| 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 "qtextmarkdownimporter_p.h" |
| 5 | #include "qtextdocumentfragment_p.h" |
| 6 | #include <QLoggingCategory> |
| 7 | #if QT_CONFIG(regularexpression) |
| 8 | #include <QRegularExpression> |
| 9 | #endif |
| 10 | #include <QTextCursor> |
| 11 | #include <QTextDocument> |
| 12 | #include <QTextDocumentFragment> |
| 13 | #include <QTextList> |
| 14 | #include <QTextTable> |
| 15 | #if QT_CONFIG(system_textmarkdownreader) |
| 16 | #include <md4c.h> |
| 17 | #else |
| 18 | #include "../../3rdparty/md4c/md4c.h" |
| 19 | #endif |
| 20 | |
| 21 | QT_BEGIN_NAMESPACE |
| 22 | |
| 23 | using namespace Qt::StringLiterals; |
| 24 | |
| 25 | Q_LOGGING_CATEGORY(lcMD, "qt.text.markdown" ) |
| 26 | |
| 27 | static const QChar qtmi_Newline = u'\n'; |
| 28 | static const QChar qtmi_Space = u' '; |
| 29 | |
| 30 | static constexpr auto markerString() noexcept { return "---"_L1 ; } |
| 31 | |
| 32 | // TODO maybe eliminate the margins after all views recognize BlockQuoteLevel, CSS can format it, etc. |
| 33 | static const int qtmi_BlockQuoteIndent = |
| 34 | 40; // pixels, same as in QTextHtmlParserNode::initializeProperties |
| 35 | |
| 36 | static_assert(int(QTextMarkdownImporter::FeatureCollapseWhitespace) == MD_FLAG_COLLAPSEWHITESPACE); |
| 37 | static_assert(int(QTextMarkdownImporter::FeaturePermissiveATXHeaders) == MD_FLAG_PERMISSIVEATXHEADERS); |
| 38 | static_assert(int(QTextMarkdownImporter::FeaturePermissiveURLAutoLinks) == MD_FLAG_PERMISSIVEURLAUTOLINKS); |
| 39 | static_assert(int(QTextMarkdownImporter::FeaturePermissiveMailAutoLinks) == MD_FLAG_PERMISSIVEEMAILAUTOLINKS); |
| 40 | static_assert(int(QTextMarkdownImporter::FeatureNoIndentedCodeBlocks) == MD_FLAG_NOINDENTEDCODEBLOCKS); |
| 41 | static_assert(int(QTextMarkdownImporter::FeatureNoHTMLBlocks) == MD_FLAG_NOHTMLBLOCKS); |
| 42 | static_assert(int(QTextMarkdownImporter::FeatureNoHTMLSpans) == MD_FLAG_NOHTMLSPANS); |
| 43 | static_assert(int(QTextMarkdownImporter::FeatureTables) == MD_FLAG_TABLES); |
| 44 | static_assert(int(QTextMarkdownImporter::FeatureStrikeThrough) == MD_FLAG_STRIKETHROUGH); |
| 45 | static_assert(int(QTextMarkdownImporter::FeatureUnderline) == MD_FLAG_UNDERLINE); |
| 46 | static_assert(int(QTextMarkdownImporter::FeaturePermissiveWWWAutoLinks) == MD_FLAG_PERMISSIVEWWWAUTOLINKS); |
| 47 | static_assert(int(QTextMarkdownImporter::FeaturePermissiveAutoLinks) == MD_FLAG_PERMISSIVEAUTOLINKS); |
| 48 | static_assert(int(QTextMarkdownImporter::FeatureTasklists) == MD_FLAG_TASKLISTS); |
| 49 | static_assert(int(QTextMarkdownImporter::FeatureNoHTML) == MD_FLAG_NOHTML); |
| 50 | static_assert(int(QTextMarkdownImporter::DialectCommonMark) == MD_DIALECT_COMMONMARK); |
| 51 | static_assert(int(QTextMarkdownImporter::DialectGitHub) == |
| 52 | (MD_DIALECT_GITHUB | MD_FLAG_UNDERLINE | QTextMarkdownImporter::FeatureFrontMatter)); |
| 53 | |
| 54 | // -------------------------------------------------------- |
| 55 | // MD4C callback function wrappers |
| 56 | |
| 57 | static int CbEnterBlock(MD_BLOCKTYPE type, void *detail, void *userdata) |
| 58 | { |
| 59 | QTextMarkdownImporter *mdi = static_cast<QTextMarkdownImporter *>(userdata); |
| 60 | return mdi->cbEnterBlock(blockType: int(type), detail); |
| 61 | } |
| 62 | |
| 63 | static int CbLeaveBlock(MD_BLOCKTYPE type, void *detail, void *userdata) |
| 64 | { |
| 65 | QTextMarkdownImporter *mdi = static_cast<QTextMarkdownImporter *>(userdata); |
| 66 | return mdi->cbLeaveBlock(blockType: int(type), detail); |
