| <!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> *<a name="line.2"></a> |
| <span class="sourceLineNo">003</span> * Licensed to the Apache Software Foundation (ASF) under one<a name="line.3"></a> |
| <span class="sourceLineNo">004</span> * or more contributor license agreements. See the NOTICE file<a name="line.4"></a> |
| <span class="sourceLineNo">005</span> * distributed with this work for additional information<a name="line.5"></a> |
| <span class="sourceLineNo">006</span> * regarding copyright ownership. The ASF licenses this file<a name="line.6"></a> |
| <span class="sourceLineNo">007</span> * to you under the Apache License, Version 2.0 (the<a name="line.7"></a> |
| <span class="sourceLineNo">008</span> * "License"); you may not use this file except in compliance<a name="line.8"></a> |
| <span class="sourceLineNo">009</span> * with the License. You may obtain a copy of the License at<a name="line.9"></a> |
| <span class="sourceLineNo">010</span> *<a name="line.10"></a> |
| <span class="sourceLineNo">011</span> * http://www.apache.org/licenses/LICENSE-2.0<a name="line.11"></a> |
| <span class="sourceLineNo">012</span> *<a name="line.12"></a> |
| <span class="sourceLineNo">013</span> * Unless required by applicable law or agreed to in writing, software<a name="line.13"></a> |
| <span class="sourceLineNo">014</span> * distributed under the License is distributed on an "AS IS" BASIS,<a name="line.14"></a> |
| <span class="sourceLineNo">015</span> * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.<a name="line.15"></a> |
| <span class="sourceLineNo">016</span> * See the License for the specific language governing permissions and<a name="line.16"></a> |
| <span class="sourceLineNo">017</span> * limitations under the License.<a name="line.17"></a> |
| <span class="sourceLineNo">018</span> */<a name="line.18"></a> |
| <span class="sourceLineNo">019</span>package org.apache.hadoop.hbase.mapreduce;<a name="line.19"></a> |
| <span class="sourceLineNo">020</span><a name="line.20"></a> |
| <span class="sourceLineNo">021</span>import java.io.IOException;<a name="line.21"></a> |
| <span class="sourceLineNo">022</span>import java.util.Arrays;<a name="line.22"></a> |
| <span class="sourceLineNo">023</span>import java.util.List;<a name="line.23"></a> |
| <span class="sourceLineNo">024</span><a name="line.24"></a> |
| <span class="sourceLineNo">025</span>import org.apache.hadoop.conf.Configuration;<a name="line.25"></a> |
| <span class="sourceLineNo">026</span>import org.apache.hadoop.fs.Path;<a name="line.26"></a> |
| <span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.CompareOperator;<a name="line.27"></a> |
| <span class="sourceLineNo">028</span>import org.apache.hadoop.hbase.HConstants;<a name="line.28"></a> |
| <span class="sourceLineNo">029</span>import org.apache.hadoop.hbase.TableName;<a name="line.29"></a> |
| <span class="sourceLineNo">030</span>import org.apache.yetus.audience.InterfaceAudience;<a name="line.30"></a> |
| <span class="sourceLineNo">031</span>import org.slf4j.Logger;<a name="line.31"></a> |
| <span class="sourceLineNo">032</span>import org.slf4j.LoggerFactory;<a name="line.32"></a> |
| <span class="sourceLineNo">033</span>import org.apache.hadoop.hbase.client.Scan;<a name="line.33"></a> |
| <span class="sourceLineNo">034</span>import org.apache.hadoop.hbase.filter.Filter;<a name="line.34"></a> |
| <span class="sourceLineNo">035</span>import org.apache.hadoop.hbase.filter.IncompatibleFilterException;<a name="line.35"></a> |
| <span class="sourceLineNo">036</span>import org.apache.hadoop.hbase.filter.PrefixFilter;<a name="line.36"></a> |
