001/*
002 * Licensed to the Apache Software Foundation (ASF) under one or more
003 * contributor license agreements.  See the NOTICE file distributed with
004 * this work for additional information regarding copyright ownership.
005 * The ASF licenses this file to You under the Apache License, Version 2.0
006 * (the "License"); you may not use this file except in compliance with
007 * the License.  You may obtain a copy of the License at
008 *
009 *     http://www.apache.org/licenses/LICENSE-2.0
010 *
011 * Unless required by applicable law or agreed to in writing, software
012 * distributed under the License is distributed on an "AS IS" BASIS,
013 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
014 * See the License for the specific language governing permissions and
015 * limitations under the License.
016 */
017package org.apache.commons.configuration2.tree;
018
019import java.util.Collection;
020import java.util.LinkedList;
021import java.util.List;
022
023import org.apache.commons.lang3.StringUtils;
024
025/**
026 * <p>
027 * A default implementation of the {@code ExpressionEngine} interface
028 * providing the &quot;native&quot; expression language for hierarchical
029 * configurations.
030 * </p>
031 * <p>
032 * This class implements a rather simple expression language for navigating
033 * through a hierarchy of configuration nodes. It supports the following
034 * operations:
035 * </p>
036 * <ul>
037 * <li>Navigating from a node to one of its children using the child node
038 * delimiter, which is by the default a dot (&quot;.&quot;).</li>
039 * <li>Navigating from a node to one of its attributes using the attribute node
040 * delimiter, which by default follows the XPATH like syntax
041 * <code>[@&lt;attributeName&gt;]</code>.</li>
042 * <li>If there are multiple child or attribute nodes with the same name, a
043 * specific node can be selected using a numerical index. By default indices are
044 * written in parenthesis.</li>
045 * </ul>
046 * <p>
047 * As an example consider the following XML document:
048 * </p>
049 *
050 * <pre>
051 *  &lt;database&gt;
052 *    &lt;tables&gt;
053 *      &lt;table type=&quot;system&quot;&gt;
054 *        &lt;name&gt;users&lt;/name&gt;
055 *        &lt;fields&gt;
056 *          &lt;field&gt;
057 *            &lt;name&gt;lid&lt;/name&gt;
058 *            &lt;type&gt;long&lt;/name&gt;
059 *          &lt;/field&gt;
060 *          &lt;field&gt;
061 *            &lt;name&gt;usrName&lt;/name&gt;
062 *            &lt;type&gt;java.lang.String&lt;/type&gt;
063 *          &lt;/field&gt;
064 *         ...
065 *        &lt;/fields&gt;
066 *      &lt;/table&gt;
067 *      &lt;table&gt;
068 *        &lt;name&gt;documents&lt;/name&gt;
069 *        &lt;fields&gt;
070 *          &lt;field&gt;
071 *            &lt;name&gt;docid&lt;/name&gt;
072 *            &lt;type&gt;long&lt;/type&gt;
073 *          &lt;/field&gt;
074 *          ...
075 *        &lt;/fields&gt;
076 *      &lt;/table&gt;
077 *      ...
078 *    &lt;/tables&gt;
079 *  &lt;/database&gt;
080 * </pre>
081 *
082 * <p>
083 * If this document is parsed and stored in a hierarchical configuration object,
084 * for instance the key {@code tables.table(0).name} can be used to find
085 * out the name of the first table. In opposite {@code tables.table.name}
086 * would return a collection with the names of all available tables. Similarly
087 * the key {@code tables.table(1).fields.field.name} returns a collection
088 * with the names of all fields of the second table. If another index is added
089 * after the {@code field} element, a single field can be accessed:
090 * {@code tables.table(1).fields.field(0).name}. The key
091 * {@code tables.table(0)[@type]} would select the type attribute of the
092 * first table.
093 * </p>
094 * <p>
095 * This example works with the default values for delimiters and index markers.
096 * It is also possible to set custom values for these properties so that you can
097 * adapt a {@code DefaultExpressionEngine} to your personal needs.
098 * </p>
099 * <p>
100 * The concrete symbols used by an instance are determined by a
101 * {@link DefaultExpressionEngineSymbols} object passed to the constructor.
102 * By providing a custom symbols object the syntax for querying properties in
103 * a hierarchical configuration can be altered.
104 * </p>
105 * <p>
106 * Instances of this class are thread-safe and can be shared between multiple
107 * hierarchical configuration objects.
