/*
 * 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
 *
 *     https://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;

import java.util.Collection;
import java.util.Collections;
import java.util.HashMap;
import java.util.Iterator;
import java.util.LinkedHashSet;
import java.util.LinkedList;
import java.util.List;
import java.util.Map;
import java.util.Objects;
import java.util.Set;
import java.util.Stack;
import java.util.stream.Collectors;

import org.apache.commons.configuration2.event.ConfigurationEvent;
import org.apache.commons.configuration2.ex.ConfigurationRuntimeException;
import org.apache.commons.configuration2.sync.NoOpSynchronizer;
import org.apache.commons.configuration2.tree.ConfigurationNodeVisitorAdapter;
import org.apache.commons.configuration2.tree.DefaultExpressionEngine;
import org.apache.commons.configuration2.tree.ExpressionEngine;
import org.apache.commons.configuration2.tree.NodeAddData;
import org.apache.commons.configuration2.tree.NodeHandler;
import org.apache.commons.configuration2.tree.NodeKeyResolver;
import org.apache.commons.configuration2.tree.NodeModel;
import org.apache.commons.configuration2.tree.NodeTreeWalker;
import org.apache.commons.configuration2.tree.NodeUpdateData;
import org.apache.commons.configuration2.tree.QueryResult;

/**
 * <p>
 * A specialized configuration class that extends its base class by the ability of keeping more structure in the stored
 * properties.
 * </p>
 * <p>
 * There are some sources of configuration data that cannot be stored very well in a {@code BaseConfiguration} object
 * because then their structure is lost. This is for instance true for XML documents. This class can deal with such
 * structured configuration sources by storing the properties in a tree-like organization. The exact storage structure
 * of the underlying data does not matter for the configuration instance; it uses a {@link NodeModel} object for
 * accessing it.
 * </p>
 * <p>
 * The hierarchical organization allows for a more sophisticated access to single properties. As an example consider the
 * following XML document:
 * </p>
 *
 * <pre>
 * &lt;database&gt;
 *   &lt;tables&gt;
 *     &lt;table&gt;
 *       &lt;name&gt;users&lt;/name&gt;
 *       &lt;fields&gt;
 *         &lt;field&gt;
 *           &lt;name&gt;lid&lt;/name&gt;
 *           &lt;type&gt;long&lt;/name&gt;
 *         &lt;/field&gt;
 *         &lt;field&gt;
 *           &lt;name&gt;usrName&lt;/name&gt;
 *           &lt;type&gt;java.lang.String&lt;/type&gt;
 *         &lt;/field&gt;
 *        ...
 *       &lt;/fields&gt;
 *     &lt;/table&gt;
 *     &lt;table&gt;
 *       &lt;name&gt;documents&lt;/name&gt;
 *       &lt;fields&gt;
 *         &lt;field&gt;
 *           &lt;name&gt;docid&lt;/name&gt;
 *           &lt;type&gt;long&lt;/type&gt;
 *         &lt;/field&gt;
 *         ...
 *       &lt;/fields&gt;
 *     &lt;/table&gt;
 *     ...
 *   &lt;/tables&gt;
 * &lt;/database&gt;
 * </pre>
 *
 * <p>
 * If this document is parsed and stored in a hierarchical configuration object (which can be done by one of the sub
 * classes), there are enhanced possibilities of accessing properties. Per default, the keys for querying information
 * can contain indices that select a specific element if there are multiple hits.
 * </p>
 * <p>
 * For instance the key {@code tables.table(0).name} can be used to find out the name of the first table. In opposite
 * {@code tables.table.name} would return a collection with the names of all available tables. Similarly the key
 * {@code tables.table(1).fields.field.name} returns a collection with the names of all fields of the second table. If
 * another index is added after the {@code field} element, a single field can be accessed:
 * {@code tables.table(1).fields.field(0).name}.
 * </p>
 * <p>
 * There is a {@code getMaxIndex()} method that returns the maximum allowed index that can be added to a given property
 * key. This method can be used to iterate over all values defined for a certain property.
 * </p>
 * <p>
 * Since the 1.3 release of <em>Commons Configuration</em> hierarchical configurations support an <em>expression
