</span> private static Table presidentsTable;<a name="line.70"></a>
<span class="sourceLineNo">071</span> private static final String TABLE_NAME = "People";<a name="line.71"></a>
<span class="sourceLineNo">072</span><a name="line.72"></a>
<span class="sourceLineNo">073</span> private static final byte[] COLUMN_FAMILY = Bytes.toBytes("info");<a name="line.73"></a>
<span class="sourceLineNo">074</span> private static final byte[] COLUMN_QUALIFIER = Bytes.toBytes("name");<a name="line.74"></a>
<span class="sourceLineNo">075</span><a name="line.75"></a>
<span class="sourceLineNo">076</span> private static ImmutableSet&lt;String&gt; presidentsRowKeys = ImmutableSet.of(<a name="line.76"></a>
<span class="sourceLineNo">077</span> "president1", "president2", "president3");<a name="line.77"></a>
<span class="sourceLineNo">078</span> private static Iterator&lt;String&gt; presidentNames = ImmutableSet.of(<a name="line.78"></a>
<span class="sourceLineNo">079</span> "John F. Kennedy", "George W. Bush", "Barack Obama").iterator();<a name="line.79"></a>
<span class="sourceLineNo">080</span><a name="line.80"></a>
<span class="sourceLineNo">081</span> private static ImmutableSet&lt;String&gt; actorsRowKeys = ImmutableSet.of("actor1",<a name="line.81"></a>
<span class="sourceLineNo">082</span> "actor2");<a name="line.82"></a>
<span class="sourceLineNo">083</span> private static Iterator&lt;String&gt; actorNames = ImmutableSet.of(<a name="line.83"></a>
<span class="sourceLineNo">084</span> "Jack Nicholson", "Martin Freeman").iterator();<a name="line.84"></a>
<span class="sourceLineNo">085</span><a name="line.85"></a>
<span class="sourceLineNo">086</span> private static String PRESIDENT_PATTERN = "president";<a name="line.86"></a>
<span class="sourceLineNo">087</span> private static String ACTOR_PATTERN = "actor";<a name="line.87"></a>
<span class="sourceLineNo">088</span> private static ImmutableMap&lt;String, ImmutableSet&lt;String&gt;&gt; relation = ImmutableMap<a name="line.88"></a>
<span class="sourceLineNo">089</span> .of(PRESIDENT_PATTERN, presidentsRowKeys, ACTOR_PATTERN, actorsRowKeys);<a name="line.89"></a>
<span class="sourceLineNo">090</span><a name="line.90"></a>
<span class="sourceLineNo">091</span> private static final HBaseTestingUtility UTIL = new HBaseTestingUtility();<a name="line.91"></a>
<span class="sourceLineNo">092</span><a name="line.92"></a>
<span class="sourceLineNo">093</span> @BeforeClass<a name="line.93"></a>
<span class="sourceLineNo">094</span> public static void beforeClass() throws Exception {<a name="line.94"></a>
<span class="sourceLineNo">095</span> UTIL.startMiniCluster();<a name="line.95"></a>
<span class="sourceLineNo">096</span> presidentsTable = createAndFillTable(TableName.valueOf(TABLE_NAME));<a name="line.96"></a>
<span class="sourceLineNo">097</span> }<a name="line.97"></a>
<span class="sourceLineNo">098</span><a name="line.98"></a>
<span class="sourceLineNo">099</span> @AfterClass<a name="line.99"></a>
<span class="sourceLineNo">100</span> public static void afterClass() throws Exception {<a name="line.100"></a>
<span class="sourceLineNo">101</span> UTIL.shutdownMiniCluster();<a name="line.101"></a>
<span class="sourceLineNo">102</span> }<a name="line.102"></a>
<span class="sourceLineNo">103</span><a name="line.103"></a>
<span class="sourceLineNo">104</span> @Before<a name="line.104"></a>
<span class="sourceLineNo">105</span> public void before() throws IOException {<a name="line.105"></a>
<span class="sourceLineNo">106</span> LOG.info("before");<a name="line.106"></a>
<span class="sourceLineNo">107</span> UTIL.ensureSomeRegionServersAvailable(1);<a name="line.107"></a>
<span class="sourceLineNo">108</span> LOG.info("before done");<a name="line.108"></a>
<span class="sourceLineNo">109</span> }<a name="line.109"></a>
<span class="sourceLineNo">110</span><a name="line.110"></a>
