blob: f65f77cbb4938b2229032bcd1c5efafa9543d190 [file] [log] [blame]
<!DOCTYPE HTML>
<html lang="en">
<head>
<!-- Generated by javadoc (17) -->
<title>Source code</title>
<meta name="viewport" content="width=device-width, initial-scale=1">
<meta name="description" content="source: package: org.apache.hadoop.hbase.mapreduce, class: HFileInputFormat">
<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 java.util.OptionalLong;</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.fs.FileStatus;</span>
<span class="source-line-no">026</span><span id="line-26">import org.apache.hadoop.fs.FileSystem;</span>
<span class="source-line-no">027</span><span id="line-27">import org.apache.hadoop.fs.Path;</span>
<span class="source-line-no">028</span><span id="line-28">import org.apache.hadoop.fs.PathFilter;</span>
<span class="source-line-no">029</span><span id="line-29">import org.apache.hadoop.hbase.Cell;</span>
<span class="source-line-no">030</span><span id="line-30">import org.apache.hadoop.hbase.PrivateCellUtil;</span>
<span class="source-line-no">031</span><span id="line-31">import org.apache.hadoop.hbase.io.hfile.HFile;</span>
<span class="source-line-no">032</span><span id="line-32">import org.apache.hadoop.hbase.io.hfile.HFile.Reader;</span>
<span class="source-line-no">033</span><span id="line-33">import org.apache.hadoop.hbase.io.hfile.HFileScanner;</span>
<span class="source-line-no">034</span><span id="line-34">import org.apache.hadoop.hbase.regionserver.StoreFileInfo;</span>
<span class="source-line-no">035</span><span id="line-35">import org.apache.hadoop.io.NullWritable;</span>
<span class="source-line-no">036</span><span id="line-36">import org.apache.hadoop.mapreduce.InputSplit;</span>
<span class="source-line-no">037</span><span id="line-37">import org.apache.hadoop.mapreduce.JobContext;</span>
<span class="source-line-no">038</span><span id="line-38">import org.apache.hadoop.mapreduce.RecordReader;</span>
<span class="source-line-no">039</span><span id="line-39">import org.apache.hadoop.mapreduce.TaskAttemptContext;</span>
<span class="source-line-no">040</span><span id="line-40">import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;</span>
<span class="source-line-no">041</span><span id="line-41">import org.apache.hadoop.mapreduce.lib.input.FileSplit;</span>
<span class="source-line-no">042</span><span id="line-42">import org.apache.yetus.audience.InterfaceAudience;</span>
<span class="source-line-no">043</span><span id="line-43">import org.slf4j.Logger;</span>
<span class="source-line-no">044</span><span id="line-44">import org.slf4j.LoggerFactory;</span>
<span class="source-line-no">045</span><span id="line-45"></span>
<span class="source-line-no">046</span><span id="line-46">/**</span>
<span class="source-line-no">047</span><span id="line-47"> * Simple MR input format for HFiles. This code was borrowed from Apache Crunch project. Updated to</span>
<span class="source-line-no">048</span><span id="line-48"> * the recent version of HBase.</span>
<span class="source-line-no">049</span><span id="line-49"> */</span>
<span class="source-line-no">050</span><span id="line-50">@InterfaceAudience.Private</span>
<span class="source-line-no">051</span><span id="line-51">public class HFileInputFormat extends FileInputFormat&lt;NullWritable, Cell&gt; {</span>
<span class="source-line-no">052</span><span id="line-52"></span>
<span class="source-line-no">053</span><span id="line-53"> private static final Logger LOG = LoggerFactory.getLogger(HFileInputFormat.class);</span>
<span class="source-line-no">054</span><span id="line-54"></span>
<span class="source-line-no">055</span><span id="line-55"> /**</span>
<span class="source-line-no">056</span><span id="line-56"> * File filter that removes all "hidden" files. This might be something worth removing from a more</span>
<span class="source-line-no">057</span><span id="line-57"> * general purpose utility; it accounts for the presence of metadata files created in the way</span>
<span class="source-line-no">058</span><span id="line-58"> * we're doing exports.</span>
<span class="source-line-no">059</span><span id="line-59"> */</span>
<span class="source-line-no">060</span><span id="line-60"> static final PathFilter HIDDEN_FILE_FILTER = new PathFilter() {</span>
<span class="source-line-no">061</span><span id="line-61"> @Override</span>
<span class="source-line-no">062</span><span id="line-62"> public boolean accept(Path p) {</span>
<span class="source-line-no">063</span><span id="line-63"> String name = p.getName();</span>
<span class="source-line-no">064</span><span id="line-64"> return !name.startsWith("_") &amp;&amp; !name.startsWith(".");</span>
<span class="source-line-no">065</span><span id="line-65"> }</span>
<span class="source-line-no">066</span><span id="line-66"> };</span>
<span class="source-line-no">067</span><span id="line-67"></span>
<span class="source-line-no">068</span><span id="line-68"> /**</span>
<span class="source-line-no">069</span><span id="line-69"> * Record reader for HFiles.</span>
<span class="source-line-no">070</span><span id="line-70"> */</span>
<span class="source-line-no">071</span><span id="line-71"> private static class HFileRecordReader extends RecordReader&lt;NullWritable, Cell&gt; {</span>
