Google Git
Sign in
apache / hbase-site / 1f0bfa6196fcf97b0302890df5bb4e21963b0077 / . / devapidocs / src-html / org / apache / hadoop / hbase / io / hfile / CompoundBloomFilter.html
blob: 7fffa5b5daf6801016c293be7a8208ebce11c63c [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.hfile;<a name="line.18"></a>
<span class="sourceLineNo">019</span><a name="line.19"></a>
<span class="sourceLineNo">020</span>import java.io.DataInput;<a name="line.20"></a>
<span class="sourceLineNo">021</span>import java.io.IOException;<a name="line.21"></a>
<span class="sourceLineNo">022</span>import org.apache.hadoop.hbase.Cell;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import org.apache.hadoop.hbase.nio.ByteBuff;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import org.apache.hadoop.hbase.regionserver.BloomType;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import org.apache.hadoop.hbase.util.BloomFilter;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.util.BloomFilterUtil;<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.Hash;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.29"></a>
<span class="sourceLineNo">030</span><a name="line.30"></a>
<span class="sourceLineNo">031</span>/**<a name="line.31"></a>
<span class="sourceLineNo">032</span> * A Bloom filter implementation built on top of<a name="line.32"></a>
<span class="sourceLineNo">033</span> * {@link org.apache.hadoop.hbase.util.BloomFilterChunk}, encapsulating a set of fixed-size Bloom<a name="line.33"></a>
<span class="sourceLineNo">034</span> * filters written out at the time of {@link org.apache.hadoop.hbase.io.hfile.HFile} generation into<a name="line.34"></a>
<span class="sourceLineNo">035</span> * the data block stream, and loaded on demand at query time. This class only provides reading<a name="line.35"></a>
<span class="sourceLineNo">036</span> * capabilities.<a name="line.36"></a>
<span class="sourceLineNo">037</span> */<a name="line.37"></a>
<span class="sourceLineNo">038</span>@InterfaceAudience.Private<a name="line.38"></a>
<span class="sourceLineNo">039</span>public class CompoundBloomFilter extends CompoundBloomFilterBase implements BloomFilter {<a name="line.39"></a>
<span class="sourceLineNo">040</span><a name="line.40"></a>
<span class="sourceLineNo">041</span> /** Used to load chunks on demand */<a name="line.41"></a>
<span class="sourceLineNo">042</span> private HFile.Reader reader;<a name="line.42"></a>
<span class="sourceLineNo">043</span> private final BloomFilterMetrics metrics;<a name="line.43"></a>
<span class="sourceLineNo">044</span><a name="line.44"></a>
<span class="sourceLineNo">045</span> private HFileBlockIndex.BlockIndexReader index;<a name="line.45"></a>
<span class="sourceLineNo">046</span><a name="line.46"></a>
<span class="sourceLineNo">047</span> private int hashCount;<a name="line.47"></a>
<span class="sourceLineNo">048</span> private Hash hash;<a name="line.48"></a>
<span class="sourceLineNo">049</span><a name="line.49"></a>
<span class="sourceLineNo">050</span> private long[] numQueriesPerChunk;<a name="line.50"></a>
<span class="sourceLineNo">051</span> private long[] numPositivesPerChunk;<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> * De-serialization for compound Bloom filter metadata. Must be consistent with what<a name="line.54"></a>
<span class="sourceLineNo">055</span> * {@link CompoundBloomFilterWriter} does.<a name="line.55"></a>
<span class="sourceLineNo">056</span> * @param meta serialized Bloom filter metadata without any magic blocks<a name="line.56"></a>
