| 1 | // Copyright (C) 2019 The Qt Company Ltd. |
| 2 | // SPDX-License-Identifier: LicenseRef-Qt-Commercial OR LGPL-3.0-only OR GPL-2.0-only OR GPL-3.0-only |
| 3 | |
| 4 | #include "qtextmarkdownwriter_p.h" |
| 5 | #include "qtextdocumentlayout_p.h" |
| 6 | #include "qfontinfo.h" |
| 7 | #include "qfontmetrics.h" |
| 8 | #include "qtextdocument_p.h" |
| 9 | #include "qtextlist.h" |
| 10 | #include "qtexttable.h" |
| 11 | #include "qtextcursor.h" |
| 12 | #include "qtextimagehandler_p.h" |
| 13 | #include "qtextmarkdownimporter_p.h" |
| 14 | #include "qloggingcategory.h" |
| 15 | #include <QtCore/QRegularExpression> |
| 16 | #if QT_CONFIG(itemmodel) |
| 17 | #include "qabstractitemmodel.h" |
| 18 | #endif |
| 19 | |
| 20 | QT_BEGIN_NAMESPACE |
| 21 | |
| 22 | using namespace Qt::StringLiterals; |
| 23 | |
| 24 | Q_LOGGING_CATEGORY(lcMDW, "qt.text.markdown.writer" ) |
| 25 | |
| 26 | static const QChar qtmw_Space = u' '; |
| 27 | static const QChar qtmw_Tab = u'\t'; |
| 28 | static const QChar qtmw_Newline = u'\n'; |
| 29 | static const QChar qtmw_CarriageReturn = u'\r'; |
| 30 | static const QChar qtmw_LineBreak = u'\x2028'; |
| 31 | static const QChar qtmw_DoubleQuote = u'"'; |
| 32 | static const QChar qtmw_Backtick = u'`'; |
| 33 | static const QChar qtmw_Backslash = u'\\'; |
| 34 | static const QChar qtmw_Period = u'.'; |
| 35 | |
| 36 | QTextMarkdownWriter::QTextMarkdownWriter(QTextStream &stream, QTextDocument::MarkdownFeatures features) |
| 37 | : m_stream(stream), m_features(features) |
| 38 | { |
| 39 | } |
| 40 | |
| 41 | bool QTextMarkdownWriter::writeAll(const QTextDocument *document) |
| 42 | { |
| 43 | writeFrontMatter(fm: document->metaInformation(info: QTextDocument::FrontMatter)); |
| 44 | writeFrame(frame: document->rootFrame()); |
| 45 | return true; |
| 46 | } |
| 47 | |
| 48 | #if QT_CONFIG(itemmodel) |
| 49 | void QTextMarkdownWriter::writeTable(const QAbstractItemModel *table) |
| 50 | { |
| 51 | QList<int> tableColumnWidths(table->columnCount()); |
| 52 | for (int col = 0; col < table->columnCount(); ++col) { |
| 53 | tableColumnWidths[col] = table->headerData(section: col, orientation: Qt::Horizontal).toString().size(); |
| 54 | for (int row = 0; row < table->rowCount(); ++row) { |
| 55 | tableColumnWidths[col] = qMax(a: tableColumnWidths[col], |
| 56 | b: table->data(index: table->index(row, column: col)).toString().size()); |
| 57 | } |
| 58 | } |
| 59 | |
| 60 | // write the header and separator |
| 61 | for (int col = 0; col < table->columnCount(); ++col) { |
| 62 | QString s = table->headerData(section: col, orientation: Qt::Horizontal).toString(); |
| 63 | m_stream << '|' << s << QString(tableColumnWidths[col] - s.size(), qtmw_Space); |
| 64 | } |
| 65 | m_stream << "|" << Qt::endl; |
| 66 | for (int col = 0; col < tableColumnWidths.size(); ++col) |
| 67 | m_stream << '|' << QString(tableColumnWidths[col], u'-'); |
| 68 | m_stream << '|'<< Qt::endl; |
| 69 | |
| 70 | // write the body |
| 71 | for (int row = 0; row < table->rowCount(); ++row) { |
| 72 | for (int col = 0; col < table->columnCount(); ++col) { |
| 73 | QString s = table->data(index: table->index(row, column: col)).toString(); |
| 74 | m_stream << '|' << s << QString(tableColumnWidths[col] - s.size(), qtmw_Space); |
| 75 | } |
| 76 | m_stream << '|'<< Qt::endl; |
| 77 | } |
| 78 | m_listInfo.clear(); |
| 79 | } |
| 80 | #endif |
| 81 | |
| 82 | void QTextMarkdownWriter::writeFrontMatter(const QString &fm) |
| 83 | { |
| 84 | const bool featureEnabled = m_features.testFlag( |
| 85 | flag: static_cast<QTextDocument::MarkdownFeature>(QTextMarkdownImporter::FeatureFrontMatter)); |
