| <!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: RowCounterMapper"> |
| <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 java.io.IOException;</span> |
| <span class="source-line-no">021</span><span id="line-21">import java.util.ArrayList;</span> |
| <span class="source-line-no">022</span><span id="line-22">import java.util.List;</span> |
| <span class="source-line-no">023</span><span id="line-23">import org.apache.commons.lang3.StringUtils;</span> |
| <span class="source-line-no">024</span><span id="line-24">import org.apache.hadoop.conf.Configuration;</span> |
| <span class="source-line-no">025</span><span id="line-25">import org.apache.hadoop.hbase.HConstants;</span> |
| <span class="source-line-no">026</span><span id="line-26">import org.apache.hadoop.hbase.client.Result;</span> |
| <span class="source-line-no">027</span><span id="line-27">import org.apache.hadoop.hbase.client.Scan;</span> |
| <span class="source-line-no">028</span><span id="line-28">import org.apache.hadoop.hbase.filter.FilterBase;</span> |
| <span class="source-line-no">029</span><span id="line-29">import org.apache.hadoop.hbase.filter.FirstKeyOnlyFilter;</span> |
| <span class="source-line-no">030</span><span id="line-30">import org.apache.hadoop.hbase.filter.MultiRowRangeFilter;</span> |
| <span class="source-line-no">031</span><span id="line-31">import org.apache.hadoop.hbase.io.ImmutableBytesWritable;</span> |
| <span class="source-line-no">032</span><span id="line-32">import org.apache.hadoop.hbase.util.AbstractHBaseTool;</span> |
| <span class="source-line-no">033</span><span id="line-33">import org.apache.hadoop.hbase.util.Bytes;</span> |
| <span class="source-line-no">034</span><span id="line-34">import org.apache.hadoop.mapreduce.Counter;</span> |
| <span class="source-line-no">035</span><span id="line-35">import org.apache.hadoop.mapreduce.Job;</span> |
| <span class="source-line-no">036</span><span id="line-36">import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;</span> |
| <span class="source-line-no">037</span><span id="line-37">import org.apache.yetus.audience.InterfaceAudience;</span> |
| <span class="source-line-no">038</span><span id="line-38">import org.slf4j.Logger;</span> |
| <span class="source-line-no">039</span><span id="line-39">import org.slf4j.LoggerFactory;</span> |
| <span class="source-line-no">040</span><span id="line-40"></span> |
| <span class="source-line-no">041</span><span id="line-41">import org.apache.hbase.thirdparty.com.google.common.base.Splitter;</span> |
| <span class="source-line-no">042</span><span id="line-42">import org.apache.hbase.thirdparty.org.apache.commons.cli.BasicParser;</span> |
| <span class="source-line-no">043</span><span id="line-43">import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLine;</span> |
| <span class="source-line-no">044</span><span id="line-44">import org.apache.hbase.thirdparty.org.apache.commons.cli.CommandLineParser;</span> |
| <span class="source-line-no">045</span><span id="line-45">import org.apache.hbase.thirdparty.org.apache.commons.cli.HelpFormatter;</span> |
| <span class="source-line-no">046</span><span id="line-46">import org.apache.hbase.thirdparty.org.apache.commons.cli.MissingOptionException;</span> |
| <span class="source-line-no">047</span><span id="line-47">import org.apache.hbase.thirdparty.org.apache.commons.cli.Option;</span> |
| <span class="source-line-no">048</span><span id="line-48"></span> |
| <span class="source-line-no">049</span><span id="line-49">/**</span> |
| <span class="source-line-no">050</span><span id="line-50"> * 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">051</span><span id="line-51"> * that have content.</span> |
| <span class="source-line-no">052</span><span id="line-52"> */</span> |
| <span class="source-line-no">053</span><span id="line-53">@InterfaceAudience.Public</span> |
| <span class="source-line-no">054</span><span id="line-54">public class RowCounter extends AbstractHBaseTool {</span> |
| <span class="source-line-no">055</span><span id="line-55"></span> |
| <span class="source-line-no">056</span><span id="line-56"> private static final Logger LOG = LoggerFactory.getLogger(RowCounter.class);</span> |
| <span class="source-line-no">057</span><span id="line-57"></span> |
| <span class="source-line-no">058</span><span id="line-58"> /** Name of this 'program'. */</span> |
| <span class="source-line-no">059</span><span id="line-59"> static final String NAME = "rowcounter";</span> |
| <span class="source-line-no">060</span><span id="line-60"></span> |
| <span class="source-line-no">061</span><span id="line-61"> private final static String JOB_NAME_CONF_KEY = "mapreduce.job.name";</span> |
| <span class="source-line-no">062</span><span id="line-62"> private final static String EXPECTED_COUNT_KEY = RowCounter.class.getName() + ".expected_count";</span> |
