| 1 | /* |
| 2 | * Copyright (C) 2006, 2007, 2014 Apple Inc. All rights reserved. |
| 3 | * Copyright (C) 2008 Nokia Corporation and/or its subsidiary(-ies) |
| 4 | * |
| 5 | * Redistribution and use in source and binary forms, with or without |
| 6 | * modification, are permitted provided that the following conditions |
| 7 | * are met: |
| 8 | * 1. Redistributions of source code must retain the above copyright |
| 9 | * notice, this list of conditions and the following disclaimer. |
| 10 | * 2. Redistributions in binary form must reproduce the above copyright |
| 11 | * notice, this list of conditions and the following disclaimer in the |
| 12 | * documentation and/or other materials provided with the distribution. |
| 13 | * |
| 14 | * THIS SOFTWARE IS PROVIDED BY APPLE INC. ``AS IS'' AND ANY |
| 15 | * EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE |
| 16 | * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR |
| 17 | * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL APPLE INC. OR |
| 18 | * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, |
| 19 | * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, |
| 20 | * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR |
| 21 | * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY |
| 22 | * OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT |
| 23 | * (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE |
| 24 | * OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. |
| 25 | */ |
| 26 | |
| 27 | #include "config.h" |
| 28 | #include "TextCheckingHelper.h" |
| 29 | |
| 30 | #include "Document.h" |
| 31 | #include "DocumentMarkerController.h" |
| 32 | #include "Frame.h" |
| 33 | #include "FrameSelection.h" |
| 34 | #include "Settings.h" |
| 35 | #include "TextCheckerClient.h" |
| 36 | #include "TextIterator.h" |
| 37 | #include "VisiblePosition.h" |
| 38 | #include "VisibleUnits.h" |
| 39 | #include <unicode/ubrk.h> |
| 40 | #include <wtf/text/StringView.h> |
| 41 | #include <wtf/text/TextBreakIterator.h> |
| 42 | |
| 43 | namespace WebCore { |
| 44 | |
| 45 | #if !USE(UNIFIED_TEXT_CHECKING) |
| 46 | |
| 47 | #if USE(GRAMMAR_CHECKING) |
| 48 | |
| 49 | static void findGrammaticalErrors(TextCheckerClient& client, StringView text, Vector<TextCheckingResult>& results) |
| 50 | { |
| 51 | for (unsigned checkLocation = 0; checkLocation < text.length(); ) { |
| 52 | int badGrammarLocation = -1; |
| 53 | int badGrammarLength = 0; |
| 54 | Vector<GrammarDetail> badGrammarDetails; |
| 55 | client.checkGrammarOfString(text.substring(checkLocation), badGrammarDetails, &badGrammarLocation, &badGrammarLength); |
| 56 | if (!badGrammarLength) |
| 57 | break; |
| 58 | |
| 59 | ASSERT(badGrammarLocation >= 0); |
| 60 | ASSERT(static_cast<unsigned>(badGrammarLocation) <= text.length() - checkLocation); |
| 61 | ASSERT(badGrammarLength > 0); |
| 62 | ASSERT(static_cast<unsigned>(badGrammarLength) <= text.length() - checkLocation - badGrammarLocation); |
| 63 | |
| 64 | TextCheckingResult badGrammar; |
| 65 | badGrammar.type = TextCheckingType::Grammar; |
| 66 | badGrammar.location = checkLocation + badGrammarLocation; |
| 67 | badGrammar.length = badGrammarLength; |
| 68 | badGrammar.details = WTFMove(badGrammarDetails); |
| 69 | results.append(badGrammar); |
| 70 | |
| 71 | checkLocation += badGrammarLocation + badGrammarLength; |
| 72 | } |
| 73 | } |
| 74 | |
| 75 | #endif |
| 76 | |
| 77 | static void findMisspellings(TextCheckerClient& client, StringView text, Vector<TextCheckingResult>& results) |
| 78 | { |
| 79 | UBreakIterator* iterator = wordBreakIterator(text); |
| 80 | if (!iterator) |
| 81 | return; |
| 82 | for (int wordStart = ubrk_current(iterator); wordStart >= 0; ) { |
| 83 | int wordEnd = ubrk_next(iterator); |
| 84 | if (wordEnd < 0) |
| 85 | break; |
| 86 | |
| 87 | int wordLength = wordEnd - wordStart; |
| 88 | int misspellingLocation = -1; |
| 89 | int misspellingLength = 0; |
| 90 | client.checkSpellingOfString(text.substring(wordStart, wordLength), &misspellingLocation, &misspellingLength); |
| 91 | |
| 92 | if (misspellingLength > 0) { |
