| <!DOCTYPE HTML> |
| <html lang="en"> |
| <head> |
| <!-- Generated by javadoc (17) --> |
| <title>Source code</title> |
| <meta name="viewport" content="width=device-width, initial-scale=1"> |
| <meta name="description" content="source: package: org.apache.hadoop.hbase.mapreduce, class: TsvImporterTextMapper"> |
| <meta name="generator" content="javadoc/SourceToHTMLConverter"> |
| <link rel="stylesheet" type="text/css" href="../../../../../../stylesheet.css" title="Style"> |
| </head> |
| <body class="source-page"> |
| <main role="main"> |
| <div class="source-container"> |
| <pre><span class="source-line-no">001</span><span id="line-1">/*</span> |
| <span class="source-line-no">002</span><span id="line-2"> * Licensed to the Apache Software Foundation (ASF) under one</span> |
| <span class="source-line-no">003</span><span id="line-3"> * or more contributor license agreements. See the NOTICE file</span> |
| <span class="source-line-no">004</span><span id="line-4"> * distributed with this work for additional information</span> |
| <span class="source-line-no">005</span><span id="line-5"> * regarding copyright ownership. The ASF licenses this file</span> |
| <span class="source-line-no">006</span><span id="line-6"> * to you under the Apache License, Version 2.0 (the</span> |
| <span class="source-line-no">007</span><span id="line-7"> * "License"); you may not use this file except in compliance</span> |
| <span class="source-line-no">008</span><span id="line-8"> * with the License. You may obtain a copy of the License at</span> |
| <span class="source-line-no">009</span><span id="line-9"> *</span> |
| <span class="source-line-no">010</span><span id="line-10"> * http://www.apache.org/licenses/LICENSE-2.0</span> |
| <span class="source-line-no">011</span><span id="line-11"> *</span> |
| <span class="source-line-no">012</span><span id="line-12"> * Unless required by applicable law or agreed to in writing, software</span> |
| <span class="source-line-no">013</span><span id="line-13"> * distributed under the License is distributed on an "AS IS" BASIS,</span> |
| <span class="source-line-no">014</span><span id="line-14"> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.</span> |
| <span class="source-line-no">015</span><span id="line-15"> * See the License for the specific language governing permissions and</span> |
| <span class="source-line-no">016</span><span id="line-16"> * limitations under the License.</span> |
| <span class="source-line-no">017</span><span id="line-17"> */</span> |
| <span class="source-line-no">018</span><span id="line-18">package org.apache.hadoop.hbase.mapreduce;</span> |
| <span class="source-line-no">019</span><span id="line-19"></span> |
| <span class="source-line-no">020</span><span id="line-20">import java.io.IOException;</span> |
| <span class="source-line-no">021</span><span id="line-21">import java.util.Base64;</span> |
| <span class="source-line-no">022</span><span id="line-22">import org.apache.hadoop.conf.Configuration;</span> |
| <span class="source-line-no">023</span><span id="line-23">import org.apache.hadoop.hbase.io.ImmutableBytesWritable;</span> |
| <span class="source-line-no">024</span><span id="line-24">import org.apache.hadoop.hbase.util.Pair;</span> |
| <span class="source-line-no">025</span><span id="line-25">import org.apache.hadoop.io.LongWritable;</span> |
| <span class="source-line-no">026</span><span id="line-26">import org.apache.hadoop.io.Text;</span> |
| <span class="source-line-no">027</span><span id="line-27">import org.apache.hadoop.mapreduce.Counter;</span> |
| <span class="source-line-no">028</span><span id="line-28">import org.apache.hadoop.mapreduce.Mapper;</span> |
| <span class="source-line-no">029</span><span id="line-29">import org.apache.yetus.audience.InterfaceAudience;</span> |
| <span class="source-line-no">030</span><span id="line-30">import org.slf4j.Logger;</span> |
| <span class="source-line-no">031</span><span id="line-31">import org.slf4j.LoggerFactory;</span> |
| <span class="source-line-no">032</span><span id="line-32"></span> |
| <span class="source-line-no">033</span><span id="line-33">/**</span> |
| <span class="source-line-no">034</span><span id="line-34"> * Write table content out to map output files.</span> |
| <span class="source-line-no">035</span><span id="line-35"> */</span> |
| <span class="source-line-no">036</span><span id="line-36">@InterfaceAudience.Public</span> |
| <span class="source-line-no">037</span><span id="line-37">public class TsvImporterTextMapper</span> |
