| <!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 java.text.MessageFormat;<a name="line.21"></a> |
| <span class="sourceLineNo">022</span>import java.util.ArrayList;<a name="line.22"></a> |
| <span class="sourceLineNo">023</span>import java.util.HashMap;<a name="line.23"></a> |
| <span class="sourceLineNo">024</span>import java.util.Iterator;<a name="line.24"></a> |
| <span class="sourceLineNo">025</span>import java.util.List;<a name="line.25"></a> |
| <span class="sourceLineNo">026</span>import java.util.Map;<a name="line.26"></a> |
| <span class="sourceLineNo">027</span><a name="line.27"></a> |
| <span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.HRegionLocation;<a name="line.28"></a> |
| <span class="sourceLineNo">029</span>import org.apache.hadoop.hbase.TableName;<a name="line.29"></a> |
| <span class="sourceLineNo">030</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.30"></a> |
| <span class="sourceLineNo">031</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.31"></a> |
| <span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.client.RegionInfo;<a name="line.32"></a> |
| <span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.client.RegionLocator;<a name="line.33"></a> |
| <span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.client.Result;<a name="line.34"></a> |
| <span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.client.Scan;<a name="line.35"></a> |
| <span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.client.Table;<a name="line.36"></a> |
| <span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.37"></a> |
| <span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.38"></a> |
| <span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.util.Pair;<a name="line.39"></a> |
| <span class="sourceLineNo">040</span>import org.apache.hadoop.mapreduce.InputFormat;<a name="line.40"></a> |
| <span class="sourceLineNo">041</span>import org.apache.hadoop.mapreduce.InputSplit;<a name="line.41"></a> |
| <span class="sourceLineNo">042</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.42"></a> |
| <span class="sourceLineNo">043</span>import org.apache.hadoop.mapreduce.RecordReader;<a name="line.43"></a> |
| <span class="sourceLineNo">044</span>import org.apache.hadoop.mapreduce.TaskAttemptContext;<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>/**<a name="line.49"></a> |
| <span class="sourceLineNo">050</span> * A base for {@link MultiTableInputFormat}s. Receives a list of<a name="line.50"></a> |
| <span class="sourceLineNo">051</span> * {@link Scan} instances that define the input tables and<a name="line.51"></a> |
| <span class="sourceLineNo">052</span> * filters etc. Subclasses may use other TableRecordReader implementations.<a name="line.52"></a> |
| <span class="sourceLineNo">053</span> */<a name="line.53"></a> |
| <span class="sourceLineNo">054</span>@InterfaceAudience.Public<a name="line.54"></a> |
| <span class="sourceLineNo">055</span>public abstract class MultiTableInputFormatBase extends<a name="line.55"></a> |
| <span class="sourceLineNo">056</span> InputFormat<ImmutableBytesWritable, Result> {<a name="line.56"></a> |
| <span class="sourceLineNo">057</span><a name="line.57"></a> |
| <span class="sourceLineNo">058</span> private static final Logger LOG = LoggerFactory.getLogger(MultiTableInputFormatBase.class);<a name="line.58"></a> |
| <span class="sourceLineNo">059</span><a name="line.59"></a> |
| <span class="sourceLineNo">060</span> /** Holds the set of scans used to define the input. */<a name="line.60"></a> |
| <span class="sourceLineNo">061</span> private List<Scan> scans;<a name="line.61"></a> |
| <span class="sourceLineNo">062</span><a name="line.62"></a> |