| 86 | qCDebug(lcMDW) << "writing FrontMatter?" << featureEnabled << "size" << fm.size(); |
| 87 | if (fm.isEmpty() || !featureEnabled) |
| 88 | return; |
| 89 | m_stream << "---\n"_L1 << fm; |
| 90 | if (!fm.endsWith(c: qtmw_Newline)) |
| 91 | m_stream << qtmw_Newline; |
| 92 | m_stream << "---\n"_L1 ; |
| 93 | } |
| 94 | |
| 95 | void QTextMarkdownWriter::writeFrame(const QTextFrame *frame) |
| 96 | { |
| 97 | Q_ASSERT(frame); |
| 98 | const QTextTable *table = qobject_cast<const QTextTable*> (object: frame); |
| 99 | QTextFrame::iterator iterator = frame->begin(); |
| 100 | QTextFrame *child = nullptr; |
| 101 | int tableRow = -1; |
| 102 | bool lastWasList = false; |
| 103 | QList<int> tableColumnWidths; |
| 104 | if (table) { |
| 105 | tableColumnWidths.resize(size: table->columns()); |
| 106 | for (int col = 0; col < table->columns(); ++col) { |
| 107 | for (int row = 0; row < table->rows(); ++ row) { |
| 108 | QTextTableCell cell = table->cellAt(row, col); |
| 109 | int cellTextLen = 0; |
| 110 | auto it = cell.begin(); |
| 111 | while (it != cell.end()) { |
| 112 | QTextBlock block = it.currentBlock(); |
| 113 | if (block.isValid()) |
| 114 | cellTextLen += block.text().size(); |
| 115 | ++it; |
| 116 | } |
| 117 | if (cell.columnSpan() == 1 && tableColumnWidths[col] < cellTextLen) |
| 118 | tableColumnWidths[col] = cellTextLen; |
| 119 | } |
| 120 | } |
| 121 | } |
| 122 | while (!iterator.atEnd()) { |
| 123 | if (iterator.currentFrame() && child != iterator.currentFrame()) |
| 124 | writeFrame(frame: iterator.currentFrame()); |
| 125 | else { // no frame, it's a block |
| 126 | QTextBlock block = iterator.currentBlock(); |
| 127 | // Look ahead and detect some cases when we should |
| 128 | // suppress needless blank lines, when there will be a big change in block format |
| 129 | bool nextIsDifferent = false; |
| 130 | bool ending = false; |
| 131 | int blockQuoteIndent = 0; |
| 132 | int nextBlockQuoteIndent = 0; |
| 133 | { |
| 134 | QTextFrame::iterator next = iterator; |
| 135 | ++next; |
| 136 | QTextBlockFormat format = iterator.currentBlock().blockFormat(); |
| 137 | QTextBlockFormat nextFormat = next.currentBlock().blockFormat(); |
| 138 | blockQuoteIndent = format.intProperty(propertyId: QTextFormat::BlockQuoteLevel); |
| 139 | nextBlockQuoteIndent = nextFormat.intProperty(propertyId: QTextFormat::BlockQuoteLevel); |
| 140 | if (next.atEnd()) { |
| 141 | nextIsDifferent = true; |
| 142 | ending = true; |
| 143 | } else { |
| 144 | if (nextFormat.indent() != format.indent() || |
| 145 | nextFormat.property(propertyId: QTextFormat::BlockCodeLanguage) != |
| 146 | format.property(propertyId: QTextFormat::BlockCodeLanguage)) |
| 147 | nextIsDifferent = true; |
| 148 | } |
| 149 | } |
| 150 | if (table) { |
| 151 | QTextTableCell cell = table->cellAt(position: block.position()); |
| 152 | if (tableRow < cell.row()) { |
| 153 | if (tableRow == 0) { |
| 154 | m_stream << qtmw_Newline; |
| 155 | for (int col = 0; col < tableColumnWidths.size(); ++col) |
| 156 | m_stream << '|' << QString(tableColumnWidths[col], u'-'); |
| 157 | m_stream << '|'; |
| 158 | } |
| 159 | m_stream << qtmw_Newline << '|'; |
| 160 | tableRow = cell.row(); |
| 161 | } |
| 162 | } else if (!block.textList()) { |
| 163 | if (lastWasList) { |
| 164 | m_stream << qtmw_Newline; |
| 165 | m_linePrefixWritten = false; |
| 166 | } |
| 167 | } |
| 168 | int endingCol = writeBlock(block, table: !table, ignoreFormat: table && tableRow == 0, |
| 169 | ignoreEmpty: nextIsDifferent && !block.textList()); |
| 170 | m_doubleNewlineWritten = false; |
| 171 | if (table) { |
| 172 | QTextTableCell cell = table->cellAt(position: block.position()); |
