| <!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.util.ArrayList;<a name="line.21"></a> |
| <span class="sourceLineNo">022</span>import java.util.List;<a name="line.22"></a> |
| <span class="sourceLineNo">023</span>import org.apache.hadoop.conf.Configuration;<a name="line.23"></a> |
| <span class="sourceLineNo">024</span>import org.apache.hadoop.fs.FileStatus;<a name="line.24"></a> |
| <span class="sourceLineNo">025</span>import org.apache.hadoop.fs.FileSystem;<a name="line.25"></a> |
| <span class="sourceLineNo">026</span>import org.apache.hadoop.fs.Path;<a name="line.26"></a> |
| <span class="sourceLineNo">027</span>import org.apache.hadoop.fs.PathFilter;<a name="line.27"></a> |
| <span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.Cell;<a name="line.28"></a> |
| <span class="sourceLineNo">029</span>import org.apache.hadoop.hbase.io.hfile.HFile;<a name="line.29"></a> |
| <span class="sourceLineNo">030</span>import org.apache.hadoop.hbase.io.hfile.HFile.Reader;<a name="line.30"></a> |
| <span class="sourceLineNo">031</span>import org.apache.hadoop.hbase.io.hfile.HFileScanner;<a name="line.31"></a> |
| <span class="sourceLineNo">032</span>import org.apache.hadoop.io.NullWritable;<a name="line.32"></a> |
| <span class="sourceLineNo">033</span>import org.apache.hadoop.mapreduce.InputSplit;<a name="line.33"></a> |
| <span class="sourceLineNo">034</span>import org.apache.hadoop.mapreduce.JobContext;<a name="line.34"></a> |
| <span class="sourceLineNo">035</span>import org.apache.hadoop.mapreduce.RecordReader;<a name="line.35"></a> |
| <span class="sourceLineNo">036</span>import org.apache.hadoop.mapreduce.TaskAttemptContext;<a name="line.36"></a> |
| <span class="sourceLineNo">037</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.37"></a> |
| <span class="sourceLineNo">038</span>import org.apache.hadoop.mapreduce.lib.input.FileSplit;<a name="line.38"></a> |
| <span class="sourceLineNo">039</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.39"></a> |
| <span class="sourceLineNo">040</span>import org.slf4j.Logger;<a name="line.40"></a> |
| <span class="sourceLineNo">041</span>import org.slf4j.LoggerFactory;<a name="line.41"></a> |
| <span class="sourceLineNo">042</span><a name="line.42"></a> |
| <span class="sourceLineNo">043</span>/**<a name="line.43"></a> |
| <span class="sourceLineNo">044</span> * Simple MR input format for HFiles.<a name="line.44"></a> |
| <span class="sourceLineNo">045</span> * This code was borrowed from Apache Crunch project.<a name="line.45"></a> |
| <span class="sourceLineNo">046</span> * Updated to the recent version of HBase.<a name="line.46"></a> |
| <span class="sourceLineNo">047</span> */<a name="line.47"></a> |
| <span class="sourceLineNo">048</span>@InterfaceAudience.Private<a name="line.48"></a> |
| <span class="sourceLineNo">049</span>public class HFileInputFormat extends FileInputFormat<NullWritable, Cell> {<a name="line.49"></a> |
| <span class="sourceLineNo">050</span><a name="line.50"></a> |
| <span class="sourceLineNo">051</span> private static final Logger LOG = LoggerFactory.getLogger(HFileInputFormat.class);<a name="line.51"></a> |
| <span class="sourceLineNo">052</span><a name="line.52"></a> |
| <span class="sourceLineNo">053</span> /**<a name="line.53"></a> |
| <span class="sourceLineNo">054</span> * File filter that removes all "hidden" files. This might be something worth removing from<a name="line.54"></a> |
| <span class="sourceLineNo">055</span> * a more general purpose utility; it accounts for the presence of metadata files created<a name="line.55"></a> |
| <span class="sourceLineNo">056</span> * in the way we're doing exports.<a name="line.56"></a> |
| <span class="sourceLineNo">057</span> */<a name="line.57"></a> |
| <span class="sourceLineNo">058</span> static final PathFilter HIDDEN_FILE_FILTER = new PathFilter() {<a name="line.58"></a> |
| <span class="sourceLineNo">059</span> @Override<a name="line.59"></a> |
| <span class="sourceLineNo">060</span> public boolean accept(Path p) {<a name="line.60"></a> |
| <span class="sourceLineNo">061</span> String name = p.getName();<a name="line.61"></a> |
| <span class="sourceLineNo">062</span> return !name.startsWith("_") && !name.startsWith(".");<a name="line.62"></a> |
| <span class="sourceLineNo">063</span> }<a name="line.63"></a> |
| <span class="sourceLineNo">064</span> };<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> * Record reader for HFiles.<a name="line.67"></a> |
| <span class="sourceLineNo">068</span> */<a name="line.68"></a> |