| <span class="sourceLineNo">063</span> /** The reader scanning the table, can be a custom one. */<a name="line.63"></a> |
| <span class="sourceLineNo">064</span> private TableRecordReader tableRecordReader = null;<a name="line.64"></a> |
| <span class="sourceLineNo">065</span><a name="line.65"></a> |
| <span class="sourceLineNo">066</span> /**<a name="line.66"></a> |
| <span class="sourceLineNo">067</span> * Builds a TableRecordReader. If no TableRecordReader was provided, uses the<a name="line.67"></a> |
| <span class="sourceLineNo">068</span> * default.<a name="line.68"></a> |
| <span class="sourceLineNo">069</span> *<a name="line.69"></a> |
| <span class="sourceLineNo">070</span> * @param split The split to work with.<a name="line.70"></a> |
| <span class="sourceLineNo">071</span> * @param context The current context.<a name="line.71"></a> |
| <span class="sourceLineNo">072</span> * @return The newly created record reader.<a name="line.72"></a> |
| <span class="sourceLineNo">073</span> * @throws IOException When creating the reader fails.<a name="line.73"></a> |
| <span class="sourceLineNo">074</span> * @throws InterruptedException when record reader initialization fails<a name="line.74"></a> |
| <span class="sourceLineNo">075</span> * @see org.apache.hadoop.mapreduce.InputFormat#createRecordReader(<a name="line.75"></a> |
| <span class="sourceLineNo">076</span> * org.apache.hadoop.mapreduce.InputSplit,<a name="line.76"></a> |
| <span class="sourceLineNo">077</span> * org.apache.hadoop.mapreduce.TaskAttemptContext)<a name="line.77"></a> |
| <span class="sourceLineNo">078</span> */<a name="line.78"></a> |
| <span class="sourceLineNo">079</span> @Override<a name="line.79"></a> |
| <span class="sourceLineNo">080</span> public RecordReader<ImmutableBytesWritable, Result> createRecordReader(<a name="line.80"></a> |
| <span class="sourceLineNo">081</span> InputSplit split, TaskAttemptContext context)<a name="line.81"></a> |
| <span class="sourceLineNo">082</span> throws IOException, InterruptedException {<a name="line.82"></a> |
| <span class="sourceLineNo">083</span> TableSplit tSplit = (TableSplit) split;<a name="line.83"></a> |
| <span class="sourceLineNo">084</span> LOG.info(MessageFormat.format("Input split length: {0} bytes.", tSplit.getLength()));<a name="line.84"></a> |
| <span class="sourceLineNo">085</span><a name="line.85"></a> |
| <span class="sourceLineNo">086</span> if (tSplit.getTable() == null) {<a name="line.86"></a> |
| <span class="sourceLineNo">087</span> throw new IOException("Cannot create a record reader because of a"<a name="line.87"></a> |
| <span class="sourceLineNo">088</span> + " previous error. Please look at the previous logs lines from"<a name="line.88"></a> |
| <span class="sourceLineNo">089</span> + " the task's full log for more details.");<a name="line.89"></a> |
| <span class="sourceLineNo">090</span> }<a name="line.90"></a> |
| <span class="sourceLineNo">091</span> final Connection connection = ConnectionFactory.createConnection(context.getConfiguration());<a name="line.91"></a> |
| <span class="sourceLineNo">092</span> Table table = connection.getTable(tSplit.getTable());<a name="line.92"></a> |
| <span class="sourceLineNo">093</span><a name="line.93"></a> |
| <span class="sourceLineNo">094</span> if (this.tableRecordReader == null) {<a name="line.94"></a> |
| <span class="sourceLineNo">095</span> this.tableRecordReader = new TableRecordReader();<a name="line.95"></a> |
| <span class="sourceLineNo">096</span> }<a name="line.96"></a> |
| <span class="sourceLineNo">097</span> final TableRecordReader trr = this.tableRecordReader;<a name="line.97"></a> |
| <span class="sourceLineNo">098</span><a name="line.98"></a> |
| <span class="sourceLineNo">099</span> try {<a name="line.99"></a> |