<span class="sourceLineNo">057</span> * @param reader reader for hfile<a name="line.57"></a>
<span class="sourceLineNo">058</span> * @param metrics for collecting bloom filter metrics. may be null<a name="line.58"></a>
<span class="sourceLineNo">059</span> */<a name="line.59"></a>
<span class="sourceLineNo">060</span> public CompoundBloomFilter(DataInput meta, HFile.Reader reader, BloomFilterMetrics metrics)<a name="line.60"></a>
<span class="sourceLineNo">061</span> throws IOException {<a name="line.61"></a>
<span class="sourceLineNo">062</span> this.reader = reader;<a name="line.62"></a>
<span class="sourceLineNo">063</span> this.metrics = metrics;<a name="line.63"></a>
<span class="sourceLineNo">064</span><a name="line.64"></a>
<span class="sourceLineNo">065</span> totalByteSize = meta.readLong();<a name="line.65"></a>
<span class="sourceLineNo">066</span> hashCount = meta.readInt();<a name="line.66"></a>
<span class="sourceLineNo">067</span> hashType = meta.readInt();<a name="line.67"></a>
<span class="sourceLineNo">068</span> totalKeyCount = meta.readLong();<a name="line.68"></a>
<span class="sourceLineNo">069</span> totalMaxKeys = meta.readLong();<a name="line.69"></a>
<span class="sourceLineNo">070</span> numChunks = meta.readInt();<a name="line.70"></a>
<span class="sourceLineNo">071</span> byte[] comparatorClassName = Bytes.readByteArray(meta);<a name="line.71"></a>
<span class="sourceLineNo">072</span> // The writer would have return 0 as the vint length for the case of<a name="line.72"></a>
<span class="sourceLineNo">073</span> // Bytes.BYTES_RAWCOMPARATOR. In such cases do not initialize comparator, it can be<a name="line.73"></a>
<span class="sourceLineNo">074</span> // null<a name="line.74"></a>
<span class="sourceLineNo">075</span> if (comparatorClassName.length != 0) {<a name="line.75"></a>
<span class="sourceLineNo">076</span> comparator = FixedFileTrailer.createComparator(Bytes.toString(comparatorClassName));<a name="line.76"></a>
<span class="sourceLineNo">077</span> }<a name="line.77"></a>
<span class="sourceLineNo">078</span><a name="line.78"></a>
<span class="sourceLineNo">079</span> hash = Hash.getInstance(hashType);<a name="line.79"></a>
<span class="sourceLineNo">080</span> if (hash == null) {<a name="line.80"></a>
<span class="sourceLineNo">081</span> throw new IllegalArgumentException("Invalid hash type: " + hashType);<a name="line.81"></a>
<span class="sourceLineNo">082</span> }<a name="line.82"></a>
<span class="sourceLineNo">083</span> // We will pass null for ROW block<a name="line.83"></a>
<span class="sourceLineNo">084</span> if (comparator == null) {<a name="line.84"></a>
<span class="sourceLineNo">085</span> index = new HFileBlockIndex.ByteArrayKeyBlockIndexReader(1);<a name="line.85"></a>
<span class="sourceLineNo">086</span> } else {<a name="line.86"></a>
<span class="sourceLineNo">087</span> index = new HFileBlockIndex.CellBasedKeyBlockIndexReader(comparator, 1);<a name="line.87"></a>
<span class="sourceLineNo">088</span> }<a name="line.88"></a>
<span class="sourceLineNo">089</span> index.readRootIndex(meta, numChunks);<a name="line.89"></a>
<span class="sourceLineNo">090</span> }<a name="line.90"></a>
<span class="sourceLineNo">091</span><a name="line.91"></a>
<span class="sourceLineNo">092</span> @Override<a name="line.92"></a>
<span class="sourceLineNo">093</span> public boolean contains(byte[] key, int keyOffset, int keyLength, ByteBuff bloom) {<a name="line.93"></a>
<span class="sourceLineNo">094</span> boolean result = containsInternal(key, keyOffset, keyLength, bloom);<a name="line.94"></a>
<span class="sourceLineNo">095</span> if (metrics != null) {<a name="line.95"></a>