| <span class="source-line-no">063</span><span id="line-63"></span> |
| <span class="source-line-no">064</span><span id="line-64"> private final static String OPT_START_TIME = "starttime";</span> |
| <span class="source-line-no">065</span><span id="line-65"> private final static String OPT_END_TIME = "endtime";</span> |
| <span class="source-line-no">066</span><span id="line-66"> private final static String OPT_RANGE = "range";</span> |
| <span class="source-line-no">067</span><span id="line-67"> private final static String OPT_EXPECTED_COUNT = "expectedCount";</span> |
| <span class="source-line-no">068</span><span id="line-68"></span> |
| <span class="source-line-no">069</span><span id="line-69"> private String tableName;</span> |
| <span class="source-line-no">070</span><span id="line-70"> private List<MultiRowRangeFilter.RowRange> rowRangeList;</span> |
| <span class="source-line-no">071</span><span id="line-71"> private long startTime;</span> |
| <span class="source-line-no">072</span><span id="line-72"> private long endTime;</span> |
| <span class="source-line-no">073</span><span id="line-73"> private long expectedCount;</span> |
| <span class="source-line-no">074</span><span id="line-74"> private List<String> columns = new ArrayList<>();</span> |
| <span class="source-line-no">075</span><span id="line-75"></span> |
| <span class="source-line-no">076</span><span id="line-76"> /**</span> |
| <span class="source-line-no">077</span><span id="line-77"> * Mapper that runs the count.</span> |
| <span class="source-line-no">078</span><span id="line-78"> */</span> |
| <span class="source-line-no">079</span><span id="line-79"> static class RowCounterMapper extends TableMapper<ImmutableBytesWritable, Result> {</span> |
| <span class="source-line-no">080</span><span id="line-80"></span> |
| <span class="source-line-no">081</span><span id="line-81"> /** Counter enumeration to count the actual rows. */</span> |
| <span class="source-line-no">082</span><span id="line-82"> public static enum Counters {</span> |
| <span class="source-line-no">083</span><span id="line-83"> ROWS</span> |
| <span class="source-line-no">084</span><span id="line-84"> }</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"> * Maps the data.</span> |
| <span class="source-line-no">088</span><span id="line-88"> * @param row The current table row key.</span> |
| <span class="source-line-no">089</span><span id="line-89"> * @param values The columns.</span> |
| <span class="source-line-no">090</span><span id="line-90"> * @param context The current context.</span> |
| <span class="source-line-no">091</span><span id="line-91"> * @throws IOException When something is broken with the data.</span> |
| <span class="source-line-no">092</span><span id="line-92"> * @see org.apache.hadoop.mapreduce.Mapper#map(Object, Object, Context)</span> |
| <span class="source-line-no">093</span><span id="line-93"> */</span> |
| <span class="source-line-no">094</span><span id="line-94"> @Override</span> |
| <span class="source-line-no">095</span><span id="line-95"> public void map(ImmutableBytesWritable row, Result values, Context context) throws IOException {</span> |
| <span class="source-line-no">096</span><span id="line-96"> // Count every row containing data, whether it's in qualifiers or values</span> |
| <span class="source-line-no">097</span><span id="line-97"> context.getCounter(Counters.ROWS).increment(1);</span> |
| <span class="source-line-no">098</span><span id="line-98"> }</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"> /**</span> |
| <span class="source-line-no">102</span><span id="line-102"> * Sets up the actual job.</span> |
| <span class="source-line-no">103</span><span id="line-103"> * @param conf The current configuration.</span> |
| <span class="source-line-no">104</span><span id="line-104"> * @return The newly created job.</span> |
| <span class="source-line-no">105</span><span id="line-105"> * @throws IOException When setting up the job fails.</span> |
| <span class="source-line-no">106</span><span id="line-106"> */</span> |
| <span class="source-line-no">107</span><span id="line-107"> public Job createSubmittableJob(Configuration conf) throws IOException {</span> |
| <span class="source-line-no">108</span><span id="line-108"> Job job = Job.getInstance(conf, conf.get(JOB_NAME_CONF_KEY, NAME + "_" + tableName));</span> |
| <span class="source-line-no">109</span><span id="line-109"> job.setJarByClass(RowCounter.class);</span> |
| <span class="source-line-no">110</span><span id="line-110"> Scan scan = new Scan();</span> |
| <span class="source-line-no">111</span><span id="line-111"> scan.setCacheBlocks(false);</span> |
| <span class="source-line-no">112</span><span id="line-112"> setScanFilter(scan, rowRangeList);</span> |
