org.apache.commons.configuration2.tree.InMemoryNodeModel Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of commons-configuration2 Show documentation
Show all versions of commons-configuration2 Show documentation
Tools to assist in the reading of configuration/preferences files in
various formats
The newest version!
/*
* 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.commons.configuration2.tree;
import java.util.ArrayList;
import java.util.Collection;
import java.util.Collections;
import java.util.HashMap;
import java.util.Iterator;
import java.util.LinkedList;
import java.util.List;
import java.util.Map;
import java.util.concurrent.atomic.AtomicReference;
import org.apache.commons.configuration2.ex.ConfigurationRuntimeException;
import org.apache.commons.lang3.mutable.Mutable;
import org.apache.commons.lang3.mutable.MutableObject;
/**
*
* A specialized node model implementation which operates on {@link ImmutableNode} structures.
*
*
* This {@code NodeModel} implementation keeps all its data as a tree of {@link ImmutableNode} objects in memory. The
* managed structure can be manipulated in a thread-safe, non-blocking way. This is achieved by using atomic variables:
* The root of the tree is stored in an atomic reference variable. Each update operation causes a new structure to be
* constructed (which reuses as much from the original structure as possible). The old root node is then replaced by the
* new one using an atomic compare-and-set operation. If this fails, the manipulation has to be done anew on the updated
* structure.
*
*
* @since 2.0
*/
public class InMemoryNodeModel implements NodeModel {
/**
* An interface used internally for handling concurrent updates. An implementation has to populate the passed in
* {@code ModelTransaction}. The transaction is then executed, and an atomic update of the model's {@code TreeData} is
* attempted. If this fails - because another update came across -, the whole operation has to be tried anew.
*/
private interface TransactionInitializer {
/**
* Initializes the specified transaction for an update operation. The return value indicates whether the transaction
* should be executed. A result of false means that the update is to be aborted (maybe another update method was
* called).
*
* @param tx the transaction to be initialized
* @return a flag whether the update should continue
*/
boolean initTransaction(ModelTransaction tx);
}
/**
* A dummy node handler instance used in operations which require only a limited functionality.
*/
private static final NodeHandler DUMMY_HANDLER = new TreeData(null, Collections.emptyMap(),
Collections.emptyMap(), null, new ReferenceTracker());
/**
* Handles an add property operation if the property to be added is an attribute.
*
* @param tx the transaction
* @param addData the {@code NodeAddData}
* @param values the collection with node values
*/
private static void addAttributeProperty(final ModelTransaction tx, final NodeAddData addData, final Iterable values) {
if (addData.getPathNodes().isEmpty()) {
tx.addAttributeOperation(addData.getParent(), addData.getNewNodeName(), values.iterator().next());
} else {
final int pathNodeCount = addData.getPathNodes().size();
final ImmutableNode childWithAttribute = new ImmutableNode.Builder().name(addData.getPathNodes().get(pathNodeCount - 1))
.addAttribute(addData.getNewNodeName(), values.iterator().next()).create();
final ImmutableNode newChild = pathNodeCount > 1
? createNodeOnPath(addData.getPathNodes().subList(0, pathNodeCount - 1).iterator(), Collections.singleton(childWithAttribute))
: childWithAttribute;
tx.addAddNodeOperation(addData.getParent(), newChild);
}
}
/**
* Handles an add property operation if the property to be added is a node.
*
* @param tx the transaction
* @param addData the {@code NodeAddData}
* @param values the collection with node values
*/
private static void addNodeProperty(final ModelTransaction tx, final NodeAddData addData, final Iterable values) {
final Collection newNodes = createNodesToAdd(addData.getNewNodeName(), values);
addNodesByAddData(tx, addData, newNodes);
}
/**
* Initializes a transaction to add a collection of nodes as described by a {@code NodeAddData} object. If necessary,
* new path nodes are created. Eventually, the new nodes are added as children to the specified target node.
*
* @param tx the transaction
* @param addData the {@code NodeAddData}
* @param newNodes the collection of new child nodes
*/
private static void addNodesByAddData(final ModelTransaction tx, final NodeAddData addData, final Collection newNodes) {
if (addData.getPathNodes().isEmpty()) {
tx.addAddNodesOperation(addData.getParent(), newNodes);
} else {
final ImmutableNode newChild = createNodeToAddWithPath(addData, newNodes);
tx.addAddNodeOperation(addData.getParent(), newChild);
}
}
/**
* Creates an exception referring to an invalid key for adding properties. Such an exception is thrown when an operation
* tries to add something to an attribute.
*
* @param key the invalid key causing this exception
* @return the exception
*/
private static IllegalArgumentException attributeKeyException(final String key) {
return new IllegalArgumentException("New nodes cannot be added to an attribute key: " + key);
}
/**
* Checks if the passed in node is defined. Result is true if the node contains any data.