| 67 | } |
| 68 | |
| 69 | static int CbEnterSpan(MD_SPANTYPE type, void *detail, void *userdata) |
| 70 | { |
| 71 | QTextMarkdownImporter *mdi = static_cast<QTextMarkdownImporter *>(userdata); |
| 72 | return mdi->cbEnterSpan(spanType: int(type), detail); |
| 73 | } |
| 74 | |
| 75 | static int CbLeaveSpan(MD_SPANTYPE type, void *detail, void *userdata) |
| 76 | { |
| 77 | QTextMarkdownImporter *mdi = static_cast<QTextMarkdownImporter *>(userdata); |
| 78 | return mdi->cbLeaveSpan(spanType: int(type), detail); |
| 79 | } |
| 80 | |
| 81 | static int CbText(MD_TEXTTYPE type, const MD_CHAR *text, MD_SIZE size, void *userdata) |
| 82 | { |
| 83 | QTextMarkdownImporter *mdi = static_cast<QTextMarkdownImporter *>(userdata); |
| 84 | return mdi->cbText(textType: int(type), text, size); |
| 85 | } |
| 86 | |
| 87 | static void CbDebugLog(const char *msg, void *userdata) |
| 88 | { |
| 89 | Q_UNUSED(userdata); |
| 90 | qCDebug(lcMD) << msg; |
| 91 | } |
| 92 | |
| 93 | // MD4C callback function wrappers |
| 94 | // -------------------------------------------------------- |
| 95 | |
| 96 | static Qt::Alignment MdAlignment(MD_ALIGN a, Qt::Alignment defaultAlignment = Qt::AlignLeft | Qt::AlignVCenter) |
| 97 | { |
| 98 | switch (a) { |
| 99 | case MD_ALIGN_LEFT: |
| 100 | return Qt::AlignLeft | Qt::AlignVCenter; |
| 101 | case MD_ALIGN_CENTER: |
| 102 | return Qt::AlignHCenter | Qt::AlignVCenter; |
| 103 | case MD_ALIGN_RIGHT: |
| 104 | return Qt::AlignRight | Qt::AlignVCenter; |
| 105 | default: // including MD_ALIGN_DEFAULT |
| 106 | return defaultAlignment; |
| 107 | } |
| 108 | } |
| 109 | |
| 110 | QTextMarkdownImporter::QTextMarkdownImporter(QTextDocument *doc, QTextMarkdownImporter::Features features) |
| 111 | : m_cursor(doc) |
| 112 | , m_monoFont(QFontDatabase::systemFont(type: QFontDatabase::FixedFont)) |
| 113 | , m_features(features) |
| 114 | { |
| 115 | } |
| 116 | |
| 117 | QTextMarkdownImporter::QTextMarkdownImporter(QTextDocument *doc, QTextDocument::MarkdownFeatures features) |
| 118 | : QTextMarkdownImporter(doc, static_cast<QTextMarkdownImporter::Features>(int(features))) |
| 119 | { |
| 120 | } |
| 121 | |
| 122 | void QTextMarkdownImporter::import(const QString &markdown) |
| 123 | { |
| 124 | MD_PARSER callbacks = { |
| 125 | .abi_version: 0, // abi_version |
| 126 | .flags: unsigned(m_features), |
| 127 | .enter_block: &CbEnterBlock, |
| 128 | .leave_block: &CbLeaveBlock, |
| 129 | .enter_span: &CbEnterSpan, |
| 130 | .leave_span: &CbLeaveSpan, |
| 131 | .text: &CbText, |
| 132 | .debug_log: &CbDebugLog, |
| 133 | .syntax: nullptr // syntax |
| 134 | }; |
| 135 | QTextDocument *doc = m_cursor.document(); |
| 136 | const auto defaultFont = doc->defaultFont(); |
| 137 | m_paragraphMargin = defaultFont.pointSize() * 2 / 3; |
| 138 | doc->clear(); |
| 139 | if (defaultFont.pointSize() != -1) |
| 140 | m_monoFont.setPointSize(defaultFont.pointSize()); |
| 141 | else |
| 142 | m_monoFont.setPixelSize(defaultFont.pixelSize()); |
| 143 | qCDebug(lcMD) << "default font" << defaultFont << "mono font" << m_monoFont; |
| 144 | QStringView md = markdown; |
| 145 | |
| 146 | if (m_features.testFlag(flag: QTextMarkdownImporter::FeatureFrontMatter) && md.startsWith(s: markerString())) { |
| 147 | qsizetype endMarkerPos = md.indexOf(s: markerString(), from: markerString().size() + 1); |
| 148 | if (endMarkerPos > 4) { |
| 149 | qsizetype firstLinePos = 4; // first line of yaml |
| 150 | while (md.at(n: firstLinePos) == '\n'_L1 || md.at(n: firstLinePos) == '\r'_L1) |