| <span class="source-line-no">113</span><span id="line-113"></span> |
| <span class="source-line-no">114</span><span id="line-114"> for (String columnName : this.columns) {</span> |
| <span class="source-line-no">115</span><span id="line-115"> String family = StringUtils.substringBefore(columnName, ":");</span> |
| <span class="source-line-no">116</span><span id="line-116"> String qualifier = StringUtils.substringAfter(columnName, ":");</span> |
| <span class="source-line-no">117</span><span id="line-117"> if (StringUtils.isBlank(qualifier)) {</span> |
| <span class="source-line-no">118</span><span id="line-118"> scan.addFamily(Bytes.toBytes(family));</span> |
| <span class="source-line-no">119</span><span id="line-119"> } else {</span> |
| <span class="source-line-no">120</span><span id="line-120"> scan.addColumn(Bytes.toBytes(family), Bytes.toBytes(qualifier));</span> |
| <span class="source-line-no">121</span><span id="line-121"> }</span> |
| <span class="source-line-no">122</span><span id="line-122"> }</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 (this.expectedCount >= 0) {</span> |
| <span class="source-line-no">125</span><span id="line-125"> conf.setLong(EXPECTED_COUNT_KEY, this.expectedCount);</span> |
| <span class="source-line-no">126</span><span id="line-126"> }</span> |
| <span class="source-line-no">127</span><span id="line-127"></span> |
| <span class="source-line-no">128</span><span id="line-128"> scan.setTimeRange(startTime, endTime);</span> |
| <span class="source-line-no">129</span><span id="line-129"> job.setOutputFormatClass(NullOutputFormat.class);</span> |
| <span class="source-line-no">130</span><span id="line-130"> TableMapReduceUtil.initTableMapperJob(tableName, scan, RowCounterMapper.class,</span> |
| <span class="source-line-no">131</span><span id="line-131"> ImmutableBytesWritable.class, Result.class, job);</span> |
| <span class="source-line-no">132</span><span id="line-132"> job.setNumReduceTasks(0);</span> |
| <span class="source-line-no">133</span><span id="line-133"> return job;</span> |
| <span class="source-line-no">134</span><span id="line-134"> }</span> |
| <span class="source-line-no">135</span><span id="line-135"></span> |
| <span class="source-line-no">136</span><span id="line-136"> /**</span> |
| <span class="source-line-no">137</span><span id="line-137"> * Sets up the actual job.</span> |
| <span class="source-line-no">138</span><span id="line-138"> * @param conf The current configuration.</span> |
| <span class="source-line-no">139</span><span id="line-139"> * @param args The command line parameters.</span> |
| <span class="source-line-no">140</span><span id="line-140"> * @return The newly created job.</span> |
| <span class="source-line-no">141</span><span id="line-141"> * @throws IOException When setting up the job fails.</span> |
| <span class="source-line-no">142</span><span id="line-142"> * @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">143</span><span id="line-143"> */</span> |
| <span class="source-line-no">144</span><span id="line-144"> @Deprecated</span> |
| <span class="source-line-no">145</span><span id="line-145"> public static Job createSubmittableJob(Configuration conf, String[] args) throws IOException {</span> |
| <span class="source-line-no">146</span><span id="line-146"> String tableName = args[0];</span> |
| <span class="source-line-no">147</span><span id="line-147"> List<MultiRowRangeFilter.RowRange> rowRangeList = null;</span> |
| <span class="source-line-no">148</span><span id="line-148"> long startTime = 0;</span> |
| <span class="source-line-no">149</span><span id="line-149"> long endTime = 0;</span> |
| <span class="source-line-no">150</span><span id="line-150"></span> |
| <span class="source-line-no">151</span><span id="line-151"> StringBuilder sb = new StringBuilder();</span> |
| <span class="source-line-no">152</span><span id="line-152"></span> |
| <span class="source-line-no">153</span><span id="line-153"> final String rangeSwitch = "--range=";</span> |
| <span class="source-line-no">154</span><span id="line-154"> final String startTimeArgKey = "--starttime=";</span> |
| <span class="source-line-no">155</span><span id="line-155"> final String endTimeArgKey = "--endtime=";</span> |
| <span class="source-line-no">156</span><span id="line-156"> final String expectedCountArg = "--expected-count=";</span> |
| <span class="source-line-no">157</span><span id="line-157"></span> |
| <span class="source-line-no">158</span><span id="line-158"> // First argument is table name, starting from second</span> |
| <span class="source-line-no">159</span><span id="line-159"> for (int i = 1; i < args.length; i++) {</span> |
| <span class="source-line-no">160</span><span id="line-160"> if (args[i].startsWith(rangeSwitch)) {</span> |
| <span class="source-line-no">161</span><span id="line-161"> try {</span> |
