Google Git
Sign in
apache / hbase-site / 2c276d06aa089b93c8b513d57d8ef6b971ea46ba / . / devapidocs / src-html / org / apache / hadoop / hbase / mapreduce / TsvImporterMapper.html
blob: 6877561d9225a8c6fa10996af24bf928d1e48032 [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.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.Base64;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import java.util.List;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import org.apache.hadoop.conf.Configuration;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import org.apache.hadoop.hbase.ArrayBackedTag;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.Cell;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.KeyValue;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.Tag;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.apache.hadoop.hbase.TagType;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.apache.hadoop.hbase.client.Put;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.31"></a>
<span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.mapreduce.ImportTsv.TsvParser.BadTsvLineException;<a name="line.32"></a>
<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.security.visibility.CellVisibility;<a name="line.33"></a>
<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.security.visibility.InvalidLabelException;<a name="line.34"></a>
<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.35"></a>
<span class="sourceLineNo">036</span>import org.apache.hadoop.io.LongWritable;<a name="line.36"></a>
<span class="sourceLineNo">037</span>import org.apache.hadoop.io.Text;<a name="line.37"></a>
<span class="sourceLineNo">038</span>import org.apache.hadoop.mapreduce.Counter;<a name="line.38"></a>
<span class="sourceLineNo">039</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.39"></a>
<span class="sourceLineNo">040</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.40"></a>
<span class="sourceLineNo">041</span>import org.slf4j.Logger;<a name="line.41"></a>
<span class="sourceLineNo">042</span>import org.slf4j.LoggerFactory;<a name="line.42"></a>
<span class="sourceLineNo">043</span><a name="line.43"></a>
<span class="sourceLineNo">044</span>/**<a name="line.44"></a>
<span class="sourceLineNo">045</span> * Write table content out to files in hdfs.<a name="line.45"></a>
<span class="sourceLineNo">046</span> */<a name="line.46"></a>
<span class="sourceLineNo">047</span>@InterfaceAudience.Public<a name="line.47"></a>
<span class="sourceLineNo">048</span>public class TsvImporterMapper extends Mapper&lt;LongWritable, Text, ImmutableBytesWritable, Put&gt; {<a name="line.48"></a>
<span class="sourceLineNo">049</span> private static final Logger LOG = LoggerFactory.getLogger(TsvImporterMapper.class);<a name="line.49"></a>
<span class="sourceLineNo">050</span><a name="line.50"></a>
<span class="sourceLineNo">051</span> /** Timestamp for all inserted rows */<a name="line.51"></a>
<span class="sourceLineNo">052</span> protected long ts;<a name="line.52"></a>
<span class="sourceLineNo">053</span><a name="line.53"></a>
<span class="sourceLineNo">054</span> /** Column seperator */<a name="line.54"></a>
<span class="sourceLineNo">055</span> private String separator;<a name="line.55"></a>
<span class="sourceLineNo">056</span><a name="line.56"></a>
<span class="sourceLineNo">057</span> /** Should skip bad lines */<a name="line.57"></a>
<span class="sourceLineNo">058</span> private boolean skipBadLines;<a name="line.58"></a>
<span class="sourceLineNo">059</span> /** Should skip empty columns */<a name="line.59"></a>
<span class="sourceLineNo">060</span> private boolean skipEmptyColumns;<a name="line.60"></a>