| 151 | ++firstLinePos; |
| 152 | auto frontMatter = md.sliced(pos: firstLinePos, n: endMarkerPos - firstLinePos); |
| 153 | firstLinePos = endMarkerPos + 4; // first line of markdown after yaml |
| 154 | while (md.size() > firstLinePos && (md.at(n: firstLinePos) == '\n'_L1 || md.at(n: firstLinePos) == '\r'_L1)) |
| 155 | ++firstLinePos; |
| 156 | md = md.sliced(pos: firstLinePos); |
| 157 | doc->setMetaInformation(info: QTextDocument::FrontMatter, frontMatter.toString()); |
| 158 | qCDebug(lcMD) << "extracted FrontMatter: size" << frontMatter.size(); |
| 159 | } |
| 160 | } |
| 161 | const auto mdUtf8 = md.toUtf8(); |
| 162 | m_cursor.beginEditBlock(); |
| 163 | md_parse(text: mdUtf8.constData(), size: MD_SIZE(mdUtf8.size()), parser: &callbacks, userdata: this); |
| 164 | m_cursor.endEditBlock(); |
| 165 | } |
| 166 | |
| 167 | int QTextMarkdownImporter::cbEnterBlock(int blockType, void *det) |
| 168 | { |
| 169 | m_blockType = blockType; |
| 170 | switch (blockType) { |
| 171 | case MD_BLOCK_P: |
| 172 | if (!m_listStack.isEmpty()) |
| 173 | qCDebug(lcMD, m_listItem ? "P of LI at level %d" : "P continuation inside LI at level %d" , int(m_listStack.size())); |
| 174 | else |
| 175 | qCDebug(lcMD, "P" ); |
| 176 | m_needsInsertBlock = true; |
| 177 | break; |
| 178 | case MD_BLOCK_QUOTE: |
| 179 | ++m_blockQuoteDepth; |
| 180 | qCDebug(lcMD, "QUOTE level %d" , m_blockQuoteDepth); |
| 181 | break; |
| 182 | case MD_BLOCK_CODE: { |
| 183 | MD_BLOCK_CODE_DETAIL *detail = static_cast<MD_BLOCK_CODE_DETAIL *>(det); |
| 184 | m_codeBlock = true; |
| 185 | m_blockCodeLanguage = QLatin1StringView(detail->lang.text, int(detail->lang.size)); |
| 186 | m_blockCodeFence = detail->fence_char; |
| 187 | QString info = QLatin1StringView(detail->info.text, int(detail->info.size)); |
| 188 | m_needsInsertBlock = true; |
| 189 | if (m_blockQuoteDepth) |
| 190 | qCDebug(lcMD, "CODE lang '%s' info '%s' fenced with '%c' inside QUOTE %d" , qPrintable(m_blockCodeLanguage), qPrintable(info), m_blockCodeFence, m_blockQuoteDepth); |
| 191 | else |
| 192 | qCDebug(lcMD, "CODE lang '%s' info '%s' fenced with '%c'" , qPrintable(m_blockCodeLanguage), qPrintable(info), m_blockCodeFence); |
| 193 | } break; |
| 194 | case MD_BLOCK_H: { |
| 195 | MD_BLOCK_H_DETAIL *detail = static_cast<MD_BLOCK_H_DETAIL *>(det); |
| 196 | QTextBlockFormat blockFmt; |
| 197 | QTextCharFormat charFmt; |
| 198 | int sizeAdjustment = 4 - int(detail->level); // H1 to H6: +3 to -2 |
| 199 | charFmt.setProperty(propertyId: QTextFormat::FontSizeAdjustment, value: sizeAdjustment); |
| 200 | charFmt.setFontWeight(QFont::Bold); |
| 201 | blockFmt.setHeadingLevel(int(detail->level)); |
| 202 | m_needsInsertBlock = false; |
| 203 | if (m_cursor.document()->isEmpty()) { |
| 204 | m_cursor.setBlockFormat(blockFmt); |
| 205 | m_cursor.setCharFormat(charFmt); |
| 206 | } else { |
| 207 | m_cursor.insertBlock(format: blockFmt, charFormat: charFmt); |
| 208 | } |
| 209 | qCDebug(lcMD, "H%d" , detail->level); |
| 210 | } break; |
| 211 | case MD_BLOCK_LI: { |
| 212 | m_needsInsertBlock = true; |
| 213 | m_listItem = true; |
| 214 | MD_BLOCK_LI_DETAIL *detail = static_cast<MD_BLOCK_LI_DETAIL *>(det); |
| 215 | m_markerType = detail->is_task ? |
| 216 | (detail->task_mark == ' ' ? QTextBlockFormat::MarkerType::Unchecked : QTextBlockFormat::MarkerType::Checked) : |
| 217 | QTextBlockFormat::MarkerType::NoMarker; |
| 218 | qCDebug(lcMD) << "LI" ; |
| 219 | } break; |
| 220 | case MD_BLOCK_UL: { |
