1 | /* |
2 | * Copyright (C) 2005 Frerich Raabe <raabe@kde.org> |
3 | * Copyright (C) 2006, 2009 Apple Inc. All rights reserved. |
4 | * Copyright (C) 2007 Alexey Proskuryakov <ap@webkit.org> |
5 | * |
6 | * Redistribution and use in source and binary forms, with or without |
7 | * modification, are permitted provided that the following conditions |
8 | * are met: |
9 | * |
10 | * 1. Redistributions of source code must retain the above copyright |
11 | * notice, this list of conditions and the following disclaimer. |
12 | * 2. Redistributions in binary form must reproduce the above copyright |
13 | * notice, this list of conditions and the following disclaimer in the |
14 | * documentation and/or other materials provided with the distribution. |
15 | * |
16 | * THIS SOFTWARE IS PROVIDED BY THE AUTHOR ``AS IS'' AND ANY EXPRESS OR |
17 | * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES |
18 | * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. |
19 | * IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR ANY DIRECT, INDIRECT, |
20 | * INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT |
21 | * NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, |
22 | * DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY |
23 | * THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
24 | * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF |
25 | * THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
26 | */ |
27 | |
28 | #include "config.h" |
29 | #include "XPathStep.h" |
30 | |
31 | #include "Attr.h" |
32 | #include "Document.h" |
33 | #include "HTMLElement.h" |
34 | #include "NodeTraversal.h" |
35 | #include "XMLNSNames.h" |
36 | #include "XPathParser.h" |
37 | #include "XPathUtil.h" |
38 | |
39 | namespace WebCore { |
40 | namespace XPath { |
41 | |
42 | Step::Step(Axis axis, NodeTest nodeTest) |
43 | : m_axis(axis) |
44 | , m_nodeTest(WTFMove(nodeTest)) |
45 | { |
46 | } |
47 | |
48 | Step::Step(Axis axis, NodeTest nodeTest, Vector<std::unique_ptr<Expression>> predicates) |
49 | : m_axis(axis) |
50 | , m_nodeTest(WTFMove(nodeTest)) |
51 | , m_predicates(WTFMove(predicates)) |
52 | { |
53 | } |
54 | |
55 | Step::~Step() = default; |
56 | |
57 | void Step::optimize() |
58 | { |
59 | // Evaluate predicates as part of node test if possible to avoid building unnecessary NodeSets. |
60 | // E.g., there is no need to build a set of all "foo" nodes to evaluate "foo[@bar]", we can check the predicate while enumerating. |
61 | // This optimization can be applied to predicates that are not context node list sensitive, or to first predicate that is only context position sensitive, e.g. foo[position() mod 2 = 0]. |
62 | Vector<std::unique_ptr<Expression>> remainingPredicates; |
63 | for (auto& predicate : m_predicates) { |
64 | if ((!predicateIsContextPositionSensitive(*predicate) || m_nodeTest.m_mergedPredicates.isEmpty()) && !predicate->isContextSizeSensitive() && remainingPredicates.isEmpty()) |
65 | m_nodeTest.m_mergedPredicates.append(WTFMove(predicate)); |
66 | else |
67 | remainingPredicates.append(WTFMove(predicate)); |
68 | } |
69 | m_predicates = WTFMove(remainingPredicates); |
70 | } |
71 | |
72 | void optimizeStepPair(Step& first, Step& second, bool& dropSecondStep) |
73 | { |
74 | dropSecondStep = false; |
75 | |
76 | if (first.m_axis != Step::DescendantOrSelfAxis) |
77 | return; |
78 | |
79 | if (first.m_nodeTest.m_kind != Step::NodeTest::AnyNodeTest) |
80 | return; |
81 | |
82 | if (!first.m_predicates.isEmpty()) |
83 | return; |
84 | |
85 | if (!first.m_nodeTest.m_mergedPredicates.isEmpty()) |
86 | return; |
87 | |
88 | ASSERT(first.m_nodeTest.m_data.isEmpty()); |
89 | ASSERT(first.m_nodeTest.m_namespaceURI.isEmpty()); |
90 | |
91 | // Optimize the common case of "//" AKA /descendant-or-self::node()/child::NodeTest to /descendant::NodeTest. |
92 | if (second.m_axis != Step::ChildAxis) |
93 | return; |
94 | |