 * engine</em>. This expression engine is responsible for evaluating the passed in configuration keys and map them to
 * the stored properties. The examples above are valid for the default expression engine, which is used when a new
 * {@code AbstractHierarchicalConfiguration} instance is created. With the {@code setExpressionEngine()} method a
 * different expression engine can be set. For instance with
 * {@link org.apache.commons.configuration2.tree.xpath.XPathExpressionEngine} there is an expression engine available
 * that supports configuration keys in XPATH syntax.
 * </p>
 * <p>
 * In addition to the events common for all configuration classes, hierarchical configurations support some more events
 * that correspond to some specific methods and features. For those events specific event type constants in
 * {@code ConfigurationEvent} exist:
 * </p>
 * <dl>
 * <dt><em>ADD_NODES</em></dt>
 * <dd>The {@code addNodes()} method was called; the event object contains the key, to which the nodes were added, and a
 * collection with the new nodes as value.</dd>
 * <dt><em>CLEAR_TREE</em></dt>
 * <dd>The {@code clearTree()} method was called; the event object stores the key of the removed sub tree.</dd>
 * <dt><em>SUBNODE_CHANGED</em></dt>
 * <dd>A {@code SubnodeConfiguration} that was created from this configuration has been changed. The value property of
 * the event object contains the original event object as it was sent by the subnode configuration.</dd>
 * </dl>
 * <p>
 * Whether an {@code AbstractHierarchicalConfiguration} object is thread-safe or not depends on the underlying
 * {@code NodeModel} and the {@link org.apache.commons.configuration2.sync.Synchronizer Synchronizer} it is associated
 * with. Some {@code NodeModel} implementations are inherently thread-safe; they do not require a special
 * {@code Synchronizer}. (Per default, a dummy {@code Synchronizer} is used which is not thread-safe!) The methods for
 * querying or updating configuration data invoke this {@code Synchronizer} accordingly. When accessing the
 * configuration's root node directly, the client application is responsible for proper synchronization. This is
 * achieved by calling the methods {@link #lock(org.apache.commons.configuration2.sync.LockMode) lock()}, and
 * {@link #unlock(org.apache.commons.configuration2.sync.LockMode) unlock()} with a proper
 * {@link org.apache.commons.configuration2.sync.LockMode LockMode} argument. In any case, it is recommended to not
 * access the root node directly, but to use corresponding methods for querying or updating configuration data instead.
 * Direct manipulations of a configuration's node structure circumvent many internal mechanisms and thus can cause
 * undesired effects. For concrete subclasses dealing with specific node structures, this situation may be different.
 * </p>
 *
 * @param <T> the type of the nodes managed by this hierarchical configuration
 * @since 2.0
 */
public abstract class AbstractHierarchicalConfiguration<T> extends AbstractConfiguration
    implements Cloneable, NodeKeyResolver<T>, HierarchicalConfiguration<T> {

    /**
     * A specialized visitor that fills a list with keys that are defined in a node hierarchy.
     */
    private final class DefinedKeysVisitor extends ConfigurationNodeVisitorAdapter<T> {

        /** Stores the list to be filled. */
        private final Set<String> keyList;

        /** A stack with the keys of the already processed nodes. */
        private final Stack<String> parentKeys;

        /**
         * Default constructor.
         */
        public DefinedKeysVisitor() {
            keyList = new LinkedHashSet<>();
            parentKeys = new Stack<>();
        }

        /**
         * Creates a new {@code DefinedKeysVisitor} instance and sets the prefix for the keys to fetch.
         *
         * @param prefix the prefix
         */
        public DefinedKeysVisitor(final String prefix) {
            this();
            parentKeys.push(prefix);
        }

        /**
         * Gets the list with all defined keys.
         *
         * @return the list with the defined keys
         */
        public Set<String> getKeyList() {
            return keyList;
        }