| 221 | if (m_needsInsertList) // list nested in an empty list |
| 222 | m_listStack.push(t: m_cursor.insertList(format: m_listFormat)); |
| 223 | else |
| 224 | m_needsInsertList = true; |
| 225 | MD_BLOCK_UL_DETAIL *detail = static_cast<MD_BLOCK_UL_DETAIL *>(det); |
| 226 | m_listFormat = QTextListFormat(); |
| 227 | m_listFormat.setIndent(m_listStack.size() + 1); |
| 228 | switch (detail->mark) { |
| 229 | case '*': |
| 230 | m_listFormat.setStyle(QTextListFormat::ListCircle); |
| 231 | break; |
| 232 | case '+': |
| 233 | m_listFormat.setStyle(QTextListFormat::ListSquare); |
| 234 | break; |
| 235 | default: // including '-' |
| 236 | m_listFormat.setStyle(QTextListFormat::ListDisc); |
| 237 | break; |
| 238 | } |
| 239 | qCDebug(lcMD, "UL %c level %d" , detail->mark, int(m_listStack.size()) + 1); |
| 240 | } break; |
| 241 | case MD_BLOCK_OL: { |
| 242 | if (m_needsInsertList) // list nested in an empty list |
| 243 | m_listStack.push(t: m_cursor.insertList(format: m_listFormat)); |
| 244 | else |
| 245 | m_needsInsertList = true; |
| 246 | MD_BLOCK_OL_DETAIL *detail = static_cast<MD_BLOCK_OL_DETAIL *>(det); |
| 247 | m_listFormat = QTextListFormat(); |
| 248 | m_listFormat.setIndent(m_listStack.size() + 1); |
| 249 | m_listFormat.setNumberSuffix(QChar::fromLatin1(c: detail->mark_delimiter)); |
| 250 | m_listFormat.setStyle(QTextListFormat::ListDecimal); |
| 251 | m_listFormat.setStart(detail->start); |
| 252 | qCDebug(lcMD, "OL xx%d level %d start %d" , detail->mark_delimiter, int(m_listStack.size()) + 1, detail->start); |
| 253 | } break; |
| 254 | case MD_BLOCK_TD: { |
| 255 | MD_BLOCK_TD_DETAIL *detail = static_cast<MD_BLOCK_TD_DETAIL *>(det); |
| 256 | ++m_tableCol; |
| 257 | // absolute movement (and storage of m_tableCol) shouldn't be necessary, but |
| 258 | // movePosition(QTextCursor::NextCell) doesn't work |
| 259 | QTextTableCell cell = m_currentTable->cellAt(row: m_tableRowCount - 1, col: m_tableCol); |
| 260 | if (!cell.isValid()) { |
| 261 | qWarning(msg: "malformed table in Markdown input" ); |
| 262 | return 1; |
| 263 | } |
| 264 | m_cursor = cell.firstCursorPosition(); |
| 265 | QTextBlockFormat blockFmt = m_cursor.blockFormat(); |
| 266 | blockFmt.setAlignment(MdAlignment(a: detail->align)); |
| 267 | m_cursor.setBlockFormat(blockFmt); |
| 268 | qCDebug(lcMD) << "TD; align" << detail->align << MdAlignment(a: detail->align) << "col" << m_tableCol; |
| 269 | } break; |
| 270 | case MD_BLOCK_TH: { |
| 271 | ++m_tableColumnCount; |
| 272 | ++m_tableCol; |
| 273 | if (m_currentTable->columns() < m_tableColumnCount) |
| 274 | m_currentTable->appendColumns(count: 1); |
| 275 | auto cell = m_currentTable->cellAt(row: m_tableRowCount - 1, col: m_tableCol); |
| 276 | if (!cell.isValid()) { |
| 277 | qWarning(msg: "malformed table in Markdown input" ); |
| 278 | return 1; |
| 279 | } |
| 280 | auto fmt = cell.format(); |
| 281 | fmt.setFontWeight(QFont::Bold); |
| 282 | cell.setFormat(fmt); |
| 283 | } break; |
| 284 | case MD_BLOCK_TR: { |
| 285 | ++m_tableRowCount; |
| 286 | m_nonEmptyTableCells.clear(); |
| 287 | if (m_currentTable->rows() < m_tableRowCount) |
| 288 | m_currentTable->appendRows(count: 1); |
| 289 | m_tableCol = -1; |
| 290 | qCDebug(lcMD) << "TR" << m_currentTable->rows(); |
| 291 | } break; |
| 292 | case MD_BLOCK_TABLE: |
| 293 | m_tableColumnCount = 0; |
| 294 | m_tableRowCount = 0; |
| 295 | m_currentTable = m_cursor.insertTable(rows: 1, cols: 1); // we don't know the dimensions yet |
| 296 | break; |
| 297 | case MD_BLOCK_HR: { |
