blob: 33ccf47d1c787b9f253d04289825398f55e75c3f [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.encoding;<a name="line.18"></a>
<span class="sourceLineNo">019</span><a name="line.19"></a>
<span class="sourceLineNo">020</span>import java.io.DataInputStream;<a name="line.20"></a>
<span class="sourceLineNo">021</span>import java.io.DataOutput;<a name="line.21"></a>
<span class="sourceLineNo">022</span>import java.io.DataOutputStream;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import java.io.IOException;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import java.nio.ByteBuffer;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import org.apache.hadoop.hbase.Cell;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.KeyValue;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.KeyValueUtil;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.PrivateCellUtil;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.apache.hadoop.hbase.nio.ByteBuff;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.apache.hadoop.hbase.util.ByteBufferUtils;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.31"></a>
<span class="sourceLineNo">032</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.32"></a>
<span class="sourceLineNo">033</span><a name="line.33"></a>
<span class="sourceLineNo">034</span>/**<a name="line.34"></a>
<span class="sourceLineNo">035</span> * Compress key by storing size of common prefix with previous KeyValue and storing raw size of<a name="line.35"></a>
<span class="sourceLineNo">036</span> * rest. Format: 1-5 bytes: compressed key length minus prefix (7-bit encoding) 1-5 bytes:<a name="line.36"></a>
<span class="sourceLineNo">037</span> * compressed value length (7-bit encoding) 1-3 bytes: compressed length of common key prefix ...<a name="line.37"></a>
<span class="sourceLineNo">038</span> * bytes: rest of key (including timestamp) ... bytes: value In a worst case compressed KeyValue<a name="line.38"></a>
<span class="sourceLineNo">039</span> * will be three bytes longer than original.<a name="line.39"></a>
<span class="sourceLineNo">040</span> */<a name="line.40"></a>
<span class="sourceLineNo">041</span>@InterfaceAudience.Private<a name="line.41"></a>
<span class="sourceLineNo">042</span>public class PrefixKeyDeltaEncoder extends BufferedDataBlockEncoder {<a name="line.42"></a>
<span class="sourceLineNo">043</span><a name="line.43"></a>
<span class="sourceLineNo">044</span> @Override<a name="line.44"></a>
<span class="sourceLineNo">045</span> public int internalEncode(Cell cell, HFileBlockDefaultEncodingContext encodingContext,<a name="line.45"></a>
<span class="sourceLineNo">046</span> DataOutputStream out) throws IOException {<a name="line.46"></a>
<span class="sourceLineNo">047</span> int klength = KeyValueUtil.keyLength(cell);<a name="line.47"></a>
<span class="sourceLineNo">048</span> int vlength = cell.getValueLength();<a name="line.48"></a>
<span class="sourceLineNo">049</span> EncodingState state = encodingContext.getEncodingState();<a name="line.49"></a>
<span class="sourceLineNo">050</span> if (state.prevCell == null) {<a name="line.50"></a>
<span class="sourceLineNo">051</span> // copy the key, there is no common prefix with none<a name="line.51"></a>
<span class="sourceLineNo">052</span> ByteBufferUtils.putCompressedInt(out, klength);<a name="line.52"></a>
<span class="sourceLineNo">053</span> ByteBufferUtils.putCompressedInt(out, vlength);<a name="line.53"></a>
<span class="sourceLineNo">054</span> ByteBufferUtils.putCompressedInt(out, 0);<a name="line.54"></a>
<span class="sourceLineNo">055</span> PrivateCellUtil.writeFlatKey(cell, (DataOutput) out);<a name="line.55"></a>
<span class="sourceLineNo">056</span> } else {<a name="line.56"></a>
<span class="sourceLineNo">057</span> // find a common prefix and skip it<a name="line.57"></a>