| 173 | int paddingLen = -endingCol; |
| 174 | int spanEndCol = cell.column() + cell.columnSpan(); |
| 175 | for (int col = cell.column(); col < spanEndCol; ++col) |
| 176 | paddingLen += tableColumnWidths[col]; |
| 177 | if (paddingLen > 0) |
| 178 | m_stream << QString(paddingLen, qtmw_Space); |
| 179 | for (int col = cell.column(); col < spanEndCol; ++col) |
| 180 | m_stream << "|" ; |
| 181 | } else if (m_fencedCodeBlock && ending) { |
| 182 | m_stream << qtmw_Newline << m_linePrefix << QString(m_wrappedLineIndent, qtmw_Space) |
| 183 | << m_codeBlockFence << qtmw_Newline << qtmw_Newline; |
| 184 | m_codeBlockFence.clear(); |
| 185 | } else if (m_indentedCodeBlock && nextIsDifferent) { |
| 186 | m_stream << qtmw_Newline << qtmw_Newline; |
| 187 | } else if (endingCol > 0) { |
| 188 | if (block.textList() || block.blockFormat().hasProperty(propertyId: QTextFormat::BlockCodeLanguage)) { |
| 189 | m_stream << qtmw_Newline; |
| 190 | if (block.textList()) { |
| 191 | m_stream << m_linePrefix; |
| 192 | m_linePrefixWritten = true; |
| 193 | } |
| 194 | } else { |
| 195 | m_stream << qtmw_Newline; |
| 196 | if (nextBlockQuoteIndent < blockQuoteIndent) |
| 197 | setLinePrefixForBlockQuote(nextBlockQuoteIndent); |
| 198 | m_stream << m_linePrefix; |
| 199 | m_stream << qtmw_Newline; |
| 200 | m_doubleNewlineWritten = true; |
| 201 | } |
| 202 | } |
| 203 | lastWasList = block.textList(); |
| 204 | } |
| 205 | child = iterator.currentFrame(); |
| 206 | ++iterator; |
| 207 | } |
| 208 | if (table) { |
| 209 | m_stream << qtmw_Newline << qtmw_Newline; |
| 210 | m_doubleNewlineWritten = true; |
| 211 | } |
| 212 | m_listInfo.clear(); |
| 213 | } |
| 214 | |
| 215 | QTextMarkdownWriter::ListInfo QTextMarkdownWriter::listInfo(QTextList *list) |
| 216 | { |
| 217 | if (!m_listInfo.contains(key: list)) { |
| 218 | // decide whether this list is loose or tight |
| 219 | ListInfo info; |
| 220 | info.loose = false; |
| 221 | if (list->count() > 1) { |
| 222 | QTextBlock first = list->item(i: 0); |
| 223 | QTextBlock last = list->item(i: list->count() - 1); |
| 224 | QTextBlock next = first.next(); |
| 225 | while (next.isValid()) { |
| 226 | if (next == last) |
| 227 | break; |
| 228 | qCDebug(lcMDW) << "next block in list" << list << next.text() << "part of list?" << next.textList(); |
| 229 | if (!next.textList()) { |
| 230 | // If we find a continuation paragraph, this list is "loose" |
| 231 | // because it will need a blank line to separate that paragraph. |
| 232 | qCDebug(lcMDW) << "decided list beginning with" << first.text() << "is loose after" << next.text(); |
| 233 | info.loose = true; |
| 234 | break; |
| 235 | } |
| 236 | next = next.next(); |
| 237 | } |
| 238 | } |
| 239 | m_listInfo.insert(key: list, value: info); |
| 240 | return info; |
| 241 | } |
| 242 | return m_listInfo.value(key: list); |
| 243 | } |
| 244 | |
| 245 | void QTextMarkdownWriter::setLinePrefixForBlockQuote(int level) |
| 246 | { |
| 247 | m_linePrefix.clear(); |
| 248 | if (level > 0) { |
| 249 | m_linePrefix.reserve(asize: level * 2); |
| 250 | for (int i = 0; i < level; ++i) |
| 251 | m_linePrefix += u"> " ; |
| 252 | } |
| 253 | } |
| 254 | |
| 255 | static int nearestWordWrapIndex(const QString &s, int before) |
| 256 | { |
| 257 | before = qMin(a: before, b: s.size()); |
| 258 | int fragBegin = qMax(a: before - 15, b: 0); |
| 259 | if (lcMDW().isDebugEnabled()) { |
| 260 | QString frag = s.mid(position: fragBegin, n: 30); |
| 261 | qCDebug(lcMDW) << frag << before; |
| 262 | qCDebug(lcMDW) << QString(before - fragBegin, qtmw_Period) + u'<'; |
| 263 | } |
| 264 | for (int i = before - 1; i >= 0; --i) { |