| 298 | qCDebug(lcMD, "HR" ); |
| 299 | QTextBlockFormat blockFmt; |
| 300 | blockFmt.setProperty(propertyId: QTextFormat::BlockTrailingHorizontalRulerWidth, value: 1); |
| 301 | m_cursor.insertBlock(format: blockFmt, charFormat: QTextCharFormat()); |
| 302 | } break; |
| 303 | default: |
| 304 | break; // nothing to do for now |
| 305 | } |
| 306 | return 0; // no error |
| 307 | } |
| 308 | |
| 309 | int QTextMarkdownImporter::cbLeaveBlock(int blockType, void *detail) |
| 310 | { |
| 311 | Q_UNUSED(detail); |
| 312 | switch (blockType) { |
| 313 | case MD_BLOCK_P: |
| 314 | m_listItem = false; |
| 315 | break; |
| 316 | case MD_BLOCK_UL: |
| 317 | case MD_BLOCK_OL: |
| 318 | if (Q_UNLIKELY(m_needsInsertList)) |
| 319 | m_listStack.push(t: m_cursor.createList(format: m_listFormat)); |
| 320 | if (Q_UNLIKELY(m_listStack.isEmpty())) { |
| 321 | qCWarning(lcMD, "list ended unexpectedly" ); |
| 322 | } else { |
| 323 | qCDebug(lcMD, "list at level %d ended" , int(m_listStack.size())); |
| 324 | m_listStack.pop(); |
| 325 | } |
| 326 | break; |
| 327 | case MD_BLOCK_TR: { |
| 328 | // https://github.com/mity/md4c/issues/29 |
| 329 | // MD4C doesn't tell us explicitly which cells are merged, so merge empty cells |
| 330 | // with previous non-empty ones |
| 331 | int mergeEnd = -1; |
| 332 | int mergeBegin = -1; |
| 333 | for (int col = m_tableCol; col >= 0; --col) { |
| 334 | if (m_nonEmptyTableCells.contains(t: col)) { |
| 335 | if (mergeEnd >= 0 && mergeBegin >= 0) { |
| 336 | qCDebug(lcMD) << "merging cells" << mergeBegin << "to" << mergeEnd << "inclusive, on row" << m_currentTable->rows() - 1; |
| 337 | m_currentTable->mergeCells(row: m_currentTable->rows() - 1, col: mergeBegin - 1, numRows: 1, numCols: mergeEnd - mergeBegin + 2); |
| 338 | } |
| 339 | mergeEnd = -1; |
| 340 | mergeBegin = -1; |
| 341 | } else { |
| 342 | if (mergeEnd < 0) |
| 343 | mergeEnd = col; |
| 344 | else |
| 345 | mergeBegin = col; |
| 346 | } |
| 347 | } |
| 348 | } break; |
| 349 | case MD_BLOCK_QUOTE: { |
| 350 | qCDebug(lcMD, "QUOTE level %d ended" , m_blockQuoteDepth); |
| 351 | --m_blockQuoteDepth; |
| 352 | m_needsInsertBlock = true; |
| 353 | } break; |
| 354 | case MD_BLOCK_TABLE: |
| 355 | qCDebug(lcMD) << "table ended with" << m_currentTable->columns() << "cols and" << m_currentTable->rows() << "rows" ; |
| 356 | m_currentTable = nullptr; |
| 357 | m_cursor.movePosition(op: QTextCursor::End); |
| 358 | break; |
| 359 | case MD_BLOCK_LI: |
| 360 | qCDebug(lcMD, "LI at level %d ended" , int(m_listStack.size())); |
| 361 | m_listItem = false; |
| 362 | break; |
| 363 | case MD_BLOCK_CODE: { |
| 364 | m_codeBlock = false; |
| 365 | m_blockCodeLanguage.clear(); |
| 366 | m_blockCodeFence = 0; |
| 367 | if (m_blockQuoteDepth) |
| 368 | qCDebug(lcMD, "CODE ended inside QUOTE %d" , m_blockQuoteDepth); |
| 369 | else |
| 370 | qCDebug(lcMD, "CODE ended" ); |
| 371 | m_needsInsertBlock = true; |
| 372 | } break; |
| 373 | case MD_BLOCK_H: |
| 374 | m_cursor.setCharFormat(QTextCharFormat()); |
| 375 | break; |
| 376 | default: |
| 377 | break; |
| 378 | } |
| 379 | return 0; // no error |
| 380 | } |
| 381 | |
| 382 | int QTextMarkdownImporter::cbEnterSpan(int spanType, void *det) |
| 383 | { |
| 384 | QTextCharFormat charFmt; |
| 385 | if (!m_spanFormatStack.isEmpty()) |
| 386 | charFmt = m_spanFormatStack.top(); |
| 387 | switch (spanType) { |
| 388 | case MD_SPAN_EM: |
| 389 | charFmt.setFontItalic(true); |
| 390 | break; |
| 391 | case MD_SPAN_STRONG: |
| 392 | charFmt.setFontWeight(QFont::Bold); |