| <span class="sourceLineNo">037</span>import org.apache.hadoop.hbase.filter.RegexStringComparator;<a name="line.37"></a> |
| <span class="sourceLineNo">038</span>import org.apache.hadoop.hbase.filter.RowFilter;<a name="line.38"></a> |
| <span class="sourceLineNo">039</span>import org.apache.hadoop.hbase.security.visibility.Authorizations;<a name="line.39"></a> |
| <span class="sourceLineNo">040</span>import org.apache.hbase.thirdparty.com.google.common.annotations.VisibleForTesting;<a name="line.40"></a> |
| <span class="sourceLineNo">041</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.41"></a> |
| <span class="sourceLineNo">042</span>import org.apache.hadoop.hbase.util.Triple;<a name="line.42"></a> |
| <span class="sourceLineNo">043</span>import org.apache.hadoop.mapreduce.lib.output.FileOutputFormat;<a name="line.43"></a> |
| <span class="sourceLineNo">044</span><a name="line.44"></a> |
| <span class="sourceLineNo">045</span>/**<a name="line.45"></a> |
| <span class="sourceLineNo">046</span> * Some helper methods are used by {@link org.apache.hadoop.hbase.mapreduce.Export}<a name="line.46"></a> |
| <span class="sourceLineNo">047</span> * and org.apache.hadoop.hbase.coprocessor.Export (in hbase-endpooint).<a name="line.47"></a> |
| <span class="sourceLineNo">048</span> */<a name="line.48"></a> |
| <span class="sourceLineNo">049</span>@InterfaceAudience.Private<a name="line.49"></a> |
| <span class="sourceLineNo">050</span>public final class ExportUtils {<a name="line.50"></a> |
| <span class="sourceLineNo">051</span> private static final Logger LOG = LoggerFactory.getLogger(ExportUtils.class);<a name="line.51"></a> |
| <span class="sourceLineNo">052</span> public static final String RAW_SCAN = "hbase.mapreduce.include.deleted.rows";<a name="line.52"></a> |
| <span class="sourceLineNo">053</span> public static final String EXPORT_BATCHING = "hbase.export.scanner.batch";<a name="line.53"></a> |
| <span class="sourceLineNo">054</span> public static final String EXPORT_CACHING = "hbase.export.scanner.caching";<a name="line.54"></a> |
| <span class="sourceLineNo">055</span> public static final String EXPORT_VISIBILITY_LABELS = "hbase.export.visibility.labels";<a name="line.55"></a> |
| <span class="sourceLineNo">056</span> /**<a name="line.56"></a> |
| <span class="sourceLineNo">057</span> * Common usage for other export tools.<a name="line.57"></a> |
| <span class="sourceLineNo">058</span> * @param errorMsg Error message. Can be null.<a name="line.58"></a> |
| <span class="sourceLineNo">059</span> */<a name="line.59"></a> |
| <span class="sourceLineNo">060</span> public static void usage(final String errorMsg) {<a name="line.60"></a> |
| <span class="sourceLineNo">061</span> if (errorMsg != null && errorMsg.length() > 0) {<a name="line.61"></a> |
| <span class="sourceLineNo">062</span> System.err.println("ERROR: " + errorMsg);<a name="line.62"></a> |
| <span class="sourceLineNo">063</span> }<a name="line.63"></a> |
| <span class="sourceLineNo">064</span> System.err.println("Usage: Export [-D <property=value>]* <tablename> <outputdir> [<versions> " +<a name="line.64"></a> |
| <span class="sourceLineNo">065</span> "[<starttime> [<endtime>]] [^[regex pattern] or [Prefix] to filter]]\n");<a name="line.65"></a> |
| <span class="sourceLineNo">066</span> System.err.println(" Note: -D properties will be applied to the conf used. ");<a name="line.66"></a> |
| <span class="sourceLineNo">067</span> System.err.println(" For example: ");<a name="line.67"></a> |
| <span class="sourceLineNo">068</span> System.err.println(" -D " + FileOutputFormat.COMPRESS + "=true");<a name="line.68"></a> |