<span class="source-line-no">072</span><span id="line-72"></span>
<span class="source-line-no">073</span><span id="line-73"> private Reader in;</span>
<span class="source-line-no">074</span><span id="line-74"> protected Configuration conf;</span>
<span class="source-line-no">075</span><span id="line-75"> private HFileScanner scanner;</span>
<span class="source-line-no">076</span><span id="line-76"></span>
<span class="source-line-no">077</span><span id="line-77"> /**</span>
<span class="source-line-no">078</span><span id="line-78"> * A private cache of the key value so it doesn't need to be loaded twice from the scanner.</span>
<span class="source-line-no">079</span><span id="line-79"> */</span>
<span class="source-line-no">080</span><span id="line-80"> private Cell value = null;</span>
<span class="source-line-no">081</span><span id="line-81"> private long count;</span>
<span class="source-line-no">082</span><span id="line-82"> private boolean seeked = false;</span>
<span class="source-line-no">083</span><span id="line-83"> private OptionalLong bulkloadSeqId;</span>
<span class="source-line-no">084</span><span id="line-84"></span>
<span class="source-line-no">085</span><span id="line-85"> @Override</span>
<span class="source-line-no">086</span><span id="line-86"> public void initialize(InputSplit split, TaskAttemptContext context)</span>
<span class="source-line-no">087</span><span id="line-87"> throws IOException, InterruptedException {</span>
<span class="source-line-no">088</span><span id="line-88"> FileSplit fileSplit = (FileSplit) split;</span>
<span class="source-line-no">089</span><span id="line-89"> conf = context.getConfiguration();</span>
<span class="source-line-no">090</span><span id="line-90"> Path path = fileSplit.getPath();</span>
<span class="source-line-no">091</span><span id="line-91"> FileSystem fs = path.getFileSystem(conf);</span>
<span class="source-line-no">092</span><span id="line-92"> LOG.info("Initialize HFileRecordReader for {}", path);</span>
<span class="source-line-no">093</span><span id="line-93"> this.in = HFile.createReader(fs, path, conf);</span>
<span class="source-line-no">094</span><span id="line-94"> this.bulkloadSeqId = StoreFileInfo.getBulkloadSeqId(path);</span>
<span class="source-line-no">095</span><span id="line-95"></span>
<span class="source-line-no">096</span><span id="line-96"> // The file info must be loaded before the scanner can be used.</span>
<span class="source-line-no">097</span><span id="line-97"> // This seems like a bug in HBase, but it's easily worked around.</span>
<span class="source-line-no">098</span><span id="line-98"> this.scanner = in.getScanner(conf, false, false);</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"> @Override</span>
<span class="source-line-no">103</span><span id="line-103"> public boolean nextKeyValue() throws IOException, InterruptedException {</span>
<span class="source-line-no">104</span><span id="line-104"> boolean hasNext;</span>
<span class="source-line-no">105</span><span id="line-105"> if (!seeked) {</span>
<span class="source-line-no">106</span><span id="line-106"> LOG.info("Seeking to start");</span>
<span class="source-line-no">107</span><span id="line-107"> hasNext = scanner.seekTo();</span>
<span class="source-line-no">108</span><span id="line-108"> seeked = true;</span>
<span class="source-line-no">109</span><span id="line-109"> } else {</span>
<span class="source-line-no">110</span><span id="line-110"> hasNext = scanner.next();</span>
<span class="source-line-no">111</span><span id="line-111"> }</span>
<span class="source-line-no">112</span><span id="line-112"> if (!hasNext) {</span>
<span class="source-line-no">113</span><span id="line-113"> return false;</span>
<span class="source-line-no">114</span><span id="line-114"> }</span>
<span class="source-line-no">115</span><span id="line-115"> value = scanner.getCell();</span>
<span class="source-line-no">116</span><span id="line-116"> if (value != null &amp;&amp; bulkloadSeqId.isPresent()) {</span>
<span class="source-line-no">117</span><span id="line-117"> PrivateCellUtil.setSequenceId(value, bulkloadSeqId.getAsLong());</span>
<span class="source-line-no">118</span><span id="line-118"> }</span>
<span class="source-line-no">119</span><span id="line-119"> count++;</span>
<span class="source-line-no">120</span><span id="line-120"> return true;</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"> @Override</span>
<span class="source-line-no">124</span><span id="line-124"> public NullWritable getCurrentKey() throws IOException, InterruptedException {</span>
<span class="source-line-no">125</span><span id="line-125"> return NullWritable.get();</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"> @Override</span>
<span class="source-line-no">129</span><span id="line-129"> public Cell getCurrentValue() throws IOException, InterruptedException {</span>
<span class="source-line-no">130</span><span id="line-130"> return value;</span>
<span class="source-line-no">131</span><span id="line-131"> }</span>
<span class="source-line-no">132</span><span id="line-132"></span>
<span class="source-line-no">133</span><span id="line-133"> @Override</span>