| 93 | ASSERT(misspellingLocation >= 0); |
| 94 | ASSERT(misspellingLocation <= wordLength); |
| 95 | ASSERT(misspellingLength > 0); |
| 96 | ASSERT(misspellingLocation + misspellingLength <= wordLength); |
| 97 | |
| 98 | TextCheckingResult misspelling; |
| 99 | misspelling.type = TextCheckingType::Spelling; |
| 100 | misspelling.location = wordStart + misspellingLocation; |
| 101 | misspelling.length = misspellingLength; |
| 102 | misspelling.replacement = client.getAutoCorrectSuggestionForMisspelledWord(text.substring(misspelling.location, misspelling.length).toStringWithoutCopying()); |
| 103 | results.append(misspelling); |
| 104 | } |
| 105 | |
| 106 | wordStart = wordEnd; |
| 107 | } |
| 108 | } |
| 109 | |
| 110 | #endif |
| 111 | |
| 112 | static Ref<Range> expandToParagraphBoundary(Range& range) |
| 113 | { |
| 114 | Ref<Range> paragraphRange = range.cloneRange(); |
| 115 | setStart(paragraphRange.ptr(), startOfParagraph(range.startPosition())); |
| 116 | setEnd(paragraphRange.ptr(), endOfParagraph(range.endPosition())); |
| 117 | return paragraphRange; |
| 118 | } |
| 119 | |
| 120 | TextCheckingParagraph::TextCheckingParagraph(Ref<Range>&& checkingAndAutomaticReplacementRange) |
| 121 | : m_checkingRange(checkingAndAutomaticReplacementRange.copyRef()) |
| 122 | , m_automaticReplacementRange(checkingAndAutomaticReplacementRange.copyRef()) |
| 123 | { |
| 124 | } |
| 125 | |
| 126 | TextCheckingParagraph::TextCheckingParagraph(Ref<Range>&& checkingRange, Ref<Range>&& automaticReplacementRange, RefPtr<Range>&& paragraphRange) |
| 127 | : m_checkingRange(WTFMove(checkingRange)) |
| 128 | , m_automaticReplacementRange(WTFMove(automaticReplacementRange)) |
| 129 | , m_paragraphRange(WTFMove(paragraphRange)) |
| 130 | { |
| 131 | } |
| 132 | |
| 133 | void TextCheckingParagraph::expandRangeToNextEnd() |
| 134 | { |
| 135 | setEnd(¶graphRange(), endOfParagraph(startOfNextParagraph(paragraphRange().startPosition()))); |
| 136 | invalidateParagraphRangeValues(); |
| 137 | } |
| 138 | |
| 139 | void TextCheckingParagraph::invalidateParagraphRangeValues() |
| 140 | { |
| 141 | m_checkingStart.reset(); |
| 142 | m_checkingEnd.reset(); |
| 143 | m_automaticReplacementStart.reset(); |
| 144 | m_automaticReplacementLength.reset(); |
| 145 | m_offsetAsRange = nullptr; |
| 146 | m_text = String(); |
| 147 | } |
| 148 | |
| 149 | int TextCheckingParagraph::rangeLength() const |
| 150 | { |
| 151 | return TextIterator::rangeLength(¶graphRange()); |
| 152 | } |
| 153 | |
| 154 | Range& TextCheckingParagraph::paragraphRange() const |
| 155 | { |
| 156 | if (!m_paragraphRange) |
| 157 | m_paragraphRange = expandToParagraphBoundary(m_checkingRange); |
| 158 | return *m_paragraphRange; |
| 159 | } |
| 160 | |
| 161 | Ref<Range> TextCheckingParagraph::subrange(int characterOffset, int characterCount) const |
| 162 | { |
| 163 | return TextIterator::subrange(paragraphRange(), characterOffset, characterCount); |
| 164 | } |
| 165 | |
| 166 | ExceptionOr<int> TextCheckingParagraph::offsetTo(const Position& position) const |
| 167 | { |
| 168 | if (!position.containerNode()) |
| 169 | return Exception { TypeError }; |
| 170 | |
| 171 | auto range = offsetAsRange().cloneRange(); |
| 172 | auto result = range->setEnd(*position.containerNode(), position.computeOffsetInContainerNode()); |
| 173 | if (result.hasException()) |
| 174 | return result.releaseException(); |
| 175 | return TextIterator::rangeLength(range.ptr()); |
| 176 | } |
| 177 | |
| 178 | bool TextCheckingParagraph::isEmpty() const |
| 179 | { |
| 180 | // Both predicates should have same result, but we check both just for sure. |
| 181 | // We need to investigate to remove this redundancy. |
| 182 | return checkingStart() >= checkingEnd() || text().isEmpty(); |
| 183 | } |
| 184 | |
| 185 | Range& TextCheckingParagraph::offsetAsRange() const |
| 186 | { |
| 187 | if (!m_offsetAsRange) |
| 188 | m_offsetAsRange = Range::create(paragraphRange().startContainer().document(), paragraphRange().startPosition(), m_checkingRange->startPosition()); |
| 189 | |
| 190 | return *m_offsetAsRange; |
| 191 | } |
