/src/qtbase/src/gui/text/qtextmarkdownimporter.cpp
Line | Count | Source (jump to first uncovered line) |
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 | | // Qt-Security score:critical reason:data-parser |
4 | | |
5 | | #include "qtextmarkdownimporter_p.h" |
6 | | #include "qtextdocumentfragment_p.h" |
7 | | #include <QLoggingCategory> |
8 | | #if QT_CONFIG(regularexpression) |
9 | | #include <QRegularExpression> |
10 | | #endif |
11 | | #include <QTextCursor> |
12 | | #include <QTextDocument> |
13 | | #include <QTextDocumentFragment> |
14 | | #include <QTextList> |
15 | | #include <QTextTable> |
16 | | #if QT_CONFIG(system_textmarkdownreader) |
17 | | #include <md4c.h> |
18 | | #else |
19 | | #include "../../3rdparty/md4c/md4c.h" |
20 | | #endif |
21 | | |
22 | | QT_BEGIN_NAMESPACE |
23 | | |
24 | | using namespace Qt::StringLiterals; |
25 | | |
26 | | Q_STATIC_LOGGING_CATEGORY(lcMD, "qt.text.markdown") |
27 | | |
28 | | static const QChar qtmi_Newline = u'\n'; |
29 | | static const QChar qtmi_Space = u' '; |
30 | | |
31 | 0 | static constexpr auto lfMarkerString() noexcept { return "---\n"_L1; } |
32 | 0 | static constexpr auto crlfMarkerString() noexcept { return "---\r\n"_L1; } |
33 | | |
34 | | // TODO maybe eliminate the margins after all views recognize BlockQuoteLevel, CSS can format it, etc. |
35 | | static const int qtmi_BlockQuoteIndent = |
36 | | 40; // pixels, same as in QTextHtmlParserNode::initializeProperties |
37 | | |
38 | | static_assert(int(QTextMarkdownImporter::FeatureCollapseWhitespace) == MD_FLAG_COLLAPSEWHITESPACE); |
39 | | static_assert(int(QTextMarkdownImporter::FeaturePermissiveATXHeaders) == MD_FLAG_PERMISSIVEATXHEADERS); |
40 | | static_assert(int(QTextMarkdownImporter::FeaturePermissiveURLAutoLinks) == MD_FLAG_PERMISSIVEURLAUTOLINKS); |
41 | | static_assert(int(QTextMarkdownImporter::FeaturePermissiveMailAutoLinks) == MD_FLAG_PERMISSIVEEMAILAUTOLINKS); |
42 | | static_assert(int(QTextMarkdownImporter::FeatureNoIndentedCodeBlocks) == MD_FLAG_NOINDENTEDCODEBLOCKS); |
43 | | static_assert(int(QTextMarkdownImporter::FeatureNoHTMLBlocks) == MD_FLAG_NOHTMLBLOCKS); |
44 | | static_assert(int(QTextMarkdownImporter::FeatureNoHTMLSpans) == MD_FLAG_NOHTMLSPANS); |
45 | | static_assert(int(QTextMarkdownImporter::FeatureTables) == MD_FLAG_TABLES); |
46 | | static_assert(int(QTextMarkdownImporter::FeatureStrikeThrough) == MD_FLAG_STRIKETHROUGH); |
47 | | static_assert(int(QTextMarkdownImporter::FeatureUnderline) == MD_FLAG_UNDERLINE); |
48 | | static_assert(int(QTextMarkdownImporter::FeaturePermissiveWWWAutoLinks) == MD_FLAG_PERMISSIVEWWWAUTOLINKS); |
49 | | static_assert(int(QTextMarkdownImporter::FeaturePermissiveAutoLinks) == MD_FLAG_PERMISSIVEAUTOLINKS); |
50 | | static_assert(int(QTextMarkdownImporter::FeatureTasklists) == MD_FLAG_TASKLISTS); |
51 | | static_assert(int(QTextMarkdownImporter::FeatureNoHTML) == MD_FLAG_NOHTML); |
52 | | static_assert(int(QTextMarkdownImporter::DialectCommonMark) == MD_DIALECT_COMMONMARK); |
53 | | static_assert(int(QTextMarkdownImporter::DialectGitHub) == |
54 | | (MD_DIALECT_GITHUB | MD_FLAG_UNDERLINE | QTextMarkdownImporter::FeatureFrontMatter)); |
55 | | |
56 | | // -------------------------------------------------------- |
57 | | // MD4C callback function wrappers |
58 | | |
59 | | static int CbEnterBlock(MD_BLOCKTYPE type, void *detail, void *userdata) |
60 | 0 | { |
61 | 0 | QTextMarkdownImporter *mdi = static_cast<QTextMarkdownImporter *>(userdata); |
62 | 0 | return mdi->cbEnterBlock(int(type), detail); |
63 | 0 | } |
64 | | |
65 | | static int CbLeaveBlock(MD_BLOCKTYPE type, void *detail, void *userdata) |