<span class="source-line-no">134</span><span id="line-134"> public float getProgress() throws IOException, InterruptedException {</span>
<span class="source-line-no">135</span><span id="line-135"> // This would be inaccurate if KVs are not uniformly-sized or we have performed a seek to</span>
<span class="source-line-no">136</span><span id="line-136"> // the start row, but better than nothing anyway.</span>
<span class="source-line-no">137</span><span id="line-137"> return 1.0f * count / in.getEntries();</span>
<span class="source-line-no">138</span><span id="line-138"> }</span>
<span class="source-line-no">139</span><span id="line-139"></span>
<span class="source-line-no">140</span><span id="line-140"> @Override</span>
<span class="source-line-no">141</span><span id="line-141"> public void close() throws IOException {</span>
<span class="source-line-no">142</span><span id="line-142"> if (in != null) {</span>
<span class="source-line-no">143</span><span id="line-143"> in.close();</span>
<span class="source-line-no">144</span><span id="line-144"> in = null;</span>
<span class="source-line-no">145</span><span id="line-145"> }</span>
<span class="source-line-no">146</span><span id="line-146"> }</span>
<span class="source-line-no">147</span><span id="line-147"> }</span>
<span class="source-line-no">148</span><span id="line-148"></span>
<span class="source-line-no">149</span><span id="line-149"> @Override</span>
<span class="source-line-no">150</span><span id="line-150"> protected List&lt;FileStatus&gt; listStatus(JobContext job) throws IOException {</span>
<span class="source-line-no">151</span><span id="line-151"> List&lt;FileStatus&gt; result = new ArrayList&lt;FileStatus&gt;();</span>
<span class="source-line-no">152</span><span id="line-152"></span>
<span class="source-line-no">153</span><span id="line-153"> // Explode out directories that match the original FileInputFormat filters</span>
<span class="source-line-no">154</span><span id="line-154"> // Typically these are &lt;regionname&gt;-level dirs, only requiring 1 level of recursion to</span>
<span class="source-line-no">155</span><span id="line-155"> // get the &lt;columnname&gt;-level dirs where the HFile are written, but in some cases</span>
<span class="source-line-no">156</span><span id="line-156"> // &lt;tablename&gt;-level dirs are provided requiring 2 levels of recursion.</span>
<span class="source-line-no">157</span><span id="line-157"> for (FileStatus status : super.listStatus(job)) {</span>
<span class="source-line-no">158</span><span id="line-158"> addFilesRecursively(job, status, result);</span>
<span class="source-line-no">159</span><span id="line-159"> }</span>
<span class="source-line-no">160</span><span id="line-160"> return result;</span>
<span class="source-line-no">161</span><span id="line-161"> }</span>
<span class="source-line-no">162</span><span id="line-162"></span>
<span class="source-line-no">163</span><span id="line-163"> private static void addFilesRecursively(JobContext job, FileStatus status,</span>
<span class="source-line-no">164</span><span id="line-164"> List&lt;FileStatus&gt; result) throws IOException {</span>
<span class="source-line-no">165</span><span id="line-165"> if (status.isDirectory()) {</span>
<span class="source-line-no">166</span><span id="line-166"> FileSystem fs = status.getPath().getFileSystem(job.getConfiguration());</span>
<span class="source-line-no">167</span><span id="line-167"> for (FileStatus fileStatus : fs.listStatus(status.getPath(), HIDDEN_FILE_FILTER)) {</span>
<span class="source-line-no">168</span><span id="line-168"> addFilesRecursively(job, fileStatus, result);</span>
<span class="source-line-no">169</span><span id="line-169"> }</span>
<span class="source-line-no">170</span><span id="line-170"> } else {</span>
<span class="source-line-no">171</span><span id="line-171"> result.add(status);</span>
<span class="source-line-no">172</span><span id="line-172"> }</span>
<span class="source-line-no">173</span><span id="line-173"> }</span>
<span class="source-line-no">174</span><span id="line-174"></span>
<span class="source-line-no">175</span><span id="line-175"> @Override</span>
<span class="source-line-no">176</span><span id="line-176"> public RecordReader&lt;NullWritable, Cell&gt; createRecordReader(InputSplit split,</span>
<span class="source-line-no">177</span><span id="line-177"> TaskAttemptContext context) throws IOException, InterruptedException {</span>
<span class="source-line-no">178</span><span id="line-178"> return new HFileRecordReader();</span>
<span class="source-line-no">179</span><span id="line-179"> }</span>
<span class="source-line-no">180</span><span id="line-180"></span>
<span class="source-line-no">181</span><span id="line-181"> @Override</span>
<span class="source-line-no">182</span><span id="line-182"> protected boolean isSplitable(JobContext context, Path filename) {</span>
<span class="source-line-no">183</span><span id="line-183"> // This file isn't splittable.</span>
<span class="source-line-no">184</span><span id="line-184"> return false;</span>
<span class="source-line-no">185</span><span id="line-185"> }</span>
<span class="source-line-no">186</span><span id="line-186">}</span>
</pre>
</div>
</main>
</body>
</html>