*
* @param node the node in question
* @return true if the node is defined, false otherwise
*/
static boolean checkIfNodeDefined(final ImmutableNode node) {
return node.getValue() != null || !node.getChildren().isEmpty() || !node.getAttributes().isEmpty();
}
/**
* Creates a new data object with a tracked child node of the given parent node. If such a child node already exists, it
* is used. Otherwise, a new one is created.
*
* @param current the current {@code TreeData} object
* @param parent the parent node
* @param childName the name of the child node
* @param resolver the {@code NodeKeyResolver}
* @param refSelector here the newly created {@code NodeSelector} is returned
* @return the new {@code TreeData} instance
*/
private static TreeData createDataWithTrackedChildNode(final TreeData current, final ImmutableNode parent, final String childName,
final NodeKeyResolver resolver, final MutableObject refSelector) {
final TreeData newData;
final List namedChildren = current.getChildren(parent, childName);
if (!namedChildren.isEmpty()) {
newData = updateDataWithNewTrackedNode(current, namedChildren.get(0), resolver, refSelector);
} else {
final ImmutableNode child = new ImmutableNode.Builder().name(childName).create();
final ModelTransaction tx = new ModelTransaction(current, null, resolver);
tx.addAddNodeOperation(parent, child);
newData = updateDataWithNewTrackedNode(tx.execute(), child, resolver, refSelector);
}
return newData;
}
/**
* Recursive helper method for creating a path node for an add operation. All path nodes except for the last have a
* single child. The last path node has the new nodes as children.
*
* @param it the iterator over the names of the path nodes
* @param newNodes the collection of new child nodes
* @return the newly created path node
*/
private static ImmutableNode createNodeOnPath(final Iterator it, final Collection newNodes) {
final String nodeName = it.next();
final ImmutableNode.Builder builder;
if (it.hasNext()) {
builder = new ImmutableNode.Builder(1);
builder.addChild(createNodeOnPath(it, newNodes));
} else {
builder = new ImmutableNode.Builder(newNodes.size());
builder.addChildren(newNodes);
}
return builder.name(nodeName).create();
}
/**
* Creates a collection with new nodes with a given name and a value from a given collection.
*
* @param newNodeName the name of the new nodes
* @param values the collection with node values
* @return the newly created collection
*/
private static Collection createNodesToAdd(final String newNodeName, final Iterable values) {
final Collection nodes = new LinkedList<>();
values.forEach(value -> nodes.add(new ImmutableNode.Builder().name(newNodeName).value(value).create()));
return nodes;
}
/**
* Creates a node structure consisting of the path nodes defined by the passed in {@code NodeAddData} instance and all
* new child nodes.
*
* @param addData the {@code NodeAddData}
* @param newNodes the collection of new child nodes
* @return the parent node of the newly created hierarchy
*/
private static ImmutableNode createNodeToAddWithPath(final NodeAddData addData, final Collection newNodes) {
return createNodeOnPath(addData.getPathNodes().iterator(), newNodes);
}
/**
* Creates tracked node entries for the specified nodes and creates the corresponding selectors.
*
* @param refSelectors the reference where to store the selectors
* @param nodes the nodes to be tracked
* @param current the current {@code TreeData} object
* @param resolver the {@code NodeKeyResolver}
* @return the updated {@code TreeData} object
*/
private static TreeData createSelectorsForTrackedNodes(final Mutable> refSelectors, final List nodes,
final TreeData current, final NodeKeyResolver resolver) {
final List selectors = new ArrayList<>(nodes.size());
final Map cache = new HashMap<>();
nodes.forEach(node -> selectors.add(new NodeSelector(resolver.nodeKey(node, cache, current))));
refSelectors.setValue(selectors);
final NodeTracker newTracker = current.getNodeTracker().trackNodes(selectors, nodes);
return current.updateNodeTracker(newTracker);
}
/**
* Determines the name of the root node for a merge operation. If a root name is provided, it is used. Otherwise, if the
* current root node has no name, the name of the node to be merged is used. A result of null means that no node
* name has to be set.
*
* @param rootNode the current root node
* @param node the node to be merged with the root node
* @param rootName the name of the resulting node
* @return the new name of the root node
*/
private static String determineRootName(final ImmutableNode rootNode, final ImmutableNode node, final String rootName) {
if (rootName != null) {
return rootName;
}
if (rootNode.getNodeName() == null) {
return node.getNodeName();
}
return null;
}
/**
* Initializes a transaction to clear the values of a property based on the passed in collection of affected results.