| <span class="sourceLineNo">069</span> private static class HFileRecordReader extends RecordReader<NullWritable, Cell> {<a name="line.69"></a> |
| <span class="sourceLineNo">070</span><a name="line.70"></a> |
| <span class="sourceLineNo">071</span> private Reader in;<a name="line.71"></a> |
| <span class="sourceLineNo">072</span> protected Configuration conf;<a name="line.72"></a> |
| <span class="sourceLineNo">073</span> private HFileScanner scanner;<a name="line.73"></a> |
| <span class="sourceLineNo">074</span><a name="line.74"></a> |
| <span class="sourceLineNo">075</span> /**<a name="line.75"></a> |
| <span class="sourceLineNo">076</span> * A private cache of the key value so it doesn't need to be loaded twice from the scanner.<a name="line.76"></a> |
| <span class="sourceLineNo">077</span> */<a name="line.77"></a> |
| <span class="sourceLineNo">078</span> private Cell value = null;<a name="line.78"></a> |
| <span class="sourceLineNo">079</span> private long count;<a name="line.79"></a> |
| <span class="sourceLineNo">080</span> private boolean seeked = false;<a name="line.80"></a> |
| <span class="sourceLineNo">081</span><a name="line.81"></a> |
| <span class="sourceLineNo">082</span> @Override<a name="line.82"></a> |
| <span class="sourceLineNo">083</span> public void initialize(InputSplit split, TaskAttemptContext context)<a name="line.83"></a> |
| <span class="sourceLineNo">084</span> throws IOException, InterruptedException {<a name="line.84"></a> |
| <span class="sourceLineNo">085</span> FileSplit fileSplit = (FileSplit) split;<a name="line.85"></a> |
| <span class="sourceLineNo">086</span> conf = context.getConfiguration();<a name="line.86"></a> |
| <span class="sourceLineNo">087</span> Path path = fileSplit.getPath();<a name="line.87"></a> |
| <span class="sourceLineNo">088</span> FileSystem fs = path.getFileSystem(conf);<a name="line.88"></a> |
| <span class="sourceLineNo">089</span> LOG.info("Initialize HFileRecordReader for {}", path);<a name="line.89"></a> |
| <span class="sourceLineNo">090</span> this.in = HFile.createReader(fs, path, conf);<a name="line.90"></a> |
| <span class="sourceLineNo">091</span><a name="line.91"></a> |
| <span class="sourceLineNo">092</span> // The file info must be loaded before the scanner can be used.<a name="line.92"></a> |
| <span class="sourceLineNo">093</span> // This seems like a bug in HBase, but it's easily worked around.<a name="line.93"></a> |
| <span class="sourceLineNo">094</span> this.scanner = in.getScanner(false, false);<a name="line.94"></a> |
| <span class="sourceLineNo">095</span><a name="line.95"></a> |
| <span class="sourceLineNo">096</span> }<a name="line.96"></a> |
| <span class="sourceLineNo">097</span><a name="line.97"></a> |
| <span class="sourceLineNo">098</span><a name="line.98"></a> |
| <span class="sourceLineNo">099</span> @Override<a name="line.99"></a> |
| <span class="sourceLineNo">100</span> public boolean nextKeyValue() throws IOException, InterruptedException {<a name="line.100"></a> |
| <span class="sourceLineNo">101</span> boolean hasNext;<a name="line.101"></a> |
| <span class="sourceLineNo">102</span> if (!seeked) {<a name="line.102"></a> |
| <span class="sourceLineNo">103</span> LOG.info("Seeking to start");<a name="line.103"></a> |
| <span class="sourceLineNo">104</span> hasNext = scanner.seekTo();<a name="line.104"></a> |
| <span class="sourceLineNo">105</span> seeked = true;<a name="line.105"></a> |
| <span class="sourceLineNo">106</span> } else {<a name="line.106"></a> |
| <span class="sourceLineNo">107</span> hasNext = scanner.next();<a name="line.107"></a> |
| <span class="sourceLineNo">108</span> }<a name="line.108"></a> |
| <span class="sourceLineNo">109</span> if (!hasNext) {<a name="line.109"></a> |
| <span class="sourceLineNo">110</span> return false;<a name="line.110"></a> |
| <span class="sourceLineNo">111</span> }<a name="line.111"></a> |
| <span class="sourceLineNo">112</span> value = scanner.getCell();<a name="line.112"></a> |
| <span class="sourceLineNo">113</span> count++;<a name="line.113"></a> |
| <span class="sourceLineNo">114</span> return true;<a name="line.114"></a> |
| <span class="sourceLineNo">115</span> }<a name="line.115"></a> |
| <span class="sourceLineNo">116</span><a name="line.116"></a> |
| <span class="sourceLineNo">117</span> @Override<a name="line.117"></a> |
| <span class="sourceLineNo">118</span> public NullWritable getCurrentKey() throws IOException, InterruptedException {<a name="line.118"></a> |
| <span class="sourceLineNo">119</span> return NullWritable.get();<a name="line.119"></a> |
| <span class="sourceLineNo">120</span> }<a name="line.120"></a> |
