| 1 | /* |
| 2 | * Copyright (C) 2015 Apple Inc. All rights reserved. |
| 3 | * |
| 4 | * Redistribution and use in source and binary forms, with or without |
| 5 | * modification, are permitted provided that the following conditions |
| 6 | * are met: |
| 7 | * 1. Redistributions of source code must retain the above copyright |
| 8 | * notice, this list of conditions and the following disclaimer. |
| 9 | * 2. Redistributions in binary form must reproduce the above copyright |
| 10 | * notice, this list of conditions and the following disclaimer in the |
| 11 | * documentation and/or other materials provided with the distribution. |
| 12 | * |
| 13 | * THIS SOFTWARE IS PROVIDED BY APPLE INC. AND ITS CONTRIBUTORS ``AS IS'' |
| 14 | * AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, |
| 15 | * THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR |
| 16 | * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL APPLE INC. OR ITS CONTRIBUTORS |
| 17 | * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR |
| 18 | * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF |
| 19 | * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS |
| 20 | * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN |
| 21 | * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) |
| 22 | * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF |
| 23 | * THE POSSIBILITY OF SUCH DAMAGE. |
| 24 | */ |
| 25 | |
| 26 | #include "config.h" |
| 27 | #include "CombinedURLFilters.h" |
| 28 | |
| 29 | #if ENABLE(CONTENT_EXTENSIONS) |
| 30 | |
| 31 | #include "HashableActionList.h" |
| 32 | #include "Term.h" |
| 33 | #include <wtf/DataLog.h> |
| 34 | #include <wtf/Vector.h> |
| 35 | #include <wtf/text/CString.h> |
| 36 | |
| 37 | namespace WebCore { |
| 38 | |
| 39 | namespace ContentExtensions { |
| 40 | |
| 41 | struct PrefixTreeEdge { |
| 42 | const Term* term; |
| 43 | std::unique_ptr<PrefixTreeVertex> child; |
| 44 | }; |
| 45 | |
| 46 | typedef Vector<PrefixTreeEdge, 0, WTF::CrashOnOverflow, 1> PrefixTreeEdges; |
| 47 | |
| 48 | struct PrefixTreeVertex { |
| 49 | WTF_MAKE_STRUCT_FAST_ALLOCATED; |
| 50 | |
| 51 | PrefixTreeEdges edges; |
| 52 | }; |
| 53 | |
| 54 | struct ReverseSuffixTreeVertex; |
| 55 | struct ReverseSuffixTreeEdge { |
| 56 | const Term* term; |
| 57 | std::unique_ptr<ReverseSuffixTreeVertex> child; |
| 58 | }; |
| 59 | typedef Vector<ReverseSuffixTreeEdge, 0, WTF::CrashOnOverflow, 1> ReverseSuffixTreeEdges; |
| 60 | |
| 61 | struct ReverseSuffixTreeVertex { |
| 62 | ReverseSuffixTreeEdges edges; |
| 63 | uint32_t nodeId; |
| 64 | }; |
| 65 | typedef HashMap<HashableActionList, ReverseSuffixTreeVertex, HashableActionListHash, HashableActionListHashTraits> ReverseSuffixTreeRoots; |
| 66 | |
| 67 | #if CONTENT_EXTENSIONS_PERFORMANCE_REPORTING |
| 68 | static size_t recursiveMemoryUsed(const PrefixTreeVertex& vertex) |
| 69 | { |
| 70 | size_t size = sizeof(PrefixTreeVertex) |
| 71 | + vertex.edges.capacity() * sizeof(PrefixTreeEdge); |
| 72 | for (const auto& edge : vertex.edges) { |
| 73 | ASSERT(edge.child); |
| 74 | size += recursiveMemoryUsed(*edge.child.get()); |
| 75 | } |
| 76 | return size; |
| 77 | } |
| 78 | |
