/*
* Licensed to the Apache Software Foundation (ASF) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* The ASF licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.apache.el.parser;
import java.io.StringReader;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.HashMap;
import java.util.concurrent.Callable;
import jakarta.el.ELBaseTest;
import jakarta.el.ELContext;
import jakarta.el.ELManager;
import jakarta.el.ExpressionFactory;
import jakarta.el.ValueExpression;
import org.junit.Assert;
import org.junit.Ignore;
import org.junit.Test;
import org.apache.tomcat.util.collections.SynchronizedStack;
/*
* This is a relative performance test so it remains part of the standard test run.
*/
public class TestELParserPerformance extends ELBaseTest {
private static final long DEFAULT_TEST_ITERATIONS = 1000000;
/*
* Test to explore if re-using Parser instances is faster.
*
* Tests on my laptop show: - overhead by introducing the stack is in the noise for parsing even the simplest
* expression - efficiency from re-using the ELParser is measurable for even a single reuse of the parser - with
* large numbers of parses (~10k) performance for a trivial parse is three times faster - around the 100 iterations
* mark GC overhead adds significant noise to the results - for consistent results you either need fewer parses to
* avoid triggering GC or more parses so the GC effects are evenly distributed between the runs
*
* Note that the test is single threaded.
*/
@Test
public void testParserInstanceReuse() throws ParseException {
final int runs = 20;
final int parseIterations = 10000;
long reinitTotalTime = 0;
long newTotalTime = 0;
for (int j = 0; j < runs; j++) {
long start = System.nanoTime();
SynchronizedStack<ELParser> stack = new SynchronizedStack<>();
for (int i = 0; i < parseIterations; i++) {
ELParser parser = stack.pop();
if (parser == null) {
parser = new ELParser(new StringReader("${'foo'}"));
} else {
parser.ReInit(new StringReader("${'foo'}"));
}
parser.CompositeExpression();
stack.push(parser);
}
long end = System.nanoTime();
reinitTotalTime += (end - start);
System.out.println(parseIterations + " iterations using ELParser.ReInit(...) took " + (end - start) + "ns");
}
for (int j = 0; j < runs; j++) {
long start = System.nanoTime();
for (int i = 0; i < parseIterations; i++) {
ELParser parser = new ELParser(new StringReader("${'foo'}"));
parser.CompositeExpression();
}
long end = System.nanoTime();
newTotalTime += (end - start);
System.out.println(parseIterations + " iterations using new ELParser(...) took " + (end - start) + "ns");
}
Assert.assertTrue("Using new ElParser() was faster then using ELParser.ReInit", reinitTotalTime < newTotalTime);
}
/*
* Ignored by default since this is an absolute test primarily for
* https://bz.apache.org/bugzilla/show_bug.cgi?id=69338
*/
@Ignore
@Test
public void testAstAnd() {
ELManager manager = new ELManager();
ELContext context = manager.getELContext();
ExpressionFactory factory = ELManager.getExpressionFactory();
long durations[] = new long[9];
for (int j = 0; j < 5; j++) {
for (int operandCount = 2; operandCount < 11; operandCount++) {
StringBuilder sb = new StringBuilder("${true");
for (int i = 2; i <= operandCount; i++) {
sb.append(" && true");
}
sb.append("}");
String expression = sb.toString();
long start = System.nanoTime();
for (int i = 0; i < 2000000; i++) {
ValueExpression ve = factory.createValueExpression(context, expression, Boolean.class);
Boolean result = ve.getValue(context);
Assert.assertEquals(Boolean.TRUE, result);
}
long duration = System.nanoTime() - start;
if (j > 0) {
durations[operandCount - 2] += duration;
}
}
}
for (int operandCount = 2; operandCount < 11; operandCount++) {
System.out.println("Operand count [" + operandCount + "], duration [" + durations[operandCount - 2] + "]");
}
System.out.println("");
}
/*
* Ignored by default since this is an absolute test primarily for
* https://bz.apache.org/bugzilla/show_bug.cgi?id=69338
*/
@Ignore
@Test
public void testAstNotEmpty() {
ELManager manager = new ELManager();
ELContext context = manager.getELContext();
ExpressionFactory factory = ELManager.getExpressionFactory();
for (int j = 0; j < 5; j++) {
String expression = "${not empty 'abc'}";
long start = System.nanoTime();
for (int i = 0; i < 10000000; i++) {
ValueExpression ve = factory.createValueExpression(context, expression, Boolean.class);
Boolean result = ve.getValue(context);
Assert.assertEquals(Boolean.TRUE, result);
}
long duration = System.nanoTime() - start;
System.out.println("duration [" + duration + "]");
}
System.out.println("");
}
/*
* Utility method for running tasks provided by the test*() methods.
*
* @param c Callable to be executed.
* @param description Label to be displayed in the results.
*
* @return Arbitrary response data provided by the Callable.
*/
private Object runTest(Callable<Object> c, String description) throws Exception {
Object result = null;
for (int j = 0; j < 5; j++) {
System.gc();
long start = System.currentTimeMillis();
for (int i = 0; i < DEFAULT_TEST_ITERATIONS; i++) {
result = c.call();
}
long duration = System.currentTimeMillis() - start;
System.out.println(description + " duration=[" + duration + "ms] or " +
(duration * 1000000 / DEFAULT_TEST_ITERATIONS + " ns each"));
}
System.out.println("Result: " + result);
return result;
}
/*
* Performance test of various expressions. This can be easily modified to explore the performance of other
* expressions.
*
* The operations tested in this method allocate many objects. For most consistent results, use runtime arguments
* such as these: -Xms3g -Xmx3g -XX:+UseParallelGC
*
* Ignored by default since this is an absolute test primarily for
* https://bz.apache.org/bugzilla/show_bug.cgi?id=69381
*/
@Ignore
@Test
public void testExpressions() throws Exception {
ELManager manager = new ELManager();
ELContext context = manager.getELContext();
ExpressionFactory factory = ELManager.getExpressionFactory();
TesterBeanD beanD = new TesterBeanD();
TesterBeanE beanE = new TesterBeanE();
beanD.setBean(beanE);
beanD.setName("name");
beanE.setName("name");
beanD.setStringMap(new HashMap<>());
beanD.setValList(new ArrayList<>(Arrays.asList("a", "b", "c")));
ValueExpression var = factory.createValueExpression(beanD, TesterBeanD.class);
context.getVariableMapper().setVariable("beanD", var);
ValueExpression varB = factory.createValueExpression(beanE, TesterBeanE.class);
context.getVariableMapper().setVariable("beanE", varB);
var = factory.createValueExpression(TesterEnum.ONE, TesterEnum.class);
context.getVariableMapper().setVariable("enumOne", var);
String[] expressions = new String[] { "${beanD.name}", "${beanD.getName()}", "${beanE.getName()}" };
for (String expression : expressions) {
runTest(() -> {
ValueExpression ve = factory.createValueExpression(context, expression, Object.class);
return ve.getValue(context);
}, expression);
}
}
}