| 192 | |
| 193 | const String& TextCheckingParagraph::text() const |
| 194 | { |
| 195 | if (m_text.isEmpty()) |
| 196 | m_text = plainText(¶graphRange()); |
| 197 | return m_text; |
| 198 | } |
| 199 | |
| 200 | int TextCheckingParagraph::checkingStart() const |
| 201 | { |
| 202 | if (!m_checkingStart) |
| 203 | m_checkingStart = TextIterator::rangeLength(&offsetAsRange()); |
| 204 | return *m_checkingStart; |
| 205 | } |
| 206 | |
| 207 | int TextCheckingParagraph::checkingEnd() const |
| 208 | { |
| 209 | if (!m_checkingEnd) |
| 210 | m_checkingEnd = checkingStart() + TextIterator::rangeLength(m_checkingRange.ptr()); |
| 211 | return *m_checkingEnd; |
| 212 | } |
| 213 | |
| 214 | int TextCheckingParagraph::checkingLength() const |
| 215 | { |
| 216 | if (!m_checkingLength) |
| 217 | m_checkingLength = TextIterator::rangeLength(m_checkingRange.ptr()); |
| 218 | return *m_checkingLength; |
| 219 | } |
| 220 | |
| 221 | int TextCheckingParagraph::automaticReplacementStart() const |
| 222 | { |
| 223 | if (m_automaticReplacementStart) |
| 224 | return *m_automaticReplacementStart; |
| 225 | |
| 226 | auto startOffsetRange = Range::create(paragraphRange().startContainer().document(), paragraphRange().startPosition(), m_automaticReplacementRange->startPosition()); |
| 227 | m_automaticReplacementStart = TextIterator::rangeLength(startOffsetRange.ptr()); |
| 228 | return *m_automaticReplacementStart; |
| 229 | } |
| 230 | |
| 231 | int TextCheckingParagraph::automaticReplacementLength() const |
| 232 | { |
| 233 | if (m_automaticReplacementLength) |
| 234 | return *m_automaticReplacementLength; |
| 235 | |
| 236 | auto endOffsetRange = Range::create(paragraphRange().startContainer().document(), paragraphRange().startPosition(), m_automaticReplacementRange->endPosition()); |
| 237 | m_automaticReplacementLength = TextIterator::rangeLength(endOffsetRange.ptr()) - automaticReplacementStart(); |
| 238 | return *m_automaticReplacementLength; |
| 239 | } |
| 240 | |
| 241 | TextCheckingHelper::TextCheckingHelper(EditorClient& client, Range& range) |
| 242 | : m_client(client) |
| 243 | , m_range(range) |
| 244 | { |
| 245 | } |
| 246 | |
| 247 | TextCheckingHelper::~TextCheckingHelper() = default; |
| 248 | |
| 249 | String TextCheckingHelper::findFirstMisspelling(int& firstMisspellingOffset, bool markAll, RefPtr<Range>& firstMisspellingRange) |
| 250 | { |
| 251 | firstMisspellingOffset = 0; |
| 252 | |
| 253 | String firstMisspelling; |
| 254 | int currentChunkOffset = 0; |
| 255 | |
| 256 | for (WordAwareIterator it(m_range); !it.atEnd(); currentChunkOffset += it.text().length(), it.advance()) { |
| 257 | StringView text = it.text(); |
| 258 | int textLength = text.length(); |
| 259 | |
| 260 | // Skip some work for one-space-char hunks. |
| 261 | if (textLength == 1 && text[0] == ' ') |
| 262 | continue; |
| 263 | |
| 264 | int misspellingLocation = -1; |
| 265 | int misspellingLength = 0; |
| 266 | m_client.textChecker()->checkSpellingOfString(text, &misspellingLocation, &misspellingLength); |
| 267 | |
| 268 | // 5490627 shows that there was some code path here where the String constructor below crashes. |
| 269 | // We don't know exactly what combination of bad input caused this, so we're making this much |
| 270 | // more robust against bad input on release builds. |
| 271 | ASSERT(misspellingLength >= 0); |
| 272 | ASSERT(misspellingLocation >= -1); |
| 273 | ASSERT(!misspellingLength || misspellingLocation >= 0); |
| 274 | ASSERT(misspellingLocation < textLength); |
| 275 | ASSERT(misspellingLength <= textLength); |
| 276 | ASSERT(misspellingLocation + misspellingLength <= textLength); |
| 277 | |
| 278 | if (misspellingLocation >= 0 && misspellingLength > 0 && misspellingLocation < textLength && misspellingLength <= textLength && misspellingLocation + misspellingLength <= textLength) { |
| 279 | // Compute range of misspelled word |
| 280 | auto misspellingRange = TextIterator::subrange(m_range, currentChunkOffset + misspellingLocation, misspellingLength); |
| 281 | |
| 282 | // Remember first-encountered misspelling and its offset. |