*
* @param tx the transaction to be initialized
* @param results a collection with results pointing to the nodes to be cleared
* @return a flag whether there are elements to be cleared
*/
private static boolean initializeClearTransaction(final ModelTransaction tx, final Collection> results) {
results.forEach(result -> {
if (result.isAttributeResult()) {
tx.addRemoveAttributeOperation(result.getNode(), result.getAttributeName());
} else {
tx.addClearNodeValueOperation(result.getNode());
}
});
return !results.isEmpty();
}
/**
* Initializes a transaction to change the values of some query results based on the passed in map.
*
* @param tx the transaction to be initialized
* @param changedValues the map defining the elements to be changed
* @return a flag whether there are elements to be updated
*/
private static boolean initializeUpdateTransaction(final ModelTransaction tx, final Map, Object> changedValues) {
changedValues.forEach((k, v) -> {
final ImmutableNode node = k.getNode();
if (k.isAttributeResult()) {
tx.addAttributeOperation(node, k.getAttributeName(), v);
} else {
tx.addChangeNodeValueOperation(node, v);
}
});
return !changedValues.isEmpty();
}
/**
* Determines the initial root node of this model. If a root node has been provided, it is used. Otherwise, an empty
* dummy root node is created.
*
* @param providedRoot the passed in root node
* @return the root node to be used
*/
private static ImmutableNode initialRootNode(final ImmutableNode providedRoot) {
return providedRoot != null ? providedRoot : new ImmutableNode.Builder().create();
}
/**
* Adds a tracked node that has already been resolved to the specified data object.
*
* @param current the current {@code TreeData} object
* @param node the node in question
* @param resolver the {@code NodeKeyResolver}
* @param refSelector here the newly created {@code NodeSelector} is returned
* @return the new {@code TreeData} instance
*/
private static TreeData updateDataWithNewTrackedNode(final TreeData current, final ImmutableNode node, final NodeKeyResolver resolver,
final MutableObject refSelector) {
final NodeSelector selector = new NodeSelector(resolver.nodeKey(node, new HashMap<>(), current));
refSelector.setValue(selector);
final NodeTracker newTracker = current.getNodeTracker().trackNodes(Collections.singleton(selector), Collections.singleton(node));
return current.updateNodeTracker(newTracker);
}
/**
* Updates the mapping from nodes to their parents for the passed in hierarchy of nodes. This method traverses all
* children and grand-children of the passed in root node. For each node in the subtree the parent relation is added to
* the map.
*
* @param parents the map with parent nodes
* @param root the root node of the current tree
*/
static void updateParentMapping(final Map parents, final ImmutableNode root) {
NodeTreeWalker.INSTANCE.walkBFS(root, new ConfigurationNodeVisitorAdapter() {
@Override
public void visitBeforeChildren(final ImmutableNode node, final NodeHandler handler) {
node.forEach(c -> parents.put(c, node));
}
}, DUMMY_HANDLER);
}
/**
* Checks whether the specified collection with values is not empty.
*
* @param values the collection with node values
* @return true if values are provided, false otherwise
*/
private static boolean valuesNotEmpty(final Iterable values) {
return values.iterator().hasNext();
}
/** Stores information about the current nodes structure. */
private final AtomicReference structure;
/**
* Creates a new instance of {@code InMemoryNodeModel} which is initialized with an empty root node.
*/
public InMemoryNodeModel() {
this(null);
}
/**
* Creates a new instance of {@code InMemoryNodeModel} and initializes it from the given root node. If the passed in
* node is null, a new, empty root node is created.
*
* @param root the new root node for this model
*/
public InMemoryNodeModel(final ImmutableNode root) {
structure = new AtomicReference<>(createTreeData(initialRootNode(root), null));
}
@Override
public void addNodes(final String key, final Collection nodes, final NodeKeyResolver resolver) {
addNodes(key, null, nodes, resolver);
}
/**
* Adds new nodes using a tracked node as root node. This method works like the normal {@code addNodes()} method, but
* the origin of the operation (also for the interpretation of the passed in key) is a tracked node identified by the
* passed in {@code NodeSelector}. The selector can be null, then the root node is assumed.
*
* @param key the key
* @param selector the {@code NodeSelector} defining the root node (or null)
* @param nodes the collection of new nodes to be added
* @param resolver the {@code NodeKeyResolver}
* @throws ConfigurationRuntimeException if the selector cannot be resolved
*/
public void addNodes(final String key, final NodeSelector selector, final Collection nodes,
final NodeKeyResolver resolver) {
if (nodes != null && !nodes.isEmpty()) {
updateModel(tx -> {
final List> results = resolver.resolveKey(tx.getQueryRoot(), key, tx.getCurrentData());
if (results.size() == 1) {
if (results.get(0).isAttributeResult()) {
throw attributeKeyException(key);
}
tx.addAddNodesOperation(results.get(0).getNode(), nodes);
} else {
final NodeAddData addData = resolver.resolveAddKey(tx.getQueryRoot(), key, tx.getCurrentData());
if (addData.isAttribute()) {
throw attributeKeyException(key);
}
final ImmutableNode newNode = new ImmutableNode.Builder(nodes.size()).name(addData.getNewNodeName()).addChildren(nodes).create();
addNodesByAddData(tx, addData, Collections.singleton(newNode));
}
return true;
}, selector, resolver);
}
}
@Override
public void addProperty(final String key, final Iterable values, final NodeKeyResolver resolver) {
addProperty(key, null, values, resolver);
}
/**
* Adds new property values using a tracked node as root node. This method works like the normal {@code addProperty()}
* method, but the origin of the operation (also for the interpretation of the passed in key) is a tracked node
* identified by the passed in {@code NodeSelector}. The selector can be null, then the root node is assumed.
