/**
* Copyright 2011 John DeRegnaucourt (jdereg@gmail.com)
*
* Licensed 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 hudson.util;
import java.lang.reflect.Array;
import java.lang.reflect.Field;
import java.lang.reflect.Modifier;
import java.util.ArrayList;
import java.util.Collection;
import java.util.HashMap;
import java.util.HashSet;
import java.util.Iterator;
import java.util.LinkedList;
import java.util.Map;
import java.util.Set;
import java.util.SortedMap;
import java.util.SortedSet;
import java.util.concurrent.ConcurrentHashMap;
/**
* Test two objects for equivalence with a 'deep' comparison. This will traverse
* the Object graph and perform either a field-by-field comparison on each
* object (if no .equals() method has been overridden from Object), or it
* will call the customized .equals() method if it exists. This method will
* allow object graphs loaded at different times (with different object ids)
* to be reliably compared. Object.equals() / Object.hashCode() rely on the
* object's identity, which would not consider two equivalent objects necessarily
* equals. This allows graphs containing instances of Classes that did not
* overide .equals() / .hashCode() to be compared. For example, testing for
* existence in a cache. Relying on an object's identity will not locate an
* equivalent object in a cache.<br/><br/>
*
* This method will handle cycles correctly, for example A->B->C->A. Suppose a and
* a' are two separate instances of A with the same values for all fields on
* A, B, and C. Then a.deepEquals(a') will return true. It uses cycle detection
* storing visited objects in a Set to prevent endless loops.
*
* @author John DeRegnaucourt (jdereg@gmail.com)
*/
public class DeepEquals
{
private static final Map<Class, Boolean> _customEquals = new ConcurrentHashMap<Class, Boolean>();
private static final Map<Class, Boolean> _customHash = new ConcurrentHashMap<Class, Boolean>();
private static final Map<Class, Collection<Field>> _reflectedFields = new ConcurrentHashMap<Class, Collection<Field>>();
private static class DualKey
{
private final Object _key1;
private final Object _key2;
private DualKey(Object k1, Object k2)
{
_key1 = k1;
_key2 = k2;
}
public boolean equals(Object other)
{
if (other == null)
{
return false;
}
if (!(other instanceof DualKey))
{
return false;
}
DualKey that = (DualKey) other;
return _key1 == that._key1 && _key2 == that._key2;
}
public int hashCode()
{
int h1 = _key1 != null ? _key1.hashCode() : 0;
int h2 = _key2 != null ? _key2.hashCode() : 0;
return h1 + h2;
}
}
/**
* Compare two objects with a 'deep' comparison. This will traverse the
* Object graph and perform either a field-by-field comparison on each
* object (if no .equals() method has been overridden from Object), or it
* will call the customized .equals() method if it exists. This method will
* allow object graphs loaded at different times (with different object ids)
* to be reliably compared. Object.equals() / Object.hashCode() rely on the
* object's identity, which would not consider to equivalent objects necessarily
* equals. This allows graphs containing instances of Classes that did no
* overide .equals() / .hashCode() to be compared. For example, testing for
* existence in a cache. Relying on an objects identity will not locate an
* object in cache, yet relying on it being equivalent will.<br/><br/>
*
* This method will handle cycles correctly, for example A->B->C->A. Suppose a and
* a' are two separate instances of the A with the same values for all fields on
* A, B, and C. Then a.deepEquals(a') will return true. It uses cycle detection
* storing visited objects in a Set to prevent endless loops.
* @param a Object one to compare
* @param b Object two to compare
* @return true if a is equivalent to b, false otherwise. Equivalent means that
* all field values of both subgraphs are the same, either at the field level
* or via the respectively encountered overridden .equals() methods during
* traversal.
*/
public static boolean deepEquals(Object a, Object b)
{
Set visited = new HashSet<DualKey>();
LinkedList<DualKey> stack = new LinkedList<DualKey>();
stack.addFirst(new DualKey(a, b));
while (!stack.isEmpty())
{
DualKey dualKey = stack.removeFirst();
visited.add(dualKey);
if (dualKey._key1 == dualKey._key2)
{ // Same instance is always equal to itself.
continue;
}
if (dualKey._key1 == null || dualKey._key2 == null)
{ // If either one is null, not equal (both can't be null, due to above comparison).
