blob: 309865b355a5ce53cecd18e4b8c15863216766ae [file] [log] [blame]
<!DOCTYPE HTML>
<html lang="en">
<head>
<!-- Generated by javadoc (17) -->
<title>Source code</title>
<meta name="viewport" content="width=device-width, initial-scale=1">
<meta name="description" content="source: package: org.apache.hadoop.hbase.mapreduce, class: RowCounter, class: RowCounterCommandLineParser">
<meta name="generator" content="javadoc/SourceToHTMLConverter">
<link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style">
</head>
<body class="source-page">
<main role="main">
<div class="source-container">
<pre><span class="source-line-no">001</span><span id="line-1">/*</span>
<span class="source-line-no">002</span><span id="line-2"> * Licensed to the Apache Software Foundation (ASF) under one</span>
<span class="source-line-no">003</span><span id="line-3"> * or more contributor license agreements. See the NOTICE file</span>
<span class="source-line-no">004</span><span id="line-4"> * distributed with this work for additional information</span>
<span class="source-line-no">005</span><span id="line-5"> * regarding copyright ownership. The ASF licenses this file</span>
<span class="source-line-no">006</span><span id="line-6"> * to you under the Apache License, Version 2.0 (the</span>
<span class="source-line-no">007</span><span id="line-7"> * "License"); you may not use this file except in compliance</span>
<span class="source-line-no">008</span><span id="line-8"> * with the License. You may obtain a copy of the License at</span>
<span class="source-line-no">009</span><span id="line-9"> *</span>
<span class="source-line-no">010</span><span id="line-10"> * http://www.apache.org/licenses/LICENSE-2.0</span>
<span class="source-line-no">011</span><span id="line-11"> *</span>
<span class="source-line-no">012</span><span id="line-12"> * Unless required by applicable law or agreed to in writing, software</span>
<span class="source-line-no">013</span><span id="line-13"> * distributed under the License is distributed on an "AS IS" BASIS,</span>
<span class="source-line-no">014</span><span id="line-14"> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span>
<span class="source-line-no">015</span><span id="line-15"> * See the License for the specific language governing permissions and</span>
<span class="source-line-no">016</span><span id="line-16"> * limitations under the License.</span>
<span class="source-line-no">017</span><span id="line-17"> */</span>
<span class="source-line-no">018</span><span id="line-18">package org.apache.hadoop.hbase.mapreduce;</span>
<span class="source-line-no">019</span><span id="line-19"></span>
<span class="source-line-no">020</span><span id="line-20">import com.google.errorprone.annotations.RestrictedApi;</span>
<span class="source-line-no">021</span><span id="line-21">import java.io.IOException;</span>
<span class="source-line-no">022</span><span id="line-22">import java.util.ArrayList;</span>
<span class="source-line-no">023</span><span id="line-23">import java.util.List;</span>
<span class="source-line-no">024</span><span id="line-24">import org.apache.commons.lang3.StringUtils;</span>
<span class="source-line-no">025</span><span id="line-25">import org.apache.hadoop.conf.Configuration;</span>
<span class="source-line-no">026</span><span id="line-26">import org.apache.hadoop.hbase.Cell;</span>
<span class="source-line-no">027</span><span id="line-27">import org.apache.hadoop.hbase.HConstants;</span>
<span class="source-line-no">028</span><span id="line-28">import org.apache.hadoop.hbase.client.Result;</span>
<span class="source-line-no">029</span><span id="line-29">import org.apache.hadoop.hbase.client.Scan;</span>
<span class="source-line-no">030</span><span id="line-30">import org.apache.hadoop.hbase.filter.Filter;</span>
<span class="source-line-no">031</span><span id="line-31">import org.apache.hadoop.hbase.filter.FilterBase;</span>
<span class="source-line-no">032</span><span id="line-32">import org.apache.hadoop.hbase.filter.FilterList;</span>
<span class="source-line-no">033</span><span id="line-33">import org.apache.hadoop.hbase.filter.FirstKeyOnlyFilter;</span>
<span class="source-line-no">034</span><span id="line-34">import org.apache.hadoop.hbase.filter.KeyOnlyFilter;</span>
<span class="source-line-no">035</span><span id="line-35">import org.apache.hadoop.hbase.filter.MultiRowRangeFilter;</span>
<span class="source-line-no">036</span><span id="line-36">import org.apache.hadoop.hbase.io.ImmutableBytesWritable;</span>
<span class="source-line-no">037</span><span id="line-37">import org.apache.hadoop.hbase.util.AbstractHBaseTool;</span>
<span class="source-line-no">038</span><span id="line-38">import org.apache.hadoop.hbase.util.Bytes;</span>
<span class="source-line-no">039</span><span id="line-39">import org.apache.hadoop.mapreduce.Counter;</span>
<span class="source-line-no">040</span><span id="line-40">import org.apache.hadoop.mapreduce.Job;</span>
<span class="source-line-no">041</span><span id="line-41">import org.apache.hadoop.mapreduce.Mapper;</span>
<span class="source-line-no">042</span><span id="line-42">import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;</span>
<span class="source-line-no">043</span><span id="line-43">import org.apache.yetus.audience.InterfaceAudience;</span>
<span class="source-line-no">044</span><span id="line-44">import org.slf4j.Logger;</span>
<span class="source-line-no">045</span><span id="line-45">import org.slf4j.LoggerFactory;</span>
<span class="source-line-no">046</span><span id="line-46"></span>
<span class="source-line-no">047</span><span id="line-47">import org.apache.hbase.thirdparty.com.google.common.base.Splitter;</span>
<span class="source-line-no">048</span><span id="line-48">import org.apache.hbase.thirdparty.org.apache.commons.cli.BasicParser;</span>
<span class="source-line-no">049</span><span id="line-49">import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;</span>
<span class="source-line-no">050</span><span id="line-50">import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLineParser;</span>
<span class="source-line-no">051</span><span id="line-51">import org.apache.hbase.thirdparty.org.apache.commons.cli.HelpFormatter;</span>
<span class="source-line-no">052</span><span id="line-52">import org.apache.hbase.thirdparty.org.apache.commons.cli.MissingOptionException;</span>