<span class="sourceLineNo">096</span> metrics.incrementRequests(result);<a name="line.96"></a>
<span class="sourceLineNo">097</span> }<a name="line.97"></a>
<span class="sourceLineNo">098</span> return result;<a name="line.98"></a>
<span class="sourceLineNo">099</span> }<a name="line.99"></a>
<span class="sourceLineNo">100</span><a name="line.100"></a>
<span class="sourceLineNo">101</span> private boolean containsInternal(byte[] key, int keyOffset, int keyLength, ByteBuff bloom) {<a name="line.101"></a>
<span class="sourceLineNo">102</span> int block = index.rootBlockContainingKey(key, keyOffset, keyLength);<a name="line.102"></a>
<span class="sourceLineNo">103</span> if (block &lt; 0) {<a name="line.103"></a>
<span class="sourceLineNo">104</span> return false; // This key is not in the file.<a name="line.104"></a>
<span class="sourceLineNo">105</span> }<a name="line.105"></a>
<span class="sourceLineNo">106</span> boolean result;<a name="line.106"></a>
<span class="sourceLineNo">107</span> HFileBlock bloomBlock = getBloomBlock(block);<a name="line.107"></a>
<span class="sourceLineNo">108</span> try {<a name="line.108"></a>
<span class="sourceLineNo">109</span> ByteBuff bloomBuf = bloomBlock.getBufferReadOnly();<a name="line.109"></a>
<span class="sourceLineNo">110</span> result = BloomFilterUtil.contains(key, keyOffset, keyLength, bloomBuf,<a name="line.110"></a>
<span class="sourceLineNo">111</span> bloomBlock.headerSize(), bloomBlock.getUncompressedSizeWithoutHeader(), hash, hashCount);<a name="line.111"></a>
<span class="sourceLineNo">112</span> } finally {<a name="line.112"></a>
<span class="sourceLineNo">113</span> // After the use, should release the block to deallocate byte buffers.<a name="line.113"></a>
<span class="sourceLineNo">114</span> bloomBlock.release();<a name="line.114"></a>
<span class="sourceLineNo">115</span> }<a name="line.115"></a>
<span class="sourceLineNo">116</span> if (numPositivesPerChunk != null &amp;&amp; result) {<a name="line.116"></a>
<span class="sourceLineNo">117</span> // Update statistics. Only used in unit tests.<a name="line.117"></a>
<span class="sourceLineNo">118</span> ++numPositivesPerChunk[block];<a name="line.118"></a>
<span class="sourceLineNo">119</span> }<a name="line.119"></a>
<span class="sourceLineNo">120</span> return result;<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> private HFileBlock getBloomBlock(int block) {<a name="line.123"></a>
<span class="sourceLineNo">124</span> HFileBlock bloomBlock;<a name="line.124"></a>
<span class="sourceLineNo">125</span> try {<a name="line.125"></a>
<span class="sourceLineNo">126</span> // We cache the block and use a positional read.<a name="line.126"></a>
<span class="sourceLineNo">127</span> bloomBlock = reader.readBlock(index.getRootBlockOffset(block),<a name="line.127"></a>
<span class="sourceLineNo">128</span> index.getRootBlockDataSize(block), true, true, false, true, BlockType.BLOOM_CHUNK, null);<a name="line.128"></a>
<span class="sourceLineNo">129</span> } catch (IOException ex) {<a name="line.129"></a>
<span class="sourceLineNo">130</span> // The Bloom filter is broken, turn it off.<a name="line.130"></a>
<span class="sourceLineNo">131</span> throw new IllegalArgumentException("Failed to load Bloom block", ex);<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> if (numQueriesPerChunk != null) {<a name="line.134"></a>
<span class="sourceLineNo">135</span> // Update statistics. Only used in unit tests.<a name="line.135"></a>
<span class="sourceLineNo">136</span> ++numQueriesPerChunk[block];<a name="line.136"></a>
