blob: e2f043c2eb526b809943de5b8855f306a2b4f9de [file] [log] [blame]
<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/html4/loose.dtd">
<html lang="en">
<head>
<title>Source code</title>
<link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style">
</head>
<body>
<div class="sourceContainer">
<pre><span class="sourceLineNo">001</span>/*<a name="line.1"></a>
<span class="sourceLineNo">002</span> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.2"></a>
<span class="sourceLineNo">003</span> * or more contributor license agreements. See the NOTICE file<a name="line.3"></a>
<span class="sourceLineNo">004</span> * distributed with this work for additional information<a name="line.4"></a>
<span class="sourceLineNo">005</span> * regarding copyright ownership. The ASF licenses this file<a name="line.5"></a>
<span class="sourceLineNo">006</span> * to you under the Apache License, Version 2.0 (the<a name="line.6"></a>
<span class="sourceLineNo">007</span> * "License"); you may not use this file except in compliance<a name="line.7"></a>
<span class="sourceLineNo">008</span> * with the License. You may obtain a copy of the License at<a name="line.8"></a>
<span class="sourceLineNo">009</span> *<a name="line.9"></a>
<span class="sourceLineNo">010</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.10"></a>
<span class="sourceLineNo">011</span> *<a name="line.11"></a>
<span class="sourceLineNo">012</span> * Unless required by applicable law or agreed to in writing, software<a name="line.12"></a>
<span class="sourceLineNo">013</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.13"></a>
<span class="sourceLineNo">014</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.14"></a>
<span class="sourceLineNo">015</span> * See the License for the specific language governing permissions and<a name="line.15"></a>
<span class="sourceLineNo">016</span> * limitations under the License.<a name="line.16"></a>
<span class="sourceLineNo">017</span> */<a name="line.17"></a>
<span class="sourceLineNo">018</span>package org.apache.hadoop.hbase.mapreduce;<a name="line.18"></a>
<span class="sourceLineNo">019</span><a name="line.19"></a>
<span class="sourceLineNo">020</span>import java.io.IOException;<a name="line.20"></a>
<span class="sourceLineNo">021</span>import org.apache.hadoop.conf.Configuration;<a name="line.21"></a>
<span class="sourceLineNo">022</span>import org.apache.hadoop.conf.Configured;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import org.apache.hadoop.fs.Path;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import org.apache.hadoop.hbase.Cell;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import org.apache.hadoop.hbase.CellUtil;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.CompareOperator;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.HConstants;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.apache.hadoop.hbase.client.Result;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.apache.hadoop.hbase.client.Scan;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import org.apache.hadoop.hbase.filter.Filter;<a name="line.31"></a>
<span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.filter.PrefixFilter;<a name="line.32"></a>
<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.filter.RegexStringComparator;<a name="line.33"></a>
<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.filter.RowFilter;<a name="line.34"></a>
<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.35"></a>
<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.36"></a>
<span class="sourceLineNo">037</span>import org.apache.hadoop.io.LongWritable;<a name="line.37"></a>
<span class="sourceLineNo">038</span>import org.apache.hadoop.io.Text;<a name="line.38"></a>
<span class="sourceLineNo">039</span>import org.apache.hadoop.mapreduce.Job;<a name="line.39"></a>
<span class="sourceLineNo">040</span>import org.apache.hadoop.mapreduce.Reducer;<a name="line.40"></a>
<span class="sourceLineNo">041</span>import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;<a name="line.41"></a>