| 79 | size_t CombinedURLFilters::memoryUsed() const |
| 80 | { |
| 81 | ASSERT(m_prefixTreeRoot); |
| 82 | |
| 83 | size_t actionsSize = 0; |
| 84 | for (const auto& slot : m_actions) |
| 85 | actionsSize += slot.value.capacity() * sizeof(uint64_t); |
| 86 | |
| 87 | return sizeof(CombinedURLFilters) |
| 88 | + m_alphabet.memoryUsed() |
| 89 | + recursiveMemoryUsed(*m_prefixTreeRoot.get()) |
| 90 | + sizeof(HashMap<PrefixTreeVertex*, ActionList>) |
| 91 | + m_actions.capacity() * (sizeof(PrefixTreeVertex*) + sizeof(ActionList)) |
| 92 | + actionsSize; |
| 93 | } |
| 94 | #endif |
| 95 | |
| 96 | #if CONTENT_EXTENSIONS_STATE_MACHINE_DEBUGGING |
| 97 | static String prefixTreeVertexToString(const PrefixTreeVertex& vertex, const HashMap<const PrefixTreeVertex*, ActionList>& actions, unsigned depth) |
| 98 | { |
| 99 | StringBuilder builder; |
| 100 | while (depth--) |
| 101 | builder.appendLiteral(" " ); |
| 102 | builder.appendLiteral("vertex actions: " ); |
| 103 | |
| 104 | auto actionsSlot = actions.find(&vertex); |
| 105 | if (actionsSlot != actions.end()) { |
| 106 | for (uint64_t action : actionsSlot->value) { |
| 107 | builder.appendNumber(action); |
| 108 | builder.append(','); |
| 109 | } |
| 110 | } |
| 111 | builder.append('\n'); |
| 112 | return builder.toString(); |
| 113 | } |
| 114 | |
| 115 | static void recursivePrint(const PrefixTreeVertex& vertex, const HashMap<const PrefixTreeVertex*, ActionList>& actions, unsigned depth) |
| 116 | { |
| 117 | dataLogF("%s" , prefixTreeVertexToString(vertex, actions, depth).utf8().data()); |
| 118 | for (const auto& edge : vertex.edges) { |
| 119 | StringBuilder builder; |
| 120 | for (unsigned i = 0; i < depth * 2; ++i) |
| 121 | builder.append(' '); |
| 122 | builder.appendLiteral("vertex edge: " ); |
| 123 | builder.append(edge.term->toString()); |
| 124 | builder.append('\n'); |
| 125 | dataLogF("%s" , builder.toString().utf8().data()); |
| 126 | ASSERT(edge.child); |
| 127 | recursivePrint(*edge.child.get(), actions, depth + 1); |
| 128 | } |
| 129 | } |
| 130 | |
| 131 | void CombinedURLFilters::print() const |
| 132 | { |
| 133 | recursivePrint(*m_prefixTreeRoot.get(), m_actions, 0); |
| 134 | } |
| 135 | #endif |
| 136 | |
| 137 | CombinedURLFilters::CombinedURLFilters() |
| 138 | : m_prefixTreeRoot(std::make_unique<PrefixTreeVertex>()) |
| 139 | { |
| 140 | } |
| 141 | |
| 142 | CombinedURLFilters::~CombinedURLFilters() = default; |
| 143 | |
| 144 | bool CombinedURLFilters::isEmpty() const |
| 145 | { |
| 146 | return m_prefixTreeRoot->edges.isEmpty(); |
| 147 | } |
| 148 | |
| 149 | void CombinedURLFilters::addDomain(uint64_t actionId, const String& domain) |
| 150 | { |
| 151 | unsigned domainLength = domain.length(); |
| 152 | if (domainLength && domain[0] == '*') { |
| 153 | // If domain starts with a '*' then it means match domain and its subdomains, like (^|.)domain$ |
| 154 | // This way a domain of "*webkit.org" will match "bugs.webkit.org" and "webkit.org". |
| 155 | Vector<Term> prependDot; |
| 156 | Vector<Term> prependBeginningOfLine; |
| 157 | prependDot.reserveInitialCapacity(domainLength + 2); |
| 158 | prependBeginningOfLine.reserveInitialCapacity(domainLength); // This is just no .* at the beginning. |