<span class="sourceLineNo">111</span> public static Table createAndFillTable(TableName tableName) throws IOException {<a name="line.111"></a>
<span class="sourceLineNo">112</span> Table table = UTIL.createTable(tableName, COLUMN_FAMILY);<a name="line.112"></a>
<span class="sourceLineNo">113</span> createPutCommand(table);<a name="line.113"></a>
<span class="sourceLineNo">114</span> return table;<a name="line.114"></a>
<span class="sourceLineNo">115</span> }<a name="line.115"></a>
<span class="sourceLineNo">116</span><a name="line.116"></a>
<span class="sourceLineNo">117</span> private static void createPutCommand(Table table) throws IOException {<a name="line.117"></a>
<span class="sourceLineNo">118</span> for (String president : presidentsRowKeys) {<a name="line.118"></a>
<span class="sourceLineNo">119</span> if (presidentNames.hasNext()) {<a name="line.119"></a>
<span class="sourceLineNo">120</span> Put p = new Put(Bytes.toBytes(president));<a name="line.120"></a>
<span class="sourceLineNo">121</span> p.addColumn(COLUMN_FAMILY, COLUMN_QUALIFIER, Bytes.toBytes(presidentNames.next()));<a name="line.121"></a>
<span class="sourceLineNo">122</span> table.put(p);<a name="line.122"></a>
<span class="sourceLineNo">123</span> }<a name="line.123"></a>
<span class="sourceLineNo">124</span> }<a name="line.124"></a>
<span class="sourceLineNo">125</span><a name="line.125"></a>
<span class="sourceLineNo">126</span> for (String actor : actorsRowKeys) {<a name="line.126"></a>
<span class="sourceLineNo">127</span> if (actorNames.hasNext()) {<a name="line.127"></a>
<span class="sourceLineNo">128</span> Put p = new Put(Bytes.toBytes(actor));<a name="line.128"></a>
<span class="sourceLineNo">129</span> p.addColumn(COLUMN_FAMILY, COLUMN_QUALIFIER, Bytes.toBytes(actorNames.next()));<a name="line.129"></a>
<span class="sourceLineNo">130</span> table.put(p);<a name="line.130"></a>
<span class="sourceLineNo">131</span> }<a name="line.131"></a>
<span class="sourceLineNo">132</span> }<a name="line.132"></a>
<span class="sourceLineNo">133</span> }<a name="line.133"></a>
<span class="sourceLineNo">134</span><a name="line.134"></a>
<span class="sourceLineNo">135</span> /**<a name="line.135"></a>
<span class="sourceLineNo">136</span> * Check what the given number of reduce tasks for the given job configuration<a name="line.136"></a>
<span class="sourceLineNo">137</span> * does not exceed the number of regions for the given table.<a name="line.137"></a>
<span class="sourceLineNo">138</span> */<a name="line.138"></a>
<span class="sourceLineNo">139</span> @Test<a name="line.139"></a>
<span class="sourceLineNo">140</span> public void shouldNumberOfReduceTaskNotExceedNumberOfRegionsForGivenTable()<a name="line.140"></a>
<span class="sourceLineNo">141</span> throws IOException {<a name="line.141"></a>
<span class="sourceLineNo">142</span> Assert.assertNotNull(presidentsTable);<a name="line.142"></a>
<span class="sourceLineNo">143</span> Configuration cfg = UTIL.getConfiguration();<a name="line.143"></a>
<span class="sourceLineNo">144</span> JobConf jobConf = new JobConf(cfg);<a name="line.144"></a>
<span class="sourceLineNo">145</span> TableMapReduceUtil.setNumReduceTasks(TABLE_NAME, jobConf);<a name="line.145"></a>
<span class="sourceLineNo">146</span> TableMapReduceUtil.limitNumReduceTasks(TABLE_NAME, jobConf);<a name="line.146"></a>
<span class="sourceLineNo">147</span> TableMapReduceUtil.setScannerCaching(jobConf, 100);<a name="line.147"></a>
<span class="sourceLineNo">148</span> assertEquals(1, jobConf.getNumReduceTasks());<a name="line.148"></a>
<span class="sourceLineNo">149</span> assertEquals(100, jobConf.getInt("hbase.client.scanner.caching", 0));<a name="line.149"></a>
<span class="sourceLineNo">150</span><a name="line.150"></a>
<span class="sourceLineNo">151</span> jobConf.setNumReduceTasks(10);<a name="line.151"></a>