| <span class="source-line-no">038</span><span id="line-38"> extends Mapper<LongWritable, Text, ImmutableBytesWritable, Text> {</span> |
| <span class="source-line-no">039</span><span id="line-39"> private static final Logger LOG = LoggerFactory.getLogger(TsvImporterTextMapper.class);</span> |
| <span class="source-line-no">040</span><span id="line-40"></span> |
| <span class="source-line-no">041</span><span id="line-41"> /** Column seperator */</span> |
| <span class="source-line-no">042</span><span id="line-42"> private String separator;</span> |
| <span class="source-line-no">043</span><span id="line-43"></span> |
| <span class="source-line-no">044</span><span id="line-44"> /** Should skip bad lines */</span> |
| <span class="source-line-no">045</span><span id="line-45"> private boolean skipBadLines;</span> |
| <span class="source-line-no">046</span><span id="line-46"> private Counter badLineCount;</span> |
| <span class="source-line-no">047</span><span id="line-47"> private boolean logBadLines;</span> |
| <span class="source-line-no">048</span><span id="line-48"></span> |
| <span class="source-line-no">049</span><span id="line-49"> private ImportTsv.TsvParser parser;</span> |
| <span class="source-line-no">050</span><span id="line-50"></span> |
| <span class="source-line-no">051</span><span id="line-51"> public boolean getSkipBadLines() {</span> |
| <span class="source-line-no">052</span><span id="line-52"> return skipBadLines;</span> |
| <span class="source-line-no">053</span><span id="line-53"> }</span> |
| <span class="source-line-no">054</span><span id="line-54"></span> |
| <span class="source-line-no">055</span><span id="line-55"> public Counter getBadLineCount() {</span> |
| <span class="source-line-no">056</span><span id="line-56"> return badLineCount;</span> |
| <span class="source-line-no">057</span><span id="line-57"> }</span> |
| <span class="source-line-no">058</span><span id="line-58"></span> |
| <span class="source-line-no">059</span><span id="line-59"> public void incrementBadLineCount(int count) {</span> |
| <span class="source-line-no">060</span><span id="line-60"> this.badLineCount.increment(count);</span> |
| <span class="source-line-no">061</span><span id="line-61"> }</span> |
| <span class="source-line-no">062</span><span id="line-62"></span> |
| <span class="source-line-no">063</span><span id="line-63"> /**</span> |
| <span class="source-line-no">064</span><span id="line-64"> * Handles initializing this class with objects specific to it (i.e., the parser). Common</span> |
| <span class="source-line-no">065</span><span id="line-65"> * initialization that might be leveraged by a subclass is done in <code>doSetup</code>. Hence a</span> |
| <span class="source-line-no">066</span><span id="line-66"> * subclass may choose to override this method and call <code>doSetup</code> as well before</span> |
| <span class="source-line-no">067</span><span id="line-67"> * handling it's own custom params.</span> |
| <span class="source-line-no">068</span><span id="line-68"> */</span> |
| <span class="source-line-no">069</span><span id="line-69"> @Override</span> |
| <span class="source-line-no">070</span><span id="line-70"> protected void setup(Context context) {</span> |
| <span class="source-line-no">071</span><span id="line-71"> doSetup(context);</span> |
| <span class="source-line-no">072</span><span id="line-72"></span> |
| <span class="source-line-no">073</span><span id="line-73"> Configuration conf = context.getConfiguration();</span> |
| <span class="source-line-no">074</span><span id="line-74"></span> |
| <span class="source-line-no">075</span><span id="line-75"> parser = new ImportTsv.TsvParser(conf.get(ImportTsv.COLUMNS_CONF_KEY), separator);</span> |
| <span class="source-line-no">076</span><span id="line-76"> if (parser.getRowKeyColumnIndex() == -1) {</span> |
| <span class="source-line-no">077</span><span id="line-77"> throw new RuntimeException("No row key column specified");</span> |
| <span class="source-line-no">078</span><span id="line-78"> }</span> |
| <span class="source-line-no">079</span><span id="line-79"> }</span> |
| <span class="source-line-no">080</span><span id="line-80"></span> |
| <span class="source-line-no">081</span><span id="line-81"> /**</span> |
| <span class="source-line-no">082</span><span id="line-82"> * Handles common parameter initialization that a subclass might want to leverage.</span> |
| <span class="source-line-no">083</span><span id="line-83"> */</span> |
| <span class="source-line-no">084</span><span id="line-84"> protected void doSetup(Context context) {</span> |
