blob: 63dca183c8bf34645ea974350107caa4ed88019e [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.filter;<a name="line.18"></a>
<span class="sourceLineNo">019</span><a name="line.19"></a>
<span class="sourceLineNo">020</span>import static org.junit.Assert.*;<a name="line.20"></a>
<span class="sourceLineNo">021</span><a name="line.21"></a>
<span class="sourceLineNo">022</span>import java.util.regex.Pattern;<a name="line.22"></a>
<span class="sourceLineNo">023</span>import org.apache.hadoop.hbase.HBaseClassTestRule;<a name="line.23"></a>
<span class="sourceLineNo">024</span>import org.apache.hadoop.hbase.filter.RegexStringComparator.EngineType;<a name="line.24"></a>
<span class="sourceLineNo">025</span>import org.apache.hadoop.hbase.testclassification.FilterTests;<a name="line.25"></a>
<span class="sourceLineNo">026</span>import org.apache.hadoop.hbase.testclassification.SmallTests;<a name="line.26"></a>
<span class="sourceLineNo">027</span>import org.apache.hadoop.hbase.util.Bytes;<a name="line.27"></a>
<span class="sourceLineNo">028</span>import org.junit.ClassRule;<a name="line.28"></a>
<span class="sourceLineNo">029</span>import org.junit.Test;<a name="line.29"></a>
<span class="sourceLineNo">030</span>import org.junit.experimental.categories.Category;<a name="line.30"></a>
<span class="sourceLineNo">031</span><a name="line.31"></a>
<span class="sourceLineNo">032</span>@Category({FilterTests.class, SmallTests.class})<a name="line.32"></a>
<span class="sourceLineNo">033</span>public class TestRegexComparator {<a name="line.33"></a>
<span class="sourceLineNo">034</span><a name="line.34"></a>
<span class="sourceLineNo">035</span> @ClassRule<a name="line.35"></a>
<span class="sourceLineNo">036</span> public static final HBaseClassTestRule CLASS_RULE =<a name="line.36"></a>
<span class="sourceLineNo">037</span> HBaseClassTestRule.forClass(TestRegexComparator.class);<a name="line.37"></a>
<span class="sourceLineNo">038</span><a name="line.38"></a>
<span class="sourceLineNo">039</span> @Test<a name="line.39"></a>
<span class="sourceLineNo">040</span> public void testSerialization() throws Exception {<a name="line.40"></a>
<span class="sourceLineNo">041</span> // Default engine is the Java engine<a name="line.41"></a>
<span class="sourceLineNo">042</span> RegexStringComparator a = new RegexStringComparator("a|b");<a name="line.42"></a>
<span class="sourceLineNo">043</span> RegexStringComparator b = RegexStringComparator.parseFrom(a.toByteArray());<a name="line.43"></a>
<span class="sourceLineNo">044</span> assertTrue(a.areSerializedFieldsEqual(b));<a name="line.44"></a>
<span class="sourceLineNo">045</span> assertTrue(b.getEngine() instanceof RegexStringComparator.JavaRegexEngine);<a name="line.45"></a>
<span class="sourceLineNo">046</span><a name="line.46"></a>
<span class="sourceLineNo">047</span> // joni engine<a name="line.47"></a>
<span class="sourceLineNo">048</span> a = new RegexStringComparator("a|b", EngineType.JONI);<a name="line.48"></a>
<span class="sourceLineNo">049</span> b = RegexStringComparator.parseFrom(a.toByteArray());<a name="line.49"></a>
<span class="sourceLineNo">050</span> assertTrue(a.areSerializedFieldsEqual(b));<a name="line.50"></a>
<span class="sourceLineNo">051</span> assertTrue(b.getEngine() instanceof RegexStringComparator.JoniRegexEngine);<a name="line.51"></a>
<span class="sourceLineNo">052</span> }<a name="line.52"></a>
<span class="sourceLineNo">053</span><a name="line.53"></a>
<span class="sourceLineNo">054</span> @Test<a name="line.54"></a>
<span class="sourceLineNo">055</span> public void testJavaEngine() throws Exception {<a name="line.55"></a>
