blob: a3d21f72232f170a798ea8c39671c976a71d1a5e [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.io.compress;<a name="line.18"></a>
<span class="sourceLineNo">019</span><a name="line.19"></a>
<span class="sourceLineNo">020</span>import static org.junit.Assert.assertTrue;<a name="line.20"></a>
<span class="sourceLineNo">021</span><a name="line.21"></a>
<span class="sourceLineNo">022</span>import java.io.ByteArrayInputStream;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import java.io.ByteArrayOutputStream;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import java.util.Arrays;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import java.util.Random;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import org.apache.hadoop.conf.Configurable;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.apache.hadoop.hbase.util.RandomDistribution;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.apache.hadoop.io.IOUtils;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import org.apache.hadoop.io.compress.CompressionCodec;<a name="line.31"></a>
<span class="sourceLineNo">032</span>import org.apache.hadoop.io.compress.CompressionInputStream;<a name="line.32"></a>
<span class="sourceLineNo">033</span>import org.apache.hadoop.io.compress.CompressionOutputStream;<a name="line.33"></a>
<span class="sourceLineNo">034</span>import org.apache.hadoop.io.compress.Compressor;<a name="line.34"></a>
<span class="sourceLineNo">035</span>import org.apache.hadoop.io.compress.Decompressor;<a name="line.35"></a>
<span class="sourceLineNo">036</span>import org.slf4j.Logger;<a name="line.36"></a>
<span class="sourceLineNo">037</span>import org.slf4j.LoggerFactory;<a name="line.37"></a>
<span class="sourceLineNo">038</span><a name="line.38"></a>
<span class="sourceLineNo">039</span>@SuppressWarnings("checkstyle:innerassignment")<a name="line.39"></a>
<span class="sourceLineNo">040</span>public class CompressionTestBase {<a name="line.40"></a>
<span class="sourceLineNo">041</span><a name="line.41"></a>
<span class="sourceLineNo">042</span> protected static final Logger LOG = LoggerFactory.getLogger(CompressionTestBase.class);<a name="line.42"></a>
<span class="sourceLineNo">043</span><a name="line.43"></a>
<span class="sourceLineNo">044</span> protected static final int LARGE_SIZE = 10 * 1024 * 1024;<a name="line.44"></a>
<span class="sourceLineNo">045</span> protected static final int VERY_LARGE_SIZE = 100 * 1024 * 1024;<a name="line.45"></a>
<span class="sourceLineNo">046</span> protected static final int BLOCK_SIZE = 4096;<a name="line.46"></a>
<span class="sourceLineNo">047</span><a name="line.47"></a>
<span class="sourceLineNo">048</span> protected static final byte[] SMALL_INPUT;<a name="line.48"></a>
<span class="sourceLineNo">049</span> static {<a name="line.49"></a>
<span class="sourceLineNo">050</span> // 1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, 233, 377, 610, 987, 1597<a name="line.50"></a>
<span class="sourceLineNo">051</span> SMALL_INPUT =<a name="line.51"></a>
<span class="sourceLineNo">052</span> new byte[1 + 1 + 2 + 3 + 5 + 8 + 13 + 21 + 34 + 55 + 89 + 144 + 233 + 377 + 610 + 987 + 1597];<a name="line.52"></a>
<span class="sourceLineNo">053</span> int off = 0;<a name="line.53"></a>
<span class="sourceLineNo">054</span> Arrays.fill(SMALL_INPUT, off, (off += 1), (byte) 'A');<a name="line.54"></a>
<span class="sourceLineNo">055</span> Arrays.fill(SMALL_INPUT, off, (off += 1), (byte) 'B');<a name="line.55"></a>
<span class="sourceLineNo">056</span> Arrays.fill(SMALL_INPUT, off, (off += 2), (byte) 'C');<a name="line.56"></a>
<span class="sourceLineNo">057</span> Arrays.fill(SMALL_INPUT, off, (off += 3), (byte) 'D');<a name="line.57"></a>
<span class="sourceLineNo">058</span> Arrays.fill(SMALL_INPUT, off, (off += 5), (byte) 'E');<a name="line.58"></a>
<span class="sourceLineNo">059</span> Arrays.fill(SMALL_INPUT, off, (off += 8), (byte) 'F');<a name="line.59"></a>
<span class="sourceLineNo">060</span> Arrays.fill(SMALL_INPUT, off, (off += 13), (byte) 'G');<a name="line.60"></a>
<span class="sourceLineNo">061</span> Arrays.fill(SMALL_INPUT, off, (off += 21), (byte) 'H');<a name="line.61"></a>
<span class="sourceLineNo">062</span> Arrays.fill(SMALL_INPUT, off, (off += 34), (byte) 'I');<a name="line.62"></a>
<span class="sourceLineNo">063</span> Arrays.fill(SMALL_INPUT, off, (off += 55), (byte) 'J');<a name="line.63"></a>
<span class="sourceLineNo">064</span> Arrays.fill(SMALL_INPUT, off, (off += 89), (byte) 'K');<a name="line.64"></a>
