blob: aa6da30ef6a986a6088c6d5d2b2f8e890a38495a [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> *<a name="line.2"></a>
<span class="sourceLineNo">003</span> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.3"></a>
<span class="sourceLineNo">004</span> * or more contributor license agreements. See the NOTICE file<a name="line.4"></a>
<span class="sourceLineNo">005</span> * distributed with this work for additional information<a name="line.5"></a>
<span class="sourceLineNo">006</span> * regarding copyright ownership. The ASF licenses this file<a name="line.6"></a>
<span class="sourceLineNo">007</span> * to you under the Apache License, Version 2.0 (the<a name="line.7"></a>
<span class="sourceLineNo">008</span> * "License"); you may not use this file except in compliance<a name="line.8"></a>
<span class="sourceLineNo">009</span> * with the License. You may obtain a copy of the License at<a name="line.9"></a>
<span class="sourceLineNo">010</span> *<a name="line.10"></a>
<span class="sourceLineNo">011</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.11"></a>
<span class="sourceLineNo">012</span> *<a name="line.12"></a>
<span class="sourceLineNo">013</span> * Unless required by applicable law or agreed to in writing, software<a name="line.13"></a>
<span class="sourceLineNo">014</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.14"></a>
<span class="sourceLineNo">015</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.15"></a>
<span class="sourceLineNo">016</span> * See the License for the specific language governing permissions and<a name="line.16"></a>
<span class="sourceLineNo">017</span> * limitations under the License.<a name="line.17"></a>
<span class="sourceLineNo">018</span> */<a name="line.18"></a>
<span class="sourceLineNo">019</span>package org.apache.hadoop.hbase.mapreduce;<a name="line.19"></a>
<span class="sourceLineNo">020</span><a name="line.20"></a>
<span class="sourceLineNo">021</span>import java.io.IOException;<a name="line.21"></a>
<span class="sourceLineNo">022</span>import java.util.List;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import java.util.ArrayList;<a name="line.23"></a>
<span class="sourceLineNo">024</span><a name="line.24"></a>
<span class="sourceLineNo">025</span>import org.apache.commons.lang3.StringUtils;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.HConstants;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.util.AbstractHBaseTool;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import org.apache.hbase.thirdparty.com.google.common.base.Splitter;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.BasicParser;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLineParser;<a name="line.31"></a>
<span class="sourceLineNo">032</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.HelpFormatter;<a name="line.32"></a>
<span class="sourceLineNo">033</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.MissingOptionException;<a name="line.33"></a>
<span class="sourceLineNo">034</span>import org.apache.hbase.thirdparty.org.apache.commons.cli.Option;<a name="line.34"></a>
<span class="sourceLineNo">035</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.35"></a>
<span class="sourceLineNo">036</span>import org.slf4j.Logger;<a name="line.36"></a>
<span class="sourceLineNo">037</span>import org.slf4j.LoggerFactory;<a name="line.37"></a>
<span class="sourceLineNo">038</span>import org.apache.hadoop.conf.Configuration;<a name="line.38"></a>
<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.client.Result;<a name="line.39"></a>
<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.client.Scan;<a name="line.40"></a>
<span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.filter.FilterBase;<a name="line.41"></a>
<span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.filter.FirstKeyOnlyFilter;<a name="line.42"></a>
<span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.filter.MultiRowRangeFilter;<a name="line.43"></a>
<span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.44"></a>
<span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.45"></a>
<span class="sourceLineNo">046</span>import org.apache.hadoop.mapreduce.Counter;<a name="line.46"></a>
<span class="sourceLineNo">047</span>import org.apache.hadoop.mapreduce.Job;<a name="line.47"></a>
<span class="sourceLineNo">048</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.48"></a>
<span class="sourceLineNo">049</span><a name="line.49"></a>
<span class="sourceLineNo">050</span>/**<a name="line.50"></a>
<span class="sourceLineNo">051</span> * A job with a just a map phase to count rows. Map outputs table rows IF the<a name="line.51"></a>