        /**
         * Appends all attribute keys of the current node.
         *
         * @param parentKey the parent key
         * @param node the current node
         * @param handler the {@code NodeHandler}
         */
        public void handleAttributeKeys(final String parentKey, final T node, final NodeHandler<T> handler) {
            handler.getAttributes(node).forEach(attr -> keyList.add(getExpressionEngine().attributeKey(parentKey, attr)));
        }

        /**
         * {@inheritDoc} This implementation removes this node's key from the stack.
         */
        @Override
        public void visitAfterChildren(final T node, final NodeHandler<T> handler) {
            parentKeys.pop();
        }

        /**
         * {@inheritDoc} If this node has a value, its key is added to the internal list.
         */
        @Override
        public void visitBeforeChildren(final T node, final NodeHandler<T> handler) {
            final String parentKey = parentKeys.isEmpty() ? null : parentKeys.peek();
            final String key = getExpressionEngine().nodeKey(node, parentKey, handler);
            parentKeys.push(key);
            if (handler.getValue(node) != null) {
                keyList.add(key);
            }
            handleAttributeKeys(key, node, handler);
        }
    }

    /**
     * A specialized visitor that checks if a node is defined. &quot;Defined&quot; in this terms means that the node or at
     * least one of its sub nodes is associated with a value.
     *
     * @param <T> the type of the nodes managed by this hierarchical configuration
     */
    private static final class DefinedVisitor<T> extends ConfigurationNodeVisitorAdapter<T> {

        /** Stores the defined flag. */
        private boolean defined;

        /**
         * Returns the defined flag.
         *
         * @return the defined flag
         */
        public boolean isDefined() {
            return defined;
        }

        /**
         * Checks if iteration should be stopped. This can be done if the first defined node is found.
         *
         * @return a flag if iteration should be stopped
         */
        @Override
        public boolean terminate() {
            return isDefined();
        }

        /**
         * Visits the node. Checks if a value is defined.
         *
         * @param node the actual node
         */
        @Override
        public void visitBeforeChildren(final T node, final NodeHandler<T> handler) {
            defined = handler.getValue(node) != null || !handler.getAttributes(node).isEmpty();
        }
    }

    /** The model for managing the data stored in this configuration. */
    private NodeModel<T> nodeModel;

    /** Stores the expression engine for this instance. */
    private ExpressionEngine expressionEngine;

    /**
     * Creates a new instance of {@code AbstractHierarchicalConfiguration} and sets the {@code NodeModel} to be used.
     *
     * @param nodeModel the {@code NodeModel}
     */
    protected AbstractHierarchicalConfiguration(final NodeModel<T> nodeModel) {
        this.nodeModel = nodeModel;
    }

    /**
     * Adds a collection of nodes at the specified position of the configuration tree. This method works similar to
     * {@code addProperty()}, but instead of a single property a whole collection of nodes can be added - and thus complete
     * configuration sub trees. E.g. with this method it is possible to add parts of another
     * {@code BaseHierarchicalConfiguration} object to this object. If the passed in key refers to an existing and unique
     * node, the new nodes are added to this node. Otherwise a new node will be created at the specified position in the
     * hierarchy. Implementation node: This method performs some book-keeping and then delegates to
     * {@code addNodesInternal()}.
     *
     * @param key the key where the nodes are to be added; can be <strong>null</strong>, then they are added to the root node
     * @param nodes a collection with the {@code Node} objects to be added
     */
    @Override
    public final void addNodes(final String key, final Collection<? extends T> nodes) {
        if (nodes == null || nodes.isEmpty()) {
            return;
        }
        syncWrite(() -> {
            fireEvent(ConfigurationEvent.ADD_NODES, key, nodes, true);
            addNodesInternal(key, nodes);
            fireEvent(ConfigurationEvent.ADD_NODES, key, nodes, false);
        }, false);
    }

    /**
     * Actually adds a collection of new nodes to this configuration. This method is called by {@code addNodes()}. It can be
     * overridden by subclasses that need to adapt this operation.
     *
     * @param key the key where the nodes are to be added; can be <strong>null</strong>, then they are added to the root node
     * @param nodes a collection with the {@code Node} objects to be added
     * @since 2.0
     */
    protected void addNodesInternal(final String key, final Collection<? extends T> nodes) {
        getModel().addNodes(key, nodes, this);
    }