95 | if (!second.predicatesAreContextListInsensitive()) |
96 | return; |
97 | |
98 | first.m_axis = Step::DescendantAxis; |
99 | first.m_nodeTest = WTFMove(second.m_nodeTest); |
100 | first.m_predicates = WTFMove(second.m_predicates); |
101 | first.optimize(); |
102 | dropSecondStep = true; |
103 | } |
104 | |
105 | bool Step::predicatesAreContextListInsensitive() const |
106 | { |
107 | for (auto& predicate : m_predicates) { |
108 | if (predicateIsContextPositionSensitive(*predicate) || predicate->isContextSizeSensitive()) |
109 | return false; |
110 | } |
111 | |
112 | for (auto& predicate : m_nodeTest.m_mergedPredicates) { |
113 | if (predicateIsContextPositionSensitive(*predicate) || predicate->isContextSizeSensitive()) |
114 | return false; |
115 | } |
116 | |
117 | return true; |
118 | } |
119 | |
120 | void Step::evaluate(Node& context, NodeSet& nodes) const |
121 | { |
122 | EvaluationContext& evaluationContext = Expression::evaluationContext(); |
123 | evaluationContext.position = 0; |
124 | |
125 | nodesInAxis(context, nodes); |
126 | |
127 | // Check predicates that couldn't be merged into node test. |
128 | for (auto& predicate : m_predicates) { |
129 | NodeSet newNodes; |
130 | if (!nodes.isSorted()) |
131 | newNodes.markSorted(false); |
132 | |
133 | for (unsigned j = 0; j < nodes.size(); j++) { |
134 | Node* node = nodes[j]; |
135 | |
136 | evaluationContext.node = node; |
137 | evaluationContext.size = nodes.size(); |
138 | evaluationContext.position = j + 1; |
139 | if (evaluatePredicate(*predicate)) |
140 | newNodes.append(node); |
141 | } |
142 | |
143 | nodes = WTFMove(newNodes); |
144 | } |
145 | } |
146 | |
147 | #if !ASSERT_DISABLED |
148 | static inline Node::NodeType primaryNodeType(Step::Axis axis) |
149 | { |
150 | switch (axis) { |
151 | case Step::AttributeAxis: |
152 | return Node::ATTRIBUTE_NODE; |
153 | default: |
154 | return Node::ELEMENT_NODE; |
155 | } |
156 | } |
157 | #endif |
158 | |
159 | // Evaluate NodeTest without considering merged predicates. |
160 | inline bool nodeMatchesBasicTest(Node& node, Step::Axis axis, const Step::NodeTest& nodeTest) |
161 | { |
162 | switch (nodeTest.m_kind) { |
163 | case Step::NodeTest::TextNodeTest: |
164 | return node.nodeType() == Node::TEXT_NODE || node.nodeType() == Node::CDATA_SECTION_NODE; |
165 | case Step::NodeTest::CommentNodeTest: |
166 | return node.nodeType() == Node::COMMENT_NODE; |
167 | case Step::NodeTest::ProcessingInstructionNodeTest: { |
168 | const AtomicString& name = nodeTest.m_data; |
169 | return node.nodeType() == Node::PROCESSING_INSTRUCTION_NODE && (name.isEmpty() || node.nodeName() == name); |
170 | } |
171 | case Step::NodeTest::AnyNodeTest: |
172 | return true; |
173 | case Step::NodeTest::NameTest: { |
174 | const AtomicString& name = nodeTest.m_data; |
175 | const AtomicString& namespaceURI = nodeTest.m_namespaceURI; |
176 | |
177 | if (axis == Step::AttributeAxis) { |
178 | ASSERT(node.isAttributeNode()); |
179 | |
180 | // In XPath land, namespace nodes are not accessible on the attribute axis. |
181 | if (node.namespaceURI() == XMLNSNames::xmlnsNamespaceURI) |
182 | return false; |
183 | |
184 | if (name == starAtom()) |
185 | return namespaceURI.isEmpty() || node.namespaceURI() == namespaceURI; |
186 | |
187 | return node.localName() == name && node.namespaceURI() == namespaceURI; |
188 | } |
189 | |
190 | // Node test on the namespace axis is not implemented yet, the caller has a check for it. |
191 | ASSERT(axis != Step::NamespaceAxis); |
192 | |
193 | // For other axes, the principal node type is element. |
194 | ASSERT(primaryNodeType(axis) == Node::ELEMENT_NODE); |
195 | if (!is<Element>(node)) |
196 | return false; |
197 | |
198 | if (name == starAtom()) |
199 | return namespaceURI.isEmpty() || namespaceURI == node.namespaceURI(); |
200 | |
201 | if (node.document().isHTMLDocument()) { |
202 | if (is<HTMLElement>(node)) { |