| 159 | |
| 160 | Term canonicalDotStar(Term::UniversalTransition); |
| 161 | canonicalDotStar.quantify(AtomQuantifier::ZeroOrMore); |
| 162 | prependDot.uncheckedAppend(canonicalDotStar); |
| 163 | prependDot.uncheckedAppend(Term('.', true)); |
| 164 | |
| 165 | for (unsigned i = 1; i < domainLength; i++) { |
| 166 | ASSERT(isASCII(domain[i])); |
| 167 | ASSERT(!isASCIIUpper(domain[i])); |
| 168 | prependDot.uncheckedAppend(Term(domain[i], true)); |
| 169 | prependBeginningOfLine.uncheckedAppend(Term(domain[i], true)); |
| 170 | } |
| 171 | prependDot.uncheckedAppend(Term::EndOfLineAssertionTerm); |
| 172 | prependBeginningOfLine.uncheckedAppend(Term::EndOfLineAssertionTerm); |
| 173 | |
| 174 | addPattern(actionId, prependDot); |
| 175 | addPattern(actionId, prependBeginningOfLine); |
| 176 | } else { |
| 177 | // This is like adding ^domain$, but interpreting domain as a series of characters, not a regular expression. |
| 178 | // "webkit.org" will match "webkit.org" but not "bugs.webkit.org". |
| 179 | Vector<Term> prependBeginningOfLine; |
| 180 | prependBeginningOfLine.reserveInitialCapacity(domainLength + 1); // This is just no .* at the beginning. |
| 181 | for (unsigned i = 0; i < domainLength; i++) { |
| 182 | ASSERT(isASCII(domain[i])); |
| 183 | ASSERT(!isASCIIUpper(domain[i])); |
| 184 | prependBeginningOfLine.uncheckedAppend(Term(domain[i], true)); |
| 185 | } |
| 186 | prependBeginningOfLine.uncheckedAppend(Term::EndOfLineAssertionTerm); |
| 187 | addPattern(actionId, prependBeginningOfLine); |
| 188 | } |
| 189 | } |
| 190 | |
| 191 | void CombinedURLFilters::addPattern(uint64_t actionId, const Vector<Term>& pattern) |
| 192 | { |
| 193 | ASSERT_WITH_MESSAGE(!pattern.isEmpty(), "The parser should have excluded empty patterns before reaching CombinedURLFilters." ); |
| 194 | |
| 195 | if (pattern.isEmpty()) |
| 196 | return; |
| 197 | |
| 198 | // Extend the prefix tree with the new pattern. |
| 199 | PrefixTreeVertex* lastPrefixTree = m_prefixTreeRoot.get(); |
| 200 | |
| 201 | for (const Term& term : pattern) { |
| 202 | size_t nextEntryIndex = WTF::notFound; |
| 203 | for (size_t i = 0; i < lastPrefixTree->edges.size(); ++i) { |
| 204 | if (*lastPrefixTree->edges[i].term == term) { |
| 205 | nextEntryIndex = i; |
| 206 | break; |
| 207 | } |
| 208 | } |
| 209 | if (nextEntryIndex != WTF::notFound) |
| 210 | lastPrefixTree = lastPrefixTree->edges[nextEntryIndex].child.get(); |
| 211 | else { |
| 212 | lastPrefixTree->edges.append(PrefixTreeEdge({m_alphabet.interned(term), std::make_unique<PrefixTreeVertex>()})); |
| 213 | lastPrefixTree = lastPrefixTree->edges.last().child.get(); |
| 214 | } |
| 215 | } |
| 216 | |
| 217 | auto addResult = m_actions.add(lastPrefixTree, ActionList()); |
| 218 | ActionList& actions = addResult.iterator->value; |
| 219 | if (actions.find(actionId) == WTF::notFound) |
| 220 | actions.append(actionId); |
| 221 | } |
| 222 | |
| 223 | struct ActiveSubtree { |
| 224 | ActiveSubtree(PrefixTreeVertex& vertex, ImmutableCharNFANodeBuilder&& nfaNode, unsigned edgeIndex) |
| 225 | : vertex(vertex) |
| 226 | , nfaNode(WTFMove(nfaNode)) |
| 227 | , edgeIndex(edgeIndex) |