| 283 | if (!firstMisspelling) { |
| 284 | firstMisspellingOffset = currentChunkOffset + misspellingLocation; |
| 285 | firstMisspelling = text.substring(misspellingLocation, misspellingLength).toString(); |
| 286 | firstMisspellingRange = misspellingRange.ptr(); |
| 287 | } |
| 288 | |
| 289 | // Store marker for misspelled word. |
| 290 | misspellingRange->startContainer().document().markers().addMarker(misspellingRange, DocumentMarker::Spelling); |
| 291 | |
| 292 | // Bail out if we're marking only the first misspelling, and not all instances. |
| 293 | if (!markAll) |
| 294 | break; |
| 295 | } |
| 296 | } |
| 297 | |
| 298 | return firstMisspelling; |
| 299 | } |
| 300 | |
| 301 | String TextCheckingHelper::findFirstMisspellingOrBadGrammar(bool checkGrammar, bool& outIsSpelling, int& outFirstFoundOffset, GrammarDetail& outGrammarDetail) |
| 302 | { |
| 303 | if (!unifiedTextCheckerEnabled()) |
| 304 | return emptyString(); |
| 305 | |
| 306 | if (platformDrivenTextCheckerEnabled()) |
| 307 | return emptyString(); |
| 308 | |
| 309 | String firstFoundItem; |
| 310 | String misspelledWord; |
| 311 | String badGrammarPhrase; |
| 312 | |
| 313 | // Initialize out parameters; these will be updated if we find something to return. |
| 314 | outIsSpelling = true; |
| 315 | outFirstFoundOffset = 0; |
| 316 | outGrammarDetail.location = -1; |
| 317 | outGrammarDetail.length = 0; |
| 318 | outGrammarDetail.guesses.clear(); |
| 319 | outGrammarDetail.userDescription = emptyString(); |
| 320 | |
| 321 | // Expand the search range to encompass entire paragraphs, since text checking needs that much context. |
| 322 | // Determine the character offset from the start of the paragraph to the start of the original search range, |
| 323 | // since we will want to ignore results in this area. |
| 324 | Ref<Range> paragraphRange = m_range->cloneRange(); |
| 325 | setStart(paragraphRange.ptr(), startOfParagraph(m_range->startPosition())); |
| 326 | int totalRangeLength = TextIterator::rangeLength(paragraphRange.ptr()); |
| 327 | setEnd(paragraphRange.ptr(), endOfParagraph(m_range->startPosition())); |
| 328 | |
| 329 | Ref<Range> offsetAsRange = Range::create(paragraphRange->startContainer().document(), paragraphRange->startPosition(), m_range->startPosition()); |
| 330 | int rangeStartOffset = TextIterator::rangeLength(offsetAsRange.ptr()); |
| 331 | int totalLengthProcessed = 0; |
| 332 | |
| 333 | bool firstIteration = true; |
| 334 | bool lastIteration = false; |
| 335 | while (totalLengthProcessed < totalRangeLength) { |
| 336 | // Iterate through the search range by paragraphs, checking each one for spelling and grammar. |
| 337 | int currentLength = TextIterator::rangeLength(paragraphRange.ptr()); |
| 338 | int currentStartOffset = firstIteration ? rangeStartOffset : 0; |
| 339 | int currentEndOffset = currentLength; |
| 340 | if (inSameParagraph(paragraphRange->startPosition(), m_range->endPosition())) { |
| 341 | // Determine the character offset from the end of the original search range to the end of the paragraph, |
| 342 | // since we will want to ignore results in this area. |
| 343 | auto endOffsetAsRange = Range::create(paragraphRange->startContainer().document(), paragraphRange->startPosition(), m_range->endPosition()); |
| 344 | currentEndOffset = TextIterator::rangeLength(endOffsetAsRange.ptr()); |
| 345 | lastIteration = true; |
| 346 | } |
| 347 | if (currentStartOffset < currentEndOffset) { |
| 348 | String paragraphString = plainText(paragraphRange.ptr()); |
| 349 | if (paragraphString.length() > 0) { |
| 350 | bool foundGrammar = false; |
| 351 | int spellingLocation = 0; |
| 352 | int grammarPhraseLocation = 0; |
| 353 | int grammarDetailLocation = 0; |
| 354 | unsigned grammarDetailIndex = 0; |
| 355 | |
| 356 | Vector<TextCheckingResult> results; |
| 357 | OptionSet<TextCheckingType> checkingTypes { TextCheckingType::Spelling }; |
| 358 | if (checkGrammar) |
| 359 | checkingTypes.add(TextCheckingType::Grammar); |
| 360 | VisibleSelection currentSelection; |
| 361 | if (Frame* frame = paragraphRange->ownerDocument().frame()) |