<span class="sourceLineNo">061</span> private Counter badLineCount;<a name="line.61"></a>
<span class="sourceLineNo">062</span> private boolean logBadLines;<a name="line.62"></a>
<span class="sourceLineNo">063</span><a name="line.63"></a>
<span class="sourceLineNo">064</span> protected ImportTsv.TsvParser parser;<a name="line.64"></a>
<span class="sourceLineNo">065</span><a name="line.65"></a>
<span class="sourceLineNo">066</span> protected Configuration conf;<a name="line.66"></a>
<span class="sourceLineNo">067</span><a name="line.67"></a>
<span class="sourceLineNo">068</span> protected String cellVisibilityExpr;<a name="line.68"></a>
<span class="sourceLineNo">069</span><a name="line.69"></a>
<span class="sourceLineNo">070</span> protected long ttl;<a name="line.70"></a>
<span class="sourceLineNo">071</span><a name="line.71"></a>
<span class="sourceLineNo">072</span> protected CellCreator kvCreator;<a name="line.72"></a>
<span class="sourceLineNo">073</span><a name="line.73"></a>
<span class="sourceLineNo">074</span> private String hfileOutPath;<a name="line.74"></a>
<span class="sourceLineNo">075</span><a name="line.75"></a>
<span class="sourceLineNo">076</span> /** List of cell tags */<a name="line.76"></a>
<span class="sourceLineNo">077</span> private List&lt;Tag&gt; tags;<a name="line.77"></a>
<span class="sourceLineNo">078</span><a name="line.78"></a>
<span class="sourceLineNo">079</span> public long getTs() {<a name="line.79"></a>
<span class="sourceLineNo">080</span> return ts;<a name="line.80"></a>
<span class="sourceLineNo">081</span> }<a name="line.81"></a>
<span class="sourceLineNo">082</span><a name="line.82"></a>
<span class="sourceLineNo">083</span> public boolean getSkipBadLines() {<a name="line.83"></a>
<span class="sourceLineNo">084</span> return skipBadLines;<a name="line.84"></a>
<span class="sourceLineNo">085</span> }<a name="line.85"></a>
<span class="sourceLineNo">086</span><a name="line.86"></a>
<span class="sourceLineNo">087</span> public Counter getBadLineCount() {<a name="line.87"></a>
<span class="sourceLineNo">088</span> return badLineCount;<a name="line.88"></a>
<span class="sourceLineNo">089</span> }<a name="line.89"></a>
<span class="sourceLineNo">090</span><a name="line.90"></a>
<span class="sourceLineNo">091</span> public void incrementBadLineCount(int count) {<a name="line.91"></a>
<span class="sourceLineNo">092</span> this.badLineCount.increment(count);<a name="line.92"></a>
<span class="sourceLineNo">093</span> }<a name="line.93"></a>
<span class="sourceLineNo">094</span><a name="line.94"></a>
<span class="sourceLineNo">095</span> /**<a name="line.95"></a>
<span class="sourceLineNo">096</span> * Handles initializing this class with objects specific to it (i.e., the parser). Common<a name="line.96"></a>
<span class="sourceLineNo">097</span> * initialization that might be leveraged by a subsclass is done in &lt;code&gt;doSetup&lt;/code&gt;. Hence a<a name="line.97"></a>
<span class="sourceLineNo">098</span> * subclass may choose to override this method and call &lt;code&gt;doSetup&lt;/code&gt; as well before<a name="line.98"></a>
<span class="sourceLineNo">099</span> * handling it's own custom params. n<a name="line.99"></a>
<span class="sourceLineNo">100</span> */<a name="line.100"></a>
<span class="sourceLineNo">101</span> @Override<a name="line.101"></a>
<span class="sourceLineNo">102</span> protected void setup(Context context) {<a name="line.102"></a>
<span class="sourceLineNo">103</span> doSetup(context);<a name="line.103"></a>
<span class="sourceLineNo">104</span><a name="line.104"></a>
<span class="sourceLineNo">105</span> conf = context.getConfiguration();<a name="line.105"></a>