<span class="sourceLineNo">065</span> Arrays.fill(SMALL_INPUT, off, (off += 144), (byte) 'L');<a name="line.65"></a>
<span class="sourceLineNo">066</span> Arrays.fill(SMALL_INPUT, off, (off += 233), (byte) 'M');<a name="line.66"></a>
<span class="sourceLineNo">067</span> Arrays.fill(SMALL_INPUT, off, (off += 377), (byte) 'N');<a name="line.67"></a>
<span class="sourceLineNo">068</span> Arrays.fill(SMALL_INPUT, off, (off += 610), (byte) 'O');<a name="line.68"></a>
<span class="sourceLineNo">069</span> Arrays.fill(SMALL_INPUT, off, (off += 987), (byte) 'P');<a name="line.69"></a>
<span class="sourceLineNo">070</span> Arrays.fill(SMALL_INPUT, off, (off += 1597), (byte) 'Q');<a name="line.70"></a>
<span class="sourceLineNo">071</span> }<a name="line.71"></a>
<span class="sourceLineNo">072</span><a name="line.72"></a>
<span class="sourceLineNo">073</span> protected void codecTest(final CompressionCodec codec, final byte[][] input) throws Exception {<a name="line.73"></a>
<span class="sourceLineNo">074</span> codecTest(codec, input, null);<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> protected void codecTest(final CompressionCodec codec, final byte[][] input,<a name="line.77"></a>
<span class="sourceLineNo">078</span> final Integer expectedCompressedSize) throws Exception {<a name="line.78"></a>
<span class="sourceLineNo">079</span> // We do this in Compression.java<a name="line.79"></a>
<span class="sourceLineNo">080</span> ((Configurable) codec).getConf().setInt("io.file.buffer.size", 32 * 1024);<a name="line.80"></a>
<span class="sourceLineNo">081</span> // Compress<a name="line.81"></a>
<span class="sourceLineNo">082</span> long start = EnvironmentEdgeManager.currentTime();<a name="line.82"></a>
<span class="sourceLineNo">083</span> Compressor compressor = codec.createCompressor();<a name="line.83"></a>
<span class="sourceLineNo">084</span> ByteArrayOutputStream baos = new ByteArrayOutputStream();<a name="line.84"></a>
<span class="sourceLineNo">085</span> CompressionOutputStream out = codec.createOutputStream(baos, compressor);<a name="line.85"></a>
<span class="sourceLineNo">086</span> int inLen = 0;<a name="line.86"></a>
<span class="sourceLineNo">087</span> for (int i = 0; i &lt; input.length; i++) {<a name="line.87"></a>
<span class="sourceLineNo">088</span> out.write(input[i]);<a name="line.88"></a>
<span class="sourceLineNo">089</span> inLen += input[i].length;<a name="line.89"></a>
<span class="sourceLineNo">090</span> }<a name="line.90"></a>
<span class="sourceLineNo">091</span> out.close();<a name="line.91"></a>
<span class="sourceLineNo">092</span> long end = EnvironmentEdgeManager.currentTime();<a name="line.92"></a>
<span class="sourceLineNo">093</span> final byte[] compressed = baos.toByteArray();<a name="line.93"></a>
<span class="sourceLineNo">094</span> LOG.info("{} compressed {} bytes to {} bytes in {} ms", codec.getClass().getSimpleName(), inLen,<a name="line.94"></a>
<span class="sourceLineNo">095</span> compressed.length, end - start);<a name="line.95"></a>
<span class="sourceLineNo">096</span> if (expectedCompressedSize != null) {<a name="line.96"></a>
<span class="sourceLineNo">097</span> assertTrue("Expected compressed size does not match: (expected=" + expectedCompressedSize<a name="line.97"></a>
<span class="sourceLineNo">098</span> + ", actual=" + compressed.length + ")", expectedCompressedSize == compressed.length);<a name="line.98"></a>
<span class="sourceLineNo">099</span> }<a name="line.99"></a>
<span class="sourceLineNo">100</span> // Decompress<a name="line.100"></a>
<span class="sourceLineNo">101</span> final byte[] plain = new byte[inLen];<a name="line.101"></a>
<span class="sourceLineNo">102</span> Decompressor decompressor = codec.createDecompressor();<a name="line.102"></a>
<span class="sourceLineNo">103</span> CompressionInputStream in =<a name="line.103"></a>
<span class="sourceLineNo">104</span> codec.createInputStream(new ByteArrayInputStream(compressed), decompressor);<a name="line.104"></a>
<span class="sourceLineNo">105</span> start = EnvironmentEdgeManager.currentTime();<a name="line.105"></a>
<span class="sourceLineNo">106</span> IOUtils.readFully(in, plain, 0, plain.length);<a name="line.106"></a>
<span class="sourceLineNo">107</span> in.close();<a name="line.107"></a>
<span class="sourceLineNo">108</span> end = EnvironmentEdgeManager.currentTime();<a name="line.108"></a>