| 393 | break; |
| 394 | case MD_SPAN_U: |
| 395 | charFmt.setFontUnderline(true); |
| 396 | break; |
| 397 | case MD_SPAN_A: { |
| 398 | MD_SPAN_A_DETAIL *detail = static_cast<MD_SPAN_A_DETAIL *>(det); |
| 399 | QString url = QString::fromUtf8(utf8: detail->href.text, size: int(detail->href.size)); |
| 400 | QString title = QString::fromUtf8(utf8: detail->title.text, size: int(detail->title.size)); |
| 401 | charFmt.setAnchor(true); |
| 402 | charFmt.setAnchorHref(url); |
| 403 | if (!title.isEmpty()) |
| 404 | charFmt.setToolTip(title); |
| 405 | charFmt.setForeground(m_palette.link()); |
| 406 | qCDebug(lcMD) << "anchor" << url << title; |
| 407 | } break; |
| 408 | case MD_SPAN_IMG: { |
| 409 | m_imageSpan = true; |
| 410 | m_imageFormat = QTextImageFormat(); |
| 411 | MD_SPAN_IMG_DETAIL *detail = static_cast<MD_SPAN_IMG_DETAIL *>(det); |
| 412 | m_imageFormat.setName(QString::fromUtf8(utf8: detail->src.text, size: int(detail->src.size))); |
| 413 | m_imageFormat.setProperty(propertyId: QTextFormat::ImageTitle, value: QString::fromUtf8(utf8: detail->title.text, size: int(detail->title.size))); |
| 414 | break; |
| 415 | } |
| 416 | case MD_SPAN_CODE: |
| 417 | charFmt.setFont(font: m_monoFont); |
| 418 | charFmt.setFontFixedPitch(true); |
| 419 | break; |
| 420 | case MD_SPAN_DEL: |
| 421 | charFmt.setFontStrikeOut(true); |
| 422 | break; |
| 423 | } |
| 424 | m_spanFormatStack.push(t: charFmt); |
| 425 | qCDebug(lcMD) << spanType << "setCharFormat" << charFmt.font().families().constFirst() |
| 426 | << charFmt.fontWeight() << (charFmt.fontItalic() ? "italic" : "" ) |
| 427 | << charFmt.foreground().color().name(); |
| 428 | m_cursor.setCharFormat(charFmt); |
| 429 | return 0; // no error |
| 430 | } |
| 431 | |
| 432 | int QTextMarkdownImporter::cbLeaveSpan(int spanType, void *detail) |
| 433 | { |
| 434 | Q_UNUSED(detail); |
| 435 | QTextCharFormat charFmt; |
| 436 | if (!m_spanFormatStack.isEmpty()) { |
| 437 | m_spanFormatStack.pop(); |
| 438 | if (!m_spanFormatStack.isEmpty()) |
| 439 | charFmt = m_spanFormatStack.top(); |
| 440 | } |
| 441 | m_cursor.setCharFormat(charFmt); |
| 442 | qCDebug(lcMD) << spanType << "setCharFormat" << charFmt.font().families().constFirst() |
| 443 | << charFmt.fontWeight() << (charFmt.fontItalic() ? "italic" : "" ) |
| 444 | << charFmt.foreground().color().name(); |
| 445 | if (spanType == int(MD_SPAN_IMG)) |
| 446 | m_imageSpan = false; |
| 447 | return 0; // no error |
| 448 | } |
| 449 | |
| 450 | int QTextMarkdownImporter::cbText(int textType, const char *text, unsigned size) |
| 451 | { |
| 452 | if (m_needsInsertBlock) |
| 453 | insertBlock(); |
| 454 | #if QT_CONFIG(regularexpression) |
| 455 | static const QRegularExpression openingBracket(QStringLiteral("<[a-zA-Z]" )); |
| 456 | static const QRegularExpression closingBracket(QStringLiteral("(/>|</)" )); |
| 457 | #endif |
| 458 | QString s = QString::fromUtf8(utf8: text, size: int(size)); |
| 459 | |
| 460 | switch (textType) { |
| 461 | case MD_TEXT_NORMAL: |
| 462 | #if QT_CONFIG(regularexpression) |
| 463 | if (m_htmlTagDepth) { |
| 464 | m_htmlAccumulator += s; |
| 465 | s = QString(); |
| 466 | } |
| 467 | #endif |
| 468 | break; |
| 469 | case MD_TEXT_NULLCHAR: |
| 470 | s = QString(QChar(u'\xFFFD')); // CommonMark-required replacement for null |
| 471 | break; |
| 472 | case MD_TEXT_BR: |
| 473 | s = QString(qtmi_Newline); |
| 474 | break; |
| 475 | case MD_TEXT_SOFTBR: |
| 476 | s = QString(qtmi_Space); |
| 477 | break; |
| 478 | case MD_TEXT_CODE: |