| 265 | if (s.at(i).isSpace()) { |
| 266 | qCDebug(lcMDW) << QString(i - fragBegin, qtmw_Period) + u'^' << i; |
| 267 | return i; |
| 268 | } |
| 269 | } |
| 270 | qCDebug(lcMDW, "not possible" ); |
| 271 | return -1; |
| 272 | } |
| 273 | |
| 274 | static int adjacentBackticksCount(const QString &s) |
| 275 | { |
| 276 | int start = -1, len = s.size(); |
| 277 | int ret = 0; |
| 278 | for (int i = 0; i < len; ++i) { |
| 279 | if (s.at(i) == qtmw_Backtick) { |
| 280 | if (start < 0) |
| 281 | start = i; |
| 282 | } else if (start >= 0) { |
| 283 | ret = qMax(a: ret, b: i - start); |
| 284 | start = -1; |
| 285 | } |
| 286 | } |
| 287 | if (s.at(i: len - 1) == qtmw_Backtick) |
| 288 | ret = qMax(a: ret, b: len - start); |
| 289 | return ret; |
| 290 | } |
| 291 | |
| 292 | /*! \internal |
| 293 | Escape anything at the beginning of a line of markdown that would be |
| 294 | misinterpreted by a markdown parser, including any period that follows a |
| 295 | number (to avoid misinterpretation as a numbered list item). |
| 296 | https://spec.commonmark.org/0.31.2/#backslash-escapes |
| 297 | */ |
| 298 | static void maybeEscapeFirstChar(QString &s) |
| 299 | { |
| 300 | static const QRegularExpression numericListRe(uR"(\d+([\.)])\s)"_s ); |
| 301 | static const QLatin1StringView specialFirstCharacters("#*+-" ); |
| 302 | |
| 303 | QString sTrimmed = s.trimmed(); |
| 304 | if (sTrimmed.isEmpty()) |
| 305 | return; |
| 306 | QChar firstChar = sTrimmed.at(i: 0); |
| 307 | if (specialFirstCharacters.contains(c: firstChar)) { |
| 308 | int i = s.indexOf(ch: firstChar); // == 0 unless s got trimmed |
| 309 | s.insert(i, c: u'\\'); |
| 310 | } else { |
| 311 | auto match = numericListRe.match(subject: s, offset: 0, matchType: QRegularExpression::NormalMatch, |
| 312 | matchOptions: QRegularExpression::AnchorAtOffsetMatchOption); |
| 313 | if (match.hasMatch()) |
| 314 | s.insert(i: match.capturedStart(nth: 1), c: qtmw_Backslash); |
| 315 | } |
| 316 | } |
| 317 | |
| 318 | /*! \internal |
| 319 | Escape all backslashes. Then escape any special character that stands |
| 320 | alone or prefixes a "word", including the \c < that starts an HTML tag. |
| 321 | https://spec.commonmark.org/0.31.2/#backslash-escapes |
| 322 | */ |
| 323 | static void escapeSpecialCharacters(QString &s) |
| 324 | { |
| 325 | static const QRegularExpression spaceRe(uR"(\s+)"_s ); |
| 326 | static const QRegularExpression specialRe(uR"([<!*[`&]+[/\w])"_s ); |
| 327 | |
| 328 | s.replace(before: "\\"_L1 , after: "\\\\"_L1 ); |
| 329 | |
| 330 | int i = 0; |
| 331 | while (i >= 0) { |
| 332 | if (int j = s.indexOf(re: specialRe, from: i); j >= 0) { |
| 333 | s.insert(i: j, c: qtmw_Backslash); |
| 334 | i = j + 3; |
| 335 | } |
| 336 | i = s.indexOf(re: spaceRe, from: i); |
| 337 | if (i >= 0) |
| 338 | ++i; // past the whitespace, if found |
| 339 | } |
| 340 | } |
| 341 | |
| 342 | struct LineEndPositions { |
| 343 | const QChar *lineEnd; |
| 344 | const QChar *nextLineBegin; |
| 345 | }; |
| 346 | |
| 347 | static LineEndPositions findLineEnd(const QChar *begin, const QChar *end) |
| 348 | { |
| 349 | LineEndPositions result{ .lineEnd: end, .nextLineBegin: end }; |
| 350 | |
| 351 | while (begin < end) { |
| 352 | if (*begin == qtmw_Newline) { |
| 353 | result.lineEnd = begin; |
| 354 | result.nextLineBegin = begin + 1; |
| 355 | break; |
| 356 | } else if (*begin == qtmw_CarriageReturn) { |
| 357 | result.lineEnd = begin; |
| 358 | result.nextLineBegin = begin + 1; |
| 359 | if (((begin + 1) < end) && begin[1] == qtmw_Newline) |
| 360 | ++result.nextLineBegin; |
| 361 | break; |
| 362 | } |
| 363 | |