| <span class="sourceLineNo">100</span> Scan sc = tSplit.getScan();<a name="line.100"></a> |
| <span class="sourceLineNo">101</span> sc.setStartRow(tSplit.getStartRow());<a name="line.101"></a> |
| <span class="sourceLineNo">102</span> sc.setStopRow(tSplit.getEndRow());<a name="line.102"></a> |
| <span class="sourceLineNo">103</span> trr.setScan(sc);<a name="line.103"></a> |
| <span class="sourceLineNo">104</span> trr.setTable(table);<a name="line.104"></a> |
| <span class="sourceLineNo">105</span> return new RecordReader<ImmutableBytesWritable, Result>() {<a name="line.105"></a> |
| <span class="sourceLineNo">106</span><a name="line.106"></a> |
| <span class="sourceLineNo">107</span> @Override<a name="line.107"></a> |
| <span class="sourceLineNo">108</span> public void close() throws IOException {<a name="line.108"></a> |
| <span class="sourceLineNo">109</span> trr.close();<a name="line.109"></a> |
| <span class="sourceLineNo">110</span> connection.close();<a name="line.110"></a> |
| <span class="sourceLineNo">111</span> }<a name="line.111"></a> |
| <span class="sourceLineNo">112</span><a name="line.112"></a> |
| <span class="sourceLineNo">113</span> @Override<a name="line.113"></a> |
| <span class="sourceLineNo">114</span> public ImmutableBytesWritable getCurrentKey() throws IOException, InterruptedException {<a name="line.114"></a> |
| <span class="sourceLineNo">115</span> return trr.getCurrentKey();<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> @Override<a name="line.118"></a> |
| <span class="sourceLineNo">119</span> public Result getCurrentValue() throws IOException, InterruptedException {<a name="line.119"></a> |
| <span class="sourceLineNo">120</span> return trr.getCurrentValue();<a name="line.120"></a> |
| <span class="sourceLineNo">121</span> }<a name="line.121"></a> |
| <span class="sourceLineNo">122</span><a name="line.122"></a> |
| <span class="sourceLineNo">123</span> @Override<a name="line.123"></a> |
| <span class="sourceLineNo">124</span> public float getProgress() throws IOException, InterruptedException {<a name="line.124"></a> |
| <span class="sourceLineNo">125</span> return trr.getProgress();<a name="line.125"></a> |
| <span class="sourceLineNo">126</span> }<a name="line.126"></a> |
| <span class="sourceLineNo">127</span><a name="line.127"></a> |
| <span class="sourceLineNo">128</span> @Override<a name="line.128"></a> |
| <span class="sourceLineNo">129</span> public void initialize(InputSplit inputsplit, TaskAttemptContext context)<a name="line.129"></a> |
| <span class="sourceLineNo">130</span> throws IOException, InterruptedException {<a name="line.130"></a> |
| <span class="sourceLineNo">131</span> trr.initialize(inputsplit, context);<a name="line.131"></a> |
| <span class="sourceLineNo">132</span> }<a name="line.132"></a> |
| <span class="sourceLineNo">133</span><a name="line.133"></a> |
| <span class="sourceLineNo">134</span> @Override<a name="line.134"></a> |
| <span class="sourceLineNo">135</span> public boolean nextKeyValue() throws IOException, InterruptedException {<a name="line.135"></a> |
| <span class="sourceLineNo">136</span> return trr.nextKeyValue();<a name="line.136"></a> |
| <span class="sourceLineNo">137</span> }<a name="line.137"></a> |
| <span class="sourceLineNo">138</span> };<a name="line.138"></a> |
| <span class="sourceLineNo">139</span> } catch (IOException ioe) {<a name="line.139"></a> |
| <span class="sourceLineNo">140</span> // If there is an exception make sure that all<a name="line.140"></a> |
| <span class="sourceLineNo">141</span> // resources are closed and released.<a name="line.141"></a> |
| <span class="sourceLineNo">142</span> trr.close();<a name="line.142"></a> |
| <span class="sourceLineNo">143</span> connection.close();<a name="line.143"></a> |
