blob: 4305733c61d86e7c5dc83babf325de4046d54017 [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 static java.lang.String.format;<a name="line.20"></a>
<span class="sourceLineNo">021</span><a name="line.21"></a>
<span class="sourceLineNo">022</span>import java.io.File;<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.util.ArrayList;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import java.util.Base64;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import java.util.HashSet;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import java.util.Set;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import org.apache.commons.lang3.StringUtils;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.apache.hadoop.conf.Configuration;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.apache.hadoop.conf.Configured;<a name="line.30"></a>
<span class="sourceLineNo">031</span>import org.apache.hadoop.fs.Path;<a name="line.31"></a>
<span class="sourceLineNo">032</span>import org.apache.hadoop.hbase.HBaseConfiguration;<a name="line.32"></a>
<span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.HConstants;<a name="line.33"></a>
<span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.TableName;<a name="line.34"></a>
<span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.TableNotEnabledException;<a name="line.35"></a>
<span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.TableNotFoundException;<a name="line.36"></a>
<span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.client.Admin;<a name="line.37"></a>
<span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptor;<a name="line.38"></a>
<span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.client.ColumnFamilyDescriptorBuilder;<a name="line.39"></a>
<span class="sourceLineNo">040</span>import org.apache.hadoop.hbase.client.Connection;<a name="line.40"></a>
<span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.client.ConnectionFactory;<a name="line.41"></a>
<span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.client.Put;<a name="line.42"></a>
<span class="sourceLineNo">043</span>import org.apache.hadoop.hbase.client.RegionLocator;<a name="line.43"></a>
<span class="sourceLineNo">044</span>import org.apache.hadoop.hbase.client.Table;<a name="line.44"></a>
<span class="sourceLineNo">045</span>import org.apache.hadoop.hbase.client.TableDescriptor;<a name="line.45"></a>
<span class="sourceLineNo">046</span>import org.apache.hadoop.hbase.client.TableDescriptorBuilder;<a name="line.46"></a>
<span class="sourceLineNo">047</span>import org.apache.hadoop.hbase.io.ImmutableBytesWritable;<a name="line.47"></a>
<span class="sourceLineNo">048</span>import org.apache.hadoop.hbase.regionserver.NoSuchColumnFamilyException;<a name="line.48"></a>
<span class="sourceLineNo">049</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.49"></a>
<span class="sourceLineNo">050</span>import org.apache.hadoop.hbase.util.EnvironmentEdgeManager;<a name="line.50"></a>
<span class="sourceLineNo">051</span>import org.apache.hadoop.hbase.util.Pair;<a name="line.51"></a>
<span class="sourceLineNo">052</span>import org.apache.hadoop.io.Text;<a name="line.52"></a>
<span class="sourceLineNo">053</span>import org.apache.hadoop.mapreduce.Job;<a name="line.53"></a>
<span class="sourceLineNo">054</span>import org.apache.hadoop.mapreduce.lib.input.FileInputFormat;<a name="line.54"></a>
<span class="sourceLineNo">055</span>import org.apache.hadoop.mapreduce.lib.input.TextInputFormat;<a name="line.55"></a>
<span class="sourceLineNo">056</span>import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;<a name="line.56"></a>
<span class="sourceLineNo">057</span>import org.apache.hadoop.mapreduce.lib.output.NullOutputFormat;<a name="line.57"></a>
<span class="sourceLineNo">058</span>import org.apache.hadoop.security.Credentials;<a name="line.58"></a>
<span class="sourceLineNo">059</span>import org.apache.hadoop.util.Tool;<a name="line.59"></a>
<span class="sourceLineNo">060</span>import org.apache.hadoop.util.ToolRunner;<a name="line.60"></a>
<span class="sourceLineNo">061</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.61"></a>
<span class="sourceLineNo">062</span>import org.slf4j.Logger;<a name="line.62"></a>
<span class="sourceLineNo">063</span>import org.slf4j.LoggerFactory;<a name="line.63"></a>
<span class="sourceLineNo">064</span><a name="line.64"></a>
<span class="sourceLineNo">065</span>import org.apache.hbase.thirdparty.com.google.common.base.Preconditions;<a name="line.65"></a>
<span class="sourceLineNo">066</span>import org.apache.hbase.thirdparty.com.google.common.base.Splitter;<a name="line.66"></a>
<span class="sourceLineNo">067</span>import org.apache.hbase.thirdparty.com.google.common.collect.Lists;<a name="line.67"></a>
<span class="sourceLineNo">068</span><a name="line.68"></a>
<span class="sourceLineNo">069</span>/**<a name="line.69"></a>
<span class="sourceLineNo">070</span> * Tool to import data from a TSV file. This tool is rather simplistic - it doesn't do any quoting<a name="line.70"></a>
<span class="sourceLineNo">071</span> * or escaping, but is useful for many data loads.<a name="line.71"></a>
<span class="sourceLineNo">072</span> * @see ImportTsv#usage(String)<a name="line.72"></a>
<span class="sourceLineNo">073</span> */<a name="line.73"></a>
<span class="sourceLineNo">074</span>@InterfaceAudience.Public<a name="line.74"></a>
<span class="sourceLineNo">075</span>public class ImportTsv extends Configured implements Tool {<a name="line.75"></a>
<span class="sourceLineNo">076</span><a name="line.76"></a>
<span class="sourceLineNo">077</span> protected static final Logger LOG = LoggerFactory.getLogger(ImportTsv.class);<a name="line.77"></a>
<span class="sourceLineNo">078</span><a name="line.78"></a>
<span class="sourceLineNo">079</span> final static String NAME = "importtsv";<a name="line.79"></a>
<span class="sourceLineNo">080</span><a name="line.80"></a>
<span class="sourceLineNo">081</span> public final static String MAPPER_CONF_KEY = "importtsv.mapper.class";<a name="line.81"></a>
<span class="sourceLineNo">082</span> public final static String BULK_OUTPUT_CONF_KEY = "importtsv.bulk.output";<a name="line.82"></a>
<span class="sourceLineNo">083</span> public final static String TIMESTAMP_CONF_KEY = "importtsv.timestamp";<a name="line.83"></a>
<span class="sourceLineNo">084</span> public final static String JOB_NAME_CONF_KEY = "mapreduce.job.name";<a name="line.84"></a>
<span class="sourceLineNo">085</span> // TODO: the rest of these configs are used exclusively by TsvImporterMapper.<a name="line.85"></a>
<span class="sourceLineNo">086</span> // Move them out of the tool and let the mapper handle its own validation.<a name="line.86"></a>
<span class="sourceLineNo">087</span> public final static String DRY_RUN_CONF_KEY = "importtsv.dry.run";<a name="line.87"></a>
<span class="sourceLineNo">088</span> // If true, bad lines are logged to stderr. Default: false.<a name="line.88"></a>
<span class="sourceLineNo">089</span> public final static String LOG_BAD_LINES_CONF_KEY = "importtsv.log.bad.lines";<a name="line.89"></a>
<span class="sourceLineNo">090</span> public final static String SKIP_LINES_CONF_KEY = "importtsv.skip.bad.lines";<a name="line.90"></a>
<span class="sourceLineNo">091</span> public final static String SKIP_EMPTY_COLUMNS = "importtsv.skip.empty.columns";<a name="line.91"></a>
<span class="sourceLineNo">092</span> public final static String COLUMNS_CONF_KEY = "importtsv.columns";<a name="line.92"></a>
<span class="sourceLineNo">093</span> public final static String SEPARATOR_CONF_KEY = "importtsv.separator";<a name="line.93"></a>
<span class="sourceLineNo">094</span> public final static String ATTRIBUTE_SEPERATOR_CONF_KEY = "attributes.seperator";<a name="line.94"></a>
<span class="sourceLineNo">095</span> // This config is used to propagate credentials from parent MR jobs which launch<a name="line.95"></a>
<span class="sourceLineNo">096</span> // ImportTSV jobs. SEE IntegrationTestImportTsv.<a name="line.96"></a>
<span class="sourceLineNo">097</span> public final static String CREDENTIALS_LOCATION = "credentials_location";<a name="line.97"></a>
<span class="sourceLineNo">098</span> final static String DEFAULT_SEPARATOR = "\t";<a name="line.98"></a>
<span class="sourceLineNo">099</span> final static String DEFAULT_ATTRIBUTES_SEPERATOR = "=&gt;";<a name="line.99"></a>
<span class="sourceLineNo">100</span> final static String DEFAULT_MULTIPLE_ATTRIBUTES_SEPERATOR = ",";<a name="line.100"></a>
<span class="sourceLineNo">101</span> final static Class DEFAULT_MAPPER = TsvImporterMapper.class;<a name="line.101"></a>
<span class="sourceLineNo">102</span> public final static String CREATE_TABLE_CONF_KEY = "create.table";<a name="line.102"></a>
<span class="sourceLineNo">103</span> public final static String NO_STRICT_COL_FAMILY = "no.strict";<a name="line.103"></a>
<span class="sourceLineNo">104</span> /**<a name="line.104"></a>
<span class="sourceLineNo">105</span> * If table didn't exist and was created in dry-run mode, this flag is flipped to delete it when<a name="line.105"></a>