<span class="sourceLineNo">056</span> for (TestCase t: TEST_CASES) {<a name="line.56"></a>
<span class="sourceLineNo">057</span> boolean result = new RegexStringComparator(t.regex, t.flags, EngineType.JAVA)<a name="line.57"></a>
<span class="sourceLineNo">058</span> .compareTo(Bytes.toBytes(t.haystack)) == 0;<a name="line.58"></a>
<span class="sourceLineNo">059</span> assertEquals("Regex '" + t.regex + "' failed test '" + t.haystack + "'", result,<a name="line.59"></a>
<span class="sourceLineNo">060</span> t.expected);<a name="line.60"></a>
<span class="sourceLineNo">061</span> }<a name="line.61"></a>
<span class="sourceLineNo">062</span> }<a name="line.62"></a>
<span class="sourceLineNo">063</span><a name="line.63"></a>
<span class="sourceLineNo">064</span> @Test<a name="line.64"></a>
<span class="sourceLineNo">065</span> public void testJoniEngine() throws Exception {<a name="line.65"></a>
<span class="sourceLineNo">066</span> for (TestCase t: TEST_CASES) {<a name="line.66"></a>
<span class="sourceLineNo">067</span> boolean result = new RegexStringComparator(t.regex, t.flags, EngineType.JONI)<a name="line.67"></a>
<span class="sourceLineNo">068</span> .compareTo(Bytes.toBytes(t.haystack)) == 0;<a name="line.68"></a>
<span class="sourceLineNo">069</span> assertEquals("Regex '" + t.regex + "' failed test '" + t.haystack + "'", result,<a name="line.69"></a>
<span class="sourceLineNo">070</span> t.expected);<a name="line.70"></a>
<span class="sourceLineNo">071</span> }<a name="line.71"></a>
<span class="sourceLineNo">072</span> }<a name="line.72"></a>
<span class="sourceLineNo">073</span><a name="line.73"></a>
<span class="sourceLineNo">074</span> private static class TestCase {<a name="line.74"></a>
<span class="sourceLineNo">075</span> String regex;<a name="line.75"></a>
<span class="sourceLineNo">076</span> String haystack;<a name="line.76"></a>
<span class="sourceLineNo">077</span> int flags;<a name="line.77"></a>
<span class="sourceLineNo">078</span> boolean expected;<a name="line.78"></a>
<span class="sourceLineNo">079</span><a name="line.79"></a>
<span class="sourceLineNo">080</span> public TestCase(String regex, String haystack, boolean expected) {<a name="line.80"></a>
<span class="sourceLineNo">081</span> this(regex, Pattern.DOTALL, haystack, expected);<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> public TestCase(String regex, int flags, String haystack, boolean expected) {<a name="line.84"></a>
<span class="sourceLineNo">085</span> this.regex = regex;<a name="line.85"></a>
<span class="sourceLineNo">086</span> this.flags = flags;<a name="line.86"></a>
<span class="sourceLineNo">087</span> this.haystack = haystack;<a name="line.87"></a>
<span class="sourceLineNo">088</span> this.expected = expected;<a name="line.88"></a>
<span class="sourceLineNo">089</span> }<a name="line.89"></a>
<span class="sourceLineNo">090</span> }<a name="line.90"></a>
<span class="sourceLineNo">091</span><a name="line.91"></a>
<span class="sourceLineNo">092</span> // These are a subset of the regex tests from OpenJDK 7<a name="line.92"></a>
<span class="sourceLineNo">093</span> private static TestCase TEST_CASES[] = {<a name="line.93"></a>
<span class="sourceLineNo">094</span> new TestCase("a|b", "a", true),<a name="line.94"></a>
<span class="sourceLineNo">095</span> new TestCase("a|b", "b", true),<a name="line.95"></a>
<span class="sourceLineNo">096</span> new TestCase("a|b", Pattern.CASE_INSENSITIVE, "A", true),<a name="line.96"></a>
<span class="sourceLineNo">097</span> new TestCase("a|b", Pattern.CASE_INSENSITIVE, "B", true),<a name="line.97"></a>