| <span class="source-line-no">162</span><span id="line-162"> rowRangeList = parseRowRangeParameter(</span> |
| <span class="source-line-no">163</span><span id="line-163"> args[i].substring(args[1].indexOf(rangeSwitch) + rangeSwitch.length()));</span> |
| <span class="source-line-no">164</span><span id="line-164"> } catch (IllegalArgumentException e) {</span> |
| <span class="source-line-no">165</span><span id="line-165"> return null;</span> |
| <span class="source-line-no">166</span><span id="line-166"> }</span> |
| <span class="source-line-no">167</span><span id="line-167"> continue;</span> |
| <span class="source-line-no">168</span><span id="line-168"> }</span> |
| <span class="source-line-no">169</span><span id="line-169"> if (args[i].startsWith(startTimeArgKey)) {</span> |
| <span class="source-line-no">170</span><span id="line-170"> startTime = Long.parseLong(args[i].substring(startTimeArgKey.length()));</span> |
| <span class="source-line-no">171</span><span id="line-171"> continue;</span> |
| <span class="source-line-no">172</span><span id="line-172"> }</span> |
| <span class="source-line-no">173</span><span id="line-173"> if (args[i].startsWith(endTimeArgKey)) {</span> |
| <span class="source-line-no">174</span><span id="line-174"> endTime = Long.parseLong(args[i].substring(endTimeArgKey.length()));</span> |
| <span class="source-line-no">175</span><span id="line-175"> continue;</span> |
| <span class="source-line-no">176</span><span id="line-176"> }</span> |
| <span class="source-line-no">177</span><span id="line-177"> if (args[i].startsWith(expectedCountArg)) {</span> |
| <span class="source-line-no">178</span><span id="line-178"> conf.setLong(EXPECTED_COUNT_KEY,</span> |
| <span class="source-line-no">179</span><span id="line-179"> Long.parseLong(args[i].substring(expectedCountArg.length())));</span> |
| <span class="source-line-no">180</span><span id="line-180"> continue;</span> |
| <span class="source-line-no">181</span><span id="line-181"> }</span> |
| <span class="source-line-no">182</span><span id="line-182"> // if no switch, assume column names</span> |
| <span class="source-line-no">183</span><span id="line-183"> sb.append(args[i]);</span> |
| <span class="source-line-no">184</span><span id="line-184"> sb.append(" ");</span> |
| <span class="source-line-no">185</span><span id="line-185"> }</span> |
| <span class="source-line-no">186</span><span id="line-186"> if (endTime < startTime) {</span> |
| <span class="source-line-no">187</span><span id="line-187"> printUsage("--endtime=" + endTime + " needs to be greater than --starttime=" + startTime);</span> |
| <span class="source-line-no">188</span><span id="line-188"> return null;</span> |
| <span class="source-line-no">189</span><span id="line-189"> }</span> |
| <span class="source-line-no">190</span><span id="line-190"></span> |
| <span class="source-line-no">191</span><span id="line-191"> Job job = Job.getInstance(conf, conf.get(JOB_NAME_CONF_KEY, NAME + "_" + tableName));</span> |
| <span class="source-line-no">192</span><span id="line-192"> job.setJarByClass(RowCounter.class);</span> |
| <span class="source-line-no">193</span><span id="line-193"> Scan scan = new Scan();</span> |
| <span class="source-line-no">194</span><span id="line-194"> scan.setCacheBlocks(false);</span> |
| <span class="source-line-no">195</span><span id="line-195"> setScanFilter(scan, rowRangeList);</span> |
| <span class="source-line-no">196</span><span id="line-196"> if (sb.length() > 0) {</span> |
| <span class="source-line-no">197</span><span id="line-197"> for (String columnName : sb.toString().trim().split(" ")) {</span> |
| <span class="source-line-no">198</span><span id="line-198"> String family = StringUtils.substringBefore(columnName, ":");</span> |
| <span class="source-line-no">199</span><span id="line-199"> String qualifier = StringUtils.substringAfter(columnName, ":");</span> |
| <span class="source-line-no">200</span><span id="line-200"></span> |
| <span class="source-line-no">201</span><span id="line-201"> if (StringUtils.isBlank(qualifier)) {</span> |
| <span class="source-line-no">202</span><span id="line-202"> scan.addFamily(Bytes.toBytes(family));</span> |
| <span class="source-line-no">203</span><span id="line-203"> } else {</span> |
| <span class="source-line-no">204</span><span id="line-204"> scan.addColumn(Bytes.toBytes(family), Bytes.toBytes(qualifier));</span> |
| <span class="source-line-no">205</span><span id="line-205"> }</span> |
| <span class="source-line-no">206</span><span id="line-206"> }</span> |
| <span class="source-line-no">207</span><span id="line-207"> }</span> |
| <span class="source-line-no">208</span><span id="line-208"> scan.setTimeRange(startTime, endTime == 0 ? HConstants.LATEST_TIMESTAMP : endTime);</span> |
| <span class="source-line-no">209</span><span id="line-209"> job.setOutputFormatClass(NullOutputFormat.class);</span> |