<span class="source-line-no">053</span><span id="line-53">import org.apache.hbase.thirdparty.org.apache.commons.cli.Option;</span>
<span class="source-line-no">054</span><span id="line-54"></span>
<span class="source-line-no">055</span><span id="line-55">/**</span>
<span class="source-line-no">056</span><span id="line-56"> * A job with a just a map phase to count rows. Map outputs table rows IF the input row has columns</span>
<span class="source-line-no">057</span><span id="line-57"> * that have content.</span>
<span class="source-line-no">058</span><span id="line-58"> */</span>
<span class="source-line-no">059</span><span id="line-59">@InterfaceAudience.Public</span>
<span class="source-line-no">060</span><span id="line-60">public class RowCounter extends AbstractHBaseTool {</span>
<span class="source-line-no">061</span><span id="line-61"></span>
<span class="source-line-no">062</span><span id="line-62"> private static final Logger LOG = LoggerFactory.getLogger(RowCounter.class);</span>
<span class="source-line-no">063</span><span id="line-63"></span>
<span class="source-line-no">064</span><span id="line-64"> /** Name of this 'program'. */</span>
<span class="source-line-no">065</span><span id="line-65"> static final String NAME = "rowcounter";</span>
<span class="source-line-no">066</span><span id="line-66"></span>
<span class="source-line-no">067</span><span id="line-67"> private final static String JOB_NAME_CONF_KEY = "mapreduce.job.name";</span>
<span class="source-line-no">068</span><span id="line-68"> private final static String EXPECTED_COUNT_KEY = RowCounter.class.getName() + ".expected_count";</span>
<span class="source-line-no">069</span><span id="line-69"></span>
<span class="source-line-no">070</span><span id="line-70"> private final static String OPT_START_TIME = "starttime";</span>
<span class="source-line-no">071</span><span id="line-71"> private final static String OPT_END_TIME = "endtime";</span>
<span class="source-line-no">072</span><span id="line-72"> private final static String OPT_RANGE = "range";</span>
<span class="source-line-no">073</span><span id="line-73"> private final static String OPT_EXPECTED_COUNT = "expectedCount";</span>
<span class="source-line-no">074</span><span id="line-74"> private final static String OPT_COUNT_DELETE_MARKERS = "countDeleteMarkers";</span>
<span class="source-line-no">075</span><span id="line-75"></span>
<span class="source-line-no">076</span><span id="line-76"> private String tableName;</span>
<span class="source-line-no">077</span><span id="line-77"> private List&lt;MultiRowRangeFilter.RowRange&gt; rowRangeList;</span>
<span class="source-line-no">078</span><span id="line-78"> private long startTime;</span>
<span class="source-line-no">079</span><span id="line-79"> private long endTime;</span>
<span class="source-line-no">080</span><span id="line-80"> private long expectedCount;</span>
<span class="source-line-no">081</span><span id="line-81"> private boolean countDeleteMarkers;</span>
<span class="source-line-no">082</span><span id="line-82"> private List&lt;String&gt; columns = new ArrayList&lt;&gt;();</span>
<span class="source-line-no">083</span><span id="line-83"></span>
<span class="source-line-no">084</span><span id="line-84"> private Job job;</span>
<span class="source-line-no">085</span><span id="line-85"></span>
<span class="source-line-no">086</span><span id="line-86"> /**</span>
<span class="source-line-no">087</span><span id="line-87"> * Mapper that runs the count.</span>
<span class="source-line-no">088</span><span id="line-88"> */</span>
<span class="source-line-no">089</span><span id="line-89"> static class RowCounterMapper extends TableMapper&lt;ImmutableBytesWritable, Result&gt; {</span>
<span class="source-line-no">090</span><span id="line-90"></span>
<span class="source-line-no">091</span><span id="line-91"> /** Counter enumeration to count the actual rows, cells and delete markers. */</span>
<span class="source-line-no">092</span><span id="line-92"> public static enum Counters {</span>
<span class="source-line-no">093</span><span id="line-93"> ROWS,</span>
<span class="source-line-no">094</span><span id="line-94"> DELETE,</span>
<span class="source-line-no">095</span><span id="line-95"> DELETE_COLUMN,</span>
<span class="source-line-no">096</span><span id="line-96"> DELETE_FAMILY,</span>
<span class="source-line-no">097</span><span id="line-97"> DELETE_FAMILY_VERSION,</span>
<span class="source-line-no">098</span><span id="line-98"> ROWS_WITH_DELETE_MARKER</span>
<span class="source-line-no">099</span><span id="line-99"> }</span>
<span class="source-line-no">100</span><span id="line-100"></span>
<span class="source-line-no">101</span><span id="line-101"> private boolean countDeleteMarkers;</span>
<span class="source-line-no">102</span><span id="line-102"></span>
<span class="source-line-no">103</span><span id="line-103"> @Override</span>
<span class="source-line-no">104</span><span id="line-104"> protected void</span>
<span class="source-line-no">105</span><span id="line-105"> setup(Mapper&lt;ImmutableBytesWritable, Result, ImmutableBytesWritable, Result&gt;.Context context)</span>
<span class="source-line-no">106</span><span id="line-106"> throws IOException, InterruptedException {</span>
<span class="source-line-no">107</span><span id="line-107"> Configuration conf = context.getConfiguration();</span>
<span class="source-line-no">108</span><span id="line-108"> countDeleteMarkers = conf.getBoolean(OPT_COUNT_DELETE_MARKERS, false);</span>
<span class="source-line-no">109</span><span id="line-109"> }</span>
<span class="source-line-no">110</span><span id="line-110"></span>
<span class="source-line-no">111</span><span id="line-111"> /**</span>
<span class="source-line-no">112</span><span id="line-112"> * Maps the data.</span>
<span class="source-line-no">113</span><span id="line-113"> * @param row The current table row key.</span>
<span class="source-line-no">114</span><span id="line-114"> * @param values The columns.</span>
<span class="source-line-no">115</span><span id="line-115"> * @param context The current context.</span>
<span class="source-line-no">116</span><span id="line-116"> * @throws IOException When something is broken with the data.</span>