| 228 | { |
| 229 | } |
| 230 | PrefixTreeVertex& vertex; |
| 231 | ImmutableCharNFANodeBuilder nfaNode; |
| 232 | unsigned edgeIndex; |
| 233 | }; |
| 234 | |
| 235 | static void generateInfixUnsuitableForReverseSuffixTree(NFA& nfa, Vector<ActiveSubtree>& stack, const HashMap<const PrefixTreeVertex*, ActionList>& actions) |
| 236 | { |
| 237 | // To avoid conflicts, we use the reverse suffix tree for subtrees that do not merge |
| 238 | // in the prefix tree. |
| 239 | // |
| 240 | // We only unify the suffixes to the actions on the leaf. |
| 241 | // If there are actions inside the tree, we generate the part of the subtree up to the action. |
| 242 | // |
| 243 | // If we accidentally insert a node with action inside the reverse-suffix-tree, we would create |
| 244 | // new actions on unrelated pattern when unifying their suffixes. |
| 245 | for (unsigned i = stack.size() - 1; i--;) { |
| 246 | ActiveSubtree& activeSubtree = stack[i]; |
| 247 | if (activeSubtree.nfaNode.isValid()) |
| 248 | return; |
| 249 | |
| 250 | RELEASE_ASSERT_WITH_MESSAGE(i > 0, "The bottom of the stack must be the root of our fixed-length subtree. It should have it the isValid() case above." ); |
| 251 | |
| 252 | auto actionsIterator = actions.find(&activeSubtree.vertex); |
| 253 | bool hasActionInsideTree = actionsIterator != actions.end(); |
| 254 | |
| 255 | // Stricto sensu, we should count the number of exit edges with fixed length. |
| 256 | // That is costly and unlikely to matter in practice. |
| 257 | bool hasSingleOutcome = activeSubtree.vertex.edges.size() == 1; |
| 258 | |
| 259 | if (hasActionInsideTree || !hasSingleOutcome) { |
| 260 | // Go back to the end of the subtree that has already been generated. |
| 261 | // From there, generate everything up to the vertex we found. |
| 262 | unsigned end = i; |
| 263 | unsigned beginning = end; |
| 264 | |
| 265 | ActiveSubtree* sourceActiveSubtree = nullptr; |
| 266 | while (beginning--) { |
| 267 | ActiveSubtree& activeSubtree = stack[beginning]; |
| 268 | if (activeSubtree.nfaNode.isValid()) { |
| 269 | sourceActiveSubtree = &activeSubtree; |
| 270 | break; |
| 271 | } |
| 272 | } |
| 273 | ASSERT_WITH_MESSAGE(sourceActiveSubtree, "The root should always have a valid generator." ); |
| 274 | |
| 275 | for (unsigned stackIndex = beginning + 1; stackIndex <= end; ++stackIndex) { |
| 276 | ImmutableCharNFANodeBuilder& sourceNode = sourceActiveSubtree->nfaNode; |
| 277 | ASSERT(sourceNode.isValid()); |
| 278 | auto& edge = sourceActiveSubtree->vertex.edges[sourceActiveSubtree->edgeIndex]; |
| 279 | |
| 280 | ActiveSubtree& destinationActiveSubtree = stack[stackIndex]; |
| 281 | destinationActiveSubtree.nfaNode = edge.term->generateGraph(nfa, sourceNode, actions.get(&destinationActiveSubtree.vertex)); |
| 282 | |
| 283 | sourceActiveSubtree = &destinationActiveSubtree; |
| 284 | } |
| 285 | |
| 286 | return; |
| 287 | } |
| 288 | } |
| 289 | } |
| 290 | |
| 291 | static void generateSuffixWithReverseSuffixTree(NFA& nfa, Vector<ActiveSubtree>& stack, const HashMap<const PrefixTreeVertex*, ActionList>& actions, ReverseSuffixTreeRoots& reverseSuffixTreeRoots) |
| 292 | { |
| 293 | ActiveSubtree& leafSubtree = stack.last(); |