    /**
     * {@inheritDoc} This method is not called in the normal way (via {@code addProperty()} for hierarchical configurations
     * because all values to be added for the property have to be passed to the model in a single step. However, to allow
     * derived classes to add an arbitrary value as an object, a special implementation is provided here. The passed in
     * object is not parsed as a list, but passed directly as only value to the model.
     */
    @Override
    protected void addPropertyDirect(final String key, final Object value) {
        addPropertyToModel(key, Collections.singleton(value));
    }

    /**
     * Adds the property with the specified key. This task will be delegated to the associated {@code ExpressionEngine}, so
     * the passed in key must match the requirements of this implementation.
     *
     * @param key the key of the new property
     * @param obj the value of the new property
     */
    @Override
    protected void addPropertyInternal(final String key, final Object obj) {
        addPropertyToModel(key, getListDelimiterHandler().parse(obj));
    }

    /**
     * Helper method for executing an add property operation on the model.
     *
     * @param key the key of the new property
     * @param values the values to be added for this property
     */
    private void addPropertyToModel(final String key, final Iterable<?> values) {
        getModel().addProperty(key, values, this);
    }

    /**
     * Clears this configuration. This is a more efficient implementation than the one inherited from the base class. It
     * delegates to the node model.
     */
    @Override
    protected void clearInternal() {
        getModel().clear(this);
    }

    /**
     * Removes the property with the given key. Properties with names that start with the given key (i.e. properties below
     * the specified key in the hierarchy) won't be affected. This implementation delegates to the node+ model.
     *
     * @param key the key of the property to be removed
     */
    @Override
    protected void clearPropertyDirect(final String key) {
        getModel().clearProperty(key, this);
    }

    /**
     * Removes all values of the property with the given name and of keys that start with this name. So if there is a
     * property with the key &quot;foo&quot; and a property with the key &quot;foo.bar&quot;, a call of
     * {@code clearTree("foo")} would remove both properties.
     *
     * @param key the key of the property to be removed
     */
    @Override
    public final void clearTree(final String key) {
        syncWrite(() -> {
            fireEvent(ConfigurationEvent.CLEAR_TREE, key, null, true);
            fireEvent(ConfigurationEvent.CLEAR_TREE, key, clearTreeInternal(key), false);
        }, false);
    }

    /**
     * Actually clears the tree of elements referenced by the given key. This method is called by {@code clearTree()}.
     * Subclasses that need to adapt this operation can override this method. This base implementation delegates to the node
     * model.
     *
     * @param key the key of the property to be removed
     * @return an object with information about the nodes that have been removed (this is needed for firing a meaningful
     *         event of type CLEAR_TREE)
     * @since 2.0
     */
    protected Object clearTreeInternal(final String key) {
        return getModel().clearTree(key, this);
    }

    /**
     * Creates a copy of this object. This new configuration object will contain copies of all nodes in the same structure.
     * Registered event listeners won't be cloned; so they are not registered at the returned copy.
     *
     * @return the copy
     * @since 1.2
     */
    @SuppressWarnings("unchecked")
    @Override
    public Object clone() {
        return syncRead(() -> {
            try {
                final AbstractHierarchicalConfiguration<T> copy = (AbstractHierarchicalConfiguration<T>) AbstractHierarchicalConfiguration.super.clone();
                copy.setSynchronizer(NoOpSynchronizer.INSTANCE);
                copy.cloneInterpolator(this);
                copy.setSynchronizer(ConfigurationUtils.cloneSynchronizer(getSynchronizer()));
                copy.nodeModel = cloneNodeModel();
                return copy;
            } catch (final CloneNotSupportedException cex) {
                // should not happen
                throw new ConfigurationRuntimeException(cex);
            }
        }, false);
    }

    /**
     * Creates a clone of the node model. This method is called by {@code clone()}.
     *
     * @return the clone of the {@code NodeModel}
     * @since 2.0
     */
    protected abstract NodeModel<T> cloneNodeModel();