| <span class="source-line-no">210</span><span id="line-210"> TableMapReduceUtil.initTableMapperJob(tableName, scan, RowCounterMapper.class,</span> |
| <span class="source-line-no">211</span><span id="line-211"> ImmutableBytesWritable.class, Result.class, job);</span> |
| <span class="source-line-no">212</span><span id="line-212"> job.setNumReduceTasks(0);</span> |
| <span class="source-line-no">213</span><span id="line-213"> return job;</span> |
| <span class="source-line-no">214</span><span id="line-214"> }</span> |
| <span class="source-line-no">215</span><span id="line-215"></span> |
| <span class="source-line-no">216</span><span id="line-216"> /**</span> |
| <span class="source-line-no">217</span><span id="line-217"> * Prints usage without error message. Note that we don't document --expected-count, because it's</span> |
| <span class="source-line-no">218</span><span id="line-218"> * intended for test.</span> |
| <span class="source-line-no">219</span><span id="line-219"> */</span> |
| <span class="source-line-no">220</span><span id="line-220"> private static void printUsage(String errorMessage) {</span> |
| <span class="source-line-no">221</span><span id="line-221"> System.err.println("ERROR: " + errorMessage);</span> |
| <span class="source-line-no">222</span><span id="line-222"> System.err.println(</span> |
| <span class="source-line-no">223</span><span id="line-223"> "Usage: hbase rowcounter [options] <tablename> " + "[--starttime=<start> --endtime=<end>] "</span> |
| <span class="source-line-no">224</span><span id="line-224"> + "[--range=[startKey],[endKey][;[startKey],[endKey]...]] [<column1> <column2>...]");</span> |
| <span class="source-line-no">225</span><span id="line-225"> System.err.println("For performance consider the following options:\n"</span> |
| <span class="source-line-no">226</span><span id="line-226"> + "-Dhbase.client.scanner.caching=100\n" + "-Dmapreduce.map.speculative=false");</span> |
| <span class="source-line-no">227</span><span id="line-227"> }</span> |
| <span class="source-line-no">228</span><span id="line-228"></span> |
| <span class="source-line-no">229</span><span id="line-229"> private static List<MultiRowRangeFilter.RowRange> parseRowRangeParameter(String arg) {</span> |
| <span class="source-line-no">230</span><span id="line-230"> final List<String> rangesSplit = Splitter.on(";").splitToList(arg);</span> |
| <span class="source-line-no">231</span><span id="line-231"> final List<MultiRowRangeFilter.RowRange> rangeList = new ArrayList<>();</span> |
| <span class="source-line-no">232</span><span id="line-232"> for (String range : rangesSplit) {</span> |
| <span class="source-line-no">233</span><span id="line-233"> if (range != null && !range.isEmpty()) {</span> |
| <span class="source-line-no">234</span><span id="line-234"> List<String> startEnd = Splitter.on(",").splitToList(range);</span> |
| <span class="source-line-no">235</span><span id="line-235"> if (startEnd.size() != 2 || startEnd.get(1).contains(",")) {</span> |
| <span class="source-line-no">236</span><span id="line-236"> throw new IllegalArgumentException("Wrong range specification: " + range);</span> |
| <span class="source-line-no">237</span><span id="line-237"> }</span> |
| <span class="source-line-no">238</span><span id="line-238"> String startKey = startEnd.get(0);</span> |
| <span class="source-line-no">239</span><span id="line-239"> String endKey = startEnd.get(1);</span> |
| <span class="source-line-no">240</span><span id="line-240"> rangeList.add(new MultiRowRangeFilter.RowRange(Bytes.toBytesBinary(startKey), true,</span> |
| <span class="source-line-no">241</span><span id="line-241"> Bytes.toBytesBinary(endKey), false));</span> |
| <span class="source-line-no">242</span><span id="line-242"> }</span> |
| <span class="source-line-no">243</span><span id="line-243"> }</span> |
| <span class="source-line-no">244</span><span id="line-244"> return rangeList;</span> |
| <span class="source-line-no">245</span><span id="line-245"> }</span> |
| <span class="source-line-no">246</span><span id="line-246"></span> |
| <span class="source-line-no">247</span><span id="line-247"> /**</span> |
| <span class="source-line-no">248</span><span id="line-248"> * Sets filter {@link FilterBase} to the {@link Scan} instance. If provided rowRangeList contains</span> |
| <span class="source-line-no">249</span><span id="line-249"> * more than one element, method sets filter which is instance of {@link MultiRowRangeFilter}.</span> |
| <span class="source-line-no">250</span><span id="line-250"> * Otherwise, method sets filter which is instance of {@link FirstKeyOnlyFilter}. If rowRangeList</span> |
| <span class="source-line-no">251</span><span id="line-251"> * contains exactly one element, startRow and stopRow are set to the scan.</span> |