| 294 | ASSERT_WITH_MESSAGE(!leafSubtree.nfaNode.isValid(), "The leaf should never be generated by the code above, it should always be inserted into the prefix tree." ); |
| 295 | |
| 296 | ActionList actionList = actions.get(&leafSubtree.vertex); |
| 297 | ASSERT_WITH_MESSAGE(!actionList.isEmpty(), "Our prefix tree should always have actions on the leaves by construction." ); |
| 298 | |
| 299 | HashableActionList hashableActionList(actionList); |
| 300 | auto rootAddResult = reverseSuffixTreeRoots.add(hashableActionList, ReverseSuffixTreeVertex()); |
| 301 | if (rootAddResult.isNewEntry) { |
| 302 | ImmutableCharNFANodeBuilder newNode(nfa); |
| 303 | newNode.setActions(actionList.begin(), actionList.end()); |
| 304 | rootAddResult.iterator->value.nodeId = newNode.nodeId(); |
| 305 | } |
| 306 | |
| 307 | ReverseSuffixTreeVertex* activeReverseSuffixTreeVertex = &rootAddResult.iterator->value; |
| 308 | uint32_t destinationNodeId = rootAddResult.iterator->value.nodeId; |
| 309 | |
| 310 | unsigned stackPosition = stack.size() - 2; |
| 311 | while (true) { |
| 312 | ActiveSubtree& source = stack[stackPosition]; |
| 313 | auto& edge = source.vertex.edges[source.edgeIndex]; |
| 314 | |
| 315 | // This is the end condition: when we meet a node that has already been generated, |
| 316 | // we just need to connect our backward tree to the forward tree. |
| 317 | // |
| 318 | // We *must not* add this last node to the reverse-suffix tree. That node can have |
| 319 | // transitions back to earlier part of the prefix tree. If the prefix tree "caches" |
| 320 | // such node, it would create new transitions that did not exist in the source language. |
| 321 | if (source.nfaNode.isValid()) { |
| 322 | stack.shrink(stackPosition + 1); |
| 323 | edge.term->generateGraph(nfa, source.nfaNode, destinationNodeId); |
| 324 | return; |
| 325 | } |
| 326 | --stackPosition; |
| 327 | |
| 328 | ASSERT_WITH_MESSAGE(!actions.contains(&source.vertex), "Any node with final actions should have been created before hitting the reverse suffix-tree." ); |
| 329 | |
| 330 | ReverseSuffixTreeEdge* existingEdge = nullptr; |
| 331 | for (ReverseSuffixTreeEdge& potentialExistingEdge : activeReverseSuffixTreeVertex->edges) { |
| 332 | if (edge.term == potentialExistingEdge.term) { |
| 333 | existingEdge = &potentialExistingEdge; |
| 334 | break; |
| 335 | } |
| 336 | } |
| 337 | |
| 338 | if (existingEdge) |
| 339 | activeReverseSuffixTreeVertex = existingEdge->child.get(); |
| 340 | else { |
| 341 | ImmutableCharNFANodeBuilder newNode(nfa); |
| 342 | edge.term->generateGraph(nfa, newNode, destinationNodeId); |
| 343 | std::unique_ptr<ReverseSuffixTreeVertex> newVertex(new ReverseSuffixTreeVertex()); |
| 344 | newVertex->nodeId = newNode.nodeId(); |
| 345 | |
| 346 | ReverseSuffixTreeVertex* newVertexAddress = newVertex.get(); |
| 347 | activeReverseSuffixTreeVertex->edges.append(ReverseSuffixTreeEdge({ edge.term, WTFMove(newVertex) })); |
| 348 | activeReverseSuffixTreeVertex = newVertexAddress; |
| 349 | } |
| 350 | destinationNodeId = activeReverseSuffixTreeVertex->nodeId; |
| 351 | |
| 352 | ASSERT(source.vertex.edges.size() == 1); |
