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 "native" 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 (".").</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 [@<attributeName>]}.</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 * <database> 052 * <tables> 053 * <table type="system"> 054 * <name>users</name> 055 * <fields> 056 * <field> 057 * <name>lid</name> 058 * <type>long</name> 059 * </field> 060 * <field> 061 * <name>usrName</name> 062 * <type>java.lang.String</type> 063 * </field> 064 * ... 065 * </fields> 066 * </table> 067 * <table> 068 * <name>documents</name> 069 * <fields> 070 * <field> 071 * <name>docid</name> 072 * <type>long</type> 073 * </field> 074 * ... 075 * </fields> 076 * </table> 077 * ... 078 * </tables> 079 * </database> 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 */ 112public class DefaultExpressionEngine implements ExpressionEngine 113{ 114 /** 115 * A default instance of this class that is used as expression engine for 116 * hierarchical configurations per default. 117 */ 118 public static final DefaultExpressionEngine INSTANCE = 119 new DefaultExpressionEngine( 120 DefaultExpressionEngineSymbols.DEFAULT_SYMBOLS); 121 122 /** The symbols used by this instance. */ 123 private final DefaultExpressionEngineSymbols symbols; 124 125 /** The matcher for node names. */ 126 private final NodeMatcher<String> nameMatcher; 127 128 /** 129 * Creates a new instance of {@code DefaultExpressionEngine} and initializes 130 * its symbols. 131 * 132 * @param syms the object with the symbols (must not be <b>null</b>) 133 * @throws IllegalArgumentException if the symbols are <b>null</b> 134 */ 135 public DefaultExpressionEngine(final DefaultExpressionEngineSymbols syms) 136 { 137 this(syms, null); 138 } 139 140 /** 141 * Creates a new instance of {@code DefaultExpressionEngine} and initializes 142 * its symbols and the matcher for comparing node names. The passed in 143 * matcher is always used when the names of nodes have to be matched against 144 * parts of configuration keys. 145 * 146 * @param syms the object with the symbols (must not be <b>null</b>) 147 * @param nodeNameMatcher the matcher for node names; can be <b>null</b>, 148 * then a default matcher is used 149 * @throws IllegalArgumentException if the symbols are <b>null</b> 150 */ 151 public DefaultExpressionEngine(final DefaultExpressionEngineSymbols syms, 152 final NodeMatcher<String> nodeNameMatcher) 153 { 154 if (syms == null) 155 { 156 throw new IllegalArgumentException("Symbols must not be null!"); 157 } 158 159 symbols = syms; 160 nameMatcher = 161 nodeNameMatcher != null ? nodeNameMatcher 162 : NodeNameMatchers.EQUALS; 163 } 164 165 /** 166 * Returns the {@code DefaultExpressionEngineSymbols} object associated with 167 * this instance. 168 * 169 * @return the {@code DefaultExpressionEngineSymbols} used by this engine 170 * @since 2.0 171 */ 172 public DefaultExpressionEngineSymbols getSymbols() 173 { 174 return symbols; 175 } 176 177 /** 178 * {@inheritDoc} This method supports the syntax as described in the class 179 * comment. 180 */ 181 @Override 182 public <T> List<QueryResult<T>> query(final T root, final String key, 183 final NodeHandler<T> handler) 184 { 185 final List<QueryResult<T>> results = new LinkedList<>(); 186 findNodesForKey(new DefaultConfigurationKey(this, key).iterator(), 187 root, results, handler); 188 return results; 189 } 190 191 /** 192 * {@inheritDoc} This implementation takes the 193 * given parent key, adds a property delimiter, and then adds the node's 194 * name. 195 * The name of the root node is a blank string. Note that no indices are 196 * returned. 197 */ 198 @Override 199 public <T> String nodeKey(final T node, final String parentKey, final NodeHandler<T> handler) 200 { 201 if (parentKey == null) 202 { 203 // this is the root node 204 return StringUtils.EMPTY; 205 } 206 final DefaultConfigurationKey key = new DefaultConfigurationKey(this, 207 parentKey); 208 key.append(handler.nodeName(node), true); 209 return key.toString(); 210 } 211 212 @Override 213 public String attributeKey(final String parentKey, final String attributeName) 214 { 215 final DefaultConfigurationKey key = 216 new DefaultConfigurationKey(this, parentKey); 217 key.appendAttribute(attributeName); 218 return key.toString(); 219 } 220 221 /** 222 * {@inheritDoc} This implementation works similar to {@code nodeKey()}; 223 * however, each key returned by this method has an index (except for the 224 * root node). The parent key is prepended to the name of the current node 225 * in any case and without further checks. If it is <b>null</b>, only the 226 * name of the current node with its index is returned. 