<span class="sourceLineNo">052</span> * input row has columns that have content.<a name="line.52"></a>
<span class="sourceLineNo">053</span> */<a name="line.53"></a>
<span class="sourceLineNo">054</span>@InterfaceAudience.Public<a name="line.54"></a>
<span class="sourceLineNo">055</span>public class RowCounter extends AbstractHBaseTool {<a name="line.55"></a>
<span class="sourceLineNo">056</span><a name="line.56"></a>
<span class="sourceLineNo">057</span> private static final Logger LOG = LoggerFactory.getLogger(RowCounter.class);<a name="line.57"></a>
<span class="sourceLineNo">058</span><a name="line.58"></a>
<span class="sourceLineNo">059</span> /** Name of this 'program'. */<a name="line.59"></a>
<span class="sourceLineNo">060</span> static final String NAME = "rowcounter";<a name="line.60"></a>
<span class="sourceLineNo">061</span><a name="line.61"></a>
<span class="sourceLineNo">062</span> private final static String JOB_NAME_CONF_KEY = "mapreduce.job.name";<a name="line.62"></a>
<span class="sourceLineNo">063</span> private final static String EXPECTED_COUNT_KEY = RowCounter.class.getName() + ".expected_count";<a name="line.63"></a>
<span class="sourceLineNo">064</span><a name="line.64"></a>
<span class="sourceLineNo">065</span> private final static String OPT_START_TIME = "starttime";<a name="line.65"></a>
<span class="sourceLineNo">066</span> private final static String OPT_END_TIME = "endtime";<a name="line.66"></a>
<span class="sourceLineNo">067</span> private final static String OPT_RANGE = "range";<a name="line.67"></a>
<span class="sourceLineNo">068</span> private final static String OPT_EXPECTED_COUNT = "expectedCount";<a name="line.68"></a>
<span class="sourceLineNo">069</span><a name="line.69"></a>
<span class="sourceLineNo">070</span> private String tableName;<a name="line.70"></a>
<span class="sourceLineNo">071</span> private List&lt;MultiRowRangeFilter.RowRange&gt; rowRangeList;<a name="line.71"></a>
<span class="sourceLineNo">072</span> private long startTime;<a name="line.72"></a>
<span class="sourceLineNo">073</span> private long endTime;<a name="line.73"></a>
<span class="sourceLineNo">074</span> private long expectedCount;<a name="line.74"></a>
<span class="sourceLineNo">075</span> private List&lt;String&gt; columns = new ArrayList&lt;&gt;();<a name="line.75"></a>
<span class="sourceLineNo">076</span><a name="line.76"></a>
<span class="sourceLineNo">077</span> /**<a name="line.77"></a>
<span class="sourceLineNo">078</span> * Mapper that runs the count.<a name="line.78"></a>
<span class="sourceLineNo">079</span> */<a name="line.79"></a>
<span class="sourceLineNo">080</span> static class RowCounterMapper<a name="line.80"></a>
<span class="sourceLineNo">081</span> extends TableMapper&lt;ImmutableBytesWritable, Result&gt; {<a name="line.81"></a>
<span class="sourceLineNo">082</span><a name="line.82"></a>
<span class="sourceLineNo">083</span> /** Counter enumeration to count the actual rows. */<a name="line.83"></a>
<span class="sourceLineNo">084</span> public static enum Counters {ROWS}<a name="line.84"></a>
<span class="sourceLineNo">085</span><a name="line.85"></a>
<span class="sourceLineNo">086</span> /**<a name="line.86"></a>
<span class="sourceLineNo">087</span> * Maps the data.<a name="line.87"></a>
<span class="sourceLineNo">088</span> *<a name="line.88"></a>
<span class="sourceLineNo">089</span> * @param row The current table row key.<a name="line.89"></a>
<span class="sourceLineNo">090</span> * @param values The columns.<a name="line.90"></a>
<span class="sourceLineNo">091</span> * @param context The current context.<a name="line.91"></a>
<span class="sourceLineNo">092</span> * @throws IOException When something is broken with the data.<a name="line.92"></a>
<span class="sourceLineNo">093</span> * @see org.apache.hadoop.mapreduce.Mapper#map(Object, Object, Context)<a name="line.93"></a>
<span class="sourceLineNo">094</span> */<a name="line.94"></a>
<span class="sourceLineNo">095</span> @Override<a name="line.95"></a>
<span class="sourceLineNo">096</span> public void map(ImmutableBytesWritable row, Result values,<a name="line.96"></a>
<span class="sourceLineNo">097</span> Context context)<a name="line.97"></a>
<span class="sourceLineNo">098</span> throws IOException {<a name="line.98"></a>
<span class="sourceLineNo">099</span> // Count every row containing data, whether it's in qualifiers or values<a name="line.99"></a>
<span class="sourceLineNo">100</span> context.getCounter(Counters.ROWS).increment(1);<a name="line.100"></a>
<span class="sourceLineNo">101</span> }<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> /**<a name="line.104"></a>