<span class="source-line-no">117</span><span id="line-117"> * @see org.apache.hadoop.mapreduce.Mapper#map(Object, Object, Context)</span>
<span class="source-line-no">118</span><span id="line-118"> */</span>
<span class="source-line-no">119</span><span id="line-119"> @Override</span>
<span class="source-line-no">120</span><span id="line-120"> public void map(ImmutableBytesWritable row, Result values, Context context) throws IOException {</span>
<span class="source-line-no">121</span><span id="line-121"> // Count every row containing data, whether it's in qualifiers or values</span>
<span class="source-line-no">122</span><span id="line-122"> context.getCounter(Counters.ROWS).increment(1);</span>
<span class="source-line-no">123</span><span id="line-123"></span>
<span class="source-line-no">124</span><span id="line-124"> if (countDeleteMarkers) {</span>
<span class="source-line-no">125</span><span id="line-125"> boolean rowContainsDeleteMarker = false;</span>
<span class="source-line-no">126</span><span id="line-126"> for (Cell cell : values.rawCells()) {</span>
<span class="source-line-no">127</span><span id="line-127"> Cell.Type type = cell.getType();</span>
<span class="source-line-no">128</span><span id="line-128"> switch (type) {</span>
<span class="source-line-no">129</span><span id="line-129"> case Delete:</span>
<span class="source-line-no">130</span><span id="line-130"> rowContainsDeleteMarker = true;</span>
<span class="source-line-no">131</span><span id="line-131"> context.getCounter(Counters.DELETE).increment(1);</span>
<span class="source-line-no">132</span><span id="line-132"> break;</span>
<span class="source-line-no">133</span><span id="line-133"> case DeleteColumn:</span>
<span class="source-line-no">134</span><span id="line-134"> rowContainsDeleteMarker = true;</span>
<span class="source-line-no">135</span><span id="line-135"> context.getCounter(Counters.DELETE_COLUMN).increment(1);</span>
<span class="source-line-no">136</span><span id="line-136"> break;</span>
<span class="source-line-no">137</span><span id="line-137"> case DeleteFamily:</span>
<span class="source-line-no">138</span><span id="line-138"> rowContainsDeleteMarker = true;</span>
<span class="source-line-no">139</span><span id="line-139"> context.getCounter(Counters.DELETE_FAMILY).increment(1);</span>
<span class="source-line-no">140</span><span id="line-140"> break;</span>
<span class="source-line-no">141</span><span id="line-141"> case DeleteFamilyVersion:</span>
<span class="source-line-no">142</span><span id="line-142"> rowContainsDeleteMarker = true;</span>
<span class="source-line-no">143</span><span id="line-143"> context.getCounter(Counters.DELETE_FAMILY_VERSION).increment(1);</span>
<span class="source-line-no">144</span><span id="line-144"> break;</span>
<span class="source-line-no">145</span><span id="line-145"> default:</span>
<span class="source-line-no">146</span><span id="line-146"> break;</span>
<span class="source-line-no">147</span><span id="line-147"> }</span>
<span class="source-line-no">148</span><span id="line-148"> }</span>
<span class="source-line-no">149</span><span id="line-149"></span>
<span class="source-line-no">150</span><span id="line-150"> if (rowContainsDeleteMarker) {</span>
<span class="source-line-no">151</span><span id="line-151"> context.getCounter(Counters.ROWS_WITH_DELETE_MARKER).increment(1);</span>
<span class="source-line-no">152</span><span id="line-152"> }</span>
<span class="source-line-no">153</span><span id="line-153"> }</span>
<span class="source-line-no">154</span><span id="line-154"> }</span>
<span class="source-line-no">155</span><span id="line-155"> }</span>
<span class="source-line-no">156</span><span id="line-156"></span>
<span class="source-line-no">157</span><span id="line-157"> /**</span>
<span class="source-line-no">158</span><span id="line-158"> * Sets up the actual job.</span>
<span class="source-line-no">159</span><span id="line-159"> * @param conf The current configuration.</span>
<span class="source-line-no">160</span><span id="line-160"> * @return The newly created job.</span>
<span class="source-line-no">161</span><span id="line-161"> * @throws IOException When setting up the job fails.</span>
<span class="source-line-no">162</span><span id="line-162"> */</span>
<span class="source-line-no">163</span><span id="line-163"> public Job createSubmittableJob(Configuration conf) throws IOException {</span>
<span class="source-line-no">164</span><span id="line-164"> conf.setBoolean(OPT_COUNT_DELETE_MARKERS, this.countDeleteMarkers);</span>
<span class="source-line-no">165</span><span id="line-165"> Job job = Job.getInstance(conf, conf.get(JOB_NAME_CONF_KEY, NAME + "_" + tableName));</span>
<span class="source-line-no">166</span><span id="line-166"> job.setJarByClass(RowCounter.class);</span>
<span class="source-line-no">167</span><span id="line-167"> Scan scan = new Scan();</span>
<span class="source-line-no">168</span><span id="line-168"> // raw scan will be needed to account for delete markers when --countDeleteMarkers flag is set</span>
<span class="source-line-no">169</span><span id="line-169"> scan.setRaw(this.countDeleteMarkers);</span>
<span class="source-line-no">170</span><span id="line-170"> scan.setCacheBlocks(false);</span>
<span class="source-line-no">171</span><span id="line-171"> setScanFilter(scan, rowRangeList, this.countDeleteMarkers);</span>
<span class="source-line-no">172</span><span id="line-172"></span>
<span class="source-line-no">173</span><span id="line-173"> for (String columnName : this.columns) {</span>
<span class="source-line-no">174</span><span id="line-174"> String family = StringUtils.substringBefore(columnName, ":");</span>
<span class="source-line-no">175</span><span id="line-175"> String qualifier = StringUtils.substringAfter(columnName, ":");</span>
<span class="source-line-no">176</span><span id="line-176"> if (StringUtils.isBlank(qualifier)) {</span>
<span class="source-line-no">177</span><span id="line-177"> scan.addFamily(Bytes.toBytes(family));</span>
<span class="source-line-no">178</span><span id="line-178"> } else {</span>
<span class="source-line-no">179</span><span id="line-179"> scan.addColumn(Bytes.toBytes(family), Bytes.toBytes(qualifier));</span>
<span class="source-line-no">180</span><span id="line-180"> }</span>
<span class="source-line-no">181</span><span id="line-181"> }</span>