<span class="sourceLineNo">106</span> * MR ends.<a name="line.106"></a>
<span class="sourceLineNo">107</span> */<a name="line.107"></a>
<span class="sourceLineNo">108</span> private static boolean DRY_RUN_TABLE_CREATED;<a name="line.108"></a>
<span class="sourceLineNo">109</span><a name="line.109"></a>
<span class="sourceLineNo">110</span> public static class TsvParser {<a name="line.110"></a>
<span class="sourceLineNo">111</span> /**<a name="line.111"></a>
<span class="sourceLineNo">112</span> * Column families and qualifiers mapped to the TSV columns<a name="line.112"></a>
<span class="sourceLineNo">113</span> */<a name="line.113"></a>
<span class="sourceLineNo">114</span> private final byte[][] families;<a name="line.114"></a>
<span class="sourceLineNo">115</span> private final byte[][] qualifiers;<a name="line.115"></a>
<span class="sourceLineNo">116</span><a name="line.116"></a>
<span class="sourceLineNo">117</span> private final byte separatorByte;<a name="line.117"></a>
<span class="sourceLineNo">118</span><a name="line.118"></a>
<span class="sourceLineNo">119</span> private int rowKeyColumnIndex;<a name="line.119"></a>
<span class="sourceLineNo">120</span><a name="line.120"></a>
<span class="sourceLineNo">121</span> private int maxColumnCount;<a name="line.121"></a>
<span class="sourceLineNo">122</span><a name="line.122"></a>
<span class="sourceLineNo">123</span> // Default value must be negative<a name="line.123"></a>
<span class="sourceLineNo">124</span> public static final int DEFAULT_TIMESTAMP_COLUMN_INDEX = -1;<a name="line.124"></a>
<span class="sourceLineNo">125</span><a name="line.125"></a>
<span class="sourceLineNo">126</span> private int timestampKeyColumnIndex = DEFAULT_TIMESTAMP_COLUMN_INDEX;<a name="line.126"></a>
<span class="sourceLineNo">127</span><a name="line.127"></a>
<span class="sourceLineNo">128</span> public static final String ROWKEY_COLUMN_SPEC = "HBASE_ROW_KEY";<a name="line.128"></a>
<span class="sourceLineNo">129</span><a name="line.129"></a>
<span class="sourceLineNo">130</span> public static final String TIMESTAMPKEY_COLUMN_SPEC = "HBASE_TS_KEY";<a name="line.130"></a>
<span class="sourceLineNo">131</span><a name="line.131"></a>
<span class="sourceLineNo">132</span> public static final String ATTRIBUTES_COLUMN_SPEC = "HBASE_ATTRIBUTES_KEY";<a name="line.132"></a>
<span class="sourceLineNo">133</span><a name="line.133"></a>
<span class="sourceLineNo">134</span> public static final String CELL_VISIBILITY_COLUMN_SPEC = "HBASE_CELL_VISIBILITY";<a name="line.134"></a>
<span class="sourceLineNo">135</span><a name="line.135"></a>
<span class="sourceLineNo">136</span> public static final String CELL_TTL_COLUMN_SPEC = "HBASE_CELL_TTL";<a name="line.136"></a>
<span class="sourceLineNo">137</span><a name="line.137"></a>
<span class="sourceLineNo">138</span> private int attrKeyColumnIndex = DEFAULT_ATTRIBUTES_COLUMN_INDEX;<a name="line.138"></a>
<span class="sourceLineNo">139</span><a name="line.139"></a>
<span class="sourceLineNo">140</span> public static final int DEFAULT_ATTRIBUTES_COLUMN_INDEX = -1;<a name="line.140"></a>
<span class="sourceLineNo">141</span><a name="line.141"></a>
<span class="sourceLineNo">142</span> public static final int DEFAULT_CELL_VISIBILITY_COLUMN_INDEX = -1;<a name="line.142"></a>
<span class="sourceLineNo">143</span><a name="line.143"></a>
<span class="sourceLineNo">144</span> public static final int DEFAULT_CELL_TTL_COLUMN_INDEX = -1;<a name="line.144"></a>
<span class="sourceLineNo">145</span><a name="line.145"></a>
<span class="sourceLineNo">146</span> private int cellVisibilityColumnIndex = DEFAULT_CELL_VISIBILITY_COLUMN_INDEX;<a name="line.146"></a>
<span class="sourceLineNo">147</span><a name="line.147"></a>
<span class="sourceLineNo">148</span> private int cellTTLColumnIndex = DEFAULT_CELL_TTL_COLUMN_INDEX;<a name="line.148"></a>
<span class="sourceLineNo">149</span><a name="line.149"></a>
<span class="sourceLineNo">150</span> /**<a name="line.150"></a>
<span class="sourceLineNo">151</span> * @param columnsSpecification the list of columns to parser out, comma separated. The row key<a name="line.151"></a>
<span class="sourceLineNo">152</span> * should be the special token TsvParser.ROWKEY_COLUMN_SPEC<a name="line.152"></a>
<span class="sourceLineNo">153</span> */<a name="line.153"></a>
<span class="sourceLineNo">154</span> public TsvParser(String columnsSpecification, String separatorStr) {<a name="line.154"></a>
<span class="sourceLineNo">155</span> // Configure separator<a name="line.155"></a>
<span class="sourceLineNo">156</span> byte[] separator = Bytes.toBytes(separatorStr);<a name="line.156"></a>
<span class="sourceLineNo">157</span> Preconditions.checkArgument(separator.length == 1,<a name="line.157"></a>
<span class="sourceLineNo">158</span> "TsvParser only supports single-byte separators");<a name="line.158"></a>
<span class="sourceLineNo">159</span> separatorByte = separator[0];<a name="line.159"></a>
<span class="sourceLineNo">160</span><a name="line.160"></a>
<span class="sourceLineNo">161</span> // Configure columns<a name="line.161"></a>
<span class="sourceLineNo">162</span> ArrayList&lt;String&gt; columnStrings =<a name="line.162"></a>
<span class="sourceLineNo">163</span> Lists.newArrayList(Splitter.on(',').trimResults().split(columnsSpecification));<a name="line.163"></a>
<span class="sourceLineNo">164</span><a name="line.164"></a>
<span class="sourceLineNo">165</span> maxColumnCount = columnStrings.size();<a name="line.165"></a>
<span class="sourceLineNo">166</span> families = new byte[maxColumnCount][];<a name="line.166"></a>
<span class="sourceLineNo">167</span> qualifiers = new byte[maxColumnCount][];<a name="line.167"></a>
<span class="sourceLineNo">168</span><a name="line.168"></a>
<span class="sourceLineNo">169</span> for (int i = 0; i &lt; columnStrings.size(); i++) {<a name="line.169"></a>
<span class="sourceLineNo">170</span> String str = columnStrings.get(i);<a name="line.170"></a>
<span class="sourceLineNo">171</span> if (ROWKEY_COLUMN_SPEC.equals(str)) {<a name="line.171"></a>
<span class="sourceLineNo">172</span> rowKeyColumnIndex = i;<a name="line.172"></a>
<span class="sourceLineNo">173</span> continue;<a name="line.173"></a>
<span class="sourceLineNo">174</span> }<a name="line.174"></a>
<span class="sourceLineNo">175</span> if (TIMESTAMPKEY_COLUMN_SPEC.equals(str)) {<a name="line.175"></a>
<span class="sourceLineNo">176</span> timestampKeyColumnIndex = i;<a name="line.176"></a>
<span class="sourceLineNo">177</span> continue;<a name="line.177"></a>
<span class="sourceLineNo">178</span> }<a name="line.178"></a>
<span class="sourceLineNo">179</span> if (ATTRIBUTES_COLUMN_SPEC.equals(str)) {<a name="line.179"></a>
<span class="sourceLineNo">180</span> attrKeyColumnIndex = i;<a name="line.180"></a>
<span class="sourceLineNo">181</span> continue;<a name="line.181"></a>
<span class="sourceLineNo">182</span> }<a name="line.182"></a>
<span class="sourceLineNo">183</span> if (CELL_VISIBILITY_COLUMN_SPEC.equals(str)) {<a name="line.183"></a>
<span class="sourceLineNo">184</span> cellVisibilityColumnIndex = i;<a name="line.184"></a>
<span class="sourceLineNo">185</span> continue;<a name="line.185"></a>
<span class="sourceLineNo">186</span> }<a name="line.186"></a>
<span class="sourceLineNo">187</span> if (CELL_TTL_COLUMN_SPEC.equals(str)) {<a name="line.187"></a>
<span class="sourceLineNo">188</span> cellTTLColumnIndex = i;<a name="line.188"></a>
<span class="sourceLineNo">189</span> continue;<a name="line.189"></a>
<span class="sourceLineNo">190</span> }<a name="line.190"></a>
<span class="sourceLineNo">191</span> String[] parts = str.split(":", 2);<a name="line.191"></a>
<span class="sourceLineNo">192</span> if (parts.length == 1) {<a name="line.192"></a>
<span class="sourceLineNo">193</span> families[i] = Bytes.toBytes(str);<a name="line.193"></a>
<span class="sourceLineNo">194</span> qualifiers[i] = HConstants.EMPTY_BYTE_ARRAY;<a name="line.194"></a>
<span class="sourceLineNo">195</span> } else {<a name="line.195"></a>
<span class="sourceLineNo">196</span> families[i] = Bytes.toBytes(parts[0]);<a name="line.196"></a>
<span class="sourceLineNo">197</span> qualifiers[i] = Bytes.toBytes(parts[1]);<a name="line.197"></a>
<span class="sourceLineNo">198</span> }<a name="line.198"></a>
<span class="sourceLineNo">199</span> }<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 boolean hasTimestamp() {<a name="line.202"></a>
<span class="sourceLineNo">203</span> return timestampKeyColumnIndex != DEFAULT_TIMESTAMP_COLUMN_INDEX;<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 int getTimestampKeyColumnIndex() {<a name="line.206"></a>
<span class="sourceLineNo">207</span> return timestampKeyColumnIndex;<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> public boolean hasAttributes() {<a name="line.210"></a>
<span class="sourceLineNo">211</span> return attrKeyColumnIndex != DEFAULT_ATTRIBUTES_COLUMN_INDEX;<a name="line.211"></a>
<span class="sourceLineNo">212</span> }<a name="line.212"></a>