| 364 | ++begin; |
| 365 | } |
| 366 | |
| 367 | return result; |
| 368 | } |
| 369 | |
| 370 | static bool isBlankLine(const QChar *begin, const QChar *end) |
| 371 | { |
| 372 | while (begin < end) { |
| 373 | if (*begin != qtmw_Space && *begin != qtmw_Tab) |
| 374 | return false; |
| 375 | ++begin; |
| 376 | } |
| 377 | return true; |
| 378 | } |
| 379 | |
| 380 | static QString createLinkTitle(const QString &title) |
| 381 | { |
| 382 | QString result; |
| 383 | result.reserve(asize: title.size() + 2); |
| 384 | result += qtmw_DoubleQuote; |
| 385 | |
| 386 | const QChar *data = title.data(); |
| 387 | const QChar *end = data + title.size(); |
| 388 | |
| 389 | while (data < end) { |
| 390 | const auto lineEndPositions = findLineEnd(begin: data, end); |
| 391 | |
| 392 | if (!isBlankLine(begin: data, end: lineEndPositions.lineEnd)) { |
| 393 | while (data < lineEndPositions.nextLineBegin) { |
| 394 | if (*data == qtmw_DoubleQuote) |
| 395 | result += qtmw_Backslash; |
| 396 | result += *data; |
| 397 | ++data; |
| 398 | } |
| 399 | } |
| 400 | |
| 401 | data = lineEndPositions.nextLineBegin; |
| 402 | } |
| 403 | |
| 404 | result += qtmw_DoubleQuote; |
| 405 | return result; |
| 406 | } |
| 407 | |
| 408 | int QTextMarkdownWriter::writeBlock(const QTextBlock &block, bool wrap, bool ignoreFormat, bool ignoreEmpty) |
| 409 | { |
| 410 | if (block.text().isEmpty() && ignoreEmpty) |
| 411 | return 0; |
| 412 | const int ColumnLimit = 80; |
| 413 | QTextBlockFormat blockFmt = block.blockFormat(); |
| 414 | bool missedBlankCodeBlockLine = false; |
| 415 | const bool codeBlock = blockFmt.hasProperty(propertyId: QTextFormat::BlockCodeFence) || |
| 416 | blockFmt.stringProperty(propertyId: QTextFormat::BlockCodeLanguage).size() > 0 || |
| 417 | blockFmt.nonBreakableLines(); |
| 418 | const int blockQuoteLevel = blockFmt.intProperty(propertyId: QTextFormat::BlockQuoteLevel); |
| 419 | if (m_fencedCodeBlock && !codeBlock) { |
| 420 | m_stream << m_linePrefix << m_codeBlockFence << qtmw_Newline; |
| 421 | m_fencedCodeBlock = false; |
| 422 | m_codeBlockFence.clear(); |
| 423 | m_linePrefixWritten = m_linePrefix.size() > 0; |
| 424 | } |
| 425 | m_linePrefix.clear(); |
| 426 | if (!blockFmt.headingLevel() && blockQuoteLevel > 0) { |
| 427 | setLinePrefixForBlockQuote(blockQuoteLevel); |
| 428 | if (!m_linePrefixWritten) { |
| 429 | m_stream << m_linePrefix; |
| 430 | m_linePrefixWritten = true; |
| 431 | } |
| 432 | } |
| 433 | if (block.textList()) { // it's a list-item |
| 434 | auto fmt = block.textList()->format(); |
| 435 | const int listLevel = fmt.indent(); |
| 436 | // Negative numbers don't start a list in Markdown, so ignore them. |
| 437 | const int start = fmt.start() >= 0 ? fmt.start() : 1; |
| 438 | const int number = block.textList()->itemNumber(block) + start; |
| 439 | QByteArray bullet = " " ; |
| 440 | bool numeric = false; |
| 441 | switch (fmt.style()) { |
| 442 | case QTextListFormat::ListDisc: |
| 443 | bullet = "-" ; |
| 444 | m_wrappedLineIndent = 2; |
| 445 | break; |
| 446 | case QTextListFormat::ListCircle: |
| 447 | bullet = "*" ; |
| 448 | m_wrappedLineIndent = 2; |
| 449 | break; |
| 450 | case QTextListFormat::ListSquare: |
| 451 | bullet = "+" ; |
| 452 | m_wrappedLineIndent = 2; |
| 453 | break; |
| 454 | case QTextListFormat::ListStyleUndefined: break; |
| 455 | case QTextListFormat::ListDecimal: |
| 456 | case QTextListFormat::ListLowerAlpha: |
| 457 | case QTextListFormat::ListUpperAlpha: |
| 458 | case QTextListFormat::ListLowerRoman: |
| 459 | case QTextListFormat::ListUpperRoman: |
| 460 | numeric = true; |
| 461 | m_wrappedLineIndent = 4; |
| 462 | break; |