<span class="source-line-no">182</span><span id="line-182"></span>
<span class="source-line-no">183</span><span id="line-183"> if (this.expectedCount &gt;= 0) {</span>
<span class="source-line-no">184</span><span id="line-184"> conf.setLong(EXPECTED_COUNT_KEY, this.expectedCount);</span>
<span class="source-line-no">185</span><span id="line-185"> }</span>
<span class="source-line-no">186</span><span id="line-186"></span>
<span class="source-line-no">187</span><span id="line-187"> scan.setTimeRange(startTime, endTime);</span>
<span class="source-line-no">188</span><span id="line-188"> job.setOutputFormatClass(NullOutputFormat.class);</span>
<span class="source-line-no">189</span><span id="line-189"> TableMapReduceUtil.initTableMapperJob(tableName, scan, RowCounterMapper.class,</span>
<span class="source-line-no">190</span><span id="line-190"> ImmutableBytesWritable.class, Result.class, job);</span>
<span class="source-line-no">191</span><span id="line-191"> job.setNumReduceTasks(0);</span>
<span class="source-line-no">192</span><span id="line-192"> return job;</span>
<span class="source-line-no">193</span><span id="line-193"> }</span>
<span class="source-line-no">194</span><span id="line-194"></span>
<span class="source-line-no">195</span><span id="line-195"> /**</span>
<span class="source-line-no">196</span><span id="line-196"> * Sets up the actual job.</span>
<span class="source-line-no">197</span><span id="line-197"> * @param conf The current configuration.</span>
<span class="source-line-no">198</span><span id="line-198"> * @param args The command line parameters.</span>
<span class="source-line-no">199</span><span id="line-199"> * @return The newly created job.</span>
<span class="source-line-no">200</span><span id="line-200"> * @throws IOException When setting up the job fails.</span>
<span class="source-line-no">201</span><span id="line-201"> * @deprecated as of release 2.3.0. Will be removed on 4.0.0. Please use main method instead.</span>
<span class="source-line-no">202</span><span id="line-202"> */</span>
<span class="source-line-no">203</span><span id="line-203"> @Deprecated</span>
<span class="source-line-no">204</span><span id="line-204"> public static Job createSubmittableJob(Configuration conf, String[] args) throws IOException {</span>
<span class="source-line-no">205</span><span id="line-205"> String tableName = args[0];</span>
<span class="source-line-no">206</span><span id="line-206"> List&lt;MultiRowRangeFilter.RowRange&gt; rowRangeList = null;</span>
<span class="source-line-no">207</span><span id="line-207"> long startTime = 0;</span>
<span class="source-line-no">208</span><span id="line-208"> long endTime = 0;</span>
<span class="source-line-no">209</span><span id="line-209"> boolean countDeleteMarkers = false;</span>
<span class="source-line-no">210</span><span id="line-210"></span>
<span class="source-line-no">211</span><span id="line-211"> StringBuilder sb = new StringBuilder();</span>
<span class="source-line-no">212</span><span id="line-212"></span>
<span class="source-line-no">213</span><span id="line-213"> final String rangeSwitch = "--range=";</span>
<span class="source-line-no">214</span><span id="line-214"> final String startTimeArgKey = "--starttime=";</span>
<span class="source-line-no">215</span><span id="line-215"> final String endTimeArgKey = "--endtime=";</span>
<span class="source-line-no">216</span><span id="line-216"> final String expectedCountArg = "--expected-count=";</span>
<span class="source-line-no">217</span><span id="line-217"> final String countDeleteMarkersArg = "--countDeleteMarkers";</span>
<span class="source-line-no">218</span><span id="line-218"></span>
<span class="source-line-no">219</span><span id="line-219"> // First argument is table name, starting from second</span>
<span class="source-line-no">220</span><span id="line-220"> for (int i = 1; i &lt; args.length; i++) {</span>
<span class="source-line-no">221</span><span id="line-221"> if (args[i].startsWith(rangeSwitch)) {</span>
<span class="source-line-no">222</span><span id="line-222"> try {</span>
<span class="source-line-no">223</span><span id="line-223"> rowRangeList = parseRowRangeParameter(</span>
<span class="source-line-no">224</span><span id="line-224"> args[i].substring(args[1].indexOf(rangeSwitch) + rangeSwitch.length()));</span>
<span class="source-line-no">225</span><span id="line-225"> } catch (IllegalArgumentException e) {</span>
<span class="source-line-no">226</span><span id="line-226"> return null;</span>
<span class="source-line-no">227</span><span id="line-227"> }</span>
<span class="source-line-no">228</span><span id="line-228"> continue;</span>
<span class="source-line-no">229</span><span id="line-229"> }</span>
<span class="source-line-no">230</span><span id="line-230"> if (args[i].startsWith(startTimeArgKey)) {</span>
<span class="source-line-no">231</span><span id="line-231"> startTime = Long.parseLong(args[i].substring(startTimeArgKey.length()));</span>
<span class="source-line-no">232</span><span id="line-232"> continue;</span>
<span class="source-line-no">233</span><span id="line-233"> }</span>
<span class="source-line-no">234</span><span id="line-234"> if (args[i].startsWith(endTimeArgKey)) {</span>
<span class="source-line-no">235</span><span id="line-235"> endTime = Long.parseLong(args[i].substring(endTimeArgKey.length()));</span>
<span class="source-line-no">236</span><span id="line-236"> continue;</span>
<span class="source-line-no">237</span><span id="line-237"> }</span>
<span class="source-line-no">238</span><span id="line-238"> if (args[i].startsWith(expectedCountArg)) {</span>
<span class="source-line-no">239</span><span id="line-239"> conf.setLong(EXPECTED_COUNT_KEY,</span>
<span class="source-line-no">240</span><span id="line-240"> Long.parseLong(args[i].substring(expectedCountArg.length())));</span>
<span class="source-line-no">241</span><span id="line-241"> continue;</span>
<span class="source-line-no">242</span><span id="line-242"> }</span>
<span class="source-line-no">243</span><span id="line-243"> if (args[i].startsWith(countDeleteMarkersArg)) {</span>
<span class="source-line-no">244</span><span id="line-244"> countDeleteMarkers = true;</span>
<span class="source-line-no">245</span><span id="line-245"> continue;</span>