<span class="sourceLineNo">213</span><a name="line.213"></a>
<span class="sourceLineNo">214</span> public boolean hasCellVisibility() {<a name="line.214"></a>
<span class="sourceLineNo">215</span> return cellVisibilityColumnIndex != DEFAULT_CELL_VISIBILITY_COLUMN_INDEX;<a name="line.215"></a>
<span class="sourceLineNo">216</span> }<a name="line.216"></a>
<span class="sourceLineNo">217</span><a name="line.217"></a>
<span class="sourceLineNo">218</span> public boolean hasCellTTL() {<a name="line.218"></a>
<span class="sourceLineNo">219</span> return cellTTLColumnIndex != DEFAULT_CELL_VISIBILITY_COLUMN_INDEX;<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>
<span class="sourceLineNo">222</span> public int getAttributesKeyColumnIndex() {<a name="line.222"></a>
<span class="sourceLineNo">223</span> return attrKeyColumnIndex;<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>
<span class="sourceLineNo">226</span> public int getCellVisibilityColumnIndex() {<a name="line.226"></a>
<span class="sourceLineNo">227</span> return cellVisibilityColumnIndex;<a name="line.227"></a>
<span class="sourceLineNo">228</span> }<a name="line.228"></a>
<span class="sourceLineNo">229</span><a name="line.229"></a>
<span class="sourceLineNo">230</span> public int getCellTTLColumnIndex() {<a name="line.230"></a>
<span class="sourceLineNo">231</span> return cellTTLColumnIndex;<a name="line.231"></a>
<span class="sourceLineNo">232</span> }<a name="line.232"></a>
<span class="sourceLineNo">233</span><a name="line.233"></a>
<span class="sourceLineNo">234</span> public int getRowKeyColumnIndex() {<a name="line.234"></a>
<span class="sourceLineNo">235</span> return rowKeyColumnIndex;<a name="line.235"></a>
<span class="sourceLineNo">236</span> }<a name="line.236"></a>
<span class="sourceLineNo">237</span><a name="line.237"></a>
<span class="sourceLineNo">238</span> public byte[] getFamily(int idx) {<a name="line.238"></a>
<span class="sourceLineNo">239</span> return families[idx];<a name="line.239"></a>
<span class="sourceLineNo">240</span> }<a name="line.240"></a>
<span class="sourceLineNo">241</span><a name="line.241"></a>
<span class="sourceLineNo">242</span> public byte[] getQualifier(int idx) {<a name="line.242"></a>
<span class="sourceLineNo">243</span> return qualifiers[idx];<a name="line.243"></a>
<span class="sourceLineNo">244</span> }<a name="line.244"></a>
<span class="sourceLineNo">245</span><a name="line.245"></a>
<span class="sourceLineNo">246</span> public ParsedLine parse(byte[] lineBytes, int length) throws BadTsvLineException {<a name="line.246"></a>
<span class="sourceLineNo">247</span> // Enumerate separator offsets<a name="line.247"></a>
<span class="sourceLineNo">248</span> ArrayList&lt;Integer&gt; tabOffsets = new ArrayList&lt;&gt;(maxColumnCount);<a name="line.248"></a>
<span class="sourceLineNo">249</span> for (int i = 0; i &lt; length; i++) {<a name="line.249"></a>
<span class="sourceLineNo">250</span> if (lineBytes[i] == separatorByte) {<a name="line.250"></a>
<span class="sourceLineNo">251</span> tabOffsets.add(i);<a name="line.251"></a>
<span class="sourceLineNo">252</span> }<a name="line.252"></a>
<span class="sourceLineNo">253</span> }<a name="line.253"></a>
<span class="sourceLineNo">254</span> if (tabOffsets.isEmpty()) {<a name="line.254"></a>
<span class="sourceLineNo">255</span> throw new BadTsvLineException("No delimiter");<a name="line.255"></a>
<span class="sourceLineNo">256</span> }<a name="line.256"></a>
<span class="sourceLineNo">257</span><a name="line.257"></a>
<span class="sourceLineNo">258</span> tabOffsets.add(length);<a name="line.258"></a>
<span class="sourceLineNo">259</span><a name="line.259"></a>
<span class="sourceLineNo">260</span> if (tabOffsets.size() &gt; maxColumnCount) {<a name="line.260"></a>
<span class="sourceLineNo">261</span> throw new BadTsvLineException("Excessive columns");<a name="line.261"></a>
<span class="sourceLineNo">262</span> } else if (tabOffsets.size() &lt;= getRowKeyColumnIndex()) {<a name="line.262"></a>
<span class="sourceLineNo">263</span> throw new BadTsvLineException("No row key");<a name="line.263"></a>
<span class="sourceLineNo">264</span> } else if (hasTimestamp() &amp;&amp; tabOffsets.size() &lt;= getTimestampKeyColumnIndex()) {<a name="line.264"></a>
<span class="sourceLineNo">265</span> throw new BadTsvLineException("No timestamp");<a name="line.265"></a>
<span class="sourceLineNo">266</span> } else if (hasAttributes() &amp;&amp; tabOffsets.size() &lt;= getAttributesKeyColumnIndex()) {<a name="line.266"></a>
<span class="sourceLineNo">267</span> throw new BadTsvLineException("No attributes specified");<a name="line.267"></a>
<span class="sourceLineNo">268</span> } else if (hasCellVisibility() &amp;&amp; tabOffsets.size() &lt;= getCellVisibilityColumnIndex()) {<a name="line.268"></a>
<span class="sourceLineNo">269</span> throw new BadTsvLineException("No cell visibility specified");<a name="line.269"></a>
<span class="sourceLineNo">270</span> } else if (hasCellTTL() &amp;&amp; tabOffsets.size() &lt;= getCellTTLColumnIndex()) {<a name="line.270"></a>
<span class="sourceLineNo">271</span> throw new BadTsvLineException("No cell TTL specified");<a name="line.271"></a>
<span class="sourceLineNo">272</span> }<a name="line.272"></a>
<span class="sourceLineNo">273</span> return new ParsedLine(tabOffsets, lineBytes);<a name="line.273"></a>
<span class="sourceLineNo">274</span> }<a name="line.274"></a>
<span class="sourceLineNo">275</span><a name="line.275"></a>
<span class="sourceLineNo">276</span> class ParsedLine {<a name="line.276"></a>
<span class="sourceLineNo">277</span> private final ArrayList&lt;Integer&gt; tabOffsets;<a name="line.277"></a>
<span class="sourceLineNo">278</span> private byte[] lineBytes;<a name="line.278"></a>
<span class="sourceLineNo">279</span><a name="line.279"></a>
<span class="sourceLineNo">280</span> ParsedLine(ArrayList&lt;Integer&gt; tabOffsets, byte[] lineBytes) {<a name="line.280"></a>
<span class="sourceLineNo">281</span> this.tabOffsets = tabOffsets;<a name="line.281"></a>
<span class="sourceLineNo">282</span> this.lineBytes = lineBytes;<a name="line.282"></a>
<span class="sourceLineNo">283</span> }<a name="line.283"></a>
<span class="sourceLineNo">284</span><a name="line.284"></a>
<span class="sourceLineNo">285</span> public int getRowKeyOffset() {<a name="line.285"></a>
<span class="sourceLineNo">286</span> return getColumnOffset(rowKeyColumnIndex);<a name="line.286"></a>
<span class="sourceLineNo">287</span> }<a name="line.287"></a>
<span class="sourceLineNo">288</span><a name="line.288"></a>
<span class="sourceLineNo">289</span> public int getRowKeyLength() {<a name="line.289"></a>
<span class="sourceLineNo">290</span> return getColumnLength(rowKeyColumnIndex);<a name="line.290"></a>
<span class="sourceLineNo">291</span> }<a name="line.291"></a>
<span class="sourceLineNo">292</span><a name="line.292"></a>
<span class="sourceLineNo">293</span> public long getTimestamp(long ts) throws BadTsvLineException {<a name="line.293"></a>
<span class="sourceLineNo">294</span> // Return ts if HBASE_TS_KEY is not configured in column spec<a name="line.294"></a>
<span class="sourceLineNo">295</span> if (!hasTimestamp()) {<a name="line.295"></a>
<span class="sourceLineNo">296</span> return ts;<a name="line.296"></a>
<span class="sourceLineNo">297</span> }<a name="line.297"></a>
<span class="sourceLineNo">298</span><a name="line.298"></a>
<span class="sourceLineNo">299</span> String timeStampStr = Bytes.toString(lineBytes, getColumnOffset(timestampKeyColumnIndex),<a name="line.299"></a>
<span class="sourceLineNo">300</span> getColumnLength(timestampKeyColumnIndex));<a name="line.300"></a>
<span class="sourceLineNo">301</span> try {<a name="line.301"></a>
<span class="sourceLineNo">302</span> return Long.parseLong(timeStampStr);<a name="line.302"></a>
<span class="sourceLineNo">303</span> } catch (NumberFormatException nfe) {<a name="line.303"></a>
<span class="sourceLineNo">304</span> // treat this record as bad record<a name="line.304"></a>
<span class="sourceLineNo">305</span> throw new BadTsvLineException("Invalid timestamp " + timeStampStr);<a name="line.305"></a>
<span class="sourceLineNo">306</span> }<a name="line.306"></a>
<span class="sourceLineNo">307</span> }<a name="line.307"></a>
<span class="sourceLineNo">308</span><a name="line.308"></a>
<span class="sourceLineNo">309</span> private String getAttributes() {<a name="line.309"></a>
<span class="sourceLineNo">310</span> if (!hasAttributes()) {<a name="line.310"></a>
<span class="sourceLineNo">311</span> return null;<a name="line.311"></a>
<span class="sourceLineNo">312</span> } else {<a name="line.312"></a>
<span class="sourceLineNo">313</span> return Bytes.toString(lineBytes, getColumnOffset(attrKeyColumnIndex),<a name="line.313"></a>
<span class="sourceLineNo">314</span> getColumnLength(attrKeyColumnIndex));<a name="line.314"></a>
<span class="sourceLineNo">315</span> }<a name="line.315"></a>
<span class="sourceLineNo">316</span> }<a name="line.316"></a>