<span class="sourceLineNo">105</span> * Sets up the actual job.<a name="line.105"></a>
<span class="sourceLineNo">106</span> *<a name="line.106"></a>
<span class="sourceLineNo">107</span> * @param conf The current configuration.<a name="line.107"></a>
<span class="sourceLineNo">108</span> * @return The newly created job.<a name="line.108"></a>
<span class="sourceLineNo">109</span> * @throws IOException When setting up the job fails.<a name="line.109"></a>
<span class="sourceLineNo">110</span> */<a name="line.110"></a>
<span class="sourceLineNo">111</span> public Job createSubmittableJob(Configuration conf) throws IOException {<a name="line.111"></a>
<span class="sourceLineNo">112</span> Job job = Job.getInstance(conf, conf.get(JOB_NAME_CONF_KEY, NAME + "_" + tableName));<a name="line.112"></a>
<span class="sourceLineNo">113</span> job.setJarByClass(RowCounter.class);<a name="line.113"></a>
<span class="sourceLineNo">114</span> Scan scan = new Scan();<a name="line.114"></a>
<span class="sourceLineNo">115</span> scan.setCacheBlocks(false);<a name="line.115"></a>
<span class="sourceLineNo">116</span> setScanFilter(scan, rowRangeList);<a name="line.116"></a>
<span class="sourceLineNo">117</span><a name="line.117"></a>
<span class="sourceLineNo">118</span> for (String columnName : this.columns) {<a name="line.118"></a>
<span class="sourceLineNo">119</span> String family = StringUtils.substringBefore(columnName, ":");<a name="line.119"></a>
<span class="sourceLineNo">120</span> String qualifier = StringUtils.substringAfter(columnName, ":");<a name="line.120"></a>
<span class="sourceLineNo">121</span> if (StringUtils.isBlank(qualifier)) {<a name="line.121"></a>
<span class="sourceLineNo">122</span> scan.addFamily(Bytes.toBytes(family));<a name="line.122"></a>
<span class="sourceLineNo">123</span> } else {<a name="line.123"></a>
<span class="sourceLineNo">124</span> scan.addColumn(Bytes.toBytes(family), Bytes.toBytes(qualifier));<a name="line.124"></a>
<span class="sourceLineNo">125</span> }<a name="line.125"></a>
<span class="sourceLineNo">126</span> }<a name="line.126"></a>
<span class="sourceLineNo">127</span><a name="line.127"></a>
<span class="sourceLineNo">128</span> if(this.expectedCount &gt;= 0) {<a name="line.128"></a>
<span class="sourceLineNo">129</span> conf.setLong(EXPECTED_COUNT_KEY, this.expectedCount);<a name="line.129"></a>
<span class="sourceLineNo">130</span> }<a name="line.130"></a>
<span class="sourceLineNo">131</span><a name="line.131"></a>
<span class="sourceLineNo">132</span> scan.setTimeRange(startTime, endTime);<a name="line.132"></a>
<span class="sourceLineNo">133</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.133"></a>
<span class="sourceLineNo">134</span> TableMapReduceUtil.initTableMapperJob(tableName, scan,<a name="line.134"></a>
<span class="sourceLineNo">135</span> RowCounterMapper.class, ImmutableBytesWritable.class, Result.class, job);<a name="line.135"></a>
<span class="sourceLineNo">136</span> job.setNumReduceTasks(0);<a name="line.136"></a>
<span class="sourceLineNo">137</span> return job;<a name="line.137"></a>
<span class="sourceLineNo">138</span> }<a name="line.138"></a>
<span class="sourceLineNo">139</span><a name="line.139"></a>
<span class="sourceLineNo">140</span> /**<a name="line.140"></a>
<span class="sourceLineNo">141</span> * Sets up the actual job.<a name="line.141"></a>
<span class="sourceLineNo">142</span> *<a name="line.142"></a>
<span class="sourceLineNo">143</span> * @param conf The current configuration.<a name="line.143"></a>
<span class="sourceLineNo">144</span> * @param args The command line parameters.<a name="line.144"></a>
<span class="sourceLineNo">145</span> * @return The newly created job.<a name="line.145"></a>
<span class="sourceLineNo">146</span> * @throws IOException When setting up the job fails.<a name="line.146"></a>
<span class="sourceLineNo">147</span> * @deprecated as of release 2.3.0. Will be removed on 4.0.0. Please use main method instead.<a name="line.147"></a>
<span class="sourceLineNo">148</span> */<a name="line.148"></a>
<span class="sourceLineNo">149</span> @Deprecated<a name="line.149"></a>
<span class="sourceLineNo">150</span> public static Job createSubmittableJob(Configuration conf, String[] args)<a name="line.150"></a>
<span class="sourceLineNo">151</span> throws IOException {<a name="line.151"></a>
<span class="sourceLineNo">152</span> String tableName = args[0];<a name="line.152"></a>
<span class="sourceLineNo">153</span> List&lt;MultiRowRangeFilter.RowRange&gt; rowRangeList = null;<a name="line.153"></a>
<span class="sourceLineNo">154</span> long startTime = 0;<a name="line.154"></a>