203 | // Paths without namespaces should match HTML elements in HTML documents despite those having an XHTML namespace. Names are compared case-insensitively. |
204 | return equalIgnoringASCIICase(downcast<HTMLElement>(node).localName(), name) && (namespaceURI.isNull() || namespaceURI == node.namespaceURI()); |
205 | } |
206 | // An expression without any prefix shouldn't match no-namespace nodes (because HTML5 says so). |
207 | return downcast<Element>(node).hasLocalName(name) && namespaceURI == node.namespaceURI() && !namespaceURI.isNull(); |
208 | } |
209 | return downcast<Element>(node).hasLocalName(name) && namespaceURI == node.namespaceURI(); |
210 | } |
211 | } |
212 | ASSERT_NOT_REACHED(); |
213 | return false; |
214 | } |
215 | |
216 | inline bool nodeMatches(Node& node, Step::Axis axis, const Step::NodeTest& nodeTest) |
217 | { |
218 | if (!nodeMatchesBasicTest(node, axis, nodeTest)) |
219 | return false; |
220 | |
221 | EvaluationContext& evaluationContext = Expression::evaluationContext(); |
222 | |
223 | // Only the first merged predicate may depend on position. |
224 | ++evaluationContext.position; |
225 | |
226 | for (auto& predicate : nodeTest.m_mergedPredicates) { |
227 | // No need to set context size - we only get here when evaluating predicates that do not depend on it. |
228 | evaluationContext.node = &node; |
229 | if (!evaluatePredicate(*predicate)) |
230 | return false; |
231 | } |
232 | |
233 | return true; |
234 | } |
235 | |
236 | // Result nodes are ordered in axis order. Node test (including merged predicates) is applied. |
237 | void Step::nodesInAxis(Node& context, NodeSet& nodes) const |
238 | { |
239 | ASSERT(nodes.isEmpty()); |
240 | switch (m_axis) { |
241 | case ChildAxis: |
242 | if (context.isAttributeNode()) // In XPath model, attribute nodes do not have children. |
243 | return; |
244 | for (Node* node = context.firstChild(); node; node = node->nextSibling()) { |
245 | if (nodeMatches(*node, ChildAxis, m_nodeTest)) |
246 | nodes.append(node); |
247 | } |
248 | return; |
249 | case DescendantAxis: |
250 | if (context.isAttributeNode()) // In XPath model, attribute nodes do not have children. |
251 | return; |
252 | for (Node* node = context.firstChild(); node; node = NodeTraversal::next(*node, &context)) { |
253 | if (nodeMatches(*node, DescendantAxis, m_nodeTest)) |
254 | nodes.append(node); |
255 | } |
256 | return; |
257 | case ParentAxis: |
258 | if (context.isAttributeNode()) { |
259 | Element* node = static_cast<Attr&>(context).ownerElement(); |
260 | if (nodeMatches(*node, ParentAxis, m_nodeTest)) |
261 | nodes.append(node); |
262 | } else { |
263 | ContainerNode* node = context.parentNode(); |
264 | if (node && nodeMatches(*node, ParentAxis, m_nodeTest)) |
265 | nodes.append(node); |
266 | } |
267 | return; |
268 | case AncestorAxis: { |
269 | Node* node = &context; |
270 | if (context.isAttributeNode()) { |
271 | node = static_cast<Attr&>(context).ownerElement(); |
272 | if (nodeMatches(*node, AncestorAxis, m_nodeTest)) |
273 | nodes.append(node); |
274 | } |
275 | for (node = node->parentNode(); node; node = node->parentNode()) { |
276 | if (nodeMatches(*node, AncestorAxis, m_nodeTest)) |
277 | nodes.append(node); |
278 | } |
279 | nodes.markSorted(false); |
280 | return; |
281 | } |
282 | case FollowingSiblingAxis: |
283 | if (context.isAttributeNode()) |
284 | return; |
285 | for (Node* node = context.nextSibling(); node; node = node->nextSibling()) { |
286 | if (nodeMatches(*node, FollowingSiblingAxis, m_nodeTest)) |
287 | nodes.append(node); |
288 | } |
289 | return; |
290 | case PrecedingSiblingAxis: |
291 | if (context.isAttributeNode()) |
292 | return; |
293 | for (Node* node = context.previousSibling(); node; node = node->previousSibling()) { |
294 | if (nodeMatches(*node, PrecedingSiblingAxis, m_nodeTest)) |
295 | nodes.append(node); |
296 | } |
297 | nodes.markSorted(false); |