<span class="sourceLineNo">042</span>import org.apache.hadoop.mapreduce.lib.output.TextOutputFormat;<a name="line.42"></a>
<span class="sourceLineNo">043</span>import org.apache.hadoop.util.Tool;<a name="line.43"></a>
<span class="sourceLineNo">044</span>import org.apache.hadoop.util.ToolRunner;<a name="line.44"></a>
<span class="sourceLineNo">045</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.45"></a>
<span class="sourceLineNo">046</span>import org.slf4j.Logger;<a name="line.46"></a>
<span class="sourceLineNo">047</span>import org.slf4j.LoggerFactory;<a name="line.47"></a>
<span class="sourceLineNo">048</span><a name="line.48"></a>
<span class="sourceLineNo">049</span>import org.apache.hbase.thirdparty.com.google.common.base.Preconditions;<a name="line.49"></a>
<span class="sourceLineNo">050</span><a name="line.50"></a>
<span class="sourceLineNo">051</span>/**<a name="line.51"></a>
<span class="sourceLineNo">052</span> * A job with a a map and reduce phase to count cells in a table. The counter lists the following<a name="line.52"></a>
<span class="sourceLineNo">053</span> * stats for a given table:<a name="line.53"></a>
<span class="sourceLineNo">054</span> *<a name="line.54"></a>
<span class="sourceLineNo">055</span> * &lt;pre&gt;<a name="line.55"></a>
<span class="sourceLineNo">056</span> * 1. Total number of rows in the table<a name="line.56"></a>
<span class="sourceLineNo">057</span> * 2. Total number of CFs across all rows<a name="line.57"></a>
<span class="sourceLineNo">058</span> * 3. Total qualifiers across all rows<a name="line.58"></a>
<span class="sourceLineNo">059</span> * 4. Total occurrence of each CF<a name="line.59"></a>
<span class="sourceLineNo">060</span> * 5. Total occurrence of each qualifier<a name="line.60"></a>
<span class="sourceLineNo">061</span> * 6. Total number of versions of each qualifier.<a name="line.61"></a>
<span class="sourceLineNo">062</span> * 7. Total size of serialized cells of each CF.<a name="line.62"></a>
<span class="sourceLineNo">063</span> * 8. Total size of serialized cells of each qualifier.<a name="line.63"></a>
<span class="sourceLineNo">064</span> * 9. Total size of serialized cells across all rows.<a name="line.64"></a>
<span class="sourceLineNo">065</span> * &lt;/pre&gt;<a name="line.65"></a>
<span class="sourceLineNo">066</span> *<a name="line.66"></a>
<span class="sourceLineNo">067</span> * The cellcounter can take optional parameters to use a user supplied row/family/qualifier string<a name="line.67"></a>
<span class="sourceLineNo">068</span> * to use in the report and second a regex based or prefix based row filter to restrict the count<a name="line.68"></a>
<span class="sourceLineNo">069</span> * operation to a limited subset of rows from the table or a start time and/or end time to limit the<a name="line.69"></a>
<span class="sourceLineNo">070</span> * count to a time range.<a name="line.70"></a>
<span class="sourceLineNo">071</span> */<a name="line.71"></a>
<span class="sourceLineNo">072</span>@InterfaceAudience.Public<a name="line.72"></a>
<span class="sourceLineNo">073</span>public class CellCounter extends Configured implements Tool {<a name="line.73"></a>
<span class="sourceLineNo">074</span> private static final Logger LOG = LoggerFactory.getLogger(CellCounter.class.getName());<a name="line.74"></a>
<span class="sourceLineNo">075</span><a name="line.75"></a>
<span class="sourceLineNo">076</span> /**<a name="line.76"></a>
<span class="sourceLineNo">077</span> * Name of this 'program'.<a name="line.77"></a>
<span class="sourceLineNo">078</span> */<a name="line.78"></a>
<span class="sourceLineNo">079</span> static final String NAME = "CellCounter";<a name="line.79"></a>
<span class="sourceLineNo">080</span><a name="line.80"></a>
<span class="sourceLineNo">081</span> private final static String JOB_NAME_CONF_KEY = "mapreduce.job.name";<a name="line.81"></a>
<span class="sourceLineNo">082</span><a name="line.82"></a>
<span class="sourceLineNo">083</span> /**<a name="line.83"></a>
<span class="sourceLineNo">084</span> * Mapper that runs the count.<a name="line.84"></a>