66 | 0 | { |
67 | 0 | QTextMarkdownImporter *mdi = static_cast<QTextMarkdownImporter *>(userdata); |
68 | 0 | return mdi->cbLeaveBlock(int(type), detail); |
69 | 0 | } |
70 | | |
71 | | static int CbEnterSpan(MD_SPANTYPE type, void *detail, void *userdata) |
72 | 0 | { |
73 | 0 | QTextMarkdownImporter *mdi = static_cast<QTextMarkdownImporter *>(userdata); |
74 | 0 | return mdi->cbEnterSpan(int(type), detail); |
75 | 0 | } |
76 | | |
77 | | static int CbLeaveSpan(MD_SPANTYPE type, void *detail, void *userdata) |
78 | 0 | { |
79 | 0 | QTextMarkdownImporter *mdi = static_cast<QTextMarkdownImporter *>(userdata); |
80 | 0 | return mdi->cbLeaveSpan(int(type), detail); |
81 | 0 | } |
82 | | |
83 | | static int CbText(MD_TEXTTYPE type, const MD_CHAR *text, MD_SIZE size, void *userdata) |
84 | 0 | { |
85 | 0 | QTextMarkdownImporter *mdi = static_cast<QTextMarkdownImporter *>(userdata); |
86 | 0 | return mdi->cbText(int(type), text, size); |
87 | 0 | } |
88 | | |
89 | | static void CbDebugLog(const char *msg, void *userdata) |
90 | 0 | { |
91 | 0 | Q_UNUSED(userdata); |
92 | 0 | qCDebug(lcMD) << msg; |
93 | 0 | } |
94 | | |
95 | | // MD4C callback function wrappers |
96 | | // -------------------------------------------------------- |
97 | | |
98 | | static Qt::Alignment MdAlignment(MD_ALIGN a, Qt::Alignment defaultAlignment = Qt::AlignLeft | Qt::AlignVCenter) |
99 | 0 | { |
100 | 0 | switch (a) { |
101 | 0 | case MD_ALIGN_LEFT: |
102 | 0 | return Qt::AlignLeft | Qt::AlignVCenter; |
103 | 0 | case MD_ALIGN_CENTER: |
104 | 0 | return Qt::AlignHCenter | Qt::AlignVCenter; |
105 | 0 | case MD_ALIGN_RIGHT: |
106 | 0 | return Qt::AlignRight | Qt::AlignVCenter; |
107 | 0 | default: // including MD_ALIGN_DEFAULT |
108 | 0 | return defaultAlignment; |
109 | 0 | } |
110 | 0 | } |
111 | | |
112 | | QTextMarkdownImporter::QTextMarkdownImporter(QTextDocument *doc, QTextMarkdownImporter::Features features) |
113 | 0 | : m_cursor(doc) |
114 | 0 | , m_monoFont(QFontDatabase::systemFont(QFontDatabase::FixedFont)) |
115 | 0 | , m_features(features) |
116 | 0 | { |
117 | 0 | } |
118 | | |
119 | | QTextMarkdownImporter::QTextMarkdownImporter(QTextDocument *doc, QTextDocument::MarkdownFeatures features) |
120 | 0 | : QTextMarkdownImporter(doc, static_cast<QTextMarkdownImporter::Features>(int(features))) |
121 | 0 | { |
122 | 0 | } |
123 | | |
124 | | /*! \internal |
125 | | Split any Front Matter from the Markdown document \a md. |
126 | | Returns a pair of QStringViews: if \a md begins with qualifying Front Matter |
127 | | (according to the specification at https://jekyllrb.com/docs/front-matter/ ), |
128 | | put it into the \c frontMatter view, omitting both markers; and put the remaining |
129 | | Markdown into \c rest. If no Front Matter is found, return all of \a md in \c rest. |
130 | | */ |
131 | | static auto splitFrontMatter(QStringView md) |
132 | 0 | { |
133 | 0 | struct R { |
134 | 0 | QStringView frontMatter, rest; |
135 | 0 | explicit operator bool() const noexcept { return !frontMatter.isEmpty(); } |
136 | 0 | }; |
137 | |
|
138 | 0 | const auto NotFound = R{{}, md}; |
139 | | |
140 | | /* Front Matter must start with '---\n' or '---\r\n' on the very first line, |
141 | | and Front Matter must end with another such line. |
142 | | If that is not the case, we return NotFound: then the whole document is |
143 | | to be passed on to the Markdown parser, in which '---\n' is interpreted |
144 | | as a "thematic break" (like <hr/> in HTML). */ |
145 | 0 | QLatin1StringView marker; |
146 | 0 | if (md.startsWith(lfMarkerString())) |
147 | 0 | marker = lfMarkerString(); |