108 * </p>
109 *
110 * @since 1.3
111 * @author <a
112 * href="http://commons.apache.org/configuration/team-list.html">Commons
113 * Configuration team</a>
114 * @version $Id: DefaultExpressionEngine.java 1790899 2017-04-10 21:56:46Z ggregory $
115 */
116public class DefaultExpressionEngine implements ExpressionEngine
117{
118    /**
119     * A default instance of this class that is used as expression engine for
120     * hierarchical configurations per default.
121     */
122    public static final DefaultExpressionEngine INSTANCE =
123            new DefaultExpressionEngine(
124                    DefaultExpressionEngineSymbols.DEFAULT_SYMBOLS);
125
126    /** The symbols used by this instance. */
127    private final DefaultExpressionEngineSymbols symbols;
128
129    /** The matcher for node names. */
130    private final NodeMatcher<String> nameMatcher;
131
132    /**
133     * Creates a new instance of {@code DefaultExpressionEngine} and initializes
134     * its symbols.
135     *
136     * @param syms the object with the symbols (must not be <b>null</b>)
137     * @throws IllegalArgumentException if the symbols are <b>null</b>
138     */
139    public DefaultExpressionEngine(DefaultExpressionEngineSymbols syms)
140    {
141        this(syms, null);
142    }
143
144    /**
145     * Creates a new instance of {@code DefaultExpressionEngine} and initializes
146     * its symbols and the matcher for comparing node names. The passed in
147     * matcher is always used when the names of nodes have to be matched against
148     * parts of configuration keys.
149     *
150     * @param syms the object with the symbols (must not be <b>null</b>)
151     * @param nodeNameMatcher the matcher for node names; can be <b>null</b>,
152     *        then a default matcher is used
153     * @throws IllegalArgumentException if the symbols are <b>null</b>
154     */
155    public DefaultExpressionEngine(DefaultExpressionEngineSymbols syms,
156            NodeMatcher<String> nodeNameMatcher)
157    {
158        if (syms == null)
159        {
160            throw new IllegalArgumentException("Symbols must not be null!");
161        }
162
163        symbols = syms;
164        nameMatcher =
165                (nodeNameMatcher != null) ? nodeNameMatcher
166                        : NodeNameMatchers.EQUALS;
167    }
168
169    /**
170     * Returns the {@code DefaultExpressionEngineSymbols} object associated with
171     * this instance.
172     *
173     * @return the {@code DefaultExpressionEngineSymbols} used by this engine
174     * @since 2.0
175     */
176    public DefaultExpressionEngineSymbols getSymbols()
177    {
178        return symbols;
179    }
180
181    /**
182     * {@inheritDoc} This method supports the syntax as described in the class
183     * comment.
184     */
185    @Override
186    public <T> List<QueryResult<T>> query(T root, String key,
187            NodeHandler<T> handler)
188    {
189        List<QueryResult<T>> results = new LinkedList<>();
190        findNodesForKey(new DefaultConfigurationKey(this, key).iterator(),
191                root, results, handler);
192        return results;
193    }
194
195    /**
196     * {@inheritDoc} This implementation takes the
197     * given parent key, adds a property delimiter, and then adds the node's
198     * name.
199     * The name of the root node is a blank string. Note that no indices are
200     * returned.
201     */
202    @Override
203    public <T> String nodeKey(T node, String parentKey, NodeHandler<T> handler)
204    {
205        if (parentKey == null)
206        {
207            // this is the root node
208            return StringUtils.EMPTY;
209        }
210
211        else
212        {
213            DefaultConfigurationKey key = new DefaultConfigurationKey(this,
214                    parentKey);
215                key.append(handler.nodeName(node), true);
216            return key.toString();
217        }
218    }
219
220    @Override
221    public String attributeKey(String parentKey, String attributeName)
222    {
223        DefaultConfigurationKey key =
224                new DefaultConfigurationKey(this, parentKey);
225        key.appendAttribute(attributeName);
226        return key.toString();
227    }
228
229    /**
230     * {@inheritDoc} This implementation works similar to {@code nodeKey()};
231     * however, each key returned by this method has an index (except for the
232     * root node). The parent key is prepended to the name of the current node
233     * in any case and without further checks. If it is <b>null</b>, only the
234     * name of the current node with its index is returned.
235     */
236    @Override
237    public <T> String canonicalKey(T node, String parentKey,
238            NodeHandler<T> handler)
239    {
240        String nodeName = handler.nodeName(node);
241        T parent = handler.getParent(node);
242        DefaultConfigurationKey key =
243                new DefaultConfigurationKey(this, parentKey);
244        key.append(StringUtils.defaultString(nodeName));
245
246        if (parent != null)
247        {
248            // this is not the root key
249            key.appendIndex(determineIndex(node, parent, nodeName, handler));
250        }
251        return key.toString();
252    }
253
254    /**
255     * <p>
256     * Prepares Adding the property with the specified key.