<span class="sourceLineNo">085</span> */<a name="line.85"></a>
<span class="sourceLineNo">086</span> static class CellCounterMapper extends TableMapper&lt;Text, LongWritable&gt; {<a name="line.86"></a>
<span class="sourceLineNo">087</span> /**<a name="line.87"></a>
<span class="sourceLineNo">088</span> * Counter enumeration to count the actual rows.<a name="line.88"></a>
<span class="sourceLineNo">089</span> */<a name="line.89"></a>
<span class="sourceLineNo">090</span> public static enum Counters {<a name="line.90"></a>
<span class="sourceLineNo">091</span> ROWS,<a name="line.91"></a>
<span class="sourceLineNo">092</span> CELLS,<a name="line.92"></a>
<span class="sourceLineNo">093</span> SIZE<a name="line.93"></a>
<span class="sourceLineNo">094</span> }<a name="line.94"></a>
<span class="sourceLineNo">095</span><a name="line.95"></a>
<span class="sourceLineNo">096</span> private Configuration conf;<a name="line.96"></a>
<span class="sourceLineNo">097</span> private String separator;<a name="line.97"></a>
<span class="sourceLineNo">098</span><a name="line.98"></a>
<span class="sourceLineNo">099</span> // state of current row, family, column needs to persist across map() invocations<a name="line.99"></a>
<span class="sourceLineNo">100</span> // in order to properly handle scanner batching, where a single qualifier may have too<a name="line.100"></a>
<span class="sourceLineNo">101</span> // many versions for a single map() call<a name="line.101"></a>
<span class="sourceLineNo">102</span> private byte[] lastRow;<a name="line.102"></a>
<span class="sourceLineNo">103</span> private String currentRowKey;<a name="line.103"></a>
<span class="sourceLineNo">104</span> byte[] currentFamily = null;<a name="line.104"></a>
<span class="sourceLineNo">105</span> String currentFamilyName = null;<a name="line.105"></a>
<span class="sourceLineNo">106</span> byte[] currentQualifier = null;<a name="line.106"></a>
<span class="sourceLineNo">107</span> // family + qualifier<a name="line.107"></a>
<span class="sourceLineNo">108</span> String currentQualifierName = null;<a name="line.108"></a>
<span class="sourceLineNo">109</span> // rowkey + family + qualifier<a name="line.109"></a>
<span class="sourceLineNo">110</span> String currentRowQualifierName = null;<a name="line.110"></a>
<span class="sourceLineNo">111</span><a name="line.111"></a>
<span class="sourceLineNo">112</span> @Override<a name="line.112"></a>
<span class="sourceLineNo">113</span> protected void setup(Context context) throws IOException, InterruptedException {<a name="line.113"></a>
<span class="sourceLineNo">114</span> conf = context.getConfiguration();<a name="line.114"></a>
<span class="sourceLineNo">115</span> separator = conf.get("ReportSeparator", ":");<a name="line.115"></a>
<span class="sourceLineNo">116</span> }<a name="line.116"></a>
<span class="sourceLineNo">117</span><a name="line.117"></a>
<span class="sourceLineNo">118</span> /**<a name="line.118"></a>
<span class="sourceLineNo">119</span> * Maps the data.<a name="line.119"></a>
<span class="sourceLineNo">120</span> * @param row The current table row key.<a name="line.120"></a>
<span class="sourceLineNo">121</span> * @param values The columns.<a name="line.121"></a>
<span class="sourceLineNo">122</span> * @param context The current context.<a name="line.122"></a>
<span class="sourceLineNo">123</span> * @throws IOException When something is broken with the data.<a name="line.123"></a>
<span class="sourceLineNo">124</span> */<a name="line.124"></a>
<span class="sourceLineNo">125</span><a name="line.125"></a>
<span class="sourceLineNo">126</span> @Override<a name="line.126"></a>
<span class="sourceLineNo">127</span> @edu.umd.cs.findbugs.annotations.SuppressWarnings(value = "NP_NULL_ON_SOME_PATH",<a name="line.127"></a>
<span class="sourceLineNo">128</span> justification = "Findbugs is blind to the Precondition null check")<a name="line.128"></a>
<span class="sourceLineNo">129</span> public void map(ImmutableBytesWritable row, Result values, Context context) throws IOException {<a name="line.129"></a>