<span class="sourceLineNo">155</span> long endTime = 0;<a name="line.155"></a>
<span class="sourceLineNo">156</span><a name="line.156"></a>
<span class="sourceLineNo">157</span> StringBuilder sb = new StringBuilder();<a name="line.157"></a>
<span class="sourceLineNo">158</span><a name="line.158"></a>
<span class="sourceLineNo">159</span> final String rangeSwitch = "--range=";<a name="line.159"></a>
<span class="sourceLineNo">160</span> final String startTimeArgKey = "--starttime=";<a name="line.160"></a>
<span class="sourceLineNo">161</span> final String endTimeArgKey = "--endtime=";<a name="line.161"></a>
<span class="sourceLineNo">162</span> final String expectedCountArg = "--expected-count=";<a name="line.162"></a>
<span class="sourceLineNo">163</span><a name="line.163"></a>
<span class="sourceLineNo">164</span> // First argument is table name, starting from second<a name="line.164"></a>
<span class="sourceLineNo">165</span> for (int i = 1; i &lt; args.length; i++) {<a name="line.165"></a>
<span class="sourceLineNo">166</span> if (args[i].startsWith(rangeSwitch)) {<a name="line.166"></a>
<span class="sourceLineNo">167</span> try {<a name="line.167"></a>
<span class="sourceLineNo">168</span> rowRangeList = parseRowRangeParameter(<a name="line.168"></a>
<span class="sourceLineNo">169</span> args[i].substring(args[1].indexOf(rangeSwitch)+rangeSwitch.length()));<a name="line.169"></a>
<span class="sourceLineNo">170</span> } catch (IllegalArgumentException e) {<a name="line.170"></a>
<span class="sourceLineNo">171</span> return null;<a name="line.171"></a>
<span class="sourceLineNo">172</span> }<a name="line.172"></a>
<span class="sourceLineNo">173</span> continue;<a name="line.173"></a>
<span class="sourceLineNo">174</span> }<a name="line.174"></a>
<span class="sourceLineNo">175</span> if (args[i].startsWith(startTimeArgKey)) {<a name="line.175"></a>
<span class="sourceLineNo">176</span> startTime = Long.parseLong(args[i].substring(startTimeArgKey.length()));<a name="line.176"></a>
<span class="sourceLineNo">177</span> continue;<a name="line.177"></a>
<span class="sourceLineNo">178</span> }<a name="line.178"></a>
<span class="sourceLineNo">179</span> if (args[i].startsWith(endTimeArgKey)) {<a name="line.179"></a>
<span class="sourceLineNo">180</span> endTime = Long.parseLong(args[i].substring(endTimeArgKey.length()));<a name="line.180"></a>
<span class="sourceLineNo">181</span> continue;<a name="line.181"></a>
<span class="sourceLineNo">182</span> }<a name="line.182"></a>
<span class="sourceLineNo">183</span> if (args[i].startsWith(expectedCountArg)) {<a name="line.183"></a>
<span class="sourceLineNo">184</span> conf.setLong(EXPECTED_COUNT_KEY,<a name="line.184"></a>
<span class="sourceLineNo">185</span> Long.parseLong(args[i].substring(expectedCountArg.length())));<a name="line.185"></a>
<span class="sourceLineNo">186</span> continue;<a name="line.186"></a>
<span class="sourceLineNo">187</span> }<a name="line.187"></a>
<span class="sourceLineNo">188</span> // if no switch, assume column names<a name="line.188"></a>
<span class="sourceLineNo">189</span> sb.append(args[i]);<a name="line.189"></a>
<span class="sourceLineNo">190</span> sb.append(" ");<a name="line.190"></a>
<span class="sourceLineNo">191</span> }<a name="line.191"></a>
<span class="sourceLineNo">192</span> if (endTime &lt; startTime) {<a name="line.192"></a>
<span class="sourceLineNo">193</span> printUsage("--endtime=" + endTime + " needs to be greater than --starttime=" + startTime);<a name="line.193"></a>
<span class="sourceLineNo">194</span> return null;<a name="line.194"></a>
<span class="sourceLineNo">195</span> }<a name="line.195"></a>
<span class="sourceLineNo">196</span><a name="line.196"></a>
<span class="sourceLineNo">197</span> Job job = Job.getInstance(conf, conf.get(JOB_NAME_CONF_KEY, NAME + "_" + tableName));<a name="line.197"></a>
<span class="sourceLineNo">198</span> job.setJarByClass(RowCounter.class);<a name="line.198"></a>
<span class="sourceLineNo">199</span> Scan scan = new Scan();<a name="line.199"></a>
<span class="sourceLineNo">200</span> scan.setCacheBlocks(false);<a name="line.200"></a>
<span class="sourceLineNo">201</span> setScanFilter(scan, rowRangeList);<a name="line.201"></a>
<span class="sourceLineNo">202</span> if (sb.length() &gt; 0) {<a name="line.202"></a>
<span class="sourceLineNo">203</span> for (String columnName : sb.toString().trim().split(" ")) {<a name="line.203"></a>
<span class="sourceLineNo">204</span> String family = StringUtils.substringBefore(columnName, ":");<a name="line.204"></a>