    /**
     * Checks if the specified key is contained in this configuration. Note that for this configuration the term
     * &quot;contained&quot; means that the key has an associated value. If there is a node for this key that has no value
     * but children (either defined or undefined), this method will still return <strong>false </strong>.
     *
     * @param key the key to be checked
     * @return a flag if this key is contained in this configuration
     */
    @Override
    protected boolean containsKeyInternal(final String key) {
        return getPropertyInternal(key) != null;
    }

    /**
     * Tests whether this configuration contains one or more matches to this value. This operation stops at first
     * match but may be more expensive than the containsKey method.
     *
     * @since 2.11.0
     */
    @Override
    protected boolean containsValueInternal(final Object value) {
        return contains(getKeys(), value);
    }

    /**
     * Helper method for resolving the specified key.
     *
     * @param key the key
     * @return a list with all results selected by this key
     */
    protected List<QueryResult<T>> fetchNodeList(final String key) {
        final NodeHandler<T> nodeHandler = getModel().getNodeHandler();
        return resolveKey(nodeHandler.getRootNode(), key, nodeHandler);
    }

    /**
     * Gets the expression engine used by this configuration. This method will never return <strong>null</strong>; if no specific
     * expression engine was set, the default expression engine will be returned.
     *
     * @return the current expression engine
     * @since 1.3
     */
    @Override
    public ExpressionEngine getExpressionEngine() {
        return expressionEngine != null ? expressionEngine : DefaultExpressionEngine.INSTANCE;
    }

    /**
     * Gets an iterator with all keys defined in this configuration. Note that the keys returned by this method will not
     * contain any indices. This means that some structure will be lost.
     *
     * @return an iterator with the defined keys in this configuration
     */
    @Override
    protected Iterator<String> getKeysInternal() {
        return visitDefinedKeys().getKeyList().iterator();
    }

    /**
     * Gets an iterator with all keys defined in this configuration that start with the given prefix. The returned keys
     * will not contain any indices. This implementation tries to locate a node whose key is the same as the passed in
     * prefix. Then the subtree of this node is traversed, and the keys of all nodes encountered (including attributes) are
     * added to the result set.
     *
     * @param prefix the prefix of the keys to start with
     * @return an iterator with the found keys
     */
    @Override
    protected Iterator<String> getKeysInternal(final String prefix) {
        return getKeysInternal(prefix, DELIMITER);
    }

    /**
     * Gets an iterator with all keys defined in this configuration that start with the given prefix. The returned keys
     * will not contain any indices. This implementation tries to locate a node whose key is the same as the passed in
     * prefix. Then the subtree of this node is traversed, and the keys of all nodes encountered (including attributes) are
     * added to the result set.
     *
     * @param prefix the prefix of the keys to start with
     * @param delimiter the prefix delimiter (unused)
     * @return an iterator with the found keys
     * @since 2.12.0
     */
    @Override
    protected Iterator<String> getKeysInternal(final String prefix, final String delimiter) {
        final DefinedKeysVisitor visitor = new DefinedKeysVisitor(prefix);
        if (containsKey(prefix)) {
            // explicitly add the prefix
            visitor.getKeyList().add(prefix);
        }

        final List<QueryResult<T>> results = fetchNodeList(prefix);
        final NodeHandler<T> handler = getModel().getNodeHandler();

        results.forEach(result -> {
            if (!result.isAttributeResult()) {
                handler.getChildren(result.getNode()).forEach(c -> NodeTreeWalker.INSTANCE.walkDFS(c, visitor, handler));
                visitor.handleAttributeKeys(prefix, result.getNode(), handler);
            }
        });

        return visitor.getKeyList().iterator();
    }

    /**
     * Gets the maximum defined index for the given key. This is useful if there are multiple values for this key. They
     * can then be addressed separately by specifying indices from 0 to the return value of this method. If the passed in
     * key is not contained in this configuration, result is -1.
     *
     * @param key the key to be checked
     * @return the maximum defined index for this key
     */
    @Override
    public final int getMaxIndex(final String key) {
        return syncRead(() -> getMaxIndexInternal(key), false);
    }