<span class="sourceLineNo">058</span> int common = PrivateCellUtil.findCommonPrefixInFlatKey(cell, state.prevCell, true, true);<a name="line.58"></a>
<span class="sourceLineNo">059</span> ByteBufferUtils.putCompressedInt(out, klength - common);<a name="line.59"></a>
<span class="sourceLineNo">060</span> ByteBufferUtils.putCompressedInt(out, vlength);<a name="line.60"></a>
<span class="sourceLineNo">061</span> ByteBufferUtils.putCompressedInt(out, common);<a name="line.61"></a>
<span class="sourceLineNo">062</span> writeKeyExcludingCommon(cell, common, out);<a name="line.62"></a>
<span class="sourceLineNo">063</span> }<a name="line.63"></a>
<span class="sourceLineNo">064</span> // Write the value part<a name="line.64"></a>
<span class="sourceLineNo">065</span> PrivateCellUtil.writeValue(out, cell, vlength);<a name="line.65"></a>
<span class="sourceLineNo">066</span> int size = klength + vlength + KeyValue.KEYVALUE_INFRASTRUCTURE_SIZE;<a name="line.66"></a>
<span class="sourceLineNo">067</span> size += afterEncodingKeyValue(cell, out, encodingContext);<a name="line.67"></a>
<span class="sourceLineNo">068</span> state.prevCell = cell;<a name="line.68"></a>
<span class="sourceLineNo">069</span> return size;<a name="line.69"></a>
<span class="sourceLineNo">070</span> }<a name="line.70"></a>
<span class="sourceLineNo">071</span><a name="line.71"></a>
<span class="sourceLineNo">072</span> private void writeKeyExcludingCommon(Cell cell, int commonPrefix, DataOutputStream out)<a name="line.72"></a>
<span class="sourceLineNo">073</span> throws IOException {<a name="line.73"></a>
<span class="sourceLineNo">074</span> short rLen = cell.getRowLength();<a name="line.74"></a>
<span class="sourceLineNo">075</span> if (commonPrefix &lt; rLen + KeyValue.ROW_LENGTH_SIZE) {<a name="line.75"></a>
<span class="sourceLineNo">076</span> // Previous and current rows are different. Need to write the differing part followed by<a name="line.76"></a>
<span class="sourceLineNo">077</span> // cf,q,ts and type<a name="line.77"></a>
<span class="sourceLineNo">078</span> PrivateCellUtil.writeRowKeyExcludingCommon(cell, rLen, commonPrefix, out);<a name="line.78"></a>
<span class="sourceLineNo">079</span> byte fLen = cell.getFamilyLength();<a name="line.79"></a>
<span class="sourceLineNo">080</span> out.writeByte(fLen);<a name="line.80"></a>
<span class="sourceLineNo">081</span> PrivateCellUtil.writeFamily(out, cell, fLen);<a name="line.81"></a>
<span class="sourceLineNo">082</span> PrivateCellUtil.writeQualifier(out, cell, cell.getQualifierLength());<a name="line.82"></a>
<span class="sourceLineNo">083</span> out.writeLong(cell.getTimestamp());<a name="line.83"></a>
<span class="sourceLineNo">084</span> out.writeByte(cell.getTypeByte());<a name="line.84"></a>
<span class="sourceLineNo">085</span> } else {<a name="line.85"></a>
<span class="sourceLineNo">086</span> // The full row key part is common. CF part will be common for sure as we deal with Cells in<a name="line.86"></a>
<span class="sourceLineNo">087</span> // same family. Just need write the differing part in q, ts and type<a name="line.87"></a>
<span class="sourceLineNo">088</span> commonPrefix = commonPrefix - (rLen + KeyValue.ROW_LENGTH_SIZE)<a name="line.88"></a>
<span class="sourceLineNo">089</span> - (cell.getFamilyLength() + KeyValue.FAMILY_LENGTH_SIZE);<a name="line.89"></a>
<span class="sourceLineNo">090</span> int qLen = cell.getQualifierLength();<a name="line.90"></a>
<span class="sourceLineNo">091</span> int commonQualPrefix = Math.min(commonPrefix, qLen);<a name="line.91"></a>
<span class="sourceLineNo">092</span> int qualPartLenToWrite = qLen - commonQualPrefix;<a name="line.92"></a>
<span class="sourceLineNo">093</span> if (qualPartLenToWrite &gt; 0) {<a name="line.93"></a>
<span class="sourceLineNo">094</span> PrivateCellUtil.writeQualifierSkippingBytes(out, cell, qLen, commonQualPrefix);<a name="line.94"></a>