<span class="sourceLineNo">137</span> }<a name="line.137"></a>
<span class="sourceLineNo">138</span> return bloomBlock;<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> @Override<a name="line.141"></a>
<span class="sourceLineNo">142</span> public boolean contains(Cell keyCell, ByteBuff bloom, BloomType type) {<a name="line.142"></a>
<span class="sourceLineNo">143</span> boolean result = containsInternal(keyCell, bloom, type);<a name="line.143"></a>
<span class="sourceLineNo">144</span> if (metrics != null) {<a name="line.144"></a>
<span class="sourceLineNo">145</span> metrics.incrementRequests(result);<a name="line.145"></a>
<span class="sourceLineNo">146</span> }<a name="line.146"></a>
<span class="sourceLineNo">147</span> return result;<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> private boolean containsInternal(Cell keyCell, ByteBuff bloom, BloomType type) {<a name="line.150"></a>
<span class="sourceLineNo">151</span> int block = index.rootBlockContainingKey(keyCell);<a name="line.151"></a>
<span class="sourceLineNo">152</span> if (block &lt; 0) {<a name="line.152"></a>
<span class="sourceLineNo">153</span> return false; // This key is not in the file.<a name="line.153"></a>
<span class="sourceLineNo">154</span> }<a name="line.154"></a>
<span class="sourceLineNo">155</span> boolean result;<a name="line.155"></a>
<span class="sourceLineNo">156</span> HFileBlock bloomBlock = getBloomBlock(block);<a name="line.156"></a>
<span class="sourceLineNo">157</span> try {<a name="line.157"></a>
<span class="sourceLineNo">158</span> ByteBuff bloomBuf = bloomBlock.getBufferReadOnly();<a name="line.158"></a>
<span class="sourceLineNo">159</span> result = BloomFilterUtil.contains(keyCell, bloomBuf, bloomBlock.headerSize(),<a name="line.159"></a>
<span class="sourceLineNo">160</span> bloomBlock.getUncompressedSizeWithoutHeader(), hash, hashCount, type);<a name="line.160"></a>
<span class="sourceLineNo">161</span> } finally {<a name="line.161"></a>
<span class="sourceLineNo">162</span> // After the use, should release the block to deallocate the byte buffers.<a name="line.162"></a>
<span class="sourceLineNo">163</span> bloomBlock.release();<a name="line.163"></a>
<span class="sourceLineNo">164</span> }<a name="line.164"></a>
<span class="sourceLineNo">165</span> if (numPositivesPerChunk != null &amp;&amp; result) {<a name="line.165"></a>
<span class="sourceLineNo">166</span> // Update statistics. Only used in unit tests.<a name="line.166"></a>
<span class="sourceLineNo">167</span> ++numPositivesPerChunk[block];<a name="line.167"></a>
<span class="sourceLineNo">168</span> }<a name="line.168"></a>
<span class="sourceLineNo">169</span> return result;<a name="line.169"></a>
<span class="sourceLineNo">170</span> }<a name="line.170"></a>
<span class="sourceLineNo">171</span><a name="line.171"></a>
<span class="sourceLineNo">172</span> @Override<a name="line.172"></a>
<span class="sourceLineNo">173</span> public boolean supportsAutoLoading() {<a name="line.173"></a>
<span class="sourceLineNo">174</span> return true;<a name="line.174"></a>
<span class="sourceLineNo">175</span> }<a name="line.175"></a>
<span class="sourceLineNo">176</span><a name="line.176"></a>
<span class="sourceLineNo">177</span> public int getNumChunks() {<a name="line.177"></a>
<span class="sourceLineNo">178</span> return numChunks;<a name="line.178"></a>
<span class="sourceLineNo">179</span> }<a name="line.179"></a>
<span class="sourceLineNo">180</span><a name="line.180"></a>
<span class="sourceLineNo">181</span> public void enableTestingStats() {<a name="line.181"></a>