    /**
     * Actually retrieves the maximum defined index for the given key. This method is called by {@code getMaxIndex()}.
     * Subclasses that need to adapt this operation have to override this method.
     *
     * @param key the key to be checked
     * @return the maximum defined index for this key
     * @since 2.0
     */
    protected int getMaxIndexInternal(final String key) {
        return fetchNodeList(key).size() - 1;
    }

    /**
     * Gets the {@code NodeModel} used by this configuration. This method is intended for internal use only. Access to
     * the model is granted without any synchronization. This is in contrast to the &quot;official&quot;
     * {@code getNodeModel()} method which is guarded by the configuration's {@code Synchronizer}.
     *
     * @return the node model
     */
    protected NodeModel<T> getModel() {
        return nodeModel;
    }

    /**
     * {@inheritDoc} This implementation returns the configuration's {@code NodeModel}. It is guarded by the current
     * {@code Synchronizer}.
     */
    @Override
    public NodeModel<T> getNodeModel() {
        return syncRead(this::getModel, false);
    }

    /**
     * Fetches the specified property. This task is delegated to the associated expression engine.
     *
     * @param key the key to be looked up
     * @return the found value
     */
    @Override
    protected Object getPropertyInternal(final String key) {
        final List<QueryResult<T>> results = fetchNodeList(key);

        if (results.isEmpty()) {
            return null;
        }
        final NodeHandler<T> handler = getModel().getNodeHandler();
        final List<Object> list = results.stream().map(r -> valueFromResult(r, handler)).filter(Objects::nonNull).collect(Collectors.toList());

        if (list.size() < 1) {
            return null;
        }
        return list.size() == 1 ? list.get(0) : list;
    }

    /**
     * {@inheritDoc} This implementation handles synchronization and delegates to {@code getRootElementNameInternal()}.
     */
    @Override
    public final String getRootElementName() {
        return syncRead(this::getRootElementNameInternal, false);
    }

    /**
     * Actually obtains the name of the root element. This method is called by {@code getRootElementName()}. It just returns
     * the name of the root node. Subclasses that treat the root element name differently can override this method.
     *
     * @return the name of this configuration's root element
     */
    protected String getRootElementNameInternal() {
        final NodeHandler<T> nodeHandler = getModel().getNodeHandler();
        return nodeHandler.nodeName(nodeHandler.getRootNode());
    }

    /**
     * Checks if this configuration is empty. Empty means that there are no keys with any values, though there can be some
     * (empty) nodes.
     *
     * @return a flag if this configuration is empty
     */
    @Override
    protected boolean isEmptyInternal() {
        return !nodeDefined(getModel().getNodeHandler().getRootNode());
    }

    /**
     * Checks if the specified node is defined.
     *
     * @param node the node to be checked
     * @return a flag if this node is defined
     */
    protected boolean nodeDefined(final T node) {
        final DefinedVisitor<T> visitor = new DefinedVisitor<>();
        NodeTreeWalker.INSTANCE.walkBFS(node, visitor, getModel().getNodeHandler());
        return visitor.isDefined();
    }

    /**
     * {@inheritDoc} This implementation uses the expression engine to generate a canonical key for the passed in node. For
     * this purpose, the path to the root node has to be traversed. The cache is used to store and access keys for nodes
     * encountered on the path.
     */
    @Override
    public String nodeKey(final T node, final Map<T, String> cache, final NodeHandler<T> handler) {
        final List<T> paths = new LinkedList<>();
        T currentNode = node;
        String key = cache.get(node);
        while (key == null && currentNode != null) {
            paths.add(0, currentNode);
            currentNode = handler.getParent(currentNode);
            key = cache.get(currentNode);
        }

        for (final T n : paths) {
            final String currentKey = getExpressionEngine().canonicalKey(n, key, handler);
            cache.put(n, currentKey);
            key = currentKey;
        }

        return key;
    }

    /**
     * {@inheritDoc} This implementation delegates to the expression engine.
     */
    @Override
    public NodeAddData<T> resolveAddKey(final T root, final String key, final NodeHandler<T> handler) {
        return getExpressionEngine().prepareAdd(root, key, handler);
    }