<span class="sourceLineNo">098</span> new TestCase("a|b", "z", false),<a name="line.98"></a>
<span class="sourceLineNo">099</span> new TestCase("a|b|cd", "cd", true),<a name="line.99"></a>
<span class="sourceLineNo">100</span> new TestCase("z(a|ac)b", "zacb", true),<a name="line.100"></a>
<span class="sourceLineNo">101</span> new TestCase("[abc]+", "ababab", true),<a name="line.101"></a>
<span class="sourceLineNo">102</span> new TestCase("[abc]+", "defg", false),<a name="line.102"></a>
<span class="sourceLineNo">103</span> new TestCase("[abc]+[def]+[ghi]+", "zzzaaddggzzz", true),<a name="line.103"></a>
<span class="sourceLineNo">104</span> new TestCase("[a-\\u4444]+", "za-9z", true),<a name="line.104"></a>
<span class="sourceLineNo">105</span> new TestCase("[^abc]+", "ababab", false),<a name="line.105"></a>
<span class="sourceLineNo">106</span> new TestCase("[^abc]+", "aaabbbcccdefg", true),<a name="line.106"></a>
<span class="sourceLineNo">107</span> new TestCase("[abc^b]", "b", true),<a name="line.107"></a>
<span class="sourceLineNo">108</span> new TestCase("[abc[def]]", "b", true),<a name="line.108"></a>
<span class="sourceLineNo">109</span> new TestCase("[abc[def]]", "e", true),<a name="line.109"></a>
<span class="sourceLineNo">110</span> new TestCase("[a-c[d-f[g-i]]]", "h", true),<a name="line.110"></a>
<span class="sourceLineNo">111</span> new TestCase("[a-c[d-f[g-i]]m]", "m", true),<a name="line.111"></a>
<span class="sourceLineNo">112</span> new TestCase("[a-c&amp;&amp;[d-f]]", "a", false),<a name="line.112"></a>
<span class="sourceLineNo">113</span> new TestCase("[a-c&amp;&amp;[d-f]]", "z", false),<a name="line.113"></a>
<span class="sourceLineNo">114</span> new TestCase("[a-m&amp;&amp;m-z&amp;&amp;a-c]", "m", false),<a name="line.114"></a>
<span class="sourceLineNo">115</span> new TestCase("[a-m&amp;&amp;m-z&amp;&amp;a-z]", "m", true),<a name="line.115"></a>
<span class="sourceLineNo">116</span> new TestCase("[[a-m]&amp;&amp;[^a-c]]", "a", false),<a name="line.116"></a>
<span class="sourceLineNo">117</span> new TestCase("[[a-m]&amp;&amp;[^a-c]]", "d", true),<a name="line.117"></a>
<span class="sourceLineNo">118</span> new TestCase("[[a-c][d-f]&amp;&amp;abc[def]]", "e", true),<a name="line.118"></a>
<span class="sourceLineNo">119</span> new TestCase("[[a-c]&amp;&amp;[b-d]&amp;&amp;[c-e]]", "c", true),<a name="line.119"></a>
<span class="sourceLineNo">120</span> new TestCase("[[a-c]&amp;&amp;[b-d][c-e]&amp;&amp;[u-z]]", "c", false),<a name="line.120"></a>
<span class="sourceLineNo">121</span> new TestCase("[[a]&amp;&amp;[b][c][a]&amp;&amp;[^d]]", "a", true),<a name="line.121"></a>
<span class="sourceLineNo">122</span> new TestCase("[[a]&amp;&amp;[b][c][a]&amp;&amp;[^d]]", "d", false),<a name="line.122"></a>
<span class="sourceLineNo">123</span> new TestCase("[[[a-d]&amp;&amp;[c-f]]&amp;&amp;[c]&amp;&amp;c&amp;&amp;[cde]]", "c", true),<a name="line.123"></a>
<span class="sourceLineNo">124</span> new TestCase("[x[[wz]abc&amp;&amp;bcd[z]]&amp;&amp;[u-z]]", "z", true),<a name="line.124"></a>
<span class="sourceLineNo">125</span> new TestCase("a.c.+", "a#c%&amp;", true),<a name="line.125"></a>
<span class="sourceLineNo">126</span> new TestCase("ab.", "ab\n", true),<a name="line.126"></a>
<span class="sourceLineNo">127</span> new TestCase("(?s)ab.", "ab\n", true),<a name="line.127"></a>
<span class="sourceLineNo">128</span> new TestCase("ab\\wc", "abcc", true),<a name="line.128"></a>
<span class="sourceLineNo">129</span> new TestCase("\\W\\w\\W", "#r#", true),<a name="line.129"></a>