*
* @param key the key
* @param selector the {@code NodeSelector} defining the root node (or null)
* @param values the values to be added
* @param resolver the {@code NodeKeyResolver}
* @throws ConfigurationRuntimeException if the selector cannot be resolved
*/
public void addProperty(final String key, final NodeSelector selector, final Iterable values, final NodeKeyResolver resolver) {
if (valuesNotEmpty(values)) {
updateModel(tx -> {
initializeAddTransaction(tx, key, values, resolver);
return true;
}, selector, resolver);
}
}
/**
* {@inheritDoc} A new empty root node is created with the same name as the current root node. Implementation note:
* Because this is a hard reset the usual dance for dealing with concurrent updates is not required here.
*
* @param resolver the {@code NodeKeyResolver}
*/
@Override
public void clear(final NodeKeyResolver resolver) {
final ImmutableNode newRoot = new ImmutableNode.Builder().name(getRootNode().getNodeName()).create();
setRootNode(newRoot);
}
/**
* {@inheritDoc} If this operation leaves an affected node in an undefined state, it is removed from the model.
*/
@Override
public void clearProperty(final String key, final NodeKeyResolver resolver) {
clearProperty(key, null, resolver);
}
/**
* Clears a property using a tracked node as root node. This method works like the normal {@code clearProperty()}
* method, but the origin of the operation (also for the interpretation of the passed in key) is a tracked node
* identified by the passed in {@code NodeSelector}. The selector can be null, then the root node is assumed.
*
* @param key the key
* @param selector the {@code NodeSelector} defining the root node (or null)
* @param resolver the {@code NodeKeyResolver}
* @throws ConfigurationRuntimeException if the selector cannot be resolved
*/
public void clearProperty(final String key, final NodeSelector selector, final NodeKeyResolver resolver) {
updateModel(tx -> {
final List> results = resolver.resolveKey(tx.getQueryRoot(), key, tx.getCurrentData());
return initializeClearTransaction(tx, results);
}, selector, resolver);
}
/**
* {@inheritDoc} This implementation checks whether nodes become undefined after subtrees have been removed. If this is
* the case, such nodes are removed, too. Return value is a collection with {@code QueryResult} objects for the elements
* to be removed from the model.
*/
@Override
public List> clearTree(final String key, final NodeKeyResolver resolver) {
return clearTree(key, null, resolver);
}
/**
* Clears a whole sub tree using a tracked node as root node. This method works like the normal {@code clearTree()}
* method, but the origin of the operation (also for the interpretation of the passed in key) is a tracked node
* identified by the passed in {@code NodeSelector}. The selector can be null, then the root node is assumed.
*
* @param key the key
* @param selector the {@code NodeSelector} defining the root node (or null)
* @param resolver the {@code NodeKeyResolver}
* @return a list with the results to be removed
* @throws ConfigurationRuntimeException if the selector cannot be resolved
*/
public List> clearTree(final String key, final NodeSelector selector, final NodeKeyResolver resolver) {
final List> removedElements = new LinkedList<>();
updateModel(tx -> {
boolean changes = false;
final TreeData currentStructure = tx.getCurrentData();
final List> results = resolver.resolveKey(tx.getQueryRoot(), key, currentStructure);
removedElements.clear();
removedElements.addAll(results);
for (final QueryResult result : results) {
if (result.isAttributeResult()) {
tx.addRemoveAttributeOperation(result.getNode(), result.getAttributeName());
} else {
if (result.getNode() == currentStructure.getRootNode()) {
// the whole model is to be cleared
clear(resolver);
return false;
}
tx.addRemoveNodeOperation(currentStructure.getParent(result.getNode()), result.getNode());
}
changes = true;
}
return changes;
}, selector, resolver);
return removedElements;
}
/**
* Creates the mapping to parent nodes for the nodes structured represented by the passed in root node. Each node is
* assigned its parent node. Here an iterative algorithm is used rather than a recursive one to avoid stack overflow for
* huge structures.
*
* @param root the root node of the structure
* @return the parent node mapping
*/
private Map createParentMapping(final ImmutableNode root) {
final Map parents = new HashMap<>();
updateParentMapping(parents, root);
return parents;
}
/**
* Creates a {@code TreeData} object for the specified root node.