<span class="sourceLineNo">152</span> TableMapReduceUtil.setNumMapTasks(TABLE_NAME, jobConf);<a name="line.152"></a>
<span class="sourceLineNo">153</span> TableMapReduceUtil.limitNumReduceTasks(TABLE_NAME, jobConf);<a name="line.153"></a>
<span class="sourceLineNo">154</span> assertEquals(1, jobConf.getNumReduceTasks());<a name="line.154"></a>
<span class="sourceLineNo">155</span> }<a name="line.155"></a>
<span class="sourceLineNo">156</span><a name="line.156"></a>
<span class="sourceLineNo">157</span> @Test<a name="line.157"></a>
<span class="sourceLineNo">158</span> public void shouldNumberOfMapTaskNotExceedNumberOfRegionsForGivenTable()<a name="line.158"></a>
<span class="sourceLineNo">159</span> throws IOException {<a name="line.159"></a>
<span class="sourceLineNo">160</span> Configuration cfg = UTIL.getConfiguration();<a name="line.160"></a>
<span class="sourceLineNo">161</span> JobConf jobConf = new JobConf(cfg);<a name="line.161"></a>
<span class="sourceLineNo">162</span> TableMapReduceUtil.setNumReduceTasks(TABLE_NAME, jobConf);<a name="line.162"></a>
<span class="sourceLineNo">163</span> TableMapReduceUtil.limitNumMapTasks(TABLE_NAME, jobConf);<a name="line.163"></a>
<span class="sourceLineNo">164</span> assertEquals(1, jobConf.getNumMapTasks());<a name="line.164"></a>
<span class="sourceLineNo">165</span><a name="line.165"></a>
<span class="sourceLineNo">166</span> jobConf.setNumMapTasks(10);<a name="line.166"></a>
<span class="sourceLineNo">167</span> TableMapReduceUtil.setNumMapTasks(TABLE_NAME, jobConf);<a name="line.167"></a>
<span class="sourceLineNo">168</span> TableMapReduceUtil.limitNumMapTasks(TABLE_NAME, jobConf);<a name="line.168"></a>
<span class="sourceLineNo">169</span> assertEquals(1, jobConf.getNumMapTasks());<a name="line.169"></a>
<span class="sourceLineNo">170</span> }<a name="line.170"></a>
<span class="sourceLineNo">171</span><a name="line.171"></a>
<span class="sourceLineNo">172</span> @Test<a name="line.172"></a>
<span class="sourceLineNo">173</span> @SuppressWarnings("deprecation")<a name="line.173"></a>
<span class="sourceLineNo">174</span> public void shoudBeValidMapReduceEvaluation() throws Exception {<a name="line.174"></a>
<span class="sourceLineNo">175</span> Configuration cfg = UTIL.getConfiguration();<a name="line.175"></a>
<span class="sourceLineNo">176</span> JobConf jobConf = new JobConf(cfg);<a name="line.176"></a>
<span class="sourceLineNo">177</span> try {<a name="line.177"></a>
<span class="sourceLineNo">178</span> jobConf.setJobName("process row task");<a name="line.178"></a>
<span class="sourceLineNo">179</span> jobConf.setNumReduceTasks(1);<a name="line.179"></a>
<span class="sourceLineNo">180</span> TableMapReduceUtil.initTableMapJob(TABLE_NAME, new String(COLUMN_FAMILY),<a name="line.180"></a>
<span class="sourceLineNo">181</span> ClassificatorMapper.class, ImmutableBytesWritable.class, Put.class,<a name="line.181"></a>
<span class="sourceLineNo">182</span> jobConf);<a name="line.182"></a>
<span class="sourceLineNo">183</span> TableMapReduceUtil.initTableReduceJob(TABLE_NAME,<a name="line.183"></a>
<span class="sourceLineNo">184</span> ClassificatorRowReduce.class, jobConf);<a name="line.184"></a>
<span class="sourceLineNo">185</span> RunningJob job = JobClient.runJob(jobConf);<a name="line.185"></a>
<span class="sourceLineNo">186</span> assertTrue(job.isSuccessful());<a name="line.186"></a>
<span class="sourceLineNo">187</span> } finally {<a name="line.187"></a>
<span class="sourceLineNo">188</span> if (jobConf != null)<a name="line.188"></a>
<span class="sourceLineNo">189</span> FileUtil.fullyDelete(new File(jobConf.get("hadoop.tmp.dir")));<a name="line.189"></a>
<span class="sourceLineNo">190</span> }<a name="line.190"></a>
<span class="sourceLineNo">191</span> }<a name="line.191"></a>
<span class="sourceLineNo">192</span><a name="line.192"></a>