<span class="sourceLineNo">317</span><a name="line.317"></a>
<span class="sourceLineNo">318</span> public String[] getIndividualAttributes() {<a name="line.318"></a>
<span class="sourceLineNo">319</span> String attributes = getAttributes();<a name="line.319"></a>
<span class="sourceLineNo">320</span> if (attributes != null) {<a name="line.320"></a>
<span class="sourceLineNo">321</span> return attributes.split(DEFAULT_MULTIPLE_ATTRIBUTES_SEPERATOR);<a name="line.321"></a>
<span class="sourceLineNo">322</span> } else {<a name="line.322"></a>
<span class="sourceLineNo">323</span> return null;<a name="line.323"></a>
<span class="sourceLineNo">324</span> }<a name="line.324"></a>
<span class="sourceLineNo">325</span> }<a name="line.325"></a>
<span class="sourceLineNo">326</span><a name="line.326"></a>
<span class="sourceLineNo">327</span> public int getAttributeKeyOffset() {<a name="line.327"></a>
<span class="sourceLineNo">328</span> if (hasAttributes()) {<a name="line.328"></a>
<span class="sourceLineNo">329</span> return getColumnOffset(attrKeyColumnIndex);<a name="line.329"></a>
<span class="sourceLineNo">330</span> } else {<a name="line.330"></a>
<span class="sourceLineNo">331</span> return DEFAULT_ATTRIBUTES_COLUMN_INDEX;<a name="line.331"></a>
<span class="sourceLineNo">332</span> }<a name="line.332"></a>
<span class="sourceLineNo">333</span> }<a name="line.333"></a>
<span class="sourceLineNo">334</span><a name="line.334"></a>
<span class="sourceLineNo">335</span> public int getAttributeKeyLength() {<a name="line.335"></a>
<span class="sourceLineNo">336</span> if (hasAttributes()) {<a name="line.336"></a>
<span class="sourceLineNo">337</span> return getColumnLength(attrKeyColumnIndex);<a name="line.337"></a>
<span class="sourceLineNo">338</span> } else {<a name="line.338"></a>
<span class="sourceLineNo">339</span> return DEFAULT_ATTRIBUTES_COLUMN_INDEX;<a name="line.339"></a>
<span class="sourceLineNo">340</span> }<a name="line.340"></a>
<span class="sourceLineNo">341</span> }<a name="line.341"></a>
<span class="sourceLineNo">342</span><a name="line.342"></a>
<span class="sourceLineNo">343</span> public int getCellVisibilityColumnOffset() {<a name="line.343"></a>
<span class="sourceLineNo">344</span> if (hasCellVisibility()) {<a name="line.344"></a>
<span class="sourceLineNo">345</span> return getColumnOffset(cellVisibilityColumnIndex);<a name="line.345"></a>
<span class="sourceLineNo">346</span> } else {<a name="line.346"></a>
<span class="sourceLineNo">347</span> return DEFAULT_CELL_VISIBILITY_COLUMN_INDEX;<a name="line.347"></a>
<span class="sourceLineNo">348</span> }<a name="line.348"></a>
<span class="sourceLineNo">349</span> }<a name="line.349"></a>
<span class="sourceLineNo">350</span><a name="line.350"></a>
<span class="sourceLineNo">351</span> public int getCellVisibilityColumnLength() {<a name="line.351"></a>
<span class="sourceLineNo">352</span> if (hasCellVisibility()) {<a name="line.352"></a>
<span class="sourceLineNo">353</span> return getColumnLength(cellVisibilityColumnIndex);<a name="line.353"></a>
<span class="sourceLineNo">354</span> } else {<a name="line.354"></a>
<span class="sourceLineNo">355</span> return DEFAULT_CELL_VISIBILITY_COLUMN_INDEX;<a name="line.355"></a>
<span class="sourceLineNo">356</span> }<a name="line.356"></a>
<span class="sourceLineNo">357</span> }<a name="line.357"></a>
<span class="sourceLineNo">358</span><a name="line.358"></a>
<span class="sourceLineNo">359</span> public String getCellVisibility() {<a name="line.359"></a>
<span class="sourceLineNo">360</span> if (!hasCellVisibility()) {<a name="line.360"></a>
<span class="sourceLineNo">361</span> return null;<a name="line.361"></a>
<span class="sourceLineNo">362</span> } else {<a name="line.362"></a>
<span class="sourceLineNo">363</span> return Bytes.toString(lineBytes, getColumnOffset(cellVisibilityColumnIndex),<a name="line.363"></a>
<span class="sourceLineNo">364</span> getColumnLength(cellVisibilityColumnIndex));<a name="line.364"></a>
<span class="sourceLineNo">365</span> }<a name="line.365"></a>
<span class="sourceLineNo">366</span> }<a name="line.366"></a>
<span class="sourceLineNo">367</span><a name="line.367"></a>
<span class="sourceLineNo">368</span> public int getCellTTLColumnOffset() {<a name="line.368"></a>
<span class="sourceLineNo">369</span> if (hasCellTTL()) {<a name="line.369"></a>
<span class="sourceLineNo">370</span> return getColumnOffset(cellTTLColumnIndex);<a name="line.370"></a>
<span class="sourceLineNo">371</span> } else {<a name="line.371"></a>
<span class="sourceLineNo">372</span> return DEFAULT_CELL_TTL_COLUMN_INDEX;<a name="line.372"></a>
<span class="sourceLineNo">373</span> }<a name="line.373"></a>
<span class="sourceLineNo">374</span> }<a name="line.374"></a>
<span class="sourceLineNo">375</span><a name="line.375"></a>
<span class="sourceLineNo">376</span> public int getCellTTLColumnLength() {<a name="line.376"></a>
<span class="sourceLineNo">377</span> if (hasCellTTL()) {<a name="line.377"></a>
<span class="sourceLineNo">378</span> return getColumnLength(cellTTLColumnIndex);<a name="line.378"></a>
<span class="sourceLineNo">379</span> } else {<a name="line.379"></a>
<span class="sourceLineNo">380</span> return DEFAULT_CELL_TTL_COLUMN_INDEX;<a name="line.380"></a>
<span class="sourceLineNo">381</span> }<a name="line.381"></a>
<span class="sourceLineNo">382</span> }<a name="line.382"></a>
<span class="sourceLineNo">383</span><a name="line.383"></a>
<span class="sourceLineNo">384</span> public long getCellTTL() {<a name="line.384"></a>
<span class="sourceLineNo">385</span> if (!hasCellTTL()) {<a name="line.385"></a>
<span class="sourceLineNo">386</span> return 0;<a name="line.386"></a>
<span class="sourceLineNo">387</span> } else {<a name="line.387"></a>
<span class="sourceLineNo">388</span> return Bytes.toLong(lineBytes, getColumnOffset(cellTTLColumnIndex),<a name="line.388"></a>
<span class="sourceLineNo">389</span> getColumnLength(cellTTLColumnIndex));<a name="line.389"></a>
<span class="sourceLineNo">390</span> }<a name="line.390"></a>
<span class="sourceLineNo">391</span> }<a name="line.391"></a>
<span class="sourceLineNo">392</span><a name="line.392"></a>
<span class="sourceLineNo">393</span> public int getColumnOffset(int idx) {<a name="line.393"></a>
<span class="sourceLineNo">394</span> if (idx &gt; 0) return tabOffsets.get(idx - 1) + 1;<a name="line.394"></a>
<span class="sourceLineNo">395</span> else return 0;<a name="line.395"></a>
<span class="sourceLineNo">396</span> }<a name="line.396"></a>
<span class="sourceLineNo">397</span><a name="line.397"></a>
<span class="sourceLineNo">398</span> public int getColumnLength(int idx) {<a name="line.398"></a>
<span class="sourceLineNo">399</span> return tabOffsets.get(idx) - getColumnOffset(idx);<a name="line.399"></a>
<span class="sourceLineNo">400</span> }<a name="line.400"></a>
<span class="sourceLineNo">401</span><a name="line.401"></a>
<span class="sourceLineNo">402</span> public int getColumnCount() {<a name="line.402"></a>
<span class="sourceLineNo">403</span> return tabOffsets.size();<a name="line.403"></a>
<span class="sourceLineNo">404</span> }<a name="line.404"></a>
<span class="sourceLineNo">405</span><a name="line.405"></a>
<span class="sourceLineNo">406</span> public byte[] getLineBytes() {<a name="line.406"></a>
<span class="sourceLineNo">407</span> return lineBytes;<a name="line.407"></a>
<span class="sourceLineNo">408</span> }<a name="line.408"></a>
<span class="sourceLineNo">409</span> }<a name="line.409"></a>
<span class="sourceLineNo">410</span><a name="line.410"></a>
<span class="sourceLineNo">411</span> public static class BadTsvLineException extends Exception {<a name="line.411"></a>
<span class="sourceLineNo">412</span> public BadTsvLineException(String err) {<a name="line.412"></a>
<span class="sourceLineNo">413</span> super(err);<a name="line.413"></a>
<span class="sourceLineNo">414</span> }<a name="line.414"></a>
<span class="sourceLineNo">415</span><a name="line.415"></a>
<span class="sourceLineNo">416</span> private static final long serialVersionUID = 1L;<a name="line.416"></a>
<span class="sourceLineNo">417</span> }<a name="line.417"></a>
<span class="sourceLineNo">418</span><a name="line.418"></a>
<span class="sourceLineNo">419</span> /**<a name="line.419"></a>
<span class="sourceLineNo">420</span> * Return starting position and length of row key from the specified line bytes.<a name="line.420"></a>
<span class="sourceLineNo">421</span> * @return Pair of row key offset and length.<a name="line.421"></a>
<span class="sourceLineNo">422</span> */<a name="line.422"></a>
<span class="sourceLineNo">423</span> public Pair&lt;Integer, Integer&gt; parseRowKey(byte[] lineBytes, int length)<a name="line.423"></a>
<span class="sourceLineNo">424</span> throws BadTsvLineException {<a name="line.424"></a>
<span class="sourceLineNo">425</span> int rkColumnIndex = 0;<a name="line.425"></a>
<span class="sourceLineNo">426</span> int startPos = 0, endPos = 0;<a name="line.426"></a>
<span class="sourceLineNo">427</span> for (int i = 0; i &lt;= length; i++) {<a name="line.427"></a>