| <span class="sourceLineNo">069</span> System.err.println(" -D " + FileOutputFormat.COMPRESS_CODEC + "=org.apache.hadoop.io.compress.GzipCodec");<a name="line.69"></a> |
| <span class="sourceLineNo">070</span> System.err.println(" -D " + FileOutputFormat.COMPRESS_TYPE + "=BLOCK");<a name="line.70"></a> |
| <span class="sourceLineNo">071</span> System.err.println(" Additionally, the following SCAN properties can be specified");<a name="line.71"></a> |
| <span class="sourceLineNo">072</span> System.err.println(" to control/limit what is exported..");<a name="line.72"></a> |
| <span class="sourceLineNo">073</span> System.err.println(" -D " + TableInputFormat.SCAN_COLUMN_FAMILY + "=<family1>,<family2>, ...");<a name="line.73"></a> |
| <span class="sourceLineNo">074</span> System.err.println(" -D " + RAW_SCAN + "=true");<a name="line.74"></a> |
| <span class="sourceLineNo">075</span> System.err.println(" -D " + TableInputFormat.SCAN_ROW_START + "=<ROWSTART>");<a name="line.75"></a> |
| <span class="sourceLineNo">076</span> System.err.println(" -D " + TableInputFormat.SCAN_ROW_STOP + "=<ROWSTOP>");<a name="line.76"></a> |
| <span class="sourceLineNo">077</span> System.err.println(" -D " + HConstants.HBASE_CLIENT_SCANNER_CACHING + "=100");<a name="line.77"></a> |
| <span class="sourceLineNo">078</span> System.err.println(" -D " + EXPORT_VISIBILITY_LABELS + "=<labels>");<a name="line.78"></a> |
| <span class="sourceLineNo">079</span> System.err.println("For tables with very wide rows consider setting the batch size as below:\n"<a name="line.79"></a> |
| <span class="sourceLineNo">080</span> + " -D " + EXPORT_BATCHING + "=10\n"<a name="line.80"></a> |
| <span class="sourceLineNo">081</span> + " -D " + EXPORT_CACHING + "=100");<a name="line.81"></a> |
| <span class="sourceLineNo">082</span> }<a name="line.82"></a> |
| <span class="sourceLineNo">083</span><a name="line.83"></a> |
| <span class="sourceLineNo">084</span> private static Filter getExportFilter(String[] args) {<a name="line.84"></a> |
| <span class="sourceLineNo">085</span> Filter exportFilter;<a name="line.85"></a> |
| <span class="sourceLineNo">086</span> String filterCriteria = (args.length > 5) ? args[5]: null;<a name="line.86"></a> |
| <span class="sourceLineNo">087</span> if (filterCriteria == null) return null;<a name="line.87"></a> |
| <span class="sourceLineNo">088</span> if (filterCriteria.startsWith("^")) {<a name="line.88"></a> |
| <span class="sourceLineNo">089</span> String regexPattern = filterCriteria.substring(1, filterCriteria.length());<a name="line.89"></a> |
| <span class="sourceLineNo">090</span> exportFilter = new RowFilter(CompareOperator.EQUAL, new RegexStringComparator(regexPattern));<a name="line.90"></a> |
| <span class="sourceLineNo">091</span> } else {<a name="line.91"></a> |
| <span class="sourceLineNo">092</span> exportFilter = new PrefixFilter(Bytes.toBytesBinary(filterCriteria));<a name="line.92"></a> |
| <span class="sourceLineNo">093</span> }<a name="line.93"></a> |
| <span class="sourceLineNo">094</span> return exportFilter;<a name="line.94"></a> |
| <span class="sourceLineNo">095</span> }<a name="line.95"></a> |
| <span class="sourceLineNo">096</span><a name="line.96"></a> |
| <span class="sourceLineNo">097</span> public static boolean isValidArguements(String[] args) {<a name="line.97"></a> |
| <span class="sourceLineNo">098</span> return args != null && args.length >= 2;<a name="line.98"></a> |
| <span class="sourceLineNo">099</span> }<a name="line.99"></a> |
| <span class="sourceLineNo">100</span><a name="line.100"></a> |