| 362 | currentSelection = frame->selection().selection(); |
| 363 | checkTextOfParagraph(*m_client.textChecker(), paragraphString, checkingTypes, results, currentSelection); |
| 364 | |
| 365 | for (auto& result : results) { |
| 366 | if (result.type == TextCheckingType::Spelling && result.location >= currentStartOffset && result.location + result.length <= currentEndOffset) { |
| 367 | ASSERT(result.length > 0); |
| 368 | ASSERT(result.location >= 0); |
| 369 | spellingLocation = result.location; |
| 370 | misspelledWord = paragraphString.substring(result.location, result.length); |
| 371 | ASSERT(misspelledWord.length()); |
| 372 | break; |
| 373 | } |
| 374 | if (checkGrammar && result.type == TextCheckingType::Grammar && result.location < currentEndOffset && result.location + result.length > currentStartOffset) { |
| 375 | ASSERT(result.length > 0); |
| 376 | ASSERT(result.location >= 0); |
| 377 | // We can't stop after the first grammar result, since there might still be a spelling result after |
| 378 | // it begins but before the first detail in it, but we can stop if we find a second grammar result. |
| 379 | if (foundGrammar) |
| 380 | break; |
| 381 | for (unsigned j = 0; j < result.details.size(); j++) { |
| 382 | const GrammarDetail* detail = &result.details[j]; |
| 383 | ASSERT(detail->length > 0); |
| 384 | ASSERT(detail->location >= 0); |
| 385 | if (result.location + detail->location >= currentStartOffset && result.location + detail->location + detail->length <= currentEndOffset && (!foundGrammar || result.location + detail->location < grammarDetailLocation)) { |
| 386 | grammarDetailIndex = j; |
| 387 | grammarDetailLocation = result.location + detail->location; |
| 388 | foundGrammar = true; |
| 389 | } |
| 390 | } |
| 391 | if (foundGrammar) { |
| 392 | grammarPhraseLocation = result.location; |
| 393 | outGrammarDetail = result.details[grammarDetailIndex]; |
| 394 | badGrammarPhrase = paragraphString.substring(result.location, result.length); |
| 395 | ASSERT(badGrammarPhrase.length()); |
| 396 | } |
| 397 | } |
| 398 | } |
| 399 | |
| 400 | if (!misspelledWord.isEmpty() && (!checkGrammar || badGrammarPhrase.isEmpty() || spellingLocation <= grammarDetailLocation)) { |
| 401 | int spellingOffset = spellingLocation - currentStartOffset; |
| 402 | if (!firstIteration) { |
| 403 | auto paragraphOffsetAsRange = Range::create(paragraphRange->startContainer().document(), m_range->startPosition(), paragraphRange->startPosition()); |
| 404 | spellingOffset += TextIterator::rangeLength(paragraphOffsetAsRange.ptr()); |
| 405 | } |
| 406 | outIsSpelling = true; |
| 407 | outFirstFoundOffset = spellingOffset; |
| 408 | firstFoundItem = misspelledWord; |
| 409 | break; |
| 410 | } |
| 411 | if (checkGrammar && !badGrammarPhrase.isEmpty()) { |
| 412 | int grammarPhraseOffset = grammarPhraseLocation - currentStartOffset; |
| 413 | if (!firstIteration) { |
| 414 | auto paragraphOffsetAsRange = Range::create(paragraphRange->startContainer().document(), m_range->startPosition(), paragraphRange->startPosition()); |
| 415 | grammarPhraseOffset += TextIterator::rangeLength(paragraphOffsetAsRange.ptr()); |
| 416 | } |
| 417 | outIsSpelling = false; |
| 418 | outFirstFoundOffset = grammarPhraseOffset; |
| 419 | firstFoundItem = badGrammarPhrase; |
| 420 | break; |
| 421 | } |
| 422 | } |
| 423 | } |
| 424 | if (lastIteration || totalLengthProcessed + currentLength >= totalRangeLength) |
| 425 | break; |
| 426 | VisiblePosition newParagraphStart = startOfNextParagraph(paragraphRange->endPosition()); |
| 427 | setStart(paragraphRange.ptr(), newParagraphStart); |
| 428 | setEnd(paragraphRange.ptr(), endOfParagraph(newParagraphStart)); |
| 429 | firstIteration = false; |
| 430 | totalLengthProcessed += currentLength; |
| 431 | } |
| 432 | return firstFoundItem; |
| 433 | } |
| 434 | |
| 435 | #if USE(GRAMMAR_CHECKING) |
| 436 | |
| 437 | int TextCheckingHelper::findFirstGrammarDetail(const Vector<GrammarDetail>& grammarDetails, int badGrammarPhraseLocation, int startOffset, int endOffset, bool markAll) const |
| 438 | { |