<span class="source-line-no">246</span><span id="line-246"> }</span>
<span class="source-line-no">247</span><span id="line-247"> // if no switch, assume column names</span>
<span class="source-line-no">248</span><span id="line-248"> sb.append(args[i]);</span>
<span class="source-line-no">249</span><span id="line-249"> sb.append(" ");</span>
<span class="source-line-no">250</span><span id="line-250"> }</span>
<span class="source-line-no">251</span><span id="line-251"> conf.setBoolean(OPT_COUNT_DELETE_MARKERS, countDeleteMarkers);</span>
<span class="source-line-no">252</span><span id="line-252"> if (endTime &lt; startTime) {</span>
<span class="source-line-no">253</span><span id="line-253"> printUsage("--endtime=" + endTime + " needs to be greater than --starttime=" + startTime);</span>
<span class="source-line-no">254</span><span id="line-254"> return null;</span>
<span class="source-line-no">255</span><span id="line-255"> }</span>
<span class="source-line-no">256</span><span id="line-256"></span>
<span class="source-line-no">257</span><span id="line-257"> Job job = Job.getInstance(conf, conf.get(JOB_NAME_CONF_KEY, NAME + "_" + tableName));</span>
<span class="source-line-no">258</span><span id="line-258"> job.setJarByClass(RowCounter.class);</span>
<span class="source-line-no">259</span><span id="line-259"> Scan scan = new Scan();</span>
<span class="source-line-no">260</span><span id="line-260"> scan.setCacheBlocks(false);</span>
<span class="source-line-no">261</span><span id="line-261"> // raw scan will be needed to account for delete markers when --countDeleteMarkers flag is set</span>
<span class="source-line-no">262</span><span id="line-262"> scan.setRaw(countDeleteMarkers);</span>
<span class="source-line-no">263</span><span id="line-263"> setScanFilter(scan, rowRangeList, countDeleteMarkers);</span>
<span class="source-line-no">264</span><span id="line-264"> if (sb.length() &gt; 0) {</span>
<span class="source-line-no">265</span><span id="line-265"> for (String columnName : sb.toString().trim().split(" ")) {</span>
<span class="source-line-no">266</span><span id="line-266"> String family = StringUtils.substringBefore(columnName, ":");</span>
<span class="source-line-no">267</span><span id="line-267"> String qualifier = StringUtils.substringAfter(columnName, ":");</span>
<span class="source-line-no">268</span><span id="line-268"></span>
<span class="source-line-no">269</span><span id="line-269"> if (StringUtils.isBlank(qualifier)) {</span>
<span class="source-line-no">270</span><span id="line-270"> scan.addFamily(Bytes.toBytes(family));</span>
<span class="source-line-no">271</span><span id="line-271"> } else {</span>
<span class="source-line-no">272</span><span id="line-272"> scan.addColumn(Bytes.toBytes(family), Bytes.toBytes(qualifier));</span>
<span class="source-line-no">273</span><span id="line-273"> }</span>
<span class="source-line-no">274</span><span id="line-274"> }</span>
<span class="source-line-no">275</span><span id="line-275"> }</span>
<span class="source-line-no">276</span><span id="line-276"> scan.setTimeRange(startTime, endTime == 0 ? HConstants.LATEST_TIMESTAMP : endTime);</span>
<span class="source-line-no">277</span><span id="line-277"> job.setOutputFormatClass(NullOutputFormat.class);</span>
<span class="source-line-no">278</span><span id="line-278"> TableMapReduceUtil.initTableMapperJob(tableName, scan, RowCounterMapper.class,</span>
<span class="source-line-no">279</span><span id="line-279"> ImmutableBytesWritable.class, Result.class, job);</span>
<span class="source-line-no">280</span><span id="line-280"> job.setNumReduceTasks(0);</span>
<span class="source-line-no">281</span><span id="line-281"> return job;</span>
<span class="source-line-no">282</span><span id="line-282"> }</span>
<span class="source-line-no">283</span><span id="line-283"></span>
<span class="source-line-no">284</span><span id="line-284"> /**</span>
<span class="source-line-no">285</span><span id="line-285"> * Prints usage without error message. Note that we don't document --expected-count, because it's</span>
<span class="source-line-no">286</span><span id="line-286"> * intended for test.</span>
<span class="source-line-no">287</span><span id="line-287"> */</span>
<span class="source-line-no">288</span><span id="line-288"> private static void printUsage(String errorMessage) {</span>
<span class="source-line-no">289</span><span id="line-289"> System.err.println("ERROR: " + errorMessage);</span>
<span class="source-line-no">290</span><span id="line-290"> System.err.println(</span>
<span class="source-line-no">291</span><span id="line-291"> "Usage: hbase rowcounter [options] &lt;tablename&gt; " + "[--starttime=&lt;start&gt; --endtime=&lt;end&gt;] "</span>
<span class="source-line-no">292</span><span id="line-292"> + "[--range=[startKey],[endKey][;[startKey],[endKey]...]] [&lt;column1&gt; &lt;column2&gt;...]");</span>
<span class="source-line-no">293</span><span id="line-293"> System.err.println("For performance consider the following options:\n"</span>
<span class="source-line-no">294</span><span id="line-294"> + "-Dhbase.client.scanner.caching=100\n" + "-Dmapreduce.map.speculative=false");</span>
<span class="source-line-no">295</span><span id="line-295"> }</span>
<span class="source-line-no">296</span><span id="line-296"></span>
<span class="source-line-no">297</span><span id="line-297"> private static List&lt;MultiRowRangeFilter.RowRange&gt; parseRowRangeParameter(String arg) {</span>
<span class="source-line-no">298</span><span id="line-298"> final List&lt;String&gt; rangesSplit = Splitter.on(";").splitToList(arg);</span>
<span class="source-line-no">299</span><span id="line-299"> final List&lt;MultiRowRangeFilter.RowRange&gt; rangeList = new ArrayList&lt;&gt;();</span>
<span class="source-line-no">300</span><span id="line-300"> for (String range : rangesSplit) {</span>
<span class="source-line-no">301</span><span id="line-301"> if (range != null &amp;&amp; !range.isEmpty()) {</span>
<span class="source-line-no">302</span><span id="line-302"> List&lt;String&gt; startEnd = Splitter.on(",").splitToList(range);</span>
<span class="source-line-no">303</span><span id="line-303"> if (startEnd.size() != 2 || startEnd.get(1).contains(",")) {</span>
<span class="source-line-no">304</span><span id="line-304"> throw new IllegalArgumentException("Wrong range specification: " + range);</span>