| <span class="sourceLineNo">101</span> public static Triple<TableName, Scan, Path> getArgumentsFromCommandLine(<a name="line.101"></a> |
| <span class="sourceLineNo">102</span> Configuration conf, String[] args) throws IOException {<a name="line.102"></a> |
| <span class="sourceLineNo">103</span> if (!isValidArguements(args)) {<a name="line.103"></a> |
| <span class="sourceLineNo">104</span> return null;<a name="line.104"></a> |
| <span class="sourceLineNo">105</span> }<a name="line.105"></a> |
| <span class="sourceLineNo">106</span> return new Triple<>(TableName.valueOf(args[0]), getScanFromCommandLine(conf, args), new Path(args[1]));<a name="line.106"></a> |
| <span class="sourceLineNo">107</span> }<a name="line.107"></a> |
| <span class="sourceLineNo">108</span><a name="line.108"></a> |
| <span class="sourceLineNo">109</span> @VisibleForTesting<a name="line.109"></a> |
| <span class="sourceLineNo">110</span> static Scan getScanFromCommandLine(Configuration conf, String[] args) throws IOException {<a name="line.110"></a> |
| <span class="sourceLineNo">111</span> Scan s = new Scan();<a name="line.111"></a> |
| <span class="sourceLineNo">112</span> // Optional arguments.<a name="line.112"></a> |
| <span class="sourceLineNo">113</span> // Set Scan Versions<a name="line.113"></a> |
| <span class="sourceLineNo">114</span> int versions = args.length > 2? Integer.parseInt(args[2]): 1;<a name="line.114"></a> |
| <span class="sourceLineNo">115</span> s.setMaxVersions(versions);<a name="line.115"></a> |
| <span class="sourceLineNo">116</span> // Set Scan Range<a name="line.116"></a> |
| <span class="sourceLineNo">117</span> long startTime = args.length > 3? Long.parseLong(args[3]): 0L;<a name="line.117"></a> |
| <span class="sourceLineNo">118</span> long endTime = args.length > 4? Long.parseLong(args[4]): Long.MAX_VALUE;<a name="line.118"></a> |
| <span class="sourceLineNo">119</span> s.setTimeRange(startTime, endTime);<a name="line.119"></a> |
| <span class="sourceLineNo">120</span> // Set cache blocks<a name="line.120"></a> |
| <span class="sourceLineNo">121</span> s.setCacheBlocks(false);<a name="line.121"></a> |
| <span class="sourceLineNo">122</span> // set Start and Stop row<a name="line.122"></a> |
| <span class="sourceLineNo">123</span> if (conf.get(TableInputFormat.SCAN_ROW_START) != null) {<a name="line.123"></a> |
| <span class="sourceLineNo">124</span> s.setStartRow(Bytes.toBytesBinary(conf.get(TableInputFormat.SCAN_ROW_START)));<a name="line.124"></a> |
| <span class="sourceLineNo">125</span> }<a name="line.125"></a> |
| <span class="sourceLineNo">126</span> if (conf.get(TableInputFormat.SCAN_ROW_STOP) != null) {<a name="line.126"></a> |
| <span class="sourceLineNo">127</span> s.setStopRow(Bytes.toBytesBinary(conf.get(TableInputFormat.SCAN_ROW_STOP)));<a name="line.127"></a> |
| <span class="sourceLineNo">128</span> }<a name="line.128"></a> |
| <span class="sourceLineNo">129</span> // Set Scan Column Family<a name="line.129"></a> |
| <span class="sourceLineNo">130</span> boolean raw = Boolean.parseBoolean(conf.get(RAW_SCAN));<a name="line.130"></a> |
| <span class="sourceLineNo">131</span> if (raw) {<a name="line.131"></a> |
| <span class="sourceLineNo">132</span> s.setRaw(raw);<a name="line.132"></a> |
| <span class="sourceLineNo">133</span> }<a name="line.133"></a> |
| <span class="sourceLineNo">134</span> for (String columnFamily : conf.getTrimmedStrings(TableInputFormat.SCAN_COLUMN_FAMILY)) {<a name="line.134"></a> |
| <span class="sourceLineNo">135</span> s.addFamily(Bytes.toBytes(columnFamily));<a name="line.135"></a> |