| <span class="sourceLineNo">144</span> throw ioe;<a name="line.144"></a> |
| <span class="sourceLineNo">145</span> }<a name="line.145"></a> |
| <span class="sourceLineNo">146</span> }<a name="line.146"></a> |
| <span class="sourceLineNo">147</span><a name="line.147"></a> |
| <span class="sourceLineNo">148</span> /**<a name="line.148"></a> |
| <span class="sourceLineNo">149</span> * Calculates the splits that will serve as input for the map tasks. The<a name="line.149"></a> |
| <span class="sourceLineNo">150</span> * number of splits matches the number of regions in a table.<a name="line.150"></a> |
| <span class="sourceLineNo">151</span> *<a name="line.151"></a> |
| <span class="sourceLineNo">152</span> * @param context The current job context.<a name="line.152"></a> |
| <span class="sourceLineNo">153</span> * @return The list of input splits.<a name="line.153"></a> |
| <span class="sourceLineNo">154</span> * @throws IOException When creating the list of splits fails.<a name="line.154"></a> |
| <span class="sourceLineNo">155</span> * @see org.apache.hadoop.mapreduce.InputFormat#getSplits(org.apache.hadoop.mapreduce.JobContext)<a name="line.155"></a> |
| <span class="sourceLineNo">156</span> */<a name="line.156"></a> |
| <span class="sourceLineNo">157</span> @Override<a name="line.157"></a> |
| <span class="sourceLineNo">158</span> public List<InputSplit> getSplits(JobContext context) throws IOException {<a name="line.158"></a> |
| <span class="sourceLineNo">159</span> if (scans.isEmpty()) {<a name="line.159"></a> |
| <span class="sourceLineNo">160</span> throw new IOException("No scans were provided.");<a name="line.160"></a> |
| <span class="sourceLineNo">161</span> }<a name="line.161"></a> |
| <span class="sourceLineNo">162</span><a name="line.162"></a> |
| <span class="sourceLineNo">163</span> Map<TableName, List<Scan>> tableMaps = new HashMap<>();<a name="line.163"></a> |
| <span class="sourceLineNo">164</span> for (Scan scan : scans) {<a name="line.164"></a> |
| <span class="sourceLineNo">165</span> byte[] tableNameBytes = scan.getAttribute(Scan.SCAN_ATTRIBUTES_TABLE_NAME);<a name="line.165"></a> |
| <span class="sourceLineNo">166</span> if (tableNameBytes == null)<a name="line.166"></a> |
| <span class="sourceLineNo">167</span> throw new IOException("A scan object did not have a table name");<a name="line.167"></a> |
| <span class="sourceLineNo">168</span><a name="line.168"></a> |
| <span class="sourceLineNo">169</span> TableName tableName = TableName.valueOf(tableNameBytes);<a name="line.169"></a> |
| <span class="sourceLineNo">170</span><a name="line.170"></a> |
| <span class="sourceLineNo">171</span> List<Scan> scanList = tableMaps.get(tableName);<a name="line.171"></a> |
| <span class="sourceLineNo">172</span> if (scanList == null) {<a name="line.172"></a> |
| <span class="sourceLineNo">173</span> scanList = new ArrayList<>();<a name="line.173"></a> |
| <span class="sourceLineNo">174</span> tableMaps.put(tableName, scanList);<a name="line.174"></a> |
| <span class="sourceLineNo">175</span> }<a name="line.175"></a> |
| <span class="sourceLineNo">176</span> scanList.add(scan);<a name="line.176"></a> |
| <span class="sourceLineNo">177</span> }<a name="line.177"></a> |
| <span class="sourceLineNo">178</span><a name="line.178"></a> |
| <span class="sourceLineNo">179</span> List<InputSplit> splits = new ArrayList<>();<a name="line.179"></a> |
| <span class="sourceLineNo">180</span> Iterator iter = tableMaps.entrySet().iterator();<a name="line.180"></a> |
| <span class="sourceLineNo">181</span> // Make a single Connection to the Cluster and use it across all tables.<a name="line.181"></a> |
| <span class="sourceLineNo">182</span> try (Connection conn = ConnectionFactory.createConnection(context.getConfiguration())) {<a name="line.182"></a> |