return false;
}
if (!dualKey._key1.getClass().equals(dualKey._key2.getClass()))
{ // Must be same class
return false;
}
// Handle all [] types. In order to be equal, the arrays must be the same
// length, be of the same type, be in the same order, and all elements within
// the array must be deeply equivalent.
if (dualKey._key1.getClass().isArray())
{
if (!compareArrays(dualKey._key1, dualKey._key2, stack, visited))
{
return false;
}
continue;
}
// Special handle SortedSets because they are fast to compare because their
// elements must be in the same order to be equivalent Sets.
if (dualKey._key1 instanceof SortedSet)
{
if (!compareOrderedCollection((Collection) dualKey._key1, (Collection) dualKey._key2, stack, visited))
{
return false;
}
continue;
}
// Handled unordered Sets. This is a slightly more expensive comparison because order cannot
// be assumed, a temporary Map must be created, however the comparison still runs in O(N) time.
if (dualKey._key1 instanceof Set)
{
if (!compareUnorderedCollection((Collection) dualKey._key1, (Collection) dualKey._key2, stack, visited))
{
return false;
}
continue;
}
// Check any Collection that is not a Set. In these cases, element order
// matters, therefore this comparison is faster than using unordered comparison.
if (dualKey._key1 instanceof Collection)
{
if (!compareOrderedCollection((Collection) dualKey._key1, (Collection) dualKey._key2, stack, visited))
{
return false;
}
continue;
}
// Compare two SortedMaps. This takes advantage of the fact that these
// Maps can be compared in O(N) time due to their ordering.
if (dualKey._key1 instanceof SortedMap)
{
if (!compareSortedMap((SortedMap) dualKey._key1, (SortedMap) dualKey._key2, stack, visited))
{
return false;
}
continue;
}
// Compare two Unordered Maps. This is a slightly more expensive comparison because
// order cannot be assumed, therefore a temporary Map must be created, however the
// comparison still runs in O(N) time.
if (dualKey._key1 instanceof Map)
{
if (!compareUnorderedMap((Map) dualKey._key1, (Map) dualKey._key2, stack, visited))
{
return false;
}
continue;
}
if (hasCustomEquals(dualKey._key1.getClass()))
{
if (!dualKey._key1.equals(dualKey._key2))
{
return false;
}
continue;
}
Collection<Field> fields = getDeepDeclaredFields(dualKey._key1.getClass());
for (Field field : fields)
{
try
{
DualKey dk = new DualKey(field.get(dualKey._key1), field.get(dualKey._key2));
if (!visited.contains(dk))
{
stack.addFirst(dk);
}
}
catch (Exception ignored)
{ }
}
}
return true;
}
/**
* Deeply compare to Arrays []. Both arrays must be of the same type, same length, and all
* elements within the arrays must be deeply equal in order to return true.
* @param array1 [] type (Object[], String[], etc.)
* @param array2 [] type (Object[], String[], etc.)
* @param stack add items to compare to the Stack (Stack versus recursion)
* @param visited Set of objects already compared (prevents cycles)
* @return true if the two arrays are the same length and contain deeply equivalent items.
*/
private static boolean compareArrays(Object array1, Object array2, LinkedList stack, Set visited)
{
// Same instance check already performed...
int len = Array.getLength(array1);
if (len != Array.getLength(array2))
{
return false;
}
for (int i = 0; i < len; i++)
{
DualKey dk = new DualKey(Array.get(array1, i), Array.get(array2, i));
if (!visited.contains(dk))
{ // push contents for further comparison
stack.addFirst(dk);
}
}
return true;
}
/**
* Deeply compare two Collections that must be same length and in same order.
* @param col1 First collection of items to compare
* @param col2 Second collection of items to compare
* @param stack add items to compare to the Stack (Stack versus recursion)
* @param visited Set of objects already compared (prevents cycles)
* value of 'true' indicates that the Collections may be equal, and the sets
* items will be added to the Stack for further comparison.
*/
private static boolean compareOrderedCollection(Collection col1, Collection col2, LinkedList stack, Set visited)
{
// Same instance check already performed...
if (col1.size() != col2.size())
{
return false;
}
Iterator i1 = col1.iterator();
Iterator i2 = col2.iterator();
while (i1.hasNext())
{
DualKey dk = new DualKey(i1.next(), i2.next());
if (!visited.contains(dk))
{ // push contents for further comparison
stack.addFirst(dk);
}
}
return true;
}
/**
* Deeply compare the two sets referenced by dualKey. This method attempts
* to quickly determine inequality by length, then if lengths match, it
* places one collection into a temporary Map by deepHashCode(), so that it
* can walk the other collection and look for each item in the map, which
* runs in O(N) time, rather than an O(N^2) lookup that would occur if each
* item from collection one was scanned for in collection two.