| 463 | } |
| 464 | switch (blockFmt.marker()) { |
| 465 | case QTextBlockFormat::MarkerType::Checked: |
| 466 | bullet += " [x]" ; |
| 467 | break; |
| 468 | case QTextBlockFormat::MarkerType::Unchecked: |
| 469 | bullet += " [ ]" ; |
| 470 | break; |
| 471 | default: |
| 472 | break; |
| 473 | } |
| 474 | int indentFirstLine = (listLevel - 1) * (numeric ? 4 : 2); |
| 475 | m_wrappedLineIndent += indentFirstLine; |
| 476 | if (m_lastListIndent != listLevel && !m_doubleNewlineWritten && listInfo(list: block.textList()).loose) |
| 477 | m_stream << qtmw_Newline; |
| 478 | m_lastListIndent = listLevel; |
| 479 | QString prefix(indentFirstLine, qtmw_Space); |
| 480 | if (numeric) { |
| 481 | QString suffix = fmt.numberSuffix(); |
| 482 | if (suffix.isEmpty()) |
| 483 | suffix = QString(qtmw_Period); |
| 484 | QString numberStr = QString::number(number) + suffix + qtmw_Space; |
| 485 | if (numberStr.size() == 3) |
| 486 | numberStr += qtmw_Space; |
| 487 | prefix += numberStr; |
| 488 | } else { |
| 489 | prefix += QLatin1StringView(bullet) + qtmw_Space; |
| 490 | } |
| 491 | m_stream << prefix; |
| 492 | } else if (blockFmt.hasProperty(propertyId: QTextFormat::BlockTrailingHorizontalRulerWidth)) { |
| 493 | m_stream << "- - -\n" ; // unambiguous horizontal rule, not an underline under a heading |
| 494 | return 0; |
| 495 | } else if (codeBlock) { |
| 496 | // It's important to preserve blank lines in code blocks. But blank lines in code blocks |
| 497 | // inside block quotes are getting preserved anyway (along with the "> " prefix). |
| 498 | if (!blockFmt.hasProperty(propertyId: QTextFormat::BlockQuoteLevel)) |
| 499 | missedBlankCodeBlockLine = true; // only if we don't get any fragments below |
| 500 | if (!m_fencedCodeBlock) { |
| 501 | QString fenceChar = blockFmt.stringProperty(propertyId: QTextFormat::BlockCodeFence); |
| 502 | if (fenceChar.isEmpty()) |
| 503 | fenceChar = "`"_L1 ; |
| 504 | m_codeBlockFence = QString(3, fenceChar.at(i: 0)); |
| 505 | if (blockFmt.hasProperty(propertyId: QTextFormat::BlockIndent)) |
| 506 | m_codeBlockFence = QString(m_wrappedLineIndent, qtmw_Space) + m_codeBlockFence; |
| 507 | // A block quote can contain an indented code block, but not vice-versa. |
| 508 | m_stream << m_codeBlockFence << blockFmt.stringProperty(propertyId: QTextFormat::BlockCodeLanguage) |
| 509 | << qtmw_Newline << m_linePrefix; |
| 510 | m_fencedCodeBlock = true; |
| 511 | } |
| 512 | wrap = false; |
| 513 | } else if (!blockFmt.indent()) { |
| 514 | m_wrappedLineIndent = 0; |
| 515 | if (blockFmt.hasProperty(propertyId: QTextFormat::BlockCodeLanguage)) { |
| 516 | // A block quote can contain an indented code block, but not vice-versa. |
| 517 | m_linePrefix += QString(4, qtmw_Space); |
| 518 | m_indentedCodeBlock = true; |
| 519 | } |
| 520 | if (!m_linePrefixWritten) { |
| 521 | m_stream << m_linePrefix; |
| 522 | m_linePrefixWritten = true; |
| 523 | } |
| 524 | } |
| 525 | if (blockFmt.headingLevel()) { |
| 526 | m_stream << QByteArray(blockFmt.headingLevel(), '#') << ' '; |
| 527 | wrap = false; |
| 528 | } |
| 529 | |
| 530 | QString wrapIndentString = m_linePrefix + QString(m_wrappedLineIndent, qtmw_Space); |
| 531 | // It would be convenient if QTextStream had a lineCharPos() accessor, |
| 532 | // to keep track of how many characters (not bytes) have been written on the current line, |
| 533 | // but it doesn't. So we have to keep track with this col variable. |
| 534 | int col = wrapIndentString.size(); |
| 535 | bool mono = false; |
| 536 | bool startsOrEndsWithBacktick = false; |
| 537 | bool bold = false; |
| 538 | bool italic = false; |
| 539 | bool underline = false; |