| <span class="sourceLineNo">183</span> while (iter.hasNext()) {<a name="line.183"></a> |
| <span class="sourceLineNo">184</span> Map.Entry<TableName, List<Scan>> entry = (Map.Entry<TableName, List<Scan>>) iter.next();<a name="line.184"></a> |
| <span class="sourceLineNo">185</span> TableName tableName = entry.getKey();<a name="line.185"></a> |
| <span class="sourceLineNo">186</span> List<Scan> scanList = entry.getValue();<a name="line.186"></a> |
| <span class="sourceLineNo">187</span> try (Table table = conn.getTable(tableName);<a name="line.187"></a> |
| <span class="sourceLineNo">188</span> RegionLocator regionLocator = conn.getRegionLocator(tableName)) {<a name="line.188"></a> |
| <span class="sourceLineNo">189</span> RegionSizeCalculator sizeCalculator = new RegionSizeCalculator(<a name="line.189"></a> |
| <span class="sourceLineNo">190</span> regionLocator, conn.getAdmin());<a name="line.190"></a> |
| <span class="sourceLineNo">191</span> Pair<byte[][], byte[][]> keys = regionLocator.getStartEndKeys();<a name="line.191"></a> |
| <span class="sourceLineNo">192</span> for (Scan scan : scanList) {<a name="line.192"></a> |
| <span class="sourceLineNo">193</span> if (keys == null || keys.getFirst() == null || keys.getFirst().length == 0) {<a name="line.193"></a> |
| <span class="sourceLineNo">194</span> throw new IOException("Expecting at least one region for table : "<a name="line.194"></a> |
| <span class="sourceLineNo">195</span> + tableName.getNameAsString());<a name="line.195"></a> |
| <span class="sourceLineNo">196</span> }<a name="line.196"></a> |
| <span class="sourceLineNo">197</span> int count = 0;<a name="line.197"></a> |
| <span class="sourceLineNo">198</span><a name="line.198"></a> |
| <span class="sourceLineNo">199</span> byte[] startRow = scan.getStartRow();<a name="line.199"></a> |
| <span class="sourceLineNo">200</span> byte[] stopRow = scan.getStopRow();<a name="line.200"></a> |
| <span class="sourceLineNo">201</span><a name="line.201"></a> |
| <span class="sourceLineNo">202</span> for (int i = 0; i < keys.getFirst().length; i++) {<a name="line.202"></a> |
| <span class="sourceLineNo">203</span> if (!includeRegionInSplit(keys.getFirst()[i], keys.getSecond()[i])) {<a name="line.203"></a> |
| <span class="sourceLineNo">204</span> continue;<a name="line.204"></a> |
| <span class="sourceLineNo">205</span> }<a name="line.205"></a> |
| <span class="sourceLineNo">206</span><a name="line.206"></a> |
| <span class="sourceLineNo">207</span> if ((startRow.length == 0 || keys.getSecond()[i].length == 0 ||<a name="line.207"></a> |
| <span class="sourceLineNo">208</span> Bytes.compareTo(startRow, keys.getSecond()[i]) < 0) &&<a name="line.208"></a> |
| <span class="sourceLineNo">209</span> (stopRow.length == 0 || Bytes.compareTo(stopRow,<a name="line.209"></a> |
| <span class="sourceLineNo">210</span> keys.getFirst()[i]) > 0)) {<a name="line.210"></a> |
| <span class="sourceLineNo">211</span> byte[] splitStart = startRow.length == 0 ||<a name="line.211"></a> |
| <span class="sourceLineNo">212</span> Bytes.compareTo(keys.getFirst()[i], startRow) >= 0 ?<a name="line.212"></a> |
| <span class="sourceLineNo">213</span> keys.getFirst()[i] : startRow;<a name="line.213"></a> |
| <span class="sourceLineNo">214</span> byte[] splitStop = (stopRow.length == 0 ||<a name="line.214"></a> |
| <span class="sourceLineNo">215</span> Bytes.compareTo(keys.getSecond()[i], stopRow) <= 0) &&<a name="line.215"></a> |
| <span class="sourceLineNo">216</span> keys.getSecond()[i].length > 0 ?<a name="line.216"></a> |
| <span class="sourceLineNo">217</span> keys.getSecond()[i] : stopRow;<a name="line.217"></a> |
| <span class="sourceLineNo">218</span><a name="line.218"></a> |
