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 */
115public class DefaultExpressionEngine implements ExpressionEngine
116{
117    /**
118     * A default instance of this class that is used as expression engine for
119     * hierarchical configurations per default.
120     */
121    public static final DefaultExpressionEngine INSTANCE =
122            new DefaultExpressionEngine(
123                    DefaultExpressionEngineSymbols.DEFAULT_SYMBOLS);
124
125    /** The symbols used by this instance. */
126    private final DefaultExpressionEngineSymbols symbols;
127
128    /** The matcher for node names. */
129    private final NodeMatcher<String> nameMatcher;
130
131    /**
132     * Creates a new instance of {@code DefaultExpressionEngine} and initializes
133     * its symbols.
134     *
135     * @param syms the object with the symbols (must not be <b>null</b>)
136     * @throws IllegalArgumentException if the symbols are <b>null</b>
137     */
138    public DefaultExpressionEngine(final DefaultExpressionEngineSymbols syms)
139    {
140        this(syms, null);
141    }
142
143    /**
144     * Creates a new instance of {@code DefaultExpressionEngine} and initializes
145     * its symbols and the matcher for comparing node names. The passed in
146     * matcher is always used when the names of nodes have to be matched against
147     * parts of configuration keys.
148     *
149     * @param syms the object with the symbols (must not be <b>null</b>)
150     * @param nodeNameMatcher the matcher for node names; can be <b>null</b>,
151     *        then a default matcher is used
152     * @throws IllegalArgumentException if the symbols are <b>null</b>
153     */
154    public DefaultExpressionEngine(final DefaultExpressionEngineSymbols syms,
155            final NodeMatcher<String> nodeNameMatcher)
156    {
157        if (syms == null)
158        {
159            throw new IllegalArgumentException("Symbols must not be null!");
160        }
161
162        symbols = syms;
163        nameMatcher =
164                (nodeNameMatcher != null) ? nodeNameMatcher
165                        : NodeNameMatchers.EQUALS;
166    }
167
168    /**
169     * Returns the {@code DefaultExpressionEngineSymbols} object associated with
170     * this instance.
171     *
172     * @return the {@code DefaultExpressionEngineSymbols} used by this engine
173     * @since 2.0
174     */
175    public DefaultExpressionEngineSymbols getSymbols()
176    {
177        return symbols;
178    }
179
180    /**
181     * {@inheritDoc} This method supports the syntax as described in the class
182     * comment.
183     */
184    @Override
185    public <T> List<QueryResult<T>> query(final T root, final String key,
186            final NodeHandler<T> handler)
187    {
188        final List<QueryResult<T>> results = new LinkedList<>();
189        findNodesForKey(new DefaultConfigurationKey(this, key).iterator(),
190                root, results, handler);
191        return results;
192    }
193
194    /**
195     * {@inheritDoc} This implementation takes the
196     * given parent key, adds a property delimiter, and then adds the node's
197     * name.
198     * The name of the root node is a blank string. Note that no indices are
199     * returned.
200     */
201    @Override
202    public <T> String nodeKey(final T node, final String parentKey, final NodeHandler<T> handler)
203    {
204        if (parentKey == null)
205        {
206            // this is the root node
207            return StringUtils.EMPTY;
208        }
209        final DefaultConfigurationKey key = new DefaultConfigurationKey(this,
210                parentKey);
211            key.append(handler.nodeName(node), true);
212        return key.toString();
213    }
214
215    @Override
216    public String attributeKey(final String parentKey, final String attributeName)
217    {
218        final DefaultConfigurationKey key =
219                new DefaultConfigurationKey(this, parentKey);
220        key.appendAttribute(attributeName);
221        return key.toString();
222    }
223
224    /**
225     * {@inheritDoc} This implementation works similar to {@code nodeKey()};
226     * however, each key returned by this method has an index (except for the
227     * root node). The parent key is prepended to the name of the current node
228     * in any case and without further checks. If it is <b>null</b>, only the
229     * name of the current node with its index is returned.
230     */
231    @Override
232    public <T> String canonicalKey(final T node, final String parentKey,
233            final NodeHandler<T> handler)
234    {
235        final String nodeName = handler.nodeName(node);
236        final T parent = handler.getParent(node);
237        final DefaultConfigurationKey key =
238                new DefaultConfigurationKey(this, parentKey);
239        key.append(StringUtils.defaultString(nodeName));
240
241        if (parent != null)
242        {
243            // this is not the root key
244            key.appendIndex(determineIndex(node, parent, nodeName, handler));
245        }
246        return key.toString();
247    }
248
249    /**
250     * <p>
251     * Prepares Adding the property with the specified key.