| <span class="source-line-no">252</span><span id="line-252"> */</span> |
| <span class="source-line-no">253</span><span id="line-253"> private static void setScanFilter(Scan scan, List<MultiRowRangeFilter.RowRange> rowRangeList) {</span> |
| <span class="source-line-no">254</span><span id="line-254"> final int size = rowRangeList == null ? 0 : rowRangeList.size();</span> |
| <span class="source-line-no">255</span><span id="line-255"> if (size <= 1) {</span> |
| <span class="source-line-no">256</span><span id="line-256"> scan.setFilter(new FirstKeyOnlyFilter());</span> |
| <span class="source-line-no">257</span><span id="line-257"> }</span> |
| <span class="source-line-no">258</span><span id="line-258"> if (size == 1) {</span> |
| <span class="source-line-no">259</span><span id="line-259"> MultiRowRangeFilter.RowRange range = rowRangeList.get(0);</span> |
| <span class="source-line-no">260</span><span id="line-260"> scan.withStartRow(range.getStartRow()); // inclusive</span> |
| <span class="source-line-no">261</span><span id="line-261"> scan.withStopRow(range.getStopRow()); // exclusive</span> |
| <span class="source-line-no">262</span><span id="line-262"> } else if (size > 1) {</span> |
| <span class="source-line-no">263</span><span id="line-263"> scan.setFilter(new MultiRowRangeFilter(rowRangeList));</span> |
| <span class="source-line-no">264</span><span id="line-264"> }</span> |
| <span class="source-line-no">265</span><span id="line-265"> }</span> |
| <span class="source-line-no">266</span><span id="line-266"></span> |
| <span class="source-line-no">267</span><span id="line-267"> @Override</span> |
| <span class="source-line-no">268</span><span id="line-268"> protected void printUsage() {</span> |
| <span class="source-line-no">269</span><span id="line-269"> StringBuilder footerBuilder = new StringBuilder();</span> |
| <span class="source-line-no">270</span><span id="line-270"> footerBuilder.append("For performance, consider the following configuration properties:\n");</span> |
| <span class="source-line-no">271</span><span id="line-271"> footerBuilder.append("-Dhbase.client.scanner.caching=100\n");</span> |
| <span class="source-line-no">272</span><span id="line-272"> footerBuilder.append("-Dmapreduce.map.speculative=false\n");</span> |
| <span class="source-line-no">273</span><span id="line-273"> printUsage("hbase rowcounter <tablename> [options] [<column1> <column2>...]", "Options:",</span> |
| <span class="source-line-no">274</span><span id="line-274"> footerBuilder.toString());</span> |
| <span class="source-line-no">275</span><span id="line-275"> }</span> |
| <span class="source-line-no">276</span><span id="line-276"></span> |
| <span class="source-line-no">277</span><span id="line-277"> @Override</span> |
| <span class="source-line-no">278</span><span id="line-278"> protected void printUsage(final String usageStr, final String usageHeader,</span> |
| <span class="source-line-no">279</span><span id="line-279"> final String usageFooter) {</span> |
| <span class="source-line-no">280</span><span id="line-280"> HelpFormatter helpFormatter = new HelpFormatter();</span> |
| <span class="source-line-no">281</span><span id="line-281"> helpFormatter.setWidth(120);</span> |
| <span class="source-line-no">282</span><span id="line-282"> helpFormatter.setOptionComparator(new AbstractHBaseTool.OptionsOrderComparator());</span> |
| <span class="source-line-no">283</span><span id="line-283"> helpFormatter.setLongOptSeparator("=");</span> |
| <span class="source-line-no">284</span><span id="line-284"> helpFormatter.printHelp(usageStr, usageHeader, options, usageFooter);</span> |
| <span class="source-line-no">285</span><span id="line-285"> }</span> |
| <span class="source-line-no">286</span><span id="line-286"></span> |
| <span class="source-line-no">287</span><span id="line-287"> @Override</span> |
| <span class="source-line-no">288</span><span id="line-288"> protected void addOptions() {</span> |
| <span class="source-line-no">289</span><span id="line-289"> Option startTimeOption = Option.builder(null).valueSeparator('=').hasArg(true)</span> |
| <span class="source-line-no">290</span><span id="line-290"> .desc("starting time filter to start counting rows from.").longOpt(OPT_START_TIME).build();</span> |
| <span class="source-line-no">291</span><span id="line-291"> Option endTimeOption = Option.builder(null).valueSeparator('=').hasArg(true)</span> |
| <span class="source-line-no">292</span><span id="line-292"> .desc("end time filter limit, to only count rows up to this timestamp.").longOpt(OPT_END_TIME)</span> |
| <span class="source-line-no">293</span><span id="line-293"> .build();</span> |
| <span class="source-line-no">294</span><span id="line-294"> Option rangeOption = Option.builder(null).valueSeparator('=').hasArg(true)</span> |