| <span class="sourceLineNo">219</span> HRegionLocation hregionLocation = regionLocator.getRegionLocation(<a name="line.219"></a> |
| <span class="sourceLineNo">220</span> keys.getFirst()[i], false);<a name="line.220"></a> |
| <span class="sourceLineNo">221</span> String regionHostname = hregionLocation.getHostname();<a name="line.221"></a> |
| <span class="sourceLineNo">222</span> RegionInfo regionInfo = hregionLocation.getRegion();<a name="line.222"></a> |
| <span class="sourceLineNo">223</span> String encodedRegionName = regionInfo.getEncodedName();<a name="line.223"></a> |
| <span class="sourceLineNo">224</span> long regionSize = sizeCalculator.getRegionSize(<a name="line.224"></a> |
| <span class="sourceLineNo">225</span> regionInfo.getRegionName());<a name="line.225"></a> |
| <span class="sourceLineNo">226</span><a name="line.226"></a> |
| <span class="sourceLineNo">227</span> TableSplit split = new TableSplit(table.getName(),<a name="line.227"></a> |
| <span class="sourceLineNo">228</span> scan, splitStart, splitStop, regionHostname,<a name="line.228"></a> |
| <span class="sourceLineNo">229</span> encodedRegionName, regionSize);<a name="line.229"></a> |
| <span class="sourceLineNo">230</span><a name="line.230"></a> |
| <span class="sourceLineNo">231</span> splits.add(split);<a name="line.231"></a> |
| <span class="sourceLineNo">232</span><a name="line.232"></a> |
| <span class="sourceLineNo">233</span> if (LOG.isDebugEnabled()) {<a name="line.233"></a> |
| <span class="sourceLineNo">234</span> LOG.debug("getSplits: split -> " + (count++) + " -> " + split);<a name="line.234"></a> |
| <span class="sourceLineNo">235</span> }<a name="line.235"></a> |
| <span class="sourceLineNo">236</span> }<a name="line.236"></a> |
| <span class="sourceLineNo">237</span> }<a name="line.237"></a> |
| <span class="sourceLineNo">238</span> }<a name="line.238"></a> |
| <span class="sourceLineNo">239</span> }<a name="line.239"></a> |
| <span class="sourceLineNo">240</span> }<a name="line.240"></a> |
| <span class="sourceLineNo">241</span> }<a name="line.241"></a> |
| <span class="sourceLineNo">242</span><a name="line.242"></a> |
| <span class="sourceLineNo">243</span> return splits;<a name="line.243"></a> |
| <span class="sourceLineNo">244</span> }<a name="line.244"></a> |
| <span class="sourceLineNo">245</span><a name="line.245"></a> |
| <span class="sourceLineNo">246</span> /**<a name="line.246"></a> |
| <span class="sourceLineNo">247</span> * Test if the given region is to be included in the InputSplit while<a name="line.247"></a> |
| <span class="sourceLineNo">248</span> * splitting the regions of a table.<a name="line.248"></a> |
| <span class="sourceLineNo">249</span> * <p><a name="line.249"></a> |
| <span class="sourceLineNo">250</span> * This optimization is effective when there is a specific reasoning to<a name="line.250"></a> |
| <span class="sourceLineNo">251</span> * exclude an entire region from the M-R job, (and hence, not contributing to<a name="line.251"></a> |
| <span class="sourceLineNo">252</span> * the InputSplit), given the start and end keys of the same. <br><a name="line.252"></a> |
| <span class="sourceLineNo">253</span> * Useful when we need to remember the last-processed top record and revisit<a name="line.253"></a> |
| <span class="sourceLineNo">254</span> * the [last, current) interval for M-R processing, continuously. In addition<a name="line.254"></a> |
| <span class="sourceLineNo">255</span> * to reducing InputSplits, reduces the load on the region server as well, due<a name="line.255"></a> |
| <span class="sourceLineNo">256</span> * to the ordering of the keys. <br><a name="line.256"></a> |
| <span class="sourceLineNo">257</span> * <br><a name="line.257"></a> |