* @param col1 First collection of items to compare
* @param col2 Second collection of items to compare
* @param stack add items to compare to the Stack (Stack versus recursion)
* @param visited Set containing items that have already been compared,
* so as to prevent cycles.
* @return boolean false if the Collections are for certain not equals. A
* value of 'true' indicates that the Collections may be equal, and the sets
* items will be added to the Stack for further comparison.
*/
private static boolean compareUnorderedCollection(Collection col1, Collection col2, LinkedList stack, Set visited)
{
// Same instance check already performed...
if (col1.size() != col2.size())
{
return false;
}
Map fastLookup = new HashMap();
for (Object o : col2)
{
fastLookup.put(deepHashCode(o), o);
}
for (Object o : col1)
{
Object other = fastLookup.get(deepHashCode(o));
if (other == null)
{ // Item not even found in other Collection, no need to continue.
return false;
}
DualKey dk = new DualKey(o, other);
if (!visited.contains(dk))
{ // Place items on 'stack' for further comparison.
stack.addFirst(dk);
}
}
return true;
}
/**
* Deeply compare two SortedMap instances. This method walks the Maps in order,
* taking advantage of the fact that they Maps are SortedMaps.
* @param map1 SortedMap one
* @param map2 SortedMap two
* @param stack add items to compare to the Stack (Stack versus recursion)
* @param visited Set containing items that have already been compared, to prevent cycles.
* @return false if the Maps are for certain not equals. 'true' indicates that 'on the surface' the maps
* are equal, however, it will place the contents of the Maps on the stack for further comparisons.
*/
private static boolean compareSortedMap(SortedMap map1, SortedMap map2, LinkedList stack, Set visited)
{
// Same instance check already performed...
if (map1.size() != map2.size())
{
return false;
}
Iterator i1 = map1.entrySet().iterator();
Iterator i2 = map2.entrySet().iterator();
while (i1.hasNext())
{
Map.Entry entry1 = (Map.Entry)i1.next();
Map.Entry entry2 = (Map.Entry)i2.next();
// Must split the Key and Value so that Map.Entry's equals() method is not used.
DualKey dk = new DualKey(entry1.getKey(), entry2.getKey());
if (!visited.contains(dk))
{ // Push Keys for further comparison
stack.addFirst(dk);
}
dk = new DualKey(entry1.getValue(), entry2.getValue());
if (!visited.contains(dk))
{ // Push values for further comparison
stack.addFirst(dk);
}
}
return true;
}
/**
* Deeply compare two Map instances. After quick short-circuit tests, this method
* uses a temporary Map so that this method can run in O(N) time.
* @param map1 Map one
* @param map2 Map two
* @param stack add items to compare to the Stack (Stack versus recursion)
* @param visited Set containing items that have already been compared, to prevent cycles.
* @return false if the Maps are for certain not equals. 'true' indicates that 'on the surface' the maps
* are equal, however, it will place the contents of the Maps on the stack for further comparisons.
*/
private static boolean compareUnorderedMap(Map map1, Map map2, LinkedList stack, Set visited)
{
// Same instance check already performed...
if (map1.size() != map2.size())
{
return false;
}
Map fastLookup = new HashMap();
for (Map.Entry entry : (Set<Map.Entry>)map2.entrySet())
{
fastLookup.put(deepHashCode(entry.getKey()), entry);
}
for (Map.Entry entry : (Set<Map.Entry>)map1.entrySet())
{
Map.Entry other = (Map.Entry)fastLookup.get(deepHashCode(entry.getKey()));
if (other == null)
{
return false;
}
DualKey dk = new DualKey(entry.getKey(), other.getKey());
if (!visited.contains(dk))
{ // Push keys for further comparison
stack.addFirst(dk);
}
dk = new DualKey(entry.getValue(), other.getValue());
if (!visited.contains(dk))
{ // Push values for further comparison
stack.addFirst(dk);
}
}
return true;
}
/**
* Determine if the passed in class has a non-Object.equals() method. This
* method caches its results in static ConcurrentHashMap to benefit
* execution performance.
* @param c Class to check.
* @return true, if the passed in Class has a .equals() method somewhere between
* itself and just below Object in it's inheritance.