<span class="sourceLineNo">193</span> @Test<a name="line.193"></a>
<span class="sourceLineNo">194</span> @SuppressWarnings("deprecation")<a name="line.194"></a>
<span class="sourceLineNo">195</span> public void shoudBeValidMapReduceWithPartitionerEvaluation()<a name="line.195"></a>
<span class="sourceLineNo">196</span> throws IOException {<a name="line.196"></a>
<span class="sourceLineNo">197</span> Configuration cfg = UTIL.getConfiguration();<a name="line.197"></a>
<span class="sourceLineNo">198</span> JobConf jobConf = new JobConf(cfg);<a name="line.198"></a>
<span class="sourceLineNo">199</span> try {<a name="line.199"></a>
<span class="sourceLineNo">200</span> jobConf.setJobName("process row task");<a name="line.200"></a>
<span class="sourceLineNo">201</span> jobConf.setNumReduceTasks(2);<a name="line.201"></a>
<span class="sourceLineNo">202</span> TableMapReduceUtil.initTableMapJob(TABLE_NAME, new String(COLUMN_FAMILY),<a name="line.202"></a>
<span class="sourceLineNo">203</span> ClassificatorMapper.class, ImmutableBytesWritable.class, Put.class,<a name="line.203"></a>
<span class="sourceLineNo">204</span> jobConf);<a name="line.204"></a>
<span class="sourceLineNo">205</span><a name="line.205"></a>
<span class="sourceLineNo">206</span> TableMapReduceUtil.initTableReduceJob(TABLE_NAME,<a name="line.206"></a>
<span class="sourceLineNo">207</span> ClassificatorRowReduce.class, jobConf, HRegionPartitioner.class);<a name="line.207"></a>
<span class="sourceLineNo">208</span> RunningJob job = JobClient.runJob(jobConf);<a name="line.208"></a>
<span class="sourceLineNo">209</span> assertTrue(job.isSuccessful());<a name="line.209"></a>
<span class="sourceLineNo">210</span> } finally {<a name="line.210"></a>
<span class="sourceLineNo">211</span> if (jobConf != null)<a name="line.211"></a>
<span class="sourceLineNo">212</span> FileUtil.fullyDelete(new File(jobConf.get("hadoop.tmp.dir")));<a name="line.212"></a>
<span class="sourceLineNo">213</span> }<a name="line.213"></a>
<span class="sourceLineNo">214</span> }<a name="line.214"></a>
<span class="sourceLineNo">215</span><a name="line.215"></a>
<span class="sourceLineNo">216</span> @SuppressWarnings("deprecation")<a name="line.216"></a>
<span class="sourceLineNo">217</span> static class ClassificatorRowReduce extends MapReduceBase implements<a name="line.217"></a>
<span class="sourceLineNo">218</span> TableReduce&lt;ImmutableBytesWritable, Put&gt; {<a name="line.218"></a>
<span class="sourceLineNo">219</span><a name="line.219"></a>
<span class="sourceLineNo">220</span> @Override<a name="line.220"></a>
<span class="sourceLineNo">221</span> public void reduce(ImmutableBytesWritable key, Iterator&lt;Put&gt; values,<a name="line.221"></a>
<span class="sourceLineNo">222</span> OutputCollector&lt;ImmutableBytesWritable, Put&gt; output, Reporter reporter)<a name="line.222"></a>
<span class="sourceLineNo">223</span> throws IOException {<a name="line.223"></a>
<span class="sourceLineNo">224</span> String strKey = Bytes.toString(key.get());<a name="line.224"></a>
<span class="sourceLineNo">225</span> List&lt;Put&gt; result = new ArrayList&lt;&gt;();<a name="line.225"></a>
<span class="sourceLineNo">226</span> while (values.hasNext())<a name="line.226"></a>
<span class="sourceLineNo">227</span> result.add(values.next());<a name="line.227"></a>
<span class="sourceLineNo">228</span><a name="line.228"></a>
<span class="sourceLineNo">229</span> if (relation.keySet().contains(strKey)) {<a name="line.229"></a>
<span class="sourceLineNo">230</span> Set&lt;String&gt; set = relation.get(strKey);<a name="line.230"></a>
<span class="sourceLineNo">231</span> if (set != null) {<a name="line.231"></a>
<span class="sourceLineNo">232</span> assertEquals(set.size(), result.size());<a name="line.232"></a>
<span class="sourceLineNo">233</span> } else {<a name="line.233"></a>