<span class="sourceLineNo">428</span> if (i == length || lineBytes[i] == separatorByte) {<a name="line.428"></a>
<span class="sourceLineNo">429</span> endPos = i - 1;<a name="line.429"></a>
<span class="sourceLineNo">430</span> if (rkColumnIndex++ == getRowKeyColumnIndex()) {<a name="line.430"></a>
<span class="sourceLineNo">431</span> if ((endPos + 1) == startPos) {<a name="line.431"></a>
<span class="sourceLineNo">432</span> throw new BadTsvLineException("Empty value for ROW KEY.");<a name="line.432"></a>
<span class="sourceLineNo">433</span> }<a name="line.433"></a>
<span class="sourceLineNo">434</span> break;<a name="line.434"></a>
<span class="sourceLineNo">435</span> } else {<a name="line.435"></a>
<span class="sourceLineNo">436</span> startPos = endPos + 2;<a name="line.436"></a>
<span class="sourceLineNo">437</span> }<a name="line.437"></a>
<span class="sourceLineNo">438</span> }<a name="line.438"></a>
<span class="sourceLineNo">439</span> if (i == length) {<a name="line.439"></a>
<span class="sourceLineNo">440</span> throw new BadTsvLineException("Row key does not exist as number of columns in the line"<a name="line.440"></a>
<span class="sourceLineNo">441</span> + " are less than row key position.");<a name="line.441"></a>
<span class="sourceLineNo">442</span> }<a name="line.442"></a>
<span class="sourceLineNo">443</span> }<a name="line.443"></a>
<span class="sourceLineNo">444</span> return new Pair&lt;&gt;(startPos, endPos - startPos + 1);<a name="line.444"></a>
<span class="sourceLineNo">445</span> }<a name="line.445"></a>
<span class="sourceLineNo">446</span> }<a name="line.446"></a>
<span class="sourceLineNo">447</span><a name="line.447"></a>
<span class="sourceLineNo">448</span> /**<a name="line.448"></a>
<span class="sourceLineNo">449</span> * Sets up the actual job.<a name="line.449"></a>
<span class="sourceLineNo">450</span> * @param conf The current configuration.<a name="line.450"></a>
<span class="sourceLineNo">451</span> * @param args The command line parameters.<a name="line.451"></a>
<span class="sourceLineNo">452</span> * @return The newly created job.<a name="line.452"></a>
<span class="sourceLineNo">453</span> * @throws IOException When setting up the job fails.<a name="line.453"></a>
<span class="sourceLineNo">454</span> */<a name="line.454"></a>
<span class="sourceLineNo">455</span> protected static Job createSubmittableJob(Configuration conf, String[] args)<a name="line.455"></a>
<span class="sourceLineNo">456</span> throws IOException, ClassNotFoundException {<a name="line.456"></a>
<span class="sourceLineNo">457</span> Job job = null;<a name="line.457"></a>
<span class="sourceLineNo">458</span> boolean isDryRun = conf.getBoolean(DRY_RUN_CONF_KEY, false);<a name="line.458"></a>
<span class="sourceLineNo">459</span> try (Connection connection = ConnectionFactory.createConnection(conf)) {<a name="line.459"></a>
<span class="sourceLineNo">460</span> try (Admin admin = connection.getAdmin()) {<a name="line.460"></a>
<span class="sourceLineNo">461</span> // Support non-XML supported characters<a name="line.461"></a>
<span class="sourceLineNo">462</span> // by re-encoding the passed separator as a Base64 string.<a name="line.462"></a>
<span class="sourceLineNo">463</span> String actualSeparator = conf.get(SEPARATOR_CONF_KEY);<a name="line.463"></a>
<span class="sourceLineNo">464</span> if (actualSeparator != null) {<a name="line.464"></a>
<span class="sourceLineNo">465</span> conf.set(SEPARATOR_CONF_KEY,<a name="line.465"></a>
<span class="sourceLineNo">466</span> Bytes.toString(Base64.getEncoder().encode(Bytes.toBytes(actualSeparator))));<a name="line.466"></a>
<span class="sourceLineNo">467</span> }<a name="line.467"></a>
<span class="sourceLineNo">468</span><a name="line.468"></a>
<span class="sourceLineNo">469</span> // See if a non-default Mapper was set<a name="line.469"></a>
<span class="sourceLineNo">470</span> String mapperClassName = conf.get(MAPPER_CONF_KEY);<a name="line.470"></a>
<span class="sourceLineNo">471</span> Class mapperClass =<a name="line.471"></a>
<span class="sourceLineNo">472</span> mapperClassName != null ? Class.forName(mapperClassName) : DEFAULT_MAPPER;<a name="line.472"></a>
<span class="sourceLineNo">473</span><a name="line.473"></a>
<span class="sourceLineNo">474</span> TableName tableName = TableName.valueOf(args[0]);<a name="line.474"></a>
<span class="sourceLineNo">475</span> Path inputDir = new Path(args[1]);<a name="line.475"></a>
<span class="sourceLineNo">476</span> String jobName = conf.get(JOB_NAME_CONF_KEY, NAME + "_" + tableName.getNameAsString());<a name="line.476"></a>
<span class="sourceLineNo">477</span> job = Job.getInstance(conf, jobName);<a name="line.477"></a>
<span class="sourceLineNo">478</span> job.setJarByClass(mapperClass);<a name="line.478"></a>
<span class="sourceLineNo">479</span> FileInputFormat.setInputPaths(job, inputDir);<a name="line.479"></a>
<span class="sourceLineNo">480</span> job.setInputFormatClass(TextInputFormat.class);<a name="line.480"></a>
<span class="sourceLineNo">481</span> job.setMapperClass(mapperClass);<a name="line.481"></a>
<span class="sourceLineNo">482</span> job.setMapOutputKeyClass(ImmutableBytesWritable.class);<a name="line.482"></a>
<span class="sourceLineNo">483</span> String hfileOutPath = conf.get(BULK_OUTPUT_CONF_KEY);<a name="line.483"></a>
<span class="sourceLineNo">484</span> String[] columns = conf.getStrings(COLUMNS_CONF_KEY);<a name="line.484"></a>
<span class="sourceLineNo">485</span> if (StringUtils.isNotEmpty(conf.get(CREDENTIALS_LOCATION))) {<a name="line.485"></a>
<span class="sourceLineNo">486</span> String fileLoc = conf.get(CREDENTIALS_LOCATION);<a name="line.486"></a>
<span class="sourceLineNo">487</span> Credentials cred = Credentials.readTokenStorageFile(new File(fileLoc), conf);<a name="line.487"></a>
<span class="sourceLineNo">488</span> job.getCredentials().addAll(cred);<a name="line.488"></a>
<span class="sourceLineNo">489</span> }<a name="line.489"></a>
<span class="sourceLineNo">490</span><a name="line.490"></a>
<span class="sourceLineNo">491</span> if (hfileOutPath != null) {<a name="line.491"></a>
<span class="sourceLineNo">492</span> if (!admin.tableExists(tableName)) {<a name="line.492"></a>
<span class="sourceLineNo">493</span> LOG.warn(format("Table '%s' does not exist.", tableName));<a name="line.493"></a>
<span class="sourceLineNo">494</span> if ("yes".equalsIgnoreCase(conf.get(CREATE_TABLE_CONF_KEY, "yes"))) {<a name="line.494"></a>
<span class="sourceLineNo">495</span> // TODO: this is backwards. Instead of depending on the existence of a table,<a name="line.495"></a>
<span class="sourceLineNo">496</span> // create a sane splits file for HFileOutputFormat based on data sampling.<a name="line.496"></a>
<span class="sourceLineNo">497</span> createTable(admin, tableName, columns);<a name="line.497"></a>
<span class="sourceLineNo">498</span> if (isDryRun) {<a name="line.498"></a>
<span class="sourceLineNo">499</span> LOG.warn("Dry run: Table will be deleted at end of dry run.");<a name="line.499"></a>
<span class="sourceLineNo">500</span> synchronized (ImportTsv.class) {<a name="line.500"></a>
<span class="sourceLineNo">501</span> DRY_RUN_TABLE_CREATED = true;<a name="line.501"></a>
<span class="sourceLineNo">502</span> }<a name="line.502"></a>
<span class="sourceLineNo">503</span> }<a name="line.503"></a>
<span class="sourceLineNo">504</span> } else {<a name="line.504"></a>
<span class="sourceLineNo">505</span> String errorMsg = format("Table '%s' does not exist and '%s' is set to no.",<a name="line.505"></a>
<span class="sourceLineNo">506</span> tableName, CREATE_TABLE_CONF_KEY);<a name="line.506"></a>
<span class="sourceLineNo">507</span> LOG.error(errorMsg);<a name="line.507"></a>
<span class="sourceLineNo">508</span> throw new TableNotFoundException(errorMsg);<a name="line.508"></a>
<span class="sourceLineNo">509</span> }<a name="line.509"></a>
<span class="sourceLineNo">510</span> }<a name="line.510"></a>
<span class="sourceLineNo">511</span> try (Table table = connection.getTable(tableName);<a name="line.511"></a>
<span class="sourceLineNo">512</span> RegionLocator regionLocator = connection.getRegionLocator(tableName)) {<a name="line.512"></a>
<span class="sourceLineNo">513</span> boolean noStrict = conf.getBoolean(NO_STRICT_COL_FAMILY, false);<a name="line.513"></a>
<span class="sourceLineNo">514</span> // if no.strict is false then check column family<a name="line.514"></a>
<span class="sourceLineNo">515</span> if (!noStrict) {<a name="line.515"></a>
<span class="sourceLineNo">516</span> ArrayList&lt;String&gt; unmatchedFamilies = new ArrayList&lt;&gt;();<a name="line.516"></a>
<span class="sourceLineNo">517</span> Set&lt;String&gt; cfSet = getColumnFamilies(columns);<a name="line.517"></a>
<span class="sourceLineNo">518</span> TableDescriptor tDesc = table.getDescriptor();<a name="line.518"></a>
<span class="sourceLineNo">519</span> for (String cf : cfSet) {<a name="line.519"></a>