<span class="sourceLineNo">182</span> numQueriesPerChunk = new long[numChunks];<a name="line.182"></a>
<span class="sourceLineNo">183</span> numPositivesPerChunk = new long[numChunks];<a name="line.183"></a>
<span class="sourceLineNo">184</span> }<a name="line.184"></a>
<span class="sourceLineNo">185</span><a name="line.185"></a>
<span class="sourceLineNo">186</span> public String formatTestingStats() {<a name="line.186"></a>
<span class="sourceLineNo">187</span> StringBuilder sb = new StringBuilder();<a name="line.187"></a>
<span class="sourceLineNo">188</span> for (int i = 0; i &lt; numChunks; ++i) {<a name="line.188"></a>
<span class="sourceLineNo">189</span> sb.append("chunk #");<a name="line.189"></a>
<span class="sourceLineNo">190</span> sb.append(i);<a name="line.190"></a>
<span class="sourceLineNo">191</span> sb.append(": queries=");<a name="line.191"></a>
<span class="sourceLineNo">192</span> sb.append(numQueriesPerChunk[i]);<a name="line.192"></a>
<span class="sourceLineNo">193</span> sb.append(", positives=");<a name="line.193"></a>
<span class="sourceLineNo">194</span> sb.append(numPositivesPerChunk[i]);<a name="line.194"></a>
<span class="sourceLineNo">195</span> sb.append(", positiveRatio=");<a name="line.195"></a>
<span class="sourceLineNo">196</span> sb.append(numPositivesPerChunk[i] * 1.0 / numQueriesPerChunk[i]);<a name="line.196"></a>
<span class="sourceLineNo">197</span> sb.append(";\n");<a name="line.197"></a>
<span class="sourceLineNo">198</span> }<a name="line.198"></a>
<span class="sourceLineNo">199</span> return sb.toString();<a name="line.199"></a>
<span class="sourceLineNo">200</span> }<a name="line.200"></a>
<span class="sourceLineNo">201</span><a name="line.201"></a>
<span class="sourceLineNo">202</span> public long getNumQueriesForTesting(int chunk) {<a name="line.202"></a>
<span class="sourceLineNo">203</span> return numQueriesPerChunk[chunk];<a name="line.203"></a>
<span class="sourceLineNo">204</span> }<a name="line.204"></a>
<span class="sourceLineNo">205</span><a name="line.205"></a>
<span class="sourceLineNo">206</span> public long getNumPositivesForTesting(int chunk) {<a name="line.206"></a>
<span class="sourceLineNo">207</span> return numPositivesPerChunk[chunk];<a name="line.207"></a>
<span class="sourceLineNo">208</span> }<a name="line.208"></a>
<span class="sourceLineNo">209</span><a name="line.209"></a>
<span class="sourceLineNo">210</span> @Override<a name="line.210"></a>
<span class="sourceLineNo">211</span> public String toString() {<a name="line.211"></a>
<span class="sourceLineNo">212</span> StringBuilder sb = new StringBuilder();<a name="line.212"></a>
<span class="sourceLineNo">213</span> sb.append(BloomFilterUtil.formatStats(this));<a name="line.213"></a>
<span class="sourceLineNo">214</span> sb.append(BloomFilterUtil.STATS_RECORD_SEP + "Number of chunks: " + numChunks);<a name="line.214"></a>
<span class="sourceLineNo">215</span> sb.append(BloomFilterUtil.STATS_RECORD_SEP + ((comparator != null)<a name="line.215"></a>
<span class="sourceLineNo">216</span> ? "Comparator: " + comparator.getClass().getSimpleName()<a name="line.216"></a>
<span class="sourceLineNo">217</span> : "Comparator: " + Bytes.BYTES_RAWCOMPARATOR.getClass().getSimpleName()));<a name="line.217"></a>
<span class="sourceLineNo">218</span> return sb.toString();<a name="line.218"></a>
<span class="sourceLineNo">219</span> }<a name="line.219"></a>
<span class="sourceLineNo">220</span><a name="line.220"></a>
<span class="sourceLineNo">221</span>}<a name="line.221"></a>
</pre>
</div>
</body>
</html>