| 353 | source.vertex.edges.clear(); |
| 354 | } |
| 355 | |
| 356 | RELEASE_ASSERT_NOT_REACHED(); |
| 357 | } |
| 358 | |
| 359 | static void clearReverseSuffixTree(ReverseSuffixTreeRoots& reverseSuffixTreeRoots) |
| 360 | { |
| 361 | // We cannot rely on the destructor being called in order from top to bottom as we may overflow |
| 362 | // the stack. Instead, we go depth first in the reverse-suffix-tree. |
| 363 | |
| 364 | for (auto& slot : reverseSuffixTreeRoots) { |
| 365 | Vector<ReverseSuffixTreeVertex*, 128> stack; |
| 366 | stack.append(&slot.value); |
| 367 | |
| 368 | while (true) { |
| 369 | ReverseSuffixTreeVertex* top = stack.last(); |
| 370 | if (top->edges.isEmpty()) { |
| 371 | stack.removeLast(); |
| 372 | if (stack.isEmpty()) |
| 373 | break; |
| 374 | stack.last()->edges.removeLast(); |
| 375 | } else |
| 376 | stack.append(top->edges.last().child.get()); |
| 377 | } |
| 378 | } |
| 379 | reverseSuffixTreeRoots.clear(); |
| 380 | } |
| 381 | |
| 382 | static void generateNFAForSubtree(NFA& nfa, ImmutableCharNFANodeBuilder&& subtreeRoot, PrefixTreeVertex& root, const HashMap<const PrefixTreeVertex*, ActionList>& actions, size_t maxNFASize) |
| 383 | { |
| 384 | // This recurses the subtree of the prefix tree. |
| 385 | // For each edge that has fixed length (no quantifiers like ?, *, or +) it generates the nfa graph, |
| 386 | // recurses into children, and deletes any processed leaf nodes. |
| 387 | |
| 388 | ReverseSuffixTreeRoots reverseSuffixTreeRoots; |
| 389 | Vector<ActiveSubtree> stack; |
| 390 | if (!root.edges.isEmpty()) |
| 391 | stack.append(ActiveSubtree(root, WTFMove(subtreeRoot), 0)); |
| 392 | |
| 393 | bool nfaTooBig = false; |
| 394 | |
| 395 | // Generate graphs for each subtree that does not contain any quantifiers. |
| 396 | while (!stack.isEmpty()) { |
| 397 | PrefixTreeVertex& vertex = stack.last().vertex; |
| 398 | const unsigned edgeIndex = stack.last().edgeIndex; |
| 399 | |
| 400 | if (edgeIndex < vertex.edges.size()) { |
| 401 | auto& edge = vertex.edges[edgeIndex]; |
| 402 | |
| 403 | // Clean up any processed leaves and return early if we are past the maxNFASize. |
| 404 | if (nfaTooBig) { |
| 405 | stack.last().edgeIndex = stack.last().vertex.edges.size(); |
| 406 | continue; |
| 407 | } |
| 408 | |
| 409 | // Quantified edges in the subtree will be a part of another NFA. |
| 410 | if (!edge.term->hasFixedLength()) { |
| 411 | stack.last().edgeIndex++; |
| 412 | continue; |
| 413 | } |
| 414 | |
| 415 | ASSERT(edge.child.get()); |
| 416 | ImmutableCharNFANodeBuilder emptyBuilder; |
| 417 | stack.append(ActiveSubtree(*edge.child.get(), WTFMove(emptyBuilder), 0)); |
| 418 | } else { |
| 419 | bool isLeaf = vertex.edges.isEmpty(); |
| 420 | |
| 421 | ASSERT(edgeIndex == vertex.edges.size()); |
| 422 | vertex.edges.removeAllMatching([](PrefixTreeEdge& edge) |
| 423 | { |
| 424 | return !edge.term; |
| 425 | }); |
| 426 | |
| 427 | if (isLeaf) { |
| 428 | generateInfixUnsuitableForReverseSuffixTree(nfa, stack, actions); |
| 429 | generateSuffixWithReverseSuffixTree(nfa, stack, actions, reverseSuffixTreeRoots); |
| 430 | |
| 431 | // Only stop generating an NFA at a leaf to ensure we have a correct NFA. We could go slightly over the maxNFASize. |