*
* @param root the root node of the current tree
* @param current the current {@code TreeData} object (may be null)
* @return the {@code TreeData} describing the current tree
*/
private TreeData createTreeData(final ImmutableNode root, final TreeData current) {
final NodeTracker newTracker = current != null ? current.getNodeTracker().detachAllTrackedNodes() : new NodeTracker();
return createTreeDataForRootAndTracker(root, newTracker);
}
/**
* Creates a {@code TreeData} object for the specified root node and {@code NodeTracker}. Other parameters are set to
* default values.
*
* @param root the new root node for this model
* @param newTracker the new {@code NodeTracker}
* @return the new {@code TreeData} object
*/
private TreeData createTreeDataForRootAndTracker(final ImmutableNode root, final NodeTracker newTracker) {
return new TreeData(root, createParentMapping(root), Collections.emptyMap(), newTracker, new ReferenceTracker());
}
/**
* Executes a transaction on the current data of this model. This method is called if an operation is to be executed on
* the model's root node or a tracked node which is not yet detached.
*
* @param txInit the {@code TransactionInitializer}
* @param selector an optional {@code NodeSelector} defining the target node
* @param currentData the current data of the model
* @param resolver the {@code NodeKeyResolver}
* @return a flag whether the operation has been completed successfully
*/
private boolean executeTransactionOnCurrentStructure(final TransactionInitializer txInit, final NodeSelector selector, final TreeData currentData,
final NodeKeyResolver resolver) {
final boolean done;
final ModelTransaction tx = new ModelTransaction(currentData, selector, resolver);
if (!txInit.initTransaction(tx)) {
done = true;
} else {
final TreeData newData = tx.execute();
done = structure.compareAndSet(tx.getCurrentData(), newData);
}
return done;
}
/**
* Tries to execute a transaction on the model of a detached tracked node. This method checks whether the target node of
* the transaction is a tracked node and if this node is already detached. If this is the case, the update operation is
* independent on this model and has to be executed on the specific model for the detached node.
*
* @param txInit the {@code TransactionInitializer}
* @param selector an optional {@code NodeSelector} defining the target node
* @param currentData the current data of the model
* @param resolver the {@code NodeKeyResolver} @return a flag whether the transaction could be executed
* @throws ConfigurationRuntimeException if the selector cannot be resolved
*/
private boolean executeTransactionOnDetachedTrackedNode(final TransactionInitializer txInit, final NodeSelector selector, final TreeData currentData,
final NodeKeyResolver resolver) {
if (selector != null) {
final InMemoryNodeModel detachedNodeModel = currentData.getNodeTracker().getDetachedNodeModel(selector);
if (detachedNodeModel != null) {
detachedNodeModel.updateModel(txInit, null, resolver);
return true;
}
}
return false;
}
/**
* {@inheritDoc} This implementation simply returns the current root node of this model.
*/
@Override
public ImmutableNode getInMemoryRepresentation() {
return getTreeData().getRootNode();
}
/**
* {@inheritDoc} {@code InMemoryNodeModel} implements the {@code NodeHandler} interface itself. So this implementation
* just returns the this reference.
*/
@Override
public NodeHandler getNodeHandler() {
return getReferenceNodeHandler();
}
/**
* Gets a {@code ReferenceNodeHandler} object for this model. This extended node handler can be used to query
* references objects stored for this model.
*
* @return the {@code ReferenceNodeHandler}
*/
public ReferenceNodeHandler getReferenceNodeHandler() {
return getTreeData();
}
/**
* Gets the root node of this mode. Note: This method should be used with care. The model may be updated concurrently
* which causes the root node to be replaced. If the root node is to be processed further (e.g. by executing queries on
* it), the model should be asked for its {@code NodeHandler}, and the root node should be obtained from there. The
* connection between a node handler and its root node remain constant because an update of the model causes the whole
* node handler to be replaced.
*
* @return the current root node
*/
public ImmutableNode getRootNode() {
return getTreeData().getRootNode();
}
/**
* Gets the current {@code ImmutableNode} instance associated with the given {@code NodeSelector}. The node must be a
* tracked node, i.e. {@link #trackNode(NodeSelector, NodeKeyResolver)} must have been called before with the given
* selector.
*
* @param selector the {@code NodeSelector} defining the desired node
* @return the current {@code ImmutableNode} associated with this selector
* @throws ConfigurationRuntimeException if the selector is unknown
*/
public ImmutableNode getTrackedNode(final NodeSelector selector) {
return structure.get().getNodeTracker().getTrackedNode(selector);
}
/**
* Gets a {@code NodeHandler} for a tracked node. Such a handler may be required for operations on a sub tree of the
* model. The handler to be returned depends on the current state of the tracked node. If it is still active, a handler
* is used which shares some data (especially the parent mapping) with this model. Detached track nodes in contrast have
* their own separate model; in this case a handler associated with this model is returned.