<span class="sourceLineNo">234</span> throwAccertionError("Test infrastructure error: set is null");<a name="line.234"></a>
<span class="sourceLineNo">235</span> }<a name="line.235"></a>
<span class="sourceLineNo">236</span> } else {<a name="line.236"></a>
<span class="sourceLineNo">237</span> throwAccertionError("Test infrastructure error: key not found in map");<a name="line.237"></a>
<span class="sourceLineNo">238</span> }<a name="line.238"></a>
<span class="sourceLineNo">239</span> }<a name="line.239"></a>
<span class="sourceLineNo">240</span><a name="line.240"></a>
<span class="sourceLineNo">241</span> private void throwAccertionError(String errorMessage) throws AssertionError {<a name="line.241"></a>
<span class="sourceLineNo">242</span> throw new AssertionError(errorMessage);<a name="line.242"></a>
<span class="sourceLineNo">243</span> }<a name="line.243"></a>
<span class="sourceLineNo">244</span> }<a name="line.244"></a>
<span class="sourceLineNo">245</span><a name="line.245"></a>
<span class="sourceLineNo">246</span> @SuppressWarnings("deprecation")<a name="line.246"></a>
<span class="sourceLineNo">247</span> static class ClassificatorMapper extends MapReduceBase implements<a name="line.247"></a>
<span class="sourceLineNo">248</span> TableMap&lt;ImmutableBytesWritable, Put&gt; {<a name="line.248"></a>
<span class="sourceLineNo">249</span><a name="line.249"></a>
<span class="sourceLineNo">250</span> @Override<a name="line.250"></a>
<span class="sourceLineNo">251</span> public void map(ImmutableBytesWritable row, Result result,<a name="line.251"></a>
<span class="sourceLineNo">252</span> OutputCollector&lt;ImmutableBytesWritable, Put&gt; outCollector,<a name="line.252"></a>
<span class="sourceLineNo">253</span> Reporter reporter) throws IOException {<a name="line.253"></a>
<span class="sourceLineNo">254</span> String rowKey = Bytes.toString(result.getRow());<a name="line.254"></a>
<span class="sourceLineNo">255</span> final ImmutableBytesWritable pKey = new ImmutableBytesWritable(<a name="line.255"></a>
<span class="sourceLineNo">256</span> Bytes.toBytes(PRESIDENT_PATTERN));<a name="line.256"></a>
<span class="sourceLineNo">257</span> final ImmutableBytesWritable aKey = new ImmutableBytesWritable(<a name="line.257"></a>
<span class="sourceLineNo">258</span> Bytes.toBytes(ACTOR_PATTERN));<a name="line.258"></a>
<span class="sourceLineNo">259</span> ImmutableBytesWritable outKey = null;<a name="line.259"></a>
<span class="sourceLineNo">260</span><a name="line.260"></a>
<span class="sourceLineNo">261</span> if (rowKey.startsWith(PRESIDENT_PATTERN)) {<a name="line.261"></a>
<span class="sourceLineNo">262</span> outKey = pKey;<a name="line.262"></a>
<span class="sourceLineNo">263</span> } else if (rowKey.startsWith(ACTOR_PATTERN)) {<a name="line.263"></a>
<span class="sourceLineNo">264</span> outKey = aKey;<a name="line.264"></a>
<span class="sourceLineNo">265</span> } else {<a name="line.265"></a>
<span class="sourceLineNo">266</span> throw new AssertionError("unexpected rowKey");<a name="line.266"></a>
<span class="sourceLineNo">267</span> }<a name="line.267"></a>
<span class="sourceLineNo">268</span><a name="line.268"></a>
<span class="sourceLineNo">269</span> String name = Bytes.toString(result.getValue(COLUMN_FAMILY,<a name="line.269"></a>
<span class="sourceLineNo">270</span> COLUMN_QUALIFIER));<a name="line.270"></a>
<span class="sourceLineNo">271</span> outCollector.collect(outKey,<a name="line.271"></a>
<span class="sourceLineNo">272</span> new Put(Bytes.toBytes("rowKey2"))<a name="line.272"></a>
<span class="sourceLineNo">273</span> .addColumn(COLUMN_FAMILY, COLUMN_QUALIFIER, Bytes.toBytes(name)));<a name="line.273"></a>
<span class="sourceLineNo">274</span> }<a name="line.274"></a>
<span class="sourceLineNo">275</span> }<a name="line.275"></a>
<span class="sourceLineNo">276</span>}<a name="line.276"></a>
</pre>
</div>
</body>
</html>