| <span class="source-line-no">295</span><span id="line-295"> .desc("[startKey],[endKey][;[startKey],[endKey]...]]").longOpt(OPT_RANGE).build();</span> |
| <span class="source-line-no">296</span><span id="line-296"> Option expectedOption = Option.builder(null).valueSeparator('=').hasArg(true)</span> |
| <span class="source-line-no">297</span><span id="line-297"> .desc("expected number of rows to be count.").longOpt(OPT_EXPECTED_COUNT).build();</span> |
| <span class="source-line-no">298</span><span id="line-298"> addOption(startTimeOption);</span> |
| <span class="source-line-no">299</span><span id="line-299"> addOption(endTimeOption);</span> |
| <span class="source-line-no">300</span><span id="line-300"> addOption(rangeOption);</span> |
| <span class="source-line-no">301</span><span id="line-301"> addOption(expectedOption);</span> |
| <span class="source-line-no">302</span><span id="line-302"> }</span> |
| <span class="source-line-no">303</span><span id="line-303"></span> |
| <span class="source-line-no">304</span><span id="line-304"> @Override</span> |
| <span class="source-line-no">305</span><span id="line-305"> protected void processOptions(CommandLine cmd) throws IllegalArgumentException {</span> |
| <span class="source-line-no">306</span><span id="line-306"> this.tableName = cmd.getArgList().get(0);</span> |
| <span class="source-line-no">307</span><span id="line-307"> if (cmd.getOptionValue(OPT_RANGE) != null) {</span> |
| <span class="source-line-no">308</span><span id="line-308"> this.rowRangeList = parseRowRangeParameter(cmd.getOptionValue(OPT_RANGE));</span> |
| <span class="source-line-no">309</span><span id="line-309"> }</span> |
| <span class="source-line-no">310</span><span id="line-310"> this.endTime = cmd.getOptionValue(OPT_END_TIME) == null</span> |
| <span class="source-line-no">311</span><span id="line-311"> ? HConstants.LATEST_TIMESTAMP</span> |
| <span class="source-line-no">312</span><span id="line-312"> : Long.parseLong(cmd.getOptionValue(OPT_END_TIME));</span> |
| <span class="source-line-no">313</span><span id="line-313"> this.expectedCount = cmd.getOptionValue(OPT_EXPECTED_COUNT) == null</span> |
| <span class="source-line-no">314</span><span id="line-314"> ? Long.MIN_VALUE</span> |
| <span class="source-line-no">315</span><span id="line-315"> : Long.parseLong(cmd.getOptionValue(OPT_EXPECTED_COUNT));</span> |
| <span class="source-line-no">316</span><span id="line-316"> this.startTime = cmd.getOptionValue(OPT_START_TIME) == null</span> |
| <span class="source-line-no">317</span><span id="line-317"> ? 0</span> |
| <span class="source-line-no">318</span><span id="line-318"> : Long.parseLong(cmd.getOptionValue(OPT_START_TIME));</span> |
| <span class="source-line-no">319</span><span id="line-319"></span> |
| <span class="source-line-no">320</span><span id="line-320"> for (int i = 1; i < cmd.getArgList().size(); i++) {</span> |
| <span class="source-line-no">321</span><span id="line-321"> String argument = cmd.getArgList().get(i);</span> |
| <span class="source-line-no">322</span><span id="line-322"> if (!argument.startsWith("-")) {</span> |
| <span class="source-line-no">323</span><span id="line-323"> this.columns.add(argument);</span> |
| <span class="source-line-no">324</span><span id="line-324"> }</span> |
| <span class="source-line-no">325</span><span id="line-325"> }</span> |
| <span class="source-line-no">326</span><span id="line-326"></span> |
| <span class="source-line-no">327</span><span id="line-327"> if (endTime < startTime) {</span> |
| <span class="source-line-no">328</span><span id="line-328"> throw new IllegalArgumentException(</span> |
| <span class="source-line-no">329</span><span id="line-329"> "--endtime=" + endTime + " needs to be greater than --starttime=" + startTime);</span> |
| <span class="source-line-no">330</span><span id="line-330"> }</span> |
| <span class="source-line-no">331</span><span id="line-331"> }</span> |
| <span class="source-line-no">332</span><span id="line-332"></span> |
| <span class="source-line-no">333</span><span id="line-333"> @Override</span> |
| <span class="source-line-no">334</span><span id="line-334"> protected void processOldArgs(List<String> args) {</span> |
| <span class="source-line-no">335</span><span id="line-335"> List<String> copiedArgs = new ArrayList<>(args);</span> |
| <span class="source-line-no">336</span><span id="line-336"> args.removeAll(copiedArgs);</span> |
| <span class="source-line-no">337</span><span id="line-337"> for (String arg : copiedArgs) {</span> |
| <span class="source-line-no">338</span><span id="line-338"> if (arg.startsWith("-") && arg.contains("=")) {</span> |
| <span class="source-line-no">339</span><span id="line-339"> String[] kv = arg.split("=");</span> |
| <span class="source-line-no">340</span><span id="line-340"> args.add(kv[0]);</span> |