<span class="sourceLineNo">095</span> }<a name="line.95"></a>
<span class="sourceLineNo">096</span> commonPrefix -= commonQualPrefix;<a name="line.96"></a>
<span class="sourceLineNo">097</span> // Common part in TS also?<a name="line.97"></a>
<span class="sourceLineNo">098</span> if (commonPrefix &gt; 0) {<a name="line.98"></a>
<span class="sourceLineNo">099</span> int commonTimestampPrefix = Math.min(commonPrefix, KeyValue.TIMESTAMP_SIZE);<a name="line.99"></a>
<span class="sourceLineNo">100</span> if (commonTimestampPrefix &lt; KeyValue.TIMESTAMP_SIZE) {<a name="line.100"></a>
<span class="sourceLineNo">101</span> byte[] curTsBuf = Bytes.toBytes(cell.getTimestamp());<a name="line.101"></a>
<span class="sourceLineNo">102</span> out.write(curTsBuf, commonTimestampPrefix,<a name="line.102"></a>
<span class="sourceLineNo">103</span> KeyValue.TIMESTAMP_SIZE - commonTimestampPrefix);<a name="line.103"></a>
<span class="sourceLineNo">104</span> }<a name="line.104"></a>
<span class="sourceLineNo">105</span> commonPrefix -= commonTimestampPrefix;<a name="line.105"></a>
<span class="sourceLineNo">106</span> if (commonPrefix == 0) {<a name="line.106"></a>
<span class="sourceLineNo">107</span> out.writeByte(cell.getTypeByte());<a name="line.107"></a>
<span class="sourceLineNo">108</span> }<a name="line.108"></a>
<span class="sourceLineNo">109</span> } else {<a name="line.109"></a>
<span class="sourceLineNo">110</span> out.writeLong(cell.getTimestamp());<a name="line.110"></a>
<span class="sourceLineNo">111</span> out.writeByte(cell.getTypeByte());<a name="line.111"></a>
<span class="sourceLineNo">112</span> }<a name="line.112"></a>
<span class="sourceLineNo">113</span> }<a name="line.113"></a>
<span class="sourceLineNo">114</span> }<a name="line.114"></a>
<span class="sourceLineNo">115</span><a name="line.115"></a>
<span class="sourceLineNo">116</span> @Override<a name="line.116"></a>
<span class="sourceLineNo">117</span> protected ByteBuffer internalDecodeKeyValues(DataInputStream source, int allocateHeaderLength,<a name="line.117"></a>
<span class="sourceLineNo">118</span> int skipLastBytes, HFileBlockDefaultDecodingContext decodingCtx) throws IOException {<a name="line.118"></a>
<span class="sourceLineNo">119</span> int decompressedSize = source.readInt();<a name="line.119"></a>
<span class="sourceLineNo">120</span> ByteBuffer buffer = ByteBuffer.allocate(decompressedSize + allocateHeaderLength);<a name="line.120"></a>
<span class="sourceLineNo">121</span> buffer.position(allocateHeaderLength);<a name="line.121"></a>
<span class="sourceLineNo">122</span> int prevKeyOffset = 0;<a name="line.122"></a>
<span class="sourceLineNo">123</span><a name="line.123"></a>
<span class="sourceLineNo">124</span> while (source.available() &gt; skipLastBytes) {<a name="line.124"></a>
<span class="sourceLineNo">125</span> prevKeyOffset = decodeKeyValue(source, buffer, prevKeyOffset);<a name="line.125"></a>
<span class="sourceLineNo">126</span> afterDecodingKeyValue(source, buffer, decodingCtx);<a name="line.126"></a>
<span class="sourceLineNo">127</span> }<a name="line.127"></a>
<span class="sourceLineNo">128</span><a name="line.128"></a>
<span class="sourceLineNo">129</span> if (source.available() != skipLastBytes) {<a name="line.129"></a>
<span class="sourceLineNo">130</span> throw new IllegalStateException("Read too many bytes.");<a name="line.130"></a>
<span class="sourceLineNo">131</span> }<a name="line.131"></a>
<span class="sourceLineNo">132</span><a name="line.132"></a>
<span class="sourceLineNo">133</span> buffer.limit(buffer.position());<a name="line.133"></a>
<span class="sourceLineNo">134</span> return buffer;<a name="line.134"></a>
<span class="sourceLineNo">135</span> }<a name="line.135"></a>
<span class="sourceLineNo">136</span><a name="line.136"></a>
<span class="sourceLineNo">137</span> private int decodeKeyValue(DataInputStream source, ByteBuffer buffer, int prevKeyOffset)<a name="line.137"></a>