    /**
     * {@inheritDoc} This implementation delegates to the expression engine.
     */
    @Override
    public List<QueryResult<T>> resolveKey(final T root, final String key, final NodeHandler<T> handler) {
        return getExpressionEngine().query(root, key, handler);
    }

    /**
     * {@inheritDoc} This implementation delegates to {@code resolveKey()} and then filters out attribute results.
     */
    @Override
    public List<T> resolveNodeKey(final T root, final String key, final NodeHandler<T> handler) {
        return resolveKey(root, key, handler).stream().filter(r -> !r.isAttributeResult()).map(QueryResult::getNode)
            .collect(Collectors.toCollection(LinkedList::new));
    }

    /**
     * {@inheritDoc} This implementation executes a query for the given key and constructs a {@code NodeUpdateData} object
     * based on the results. It determines which nodes need to be changed and whether new ones need to be added or existing
     * ones need to be removed.
     */
    @Override
    public NodeUpdateData<T> resolveUpdateKey(final T root, final String key, final Object newValue, final NodeHandler<T> handler) {
        final Iterator<QueryResult<T>> itNodes = fetchNodeList(key).iterator();
        final Iterator<?> itValues = getListDelimiterHandler().parse(newValue).iterator();
        final Map<QueryResult<T>, Object> changedValues = new HashMap<>();
        Collection<Object> additionalValues = null;
        Collection<QueryResult<T>> removedItems = null;

        while (itNodes.hasNext() && itValues.hasNext()) {
            changedValues.put(itNodes.next(), itValues.next());
        }

        // Add additional nodes if necessary
        if (itValues.hasNext()) {
            additionalValues = new LinkedList<>();
            itValues.forEachRemaining(additionalValues::add);
        }

        // Remove remaining nodes
        if (itNodes.hasNext()) {
            removedItems = new LinkedList<>();
            itNodes.forEachRemaining(removedItems::add);
        }

        return new NodeUpdateData<>(changedValues, additionalValues, removedItems, key);
    }

    /**
     * Sets the expression engine to be used by this configuration. All property keys this configuration has to deal with
     * will be interpreted by this engine.
     *
     * @param expressionEngine the new expression engine; can be <strong>null</strong>, then the default expression engine will be
     *        used
     * @since 1.3
     */
    @Override
    public void setExpressionEngine(final ExpressionEngine expressionEngine) {
        this.expressionEngine = expressionEngine;
    }

    /**
     * Sets the value of the specified property.
     *
     * @param key the key of the property to set
     * @param value the new value of this property
     */
    @Override
    protected void setPropertyInternal(final String key, final Object value) {
        getModel().setProperty(key, value, this);
    }

    /**
     * {@inheritDoc} This implementation is slightly more efficient than the default implementation. It does not iterate
     * over the key set, but directly queries its size after it has been constructed. Note that constructing the key set is
     * still an O(n) operation.
     */
    @Override
    protected int sizeInternal() {
        return visitDefinedKeys().getKeyList().size();
    }

    @Override
    public String toString() {
        return super.toString() + "(" + getRootElementNameInternal() + ")";
    }

    /**
     * Extracts the value from a query result.
     *
     * @param result the {@code QueryResult}
     * @param handler the {@code NodeHandler}
     * @return the value of this result (may be <strong>null</strong>)
     */
    private Object valueFromResult(final QueryResult<T> result, final NodeHandler<T> handler) {
        return result.isAttributeResult() ? result.getAttributeValue(handler) : handler.getValue(result.getNode());
    }

    /**
     * Creates a {@code DefinedKeysVisitor} and visits all defined keys with it.
     *
     * @return the visitor after all keys have been visited
     */
    private DefinedKeysVisitor visitDefinedKeys() {
        final DefinedKeysVisitor visitor = new DefinedKeysVisitor();
        final NodeHandler<T> nodeHandler = getModel().getNodeHandler();
        NodeTreeWalker.INSTANCE.walkDFS(nodeHandler.getRootNode(), visitor, nodeHandler);
        return visitor;
    }
}