<span class="sourceLineNo">106</span> parser = new ImportTsv.TsvParser(conf.get(ImportTsv.COLUMNS_CONF_KEY), separator);<a name="line.106"></a>
<span class="sourceLineNo">107</span> if (parser.getRowKeyColumnIndex() == -1) {<a name="line.107"></a>
<span class="sourceLineNo">108</span> throw new RuntimeException("No row key column specified");<a name="line.108"></a>
<span class="sourceLineNo">109</span> }<a name="line.109"></a>
<span class="sourceLineNo">110</span> this.kvCreator = new CellCreator(conf);<a name="line.110"></a>
<span class="sourceLineNo">111</span> tags = new ArrayList&lt;&gt;();<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> * Handles common parameter initialization that a subclass might want to leverage. n<a name="line.115"></a>
<span class="sourceLineNo">116</span> */<a name="line.116"></a>
<span class="sourceLineNo">117</span> protected void doSetup(Context context) {<a name="line.117"></a>
<span class="sourceLineNo">118</span> Configuration conf = context.getConfiguration();<a name="line.118"></a>
<span class="sourceLineNo">119</span><a name="line.119"></a>
<span class="sourceLineNo">120</span> // If a custom separator has been used,<a name="line.120"></a>
<span class="sourceLineNo">121</span> // decode it back from Base64 encoding.<a name="line.121"></a>
<span class="sourceLineNo">122</span> separator = conf.get(ImportTsv.SEPARATOR_CONF_KEY);<a name="line.122"></a>
<span class="sourceLineNo">123</span> if (separator == null) {<a name="line.123"></a>
<span class="sourceLineNo">124</span> separator = ImportTsv.DEFAULT_SEPARATOR;<a name="line.124"></a>
<span class="sourceLineNo">125</span> } else {<a name="line.125"></a>
<span class="sourceLineNo">126</span> separator = new String(Base64.getDecoder().decode(separator));<a name="line.126"></a>
<span class="sourceLineNo">127</span> }<a name="line.127"></a>
<span class="sourceLineNo">128</span> // Should never get 0 as we are setting this to a valid value in job<a name="line.128"></a>
<span class="sourceLineNo">129</span> // configuration.<a name="line.129"></a>
<span class="sourceLineNo">130</span> ts = conf.getLong(ImportTsv.TIMESTAMP_CONF_KEY, 0);<a name="line.130"></a>
<span class="sourceLineNo">131</span><a name="line.131"></a>
<span class="sourceLineNo">132</span> skipEmptyColumns = context.getConfiguration().getBoolean(ImportTsv.SKIP_EMPTY_COLUMNS, false);<a name="line.132"></a>
<span class="sourceLineNo">133</span> skipBadLines = context.getConfiguration().getBoolean(ImportTsv.SKIP_LINES_CONF_KEY, true);<a name="line.133"></a>
<span class="sourceLineNo">134</span> badLineCount = context.getCounter("ImportTsv", "Bad Lines");<a name="line.134"></a>
<span class="sourceLineNo">135</span> logBadLines = context.getConfiguration().getBoolean(ImportTsv.LOG_BAD_LINES_CONF_KEY, false);<a name="line.135"></a>
<span class="sourceLineNo">136</span> hfileOutPath = conf.get(ImportTsv.BULK_OUTPUT_CONF_KEY);<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> /**<a name="line.139"></a>
<span class="sourceLineNo">140</span> * Convert a line of TSV text into an HBase table row.<a name="line.140"></a>
<span class="sourceLineNo">141</span> */<a name="line.141"></a>
<span class="sourceLineNo">142</span> @Override<a name="line.142"></a>
<span class="sourceLineNo">143</span> public void map(LongWritable offset, Text value, Context context) throws IOException {<a name="line.143"></a>
<span class="sourceLineNo">144</span> byte[] lineBytes = value.getBytes();<a name="line.144"></a>
<span class="sourceLineNo">145</span><a name="line.145"></a>
<span class="sourceLineNo">146</span> try {<a name="line.146"></a>