*/
public static boolean hasCustomEquals(Class c)
{
Class origClass = c;
if (_customEquals.containsKey(c))
{
return _customEquals.get(c);
}
while (!Object.class.equals(c))
{
try
{
c.getDeclaredMethod("equals", Object.class);
_customEquals.put(origClass, true);
return true;
}
catch (Exception ignored) { }
c = c.getSuperclass();
}
_customEquals.put(origClass, false);
return false;
}
/**
* Get a deterministic hashCode (int) value for an Object, regardless of
* when it was created or where it was loaded into memory. The problem
* with java.lang.Object.hashCode() is that it essentially relies on
* memory location of an object (what identity it was assigned), whereas
* this method will produce the same hashCode for any object graph, regardless
* of how many times it is created.<br/><br/>
*
* This method will handle cycles correctly (A->B->C->A). In this case,
* Starting with object A, B, or C would yield the same hashCode. If an
* object encountered (root, suboject, etc.) has a hashCode() method on it
* (that is not Object.hashCode()), that hashCode() method will be called
* and it will stop traversal on that branch.
* @param obj Object who hashCode is desired.
* @return the 'deep' hashCode value for the passed in object.
*/
public static int deepHashCode(Object obj)
{
Set visited = new HashSet();
LinkedList<Object> stack = new LinkedList<Object>();
stack.addFirst(obj);
int hash = 0;
while (!stack.isEmpty())
{
obj = stack.removeFirst();
if (obj == null || visited.contains(obj))
{
continue;
}
visited.add(obj);
if (obj.getClass().isArray())
{
int len = Array.getLength(obj);
for (int i = 0; i < len; i++)
{
stack.addFirst(Array.get(obj, i));
}
continue;
}
if (obj instanceof Collection)
{
stack.addAll(0, (Collection)obj);
continue;
}
if (obj instanceof Map)
{
stack.addAll(0, ((Map)obj).keySet());
stack.addAll(0, ((Map)obj).values());
continue;
}
if (hasCustomHashCode(obj.getClass()))
{ // A real hashCode() method exists, call it.
hash += obj.hashCode();
continue;
}
Collection<Field> fields = getDeepDeclaredFields(obj.getClass());
for (Field field : fields)
{
try
{
stack.addFirst(field.get(obj));
}
catch (Exception ignored) { }
}
}
return hash;
}
/**
* Determine if the passed in class has a non-Object.hashCode() method. This
* method caches its results in static ConcurrentHashMap to benefit
* execution performance.
* @param c Class to check.
* @return true, if the passed in Class has a .hashCode() method somewhere between
* itself and just below Object in it's inheritance.
*/
public static boolean hasCustomHashCode(Class c)
{
Class origClass = c;
if (_customHash.containsKey(c))
{
return _customHash.get(c);
}
while (!Object.class.equals(c))
{
try
{
c.getDeclaredMethod("hashCode");
_customHash.put(origClass, true);
return true;
}
catch (Exception ignored) { }
c = c.getSuperclass();
}
_customHash.put(origClass, false);
return false;
}
/**
* Get all non static, non transient, fields of the passed in class.
* The special this$ field is also not returned. The result is cached
* in a static ConcurrentHashMap to benefit execution performance.
* @param c Class instance
* @return Collection of only the fields in the passed in class
* that would need further processing (reference fields). This
* makes field traversal on a class faster as it does not need to
* continually process known fields like primitives.
*/
public static Collection<Field> getDeepDeclaredFields(Class c)
{
if (_reflectedFields.containsKey(c))
{
return _reflectedFields.get(c);
}
Collection<Field> fields = new ArrayList<Field>();
Class curr = c;
while (curr != null)
{
try
{
Field[] local = curr.getDeclaredFields();
for (Field field : local)
{
if (!field.isAccessible())
{
try
{
field.setAccessible(true);
}
catch (Exception ignored) { }
}
int modifiers = field.getModifiers();
if (!Modifier.isStatic(modifiers) &&
!field.getName().startsWith("this$") &&
!Modifier.isTransient(modifiers))
{ // speed up: do not count static fields, not go back up to enclosing object in nested case
fields.add(field);
}
}
}
catch (ThreadDeath t)
{
throw t;
}
catch (Throwable ignored)
{ }
curr = curr.getSuperclass();
}
_reflectedFields.put(c, fields);
return fields;
}
}