<span class="sourceLineNo">130</span> new TestCase("\\W\\w\\W", "rrrr#ggg", false),<a name="line.130"></a>
<span class="sourceLineNo">131</span> new TestCase("abc[\\sdef]*", "abc def", true),<a name="line.131"></a>
<span class="sourceLineNo">132</span> new TestCase("abc[\\sy-z]*", "abc y z", true),<a name="line.132"></a>
<span class="sourceLineNo">133</span> new TestCase("abc[a-d\\sm-p]*", "abcaa mn p", true),<a name="line.133"></a>
<span class="sourceLineNo">134</span> new TestCase("\\s\\s\\s", "blah err", false),<a name="line.134"></a>
<span class="sourceLineNo">135</span> new TestCase("\\S\\S\\s", "blah err", true),<a name="line.135"></a>
<span class="sourceLineNo">136</span> new TestCase("ab\\dc", "ab9c", true),<a name="line.136"></a>
<span class="sourceLineNo">137</span> new TestCase("\\d\\d\\d", "blah45", false),<a name="line.137"></a>
<span class="sourceLineNo">138</span> new TestCase("^abc", "abcdef", true),<a name="line.138"></a>
<span class="sourceLineNo">139</span> new TestCase("^abc", "bcdabc", false),<a name="line.139"></a>
<span class="sourceLineNo">140</span> new TestCase("^(a)?a", "a", true),<a name="line.140"></a>
<span class="sourceLineNo">141</span> new TestCase("^(aa(bb)?)+$", "aabbaa", true),<a name="line.141"></a>
<span class="sourceLineNo">142</span> new TestCase("((a|b)?b)+", "b", true),<a name="line.142"></a>
<span class="sourceLineNo">143</span> new TestCase("^(a(b)?)+$", "aba", true),<a name="line.143"></a>
<span class="sourceLineNo">144</span> new TestCase("^(a(b(c)?)?)?abc", "abc", true),<a name="line.144"></a>
<span class="sourceLineNo">145</span> new TestCase("^(a(b(c))).*", "abc", true),<a name="line.145"></a>
<span class="sourceLineNo">146</span> new TestCase("a?b", "aaaab", true),<a name="line.146"></a>
<span class="sourceLineNo">147</span> new TestCase("a?b", "aaacc", false),<a name="line.147"></a>
<span class="sourceLineNo">148</span> new TestCase("a??b", "aaaab", true),<a name="line.148"></a>
<span class="sourceLineNo">149</span> new TestCase("a??b", "aaacc", false),<a name="line.149"></a>
<span class="sourceLineNo">150</span> new TestCase("a?+b", "aaaab", true),<a name="line.150"></a>
<span class="sourceLineNo">151</span> new TestCase("a?+b", "aaacc", false),<a name="line.151"></a>
<span class="sourceLineNo">152</span> new TestCase("a+b", "aaaab", true),<a name="line.152"></a>
<span class="sourceLineNo">153</span> new TestCase("a+b", "aaacc", false),<a name="line.153"></a>
<span class="sourceLineNo">154</span> new TestCase("a+?b", "aaaab", true),<a name="line.154"></a>
<span class="sourceLineNo">155</span> new TestCase("a+?b", "aaacc", false),<a name="line.155"></a>
<span class="sourceLineNo">156</span> new TestCase("a++b", "aaaab", true),<a name="line.156"></a>
<span class="sourceLineNo">157</span> new TestCase("a++b", "aaacc", false),<a name="line.157"></a>
<span class="sourceLineNo">158</span> new TestCase("a{2,3}", "a", false),<a name="line.158"></a>
<span class="sourceLineNo">159</span> new TestCase("a{2,3}", "aa", true),<a name="line.159"></a>
<span class="sourceLineNo">160</span> new TestCase("a{2,3}", "aaa", true),<a name="line.160"></a>
<span class="sourceLineNo">161</span> new TestCase("a{3,}", "zzzaaaazzz", true),<a name="line.161"></a>
<span class="sourceLineNo">162</span> new TestCase("a{3,}", "zzzaazzz", false),<a name="line.162"></a>
<span class="sourceLineNo">163</span> new TestCase("abc(?=d)", "zzzabcd", true),<a name="line.163"></a>
<span class="sourceLineNo">164</span> new TestCase("abc(?=d)", "zzzabced", false),<a name="line.164"></a>
<span class="sourceLineNo">165</span> new TestCase("abc(?!d)", "zzabcd", false),<a name="line.165"></a>