| <span class="sourceLineNo">121</span><a name="line.121"></a> |
| <span class="sourceLineNo">122</span> @Override<a name="line.122"></a> |
| <span class="sourceLineNo">123</span> public Cell getCurrentValue() throws IOException, InterruptedException {<a name="line.123"></a> |
| <span class="sourceLineNo">124</span> return value;<a name="line.124"></a> |
| <span class="sourceLineNo">125</span> }<a name="line.125"></a> |
| <span class="sourceLineNo">126</span><a name="line.126"></a> |
| <span class="sourceLineNo">127</span> @Override<a name="line.127"></a> |
| <span class="sourceLineNo">128</span> public float getProgress() throws IOException, InterruptedException {<a name="line.128"></a> |
| <span class="sourceLineNo">129</span> // This would be inaccurate if KVs are not uniformly-sized or we have performed a seek to<a name="line.129"></a> |
| <span class="sourceLineNo">130</span> // the start row, but better than nothing anyway.<a name="line.130"></a> |
| <span class="sourceLineNo">131</span> return 1.0f * count / in.getEntries();<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 void close() throws IOException {<a name="line.135"></a> |
| <span class="sourceLineNo">136</span> if (in != null) {<a name="line.136"></a> |
| <span class="sourceLineNo">137</span> in.close();<a name="line.137"></a> |
| <span class="sourceLineNo">138</span> in = null;<a name="line.138"></a> |
| <span class="sourceLineNo">139</span> }<a name="line.139"></a> |
| <span class="sourceLineNo">140</span> }<a name="line.140"></a> |
| <span class="sourceLineNo">141</span> }<a name="line.141"></a> |
| <span class="sourceLineNo">142</span><a name="line.142"></a> |
| <span class="sourceLineNo">143</span> @Override<a name="line.143"></a> |
| <span class="sourceLineNo">144</span> protected List<FileStatus> listStatus(JobContext job) throws IOException {<a name="line.144"></a> |
| <span class="sourceLineNo">145</span> List<FileStatus> result = new ArrayList<FileStatus>();<a name="line.145"></a> |
| <span class="sourceLineNo">146</span><a name="line.146"></a> |
| <span class="sourceLineNo">147</span> // Explode out directories that match the original FileInputFormat filters<a name="line.147"></a> |
| <span class="sourceLineNo">148</span> // since HFiles are written to directories where the<a name="line.148"></a> |
| <span class="sourceLineNo">149</span> // directory name is the column name<a name="line.149"></a> |
| <span class="sourceLineNo">150</span> for (FileStatus status : super.listStatus(job)) {<a name="line.150"></a> |
| <span class="sourceLineNo">151</span> if (status.isDirectory()) {<a name="line.151"></a> |
| <span class="sourceLineNo">152</span> FileSystem fs = status.getPath().getFileSystem(job.getConfiguration());<a name="line.152"></a> |
| <span class="sourceLineNo">153</span> for (FileStatus match : fs.listStatus(status.getPath(), HIDDEN_FILE_FILTER)) {<a name="line.153"></a> |
| <span class="sourceLineNo">154</span> result.add(match);<a name="line.154"></a> |
| <span class="sourceLineNo">155</span> }<a name="line.155"></a> |
| <span class="sourceLineNo">156</span> } else {<a name="line.156"></a> |
| <span class="sourceLineNo">157</span> result.add(status);<a name="line.157"></a> |
| <span class="sourceLineNo">158</span> }<a name="line.158"></a> |
| <span class="sourceLineNo">159</span> }<a name="line.159"></a> |
| <span class="sourceLineNo">160</span> return result;<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> @Override<a name="line.163"></a> |
| <span class="sourceLineNo">164</span> public RecordReader<NullWritable, Cell> createRecordReader(InputSplit split, TaskAttemptContext context)<a name="line.164"></a> |
| <span class="sourceLineNo">165</span> throws IOException, InterruptedException {<a name="line.165"></a> |
| <span class="sourceLineNo">166</span> return new HFileRecordReader();<a name="line.166"></a> |
| <span class="sourceLineNo">167</span> }<a name="line.167"></a> |
| <span class="sourceLineNo">168</span><a name="line.168"></a> |
| <span class="sourceLineNo">169</span> @Override<a name="line.169"></a> |
| <span class="sourceLineNo">170</span> protected boolean isSplitable(JobContext context, Path filename) {<a name="line.170"></a> |
| <span class="sourceLineNo">171</span> // This file isn't splittable.<a name="line.171"></a> |
| <span class="sourceLineNo">172</span> return false;<a name="line.172"></a> |
| <span class="sourceLineNo">173</span> }<a name="line.173"></a> |
| <span class="sourceLineNo">174</span>}<a name="line.174"></a> |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| </pre> |
| </div> |
| </body> |
| </html> |