<span class="sourceLineNo">130</span> Preconditions.checkState(values != null, "values passed to the map is null");<a name="line.130"></a>
<span class="sourceLineNo">131</span><a name="line.131"></a>
<span class="sourceLineNo">132</span> try {<a name="line.132"></a>
<span class="sourceLineNo">133</span> byte[] currentRow = values.getRow();<a name="line.133"></a>
<span class="sourceLineNo">134</span> if (lastRow == null || !Bytes.equals(lastRow, currentRow)) {<a name="line.134"></a>
<span class="sourceLineNo">135</span> lastRow = currentRow;<a name="line.135"></a>
<span class="sourceLineNo">136</span> currentRowKey = Bytes.toStringBinary(currentRow);<a name="line.136"></a>
<span class="sourceLineNo">137</span> currentFamily = null;<a name="line.137"></a>
<span class="sourceLineNo">138</span> currentQualifier = null;<a name="line.138"></a>
<span class="sourceLineNo">139</span> context.getCounter(Counters.ROWS).increment(1);<a name="line.139"></a>
<span class="sourceLineNo">140</span> context.write(new Text("Total ROWS"), new LongWritable(1));<a name="line.140"></a>
<span class="sourceLineNo">141</span> }<a name="line.141"></a>
<span class="sourceLineNo">142</span> if (!values.isEmpty()) {<a name="line.142"></a>
<span class="sourceLineNo">143</span> int cellCount = 0;<a name="line.143"></a>
<span class="sourceLineNo">144</span> for (Cell value : values.listCells()) {<a name="line.144"></a>
<span class="sourceLineNo">145</span> cellCount++;<a name="line.145"></a>
<span class="sourceLineNo">146</span> long size = value.getSerializedSize();<a name="line.146"></a>
<span class="sourceLineNo">147</span> if (currentFamily == null || !CellUtil.matchingFamily(value, currentFamily)) {<a name="line.147"></a>
<span class="sourceLineNo">148</span> currentFamily = CellUtil.cloneFamily(value);<a name="line.148"></a>
<span class="sourceLineNo">149</span> currentFamilyName = Bytes.toStringBinary(currentFamily);<a name="line.149"></a>
<span class="sourceLineNo">150</span> currentQualifier = null;<a name="line.150"></a>
<span class="sourceLineNo">151</span> context.getCounter("CF", currentFamilyName).increment(1);<a name="line.151"></a>
<span class="sourceLineNo">152</span> if (1 == context.getCounter("CF", currentFamilyName).getValue()) {<a name="line.152"></a>
<span class="sourceLineNo">153</span> context.write(new Text("Total Families Across all Rows"), new LongWritable(1));<a name="line.153"></a>
<span class="sourceLineNo">154</span> context.write(new Text(currentFamily), new LongWritable(1));<a name="line.154"></a>
<span class="sourceLineNo">155</span> }<a name="line.155"></a>
<span class="sourceLineNo">156</span> context.getCounter(Counters.SIZE).increment(size);<a name="line.156"></a>
<span class="sourceLineNo">157</span> context.write(new Text("Total SIZE"), new LongWritable(size));<a name="line.157"></a>
<span class="sourceLineNo">158</span> context.getCounter("CF", currentFamilyName + "_Size").increment(size);<a name="line.158"></a>
<span class="sourceLineNo">159</span> context.write(new Text(currentFamilyName + "_Size"), new LongWritable(size));<a name="line.159"></a>
<span class="sourceLineNo">160</span> }<a name="line.160"></a>
<span class="sourceLineNo">161</span> if (currentQualifier == null || !CellUtil.matchingQualifier(value, currentQualifier)) {<a name="line.161"></a>
<span class="sourceLineNo">162</span> currentQualifier = CellUtil.cloneQualifier(value);<a name="line.162"></a>
<span class="sourceLineNo">163</span> currentQualifierName =<a name="line.163"></a>
<span class="sourceLineNo">164</span> currentFamilyName + separator + Bytes.toStringBinary(currentQualifier);<a name="line.164"></a>
<span class="sourceLineNo">165</span> currentRowQualifierName = currentRowKey + separator + currentQualifierName;<a name="line.165"></a>
<span class="sourceLineNo">166</span><a name="line.166"></a>
<span class="sourceLineNo">167</span> context.write(new Text("Total Qualifiers across all Rows"), new LongWritable(1));<a name="line.167"></a>
<span class="sourceLineNo">168</span> context.write(new Text(currentQualifierName), new LongWritable(1));<a name="line.168"></a>