298 | return; |
299 | case FollowingAxis: |
300 | if (context.isAttributeNode()) { |
301 | Node* node = static_cast<Attr&>(context).ownerElement(); |
302 | while ((node = NodeTraversal::next(*node))) { |
303 | if (nodeMatches(*node, FollowingAxis, m_nodeTest)) |
304 | nodes.append(node); |
305 | } |
306 | } else { |
307 | for (Node* parent = &context; !isRootDomNode(parent); parent = parent->parentNode()) { |
308 | for (Node* node = parent->nextSibling(); node; node = node->nextSibling()) { |
309 | if (nodeMatches(*node, FollowingAxis, m_nodeTest)) |
310 | nodes.append(node); |
311 | for (Node* child = node->firstChild(); child; child = NodeTraversal::next(*child, node)) { |
312 | if (nodeMatches(*child, FollowingAxis, m_nodeTest)) |
313 | nodes.append(child); |
314 | } |
315 | } |
316 | } |
317 | } |
318 | return; |
319 | case PrecedingAxis: { |
320 | Node* node; |
321 | if (context.isAttributeNode()) |
322 | node = static_cast<Attr&>(context).ownerElement(); |
323 | else |
324 | node = &context; |
325 | while (ContainerNode* parent = node->parentNode()) { |
326 | for (node = NodeTraversal::previous(*node); node != parent; node = NodeTraversal::previous(*node)) { |
327 | if (nodeMatches(*node, PrecedingAxis, m_nodeTest)) |
328 | nodes.append(node); |
329 | } |
330 | node = parent; |
331 | } |
332 | nodes.markSorted(false); |
333 | return; |
334 | } |
335 | case AttributeAxis: { |
336 | if (!is<Element>(context)) |
337 | return; |
338 | |
339 | Element& contextElement = downcast<Element>(context); |
340 | |
341 | // Avoid lazily creating attribute nodes for attributes that we do not need anyway. |
342 | if (m_nodeTest.m_kind == NodeTest::NameTest && m_nodeTest.m_data != starAtom()) { |
343 | auto attr = contextElement.getAttributeNodeNS(m_nodeTest.m_namespaceURI, m_nodeTest.m_data); |
344 | if (attr && attr->namespaceURI() != XMLNSNames::xmlnsNamespaceURI) { // In XPath land, namespace nodes are not accessible on the attribute axis. |
345 | if (nodeMatches(*attr, AttributeAxis, m_nodeTest)) // Still need to check merged predicates. |
346 | nodes.append(WTFMove(attr)); |
347 | } |
348 | return; |
349 | } |
350 | |
351 | if (!contextElement.hasAttributes()) |
352 | return; |
353 | |
354 | for (const Attribute& attribute : contextElement.attributesIterator()) { |
355 | auto attr = contextElement.ensureAttr(attribute.name()); |
356 | if (nodeMatches(attr.get(), AttributeAxis, m_nodeTest)) |
357 | nodes.append(WTFMove(attr)); |
358 | } |
359 | return; |
360 | } |
361 | case NamespaceAxis: |
362 | // XPath namespace nodes are not implemented yet. |
363 | return; |
364 | case SelfAxis: |
365 | if (nodeMatches(context, SelfAxis, m_nodeTest)) |
366 | nodes.append(&context); |
367 | return; |
368 | case DescendantOrSelfAxis: |
369 | if (nodeMatches(context, DescendantOrSelfAxis, m_nodeTest)) |
370 | nodes.append(&context); |
371 | if (context.isAttributeNode()) // In XPath model, attribute nodes do not have children. |
372 | return; |
373 | for (Node* node = context.firstChild(); node; node = NodeTraversal::next(*node, &context)) { |
374 | if (nodeMatches(*node, DescendantOrSelfAxis, m_nodeTest)) |
375 | nodes.append(node); |
376 | } |
377 | return; |
378 | case AncestorOrSelfAxis: { |
379 | if (nodeMatches(context, AncestorOrSelfAxis, m_nodeTest)) |
380 | nodes.append(&context); |
381 | Node* node = &context; |
382 | if (context.isAttributeNode()) { |
383 | node = static_cast<Attr&>(context).ownerElement(); |
384 | if (nodeMatches(*node, AncestorOrSelfAxis, m_nodeTest)) |
385 | nodes.append(node); |
386 | } |
387 | for (node = node->parentNode(); node; node = node->parentNode()) { |
388 | if (nodeMatches(*node, AncestorOrSelfAxis, m_nodeTest)) |
389 | nodes.append(node); |
390 | } |
391 | nodes.markSorted(false); |
392 | return; |
393 | } |
394 | } |
395 | ASSERT_NOT_REACHED(); |
396 | } |
397 | |
398 | } |
399 | } |
400 | |