<span class="sourceLineNo">205</span> String qualifier = StringUtils.substringAfter(columnName, ":");<a name="line.205"></a>
<span class="sourceLineNo">206</span><a name="line.206"></a>
<span class="sourceLineNo">207</span> if (StringUtils.isBlank(qualifier)) {<a name="line.207"></a>
<span class="sourceLineNo">208</span> scan.addFamily(Bytes.toBytes(family));<a name="line.208"></a>
<span class="sourceLineNo">209</span> }<a name="line.209"></a>
<span class="sourceLineNo">210</span> else {<a name="line.210"></a>
<span class="sourceLineNo">211</span> scan.addColumn(Bytes.toBytes(family), Bytes.toBytes(qualifier));<a name="line.211"></a>
<span class="sourceLineNo">212</span> }<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> scan.setTimeRange(startTime, endTime == 0 ? HConstants.LATEST_TIMESTAMP : endTime);<a name="line.215"></a>
<span class="sourceLineNo">216</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.216"></a>
<span class="sourceLineNo">217</span> TableMapReduceUtil.initTableMapperJob(tableName, scan,<a name="line.217"></a>
<span class="sourceLineNo">218</span> RowCounterMapper.class, ImmutableBytesWritable.class, Result.class, job);<a name="line.218"></a>
<span class="sourceLineNo">219</span> job.setNumReduceTasks(0);<a name="line.219"></a>
<span class="sourceLineNo">220</span> return job;<a name="line.220"></a>
<span class="sourceLineNo">221</span> }<a name="line.221"></a>
<span class="sourceLineNo">222</span><a name="line.222"></a>
<span class="sourceLineNo">223</span> /**<a name="line.223"></a>
<span class="sourceLineNo">224</span> * Prints usage without error message.<a name="line.224"></a>
<span class="sourceLineNo">225</span> * Note that we don't document --expected-count, because it's intended for test.<a name="line.225"></a>
<span class="sourceLineNo">226</span> */<a name="line.226"></a>
<span class="sourceLineNo">227</span> private static void printUsage(String errorMessage) {<a name="line.227"></a>
<span class="sourceLineNo">228</span> System.err.println("ERROR: " + errorMessage);<a name="line.228"></a>
<span class="sourceLineNo">229</span> System.err.println("Usage: hbase rowcounter [options] &lt;tablename&gt; "<a name="line.229"></a>
<span class="sourceLineNo">230</span> + "[--starttime=&lt;start&gt; --endtime=&lt;end&gt;] "<a name="line.230"></a>
<span class="sourceLineNo">231</span> + "[--range=[startKey],[endKey][;[startKey],[endKey]...]] [&lt;column1&gt; &lt;column2&gt;...]");<a name="line.231"></a>
<span class="sourceLineNo">232</span> System.err.println("For performance consider the following options:\n"<a name="line.232"></a>
<span class="sourceLineNo">233</span> + "-Dhbase.client.scanner.caching=100\n"<a name="line.233"></a>
<span class="sourceLineNo">234</span> + "-Dmapreduce.map.speculative=false");<a name="line.234"></a>
<span class="sourceLineNo">235</span> }<a name="line.235"></a>
<span class="sourceLineNo">236</span><a name="line.236"></a>
<span class="sourceLineNo">237</span> private static List&lt;MultiRowRangeFilter.RowRange&gt; parseRowRangeParameter(String arg) {<a name="line.237"></a>
<span class="sourceLineNo">238</span> final List&lt;String&gt; rangesSplit = Splitter.on(";").splitToList(arg);<a name="line.238"></a>
<span class="sourceLineNo">239</span> final List&lt;MultiRowRangeFilter.RowRange&gt; rangeList = new ArrayList&lt;&gt;();<a name="line.239"></a>
<span class="sourceLineNo">240</span> for (String range : rangesSplit) {<a name="line.240"></a>
<span class="sourceLineNo">241</span> if(range!=null &amp;&amp; !range.isEmpty()) {<a name="line.241"></a>
<span class="sourceLineNo">242</span> List&lt;String&gt; startEnd = Splitter.on(",").splitToList(range);<a name="line.242"></a>
<span class="sourceLineNo">243</span> if (startEnd.size() != 2 || startEnd.get(1).contains(",")) {<a name="line.243"></a>
<span class="sourceLineNo">244</span> throw new IllegalArgumentException("Wrong range specification: " + range);<a name="line.244"></a>
<span class="sourceLineNo">245</span> }<a name="line.245"></a>
<span class="sourceLineNo">246</span> String startKey = startEnd.get(0);<a name="line.246"></a>
<span class="sourceLineNo">247</span> String endKey = startEnd.get(1);<a name="line.247"></a>
<span class="sourceLineNo">248</span> rangeList.add(new MultiRowRangeFilter.RowRange(Bytes.toBytesBinary(startKey),<a name="line.248"></a>
<span class="sourceLineNo">249</span> true, Bytes.toBytesBinary(endKey), false));<a name="line.249"></a>
<span class="sourceLineNo">250</span> }<a name="line.250"></a>
<span class="sourceLineNo">251</span> }<a name="line.251"></a>
<span class="sourceLineNo">252</span> return rangeList;<a name="line.252"></a>