<span class="sourceLineNo">520</span> if (!tDesc.hasColumnFamily(Bytes.toBytes(cf))) {<a name="line.520"></a>
<span class="sourceLineNo">521</span> unmatchedFamilies.add(cf);<a name="line.521"></a>
<span class="sourceLineNo">522</span> }<a name="line.522"></a>
<span class="sourceLineNo">523</span> }<a name="line.523"></a>
<span class="sourceLineNo">524</span> if (unmatchedFamilies.size() &gt; 0) {<a name="line.524"></a>
<span class="sourceLineNo">525</span> ArrayList&lt;String&gt; familyNames = new ArrayList&lt;&gt;();<a name="line.525"></a>
<span class="sourceLineNo">526</span> for (ColumnFamilyDescriptor family : table.getDescriptor().getColumnFamilies()) {<a name="line.526"></a>
<span class="sourceLineNo">527</span> familyNames.add(family.getNameAsString());<a name="line.527"></a>
<span class="sourceLineNo">528</span> }<a name="line.528"></a>
<span class="sourceLineNo">529</span> String msg = "Column Families " + unmatchedFamilies + " specified in "<a name="line.529"></a>
<span class="sourceLineNo">530</span> + COLUMNS_CONF_KEY + " does not match with any of the table " + tableName<a name="line.530"></a>
<span class="sourceLineNo">531</span> + " column families " + familyNames + ".\n"<a name="line.531"></a>
<span class="sourceLineNo">532</span> + "To disable column family check, use -D" + NO_STRICT_COL_FAMILY + "=true.\n";<a name="line.532"></a>
<span class="sourceLineNo">533</span> usage(msg);<a name="line.533"></a>
<span class="sourceLineNo">534</span> System.exit(-1);<a name="line.534"></a>
<span class="sourceLineNo">535</span> }<a name="line.535"></a>
<span class="sourceLineNo">536</span> }<a name="line.536"></a>
<span class="sourceLineNo">537</span> if (mapperClass.equals(TsvImporterTextMapper.class)) {<a name="line.537"></a>
<span class="sourceLineNo">538</span> job.setMapOutputValueClass(Text.class);<a name="line.538"></a>
<span class="sourceLineNo">539</span> job.setReducerClass(TextSortReducer.class);<a name="line.539"></a>
<span class="sourceLineNo">540</span> } else {<a name="line.540"></a>
<span class="sourceLineNo">541</span> job.setMapOutputValueClass(Put.class);<a name="line.541"></a>
<span class="sourceLineNo">542</span> job.setCombinerClass(PutCombiner.class);<a name="line.542"></a>
<span class="sourceLineNo">543</span> job.setReducerClass(PutSortReducer.class);<a name="line.543"></a>
<span class="sourceLineNo">544</span> }<a name="line.544"></a>
<span class="sourceLineNo">545</span> if (!isDryRun) {<a name="line.545"></a>
<span class="sourceLineNo">546</span> Path outputDir = new Path(hfileOutPath);<a name="line.546"></a>
<span class="sourceLineNo">547</span> FileOutputFormat.setOutputPath(job, outputDir);<a name="line.547"></a>
<span class="sourceLineNo">548</span> HFileOutputFormat2.configureIncrementalLoad(job, table.getDescriptor(),<a name="line.548"></a>
<span class="sourceLineNo">549</span> regionLocator);<a name="line.549"></a>
<span class="sourceLineNo">550</span> }<a name="line.550"></a>
<span class="sourceLineNo">551</span> }<a name="line.551"></a>
<span class="sourceLineNo">552</span> } else {<a name="line.552"></a>
<span class="sourceLineNo">553</span> if (!admin.tableExists(tableName)) {<a name="line.553"></a>
<span class="sourceLineNo">554</span> String errorMsg = format("Table '%s' does not exist.", tableName);<a name="line.554"></a>
<span class="sourceLineNo">555</span> LOG.error(errorMsg);<a name="line.555"></a>
<span class="sourceLineNo">556</span> throw new TableNotFoundException(errorMsg);<a name="line.556"></a>
<span class="sourceLineNo">557</span> }<a name="line.557"></a>
<span class="sourceLineNo">558</span> try (Table table = connection.getTable(tableName)) {<a name="line.558"></a>
<span class="sourceLineNo">559</span> ArrayList&lt;String&gt; unmatchedFamilies = new ArrayList&lt;&gt;();<a name="line.559"></a>
<span class="sourceLineNo">560</span> Set&lt;String&gt; cfSet = getColumnFamilies(columns);<a name="line.560"></a>
<span class="sourceLineNo">561</span> TableDescriptor tDesc = table.getDescriptor();<a name="line.561"></a>
<span class="sourceLineNo">562</span> for (String cf : cfSet) {<a name="line.562"></a>
<span class="sourceLineNo">563</span> if (!tDesc.hasColumnFamily(Bytes.toBytes(cf))) {<a name="line.563"></a>
<span class="sourceLineNo">564</span> unmatchedFamilies.add(cf);<a name="line.564"></a>
<span class="sourceLineNo">565</span> }<a name="line.565"></a>
<span class="sourceLineNo">566</span> }<a name="line.566"></a>
<span class="sourceLineNo">567</span> if (unmatchedFamilies.size() &gt; 0) {<a name="line.567"></a>
<span class="sourceLineNo">568</span> String noSuchColumnFamiliesMsg =<a name="line.568"></a>
<span class="sourceLineNo">569</span> format("Column families: %s do not exist.", unmatchedFamilies);<a name="line.569"></a>
<span class="sourceLineNo">570</span> LOG.error(noSuchColumnFamiliesMsg);<a name="line.570"></a>
<span class="sourceLineNo">571</span> throw new NoSuchColumnFamilyException(noSuchColumnFamiliesMsg);<a name="line.571"></a>
<span class="sourceLineNo">572</span> }<a name="line.572"></a>
<span class="sourceLineNo">573</span> }<a name="line.573"></a>
<span class="sourceLineNo">574</span> if (mapperClass.equals(TsvImporterTextMapper.class)) {<a name="line.574"></a>
<span class="sourceLineNo">575</span> usage(TsvImporterTextMapper.class.toString()<a name="line.575"></a>
<span class="sourceLineNo">576</span> + " should not be used for non bulkloading case. use "<a name="line.576"></a>
<span class="sourceLineNo">577</span> + TsvImporterMapper.class.toString() + " or custom mapper whose value type is Put.");<a name="line.577"></a>
<span class="sourceLineNo">578</span> System.exit(-1);<a name="line.578"></a>
<span class="sourceLineNo">579</span> }<a name="line.579"></a>
<span class="sourceLineNo">580</span> if (!isDryRun) {<a name="line.580"></a>
<span class="sourceLineNo">581</span> // No reducers. Just write straight to table. Call initTableReducerJob<a name="line.581"></a>
<span class="sourceLineNo">582</span> // to set up the TableOutputFormat.<a name="line.582"></a>
<span class="sourceLineNo">583</span> TableMapReduceUtil.initTableReducerJob(tableName.getNameAsString(), null, job);<a name="line.583"></a>
<span class="sourceLineNo">584</span> }<a name="line.584"></a>
<span class="sourceLineNo">585</span> job.setNumReduceTasks(0);<a name="line.585"></a>
<span class="sourceLineNo">586</span> }<a name="line.586"></a>
<span class="sourceLineNo">587</span> if (isDryRun) {<a name="line.587"></a>
<span class="sourceLineNo">588</span> job.setOutputFormatClass(NullOutputFormat.class);<a name="line.588"></a>
<span class="sourceLineNo">589</span> job.getConfiguration().setStrings("io.serializations",<a name="line.589"></a>
<span class="sourceLineNo">590</span> job.getConfiguration().get("io.serializations"), MutationSerialization.class.getName(),<a name="line.590"></a>
<span class="sourceLineNo">591</span> ResultSerialization.class.getName(), CellSerialization.class.getName());<a name="line.591"></a>
<span class="sourceLineNo">592</span> }<a name="line.592"></a>
<span class="sourceLineNo">593</span> TableMapReduceUtil.addDependencyJars(job);<a name="line.593"></a>
<span class="sourceLineNo">594</span> TableMapReduceUtil.addDependencyJarsForClasses(job.getConfiguration(),<a name="line.594"></a>
<span class="sourceLineNo">595</span> org.apache.hbase.thirdparty.com.google.common.base.Function.class /*<a name="line.595"></a>
<span class="sourceLineNo">596</span> * Guava used by<a name="line.596"></a>
<span class="sourceLineNo">597</span> * TsvParser<a name="line.597"></a>
<span class="sourceLineNo">598</span> */);<a name="line.598"></a>
<span class="sourceLineNo">599</span> }<a name="line.599"></a>
<span class="sourceLineNo">600</span> }<a name="line.600"></a>
<span class="sourceLineNo">601</span> return job;<a name="line.601"></a>
<span class="sourceLineNo">602</span> }<a name="line.602"></a>
<span class="sourceLineNo">603</span><a name="line.603"></a>
<span class="sourceLineNo">604</span> private static void createTable(Admin admin, TableName tableName, String[] columns)<a name="line.604"></a>
<span class="sourceLineNo">605</span> throws IOException {<a name="line.605"></a>
<span class="sourceLineNo">606</span> TableDescriptorBuilder builder = TableDescriptorBuilder.newBuilder(tableName);<a name="line.606"></a>
<span class="sourceLineNo">607</span> Set&lt;String&gt; cfSet = getColumnFamilies(columns);<a name="line.607"></a>
<span class="sourceLineNo">608</span> for (String cf : cfSet) {<a name="line.608"></a>
<span class="sourceLineNo">609</span> builder.setColumnFamily(ColumnFamilyDescriptorBuilder.of(cf));<a name="line.609"></a>
<span class="sourceLineNo">610</span> }<a name="line.610"></a>
<span class="sourceLineNo">611</span> LOG.warn(<a name="line.611"></a>
<span class="sourceLineNo">612</span> format("Creating table '%s' with '%s' columns and default descriptors.", tableName, cfSet));<a name="line.612"></a>
<span class="sourceLineNo">613</span> admin.createTable(builder.build());<a name="line.613"></a>
<span class="sourceLineNo">614</span> }<a name="line.614"></a>