252     * </p>
253     * <p>
254     * To be able to deal with the structure supported by hierarchical
255     * configuration implementations the passed in key is of importance,
256     * especially the indices it might contain. The following example should
257     * clarify this: Suppose the current node structure looks like the
258     * following:
259     * </p>
260     * <pre>
261     *  tables
262     *     +-- table
263     *             +-- name = user
264     *             +-- fields
265     *                     +-- field
266     *                             +-- name = uid
267     *                     +-- field
268     *                             +-- name = firstName
269     *                     ...
270     *     +-- table
271     *             +-- name = documents
272     *             +-- fields
273     *                    ...
274     * </pre>
275     * <p>
276     * In this example a database structure is defined, e.g. all fields of the
277     * first table could be accessed using the key
278     * {@code tables.table(0).fields.field.name}. If now properties are
279     * to be added, it must be exactly specified at which position in the
280     * hierarchy the new property is to be inserted. So to add a new field name
281     * to a table it is not enough to say just
282     * </p>
283     * <pre>
284     * config.addProperty(&quot;tables.table.fields.field.name&quot;, &quot;newField&quot;);
285     * </pre>
286     * <p>
287     * The statement given above contains some ambiguity. For instance it is not
288     * clear, to which table the new field should be added. If this method finds
289     * such an ambiguity, it is resolved by following the last valid path. Here
290     * this would be the last table. The same is true for the {@code field};
291     * because there are multiple fields and no explicit index is provided, a
292     * new {@code name} property would be added to the last field - which
293     * is probably not what was desired.
294     * </p>
295     * <p>
296     * To make things clear explicit indices should be provided whenever
297     * possible. In the example above the exact table could be specified by
298     * providing an index for the {@code table} element as in
299     * {@code tables.table(1).fields}. By specifying an index it can
300     * also be expressed that at a given position in the configuration tree a
301     * new branch should be added. In the example above we did not want to add
302     * an additional {@code name} element to the last field of the table,
303     * but we want a complete new {@code field} element. This can be
304     * achieved by specifying an invalid index (like -1) after the element where
305     * a new branch should be created. Given this our example would run:
306     * </p>
307     * <pre>
308     * config.addProperty(&quot;tables.table(1).fields.field(-1).name&quot;, &quot;newField&quot;);
309     * </pre>
310     * <p>
311     * With this notation it is possible to add new branches everywhere. We
312     * could for instance create a new {@code table} element by
313     * specifying
314     * </p>
315     * <pre>
316     * config.addProperty(&quot;tables.table(-1).fields.field.name&quot;, &quot;newField2&quot;);
317     * </pre>
318     * <p>
319     * (Note that because after the {@code table} element a new branch is
320     * created indices in following elements are not relevant; the branch is new
321     * so there cannot be any ambiguities.)
322     * </p>
323     *
324     * @param <T> the type of the nodes to be dealt with
325     * @param root the root node of the nodes hierarchy
326     * @param key the key of the new property
327     * @param handler the node handler
328     * @return a data object with information needed for the add operation
329     */
330    @Override
331    public <T> NodeAddData<T> prepareAdd(final T root, final String key, final NodeHandler<T> handler)
332    {
333        final DefaultConfigurationKey.KeyIterator it = new DefaultConfigurationKey(
334                this, key).iterator();
335        if (!it.hasNext())
336        {
337            throw new IllegalArgumentException(
338                    "Key for add operation must be defined!");
339        }
340
341        final T parent = findLastPathNode(it, root, handler);
342        final List<String> pathNodes = new LinkedList<>();
343
344        while (it.hasNext())
345        {
346            if (!it.isPropertyKey())
347            {
348                throw new IllegalArgumentException(
349                        "Invalid key for add operation: " + key
350                                + " (Attribute key in the middle.)");
351            }
352            pathNodes.add(it.currentKey());
353            it.next();
354        }
355
356        return new NodeAddData<>(parent, it.currentKey(), !it.isPropertyKey(),
357                pathNodes);
358    }
359
360    /**
361     * Recursive helper method for evaluating a key. This method processes all
362     * facets of a configuration key, traverses the tree of properties and
363     * fetches the results of all matching properties.