<span class="sourceLineNo">169</span> context.getCounter("Q", currentQualifierName + "_Size").increment(size);<a name="line.169"></a>
<span class="sourceLineNo">170</span> context.write(new Text(currentQualifierName + "_Size"), new LongWritable(size));<a name="line.170"></a>
<span class="sourceLineNo">171</span> }<a name="line.171"></a>
<span class="sourceLineNo">172</span> // Increment versions<a name="line.172"></a>
<span class="sourceLineNo">173</span> context.write(new Text(currentRowQualifierName + "_Versions"), new LongWritable(1));<a name="line.173"></a>
<span class="sourceLineNo">174</span> }<a name="line.174"></a>
<span class="sourceLineNo">175</span> context.getCounter(Counters.CELLS).increment(cellCount);<a name="line.175"></a>
<span class="sourceLineNo">176</span> }<a name="line.176"></a>
<span class="sourceLineNo">177</span> } catch (InterruptedException e) {<a name="line.177"></a>
<span class="sourceLineNo">178</span> LOG.error("Interrupted while writing cellCount", e);<a name="line.178"></a>
<span class="sourceLineNo">179</span> Thread.currentThread().interrupt();<a name="line.179"></a>
<span class="sourceLineNo">180</span> }<a name="line.180"></a>
<span class="sourceLineNo">181</span> }<a name="line.181"></a>
<span class="sourceLineNo">182</span> }<a name="line.182"></a>
<span class="sourceLineNo">183</span><a name="line.183"></a>
<span class="sourceLineNo">184</span> static class LongSumReducer&lt;Key&gt; extends Reducer&lt;Key, LongWritable, Key, LongWritable&gt; {<a name="line.184"></a>
<span class="sourceLineNo">185</span><a name="line.185"></a>
<span class="sourceLineNo">186</span> private LongWritable result = new LongWritable();<a name="line.186"></a>
<span class="sourceLineNo">187</span><a name="line.187"></a>
<span class="sourceLineNo">188</span> public void reduce(Key key, Iterable&lt;LongWritable&gt; values, Context context)<a name="line.188"></a>
<span class="sourceLineNo">189</span> throws IOException, InterruptedException {<a name="line.189"></a>
<span class="sourceLineNo">190</span> long sum = 0;<a name="line.190"></a>
<span class="sourceLineNo">191</span> for (LongWritable val : values) {<a name="line.191"></a>
<span class="sourceLineNo">192</span> sum += val.get();<a name="line.192"></a>
<span class="sourceLineNo">193</span> }<a name="line.193"></a>
<span class="sourceLineNo">194</span> result.set(sum);<a name="line.194"></a>
<span class="sourceLineNo">195</span> context.write(key, result);<a name="line.195"></a>
<span class="sourceLineNo">196</span> }<a name="line.196"></a>
<span class="sourceLineNo">197</span><a name="line.197"></a>
<span class="sourceLineNo">198</span> }<a name="line.198"></a>
<span class="sourceLineNo">199</span><a name="line.199"></a>
<span class="sourceLineNo">200</span> /**<a name="line.200"></a>
<span class="sourceLineNo">201</span> * Sets up the actual job.<a name="line.201"></a>
<span class="sourceLineNo">202</span> * @param conf The current configuration.<a name="line.202"></a>
<span class="sourceLineNo">203</span> * @param args The command line parameters.<a name="line.203"></a>
<span class="sourceLineNo">204</span> * @return The newly created job.<a name="line.204"></a>
<span class="sourceLineNo">205</span> * @throws IOException When setting up the job fails.<a name="line.205"></a>
<span class="sourceLineNo">206</span> */<a name="line.206"></a>
<span class="sourceLineNo">207</span> public static Job createSubmittableJob(Configuration conf, String[] args) throws IOException {<a name="line.207"></a>
<span class="sourceLineNo">208</span> String tableName = args[0];<a name="line.208"></a>
<span class="sourceLineNo">209</span> Path outputDir = new Path(args[1]);<a name="line.209"></a>
<span class="sourceLineNo">210</span> String reportSeparatorString = (args.length &gt; 2) ? args[2] : ":";<a name="line.210"></a>
<span class="sourceLineNo">211</span> conf.set("ReportSeparator", reportSeparatorString);<a name="line.211"></a>
<span class="sourceLineNo">212</span> Job job = Job.getInstance(conf, conf.get(JOB_NAME_CONF_KEY, NAME + "_" + tableName));<a name="line.212"></a>