<span class="sourceLineNo">253</span> }<a name="line.253"></a>
<span class="sourceLineNo">254</span><a name="line.254"></a>
<span class="sourceLineNo">255</span> /**<a name="line.255"></a>
<span class="sourceLineNo">256</span> * Sets filter {@link FilterBase} to the {@link Scan} instance.<a name="line.256"></a>
<span class="sourceLineNo">257</span> * If provided rowRangeList contains more than one element,<a name="line.257"></a>
<span class="sourceLineNo">258</span> * method sets filter which is instance of {@link MultiRowRangeFilter}.<a name="line.258"></a>
<span class="sourceLineNo">259</span> * Otherwise, method sets filter which is instance of {@link FirstKeyOnlyFilter}.<a name="line.259"></a>
<span class="sourceLineNo">260</span> * If rowRangeList contains exactly one element, startRow and stopRow are set to the scan.<a name="line.260"></a>
<span class="sourceLineNo">261</span> * @param scan<a name="line.261"></a>
<span class="sourceLineNo">262</span> * @param rowRangeList<a name="line.262"></a>
<span class="sourceLineNo">263</span> */<a name="line.263"></a>
<span class="sourceLineNo">264</span> private static void setScanFilter(Scan scan, List&lt;MultiRowRangeFilter.RowRange&gt; rowRangeList) {<a name="line.264"></a>
<span class="sourceLineNo">265</span> final int size = rowRangeList == null ? 0 : rowRangeList.size();<a name="line.265"></a>
<span class="sourceLineNo">266</span> if (size &lt;= 1) {<a name="line.266"></a>
<span class="sourceLineNo">267</span> scan.setFilter(new FirstKeyOnlyFilter());<a name="line.267"></a>
<span class="sourceLineNo">268</span> }<a name="line.268"></a>
<span class="sourceLineNo">269</span> if (size == 1) {<a name="line.269"></a>
<span class="sourceLineNo">270</span> MultiRowRangeFilter.RowRange range = rowRangeList.get(0);<a name="line.270"></a>
<span class="sourceLineNo">271</span> scan.setStartRow(range.getStartRow()); //inclusive<a name="line.271"></a>
<span class="sourceLineNo">272</span> scan.setStopRow(range.getStopRow()); //exclusive<a name="line.272"></a>
<span class="sourceLineNo">273</span> } else if (size &gt; 1) {<a name="line.273"></a>
<span class="sourceLineNo">274</span> scan.setFilter(new MultiRowRangeFilter(rowRangeList));<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>
<span class="sourceLineNo">277</span><a name="line.277"></a>
<span class="sourceLineNo">278</span> @Override<a name="line.278"></a>
<span class="sourceLineNo">279</span> protected void printUsage() {<a name="line.279"></a>
<span class="sourceLineNo">280</span> StringBuilder footerBuilder = new StringBuilder();<a name="line.280"></a>
<span class="sourceLineNo">281</span> footerBuilder.append("For performance, consider the following configuration properties:\n");<a name="line.281"></a>
<span class="sourceLineNo">282</span> footerBuilder.append("-Dhbase.client.scanner.caching=100\n");<a name="line.282"></a>
<span class="sourceLineNo">283</span> footerBuilder.append("-Dmapreduce.map.speculative=false\n");<a name="line.283"></a>
<span class="sourceLineNo">284</span> printUsage("hbase rowcounter &lt;tablename&gt; [options] [&lt;column1&gt; &lt;column2&gt;...]",<a name="line.284"></a>
<span class="sourceLineNo">285</span> "Options:", footerBuilder.toString());<a name="line.285"></a>
<span class="sourceLineNo">286</span> }<a name="line.286"></a>
<span class="sourceLineNo">287</span><a name="line.287"></a>
<span class="sourceLineNo">288</span> @Override<a name="line.288"></a>
<span class="sourceLineNo">289</span> protected void printUsage(final String usageStr, final String usageHeader,<a name="line.289"></a>
<span class="sourceLineNo">290</span> final String usageFooter) {<a name="line.290"></a>
<span class="sourceLineNo">291</span> HelpFormatter helpFormatter = new HelpFormatter();<a name="line.291"></a>
<span class="sourceLineNo">292</span> helpFormatter.setWidth(120);<a name="line.292"></a>
<span class="sourceLineNo">293</span> helpFormatter.setOptionComparator(new AbstractHBaseTool.OptionsOrderComparator());<a name="line.293"></a>
<span class="sourceLineNo">294</span> helpFormatter.setLongOptSeparator("=");<a name="line.294"></a>
<span class="sourceLineNo">295</span> helpFormatter.printHelp(usageStr, usageHeader, options, usageFooter);<a name="line.295"></a>
<span class="sourceLineNo">296</span> }<a name="line.296"></a>
<span class="sourceLineNo">297</span><a name="line.297"></a>
<span class="sourceLineNo">298</span> @Override<a name="line.298"></a>
<span class="sourceLineNo">299</span> protected void addOptions() {<a name="line.299"></a>
<span class="sourceLineNo">300</span> Option startTimeOption = Option.builder(null).valueSeparator('=').hasArg(true).<a name="line.300"></a>