227 */ 228 @Override 229 public <T> String canonicalKey(final T node, final String parentKey, 230 final NodeHandler<T> handler) 231 { 232 final String nodeName = handler.nodeName(node); 233 final T parent = handler.getParent(node); 234 final DefaultConfigurationKey key = 235 new DefaultConfigurationKey(this, parentKey); 236 key.append(StringUtils.defaultString(nodeName)); 237 238 if (parent != null) 239 { 240 // this is not the root key 241 key.appendIndex(determineIndex(node, parent, nodeName, handler)); 242 } 243 return key.toString(); 244 } 245 246 /** 247 * <p> 248 * Prepares Adding the property with the specified key. 249 * </p> 250 * <p> 251 * To be able to deal with the structure supported by hierarchical 252 * configuration implementations the passed in key is of importance, 253 * especially the indices it might contain. The following example should 254 * clarify this: Suppose the current node structure looks like the 255 * following: 256 * </p> 257 * <pre> 258 * tables 259 * +-- table 260 * +-- name = user 261 * +-- fields 262 * +-- field 263 * +-- name = uid 264 * +-- field 265 * +-- name = firstName 266 * ... 267 * +-- table 268 * +-- name = documents 269 * +-- fields 270 * ... 271 * </pre> 272 * <p> 273 * In this example a database structure is defined, e.g. all fields of the 274 * first table could be accessed using the key 275 * {@code tables.table(0).fields.field.name}. If now properties are 276 * to be added, it must be exactly specified at which position in the 277 * hierarchy the new property is to be inserted. So to add a new field name 278 * to a table it is not enough to say just 279 * </p> 280 * <pre> 281 * config.addProperty("tables.table.fields.field.name", "newField"); 282 * </pre> 283 * <p> 284 * The statement given above contains some ambiguity. For instance it is not 285 * clear, to which table the new field should be added. If this method finds 286 * such an ambiguity, it is resolved by following the last valid path. Here 287 * this would be the last table. The same is true for the {@code field}; 288 * because there are multiple fields and no explicit index is provided, a 289 * new {@code name} property would be added to the last field - which 290 * is probably not what was desired. 291 * </p> 292 * <p> 293 * To make things clear explicit indices should be provided whenever 294 * possible. In the example above the exact table could be specified by 295 * providing an index for the {@code table} element as in 296 * {@code tables.table(1).fields}. By specifying an index it can 297 * also be expressed that at a given position in the configuration tree a 298 * new branch should be added. In the example above we did not want to add 299 * an additional {@code name} element to the last field of the table, 300 * but we want a complete new {@code field} element. This can be 301 * achieved by specifying an invalid index (like -1) after the element where 302 * a new branch should be created. Given this our example would run: 303 * </p> 304 * <pre> 305 * config.addProperty("tables.table(1).fields.field(-1).name", "newField"); 306 * </pre> 307 * <p> 308 * With this notation it is possible to add new branches everywhere. We 309 * could for instance create a new {@code table} element by 310 * specifying 311 * </p> 312 * <pre> 313 * config.addProperty("tables.table(-1).fields.field.name", "newField2"); 314 * </pre> 315 * <p> 316 * (Note that because after the {@code table} element a new branch is 317 * created indices in following elements are not relevant; the branch is new 318 * so there cannot be any ambiguities.) 319 * </p> 320 * 321 * @param <T> the type of the nodes to be dealt with 322 * @param root the root node of the nodes hierarchy 323 * @param key the key of the new property 324 * @param handler the node handler 325 * @return a data object with information needed for the add operation 326 */ 327 @Override 328 public <T> NodeAddData<T> prepareAdd(final T root, final String key, final NodeHandler<T> handler) 329 { 330 final DefaultConfigurationKey.KeyIterator it = new DefaultConfigurationKey( 331 this, key).iterator(); 332 if (!it.hasNext()) 333 { 334 throw new IllegalArgumentException( 335 "Key for add operation must be defined!"); 336 } 337 338 final T parent = findLastPathNode(it, root, handler); 339 final List<String> pathNodes = new LinkedList<>(); 340 341 while (it.hasNext()) 342 { 343 if (!it.isPropertyKey()) 344 { 345 throw new IllegalArgumentException( 346 "Invalid key for add operation: " + key 347 + " (Attribute key in the middle.)"); 348 } 349 pathNodes.add(it.currentKey()); 350 it.next(); 351 } 352 353 return new NodeAddData<>(parent, it.currentKey(), !it.isPropertyKey(), 354 pathNodes); 355 } 356 357 /** 358 * Recursive helper method for evaluating a key. This method processes all 359 * facets of a configuration key, traverses the tree of properties and 360 * fetches the results of all matching properties. 