| 439 | // Found some bad grammar. Find the earliest detail range that starts in our search range (if any). |
| 440 | // Optionally add a DocumentMarker for each detail in the range. |
| 441 | int earliestDetailLocationSoFar = -1; |
| 442 | int earliestDetailIndex = -1; |
| 443 | for (unsigned i = 0; i < grammarDetails.size(); i++) { |
| 444 | const GrammarDetail* detail = &grammarDetails[i]; |
| 445 | ASSERT(detail->length > 0); |
| 446 | ASSERT(detail->location >= 0); |
| 447 | |
| 448 | int detailStartOffsetInParagraph = badGrammarPhraseLocation + detail->location; |
| 449 | |
| 450 | // Skip this detail if it starts before the original search range |
| 451 | if (detailStartOffsetInParagraph < startOffset) |
| 452 | continue; |
| 453 | |
| 454 | // Skip this detail if it starts after the original search range |
| 455 | if (detailStartOffsetInParagraph >= endOffset) |
| 456 | continue; |
| 457 | |
| 458 | if (markAll) { |
| 459 | auto badGrammarRange = TextIterator::subrange(m_range, badGrammarPhraseLocation - startOffset + detail->location, detail->length); |
| 460 | badGrammarRange->startContainer().document().markers().addMarker(badGrammarRange, DocumentMarker::Grammar, detail->userDescription); |
| 461 | } |
| 462 | |
| 463 | // Remember this detail only if it's earlier than our current candidate (the details aren't in a guaranteed order) |
| 464 | if (earliestDetailIndex < 0 || earliestDetailLocationSoFar > detail->location) { |
| 465 | earliestDetailIndex = i; |
| 466 | earliestDetailLocationSoFar = detail->location; |
| 467 | } |
| 468 | } |
| 469 | |
| 470 | return earliestDetailIndex; |
| 471 | } |
| 472 | |
| 473 | String TextCheckingHelper::findFirstBadGrammar(GrammarDetail& outGrammarDetail, int& outGrammarPhraseOffset, bool markAll) const |
| 474 | { |
| 475 | // Initialize out parameters; these will be updated if we find something to return. |
| 476 | outGrammarDetail.location = -1; |
| 477 | outGrammarDetail.length = 0; |
| 478 | outGrammarDetail.guesses.clear(); |
| 479 | outGrammarDetail.userDescription = emptyString(); |
| 480 | outGrammarPhraseOffset = 0; |
| 481 | |
| 482 | String firstBadGrammarPhrase; |
| 483 | |
| 484 | // Expand the search range to encompass entire paragraphs, since grammar checking needs that much context. |
| 485 | // Determine the character offset from the start of the paragraph to the start of the original search range, |
| 486 | // since we will want to ignore results in this area. |
| 487 | TextCheckingParagraph paragraph(m_range.copyRef()); |
| 488 | |
| 489 | // Start checking from beginning of paragraph, but skip past results that occur before the start of the original search range. |
| 490 | for (int startOffset = 0; startOffset < paragraph.checkingEnd(); ) { |
| 491 | Vector<GrammarDetail> grammarDetails; |
| 492 | int badGrammarPhraseLocation = -1; |
| 493 | int badGrammarPhraseLength = 0; |
| 494 | m_client.textChecker()->checkGrammarOfString(StringView(paragraph.text()).substring(startOffset), grammarDetails, &badGrammarPhraseLocation, &badGrammarPhraseLength); |
| 495 | |
| 496 | if (!badGrammarPhraseLength) { |
| 497 | ASSERT(badGrammarPhraseLocation == -1); |
| 498 | return String(); |
| 499 | } |
| 500 | |
| 501 | ASSERT(badGrammarPhraseLocation >= 0); |
| 502 | badGrammarPhraseLocation += startOffset; |
| 503 | |
| 504 | // Found some bad grammar. Find the earliest detail range that starts in our search range (if any). |
| 505 | int badGrammarIndex = findFirstGrammarDetail(grammarDetails, badGrammarPhraseLocation, paragraph.checkingStart(), paragraph.checkingEnd(), markAll); |
| 506 | if (badGrammarIndex >= 0) { |
| 507 | ASSERT(static_cast<unsigned>(badGrammarIndex) < grammarDetails.size()); |
| 508 | outGrammarDetail = grammarDetails[badGrammarIndex]; |
| 509 | } |
| 510 | |
| 511 | // If we found a detail in range, then we have found the first bad phrase (unless we found one earlier but |
| 512 | // kept going so we could mark all instances). |
| 513 | if (badGrammarIndex >= 0 && firstBadGrammarPhrase.isEmpty()) { |
| 514 | outGrammarPhraseOffset = badGrammarPhraseLocation - paragraph.checkingStart(); |