*
* @param selector the {@code NodeSelector} defining the tracked node
* @return a {@code NodeHandler} for this tracked node
* @throws ConfigurationRuntimeException if the selector is unknown
*/
public NodeHandler getTrackedNodeHandler(final NodeSelector selector) {
final TreeData currentData = structure.get();
final InMemoryNodeModel detachedNodeModel = currentData.getNodeTracker().getDetachedNodeModel(selector);
return detachedNodeModel != null ? detachedNodeModel.getNodeHandler()
: new TrackedNodeHandler(currentData.getNodeTracker().getTrackedNode(selector), currentData);
}
/**
* Gets the current {@code TreeData} object. This object contains all information about the current node structure.
*
* @return the current {@code TreeData} object
*/
TreeData getTreeData() {
return structure.get();
}
/**
* Initializes a transaction for an add operation.
*
* @param tx the transaction to be initialized
* @param key the key
* @param values the collection with node values
* @param resolver the {@code NodeKeyResolver}
*/
private void initializeAddTransaction(final ModelTransaction tx, final String key, final Iterable values,
final NodeKeyResolver resolver) {
final NodeAddData addData = resolver.resolveAddKey(tx.getQueryRoot(), key, tx.getCurrentData());
if (addData.isAttribute()) {
addAttributeProperty(tx, addData, values);
} else {
addNodeProperty(tx, addData, values);
}
}
/**
* Returns a flag whether the specified tracked node is detached. As long as the {@code NodeSelector} associated with
* that node returns a single instance, the tracked node is said to be life. If now an update of the model
* happens which invalidates the selector (maybe the target node was removed), the tracked node becomes detached. It is
* still possible to query the node; here the latest valid instance is returned. But further changes on the node model
* are no longer tracked for this node. So even if there are further changes which would make the {@code NodeSelector}
* valid again, the tracked node stays in detached state.
*
* @param selector the {@code NodeSelector} defining the desired node
* @return a flag whether this tracked node is in detached state
* @throws ConfigurationRuntimeException if the selector is unknown
*/
public boolean isTrackedNodeDetached(final NodeSelector selector) {
return structure.get().getNodeTracker().isTrackedNodeDetached(selector);
}
/**
* Merges the root node of this model with the specified node. This method is typically caused by configuration
* implementations when a configuration source is loaded, and its data has to be added to the model. It is possible to
* define the new name of the root node and to pass in a map with reference objects.
*
* @param node the node to be merged with the root node
* @param rootName the new name of the root node; can be null, then the name of the root node is not changed
* unless it is null
* @param references an optional map with reference objects
* @param rootRef an optional reference object for the new root node
* @param resolver the {@code NodeKeyResolver}
*/
public void mergeRoot(final ImmutableNode node, final String rootName, final Map references, final Object rootRef,
final NodeKeyResolver resolver) {
updateModel(tx -> {
final TreeData current = tx.getCurrentData();
final String newRootName = determineRootName(current.getRootNode(), node, rootName);
if (newRootName != null) {
tx.addChangeNodeNameOperation(current.getRootNode(), newRootName);
}
tx.addAddNodesOperation(current.getRootNode(), node.getChildren());
tx.addAttributesOperation(current.getRootNode(), node.getAttributes());
if (node.getValue() != null) {
tx.addChangeNodeValueOperation(current.getRootNode(), node.getValue());
}
if (references != null) {
tx.addNewReferences(references);
}
if (rootRef != null) {
tx.addNewReference(current.getRootNode(), rootRef);
}
return true;
}, null, resolver);
}
/**
* Replaces an active tracked node. The node then becomes detached.
*
* @param currentData the current data of the model
* @param selector the {@code NodeSelector} defining the tracked node
* @param newNode the node replacing the tracked node
* @return a flag whether the operation was successful
*/
private boolean replaceActiveTrackedNode(final TreeData currentData, final NodeSelector selector, final ImmutableNode newNode) {
final NodeTracker newTracker = currentData.getNodeTracker().replaceAndDetachTrackedNode(selector, newNode);
return structure.compareAndSet(currentData, currentData.updateNodeTracker(newTracker));
}
/**
* Replaces a tracked node if it is already detached.