<span class="sourceLineNo">147</span> ImportTsv.TsvParser.ParsedLine parsed = parser.parse(lineBytes, value.getLength());<a name="line.147"></a>
<span class="sourceLineNo">148</span> ImmutableBytesWritable rowKey =<a name="line.148"></a>
<span class="sourceLineNo">149</span> new ImmutableBytesWritable(lineBytes, parsed.getRowKeyOffset(), parsed.getRowKeyLength());<a name="line.149"></a>
<span class="sourceLineNo">150</span> // Retrieve timestamp if exists<a name="line.150"></a>
<span class="sourceLineNo">151</span> ts = parsed.getTimestamp(ts);<a name="line.151"></a>
<span class="sourceLineNo">152</span> cellVisibilityExpr = parsed.getCellVisibility();<a name="line.152"></a>
<span class="sourceLineNo">153</span> ttl = parsed.getCellTTL();<a name="line.153"></a>
<span class="sourceLineNo">154</span><a name="line.154"></a>
<span class="sourceLineNo">155</span> // create tags for the parsed line<a name="line.155"></a>
<span class="sourceLineNo">156</span> if (hfileOutPath != null) {<a name="line.156"></a>
<span class="sourceLineNo">157</span> tags.clear();<a name="line.157"></a>
<span class="sourceLineNo">158</span> if (cellVisibilityExpr != null) {<a name="line.158"></a>
<span class="sourceLineNo">159</span> tags.addAll(kvCreator.getVisibilityExpressionResolver()<a name="line.159"></a>
<span class="sourceLineNo">160</span> .createVisibilityExpTags(cellVisibilityExpr));<a name="line.160"></a>
<span class="sourceLineNo">161</span> }<a name="line.161"></a>
<span class="sourceLineNo">162</span> // Add TTL directly to the KV so we can vary them when packing more than one KV<a name="line.162"></a>
<span class="sourceLineNo">163</span> // into puts<a name="line.163"></a>
<span class="sourceLineNo">164</span> if (ttl &gt; 0) {<a name="line.164"></a>
<span class="sourceLineNo">165</span> tags.add(new ArrayBackedTag(TagType.TTL_TAG_TYPE, Bytes.toBytes(ttl)));<a name="line.165"></a>
<span class="sourceLineNo">166</span> }<a name="line.166"></a>
<span class="sourceLineNo">167</span> }<a name="line.167"></a>
<span class="sourceLineNo">168</span> Put put = new Put(rowKey.copyBytes());<a name="line.168"></a>
<span class="sourceLineNo">169</span> for (int i = 0; i &lt; parsed.getColumnCount(); i++) {<a name="line.169"></a>
<span class="sourceLineNo">170</span> if (<a name="line.170"></a>
<span class="sourceLineNo">171</span> i == parser.getRowKeyColumnIndex() || i == parser.getTimestampKeyColumnIndex()<a name="line.171"></a>
<span class="sourceLineNo">172</span> || i == parser.getAttributesKeyColumnIndex()<a name="line.172"></a>
<span class="sourceLineNo">173</span> || i == parser.getCellVisibilityColumnIndex() || i == parser.getCellTTLColumnIndex()<a name="line.173"></a>
<span class="sourceLineNo">174</span> || (skipEmptyColumns &amp;&amp; parsed.getColumnLength(i) == 0)<a name="line.174"></a>
<span class="sourceLineNo">175</span> ) {<a name="line.175"></a>
<span class="sourceLineNo">176</span> continue;<a name="line.176"></a>
<span class="sourceLineNo">177</span> }<a name="line.177"></a>
<span class="sourceLineNo">178</span> populatePut(lineBytes, parsed, put, i);<a name="line.178"></a>
<span class="sourceLineNo">179</span> }<a name="line.179"></a>
<span class="sourceLineNo">180</span> context.write(rowKey, put);<a name="line.180"></a>
<span class="sourceLineNo">181</span> } catch (ImportTsv.TsvParser.BadTsvLineException | IllegalArgumentException<a name="line.181"></a>
<span class="sourceLineNo">182</span> | InvalidLabelException badLine) {<a name="line.182"></a>
<span class="sourceLineNo">183</span> if (logBadLines) {<a name="line.183"></a>
<span class="sourceLineNo">184</span> System.err.println(value);<a name="line.184"></a>
