All Downloads are FREE. Search and download functionalities are using the official Maven repository.

com.sun.ejb.containers.util.MethodMap Maven / Gradle / Ivy

The newest version!
/*
 * Copyright (c) 2022 Contributors to the Eclipse Foundation
 * Copyright (c) 1997, 2018 Oracle and/or its affiliates. All rights reserved.
 *
 * This program and the accompanying materials are made available under the
 * terms of the Eclipse Public License v. 2.0, which is available at
 * http://www.eclipse.org/legal/epl-2.0.
 *
 * This Source Code may also be made available under the following Secondary
 * Licenses when the conditions for such availability set forth in the
 * Eclipse Public License v. 2.0 are satisfied: GNU General Public License,
 * version 2 with the GNU Classpath Exception, which is available at
 * https://www.gnu.org/software/classpath/license.html.
 *
 * SPDX-License-Identifier: EPL-2.0 OR GPL-2.0 WITH Classpath-exception-2.0
 */

package com.sun.ejb.containers.util;

import com.sun.ejb.InvocationInfo;

import java.lang.reflect.Method;
import java.util.HashMap;
import java.util.HashSet;
import java.util.Map;
import java.util.Set;

/**
 * This is an optimized map for resolving {@link Method} objects.
 * 

* Doing a method lookup, even on an unsynchronized Map, can be an * expensive operation, in many cases taking multiple microseconds. * In most situations this overhead is negligible, but it can be noticeable * when performed in the common path of a local ejb invocation, where our * goal is to be as fast as a raw java method call. *

* A MethodMap must be created with an existing Map and is immutable after * construction(except for clear()). *

* It does not support the optional Map operations * put, putAll, and remove. NOTE that these operations could * be implemented but are not necessary at this point since the main use * is for the container's method info, which is invariant after initialization. *

* As this is a map for Method objects, null keys are not supported. * This map is unsynchronized. */ public final class MethodMap extends HashMap { private static final long serialVersionUID = 1L; // If bucket size is not specified by caller, this is the number // of buckets per method that will be created. private static final int DEFAULT_BUCKET_MULTIPLIER = 20; private final int numBuckets_; // Sparse array of method info. Each element represents one method // or is null. Array is hashed by a combination of the // method name's hashcode and its parameter length. See // getBucket() below for more details. // // Note that reference equality is not very useful on Method since // it defines the equals() method and each call to Class.getMethods() // returns new Method instances. private MethodInfo[] methodInfo_; public MethodMap(Map methodMap) { super(methodMap); numBuckets_ = methodMap.size() * DEFAULT_BUCKET_MULTIPLIER; buildLookupTable(methodMap); } public MethodMap(Map methodMap, int numBuckets) { super(methodMap); if (numBuckets <= 0) { throw new IllegalArgumentException("Invalid value of numBuckets = " + numBuckets); } numBuckets_ = numBuckets; buildLookupTable(methodMap); } @Override public InvocationInfo put(Method key, InvocationInfo value) { throw new UnsupportedOperationException(); } @Override public void putAll(Map t) { throw new UnsupportedOperationException(); } @Override public InvocationInfo remove(Object key) { throw new UnsupportedOperationException(); } @Override public InvocationInfo get(Object key) { if (key instanceof Method) { Method m = (Method) key; Class[] paramTypes = m.getParameterTypes(); return get(m, paramTypes.length); } return null; } public InvocationInfo get(Method m, int numParams) { if (methodInfo_ == null) { return null; } else if (numParams < 0) { throw new IllegalStateException("invalid numParams = " + numParams); } InvocationInfo value = null; MethodInfo methodInfo = methodInfo_[getBucket(m, numParams)]; if (methodInfo != null) { // Declaring classes must be the same for methods to be equal. if (methodInfo.declaringClass == m.getDeclaringClass()) { value = methodInfo.value; } } return value == null ? super.get(m) : value; } @Override public void clear() { if (methodInfo_ != null) { methodInfo_ = null; super.clear(); } } private void buildLookupTable(Map methodMap) { methodInfo_ = new MethodInfo[numBuckets_]; Set occupied = new HashSet<>(); for (Entry entry : methodMap.entrySet()) { Object nextObj = entry.getKey(); Method next = null; if (nextObj == null) { throw new IllegalStateException("null keys not supported"); } else if (nextObj instanceof Method) { next = (Method) nextObj; } else { throw new IllegalStateException( "invalid key type = " + nextObj.getClass() + " key must be of type java.lang.reflect.Method"); } int bucket = getBucket(next); if (occupied.contains(bucket)) { // there's a clash for this bucket, so null it out and // defer to backing HashMap for results. methodInfo_[bucket] = null; } else { MethodInfo methodInfo = new MethodInfo(); methodInfo.value = entry.getValue(); // cache declaring class so we can avoid the method call // during lookup operation. methodInfo.declaringClass = next.getDeclaringClass(); methodInfo_[bucket] = methodInfo; occupied.add(bucket); } } } private int getBucket(Method m) { // note : getParameterTypes is guaranteed to be 0-length array // (as opposed to null) for a method with no arguments. Class[] paramTypes = m.getParameterTypes(); return getBucket(m, paramTypes.length); } private int getBucket(Method m, int numParams) { String methodName = m.getName(); // The normal Method.hashCode() method makes 5 method calls // and does not cache the result. Here, we use the method name's // hashCode since String.hashCode() makes 0 method calls *and* caches // the result. The tradeoff is that using only method name will // not account for overloaded methods, so we also add the number of // parameters to the calculation. In many cases, the caller // already knows the number of parameters, so it can be passed in // to the lookup. This gives up some encapsulation for // speed. It will result in better performance because // we can skip the call to m.getClass().getParameterTypes(), // which results in multiple method calls and can involve some // expensive copying depending of the types themselves. // Of course, this still won't account for the case where methods // are overloaded with the same number of parameters but different // types. However, the cache miss penalty should be small enough // in this case that it's a fair tradeoff. Adding anything else // to the hashcode calculation will have too large an impact on the // common case. int hashCode = methodName.hashCode(); // account for negative hashcodes hashCode = (hashCode >= 0) ? hashCode : (hashCode * -1); hashCode = (hashCode > numParams) ? (hashCode - numParams) : (hashCode + numParams); return (hashCode % numBuckets_); } private static class MethodInfo { public Class declaringClass; public InvocationInfo value; } }





© 2015 - 2024 Weber Informatics LLC | Privacy Policy