| 515 | firstBadGrammarPhrase = paragraph.textSubstring(badGrammarPhraseLocation, badGrammarPhraseLength); |
| 516 | |
| 517 | // Found one. We're done now, unless we're marking each instance. |
| 518 | if (!markAll) |
| 519 | break; |
| 520 | } |
| 521 | |
| 522 | // These results were all between the start of the paragraph and the start of the search range; look |
| 523 | // beyond this phrase. |
| 524 | startOffset = badGrammarPhraseLocation + badGrammarPhraseLength; |
| 525 | } |
| 526 | |
| 527 | return firstBadGrammarPhrase; |
| 528 | } |
| 529 | |
| 530 | bool TextCheckingHelper::isUngrammatical() const |
| 531 | { |
| 532 | if (m_range->collapsed()) |
| 533 | return false; |
| 534 | |
| 535 | // Returns true only if the passed range exactly corresponds to a bad grammar detail range. This is analogous |
| 536 | // to isSelectionMisspelled. It's not good enough for there to be some bad grammar somewhere in the range, |
| 537 | // or overlapping the range; the ranges must exactly match. |
| 538 | int grammarPhraseOffset; |
| 539 | |
| 540 | GrammarDetail grammarDetail; |
| 541 | String badGrammarPhrase = findFirstBadGrammar(grammarDetail, grammarPhraseOffset, false); |
| 542 | |
| 543 | // No bad grammar in these parts at all. |
| 544 | if (badGrammarPhrase.isEmpty()) |
| 545 | return false; |
| 546 | |
| 547 | // Bad grammar, but phrase (e.g. sentence) starts beyond start of range. |
| 548 | if (grammarPhraseOffset > 0) |
| 549 | return false; |
| 550 | |
| 551 | ASSERT(grammarDetail.location >= 0); |
| 552 | ASSERT(grammarDetail.length > 0); |
| 553 | |
| 554 | // Bad grammar, but start of detail (e.g. ungrammatical word) doesn't match start of range |
| 555 | if (grammarDetail.location + grammarPhraseOffset) |
| 556 | return false; |
| 557 | |
| 558 | // Bad grammar at start of range, but end of bad grammar is before or after end of range |
| 559 | if (grammarDetail.length != TextIterator::rangeLength(m_range.ptr())) |
| 560 | return false; |
| 561 | |
| 562 | // Update the spelling panel to be displaying this error (whether or not the spelling panel is on screen). |
| 563 | // This is necessary to make a subsequent call to [NSSpellChecker ignoreWord:inSpellDocumentWithTag:] work |
| 564 | // correctly; that call behaves differently based on whether the spelling panel is displaying a misspelling |
| 565 | // or a grammar error. |
| 566 | m_client.updateSpellingUIWithGrammarString(badGrammarPhrase, grammarDetail); |
| 567 | |
| 568 | return true; |
| 569 | } |
| 570 | |
| 571 | #endif // USE(GRAMMAR_CHECKING) |
| 572 | |
| 573 | Vector<String> TextCheckingHelper::guessesForMisspelledOrUngrammaticalRange(bool checkGrammar, bool& misspelled, bool& ungrammatical) const |
| 574 | { |
| 575 | if (!unifiedTextCheckerEnabled()) |
| 576 | return Vector<String>(); |
| 577 | |
| 578 | if (platformDrivenTextCheckerEnabled()) |
| 579 | return Vector<String>(); |
| 580 | |
| 581 | Vector<String> guesses; |
| 582 | misspelled = false; |
| 583 | ungrammatical = false; |
| 584 | |
| 585 | if (m_range->collapsed()) |
| 586 | return guesses; |
| 587 | |
| 588 | // Expand the range to encompass entire paragraphs, since text checking needs that much context. |
| 589 | TextCheckingParagraph paragraph(m_range.copyRef()); |
| 590 | if (paragraph.isEmpty()) |
| 591 | return guesses; |
| 592 | |
| 593 | Vector<TextCheckingResult> results; |
| 594 | OptionSet<TextCheckingType> checkingTypes { TextCheckingType::Spelling }; |
| 595 | if (checkGrammar) |
| 596 | checkingTypes.add(TextCheckingType::Grammar); |
| 597 | VisibleSelection currentSelection; |
| 598 | if (Frame* frame = m_range->ownerDocument().frame()) |
| 599 | currentSelection = frame->selection().selection(); |
| 600 | checkTextOfParagraph(*m_client.textChecker(), paragraph.text(), checkingTypes, results, currentSelection); |
| 601 | |
| 602 | for (auto& result : results) { |
| 603 | if (result.type == TextCheckingType::Spelling && paragraph.checkingRangeMatches(result.location, result.length)) { |
| 604 | String misspelledWord = paragraph.checkingSubstring(); |