<span class="sourceLineNo">185</span> }<a name="line.185"></a>
<span class="sourceLineNo">186</span> System.err.println("Bad line at offset: " + offset.get() + ":\n" + badLine.getMessage());<a name="line.186"></a>
<span class="sourceLineNo">187</span> if (skipBadLines) {<a name="line.187"></a>
<span class="sourceLineNo">188</span> incrementBadLineCount(1);<a name="line.188"></a>
<span class="sourceLineNo">189</span> return;<a name="line.189"></a>
<span class="sourceLineNo">190</span> }<a name="line.190"></a>
<span class="sourceLineNo">191</span> throw new IOException(badLine);<a name="line.191"></a>
<span class="sourceLineNo">192</span> } catch (InterruptedException e) {<a name="line.192"></a>
<span class="sourceLineNo">193</span> LOG.error("Interrupted while emitting put", e);<a name="line.193"></a>
<span class="sourceLineNo">194</span> Thread.currentThread().interrupt();<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><a name="line.197"></a>
<span class="sourceLineNo">198</span> protected void populatePut(byte[] lineBytes, ImportTsv.TsvParser.ParsedLine parsed, Put put,<a name="line.198"></a>
<span class="sourceLineNo">199</span> int i) throws BadTsvLineException, IOException {<a name="line.199"></a>
<span class="sourceLineNo">200</span> Cell cell = null;<a name="line.200"></a>
<span class="sourceLineNo">201</span> if (hfileOutPath == null) {<a name="line.201"></a>
<span class="sourceLineNo">202</span> cell = new KeyValue(lineBytes, parsed.getRowKeyOffset(), parsed.getRowKeyLength(),<a name="line.202"></a>
<span class="sourceLineNo">203</span> parser.getFamily(i), 0, parser.getFamily(i).length, parser.getQualifier(i), 0,<a name="line.203"></a>
<span class="sourceLineNo">204</span> parser.getQualifier(i).length, ts, KeyValue.Type.Put, lineBytes, parsed.getColumnOffset(i),<a name="line.204"></a>
<span class="sourceLineNo">205</span> parsed.getColumnLength(i));<a name="line.205"></a>
<span class="sourceLineNo">206</span> if (cellVisibilityExpr != null) {<a name="line.206"></a>
<span class="sourceLineNo">207</span> // We won't be validating the expression here. The Visibility CP will do<a name="line.207"></a>
<span class="sourceLineNo">208</span> // the validation<a name="line.208"></a>
<span class="sourceLineNo">209</span> put.setCellVisibility(new CellVisibility(cellVisibilityExpr));<a name="line.209"></a>
<span class="sourceLineNo">210</span> }<a name="line.210"></a>
<span class="sourceLineNo">211</span> if (ttl &gt; 0) {<a name="line.211"></a>
<span class="sourceLineNo">212</span> put.setTTL(ttl);<a name="line.212"></a>
<span class="sourceLineNo">213</span> }<a name="line.213"></a>
<span class="sourceLineNo">214</span> } else {<a name="line.214"></a>
<span class="sourceLineNo">215</span> // Creating the KV which needs to be directly written to HFiles. Using the Facade<a name="line.215"></a>
<span class="sourceLineNo">216</span> // KVCreator for creation of kvs.<a name="line.216"></a>
<span class="sourceLineNo">217</span> cell = this.kvCreator.create(lineBytes, parsed.getRowKeyOffset(), parsed.getRowKeyLength(),<a name="line.217"></a>
<span class="sourceLineNo">218</span> parser.getFamily(i), 0, parser.getFamily(i).length, parser.getQualifier(i), 0,<a name="line.218"></a>
<span class="sourceLineNo">219</span> parser.getQualifier(i).length, ts, lineBytes, parsed.getColumnOffset(i),<a name="line.219"></a>
<span class="sourceLineNo">220</span> parsed.getColumnLength(i), tags);<a name="line.220"></a>
<span class="sourceLineNo">221</span> }<a name="line.221"></a>
<span class="sourceLineNo">222</span> put.add(cell);<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>
</pre>
</div>
</body>
</html>