blob: 83594490fd3d9035063e14d1c3bb29dab3c265bd [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.Base64;<a name="line.21"></a>
<span class="sourceLineNo">022</span><a name="line.22"></a>
<span class="sourceLineNo">023</span>import org.apache.hadoop.io.LongWritable;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import org.apache.hadoop.io.Text;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.util.Pair;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import org.apache.hadoop.mapreduce.Mapper;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import org.apache.hadoop.mapreduce.Counter;<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>import org.apache.hadoop.conf.Configuration;<a name="line.30"></a>
<span class="sourceLineNo">031</span><a name="line.31"></a>
<span class="sourceLineNo">032</span>/**<a name="line.32"></a>
<span class="sourceLineNo">033</span> * Write table content out to map output files.<a name="line.33"></a>
<span class="sourceLineNo">034</span> */<a name="line.34"></a>
<span class="sourceLineNo">035</span>@InterfaceAudience.Public<a name="line.35"></a>
<span class="sourceLineNo">036</span>public class TsvImporterTextMapper<a name="line.36"></a>
<span class="sourceLineNo">037</span>extends Mapper&lt;LongWritable, Text, ImmutableBytesWritable, Text&gt;<a name="line.37"></a>
<span class="sourceLineNo">038</span>{<a name="line.38"></a>
<span class="sourceLineNo">039</span><a name="line.39"></a>
<span class="sourceLineNo">040</span> /** Column seperator */<a name="line.40"></a>
<span class="sourceLineNo">041</span> private String separator;<a name="line.41"></a>
<span class="sourceLineNo">042</span><a name="line.42"></a>
<span class="sourceLineNo">043</span> /** Should skip bad lines */<a name="line.43"></a>
<span class="sourceLineNo">044</span> private boolean skipBadLines;<a name="line.44"></a>
<span class="sourceLineNo">045</span> private Counter badLineCount;<a name="line.45"></a>
<span class="sourceLineNo">046</span> private boolean logBadLines;<a name="line.46"></a>
<span class="sourceLineNo">047</span><a name="line.47"></a>
<span class="sourceLineNo">048</span> private ImportTsv.TsvParser parser;<a name="line.48"></a>
<span class="sourceLineNo">049</span><a name="line.49"></a>
<span class="sourceLineNo">050</span> public boolean getSkipBadLines() {<a name="line.50"></a>
<span class="sourceLineNo">051</span> return skipBadLines;<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> public Counter getBadLineCount() {<a name="line.54"></a>
<span class="sourceLineNo">055</span> return badLineCount;<a name="line.55"></a>
<span class="sourceLineNo">056</span> }<a name="line.56"></a>
<span class="sourceLineNo">057</span><a name="line.57"></a>
<span class="sourceLineNo">058</span> public void incrementBadLineCount(int count) {<a name="line.58"></a>
<span class="sourceLineNo">059</span> this.badLineCount.increment(count);<a name="line.59"></a>
<span class="sourceLineNo">060</span> }<a name="line.60"></a>
<span class="sourceLineNo">061</span><a name="line.61"></a>
<span class="sourceLineNo">062</span> /**<a name="line.62"></a>
<span class="sourceLineNo">063</span> * Handles initializing this class with objects specific to it (i.e., the parser).<a name="line.63"></a>
<span class="sourceLineNo">064</span> * Common initialization that might be leveraged by a subclass is done in<a name="line.64"></a>
<span class="sourceLineNo">065</span> * &lt;code&gt;doSetup&lt;/code&gt;. Hence a subclass may choose to override this method<a name="line.65"></a>
<span class="sourceLineNo">066</span> * and call &lt;code&gt;doSetup&lt;/code&gt; as well before handling it's own custom params.<a name="line.66"></a>
<span class="sourceLineNo">067</span> *<a name="line.67"></a>
<span class="sourceLineNo">068</span> * @param context<a name="line.68"></a>
<span class="sourceLineNo">069</span> */<a name="line.69"></a>
<span class="sourceLineNo">070</span> @Override<a name="line.70"></a>
<span class="sourceLineNo">071</span> protected void setup(Context context) {<a name="line.71"></a>
<span class="sourceLineNo">072</span> doSetup(context);<a name="line.72"></a>
<span class="sourceLineNo">073</span><a name="line.73"></a>
<span class="sourceLineNo">074</span> Configuration conf = context.getConfiguration();<a name="line.74"></a>
<span class="sourceLineNo">075</span><a name="line.75"></a>
<span class="sourceLineNo">076</span> parser = new ImportTsv.TsvParser(conf.get(ImportTsv.COLUMNS_CONF_KEY), separator);<a name="line.76"></a>