257     * </p>
258     * <p>
259     * To be able to deal with the structure supported by hierarchical
260     * configuration implementations the passed in key is of importance,
261     * especially the indices it might contain. The following example should
262     * clarify this: Suppose the current node structure looks like the
263     * following:
264     * </p>
265     * <pre>
266     *  tables
267     *     +-- table
268     *             +-- name = user
269     *             +-- fields
270     *                     +-- field
271     *                             +-- name = uid
272     *                     +-- field
273     *                             +-- name = firstName
274     *                     ...
275     *     +-- table
276     *             +-- name = documents
277     *             +-- fields
278     *                    ...
279     * </pre>
280     * <p>
281     * In this example a database structure is defined, e.g. all fields of the
282     * first table could be accessed using the key
283     * {@code tables.table(0).fields.field.name}. If now properties are
284     * to be added, it must be exactly specified at which position in the
285     * hierarchy the new property is to be inserted. So to add a new field name
286     * to a table it is not enough to say just
287     * </p>
288     * <pre>
289     * config.addProperty(&quot;tables.table.fields.field.name&quot;, &quot;newField&quot;);
290     * </pre>
291     * <p>
292     * The statement given above contains some ambiguity. For instance it is not
293     * clear, to which table the new field should be added. If this method finds
294     * such an ambiguity, it is resolved by following the last valid path. Here
295     * this would be the last table. The same is true for the {@code field};
296     * because there are multiple fields and no explicit index is provided, a
297     * new {@code name} property would be added to the last field - which
298     * is probably not what was desired.
299     * </p>
300     * <p>
301     * To make things clear explicit indices should be provided whenever
302     * possible. In the example above the exact table could be specified by
303     * providing an index for the {@code table} element as in
304     * {@code tables.table(1).fields}. By specifying an index it can
305     * also be expressed that at a given position in the configuration tree a
306     * new branch should be added. In the example above we did not want to add
307     * an additional {@code name} element to the last field of the table,
308     * but we want a complete new {@code field} element. This can be
309     * achieved by specifying an invalid index (like -1) after the element where
310     * a new branch should be created. Given this our example would run:
311     * </p>
312     * <pre>
313     * config.addProperty(&quot;tables.table(1).fields.field(-1).name&quot;, &quot;newField&quot;);
314     * </pre>
315     * <p>
316     * With this notation it is possible to add new branches everywhere. We
317     * could for instance create a new {@code table} element by
318     * specifying
319     * </p>
320     * <pre>
321     * config.addProperty(&quot;tables.table(-1).fields.field.name&quot;, &quot;newField2&quot;);
322     * </pre>
323     * <p>
324     * (Note that because after the {@code table} element a new branch is
325     * created indices in following elements are not relevant; the branch is new
326     * so there cannot be any ambiguities.)
327     * </p>
328     *
329     * @param <T> the type of the nodes to be dealt with
330     * @param root the root node of the nodes hierarchy
331     * @param key the key of the new property
332     * @param handler the node handler
333     * @return a data object with information needed for the add operation
334     */
335    @Override
336    public <T> NodeAddData<T> prepareAdd(T root, String key, NodeHandler<T> handler)
337    {
338        DefaultConfigurationKey.KeyIterator it = new DefaultConfigurationKey(
339                this, key).iterator();
340        if (!it.hasNext())
341        {
342            throw new IllegalArgumentException(
343                    "Key for add operation must be defined!");
344        }
345
346        T parent = findLastPathNode(it, root, handler);
347        List<String> pathNodes = new LinkedList<>();
348
349        while (it.hasNext())
350        {
351            if (!it.isPropertyKey())
352            {
353                throw new IllegalArgumentException(
354                        "Invalid key for add operation: " + key
355                                + " (Attribute key in the middle.)");
356            }
357            pathNodes.add(it.currentKey());
358            it.next();
359        }
360
361        return new NodeAddData<>(parent, it.currentKey(), !it.isPropertyKey(),
362                pathNodes);
363    }
364
365    /**
366     * Recursive helper method for evaluating a key. This method processes all
367     * facets of a configuration key, traverses the tree of properties and
368     * fetches the results of all matching properties.