| 540 | bool strikeOut = false; |
| 541 | bool endingMarkers = false; |
| 542 | QString backticks(qtmw_Backtick); |
| 543 | for (QTextBlock::Iterator frag = block.begin(); !frag.atEnd(); ++frag) { |
| 544 | missedBlankCodeBlockLine = false; |
| 545 | QString fragmentText = frag.fragment().text(); |
| 546 | while (fragmentText.endsWith(c: qtmw_Newline)) |
| 547 | fragmentText.chop(n: 1); |
| 548 | if (!(m_fencedCodeBlock || m_indentedCodeBlock)) { |
| 549 | escapeSpecialCharacters(s&: fragmentText); |
| 550 | maybeEscapeFirstChar(s&: fragmentText); |
| 551 | } |
| 552 | if (block.textList()) { // <li>first line</br>continuation</li> |
| 553 | QString newlineIndent = |
| 554 | QString(qtmw_Newline) + QString(m_wrappedLineIndent, qtmw_Space); |
| 555 | fragmentText.replace(before: QString(qtmw_LineBreak), after: newlineIndent); |
| 556 | } else if (blockFmt.indent() > 0) { // <li>first line<p>continuation</p></li> |
| 557 | m_stream << QString(m_wrappedLineIndent, qtmw_Space); |
| 558 | } else { |
| 559 | fragmentText.replace(before: qtmw_LineBreak, after: qtmw_Newline); |
| 560 | } |
| 561 | startsOrEndsWithBacktick |= |
| 562 | fragmentText.startsWith(c: qtmw_Backtick) || fragmentText.endsWith(c: qtmw_Backtick); |
| 563 | QTextCharFormat fmt = frag.fragment().charFormat(); |
| 564 | if (fmt.isImageFormat()) { |
| 565 | QTextImageFormat ifmt = fmt.toImageFormat(); |
| 566 | QString desc = ifmt.stringProperty(propertyId: QTextFormat::ImageAltText); |
| 567 | if (desc.isEmpty()) |
| 568 | desc = "image"_L1 ; |
| 569 | QString s = "; |
| 570 | QString title = ifmt.stringProperty(propertyId: QTextFormat::ImageTitle); |
| 571 | if (!title.isEmpty()) |
| 572 | s += qtmw_Space + qtmw_DoubleQuote + title + qtmw_DoubleQuote; |
| 573 | s += u')'; |
| 574 | if (wrap && col + s.size() > ColumnLimit) { |
| 575 | m_stream << qtmw_Newline << wrapIndentString; |
| 576 | col = m_wrappedLineIndent; |
| 577 | } |
| 578 | m_stream << s; |
| 579 | col += s.size(); |
| 580 | } else if (fmt.hasProperty(propertyId: QTextFormat::AnchorHref)) { |
| 581 | const auto href = fmt.property(propertyId: QTextFormat::AnchorHref).toString(); |
| 582 | const bool hasToolTip = fmt.hasProperty(propertyId: QTextFormat::TextToolTip); |
| 583 | QString s; |
| 584 | if (!hasToolTip && href == fragmentText && !QUrl(href, QUrl::StrictMode).scheme().isEmpty()) { |
| 585 | s = u'<' + href + u'>'; |
| 586 | } else { |
| 587 | s = u'[' + fragmentText + "]("_L1 + href; |
| 588 | if (hasToolTip) { |
| 589 | s += qtmw_Space; |
| 590 | s += createLinkTitle(title: fmt.property(propertyId: QTextFormat::TextToolTip).toString()); |
| 591 | } |
| 592 | s += u')'; |
| 593 | } |
| 594 | if (wrap && col + s.size() > ColumnLimit) { |
| 595 | m_stream << qtmw_Newline << wrapIndentString; |
| 596 | col = m_wrappedLineIndent; |
| 597 | } |
| 598 | m_stream << s; |
| 599 | col += s.size(); |
| 600 | } else { |
| 601 | QFontInfo fontInfo(fmt.font()); |
| 602 | bool monoFrag = fontInfo.fixedPitch() || fmt.fontFixedPitch(); |
| 603 | QString markers; |
| 604 | if (!ignoreFormat) { |
| 605 | if (monoFrag != mono && !m_indentedCodeBlock && !m_fencedCodeBlock) { |
| 606 | if (monoFrag) |
| 607 | backticks = |
| 608 | QString(adjacentBackticksCount(s: fragmentText) + 1, qtmw_Backtick); |
| 609 | markers += backticks; |
| 610 | if (startsOrEndsWithBacktick) |
| 611 | markers += qtmw_Space; |
| 612 | mono = monoFrag; |
| 613 | if (!mono) |
| 614 | endingMarkers = true; |
| 615 | } |
| 616 | if (!blockFmt.headingLevel() && !mono) { |
| 617 | if (fontInfo.bold() != bold) { |
| 618 | markers += "**"_L1 ; |
| 619 | bold = fontInfo.bold(); |