| <span class="source-line-no">085</span><span id="line-85"> Configuration conf = context.getConfiguration();</span> |
| <span class="source-line-no">086</span><span id="line-86"></span> |
| <span class="source-line-no">087</span><span id="line-87"> // If a custom separator has been used,</span> |
| <span class="source-line-no">088</span><span id="line-88"> // decode it back from Base64 encoding.</span> |
| <span class="source-line-no">089</span><span id="line-89"> separator = conf.get(ImportTsv.SEPARATOR_CONF_KEY);</span> |
| <span class="source-line-no">090</span><span id="line-90"> if (separator == null) {</span> |
| <span class="source-line-no">091</span><span id="line-91"> separator = ImportTsv.DEFAULT_SEPARATOR;</span> |
| <span class="source-line-no">092</span><span id="line-92"> } else {</span> |
| <span class="source-line-no">093</span><span id="line-93"> separator = new String(Base64.getDecoder().decode(separator));</span> |
| <span class="source-line-no">094</span><span id="line-94"> }</span> |
| <span class="source-line-no">095</span><span id="line-95"></span> |
| <span class="source-line-no">096</span><span id="line-96"> skipBadLines = context.getConfiguration().getBoolean(ImportTsv.SKIP_LINES_CONF_KEY, true);</span> |
| <span class="source-line-no">097</span><span id="line-97"> logBadLines = context.getConfiguration().getBoolean(ImportTsv.LOG_BAD_LINES_CONF_KEY, false);</span> |
| <span class="source-line-no">098</span><span id="line-98"> badLineCount = context.getCounter("ImportTsv", "Bad Lines");</span> |
| <span class="source-line-no">099</span><span id="line-99"> }</span> |
| <span class="source-line-no">100</span><span id="line-100"></span> |
| <span class="source-line-no">101</span><span id="line-101"> /**</span> |
| <span class="source-line-no">102</span><span id="line-102"> * Convert a line of TSV text into an HBase table row.</span> |
| <span class="source-line-no">103</span><span id="line-103"> */</span> |
| <span class="source-line-no">104</span><span id="line-104"> @Override</span> |
| <span class="source-line-no">105</span><span id="line-105"> public void map(LongWritable offset, Text value, Context context) throws IOException {</span> |
| <span class="source-line-no">106</span><span id="line-106"> try {</span> |
| <span class="source-line-no">107</span><span id="line-107"> Pair<Integer, Integer> rowKeyOffests =</span> |
| <span class="source-line-no">108</span><span id="line-108"> parser.parseRowKey(value.getBytes(), value.getLength());</span> |
| <span class="source-line-no">109</span><span id="line-109"> ImmutableBytesWritable rowKey = new ImmutableBytesWritable(value.getBytes(),</span> |
| <span class="source-line-no">110</span><span id="line-110"> rowKeyOffests.getFirst(), rowKeyOffests.getSecond());</span> |
| <span class="source-line-no">111</span><span id="line-111"> context.write(rowKey, value);</span> |
| <span class="source-line-no">112</span><span id="line-112"> } catch (ImportTsv.TsvParser.BadTsvLineException | IllegalArgumentException badLine) {</span> |
| <span class="source-line-no">113</span><span id="line-113"> if (logBadLines) {</span> |
| <span class="source-line-no">114</span><span id="line-114"> System.err.println(value);</span> |
| <span class="source-line-no">115</span><span id="line-115"> }</span> |
| <span class="source-line-no">116</span><span id="line-116"> System.err.println("Bad line at offset: " + offset.get() + ":\n" + badLine.getMessage());</span> |
| <span class="source-line-no">117</span><span id="line-117"> if (skipBadLines) {</span> |
| <span class="source-line-no">118</span><span id="line-118"> incrementBadLineCount(1);</span> |
| <span class="source-line-no">119</span><span id="line-119"> return;</span> |
| <span class="source-line-no">120</span><span id="line-120"> }</span> |
| <span class="source-line-no">121</span><span id="line-121"> throw new IOException(badLine);</span> |
| <span class="source-line-no">122</span><span id="line-122"> } catch (InterruptedException e) {</span> |
| <span class="source-line-no">123</span><span id="line-123"> LOG.error("Interrupted while emitting TSV text", e);</span> |
| <span class="source-line-no">124</span><span id="line-124"> Thread.currentThread().interrupt();</span> |
| <span class="source-line-no">125</span><span id="line-125"> }</span> |
| <span class="source-line-no">126</span><span id="line-126"> }</span> |
| <span class="source-line-no">127</span><span id="line-127">}</span> |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| </pre> |
| </div> |
| </main> |
| </body> |
| </html> |