<span class="sourceLineNo">138</span> throws IOException, EncoderBufferTooSmallException {<a name="line.138"></a>
<span class="sourceLineNo">139</span> int keyLength = ByteBufferUtils.readCompressedInt(source);<a name="line.139"></a>
<span class="sourceLineNo">140</span> int valueLength = ByteBufferUtils.readCompressedInt(source);<a name="line.140"></a>
<span class="sourceLineNo">141</span> int commonLength = ByteBufferUtils.readCompressedInt(source);<a name="line.141"></a>
<span class="sourceLineNo">142</span> int keyOffset;<a name="line.142"></a>
<span class="sourceLineNo">143</span> keyLength += commonLength;<a name="line.143"></a>
<span class="sourceLineNo">144</span><a name="line.144"></a>
<span class="sourceLineNo">145</span> ensureSpace(buffer, keyLength + valueLength + KeyValue.ROW_OFFSET);<a name="line.145"></a>
<span class="sourceLineNo">146</span><a name="line.146"></a>
<span class="sourceLineNo">147</span> buffer.putInt(keyLength);<a name="line.147"></a>
<span class="sourceLineNo">148</span> buffer.putInt(valueLength);<a name="line.148"></a>
<span class="sourceLineNo">149</span><a name="line.149"></a>
<span class="sourceLineNo">150</span> // copy the prefix<a name="line.150"></a>
<span class="sourceLineNo">151</span> if (commonLength &gt; 0) {<a name="line.151"></a>
<span class="sourceLineNo">152</span> keyOffset = buffer.position();<a name="line.152"></a>
<span class="sourceLineNo">153</span> ByteBufferUtils.copyFromBufferToBuffer(buffer, buffer, prevKeyOffset, commonLength);<a name="line.153"></a>
<span class="sourceLineNo">154</span> } else {<a name="line.154"></a>
<span class="sourceLineNo">155</span> keyOffset = buffer.position();<a name="line.155"></a>
<span class="sourceLineNo">156</span> }<a name="line.156"></a>
<span class="sourceLineNo">157</span><a name="line.157"></a>
<span class="sourceLineNo">158</span> // copy rest of the key and value<a name="line.158"></a>
<span class="sourceLineNo">159</span> int len = keyLength - commonLength + valueLength;<a name="line.159"></a>
<span class="sourceLineNo">160</span> ByteBufferUtils.copyFromStreamToBuffer(buffer, source, len);<a name="line.160"></a>
<span class="sourceLineNo">161</span> return keyOffset;<a name="line.161"></a>
<span class="sourceLineNo">162</span> }<a name="line.162"></a>
<span class="sourceLineNo">163</span><a name="line.163"></a>
<span class="sourceLineNo">164</span> @Override<a name="line.164"></a>
<span class="sourceLineNo">165</span> public Cell getFirstKeyCellInBlock(ByteBuff block) {<a name="line.165"></a>
<span class="sourceLineNo">166</span> block.mark();<a name="line.166"></a>
<span class="sourceLineNo">167</span> block.position(Bytes.SIZEOF_INT);<a name="line.167"></a>
<span class="sourceLineNo">168</span> int keyLength = ByteBuff.readCompressedInt(block);<a name="line.168"></a>
<span class="sourceLineNo">169</span> // TODO : See if we can avoid these reads as the read values are not getting used<a name="line.169"></a>
<span class="sourceLineNo">170</span> ByteBuff.readCompressedInt(block);<a name="line.170"></a>
<span class="sourceLineNo">171</span> int commonLength = ByteBuff.readCompressedInt(block);<a name="line.171"></a>
<span class="sourceLineNo">172</span> if (commonLength != 0) {<a name="line.172"></a>
<span class="sourceLineNo">173</span> throw new AssertionError(<a name="line.173"></a>
<span class="sourceLineNo">174</span> "Nonzero common length in the first key in " + "block: " + commonLength);<a name="line.174"></a>
<span class="sourceLineNo">175</span> }<a name="line.175"></a>
<span class="sourceLineNo">176</span> ByteBuffer key = block.asSubByteBuffer(keyLength).duplicate();<a name="line.176"></a>
<span class="sourceLineNo">177</span> block.reset();<a name="line.177"></a>
<span class="sourceLineNo">178</span> return createFirstKeyCell(key, keyLength);<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> @Override<a name="line.181"></a>