<span class="sourceLineNo">166</span> new TestCase("abc(?!d)", "zzabced", true),<a name="line.166"></a>
<span class="sourceLineNo">167</span> new TestCase("\\w(?&lt;=a)", "###abc###", true),<a name="line.167"></a>
<span class="sourceLineNo">168</span> new TestCase("\\w(?&lt;=a)", "###ert###", false),<a name="line.168"></a>
<span class="sourceLineNo">169</span> new TestCase("(?&lt;!a)c", "bc", true),<a name="line.169"></a>
<span class="sourceLineNo">170</span> new TestCase("(?&lt;!a)c", "ac", false),<a name="line.170"></a>
<span class="sourceLineNo">171</span> new TestCase("(a+b)+", "ababab", true),<a name="line.171"></a>
<span class="sourceLineNo">172</span> new TestCase("(a+b)+", "accccd", false),<a name="line.172"></a>
<span class="sourceLineNo">173</span> new TestCase("(ab)+", "ababab", true),<a name="line.173"></a>
<span class="sourceLineNo">174</span> new TestCase("(ab)+", "accccd", false),<a name="line.174"></a>
<span class="sourceLineNo">175</span> new TestCase("(ab)(cd*)", "zzzabczzz", true),<a name="line.175"></a>
<span class="sourceLineNo">176</span> new TestCase("abc(d)*abc", "abcdddddabc", true),<a name="line.176"></a>
<span class="sourceLineNo">177</span> new TestCase("a*b", "aaaab", true),<a name="line.177"></a>
<span class="sourceLineNo">178</span> new TestCase("a*b", "b", true),<a name="line.178"></a>
<span class="sourceLineNo">179</span> new TestCase("a*b", "aaaac", false),<a name="line.179"></a>
<span class="sourceLineNo">180</span> new TestCase(".*?b", "aaaab", true),<a name="line.180"></a>
<span class="sourceLineNo">181</span> new TestCase("a*+b", "aaaab", true),<a name="line.181"></a>
<span class="sourceLineNo">182</span> new TestCase("a*+b", "b", true),<a name="line.182"></a>
<span class="sourceLineNo">183</span> new TestCase("a*+b", "aaaac", false),<a name="line.183"></a>
<span class="sourceLineNo">184</span> new TestCase("(?i)foobar", "fOobAr", true),<a name="line.184"></a>
<span class="sourceLineNo">185</span> new TestCase("f(?i)oobar", "fOobAr", true),<a name="line.185"></a>
<span class="sourceLineNo">186</span> new TestCase("f(?i)oobar", "FOobAr", false),<a name="line.186"></a>
<span class="sourceLineNo">187</span> new TestCase("foo(?i)bar", "fOobAr", false),<a name="line.187"></a>
<span class="sourceLineNo">188</span> new TestCase("(?i)foo[bar]+", "foObAr", true),<a name="line.188"></a>
<span class="sourceLineNo">189</span> new TestCase("(?i)foo[a-r]+", "foObAr", true),<a name="line.189"></a>
<span class="sourceLineNo">190</span> new TestCase("abc(?x)blah", "abcblah", true),<a name="line.190"></a>
<span class="sourceLineNo">191</span> new TestCase("abc(?x) blah", "abcblah", true),<a name="line.191"></a>
<span class="sourceLineNo">192</span> new TestCase("abc(?x) blah blech", "abcblahblech", true),<a name="line.192"></a>
<span class="sourceLineNo">193</span> new TestCase("[\\n-#]", "!", true),<a name="line.193"></a>
<span class="sourceLineNo">194</span> new TestCase("[\\n-#]", "-", false),<a name="line.194"></a>
<span class="sourceLineNo">195</span> new TestCase("[\\043]+", "blahblah#blech", true),<a name="line.195"></a>
<span class="sourceLineNo">196</span> new TestCase("[\\042-\\044]+", "blahblah#blech", true),<a name="line.196"></a>
<span class="sourceLineNo">197</span> new TestCase("[\\u1234-\\u1236]", "blahblah\u1235blech", true),<a name="line.197"></a>
<span class="sourceLineNo">198</span> new TestCase("[^\043]*", "blahblah#blech", true),<a name="line.198"></a>
<span class="sourceLineNo">199</span> new TestCase("(|f)?+", "foo", true),<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>
</pre>
</div>
</body>
</html>