<span class="sourceLineNo">301</span> desc("starting time filter to start counting rows from.").longOpt(OPT_START_TIME).build();<a name="line.301"></a>
<span class="sourceLineNo">302</span> Option endTimeOption = Option.builder(null).valueSeparator('=').hasArg(true).<a name="line.302"></a>
<span class="sourceLineNo">303</span> desc("end time filter limit, to only count rows up to this timestamp.").<a name="line.303"></a>
<span class="sourceLineNo">304</span> longOpt(OPT_END_TIME).build();<a name="line.304"></a>
<span class="sourceLineNo">305</span> Option rangeOption = Option.builder(null).valueSeparator('=').hasArg(true).<a name="line.305"></a>
<span class="sourceLineNo">306</span> desc("[startKey],[endKey][;[startKey],[endKey]...]]").longOpt(OPT_RANGE).build();<a name="line.306"></a>
<span class="sourceLineNo">307</span> Option expectedOption = Option.builder(null).valueSeparator('=').hasArg(true).<a name="line.307"></a>
<span class="sourceLineNo">308</span> desc("expected number of rows to be count.").longOpt(OPT_EXPECTED_COUNT).build();<a name="line.308"></a>
<span class="sourceLineNo">309</span> addOption(startTimeOption);<a name="line.309"></a>
<span class="sourceLineNo">310</span> addOption(endTimeOption);<a name="line.310"></a>
<span class="sourceLineNo">311</span> addOption(rangeOption);<a name="line.311"></a>
<span class="sourceLineNo">312</span> addOption(expectedOption);<a name="line.312"></a>
<span class="sourceLineNo">313</span> }<a name="line.313"></a>
<span class="sourceLineNo">314</span><a name="line.314"></a>
<span class="sourceLineNo">315</span> @Override<a name="line.315"></a>
<span class="sourceLineNo">316</span> protected void processOptions(CommandLine cmd) throws IllegalArgumentException{<a name="line.316"></a>
<span class="sourceLineNo">317</span> this.tableName = cmd.getArgList().get(0);<a name="line.317"></a>
<span class="sourceLineNo">318</span> if(cmd.getOptionValue(OPT_RANGE)!=null) {<a name="line.318"></a>
<span class="sourceLineNo">319</span> this.rowRangeList = parseRowRangeParameter(cmd.getOptionValue(OPT_RANGE));<a name="line.319"></a>
<span class="sourceLineNo">320</span> }<a name="line.320"></a>
<span class="sourceLineNo">321</span> this.endTime = cmd.getOptionValue(OPT_END_TIME) == null ? HConstants.LATEST_TIMESTAMP :<a name="line.321"></a>
<span class="sourceLineNo">322</span> Long.parseLong(cmd.getOptionValue(OPT_END_TIME));<a name="line.322"></a>
<span class="sourceLineNo">323</span> this.expectedCount = cmd.getOptionValue(OPT_EXPECTED_COUNT) == null ? Long.MIN_VALUE :<a name="line.323"></a>
<span class="sourceLineNo">324</span> Long.parseLong(cmd.getOptionValue(OPT_EXPECTED_COUNT));<a name="line.324"></a>
<span class="sourceLineNo">325</span> this.startTime = cmd.getOptionValue(OPT_START_TIME) == null ? 0 :<a name="line.325"></a>
<span class="sourceLineNo">326</span> Long.parseLong(cmd.getOptionValue(OPT_START_TIME));<a name="line.326"></a>
<span class="sourceLineNo">327</span><a name="line.327"></a>
<span class="sourceLineNo">328</span> for(int i=1; i&lt;cmd.getArgList().size(); i++){<a name="line.328"></a>
<span class="sourceLineNo">329</span> String argument = cmd.getArgList().get(i);<a name="line.329"></a>
<span class="sourceLineNo">330</span> if(!argument.startsWith("-")){<a name="line.330"></a>
<span class="sourceLineNo">331</span> this.columns.add(argument);<a name="line.331"></a>
<span class="sourceLineNo">332</span> }<a name="line.332"></a>
<span class="sourceLineNo">333</span> }<a name="line.333"></a>
<span class="sourceLineNo">334</span><a name="line.334"></a>
<span class="sourceLineNo">335</span> if (endTime &lt; startTime) {<a name="line.335"></a>
<span class="sourceLineNo">336</span> throw new IllegalArgumentException("--endtime=" + endTime +<a name="line.336"></a>
<span class="sourceLineNo">337</span> " needs to be greater than --starttime=" + startTime);<a name="line.337"></a>
<span class="sourceLineNo">338</span> }<a name="line.338"></a>
<span class="sourceLineNo">339</span> }<a name="line.339"></a>
<span class="sourceLineNo">340</span><a name="line.340"></a>
<span class="sourceLineNo">341</span> @Override<a name="line.341"></a>
<span class="sourceLineNo">342</span> protected void processOldArgs(List&lt;String&gt; args) {<a name="line.342"></a>
<span class="sourceLineNo">343</span> List&lt;String&gt; copiedArgs = new ArrayList&lt;&gt;(args);<a name="line.343"></a>
<span class="sourceLineNo">344</span> args.removeAll(copiedArgs);<a name="line.344"></a>
<span class="sourceLineNo">345</span> for(String arg : copiedArgs){<a name="line.345"></a>