<span class="sourceLineNo">213</span> job.setJarByClass(CellCounter.class);<a name="line.213"></a>
<span class="sourceLineNo">214</span> Scan scan = getConfiguredScanForJob(conf, args);<a name="line.214"></a>
<span class="sourceLineNo">215</span> TableMapReduceUtil.initTableMapperJob(tableName, scan, CellCounterMapper.class,<a name="line.215"></a>
<span class="sourceLineNo">216</span> ImmutableBytesWritable.class, Result.class, job);<a name="line.216"></a>
<span class="sourceLineNo">217</span> job.setMapOutputKeyClass(Text.class);<a name="line.217"></a>
<span class="sourceLineNo">218</span> job.setMapOutputValueClass(LongWritable.class);<a name="line.218"></a>
<span class="sourceLineNo">219</span> job.setOutputFormatClass(TextOutputFormat.class);<a name="line.219"></a>
<span class="sourceLineNo">220</span> job.setOutputKeyClass(Text.class);<a name="line.220"></a>
<span class="sourceLineNo">221</span> job.setOutputValueClass(LongWritable.class);<a name="line.221"></a>
<span class="sourceLineNo">222</span> FileOutputFormat.setOutputPath(job, outputDir);<a name="line.222"></a>
<span class="sourceLineNo">223</span> job.setReducerClass(LongSumReducer.class);<a name="line.223"></a>
<span class="sourceLineNo">224</span> job.setCombinerClass(LongSumReducer.class);<a name="line.224"></a>
<span class="sourceLineNo">225</span> return job;<a name="line.225"></a>
<span class="sourceLineNo">226</span> }<a name="line.226"></a>
<span class="sourceLineNo">227</span><a name="line.227"></a>
<span class="sourceLineNo">228</span> private static Scan getConfiguredScanForJob(Configuration conf, String[] args)<a name="line.228"></a>
<span class="sourceLineNo">229</span> throws IOException {<a name="line.229"></a>
<span class="sourceLineNo">230</span> // create scan with any properties set from TableInputFormat<a name="line.230"></a>
<span class="sourceLineNo">231</span> Scan s = TableInputFormat.createScanFromConfiguration(conf);<a name="line.231"></a>
<span class="sourceLineNo">232</span> // Set Scan Versions<a name="line.232"></a>
<span class="sourceLineNo">233</span> if (conf.get(TableInputFormat.SCAN_MAXVERSIONS) == null) {<a name="line.233"></a>
<span class="sourceLineNo">234</span> // default to all versions unless explicitly set<a name="line.234"></a>
<span class="sourceLineNo">235</span> s.readVersions(Integer.MAX_VALUE);<a name="line.235"></a>
<span class="sourceLineNo">236</span> }<a name="line.236"></a>
<span class="sourceLineNo">237</span> s.setCacheBlocks(false);<a name="line.237"></a>
<span class="sourceLineNo">238</span> // Set RowFilter or Prefix Filter if applicable.<a name="line.238"></a>
<span class="sourceLineNo">239</span> Filter rowFilter = getRowFilter(args);<a name="line.239"></a>
<span class="sourceLineNo">240</span> if (rowFilter != null) {<a name="line.240"></a>
<span class="sourceLineNo">241</span> LOG.info("Setting Row Filter for counter.");<a name="line.241"></a>
<span class="sourceLineNo">242</span> s.setFilter(rowFilter);<a name="line.242"></a>
<span class="sourceLineNo">243</span> }<a name="line.243"></a>
<span class="sourceLineNo">244</span> // Set TimeRange if defined<a name="line.244"></a>
<span class="sourceLineNo">245</span> long timeRange[] = getTimeRange(args);<a name="line.245"></a>
<span class="sourceLineNo">246</span> if (timeRange != null) {<a name="line.246"></a>
<span class="sourceLineNo">247</span> LOG.info("Setting TimeRange for counter.");<a name="line.247"></a>
<span class="sourceLineNo">248</span> s.setTimeRange(timeRange[0], timeRange[1]);<a name="line.248"></a>
<span class="sourceLineNo">249</span> }<a name="line.249"></a>
<span class="sourceLineNo">250</span> return s;<a name="line.250"></a>
<span class="sourceLineNo">251</span> }<a name="line.251"></a>
<span class="sourceLineNo">252</span><a name="line.252"></a>
<span class="sourceLineNo">253</span> private static Filter getRowFilter(String[] args) {<a name="line.253"></a>
<span class="sourceLineNo">254</span> Filter rowFilter = null;<a name="line.254"></a>