361 * 362 * @param <T> the type of nodes to be dealt with 363 * @param keyPart the configuration key iterator 364 * @param node the current node 365 * @param results here the found results are stored 366 * @param handler the node handler 367 */ 368 protected <T> void findNodesForKey( 369 final DefaultConfigurationKey.KeyIterator keyPart, final T node, 370 final Collection<QueryResult<T>> results, final NodeHandler<T> handler) 371 { 372 if (!keyPart.hasNext()) 373 { 374 results.add(QueryResult.createNodeResult(node)); 375 } 376 377 else 378 { 379 final String key = keyPart.nextKey(false); 380 if (keyPart.isPropertyKey()) 381 { 382 processSubNodes(keyPart, findChildNodesByName(handler, node, key), 383 results, handler); 384 } 385 if (keyPart.isAttribute() && !keyPart.hasNext()) 386 { 387 if (handler.getAttributeValue(node, key) != null) 388 { 389 results.add(QueryResult.createAttributeResult(node, key)); 390 } 391 } 392 } 393 } 394 395 /** 396 * Finds the last existing node for an add operation. This method traverses 397 * the node tree along the specified key. The last existing node on this 398 * path is returned. 399 * 400 * @param <T> the type of the nodes to be dealt with 401 * @param keyIt the key iterator 402 * @param node the current node 403 * @param handler the node handler 404 * @return the last existing node on the given path 405 */ 406 protected <T> T findLastPathNode(final DefaultConfigurationKey.KeyIterator keyIt, 407 final T node, final NodeHandler<T> handler) 408 { 409 final String keyPart = keyIt.nextKey(false); 410 411 if (keyIt.hasNext()) 412 { 413 if (!keyIt.isPropertyKey()) 414 { 415 // Attribute keys can only appear as last elements of the path 416 throw new IllegalArgumentException( 417 "Invalid path for add operation: " 418 + "Attribute key in the middle!"); 419 } 420 final int idx = 421 keyIt.hasIndex() ? keyIt.getIndex() : handler 422 .getMatchingChildrenCount(node, nameMatcher, 423 keyPart) - 1; 424 if (idx < 0 425 || idx >= handler.getMatchingChildrenCount(node, 426 nameMatcher, keyPart)) 427 { 428 return node; 429 } 430 return findLastPathNode(keyIt, 431 findChildNodesByName(handler, node, keyPart).get(idx), 432 handler); 433 } 434 return node; 435 } 436 437 /** 438 * Called by {@code findNodesForKey()} to process the sub nodes of 439 * the current node depending on the type of the current key part (children, 440 * attributes, or both). 441 * 442 * @param <T> the type of the nodes to be dealt with 443 * @param keyPart the key part 444 * @param subNodes a list with the sub nodes to process 445 * @param nodes the target collection 446 * @param handler the node handler 447 */ 448 private <T> void processSubNodes(final DefaultConfigurationKey.KeyIterator keyPart, 449 final List<T> subNodes, final Collection<QueryResult<T>> nodes, final NodeHandler<T> handler) 450 { 451 if (keyPart.hasIndex()) 452 { 453 if (keyPart.getIndex() >= 0 && keyPart.getIndex() < subNodes.size()) 454 { 455 findNodesForKey((DefaultConfigurationKey.KeyIterator) keyPart 456 .clone(), subNodes.get(keyPart.getIndex()), nodes, handler); 457 } 458 } 459 else 460 { 461 for (final T node : subNodes) 462 { 463 findNodesForKey((DefaultConfigurationKey.KeyIterator) keyPart 464 .clone(), node, nodes, handler); 465 } 466 } 467 } 468 469 /** 470 * Determines the index of the given node based on its parent node. 471 * 472 * @param node the current node 473 * @param parent the parent node 474 * @param nodeName the name of the current node 475 * @param handler the node handler 476 * @param <T> the type of the nodes to be dealt with 477 * @return the index of this node 478 */ 479 private <T> int determineIndex(final T node, final T parent, final String nodeName, 480 final NodeHandler<T> handler) 481 { 482 return findChildNodesByName(handler, parent, nodeName).indexOf(node); 483 } 484 485 /** 486 * Returns a list with all child nodes of the given parent node which match 487 * the specified node name. The match is done using the current node name 488 * matcher. 489 * 490 * @param handler the {@code NodeHandler} 491 * @param parent the parent node 492 * @param nodeName the name of the current node 493 * @param <T> the type of the nodes to be dealt with 494 * @return a list with all matching child nodes 495 */ 496 private <T> List<T> findChildNodesByName(final NodeHandler<T> handler, final T parent, 497 final String nodeName) 498 { 499 return handler.getMatchingChildren(parent, nameMatcher, nodeName); 500 } 501}