<span class="sourceLineNo">346</span> if(arg.startsWith("-") &amp;&amp; arg.contains("=")){<a name="line.346"></a>
<span class="sourceLineNo">347</span> String[] kv = arg.split("=");<a name="line.347"></a>
<span class="sourceLineNo">348</span> args.add(kv[0]);<a name="line.348"></a>
<span class="sourceLineNo">349</span> args.add(kv[1]);<a name="line.349"></a>
<span class="sourceLineNo">350</span> } else {<a name="line.350"></a>
<span class="sourceLineNo">351</span> args.add(arg);<a name="line.351"></a>
<span class="sourceLineNo">352</span> }<a name="line.352"></a>
<span class="sourceLineNo">353</span> }<a name="line.353"></a>
<span class="sourceLineNo">354</span> }<a name="line.354"></a>
<span class="sourceLineNo">355</span><a name="line.355"></a>
<span class="sourceLineNo">356</span> @Override<a name="line.356"></a>
<span class="sourceLineNo">357</span> protected int doWork() throws Exception {<a name="line.357"></a>
<span class="sourceLineNo">358</span> Job job = createSubmittableJob(getConf());<a name="line.358"></a>
<span class="sourceLineNo">359</span> if (job == null) {<a name="line.359"></a>
<span class="sourceLineNo">360</span> return -1;<a name="line.360"></a>
<span class="sourceLineNo">361</span> }<a name="line.361"></a>
<span class="sourceLineNo">362</span> boolean success = job.waitForCompletion(true);<a name="line.362"></a>
<span class="sourceLineNo">363</span> final long expectedCount = getConf().getLong(EXPECTED_COUNT_KEY, -1);<a name="line.363"></a>
<span class="sourceLineNo">364</span> if (success &amp;&amp; expectedCount != -1) {<a name="line.364"></a>
<span class="sourceLineNo">365</span> final Counter counter = job.getCounters().findCounter(RowCounterMapper.Counters.ROWS);<a name="line.365"></a>
<span class="sourceLineNo">366</span> success = expectedCount == counter.getValue();<a name="line.366"></a>
<span class="sourceLineNo">367</span> if (!success) {<a name="line.367"></a>
<span class="sourceLineNo">368</span> LOG.error("Failing job because count of '" + counter.getValue() +<a name="line.368"></a>
<span class="sourceLineNo">369</span> "' does not match expected count of '" + expectedCount + "'");<a name="line.369"></a>
<span class="sourceLineNo">370</span> }<a name="line.370"></a>
<span class="sourceLineNo">371</span> }<a name="line.371"></a>
<span class="sourceLineNo">372</span> return (success ? 0 : 1);<a name="line.372"></a>
<span class="sourceLineNo">373</span> }<a name="line.373"></a>
<span class="sourceLineNo">374</span><a name="line.374"></a>
<span class="sourceLineNo">375</span> /**<a name="line.375"></a>
<span class="sourceLineNo">376</span> * Main entry point.<a name="line.376"></a>
<span class="sourceLineNo">377</span> * @param args The command line parameters.<a name="line.377"></a>
<span class="sourceLineNo">378</span> * @throws Exception When running the job fails.<a name="line.378"></a>
<span class="sourceLineNo">379</span> */<a name="line.379"></a>
<span class="sourceLineNo">380</span> public static void main(String[] args) throws Exception {<a name="line.380"></a>
<span class="sourceLineNo">381</span> new RowCounter().doStaticMain(args);<a name="line.381"></a>
<span class="sourceLineNo">382</span> }<a name="line.382"></a>
<span class="sourceLineNo">383</span><a name="line.383"></a>
<span class="sourceLineNo">384</span> static class RowCounterCommandLineParser extends BasicParser {<a name="line.384"></a>
<span class="sourceLineNo">385</span><a name="line.385"></a>
<span class="sourceLineNo">386</span> @Override<a name="line.386"></a>
<span class="sourceLineNo">387</span> protected void checkRequiredOptions() throws MissingOptionException {<a name="line.387"></a>
<span class="sourceLineNo">388</span> if(this.cmd.getArgList().size()&lt;1 || this.cmd.getArgList().get(0).startsWith("-")){<a name="line.388"></a>
<span class="sourceLineNo">389</span> throw new MissingOptionException("First argument must be a valid table name.");<a name="line.389"></a>
<span class="sourceLineNo">390</span> }<a name="line.390"></a>
<span class="sourceLineNo">391</span> }<a name="line.391"></a>
<span class="sourceLineNo">392</span> }<a name="line.392"></a>
<span class="sourceLineNo">393</span><a name="line.393"></a>
<span class="sourceLineNo">394</span> @Override<a name="line.394"></a>
<span class="sourceLineNo">395</span> protected CommandLineParser newParser() {<a name="line.395"></a>
<span class="sourceLineNo">396</span> return new RowCounterCommandLineParser();<a name="line.396"></a>
<span class="sourceLineNo">397</span> }<a name="line.397"></a>
<span class="sourceLineNo">398</span><a name="line.398"></a>
<span class="sourceLineNo">399</span>}<a name="line.399"></a>
</pre>
</div>
</body>
</html>