<span class="sourceLineNo">255</span> String filterCriteria = (args.length &gt; 3) ? args[3] : null;<a name="line.255"></a>
<span class="sourceLineNo">256</span> if (filterCriteria == null) return null;<a name="line.256"></a>
<span class="sourceLineNo">257</span> if (filterCriteria.startsWith("^")) {<a name="line.257"></a>
<span class="sourceLineNo">258</span> String regexPattern = filterCriteria.substring(1, filterCriteria.length());<a name="line.258"></a>
<span class="sourceLineNo">259</span> rowFilter = new RowFilter(CompareOperator.EQUAL, new RegexStringComparator(regexPattern));<a name="line.259"></a>
<span class="sourceLineNo">260</span> } else {<a name="line.260"></a>
<span class="sourceLineNo">261</span> rowFilter = new PrefixFilter(Bytes.toBytesBinary(filterCriteria));<a name="line.261"></a>
<span class="sourceLineNo">262</span> }<a name="line.262"></a>
<span class="sourceLineNo">263</span> return rowFilter;<a name="line.263"></a>
<span class="sourceLineNo">264</span> }<a name="line.264"></a>
<span class="sourceLineNo">265</span><a name="line.265"></a>
<span class="sourceLineNo">266</span> private static long[] getTimeRange(String[] args) throws IOException {<a name="line.266"></a>
<span class="sourceLineNo">267</span> final String startTimeArgKey = "--starttime=";<a name="line.267"></a>
<span class="sourceLineNo">268</span> final String endTimeArgKey = "--endtime=";<a name="line.268"></a>
<span class="sourceLineNo">269</span> long startTime = 0L;<a name="line.269"></a>
<span class="sourceLineNo">270</span> long endTime = 0L;<a name="line.270"></a>
<span class="sourceLineNo">271</span><a name="line.271"></a>
<span class="sourceLineNo">272</span> for (int i = 1; i &lt; args.length; i++) {<a name="line.272"></a>
<span class="sourceLineNo">273</span> System.out.println("i:" + i + "arg[i]" + args[i]);<a name="line.273"></a>
<span class="sourceLineNo">274</span> if (args[i].startsWith(startTimeArgKey)) {<a name="line.274"></a>
<span class="sourceLineNo">275</span> startTime = Long.parseLong(args[i].substring(startTimeArgKey.length()));<a name="line.275"></a>
<span class="sourceLineNo">276</span> }<a name="line.276"></a>
<span class="sourceLineNo">277</span> if (args[i].startsWith(endTimeArgKey)) {<a name="line.277"></a>
<span class="sourceLineNo">278</span> endTime = Long.parseLong(args[i].substring(endTimeArgKey.length()));<a name="line.278"></a>
<span class="sourceLineNo">279</span> }<a name="line.279"></a>
<span class="sourceLineNo">280</span> }<a name="line.280"></a>
<span class="sourceLineNo">281</span><a name="line.281"></a>
<span class="sourceLineNo">282</span> if (startTime == 0 &amp;&amp; endTime == 0) return null;<a name="line.282"></a>
<span class="sourceLineNo">283</span><a name="line.283"></a>
<span class="sourceLineNo">284</span> endTime = endTime == 0 ? HConstants.LATEST_TIMESTAMP : endTime;<a name="line.284"></a>
<span class="sourceLineNo">285</span> return new long[] { startTime, endTime };<a name="line.285"></a>
<span class="sourceLineNo">286</span> }<a name="line.286"></a>
<span class="sourceLineNo">287</span><a name="line.287"></a>
<span class="sourceLineNo">288</span> @Override<a name="line.288"></a>
<span class="sourceLineNo">289</span> public int run(String[] args) throws Exception {<a name="line.289"></a>
<span class="sourceLineNo">290</span> if (args.length &lt; 2) {<a name="line.290"></a>
<span class="sourceLineNo">291</span> printUsage(args.length);<a name="line.291"></a>
<span class="sourceLineNo">292</span> return -1;<a name="line.292"></a>
<span class="sourceLineNo">293</span> }<a name="line.293"></a>
<span class="sourceLineNo">294</span> Job job = createSubmittableJob(getConf(), args);<a name="line.294"></a>
<span class="sourceLineNo">295</span> return (job.waitForCompletion(true) ? 0 : 1);<a name="line.295"></a>
<span class="sourceLineNo">296</span> }<a name="line.296"></a>
<span class="sourceLineNo">297</span><a name="line.297"></a>
<span class="sourceLineNo">298</span> private void printUsage(int parameterCount) {<a name="line.298"></a>
<span class="sourceLineNo">299</span> System.err.println("ERROR: Wrong number of parameters: " + parameterCount);<a name="line.299"></a>