| <span class="source-line-no">341</span><span id="line-341"> args.add(kv[1]);</span> |
| <span class="source-line-no">342</span><span id="line-342"> } else {</span> |
| <span class="source-line-no">343</span><span id="line-343"> args.add(arg);</span> |
| <span class="source-line-no">344</span><span id="line-344"> }</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"></span> |
| <span class="source-line-no">348</span><span id="line-348"> @Override</span> |
| <span class="source-line-no">349</span><span id="line-349"> protected int doWork() throws Exception {</span> |
| <span class="source-line-no">350</span><span id="line-350"> Job job = createSubmittableJob(getConf());</span> |
| <span class="source-line-no">351</span><span id="line-351"> if (job == null) {</span> |
| <span class="source-line-no">352</span><span id="line-352"> return -1;</span> |
| <span class="source-line-no">353</span><span id="line-353"> }</span> |
| <span class="source-line-no">354</span><span id="line-354"> boolean success = job.waitForCompletion(true);</span> |
| <span class="source-line-no">355</span><span id="line-355"> final long expectedCount = getConf().getLong(EXPECTED_COUNT_KEY, -1);</span> |
| <span class="source-line-no">356</span><span id="line-356"> if (success && expectedCount != -1) {</span> |
| <span class="source-line-no">357</span><span id="line-357"> final Counter counter = job.getCounters().findCounter(RowCounterMapper.Counters.ROWS);</span> |
| <span class="source-line-no">358</span><span id="line-358"> success = expectedCount == counter.getValue();</span> |
| <span class="source-line-no">359</span><span id="line-359"> if (!success) {</span> |
| <span class="source-line-no">360</span><span id="line-360"> LOG.error("Failing job because count of '" + counter.getValue()</span> |
| <span class="source-line-no">361</span><span id="line-361"> + "' does not match expected count of '" + expectedCount + "'");</span> |
| <span class="source-line-no">362</span><span id="line-362"> }</span> |
| <span class="source-line-no">363</span><span id="line-363"> }</span> |
| <span class="source-line-no">364</span><span id="line-364"> return (success ? 0 : 1);</span> |
| <span class="source-line-no">365</span><span id="line-365"> }</span> |
| <span class="source-line-no">366</span><span id="line-366"></span> |
| <span class="source-line-no">367</span><span id="line-367"> /**</span> |
| <span class="source-line-no">368</span><span id="line-368"> * Main entry point.</span> |
| <span class="source-line-no">369</span><span id="line-369"> * @param args The command line parameters.</span> |
| <span class="source-line-no">370</span><span id="line-370"> * @throws Exception When running the job fails.</span> |
| <span class="source-line-no">371</span><span id="line-371"> */</span> |
| <span class="source-line-no">372</span><span id="line-372"> public static void main(String[] args) throws Exception {</span> |
| <span class="source-line-no">373</span><span id="line-373"> new RowCounter().doStaticMain(args);</span> |
| <span class="source-line-no">374</span><span id="line-374"> }</span> |
| <span class="source-line-no">375</span><span id="line-375"></span> |
| <span class="source-line-no">376</span><span id="line-376"> static class RowCounterCommandLineParser extends BasicParser {</span> |
| <span class="source-line-no">377</span><span id="line-377"></span> |
| <span class="source-line-no">378</span><span id="line-378"> @Override</span> |
| <span class="source-line-no">379</span><span id="line-379"> protected void checkRequiredOptions() throws MissingOptionException {</span> |
| <span class="source-line-no">380</span><span id="line-380"> if (this.cmd.getArgList().size() < 1 || this.cmd.getArgList().get(0).startsWith("-")) {</span> |
| <span class="source-line-no">381</span><span id="line-381"> throw new MissingOptionException("First argument must be a valid table name.");</span> |
| <span class="source-line-no">382</span><span id="line-382"> }</span> |
| <span class="source-line-no">383</span><span id="line-383"> }</span> |
| <span class="source-line-no">384</span><span id="line-384"> }</span> |
| <span class="source-line-no">385</span><span id="line-385"></span> |
| <span class="source-line-no">386</span><span id="line-386"> @Override</span> |
| <span class="source-line-no">387</span><span id="line-387"> protected CommandLineParser newParser() {</span> |
| <span class="source-line-no">388</span><span id="line-388"> return new RowCounterCommandLineParser();</span> |
| <span class="source-line-no">389</span><span id="line-389"> }</span> |
| <span class="source-line-no">390</span><span id="line-390"></span> |
| <span class="source-line-no">391</span><span id="line-391">}</span> |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| </pre> |
| </div> |
| </main> |
| </body> |
| </html> |