| <span class="sourceLineNo">258</span> * Note: It is possible that <code>endKey.length() == 0 </code> , for the last<a name="line.258"></a> |
| <span class="sourceLineNo">259</span> * (recent) region. <br><a name="line.259"></a> |
| <span class="sourceLineNo">260</span> * Override this method, if you want to bulk exclude regions altogether from<a name="line.260"></a> |
| <span class="sourceLineNo">261</span> * M-R. By default, no region is excluded( i.e. all regions are included).<a name="line.261"></a> |
| <span class="sourceLineNo">262</span> *<a name="line.262"></a> |
| <span class="sourceLineNo">263</span> * @param startKey Start key of the region<a name="line.263"></a> |
| <span class="sourceLineNo">264</span> * @param endKey End key of the region<a name="line.264"></a> |
| <span class="sourceLineNo">265</span> * @return true, if this region needs to be included as part of the input<a name="line.265"></a> |
| <span class="sourceLineNo">266</span> * (default).<a name="line.266"></a> |
| <span class="sourceLineNo">267</span> */<a name="line.267"></a> |
| <span class="sourceLineNo">268</span> protected boolean includeRegionInSplit(final byte[] startKey,<a name="line.268"></a> |
| <span class="sourceLineNo">269</span> final byte[] endKey) {<a name="line.269"></a> |
| <span class="sourceLineNo">270</span> return true;<a name="line.270"></a> |
| <span class="sourceLineNo">271</span> }<a name="line.271"></a> |
| <span class="sourceLineNo">272</span><a name="line.272"></a> |
| <span class="sourceLineNo">273</span> /**<a name="line.273"></a> |
| <span class="sourceLineNo">274</span> * Allows subclasses to get the list of {@link Scan} objects.<a name="line.274"></a> |
| <span class="sourceLineNo">275</span> */<a name="line.275"></a> |
| <span class="sourceLineNo">276</span> protected List<Scan> getScans() {<a name="line.276"></a> |
| <span class="sourceLineNo">277</span> return this.scans;<a name="line.277"></a> |
| <span class="sourceLineNo">278</span> }<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> * Allows subclasses to set the list of {@link Scan} objects.<a name="line.281"></a> |
| <span class="sourceLineNo">282</span> *<a name="line.282"></a> |
| <span class="sourceLineNo">283</span> * @param scans The list of {@link Scan} used to define the input<a name="line.283"></a> |
| <span class="sourceLineNo">284</span> */<a name="line.284"></a> |
| <span class="sourceLineNo">285</span> protected void setScans(List<Scan> scans) {<a name="line.285"></a> |
| <span class="sourceLineNo">286</span> this.scans = scans;<a name="line.286"></a> |
| <span class="sourceLineNo">287</span> }<a name="line.287"></a> |
| <span class="sourceLineNo">288</span><a name="line.288"></a> |
| <span class="sourceLineNo">289</span> /**<a name="line.289"></a> |
| <span class="sourceLineNo">290</span> * Allows subclasses to set the {@link TableRecordReader}.<a name="line.290"></a> |
| <span class="sourceLineNo">291</span> *<a name="line.291"></a> |
| <span class="sourceLineNo">292</span> * @param tableRecordReader A different {@link TableRecordReader}<a name="line.292"></a> |
| <span class="sourceLineNo">293</span> * implementation.<a name="line.293"></a> |
| <span class="sourceLineNo">294</span> */<a name="line.294"></a> |
| <span class="sourceLineNo">295</span> protected void setTableRecordReader(TableRecordReader tableRecordReader) {<a name="line.295"></a> |
| <span class="sourceLineNo">296</span> this.tableRecordReader = tableRecordReader;<a name="line.296"></a> |
| <span class="sourceLineNo">297</span> }<a name="line.297"></a> |
| <span class="sourceLineNo">298</span>}<a name="line.298"></a> |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| </pre> |
| </div> |
| </body> |
| </html> |