<span class="sourceLineNo">300</span> System.err.println("Usage: hbase cellcounter &lt;tablename&gt; &lt;outputDir&gt; [reportSeparator] "<a name="line.300"></a>
<span class="sourceLineNo">301</span> + "[^[regex pattern] or [Prefix]] [--starttime=&lt;starttime&gt; --endtime=&lt;endtime&gt;]");<a name="line.301"></a>
<span class="sourceLineNo">302</span> System.err.println(" Note: -D properties will be applied to the conf used.");<a name="line.302"></a>
<span class="sourceLineNo">303</span> System.err.println(" Additionally, all of the SCAN properties from TableInputFormat can be "<a name="line.303"></a>
<span class="sourceLineNo">304</span> + "specified to get fine grained control on what is counted.");<a name="line.304"></a>
<span class="sourceLineNo">305</span> System.err.println(" -D" + TableInputFormat.SCAN_ROW_START + "=&lt;rowkey&gt;");<a name="line.305"></a>
<span class="sourceLineNo">306</span> System.err.println(" -D" + TableInputFormat.SCAN_ROW_STOP + "=&lt;rowkey&gt;");<a name="line.306"></a>
<span class="sourceLineNo">307</span> System.err.println(" -D" + TableInputFormat.SCAN_COLUMNS + "=\"&lt;col1&gt; &lt;col2&gt;...\"");<a name="line.307"></a>
<span class="sourceLineNo">308</span> System.err.println(" -D" + TableInputFormat.SCAN_COLUMN_FAMILY + "=&lt;family1&gt;,&lt;family2&gt;, ...");<a name="line.308"></a>
<span class="sourceLineNo">309</span> System.err.println(" -D" + TableInputFormat.SCAN_TIMESTAMP + "=&lt;timestamp&gt;");<a name="line.309"></a>
<span class="sourceLineNo">310</span> System.err.println(" -D" + TableInputFormat.SCAN_TIMERANGE_START + "=&lt;timestamp&gt;");<a name="line.310"></a>
<span class="sourceLineNo">311</span> System.err.println(" -D" + TableInputFormat.SCAN_TIMERANGE_END + "=&lt;timestamp&gt;");<a name="line.311"></a>
<span class="sourceLineNo">312</span> System.err.println(" -D" + TableInputFormat.SCAN_MAXVERSIONS + "=&lt;count&gt;");<a name="line.312"></a>
<span class="sourceLineNo">313</span> System.err.println(" -D" + TableInputFormat.SCAN_CACHEDROWS + "=&lt;count&gt;");<a name="line.313"></a>
<span class="sourceLineNo">314</span> System.err.println(" -D" + TableInputFormat.SCAN_BATCHSIZE + "=&lt;count&gt;");<a name="line.314"></a>
<span class="sourceLineNo">315</span> System.err.println(" &lt;reportSeparator&gt; parameter can be used to override the default report "<a name="line.315"></a>
<span class="sourceLineNo">316</span> + "separator string : used to separate the rowId/column family name and qualifier name.");<a name="line.316"></a>
<span class="sourceLineNo">317</span> System.err.println(" [^[regex pattern] or [Prefix] parameter can be used to limit the cell "<a name="line.317"></a>
<span class="sourceLineNo">318</span> + "counter count operation to a limited subset of rows from the table based on regex or "<a name="line.318"></a>
<span class="sourceLineNo">319</span> + "prefix pattern.");<a name="line.319"></a>
<span class="sourceLineNo">320</span> }<a name="line.320"></a>
<span class="sourceLineNo">321</span><a name="line.321"></a>
<span class="sourceLineNo">322</span> /**<a name="line.322"></a>
<span class="sourceLineNo">323</span> * Main entry point.<a name="line.323"></a>
<span class="sourceLineNo">324</span> * @param args The command line parameters.<a name="line.324"></a>
<span class="sourceLineNo">325</span> * @throws Exception When running the job fails.<a name="line.325"></a>
<span class="sourceLineNo">326</span> */<a name="line.326"></a>
<span class="sourceLineNo">327</span> public static void main(String[] args) throws Exception {<a name="line.327"></a>
<span class="sourceLineNo">328</span> int errCode = ToolRunner.run(HBaseConfiguration.create(), new CellCounter(), args);<a name="line.328"></a>
<span class="sourceLineNo">329</span> System.exit(errCode);<a name="line.329"></a>
<span class="sourceLineNo">330</span> }<a name="line.330"></a>
<span class="sourceLineNo">331</span><a name="line.331"></a>
<span class="sourceLineNo">332</span>}<a name="line.332"></a>
</pre>
</div>
</body>
</html>