| 479 | // We'll see MD_SPAN_CODE too, which will set the char format, and that's enough. |
| 480 | break; |
| 481 | #if QT_CONFIG(texthtmlparser) |
| 482 | case MD_TEXT_ENTITY: |
| 483 | if (m_htmlTagDepth) |
| 484 | m_htmlAccumulator += s; |
| 485 | else |
| 486 | m_cursor.insertHtml(html: s); |
| 487 | s = QString(); |
| 488 | break; |
| 489 | #endif |
| 490 | case MD_TEXT_HTML: |
| 491 | // count how many tags are opened and how many are closed |
| 492 | #if QT_CONFIG(regularexpression) && QT_CONFIG(texthtmlparser) |
| 493 | { |
| 494 | int startIdx = 0; |
| 495 | while ((startIdx = s.indexOf(re: openingBracket, from: startIdx)) >= 0) { |
| 496 | ++m_htmlTagDepth; |
| 497 | startIdx += 2; |
| 498 | } |
| 499 | startIdx = 0; |
| 500 | while ((startIdx = s.indexOf(re: closingBracket, from: startIdx)) >= 0) { |
| 501 | --m_htmlTagDepth; |
| 502 | startIdx += 2; |
| 503 | } |
| 504 | } |
| 505 | m_htmlAccumulator += s; |
| 506 | if (!m_htmlTagDepth) { // all open tags are now closed |
| 507 | qCDebug(lcMD) << "HTML" << m_htmlAccumulator; |
| 508 | m_cursor.insertHtml(html: m_htmlAccumulator); |
| 509 | if (m_spanFormatStack.isEmpty()) |
| 510 | m_cursor.setCharFormat(QTextCharFormat()); |
| 511 | else |
| 512 | m_cursor.setCharFormat(m_spanFormatStack.top()); |
| 513 | m_htmlAccumulator = QString(); |
| 514 | } |
| 515 | #endif |
| 516 | s = QString(); |
| 517 | break; |
| 518 | } |
| 519 | |
| 520 | switch (m_blockType) { |
| 521 | case MD_BLOCK_TD: |
| 522 | m_nonEmptyTableCells.append(t: m_tableCol); |
| 523 | break; |
| 524 | case MD_BLOCK_CODE: |
| 525 | if (s == qtmi_Newline) { |
| 526 | // defer a blank line until we see something else in the code block, |
| 527 | // to avoid ending every code block with a gratuitous blank line |
| 528 | m_needsInsertBlock = true; |
| 529 | s = QString(); |
| 530 | } |
| 531 | break; |
| 532 | default: |
| 533 | break; |
| 534 | } |
| 535 | |
| 536 | if (m_imageSpan) { |
| 537 | // TODO we don't yet support alt text with formatting, because of the cases where m_cursor |
| 538 | // already inserted the text above. Rather need to accumulate it in case we need it here. |
| 539 | m_imageFormat.setProperty(propertyId: QTextFormat::ImageAltText, value: s); |
| 540 | qCDebug(lcMD) << "image" << m_imageFormat.name() |
| 541 | << "title" << m_imageFormat.stringProperty(propertyId: QTextFormat::ImageTitle) |
| 542 | << "alt" << s << "relative to" << m_cursor.document()->baseUrl(); |
| 543 | m_cursor.insertImage(format: m_imageFormat); |
| 544 | return 0; // no error |
| 545 | } |
| 546 | |
| 547 | if (!s.isEmpty()) |
| 548 | m_cursor.insertText(text: s); |
| 549 | if (m_cursor.currentList()) { |
| 550 | // The list item will indent the list item's text, so we don't need indentation on the block. |
| 551 | QTextBlockFormat bfmt = m_cursor.blockFormat(); |
| 552 | bfmt.setIndent(0); |
| 553 | m_cursor.setBlockFormat(bfmt); |
| 554 | } |
| 555 | if (lcMD().isEnabled(type: QtDebugMsg)) { |
| 556 | QTextBlockFormat bfmt = m_cursor.blockFormat(); |
| 557 | QString debugInfo; |
| 558 | if (m_cursor.currentList()) |
| 559 | debugInfo = "in list at depth "_L1 + QString::number(m_cursor.currentList()->format().indent()); |
| 560 | if (bfmt.hasProperty(propertyId: QTextFormat::BlockQuoteLevel)) |
| 561 | debugInfo += "in blockquote at depth "_L1 + |
| 562 | QString::number(bfmt.intProperty(propertyId: QTextFormat::BlockQuoteLevel)); |
| 563 | if (bfmt.hasProperty(propertyId: QTextFormat::BlockCodeLanguage)) |