| 620 | if (!bold) |
| 621 | endingMarkers = true; |
| 622 | } |
| 623 | if (fontInfo.italic() != italic) { |
| 624 | markers += u'*'; |
| 625 | italic = fontInfo.italic(); |
| 626 | if (!italic) |
| 627 | endingMarkers = true; |
| 628 | } |
| 629 | if (fontInfo.strikeOut() != strikeOut) { |
| 630 | markers += "~~"_L1 ; |
| 631 | strikeOut = fontInfo.strikeOut(); |
| 632 | if (!strikeOut) |
| 633 | endingMarkers = true; |
| 634 | } |
| 635 | if (fontInfo.underline() != underline) { |
| 636 | // CommonMark specifies underline as another way to get emphasis (italics): |
| 637 | // https://spec.commonmark.org/0.31.2/#example-148 |
| 638 | // but md4c allows us to distinguish them; so we support underlining (in GitHub dialect). |
| 639 | markers += u'_'; |
| 640 | underline = fontInfo.underline(); |
| 641 | if (!underline) |
| 642 | endingMarkers = true; |
| 643 | } |
| 644 | } |
| 645 | } |
| 646 | if (wrap && col + markers.size() * 2 + fragmentText.size() > ColumnLimit) { |
| 647 | int i = 0; |
| 648 | const int fragLen = fragmentText.size(); |
| 649 | bool breakingLine = false; |
| 650 | while (i < fragLen) { |
| 651 | if (col >= ColumnLimit) { |
| 652 | m_stream << markers << qtmw_Newline << wrapIndentString; |
| 653 | markers.clear(); |
| 654 | col = m_wrappedLineIndent; |
| 655 | while (i < fragLen && fragmentText[i].isSpace()) |
| 656 | ++i; |
| 657 | } |
| 658 | int j = i + ColumnLimit - col; |
| 659 | if (j < fragLen) { |
| 660 | int wi = nearestWordWrapIndex(s: fragmentText, before: j); |
| 661 | if (wi < 0) { |
| 662 | j = fragLen; |
| 663 | // can't break within the fragment: we need to break already _before_ it |
| 664 | if (endingMarkers) { |
| 665 | m_stream << markers; |
| 666 | markers.clear(); |
| 667 | } |
| 668 | m_stream << qtmw_Newline << wrapIndentString; |
| 669 | col = m_wrappedLineIndent; |
| 670 | } else if (wi >= i) { |
| 671 | j = wi; |
| 672 | breakingLine = true; |
| 673 | } |
| 674 | } else { |
| 675 | j = fragLen; |
| 676 | breakingLine = false; |
| 677 | } |
| 678 | QString subfrag = fragmentText.mid(position: i, n: j - i); |
| 679 | if (!i) { |
| 680 | m_stream << markers; |
| 681 | col += markers.size(); |
| 682 | } |
| 683 | if (col == m_wrappedLineIndent) |
| 684 | maybeEscapeFirstChar(s&: subfrag); |
| 685 | m_stream << subfrag; |
| 686 | if (breakingLine) { |
| 687 | m_stream << qtmw_Newline << wrapIndentString; |
| 688 | col = m_wrappedLineIndent; |
| 689 | } else { |
| 690 | col += subfrag.size(); |
| 691 | } |
| 692 | i = j + 1; |
| 693 | } // loop over fragment characters (we know we need to break somewhere) |
| 694 | } else { |
| 695 | if (!m_linePrefixWritten && col == wrapIndentString.size()) { |
| 696 | m_stream << m_linePrefix; |
| 697 | col += m_linePrefix.size(); |
| 698 | } |
| 699 | m_stream << markers << fragmentText; |
| 700 | col += markers.size() + fragmentText.size(); |
| 701 | } |
| 702 | } |
| 703 | } |
| 704 | if (mono) { |
| 705 | if (startsOrEndsWithBacktick) { |
| 706 | m_stream << qtmw_Space; |
| 707 | col += 1; |
| 708 | } |
| 709 | m_stream << backticks; |
| 710 | col += backticks.size(); |
| 711 | } |
| 712 | if (bold) { |
| 713 | m_stream << "**" ; |
| 714 | col += 2; |
| 715 | } |
| 716 | if (italic) { |
| 717 | m_stream << "*" ; |
| 718 | col += 1; |
| 719 | } |
| 720 | if (underline) { |
| 721 | m_stream << "_" ; |
| 722 | col += 1; |
| 723 | } |
| 724 | if (strikeOut) { |
| 725 | m_stream << "~~" ; |
| 726 | col += 2; |
| 727 | } |
| 728 | if (missedBlankCodeBlockLine) |
| 729 | m_stream << qtmw_Newline; |
| 730 | m_linePrefixWritten = false; |
| 731 | return col; |
| 732 | } |
| 733 | |
| 734 | QT_END_NAMESPACE |
| 735 | |