369     *
370     * @param <T> the type of nodes to be dealt with
371     * @param keyPart the configuration key iterator
372     * @param node the current node
373     * @param results here the found results are stored
374     * @param handler the node handler
375     */
376    protected <T> void findNodesForKey(
377            DefaultConfigurationKey.KeyIterator keyPart, T node,
378            Collection<QueryResult<T>> results, NodeHandler<T> handler)
379    {
380        if (!keyPart.hasNext())
381        {
382            results.add(QueryResult.createNodeResult(node));
383        }
384
385        else
386        {
387            String key = keyPart.nextKey(false);
388            if (keyPart.isPropertyKey())
389            {
390                processSubNodes(keyPart, findChildNodesByName(handler, node, key),
391                        results, handler);
392            }
393            if (keyPart.isAttribute() && !keyPart.hasNext())
394            {
395                if (handler.getAttributeValue(node, key) != null)
396                {
397                    results.add(QueryResult.createAttributeResult(node, key));
398                }
399            }
400        }
401    }
402
403    /**
404     * Finds the last existing node for an add operation. This method traverses
405     * the node tree along the specified key. The last existing node on this
406     * path is returned.
407     *
408     * @param <T> the type of the nodes to be dealt with
409     * @param keyIt the key iterator
410     * @param node the current node
411     * @param handler the node handler
412     * @return the last existing node on the given path
413     */
414    protected <T> T findLastPathNode(DefaultConfigurationKey.KeyIterator keyIt,
415            T node, NodeHandler<T> handler)
416    {
417        String keyPart = keyIt.nextKey(false);
418
419        if (keyIt.hasNext())
420        {
421            if (!keyIt.isPropertyKey())
422            {
423                // Attribute keys can only appear as last elements of the path
424                throw new IllegalArgumentException(
425                        "Invalid path for add operation: "
426                                + "Attribute key in the middle!");
427            }
428            int idx =
429                    keyIt.hasIndex() ? keyIt.getIndex() : handler
430                            .getMatchingChildrenCount(node, nameMatcher,
431                                    keyPart) - 1;
432            if (idx < 0
433                    || idx >= handler.getMatchingChildrenCount(node,
434                            nameMatcher, keyPart))
435            {
436                return node;
437            }
438            else
439            {
440                return findLastPathNode(keyIt,
441                        findChildNodesByName(handler, node, keyPart).get(idx),
442                        handler);
443            }
444        }
445
446        else
447        {
448            return node;
449        }
450    }
451
452    /**
453     * Called by {@code findNodesForKey()} to process the sub nodes of
454     * the current node depending on the type of the current key part (children,
455     * attributes, or both).
456     *
457     * @param <T> the type of the nodes to be dealt with
458     * @param keyPart the key part
459     * @param subNodes a list with the sub nodes to process
460     * @param nodes the target collection
461     * @param handler the node handler
462     */
463    private <T> void processSubNodes(DefaultConfigurationKey.KeyIterator keyPart,
464            List<T> subNodes, Collection<QueryResult<T>> nodes, NodeHandler<T> handler)
465    {
466        if (keyPart.hasIndex())
467        {
468            if (keyPart.getIndex() >= 0 && keyPart.getIndex() < subNodes.size())
469            {
470                findNodesForKey((DefaultConfigurationKey.KeyIterator) keyPart
471                        .clone(), subNodes.get(keyPart.getIndex()), nodes, handler);
472            }
473        }
474        else
475        {
476            for (T node : subNodes)
477            {
478                findNodesForKey((DefaultConfigurationKey.KeyIterator) keyPart
479                        .clone(), node, nodes, handler);
480            }
481        }
482    }
483
484    /**
485     * Determines the index of the given node based on its parent node.
486     *
487     * @param node the current node
488     * @param parent the parent node
489     * @param nodeName the name of the current node
490     * @param handler the node handler
491     * @param <T> the type of the nodes to be dealt with
492     * @return the index of this node
493     */
494    private <T> int determineIndex(T node, T parent, String nodeName,
495                                          NodeHandler<T> handler)
496    {
497        return findChildNodesByName(handler, parent, nodeName).indexOf(node);
498    }
499
500    /**
501     * Returns a list with all child nodes of the given parent node which match
502     * the specified node name. The match is done using the current node name
503     * matcher.
504     *
505     * @param handler the {@code NodeHandler}
506     * @param parent the parent node
507     * @param nodeName the name of the current node
508     * @param <T> the type of the nodes to be dealt with
509     * @return a list with all matching child nodes
510     */
511    private <T> List<T> findChildNodesByName(NodeHandler<T> handler, T parent,
512            String nodeName)
513    {
514        return handler.getMatchingChildren(parent, nameMatcher, nodeName);
515    }
516}