<span class="sourceLineNo">615</span><a name="line.615"></a>
<span class="sourceLineNo">616</span> private static void deleteTable(Configuration conf, String[] args) {<a name="line.616"></a>
<span class="sourceLineNo">617</span> TableName tableName = TableName.valueOf(args[0]);<a name="line.617"></a>
<span class="sourceLineNo">618</span> try (Connection connection = ConnectionFactory.createConnection(conf);<a name="line.618"></a>
<span class="sourceLineNo">619</span> Admin admin = connection.getAdmin()) {<a name="line.619"></a>
<span class="sourceLineNo">620</span> try {<a name="line.620"></a>
<span class="sourceLineNo">621</span> admin.disableTable(tableName);<a name="line.621"></a>
<span class="sourceLineNo">622</span> } catch (TableNotEnabledException e) {<a name="line.622"></a>
<span class="sourceLineNo">623</span> LOG.debug("Dry mode: Table: " + tableName + " already disabled, so just deleting it.");<a name="line.623"></a>
<span class="sourceLineNo">624</span> }<a name="line.624"></a>
<span class="sourceLineNo">625</span> admin.deleteTable(tableName);<a name="line.625"></a>
<span class="sourceLineNo">626</span> } catch (IOException e) {<a name="line.626"></a>
<span class="sourceLineNo">627</span> LOG.error(format("***Dry run: Failed to delete table '%s'.***%n%s", tableName, e.toString()));<a name="line.627"></a>
<span class="sourceLineNo">628</span> return;<a name="line.628"></a>
<span class="sourceLineNo">629</span> }<a name="line.629"></a>
<span class="sourceLineNo">630</span> LOG.info(format("Dry run: Deleted table '%s'.", tableName));<a name="line.630"></a>
<span class="sourceLineNo">631</span> }<a name="line.631"></a>
<span class="sourceLineNo">632</span><a name="line.632"></a>
<span class="sourceLineNo">633</span> private static Set&lt;String&gt; getColumnFamilies(String[] columns) {<a name="line.633"></a>
<span class="sourceLineNo">634</span> Set&lt;String&gt; cfSet = new HashSet&lt;&gt;();<a name="line.634"></a>
<span class="sourceLineNo">635</span> for (String aColumn : columns) {<a name="line.635"></a>
<span class="sourceLineNo">636</span> if (<a name="line.636"></a>
<span class="sourceLineNo">637</span> TsvParser.ROWKEY_COLUMN_SPEC.equals(aColumn)<a name="line.637"></a>
<span class="sourceLineNo">638</span> || TsvParser.TIMESTAMPKEY_COLUMN_SPEC.equals(aColumn)<a name="line.638"></a>
<span class="sourceLineNo">639</span> || TsvParser.CELL_VISIBILITY_COLUMN_SPEC.equals(aColumn)<a name="line.639"></a>
<span class="sourceLineNo">640</span> || TsvParser.CELL_TTL_COLUMN_SPEC.equals(aColumn)<a name="line.640"></a>
<span class="sourceLineNo">641</span> || TsvParser.ATTRIBUTES_COLUMN_SPEC.equals(aColumn)<a name="line.641"></a>
<span class="sourceLineNo">642</span> ) continue;<a name="line.642"></a>
<span class="sourceLineNo">643</span> // we are only concerned with the first one (in case this is a cf:cq)<a name="line.643"></a>
<span class="sourceLineNo">644</span> cfSet.add(aColumn.split(":", 2)[0]);<a name="line.644"></a>
<span class="sourceLineNo">645</span> }<a name="line.645"></a>
<span class="sourceLineNo">646</span> return cfSet;<a name="line.646"></a>
<span class="sourceLineNo">647</span> }<a name="line.647"></a>
<span class="sourceLineNo">648</span><a name="line.648"></a>
<span class="sourceLineNo">649</span> /*<a name="line.649"></a>
<span class="sourceLineNo">650</span> * @param errorMsg Error message. Can be null.<a name="line.650"></a>
<span class="sourceLineNo">651</span> */<a name="line.651"></a>
<span class="sourceLineNo">652</span> private static void usage(final String errorMsg) {<a name="line.652"></a>
<span class="sourceLineNo">653</span> if (errorMsg != null &amp;&amp; errorMsg.length() &gt; 0) {<a name="line.653"></a>
<span class="sourceLineNo">654</span> System.err.println("ERROR: " + errorMsg);<a name="line.654"></a>
<span class="sourceLineNo">655</span> }<a name="line.655"></a>
<span class="sourceLineNo">656</span> String usage = "Usage: " + NAME + " -D" + COLUMNS_CONF_KEY + "=a,b,c &lt;tablename&gt; &lt;inputdir&gt;\n"<a name="line.656"></a>
<span class="sourceLineNo">657</span> + "\n" + "Imports the given input directory of TSV data into the specified table.\n" + "\n"<a name="line.657"></a>
<span class="sourceLineNo">658</span> + "The column names of the TSV data must be specified using the -D" + COLUMNS_CONF_KEY + "\n"<a name="line.658"></a>
<span class="sourceLineNo">659</span> + "option. This option takes the form of comma-separated column names, where each\n"<a name="line.659"></a>
<span class="sourceLineNo">660</span> + "column name is either a simple column family, or a columnfamily:qualifier. The special\n"<a name="line.660"></a>
<span class="sourceLineNo">661</span> + "column name " + TsvParser.ROWKEY_COLUMN_SPEC<a name="line.661"></a>
<span class="sourceLineNo">662</span> + " is used to designate that this column should be used\n"<a name="line.662"></a>
<span class="sourceLineNo">663</span> + "as the row key for each imported record. You must specify exactly one column\n"<a name="line.663"></a>
<span class="sourceLineNo">664</span> + "to be the row key, and you must specify a column name for every column that exists in the\n"<a name="line.664"></a>
<span class="sourceLineNo">665</span> + "input data. Another special column" + TsvParser.TIMESTAMPKEY_COLUMN_SPEC<a name="line.665"></a>
<span class="sourceLineNo">666</span> + " designates that this column should be\n" + "used as timestamp for each record. Unlike "<a name="line.666"></a>
<span class="sourceLineNo">667</span> + TsvParser.ROWKEY_COLUMN_SPEC + ", " + TsvParser.TIMESTAMPKEY_COLUMN_SPEC + " is optional."<a name="line.667"></a>
<span class="sourceLineNo">668</span> + "\n" + "You must specify at most one column as timestamp key for each imported record.\n"<a name="line.668"></a>
<span class="sourceLineNo">669</span> + "Record with invalid timestamps (blank, non-numeric) will be treated as bad record.\n"<a name="line.669"></a>
<span class="sourceLineNo">670</span> + "Note: if you use this option, then '" + TIMESTAMP_CONF_KEY + "' option will be ignored.\n"<a name="line.670"></a>
<span class="sourceLineNo">671</span> + "\n" + "Other special columns that can be specified are " + TsvParser.CELL_TTL_COLUMN_SPEC<a name="line.671"></a>
<span class="sourceLineNo">672</span> + " and " + TsvParser.CELL_VISIBILITY_COLUMN_SPEC + ".\n" + TsvParser.CELL_TTL_COLUMN_SPEC<a name="line.672"></a>
<span class="sourceLineNo">673</span> + " designates that this column will be used " + "as a Cell's Time To Live (TTL) attribute.\n"<a name="line.673"></a>
<span class="sourceLineNo">674</span> + TsvParser.CELL_VISIBILITY_COLUMN_SPEC + " designates that this column contains the "<a name="line.674"></a>
<span class="sourceLineNo">675</span> + "visibility label expression.\n" + "\n" + TsvParser.ATTRIBUTES_COLUMN_SPEC<a name="line.675"></a>
<span class="sourceLineNo">676</span> + " can be used to specify Operation Attributes per record.\n"<a name="line.676"></a>
<span class="sourceLineNo">677</span> + " Should be specified as key=&gt;value where " + TsvParser.DEFAULT_ATTRIBUTES_COLUMN_INDEX<a name="line.677"></a>
<span class="sourceLineNo">678</span> + " is used \n"<a name="line.678"></a>
<span class="sourceLineNo">679</span> + " as the seperator. Note that more than one OperationAttributes can be specified.\n"<a name="line.679"></a>
<span class="sourceLineNo">680</span> + "By default importtsv will load data directly into HBase. To instead generate\n"<a name="line.680"></a>
<span class="sourceLineNo">681</span> + "HFiles of data to prepare for a bulk data load, pass the option:\n" + " -D"<a name="line.681"></a>
<span class="sourceLineNo">682</span> + BULK_OUTPUT_CONF_KEY + "=/path/for/output\n"<a name="line.682"></a>
<span class="sourceLineNo">683</span> + " Note: if you do not use this option, then the target table must already exist in HBase\n"<a name="line.683"></a>
<span class="sourceLineNo">684</span> + "\n" + "Other options that may be specified with -D include:\n" + " -D" + DRY_RUN_CONF_KEY<a name="line.684"></a>
<span class="sourceLineNo">685</span> + "=true - Dry run mode. Data is not actually populated into"<a name="line.685"></a>
<span class="sourceLineNo">686</span> + " table. If table does not exist, it is created but deleted in the end.\n" + " -D"<a name="line.686"></a>
<span class="sourceLineNo">687</span> + SKIP_LINES_CONF_KEY + "=false - fail if encountering an invalid line\n" + " -D"<a name="line.687"></a>
<span class="sourceLineNo">688</span> + LOG_BAD_LINES_CONF_KEY + "=true - logs invalid lines to stderr\n" + " -D"<a name="line.688"></a>
<span class="sourceLineNo">689</span> + SKIP_EMPTY_COLUMNS + "=false - If true then skip empty columns in bulk import\n" + " '-D"<a name="line.689"></a>
<span class="sourceLineNo">690</span> + SEPARATOR_CONF_KEY + "=|' - eg separate on pipes instead of tabs\n" + " -D"<a name="line.690"></a>
<span class="sourceLineNo">691</span> + TIMESTAMP_CONF_KEY + "=currentTimeAsLong - use the specified timestamp for the import\n"<a name="line.691"></a>
<span class="sourceLineNo">692</span> + " -D" + MAPPER_CONF_KEY + "=my.Mapper - A user-defined Mapper to use instead of "<a name="line.692"></a>