| 605 | ASSERT(misspelledWord.length()); |
| 606 | m_client.textChecker()->getGuessesForWord(misspelledWord, String(), currentSelection, guesses); |
| 607 | m_client.updateSpellingUIWithMisspelledWord(misspelledWord); |
| 608 | misspelled = true; |
| 609 | return guesses; |
| 610 | } |
| 611 | } |
| 612 | |
| 613 | if (!checkGrammar) |
| 614 | return guesses; |
| 615 | |
| 616 | for (auto& result : results) { |
| 617 | if (result.type == TextCheckingType::Grammar && paragraph.isCheckingRangeCoveredBy(result.location, result.length)) { |
| 618 | for (auto& detail : result.details) { |
| 619 | ASSERT(detail.length > 0); |
| 620 | ASSERT(detail.location >= 0); |
| 621 | if (paragraph.checkingRangeMatches(result.location + detail.location, detail.length)) { |
| 622 | String badGrammarPhrase = paragraph.textSubstring(result.location, result.length); |
| 623 | ASSERT(badGrammarPhrase.length()); |
| 624 | for (auto& guess : detail.guesses) |
| 625 | guesses.append(guess); |
| 626 | m_client.updateSpellingUIWithGrammarString(badGrammarPhrase, detail); |
| 627 | ungrammatical = true; |
| 628 | return guesses; |
| 629 | } |
| 630 | } |
| 631 | } |
| 632 | } |
| 633 | return guesses; |
| 634 | } |
| 635 | |
| 636 | void TextCheckingHelper::markAllMisspellings(RefPtr<Range>& firstMisspellingRange) |
| 637 | { |
| 638 | // Use the "markAll" feature of findFirstMisspelling. Ignore the return value and the "out parameter"; |
| 639 | // all we need to do is mark every instance. |
| 640 | int ignoredOffset; |
| 641 | findFirstMisspelling(ignoredOffset, true, firstMisspellingRange); |
| 642 | } |
| 643 | |
| 644 | #if USE(GRAMMAR_CHECKING) |
| 645 | void TextCheckingHelper::markAllBadGrammar() |
| 646 | { |
| 647 | // Use the "markAll" feature of ofindFirstBadGrammar. Ignore the return value and "out parameters"; all we need to |
| 648 | // do is mark every instance. |
| 649 | GrammarDetail ignoredGrammarDetail; |
| 650 | int ignoredOffset; |
| 651 | findFirstBadGrammar(ignoredGrammarDetail, ignoredOffset, true); |
| 652 | } |
| 653 | #endif |
| 654 | |
| 655 | bool TextCheckingHelper::unifiedTextCheckerEnabled() const |
| 656 | { |
| 657 | return WebCore::unifiedTextCheckerEnabled(m_range->ownerDocument().frame()); |
| 658 | } |
| 659 | |
| 660 | void checkTextOfParagraph(TextCheckerClient& client, StringView text, OptionSet<TextCheckingType> checkingTypes, Vector<TextCheckingResult>& results, const VisibleSelection& currentSelection) |
| 661 | { |
| 662 | #if USE(UNIFIED_TEXT_CHECKING) |
| 663 | results = client.checkTextOfParagraph(text, checkingTypes, currentSelection); |
| 664 | #else |
| 665 | UNUSED_PARAM(currentSelection); |
| 666 | |
| 667 | Vector<TextCheckingResult> mispellings; |
| 668 | if (checkingTypes.contains(TextCheckingType::Spelling)) |
| 669 | findMisspellings(client, text, mispellings); |
| 670 | |
| 671 | #if USE(GRAMMAR_CHECKING) |
| 672 | // Look for grammatical errors that occur before the first misspelling. |
| 673 | Vector<TextCheckingResult> grammaticalErrors; |
| 674 | if (checkingTypes.contains(TextCheckingType::Grammar)) { |
| 675 | unsigned grammarCheckLength = text.length(); |
| 676 | for (auto& mispelling : mispellings) |
| 677 | grammarCheckLength = std::min<unsigned>(grammarCheckLength, mispelling.location); |
| 678 | findGrammaticalErrors(client, text.substring(0, grammarCheckLength), grammaticalErrors); |
| 679 | } |
| 680 | |
| 681 | results = WTFMove(grammaticalErrors); |
| 682 | #endif |
| 683 | |
| 684 | if (results.isEmpty()) |
| 685 | results = WTFMove(mispellings); |
| 686 | else |
| 687 | results.appendVector(mispellings); |
| 688 | #endif // USE(UNIFIED_TEXT_CHECKING) |
| 689 | } |
| 690 | |
| 691 | bool unifiedTextCheckerEnabled(const Frame* frame) |
| 692 | { |
| 693 | if (!frame) |
| 694 | return false; |
| 695 | return frame->settings().unifiedTextCheckerEnabled(); |
| 696 | } |
| 697 | |
| 698 | bool platformDrivenTextCheckerEnabled() |
| 699 | { |
| 700 | #if ENABLE(PLATFORM_DRIVEN_TEXT_CHECKING) |
| 701 | return true; |
| 702 | #else |
| 703 | return false; |
| 704 | #endif |
| 705 | } |
| 706 | |
| 707 | } |
| 708 | |