*
* @param currentData the current data of the model
* @param selector the {@code NodeSelector} defining the tracked node
* @param newNode the node replacing the tracked node
* @return a flag whether the operation was successful
*/
private boolean replaceDetachedTrackedNode(final TreeData currentData, final NodeSelector selector, final ImmutableNode newNode) {
final InMemoryNodeModel detachedNodeModel = currentData.getNodeTracker().getDetachedNodeModel(selector);
if (detachedNodeModel != null) {
detachedNodeModel.setRootNode(newNode);
return true;
}
return false;
}
/**
* Replaces the root node of this model. This method is similar to {@link #setRootNode(ImmutableNode)}; however, tracked
* nodes will not get lost. The model applies the selectors of all tracked nodes on the new nodes hierarchy, so that
* corresponding nodes are selected (this may cause nodes to become detached if a select operation fails). This
* operation is useful if the new nodes hierarchy to be set is known to be similar to the old one. Note that reference
* objects are lost; there is no way to automatically match nodes between the old and the new nodes hierarchy.
*
* @param newRoot the new root node to be set (must not be null)
* @param resolver the {@code NodeKeyResolver}
* @throws IllegalArgumentException if the new root node is null
*/
public void replaceRoot(final ImmutableNode newRoot, final NodeKeyResolver resolver) {
if (newRoot == null) {
throw new IllegalArgumentException("Replaced root node must not be null!");
}
final TreeData current = structure.get();
// this step is needed to get a valid NodeHandler
final TreeData temp = createTreeDataForRootAndTracker(newRoot, current.getNodeTracker());
structure.set(temp.updateNodeTracker(temp.getNodeTracker().update(newRoot, null, resolver, temp)));
}
/**
* Replaces a tracked node by another node. If the tracked node is not yet detached, it becomes now detached. The passed
* in node (which must not be null) becomes the new root node of an independent model for this tracked node.
* Further updates of this model do not affect the tracked node's model and vice versa.
*
* @param selector the {@code NodeSelector} defining the tracked node
* @param newNode the node replacing the tracked node (must not be null)
* @throws ConfigurationRuntimeException if the selector cannot be resolved
* @throws IllegalArgumentException if the replacement node is null
*/
public void replaceTrackedNode(final NodeSelector selector, final ImmutableNode newNode) {
if (newNode == null) {
throw new IllegalArgumentException("Replacement node must not be null!");
}
boolean done;
do {
final TreeData currentData = structure.get();
done = replaceDetachedTrackedNode(currentData, selector, newNode) || replaceActiveTrackedNode(currentData, selector, newNode);
} while (!done);
}
/**
* Allows tracking all nodes selected by a key. This method evaluates the specified key on the current nodes structure.
* For all selected nodes corresponding {@code NodeSelector} objects are created, and they are tracked. The returned
* collection of {@code NodeSelector} objects can be used for interacting with the selected nodes.
*
* @param key the key for selecting the nodes to track
* @param resolver the {@code NodeKeyResolver}
* @return a collection with the {@code NodeSelector} objects for the new tracked nodes
*/
public Collection selectAndTrackNodes(final String key, final NodeKeyResolver resolver) {
final Mutable> refSelectors = new MutableObject<>();
boolean done;
do {
final TreeData current = structure.get();
final List nodes = resolver.resolveNodeKey(current.getRootNode(), key, current);
if (nodes.isEmpty()) {
return Collections.emptyList();
}
done = structure.compareAndSet(current, createSelectorsForTrackedNodes(refSelectors, nodes, current, resolver));
} while (!done);
return refSelectors.getValue();
}
/**
* Sets the value of a property using a tracked node as root node. This method works like the normal
* {@code setProperty()} method, but the origin of the operation (also for the interpretation of the passed in key) is a
* tracked node identified by the passed in {@code NodeSelector}. The selector can be null, then the root node is
* assumed.
*
* @param key the key
* @param selector the {@code NodeSelector} defining the root node (or null)
* @param value the new value for this property
* @param resolver the {@code NodeKeyResolver}
* @throws ConfigurationRuntimeException if the selector cannot be resolved
*/
public void setProperty(final String key, final NodeSelector selector, final Object value, final NodeKeyResolver resolver) {
updateModel(tx -> {
boolean added = false;
final NodeUpdateData updateData = resolver.resolveUpdateKey(tx.getQueryRoot(), key, value, tx.getCurrentData());
if (!updateData.getNewValues().isEmpty()) {
initializeAddTransaction(tx, key, updateData.getNewValues(), resolver);
added = true;
}
final boolean cleared = initializeClearTransaction(tx, updateData.getRemovedNodes());
final boolean updated = initializeUpdateTransaction(tx, updateData.getChangedValues());
return added || cleared || updated;
}, selector, resolver);
}
@Override
public void setProperty(final String key, final Object value, final NodeKeyResolver resolver) {
setProperty(key, null, value, resolver);
}
/**
* {@inheritDoc} All tracked nodes and reference objects managed by this model are cleared.Care has to be taken when
* this method is used and the model is accessed by multiple threads. It is not deterministic which concurrent
* operations see the old root and which see the new root node.