364     *
365     * @param <T> the type of nodes to be dealt with
366     * @param keyPart the configuration key iterator
367     * @param node the current node
368     * @param results here the found results are stored
369     * @param handler the node handler
370     */
371    protected <T> void findNodesForKey(
372            final DefaultConfigurationKey.KeyIterator keyPart, final T node,
373            final Collection<QueryResult<T>> results, final NodeHandler<T> handler)
374    {
375        if (!keyPart.hasNext())
376        {
377            results.add(QueryResult.createNodeResult(node));
378        }
379
380        else
381        {
382            final String key = keyPart.nextKey(false);
383            if (keyPart.isPropertyKey())
384            {
385                processSubNodes(keyPart, findChildNodesByName(handler, node, key),
386                        results, handler);
387            }
388            if (keyPart.isAttribute() && !keyPart.hasNext())
389            {
390                if (handler.getAttributeValue(node, key) != null)
391                {
392                    results.add(QueryResult.createAttributeResult(node, key));
393                }
394            }
395        }
396    }
397
398    /**
399     * Finds the last existing node for an add operation. This method traverses
400     * the node tree along the specified key. The last existing node on this
401     * path is returned.
402     *
403     * @param <T> the type of the nodes to be dealt with
404     * @param keyIt the key iterator
405     * @param node the current node
406     * @param handler the node handler
407     * @return the last existing node on the given path
408     */
409    protected <T> T findLastPathNode(final DefaultConfigurationKey.KeyIterator keyIt,
410            final T node, final NodeHandler<T> handler)
411    {
412        final String keyPart = keyIt.nextKey(false);
413
414        if (keyIt.hasNext())
415        {
416            if (!keyIt.isPropertyKey())
417            {
418                // Attribute keys can only appear as last elements of the path
419                throw new IllegalArgumentException(
420                        "Invalid path for add operation: "
421                                + "Attribute key in the middle!");
422            }
423            final int idx =
424                    keyIt.hasIndex() ? keyIt.getIndex() : handler
425                            .getMatchingChildrenCount(node, nameMatcher,
426                                    keyPart) - 1;
427            if (idx < 0
428                    || idx >= handler.getMatchingChildrenCount(node,
429                            nameMatcher, keyPart))
430            {
431                return node;
432            }
433            return findLastPathNode(keyIt,
434                    findChildNodesByName(handler, node, keyPart).get(idx),
435                    handler);
436        }
437        return node;
438    }
439
440    /**
441     * Called by {@code findNodesForKey()} to process the sub nodes of
442     * the current node depending on the type of the current key part (children,
443     * attributes, or both).
444     *
445     * @param <T> the type of the nodes to be dealt with
446     * @param keyPart the key part
447     * @param subNodes a list with the sub nodes to process
448     * @param nodes the target collection
449     * @param handler the node handler
450     */
451    private <T> void processSubNodes(final DefaultConfigurationKey.KeyIterator keyPart,
452            final List<T> subNodes, final Collection<QueryResult<T>> nodes, final NodeHandler<T> handler)
453    {
454        if (keyPart.hasIndex())
455        {
456            if (keyPart.getIndex() >= 0 && keyPart.getIndex() < subNodes.size())
457            {
458                findNodesForKey((DefaultConfigurationKey.KeyIterator) keyPart
459                        .clone(), subNodes.get(keyPart.getIndex()), nodes, handler);
460            }
461        }
462        else
463        {
464            for (final T node : subNodes)
465            {
466                findNodesForKey((DefaultConfigurationKey.KeyIterator) keyPart
467                        .clone(), node, nodes, handler);
468            }
469        }
470    }
471
472    /**
473     * Determines the index of the given node based on its parent node.
474     *
475     * @param node the current node
476     * @param parent the parent node
477     * @param nodeName the name of the current node
478     * @param handler the node handler
479     * @param <T> the type of the nodes to be dealt with
480     * @return the index of this node
481     */
482    private <T> int determineIndex(final T node, final T parent, final String nodeName,
483                                          final NodeHandler<T> handler)
484    {
485        return findChildNodesByName(handler, parent, nodeName).indexOf(node);
486    }
487
488    /**
489     * Returns a list with all child nodes of the given parent node which match
490     * the specified node name. The match is done using the current node name
491     * matcher.
492     *
493     * @param handler the {@code NodeHandler}
494     * @param parent the parent node
495     * @param nodeName the name of the current node
496     * @param <T> the type of the nodes to be dealt with
497     * @return a list with all matching child nodes
498     */
499    private <T> List<T> findChildNodesByName(final NodeHandler<T> handler, final T parent,
500            final String nodeName)
501    {
502        return handler.getMatchingChildren(parent, nameMatcher, nodeName);
503    }
504}