| <span class="sourceLineNo">136</span> }<a name="line.136"></a> |
| <span class="sourceLineNo">137</span> // Set RowFilter or Prefix Filter if applicable.<a name="line.137"></a> |
| <span class="sourceLineNo">138</span> Filter exportFilter = getExportFilter(args);<a name="line.138"></a> |
| <span class="sourceLineNo">139</span> if (exportFilter!= null) {<a name="line.139"></a> |
| <span class="sourceLineNo">140</span> LOG.info("Setting Scan Filter for Export.");<a name="line.140"></a> |
| <span class="sourceLineNo">141</span> s.setFilter(exportFilter);<a name="line.141"></a> |
| <span class="sourceLineNo">142</span> }<a name="line.142"></a> |
| <span class="sourceLineNo">143</span> List<String> labels = null;<a name="line.143"></a> |
| <span class="sourceLineNo">144</span> if (conf.get(EXPORT_VISIBILITY_LABELS) != null) {<a name="line.144"></a> |
| <span class="sourceLineNo">145</span> labels = Arrays.asList(conf.getStrings(EXPORT_VISIBILITY_LABELS));<a name="line.145"></a> |
| <span class="sourceLineNo">146</span> if (!labels.isEmpty()) {<a name="line.146"></a> |
| <span class="sourceLineNo">147</span> s.setAuthorizations(new Authorizations(labels));<a name="line.147"></a> |
| <span class="sourceLineNo">148</span> }<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> int batching = conf.getInt(EXPORT_BATCHING, -1);<a name="line.151"></a> |
| <span class="sourceLineNo">152</span> if (batching != -1) {<a name="line.152"></a> |
| <span class="sourceLineNo">153</span> try {<a name="line.153"></a> |
| <span class="sourceLineNo">154</span> s.setBatch(batching);<a name="line.154"></a> |
| <span class="sourceLineNo">155</span> } catch (IncompatibleFilterException e) {<a name="line.155"></a> |
| <span class="sourceLineNo">156</span> LOG.error("Batching could not be set", e);<a name="line.156"></a> |
| <span class="sourceLineNo">157</span> }<a name="line.157"></a> |
| <span class="sourceLineNo">158</span> }<a name="line.158"></a> |
| <span class="sourceLineNo">159</span><a name="line.159"></a> |
| <span class="sourceLineNo">160</span> int caching = conf.getInt(EXPORT_CACHING, 100);<a name="line.160"></a> |
| <span class="sourceLineNo">161</span> if (caching != -1) {<a name="line.161"></a> |
| <span class="sourceLineNo">162</span> try {<a name="line.162"></a> |
| <span class="sourceLineNo">163</span> s.setCaching(caching);<a name="line.163"></a> |
| <span class="sourceLineNo">164</span> } catch (IncompatibleFilterException e) {<a name="line.164"></a> |
| <span class="sourceLineNo">165</span> LOG.error("Caching could not be set", e);<a name="line.165"></a> |
| <span class="sourceLineNo">166</span> }<a name="line.166"></a> |
| <span class="sourceLineNo">167</span> }<a name="line.167"></a> |
| <span class="sourceLineNo">168</span> LOG.info("versions=" + versions + ", starttime=" + startTime<a name="line.168"></a> |
| <span class="sourceLineNo">169</span> + ", endtime=" + endTime + ", keepDeletedCells=" + raw<a name="line.169"></a> |
| <span class="sourceLineNo">170</span> + ", visibility labels=" + labels);<a name="line.170"></a> |
| <span class="sourceLineNo">171</span> return s;<a name="line.171"></a> |
| <span class="sourceLineNo">172</span> }<a name="line.172"></a> |
| <span class="sourceLineNo">173</span><a name="line.173"></a> |
| <span class="sourceLineNo">174</span> private ExportUtils() {<a name="line.174"></a> |
| <span class="sourceLineNo">175</span> }<a name="line.175"></a> |
| <span class="sourceLineNo">176</span>}<a name="line.176"></a> |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| |
| </pre> |
| </div> |
| </body> |
| </html> |