*
* @param newRoot the new root node to be set (can be null, then an empty root node is set)
*/
@Override
public void setRootNode(final ImmutableNode newRoot) {
structure.set(createTreeData(initialRootNode(newRoot), structure.get()));
}
/**
* Tracks all nodes which are children of the node selected by the passed in key. If the key selects exactly one node,
* for all children of this node {@code NodeSelector} objects are created, and they become tracked nodes. The returned
* collection of {@code NodeSelector} objects can be used for interacting with the selected nodes.
*
* @param key the key for selecting the parent node whose children are to be tracked
* @param resolver the {@code NodeKeyResolver}
* @return a collection with the {@code NodeSelector} objects for the new tracked nodes
*/
public Collection trackChildNodes(final String key, final NodeKeyResolver resolver) {
final Mutable> refSelectors = new MutableObject<>();
boolean done;
do {
refSelectors.setValue(Collections.emptyList());
final TreeData current = structure.get();
final List nodes = resolver.resolveNodeKey(current.getRootNode(), key, current);
if (nodes.size() == 1) {
final ImmutableNode node = nodes.get(0);
done = node.getChildren().isEmpty()
|| structure.compareAndSet(current, createSelectorsForTrackedNodes(refSelectors, node.getChildren(), current, resolver));
} else {
done = true;
}
} while (!done);
return refSelectors.getValue();
}
/**
* Tracks a node which is a child of another node selected by the passed in key. If the selected node has a child node
* with this name, it is tracked and its selector is returned. Otherwise, a new child node with this name is created
* first.
*
* @param key the key for selecting the parent node
* @param childName the name of the child node
* @param resolver the {@code NodeKeyResolver}
* @return the {@code NodeSelector} for the tracked child node
* @throws ConfigurationRuntimeException if the passed in key does not select a single node
*/
public NodeSelector trackChildNodeWithCreation(final String key, final String childName, final NodeKeyResolver resolver) {
final MutableObject refSelector = new MutableObject<>();
boolean done;
do {
final TreeData current = structure.get();
final List nodes = resolver.resolveNodeKey(current.getRootNode(), key, current);
if (nodes.size() != 1) {
throw new ConfigurationRuntimeException("Key does not select a single node: " + key);
}
final ImmutableNode parent = nodes.get(0);
final TreeData newData = createDataWithTrackedChildNode(current, parent, childName, resolver, refSelector);
done = structure.compareAndSet(current, newData);
} while (!done);
return refSelector.getValue();
}
/**
* Adds a node to be tracked. After this method has been called with a specific {@code NodeSelector}, the node
* associated with this key can be always obtained using {@link #getTrackedNode(NodeSelector)} with the same selector.
* This is useful because during updates of a model parts of the structure are replaced. Therefore, it is not a good
* idea to simply hold a reference to a node; this might become outdated soon. Rather, the node should be tracked. This
* mechanism ensures that always the correct node reference can be obtained.
*
* @param selector the {@code NodeSelector} defining the desired node
* @param resolver the {@code NodeKeyResolver}
* @throws ConfigurationRuntimeException if the selector does not select a single node
*/
public void trackNode(final NodeSelector selector, final NodeKeyResolver resolver) {
boolean done;
do {
final TreeData current = structure.get();
final NodeTracker newTracker = current.getNodeTracker().trackNode(current.getRootNode(), selector, resolver, current);
done = structure.compareAndSet(current, current.updateNodeTracker(newTracker));
} while (!done);
}
/**
* Removes a tracked node. This method is the opposite of {@code trackNode()}. It has to be called if there is no longer
* the need to track a specific node. Note that for each call of {@code trackNode()} there has to be a corresponding
* {@code untrackNode()} call. This ensures that multiple observers can track the same node.
*
* @param selector the {@code NodeSelector} defining the desired node
* @throws ConfigurationRuntimeException if the specified node is not tracked
*/
public void untrackNode(final NodeSelector selector) {
boolean done;
do {
final TreeData current = structure.get();
final NodeTracker newTracker = current.getNodeTracker().untrackNode(selector);
done = structure.compareAndSet(current, current.updateNodeTracker(newTracker));
} while (!done);
}
/**
* Performs a non-blocking, thread-safe update of this model based on a transaction initialized by the passed in
* initializer. This method uses the atomic reference for the model's current data to ensure that an update was
* successful even if the model is concurrently accessed.
*
* @param txInit the {@code TransactionInitializer}
* @param selector an optional {@code NodeSelector} defining the target node of the transaction
* @param resolver the {@code NodeKeyResolver}
*/
private void updateModel(final TransactionInitializer txInit, final NodeSelector selector, final NodeKeyResolver resolver) {
boolean done;
do {
final TreeData currentData = getTreeData();
done = executeTransactionOnDetachedTrackedNode(txInit, selector, currentData, resolver)
|| executeTransactionOnCurrentStructure(txInit, selector, currentData, resolver);
} while (!done);
}
}