<span class="sourceLineNo">182</span> public String toString() {<a name="line.182"></a>
<span class="sourceLineNo">183</span> return PrefixKeyDeltaEncoder.class.getSimpleName();<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> @Override<a name="line.186"></a>
<span class="sourceLineNo">187</span> public EncodedSeeker createSeeker(final HFileBlockDecodingContext decodingCtx) {<a name="line.187"></a>
<span class="sourceLineNo">188</span> return new SeekerStateBufferedEncodedSeeker(decodingCtx);<a name="line.188"></a>
<span class="sourceLineNo">189</span> }<a name="line.189"></a>
<span class="sourceLineNo">190</span><a name="line.190"></a>
<span class="sourceLineNo">191</span> private static class SeekerStateBufferedEncodedSeeker extends BufferedEncodedSeeker&lt;SeekerState&gt; {<a name="line.191"></a>
<span class="sourceLineNo">192</span><a name="line.192"></a>
<span class="sourceLineNo">193</span> private SeekerStateBufferedEncodedSeeker(HFileBlockDecodingContext decodingCtx) {<a name="line.193"></a>
<span class="sourceLineNo">194</span> super(decodingCtx);<a name="line.194"></a>
<span class="sourceLineNo">195</span> }<a name="line.195"></a>
<span class="sourceLineNo">196</span><a name="line.196"></a>
<span class="sourceLineNo">197</span> @Override<a name="line.197"></a>
<span class="sourceLineNo">198</span> protected void decodeNext() {<a name="line.198"></a>
<span class="sourceLineNo">199</span> current.keyLength = ByteBuff.readCompressedInt(currentBuffer);<a name="line.199"></a>
<span class="sourceLineNo">200</span> current.valueLength = ByteBuff.readCompressedInt(currentBuffer);<a name="line.200"></a>
<span class="sourceLineNo">201</span> current.lastCommonPrefix = ByteBuff.readCompressedInt(currentBuffer);<a name="line.201"></a>
<span class="sourceLineNo">202</span> current.keyLength += current.lastCommonPrefix;<a name="line.202"></a>
<span class="sourceLineNo">203</span> current.ensureSpaceForKey();<a name="line.203"></a>
<span class="sourceLineNo">204</span> currentBuffer.get(current.keyBuffer, current.lastCommonPrefix,<a name="line.204"></a>
<span class="sourceLineNo">205</span> current.keyLength - current.lastCommonPrefix);<a name="line.205"></a>
<span class="sourceLineNo">206</span> current.valueOffset = currentBuffer.position();<a name="line.206"></a>
<span class="sourceLineNo">207</span> currentBuffer.skip(current.valueLength);<a name="line.207"></a>
<span class="sourceLineNo">208</span> if (includesTags()) {<a name="line.208"></a>
<span class="sourceLineNo">209</span> decodeTags();<a name="line.209"></a>
<span class="sourceLineNo">210</span> }<a name="line.210"></a>
<span class="sourceLineNo">211</span> if (includesMvcc()) {<a name="line.211"></a>
<span class="sourceLineNo">212</span> current.memstoreTS = ByteBufferUtils.readVLong(currentBuffer);<a name="line.212"></a>
<span class="sourceLineNo">213</span> } else {<a name="line.213"></a>
<span class="sourceLineNo">214</span> current.memstoreTS = 0;<a name="line.214"></a>
<span class="sourceLineNo">215</span> }<a name="line.215"></a>
<span class="sourceLineNo">216</span> current.nextKvOffset = currentBuffer.position();<a name="line.216"></a>
<span class="sourceLineNo">217</span> }<a name="line.217"></a>
<span class="sourceLineNo">218</span><a name="line.218"></a>
<span class="sourceLineNo">219</span> @Override<a name="line.219"></a>
<span class="sourceLineNo">220</span> protected void decodeFirst() {<a name="line.220"></a>
<span class="sourceLineNo">221</span> currentBuffer.skip(Bytes.SIZEOF_INT);<a name="line.221"></a>
<span class="sourceLineNo">222</span> decodeNext();<a name="line.222"></a>
<span class="sourceLineNo">223</span> }<a name="line.223"></a>
<span class="sourceLineNo">224</span> }<a name="line.224"></a>
<span class="sourceLineNo">225</span>}<a name="line.225"></a>
</pre>
</div>
</body>
</html>