<span class="sourceLineNo">077</span> if (parser.getRowKeyColumnIndex() == -1) {<a name="line.77"></a>
<span class="sourceLineNo">078</span> throw new RuntimeException("No row key column specified");<a name="line.78"></a>
<span class="sourceLineNo">079</span> }<a name="line.79"></a>
<span class="sourceLineNo">080</span> }<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> * Handles common parameter initialization that a subclass might want to leverage.<a name="line.83"></a>
<span class="sourceLineNo">084</span> * @param context<a name="line.84"></a>
<span class="sourceLineNo">085</span> */<a name="line.85"></a>
<span class="sourceLineNo">086</span> protected void doSetup(Context context) {<a name="line.86"></a>
<span class="sourceLineNo">087</span> Configuration conf = context.getConfiguration();<a name="line.87"></a>
<span class="sourceLineNo">088</span><a name="line.88"></a>
<span class="sourceLineNo">089</span> // If a custom separator has been used,<a name="line.89"></a>
<span class="sourceLineNo">090</span> // decode it back from Base64 encoding.<a name="line.90"></a>
<span class="sourceLineNo">091</span> separator = conf.get(ImportTsv.SEPARATOR_CONF_KEY);<a name="line.91"></a>
<span class="sourceLineNo">092</span> if (separator == null) {<a name="line.92"></a>
<span class="sourceLineNo">093</span> separator = ImportTsv.DEFAULT_SEPARATOR;<a name="line.93"></a>
<span class="sourceLineNo">094</span> } else {<a name="line.94"></a>
<span class="sourceLineNo">095</span> separator = new String(Base64.getDecoder().decode(separator));<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> skipBadLines = context.getConfiguration().getBoolean(ImportTsv.SKIP_LINES_CONF_KEY, true);<a name="line.98"></a>
<span class="sourceLineNo">099</span> logBadLines = context.getConfiguration().getBoolean(ImportTsv.LOG_BAD_LINES_CONF_KEY, false);<a name="line.99"></a>
<span class="sourceLineNo">100</span> badLineCount = context.getCounter("ImportTsv", "Bad Lines");<a name="line.100"></a>
<span class="sourceLineNo">101</span> }<a name="line.101"></a>
<span class="sourceLineNo">102</span><a name="line.102"></a>
<span class="sourceLineNo">103</span> /**<a name="line.103"></a>
<span class="sourceLineNo">104</span> * Convert a line of TSV text into an HBase table row.<a name="line.104"></a>
<span class="sourceLineNo">105</span> */<a name="line.105"></a>
<span class="sourceLineNo">106</span> @Override<a name="line.106"></a>
<span class="sourceLineNo">107</span> public void map(LongWritable offset, Text value, Context context) throws IOException {<a name="line.107"></a>
<span class="sourceLineNo">108</span> try {<a name="line.108"></a>
<span class="sourceLineNo">109</span> Pair&lt;Integer,Integer&gt; rowKeyOffests = parser.parseRowKey(value.getBytes(), value.getLength());<a name="line.109"></a>
<span class="sourceLineNo">110</span> ImmutableBytesWritable rowKey = new ImmutableBytesWritable(<a name="line.110"></a>
<span class="sourceLineNo">111</span> value.getBytes(), rowKeyOffests.getFirst(), rowKeyOffests.getSecond());<a name="line.111"></a>
<span class="sourceLineNo">112</span> context.write(rowKey, value);<a name="line.112"></a>
<span class="sourceLineNo">113</span> } catch (ImportTsv.TsvParser.BadTsvLineException|IllegalArgumentException badLine) {<a name="line.113"></a>
<span class="sourceLineNo">114</span> if (logBadLines) {<a name="line.114"></a>
<span class="sourceLineNo">115</span> System.err.println(value);<a name="line.115"></a>
<span class="sourceLineNo">116</span> }<a name="line.116"></a>
<span class="sourceLineNo">117</span> System.err.println("Bad line at offset: " + offset.get() + ":\n" + badLine.getMessage());<a name="line.117"></a>
<span class="sourceLineNo">118</span> if (skipBadLines) {<a name="line.118"></a>
<span class="sourceLineNo">119</span> incrementBadLineCount(1);<a name="line.119"></a>
<span class="sourceLineNo">120</span> return;<a name="line.120"></a>
<span class="sourceLineNo">121</span> }<a name="line.121"></a>
<span class="sourceLineNo">122</span> throw new IOException(badLine);<a name="line.122"></a>
<span class="sourceLineNo">123</span> } catch (InterruptedException e) {<a name="line.123"></a>
<span class="sourceLineNo">124</span> e.printStackTrace();<a name="line.124"></a>
<span class="sourceLineNo">125</span> Thread.currentThread().interrupt();<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>}<a name="line.128"></a>
</pre>
</div>
</body>
</html>