<span class="sourceLineNo">109</span> LOG.info("{} decompressed {} bytes to {} bytes in {} ms", codec.getClass().getSimpleName(),<a name="line.109"></a>
<span class="sourceLineNo">110</span> compressed.length, plain.length, end - start);<a name="line.110"></a>
<span class="sourceLineNo">111</span> // Decompressed bytes should equal the original<a name="line.111"></a>
<span class="sourceLineNo">112</span> int offset = 0;<a name="line.112"></a>
<span class="sourceLineNo">113</span> for (int i = 0; i &lt; input.length; i++) {<a name="line.113"></a>
<span class="sourceLineNo">114</span> assertTrue("Comparison failed at offset " + offset,<a name="line.114"></a>
<span class="sourceLineNo">115</span> Bytes.compareTo(plain, offset, input[i].length, input[i], 0, input[i].length) == 0);<a name="line.115"></a>
<span class="sourceLineNo">116</span> offset += input[i].length;<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><a name="line.119"></a>
<span class="sourceLineNo">120</span> /**<a name="line.120"></a>
<span class="sourceLineNo">121</span> * Test with one smallish input buffer<a name="line.121"></a>
<span class="sourceLineNo">122</span> */<a name="line.122"></a>
<span class="sourceLineNo">123</span> protected void codecSmallTest(final CompressionCodec codec) throws Exception {<a name="line.123"></a>
<span class="sourceLineNo">124</span> codecTest(codec, new byte[][] { SMALL_INPUT });<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> /**<a name="line.127"></a>
<span class="sourceLineNo">128</span> * Test with a large input (1MB) divided into blocks of 4KB.<a name="line.128"></a>
<span class="sourceLineNo">129</span> */<a name="line.129"></a>
<span class="sourceLineNo">130</span> protected void codecLargeTest(final CompressionCodec codec, final double sigma) throws Exception {<a name="line.130"></a>
<span class="sourceLineNo">131</span> RandomDistribution.DiscreteRNG rng =<a name="line.131"></a>
<span class="sourceLineNo">132</span> new RandomDistribution.Zipf(new Random(), 0, Byte.MAX_VALUE, sigma);<a name="line.132"></a>
<span class="sourceLineNo">133</span> final byte[][] input = new byte[LARGE_SIZE / BLOCK_SIZE][BLOCK_SIZE];<a name="line.133"></a>
<span class="sourceLineNo">134</span> fill(rng, input);<a name="line.134"></a>
<span class="sourceLineNo">135</span> codecTest(codec, input);<a name="line.135"></a>
<span class="sourceLineNo">136</span> }<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> * Test with a very large input (100MB) as a single input buffer.<a name="line.139"></a>
<span class="sourceLineNo">140</span> */<a name="line.140"></a>
<span class="sourceLineNo">141</span> protected void codecVeryLargeTest(final CompressionCodec codec, final double sigma)<a name="line.141"></a>
<span class="sourceLineNo">142</span> throws Exception {<a name="line.142"></a>
<span class="sourceLineNo">143</span> RandomDistribution.DiscreteRNG rng =<a name="line.143"></a>
<span class="sourceLineNo">144</span> new RandomDistribution.Zipf(new Random(), 0, Byte.MAX_VALUE, sigma);<a name="line.144"></a>
<span class="sourceLineNo">145</span> final byte[][] input = new byte[1][VERY_LARGE_SIZE];<a name="line.145"></a>
<span class="sourceLineNo">146</span> fill(rng, input);<a name="line.146"></a>
<span class="sourceLineNo">147</span> codecTest(codec, input);<a name="line.147"></a>
<span class="sourceLineNo">148</span> }<a name="line.148"></a>
<span class="sourceLineNo">149</span><a name="line.149"></a>
<span class="sourceLineNo">150</span> protected static void fill(RandomDistribution.DiscreteRNG rng, byte[][] input) {<a name="line.150"></a>
<span class="sourceLineNo">151</span> for (int i = 0; i &lt; input.length; i++) {<a name="line.151"></a>
<span class="sourceLineNo">152</span> fill(rng, input[i]);<a name="line.152"></a>
<span class="sourceLineNo">153</span> }<a name="line.153"></a>
<span class="sourceLineNo">154</span> }<a name="line.154"></a>
<span class="sourceLineNo">155</span><a name="line.155"></a>
<span class="sourceLineNo">156</span> protected static void fill(RandomDistribution.DiscreteRNG rng, byte[] input) {<a name="line.156"></a>
<span class="sourceLineNo">157</span> for (int i = 0; i &lt; input.length; i++) {<a name="line.157"></a>
<span class="sourceLineNo">158</span> input[i] = (byte) rng.nextInt();<a name="line.158"></a>
<span class="sourceLineNo">159</span> }<a name="line.159"></a>
<span class="sourceLineNo">160</span> }<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>
</pre>
</div>
</body>
</html>