<span class="sourceLineNo">693</span> + DEFAULT_MAPPER.getName() + "\n" + " -D" + JOB_NAME_CONF_KEY<a name="line.693"></a>
<span class="sourceLineNo">694</span> + "=jobName - use the specified mapreduce job name for the import\n" + " -D"<a name="line.694"></a>
<span class="sourceLineNo">695</span> + CREATE_TABLE_CONF_KEY + "=no - can be used to avoid creation of table by this tool\n"<a name="line.695"></a>
<span class="sourceLineNo">696</span> + " Note: if you set this to 'no', then the target table must already exist in HBase\n"<a name="line.696"></a>
<span class="sourceLineNo">697</span> + " -D" + NO_STRICT_COL_FAMILY + "=true - ignore column family check in hbase table. "<a name="line.697"></a>
<span class="sourceLineNo">698</span> + "Default is false\n\n" + "For performance consider the following options:\n"<a name="line.698"></a>
<span class="sourceLineNo">699</span> + " -Dmapreduce.map.speculative=false\n" + " -Dmapreduce.reduce.speculative=false";<a name="line.699"></a>
<span class="sourceLineNo">700</span><a name="line.700"></a>
<span class="sourceLineNo">701</span> System.err.println(usage);<a name="line.701"></a>
<span class="sourceLineNo">702</span> }<a name="line.702"></a>
<span class="sourceLineNo">703</span><a name="line.703"></a>
<span class="sourceLineNo">704</span> @Override<a name="line.704"></a>
<span class="sourceLineNo">705</span> public int run(String[] args) throws Exception {<a name="line.705"></a>
<span class="sourceLineNo">706</span> if (args.length &lt; 2) {<a name="line.706"></a>
<span class="sourceLineNo">707</span> usage("Wrong number of arguments: " + args.length);<a name="line.707"></a>
<span class="sourceLineNo">708</span> return -1;<a name="line.708"></a>
<span class="sourceLineNo">709</span> }<a name="line.709"></a>
<span class="sourceLineNo">710</span><a name="line.710"></a>
<span class="sourceLineNo">711</span> // When MAPPER_CONF_KEY is null, the user wants to use the provided TsvImporterMapper, so<a name="line.711"></a>
<span class="sourceLineNo">712</span> // perform validation on these additional args. When it's not null, user has provided their<a name="line.712"></a>
<span class="sourceLineNo">713</span> // own mapper, thus these validation are not relevant.<a name="line.713"></a>
<span class="sourceLineNo">714</span> // TODO: validation for TsvImporterMapper, not this tool. Move elsewhere.<a name="line.714"></a>
<span class="sourceLineNo">715</span> if (null == getConf().get(MAPPER_CONF_KEY)) {<a name="line.715"></a>
<span class="sourceLineNo">716</span> // Make sure columns are specified<a name="line.716"></a>
<span class="sourceLineNo">717</span> String[] columns = getConf().getStrings(COLUMNS_CONF_KEY);<a name="line.717"></a>
<span class="sourceLineNo">718</span> if (columns == null) {<a name="line.718"></a>
<span class="sourceLineNo">719</span> usage("No columns specified. Please specify with -D" + COLUMNS_CONF_KEY + "=...");<a name="line.719"></a>
<span class="sourceLineNo">720</span> return -1;<a name="line.720"></a>
<span class="sourceLineNo">721</span> }<a name="line.721"></a>
<span class="sourceLineNo">722</span><a name="line.722"></a>
<span class="sourceLineNo">723</span> // Make sure they specify exactly one column as the row key<a name="line.723"></a>
<span class="sourceLineNo">724</span> int rowkeysFound = 0;<a name="line.724"></a>
<span class="sourceLineNo">725</span> for (String col : columns) {<a name="line.725"></a>
<span class="sourceLineNo">726</span> if (col.equals(TsvParser.ROWKEY_COLUMN_SPEC)) rowkeysFound++;<a name="line.726"></a>
<span class="sourceLineNo">727</span> }<a name="line.727"></a>
<span class="sourceLineNo">728</span> if (rowkeysFound != 1) {<a name="line.728"></a>
<span class="sourceLineNo">729</span> usage("Must specify exactly one column as " + TsvParser.ROWKEY_COLUMN_SPEC);<a name="line.729"></a>
<span class="sourceLineNo">730</span> return -1;<a name="line.730"></a>
<span class="sourceLineNo">731</span> }<a name="line.731"></a>
<span class="sourceLineNo">732</span><a name="line.732"></a>
<span class="sourceLineNo">733</span> // Make sure we have at most one column as the timestamp key<a name="line.733"></a>
<span class="sourceLineNo">734</span> int tskeysFound = 0;<a name="line.734"></a>
<span class="sourceLineNo">735</span> for (String col : columns) {<a name="line.735"></a>
<span class="sourceLineNo">736</span> if (col.equals(TsvParser.TIMESTAMPKEY_COLUMN_SPEC)) tskeysFound++;<a name="line.736"></a>
<span class="sourceLineNo">737</span> }<a name="line.737"></a>
<span class="sourceLineNo">738</span> if (tskeysFound &gt; 1) {<a name="line.738"></a>
<span class="sourceLineNo">739</span> usage("Must specify at most one column as " + TsvParser.TIMESTAMPKEY_COLUMN_SPEC);<a name="line.739"></a>
<span class="sourceLineNo">740</span> return -1;<a name="line.740"></a>
<span class="sourceLineNo">741</span> }<a name="line.741"></a>
<span class="sourceLineNo">742</span><a name="line.742"></a>
<span class="sourceLineNo">743</span> int attrKeysFound = 0;<a name="line.743"></a>
<span class="sourceLineNo">744</span> for (String col : columns) {<a name="line.744"></a>
<span class="sourceLineNo">745</span> if (col.equals(TsvParser.ATTRIBUTES_COLUMN_SPEC)) attrKeysFound++;<a name="line.745"></a>
<span class="sourceLineNo">746</span> }<a name="line.746"></a>
<span class="sourceLineNo">747</span> if (attrKeysFound &gt; 1) {<a name="line.747"></a>
<span class="sourceLineNo">748</span> usage("Must specify at most one column as " + TsvParser.ATTRIBUTES_COLUMN_SPEC);<a name="line.748"></a>
<span class="sourceLineNo">749</span> return -1;<a name="line.749"></a>
<span class="sourceLineNo">750</span> }<a name="line.750"></a>
<span class="sourceLineNo">751</span><a name="line.751"></a>
<span class="sourceLineNo">752</span> // Make sure one or more columns are specified excluding rowkey and<a name="line.752"></a>
<span class="sourceLineNo">753</span> // timestamp key<a name="line.753"></a>
<span class="sourceLineNo">754</span> if (columns.length - (rowkeysFound + tskeysFound + attrKeysFound) &lt; 1) {<a name="line.754"></a>
<span class="sourceLineNo">755</span> usage(<a name="line.755"></a>
<span class="sourceLineNo">756</span> "One or more columns in addition to the row key and timestamp(optional) are required");<a name="line.756"></a>
<span class="sourceLineNo">757</span> return -1;<a name="line.757"></a>
<span class="sourceLineNo">758</span> }<a name="line.758"></a>
<span class="sourceLineNo">759</span> }<a name="line.759"></a>
<span class="sourceLineNo">760</span><a name="line.760"></a>
<span class="sourceLineNo">761</span> // If timestamp option is not specified, use current system time.<a name="line.761"></a>
<span class="sourceLineNo">762</span> long timstamp = getConf().getLong(TIMESTAMP_CONF_KEY, EnvironmentEdgeManager.currentTime());<a name="line.762"></a>
<span class="sourceLineNo">763</span><a name="line.763"></a>
<span class="sourceLineNo">764</span> // Set it back to replace invalid timestamp (non-numeric) with current<a name="line.764"></a>
<span class="sourceLineNo">765</span> // system time<a name="line.765"></a>
<span class="sourceLineNo">766</span> getConf().setLong(TIMESTAMP_CONF_KEY, timstamp);<a name="line.766"></a>
<span class="sourceLineNo">767</span><a name="line.767"></a>
<span class="sourceLineNo">768</span> synchronized (ImportTsv.class) {<a name="line.768"></a>
<span class="sourceLineNo">769</span> DRY_RUN_TABLE_CREATED = false;<a name="line.769"></a>
<span class="sourceLineNo">770</span> }<a name="line.770"></a>
<span class="sourceLineNo">771</span> Job job = createSubmittableJob(getConf(), args);<a name="line.771"></a>
<span class="sourceLineNo">772</span> boolean success = job.waitForCompletion(true);<a name="line.772"></a>
<span class="sourceLineNo">773</span> boolean delete = false;<a name="line.773"></a>
<span class="sourceLineNo">774</span> synchronized (ImportTsv.class) {<a name="line.774"></a>
<span class="sourceLineNo">775</span> delete = DRY_RUN_TABLE_CREATED;<a name="line.775"></a>
<span class="sourceLineNo">776</span> }<a name="line.776"></a>
<span class="sourceLineNo">777</span> if (delete) {<a name="line.777"></a>
<span class="sourceLineNo">778</span> deleteTable(getConf(), args);<a name="line.778"></a>
<span class="sourceLineNo">779</span> }<a name="line.779"></a>
<span class="sourceLineNo">780</span> return success ? 0 : 1;<a name="line.780"></a>
<span class="sourceLineNo">781</span> }<a name="line.781"></a>
<span class="sourceLineNo">782</span><a name="line.782"></a>
<span class="sourceLineNo">783</span> public static void main(String[] args) throws Exception {<a name="line.783"></a>
<span class="sourceLineNo">784</span> int status = ToolRunner.run(HBaseConfiguration.create(), new ImportTsv(), args);<a name="line.784"></a>
<span class="sourceLineNo">785</span> System.exit(status);<a name="line.785"></a>
<span class="sourceLineNo">786</span> }<a name="line.786"></a>
<span class="sourceLineNo">787</span>}<a name="line.787"></a>
</pre>
</div>
</body>
</html>