<span class="source-line-no">305</span><span id="line-305"> }</span>
<span class="source-line-no">306</span><span id="line-306"> String startKey = startEnd.get(0);</span>
<span class="source-line-no">307</span><span id="line-307"> String endKey = startEnd.get(1);</span>
<span class="source-line-no">308</span><span id="line-308"> rangeList.add(new MultiRowRangeFilter.RowRange(Bytes.toBytesBinary(startKey), true,</span>
<span class="source-line-no">309</span><span id="line-309"> Bytes.toBytesBinary(endKey), false));</span>
<span class="source-line-no">310</span><span id="line-310"> }</span>
<span class="source-line-no">311</span><span id="line-311"> }</span>
<span class="source-line-no">312</span><span id="line-312"> return rangeList;</span>
<span class="source-line-no">313</span><span id="line-313"> }</span>
<span class="source-line-no">314</span><span id="line-314"></span>
<span class="source-line-no">315</span><span id="line-315"> /**</span>
<span class="source-line-no">316</span><span id="line-316"> * Sets filter {@link FilterBase} to the {@link Scan} instance. If provided rowRangeList contains</span>
<span class="source-line-no">317</span><span id="line-317"> * more than one element, method sets filter which is instance of {@link MultiRowRangeFilter}. If</span>
<span class="source-line-no">318</span><span id="line-318"> * rowRangeList contains exactly one element, startRow and stopRow are set to the scan. Also,</span>
<span class="source-line-no">319</span><span id="line-319"> * method may apply {@link FirstKeyOnlyFilter} an {@link KeyOnlyFilter} for better performance.</span>
<span class="source-line-no">320</span><span id="line-320"> */</span>
<span class="source-line-no">321</span><span id="line-321"> private static void setScanFilter(Scan scan, List&lt;MultiRowRangeFilter.RowRange&gt; rowRangeList,</span>
<span class="source-line-no">322</span><span id="line-322"> boolean countDeleteMarkers) {</span>
<span class="source-line-no">323</span><span id="line-323"> List&lt;Filter&gt; filters = new ArrayList&lt;&gt;();</span>
<span class="source-line-no">324</span><span id="line-324"></span>
<span class="source-line-no">325</span><span id="line-325"> // Apply filters for better performance.</span>
<span class="source-line-no">326</span><span id="line-326"> if (!countDeleteMarkers) {</span>
<span class="source-line-no">327</span><span id="line-327"> // We only need one cell per row, unless --countDeleteMarkers flag is set.</span>
<span class="source-line-no">328</span><span id="line-328"> filters.add(new FirstKeyOnlyFilter());</span>
<span class="source-line-no">329</span><span id="line-329"></span>
<span class="source-line-no">330</span><span id="line-330"> // We're not interested in values. Use KeyOnlyFilter.</span>
<span class="source-line-no">331</span><span id="line-331"> // NOTE: Logically, KeyOnlyFilter should be okay for --countDeleteMarkers, because it should</span>
<span class="source-line-no">332</span><span id="line-332"> // only empty the values, but currently, having a filter changes the behavior of a raw scan.</span>
<span class="source-line-no">333</span><span id="line-333"> // So we don't use it. See {@link UserScanQueryMatcher#mergeFilterResponse} for more details.</span>
<span class="source-line-no">334</span><span id="line-334"> filters.add(new KeyOnlyFilter());</span>
<span class="source-line-no">335</span><span id="line-335"> }</span>
<span class="source-line-no">336</span><span id="line-336"></span>
<span class="source-line-no">337</span><span id="line-337"> // Depending on the number of ranges, set start/stop row or apply MultiRowRangeFilter.</span>
<span class="source-line-no">338</span><span id="line-338"> final int size = rowRangeList == null ? 0 : rowRangeList.size();</span>
<span class="source-line-no">339</span><span id="line-339"> if (size == 1) {</span>
<span class="source-line-no">340</span><span id="line-340"> MultiRowRangeFilter.RowRange range = rowRangeList.get(0);</span>
<span class="source-line-no">341</span><span id="line-341"> scan.withStartRow(range.getStartRow()); // inclusive</span>
<span class="source-line-no">342</span><span id="line-342"> scan.withStopRow(range.getStopRow()); // exclusive</span>
<span class="source-line-no">343</span><span id="line-343"> } else if (size &gt; 1) {</span>
<span class="source-line-no">344</span><span id="line-344"> filters.add(new MultiRowRangeFilter(rowRangeList));</span>
<span class="source-line-no">345</span><span id="line-345"> }</span>
<span class="source-line-no">346</span><span id="line-346"></span>
<span class="source-line-no">347</span><span id="line-347"> if (filters.isEmpty()) {</span>
<span class="source-line-no">348</span><span id="line-348"> return;</span>
<span class="source-line-no">349</span><span id="line-349"> }</span>
<span class="source-line-no">350</span><span id="line-350"> scan.setFilter(filters.size() == 1 ? filters.get(0) : new FilterList(filters));</span>
<span class="source-line-no">351</span><span id="line-351"> }</span>
<span class="source-line-no">352</span><span id="line-352"></span>
<span class="source-line-no">353</span><span id="line-353"> @Override</span>
<span class="source-line-no">354</span><span id="line-354"> protected void printUsage() {</span>
<span class="source-line-no">355</span><span id="line-355"> StringBuilder footerBuilder = new StringBuilder();</span>
<span class="source-line-no">356</span><span id="line-356"> footerBuilder.append("For performance, consider the following configuration properties:\n");</span>
<span class="source-line-no">357</span><span id="line-357"> footerBuilder.append("-Dhbase.client.scanner.caching=100\n");</span>
<span class="source-line-no">358</span><span id="line-358"> footerBuilder.append("-Dmapreduce.map.speculative=false\n");</span>
<span class="source-line-no">359</span><span id="line-359"> printUsage("hbase rowcounter &lt;tablename&gt; [options] [&lt;column1&gt; &lt;column2&gt;...]", "Options:",</span>
<span class="source-line-no">360</span><span id="line-360"> footerBuilder.toString());</span>
<span class="source-line-no">361</span><span id="line-361"> }</span>
<span class="source-line-no">362</span><span id="line-362"></span>
<span class="source-line-no">363</span><span id="line-363"> @Override</span>