| 564 | debugInfo += "in a code block"_L1 ; |
| 565 | qCDebug(lcMD) << textType << "in block" << m_blockType << s << qPrintable(debugInfo) |
| 566 | << "bindent" << bfmt.indent() << "tindent" << bfmt.textIndent() |
| 567 | << "margins" << bfmt.leftMargin() << bfmt.topMargin() << bfmt.bottomMargin() << bfmt.rightMargin(); |
| 568 | } |
| 569 | return 0; // no error |
| 570 | } |
| 571 | |
| 572 | /*! |
| 573 | Insert a new block based on stored state. |
| 574 | |
| 575 | m_cursor cannot store the state for the _next_ block ahead of time, because |
| 576 | m_cursor.setBlockFormat() controls the format of the block that the cursor |
| 577 | is already in; so cbLeaveBlock() cannot call setBlockFormat() without |
| 578 | altering the block that was just added. Therefore cbLeaveBlock() and the |
| 579 | following cbEnterBlock() set variables to remember what formatting should |
| 580 | come next, and insertBlock() is called just before the actual text |
| 581 | insertion, to create a new block with the right formatting. |
| 582 | */ |
| 583 | void QTextMarkdownImporter::insertBlock() |
| 584 | { |
| 585 | QTextCharFormat charFormat; |
| 586 | if (!m_spanFormatStack.isEmpty()) |
| 587 | charFormat = m_spanFormatStack.top(); |
| 588 | QTextBlockFormat blockFormat; |
| 589 | if (!m_listStack.isEmpty() && !m_needsInsertList && m_listItem) { |
| 590 | QTextList *list = m_listStack.top(); |
| 591 | if (list) |
| 592 | blockFormat = list->item(i: list->count() - 1).blockFormat(); |
| 593 | else |
| 594 | qWarning() << "attempted to insert into a list that no longer exists" ; |
| 595 | } |
| 596 | if (m_blockQuoteDepth) { |
| 597 | blockFormat.setProperty(propertyId: QTextFormat::BlockQuoteLevel, value: m_blockQuoteDepth); |
| 598 | blockFormat.setLeftMargin(qtmi_BlockQuoteIndent * m_blockQuoteDepth); |
| 599 | blockFormat.setRightMargin(qtmi_BlockQuoteIndent); |
| 600 | } |
| 601 | if (m_codeBlock) { |
| 602 | blockFormat.setProperty(propertyId: QTextFormat::BlockCodeLanguage, value: m_blockCodeLanguage); |
| 603 | if (m_blockCodeFence) { |
| 604 | blockFormat.setNonBreakableLines(true); |
| 605 | blockFormat.setProperty(propertyId: QTextFormat::BlockCodeFence, value: QString(QLatin1Char(m_blockCodeFence))); |
| 606 | } |
| 607 | charFormat.setFont(font: m_monoFont); |
| 608 | } else { |
| 609 | blockFormat.setTopMargin(m_paragraphMargin); |
| 610 | blockFormat.setBottomMargin(m_paragraphMargin); |
| 611 | } |
| 612 | if (m_markerType == QTextBlockFormat::MarkerType::NoMarker) |
| 613 | blockFormat.clearProperty(propertyId: QTextFormat::BlockMarker); |
| 614 | else |
| 615 | blockFormat.setMarker(m_markerType); |
| 616 | if (!m_listStack.isEmpty()) |
| 617 | blockFormat.setIndent(m_listStack.size()); |
| 618 | if (m_cursor.document()->isEmpty()) { |
| 619 | m_cursor.setBlockFormat(blockFormat); |
| 620 | m_cursor.setCharFormat(charFormat); |
| 621 | } else if (m_listItem) { |
| 622 | m_cursor.insertBlock(format: blockFormat, charFormat: QTextCharFormat()); |
| 623 | m_cursor.setCharFormat(charFormat); |
| 624 | } else { |
| 625 | m_cursor.insertBlock(format: blockFormat, charFormat); |
| 626 | } |
| 627 | if (m_needsInsertList) { |
| 628 | m_listStack.push(t: m_cursor.createList(format: m_listFormat)); |
| 629 | } else if (!m_listStack.isEmpty() && m_listItem && m_listStack.top()) { |
| 630 | m_listStack.top()->add(block: m_cursor.block()); |
| 631 | } |
| 632 | m_needsInsertList = false; |
| 633 | m_needsInsertBlock = false; |
| 634 | } |
| 635 | |
| 636 | QT_END_NAMESPACE |
| 637 | |