| 432 | if (nfa.nodes.size() > maxNFASize) |
| 433 | nfaTooBig = true; |
| 434 | } else |
| 435 | stack.removeLast(); |
| 436 | |
| 437 | if (!stack.isEmpty()) { |
| 438 | auto& activeSubtree = stack.last(); |
| 439 | auto& edge = activeSubtree.vertex.edges[stack.last().edgeIndex]; |
| 440 | if (edge.child->edges.isEmpty()) |
| 441 | edge.term = nullptr; // Mark this leaf for deleting. |
| 442 | activeSubtree.edgeIndex++; |
| 443 | } |
| 444 | } |
| 445 | } |
| 446 | clearReverseSuffixTree(reverseSuffixTreeRoots); |
| 447 | } |
| 448 | |
| 449 | void CombinedURLFilters::processNFAs(size_t maxNFASize, const WTF::Function<void(NFA&&)>& handler) |
| 450 | { |
| 451 | #if CONTENT_EXTENSIONS_STATE_MACHINE_DEBUGGING |
| 452 | print(); |
| 453 | #endif |
| 454 | while (true) { |
| 455 | // Traverse out to a leaf. |
| 456 | Vector<PrefixTreeVertex*, 128> stack; |
| 457 | PrefixTreeVertex* vertex = m_prefixTreeRoot.get(); |
| 458 | while (true) { |
| 459 | ASSERT(vertex); |
| 460 | stack.append(vertex); |
| 461 | if (vertex->edges.isEmpty()) |
| 462 | break; |
| 463 | vertex = vertex->edges.last().child.get(); |
| 464 | } |
| 465 | if (stack.size() == 1) |
| 466 | break; // We're done once we have processed and removed all the edges in the prefix tree. |
| 467 | |
| 468 | // Find the prefix root for this NFA. This is the vertex after the last term with a quantifier if there is one, |
| 469 | // or the root if there are no quantifiers left. |
| 470 | while (stack.size() > 1) { |
| 471 | if (!stack[stack.size() - 2]->edges.last().term->hasFixedLength()) |
| 472 | break; |
| 473 | stack.removeLast(); |
| 474 | } |
| 475 | ASSERT_WITH_MESSAGE(!stack.isEmpty(), "At least the root should be in the stack" ); |
| 476 | |
| 477 | // Make an NFA with the subtrees for whom this is also the last quantifier (or who also have no quantifier). |
| 478 | NFA nfa; |
| 479 | { |
| 480 | // Put the prefix into the NFA. |
| 481 | ImmutableCharNFANodeBuilder lastNode(nfa); |
| 482 | for (unsigned i = 0; i < stack.size() - 1; ++i) { |
| 483 | const PrefixTreeEdge& edge = stack[i]->edges.last(); |
| 484 | ImmutableCharNFANodeBuilder newNode = edge.term->generateGraph(nfa, lastNode, m_actions.get(edge.child.get())); |
| 485 | lastNode = WTFMove(newNode); |
| 486 | } |
| 487 | |
| 488 | // Put the non-quantified vertices in the subtree into the NFA and delete them. |
| 489 | ASSERT(stack.last()); |
| 490 | generateNFAForSubtree(nfa, WTFMove(lastNode), *stack.last(), m_actions, maxNFASize); |
| 491 | } |
| 492 | nfa.finalize(); |
| 493 | |
| 494 | handler(WTFMove(nfa)); |
| 495 | |
| 496 | // Clean up any processed leaf nodes. |
| 497 | while (true) { |
| 498 | if (stack.size() > 1) { |
| 499 | if (stack[stack.size() - 1]->edges.isEmpty()) { |
| 500 | stack[stack.size() - 2]->edges.removeLast(); |
| 501 | stack.removeLast(); |
| 502 | } else |
| 503 | break; // Vertex is not a leaf. |
| 504 | } else |
| 505 | break; // Leave the empty root. |
| 506 | } |
| 507 | } |
| 508 | } |
| 509 | |
| 510 | } // namespace ContentExtensions |
| 511 | } // namespace WebCore |
| 512 | |
| 513 | #endif // ENABLE(CONTENT_EXTENSIONS) |
| 514 | |