<span class="source-line-no">364</span><span id="line-364"> protected void printUsage(final String usageStr, final String usageHeader,</span>
<span class="source-line-no">365</span><span id="line-365"> final String usageFooter) {</span>
<span class="source-line-no">366</span><span id="line-366"> HelpFormatter helpFormatter = new HelpFormatter();</span>
<span class="source-line-no">367</span><span id="line-367"> helpFormatter.setWidth(120);</span>
<span class="source-line-no">368</span><span id="line-368"> helpFormatter.setOptionComparator(new AbstractHBaseTool.OptionsOrderComparator());</span>
<span class="source-line-no">369</span><span id="line-369"> helpFormatter.setLongOptSeparator("=");</span>
<span class="source-line-no">370</span><span id="line-370"> helpFormatter.printHelp(usageStr, usageHeader, options, usageFooter);</span>
<span class="source-line-no">371</span><span id="line-371"> }</span>
<span class="source-line-no">372</span><span id="line-372"></span>
<span class="source-line-no">373</span><span id="line-373"> @Override</span>
<span class="source-line-no">374</span><span id="line-374"> protected void addOptions() {</span>
<span class="source-line-no">375</span><span id="line-375"> Option startTimeOption = Option.builder(null).valueSeparator('=').hasArg(true)</span>
<span class="source-line-no">376</span><span id="line-376"> .desc("starting time filter to start counting rows from.").longOpt(OPT_START_TIME).build();</span>
<span class="source-line-no">377</span><span id="line-377"> Option endTimeOption = Option.builder(null).valueSeparator('=').hasArg(true)</span>
<span class="source-line-no">378</span><span id="line-378"> .desc("end time filter limit, to only count rows up to this timestamp.").longOpt(OPT_END_TIME)</span>
<span class="source-line-no">379</span><span id="line-379"> .build();</span>
<span class="source-line-no">380</span><span id="line-380"> Option rangeOption = Option.builder(null).valueSeparator('=').hasArg(true)</span>
<span class="source-line-no">381</span><span id="line-381"> .desc("[startKey],[endKey][;[startKey],[endKey]...]]").longOpt(OPT_RANGE).build();</span>
<span class="source-line-no">382</span><span id="line-382"> Option expectedOption = Option.builder(null).valueSeparator('=').hasArg(true)</span>
<span class="source-line-no">383</span><span id="line-383"> .desc("expected number of rows to be count.").longOpt(OPT_EXPECTED_COUNT).build();</span>
<span class="source-line-no">384</span><span id="line-384"> Option countDeleteMarkersOption = Option.builder(null).hasArg(false)</span>
<span class="source-line-no">385</span><span id="line-385"> .desc("counts the number of Delete Markers of all types, i.e. "</span>
<span class="source-line-no">386</span><span id="line-386"> + "(DELETE, DELETE_COLUMN, DELETE_FAMILY, DELETE_FAMILY_VERSION)")</span>
<span class="source-line-no">387</span><span id="line-387"> .longOpt(OPT_COUNT_DELETE_MARKERS).build();</span>
<span class="source-line-no">388</span><span id="line-388"> addOption(startTimeOption);</span>
<span class="source-line-no">389</span><span id="line-389"> addOption(endTimeOption);</span>
<span class="source-line-no">390</span><span id="line-390"> addOption(rangeOption);</span>
<span class="source-line-no">391</span><span id="line-391"> addOption(expectedOption);</span>
<span class="source-line-no">392</span><span id="line-392"> addOption(countDeleteMarkersOption);</span>
<span class="source-line-no">393</span><span id="line-393"> }</span>
<span class="source-line-no">394</span><span id="line-394"></span>
<span class="source-line-no">395</span><span id="line-395"> @Override</span>
<span class="source-line-no">396</span><span id="line-396"> protected void processOptions(CommandLine cmd) throws IllegalArgumentException {</span>
<span class="source-line-no">397</span><span id="line-397"> this.tableName = cmd.getArgList().get(0);</span>
<span class="source-line-no">398</span><span id="line-398"> if (cmd.getOptionValue(OPT_RANGE) != null) {</span>
<span class="source-line-no">399</span><span id="line-399"> this.rowRangeList = parseRowRangeParameter(cmd.getOptionValue(OPT_RANGE));</span>
<span class="source-line-no">400</span><span id="line-400"> }</span>
<span class="source-line-no">401</span><span id="line-401"> this.endTime = cmd.getOptionValue(OPT_END_TIME) == null</span>
<span class="source-line-no">402</span><span id="line-402"> ? HConstants.LATEST_TIMESTAMP</span>
<span class="source-line-no">403</span><span id="line-403"> : Long.parseLong(cmd.getOptionValue(OPT_END_TIME));</span>
<span class="source-line-no">404</span><span id="line-404"> this.expectedCount = cmd.getOptionValue(OPT_EXPECTED_COUNT) == null</span>
<span class="source-line-no">405</span><span id="line-405"> ? Long.MIN_VALUE</span>
<span class="source-line-no">406</span><span id="line-406"> : Long.parseLong(cmd.getOptionValue(OPT_EXPECTED_COUNT));</span>
<span class="source-line-no">407</span><span id="line-407"> this.startTime = cmd.getOptionValue(OPT_START_TIME) == null</span>
<span class="source-line-no">408</span><span id="line-408"> ? 0</span>
<span class="source-line-no">409</span><span id="line-409"> : Long.parseLong(cmd.getOptionValue(OPT_START_TIME));</span>
<span class="source-line-no">410</span><span id="line-410"> this.countDeleteMarkers = cmd.hasOption(OPT_COUNT_DELETE_MARKERS);</span>
<span class="source-line-no">411</span><span id="line-411"></span>
<span class="source-line-no">412</span><span id="line-412"> for (int i = 1; i &lt; cmd.getArgList().size(); i++) {</span>
<span class="source-line-no">413</span><span id="line-413"> String argument = cmd.getArgList().get(i);</span>
<span class="source-line-no">414</span><span id="line-414"> if (!argument.startsWith("-")) {</span>
<span class="source-line-no">415</span><span id="line-415"> this.columns.add(argument);</span>
<span class="source-line-no">416</span><span id="line-416"> }</span>
<span class="source-line-no">417</span><span id="line-417"> }</span>
<span class="source-line-no">418</span><span id="line-418"></span>
<span class="source-line-no">419</span><span id="line-419"> if (endTime &lt; startTime) {</span>
<span class="source-line-no">420</span><span id="line-420"> throw new IllegalArgumentException(</span>
<span class="source-line-no">421</span><span id="line-421"> "--endtime=" + endTime + " needs to be greater than --starttime=" + startTime);</span>
<span class="source-line-no">422</span><span id="line-422"> }</span>
<span class="source-line-no">423</span><span id="line-423"> }</span>
<span class="source-line-no">424</span><span id="line-424"></span>
<span class="source-line-no">425</span><span id="line-425"> @Override</span>
<span class="source-line-no">426</span><span id="line-426"> protected void processOldArgs(List&lt;String&gt; args) {</span>
<span class="source-line-no">427</span><span id="line-427"> List&lt;String&gt; copiedArgs = new ArrayList&lt;&gt;(args);</span>
<span class="source-line-no">428</span><span id="line-428"> args.removeAll(copiedArgs);</span>
<span class="source-line-no">429</span><span id="line-429"> for (String arg : copiedArgs) {</span>
<span class="source-line-no">430</span><span id="line-430"> if (arg.startsWith("-") &amp;&amp; arg.contains("=")) {</span>
<span class="source-line-no">431</span><span id="line-431"> String[] kv = arg.split("=");</span>
<span class="source-line-no">432</span><span id="line-432"> args.add(kv[0]);</span>
<span class="source-line-no">433</span><span id="line-433"> args.add(kv[1]);</span>
<span class="source-line-no">434</span><span id="line-434"> } else {</span>
<span class="source-line-no">435</span><span id="line-435"> args.add(arg);</span>
<span class="source-line-no">436</span><span id="line-436"> }</span>
<span class="source-line-no">437</span><span id="line-437"> }</span>
<span class="source-line-no">438</span><span id="line-438"> }</span>
<span class="source-line-no">439</span><span id="line-439"></span>
<span class="source-line-no">440</span><span id="line-440"> @Override</span>
<span class="source-line-no">441</span><span id="line-441"> protected int doWork() throws Exception {</span>
<span class="source-line-no">442</span><span id="line-442"> job = createSubmittableJob(getConf());</span>
<span class="source-line-no">443</span><span id="line-443"> if (job == null) {</span>
<span class="source-line-no">444</span><span id="line-444"> return -1;</span>
<span class="source-line-no">445</span><span id="line-445"> }</span>
<span class="source-line-no">446</span><span id="line-446"> boolean success = job.waitForCompletion(true);</span>
<span class="source-line-no">447</span><span id="line-447"> final long expectedCount = getConf().getLong(EXPECTED_COUNT_KEY, -1);</span>
<span class="source-line-no">448</span><span id="line-448"> if (success &amp;&amp; expectedCount != -1) {</span>
<span class="source-line-no">449</span><span id="line-449"> final Counter counter = job.getCounters().findCounter(RowCounterMapper.Counters.ROWS);</span>
<span class="source-line-no">450</span><span id="line-450"> success = expectedCount == counter.getValue();</span>
<span class="source-line-no">451</span><span id="line-451"> if (!success) {</span>
<span class="source-line-no">452</span><span id="line-452"> LOG.error("Failing job because count of '" + counter.getValue()</span>
<span class="source-line-no">453</span><span id="line-453"> + "' does not match expected count of '" + expectedCount + "'");</span>
<span class="source-line-no">454</span><span id="line-454"> }</span>
<span class="source-line-no">455</span><span id="line-455"> }</span>
<span class="source-line-no">456</span><span id="line-456"> return (success ? 0 : 1);</span>
<span class="source-line-no">457</span><span id="line-457"> }</span>
<span class="source-line-no">458</span><span id="line-458"></span>
<span class="source-line-no">459</span><span id="line-459"> /**</span>
<span class="source-line-no">460</span><span id="line-460"> * Main entry point.</span>
<span class="source-line-no">461</span><span id="line-461"> * @param args The command line parameters.</span>
<span class="source-line-no">462</span><span id="line-462"> * @throws Exception When running the job fails.</span>
<span class="source-line-no">463</span><span id="line-463"> */</span>
<span class="source-line-no">464</span><span id="line-464"> public static void main(String[] args) throws Exception {</span>
<span class="source-line-no">465</span><span id="line-465"> new RowCounter().doStaticMain(args);</span>
<span class="source-line-no">466</span><span id="line-466"> }</span>
<span class="source-line-no">467</span><span id="line-467"></span>
<span class="source-line-no">468</span><span id="line-468"> static class RowCounterCommandLineParser extends BasicParser {</span>
<span class="source-line-no">469</span><span id="line-469"></span>
<span class="source-line-no">470</span><span id="line-470"> @Override</span>
<span class="source-line-no">471</span><span id="line-471"> protected void checkRequiredOptions() throws MissingOptionException {</span>
<span class="source-line-no">472</span><span id="line-472"> if (this.cmd.getArgList().size() &lt; 1 || this.cmd.getArgList().get(0).startsWith("-")) {</span>
<span class="source-line-no">473</span><span id="line-473"> throw new MissingOptionException("First argument must be a valid table name.");</span>
<span class="source-line-no">474</span><span id="line-474"> }</span>
<span class="source-line-no">475</span><span id="line-475"> }</span>
<span class="source-line-no">476</span><span id="line-476"> }</span>
<span class="source-line-no">477</span><span id="line-477"></span>
<span class="source-line-no">478</span><span id="line-478"> @Override</span>
<span class="source-line-no">479</span><span id="line-479"> protected CommandLineParser newParser() {</span>
<span class="source-line-no">480</span><span id="line-480"> return new RowCounterCommandLineParser();</span>
<span class="source-line-no">481</span><span id="line-481"> }</span>
<span class="source-line-no">482</span><span id="line-482"></span>
<span class="source-line-no">483</span><span id="line-483"> @RestrictedApi(explanation = "Only visible for testing", link = "",</span>
<span class="source-line-no">484</span><span id="line-484"> allowedOnPath = ".*/src/test/.*")</span>
<span class="source-line-no">485</span><span id="line-485"> Job getMapReduceJob() {</span>
<span class="source-line-no">486</span><span id="line-486"> return job;</span>
<span class="source-line-no">487</span><span id="line-487"> }</span>
<span class="source-line-no">488</span><span id="line-488"></span>
<span class="source-line-no">489</span><span id="line-489">}</span>
</pre>
</div>
</main>
</body>
</html>