148 | 0 | else if (md.startsWith(crlfMarkerString())) |
149 | 0 | marker = crlfMarkerString(); |
150 | 0 | else |
151 | 0 | return NotFound; |
152 | | |
153 | 0 | const auto frontMatterStart = marker.size(); |
154 | 0 | const auto endMarkerPos = md.indexOf(marker, frontMatterStart); |
155 | |
|
156 | 0 | if (endMarkerPos < 0 || md[endMarkerPos - 1] != QChar::LineFeed) |
157 | 0 | return NotFound; |
158 | | |
159 | 0 | Q_ASSERT(frontMatterStart < md.size()); |
160 | 0 | Q_ASSERT(endMarkerPos < md.size()); |
161 | 0 | const auto frontMatter = md.sliced(frontMatterStart, endMarkerPos - frontMatterStart); |
162 | 0 | return R{frontMatter, md.sliced(endMarkerPos + marker.size())}; |
163 | 0 | } |
164 | | |
165 | | void QTextMarkdownImporter::import(const QString &markdown) |
166 | 0 | { |
167 | 0 | MD_PARSER callbacks = { |
168 | 0 | 0, // abi_version |
169 | 0 | unsigned(m_features), |
170 | 0 | &CbEnterBlock, |
171 | 0 | &CbLeaveBlock, |
172 | 0 | &CbEnterSpan, |
173 | 0 | &CbLeaveSpan, |
174 | 0 | &CbText, |
175 | 0 | &CbDebugLog, |
176 | 0 | nullptr // syntax |
177 | 0 | }; |
178 | 0 | QTextDocument *doc = m_cursor.document(); |
179 | 0 | const auto defaultFont = doc->defaultFont(); |
180 | 0 | m_paragraphMargin = defaultFont.pointSize() * 2 / 3; |
181 | 0 | doc->clear(); |
182 | 0 | if (defaultFont.pointSize() != -1) |
183 | 0 | m_monoFont.setPointSize(defaultFont.pointSize()); |
184 | 0 | else |
185 | 0 | m_monoFont.setPixelSize(defaultFont.pixelSize()); |
186 | 0 | qCDebug(lcMD) << "default font" << defaultFont << "mono font" << m_monoFont; |
187 | 0 | QStringView md = markdown; |
188 | |
|
189 | 0 | if (m_features.testFlag(QTextMarkdownImporter::FeatureFrontMatter)) { |
190 | 0 | if (const auto split = splitFrontMatter(md)) { |
191 | 0 | doc->setMetaInformation(QTextDocument::FrontMatter, split.frontMatter.toString()); |
192 | 0 | qCDebug(lcMD) << "extracted FrontMatter: size" << split.frontMatter.size(); |
193 | 0 | md = split.rest; |
194 | 0 | } |
195 | 0 | } |
196 | |
|
197 | 0 | const auto mdUtf8 = md.toUtf8(); |
198 | 0 | m_cursor.beginEditBlock(); |
199 | 0 | md_parse(mdUtf8.constData(), MD_SIZE(mdUtf8.size()), &callbacks, this); |
200 | 0 | m_cursor.endEditBlock(); |
201 | 0 | } |
202 | | |
203 | | int QTextMarkdownImporter::cbEnterBlock(int blockType, void *det) |
204 | 0 | { |
205 | 0 | m_blockType = blockType; |
206 | 0 | switch (blockType) { |
207 | 0 | case MD_BLOCK_P: |
208 | 0 | if (!m_listStack.isEmpty()) |
209 | 0 | qCDebug(lcMD, m_listItem ? "P of LI at level %d" : "P continuation inside LI at level %d", int(m_listStack.size())); |
210 | 0 | else |
211 | 0 | qCDebug(lcMD, "P"); |
212 | 0 | m_needsInsertBlock = true; |
213 | 0 | break; |
214 | 0 | case MD_BLOCK_QUOTE: |
215 | 0 | ++m_blockQuoteDepth; |
216 | 0 | qCDebug(lcMD, "QUOTE level %d", m_blockQuoteDepth); |
217 | 0 | break; |
218 | 0 | case MD_BLOCK_CODE: { |
219 | 0 | MD_BLOCK_CODE_DETAIL *detail = static_cast<MD_BLOCK_CODE_DETAIL *>(det); |
220 | 0 | m_codeBlock = true; |
221 | 0 | m_blockCodeLanguage = QLatin1StringView(detail->lang.text, int(detail->lang.size)); |
222 | 0 | m_blockCodeFence = detail->fence_char; |
223 | 0 | QString info = QLatin1StringView(detail->info.text, int(detail->info.size)); |
224 | 0 | m_needsInsertBlock = true; |
225 | 0 | if (m_blockQuoteDepth) |
226 | 0 | qCDebug(lcMD, "CODE lang '%s' info '%s' fenced with '%c' inside QUOTE %d", qPrintable(m_blockCodeLanguage), qPrintable(info), m_blockCodeFence, m_blockQuoteDepth); |
227 | 0 | else |
228 | 0 | qCDebug(lcMD, "CODE lang '%s' info '%s' fenced with '%c'", qPrintable(m_blockCodeLanguage), qPrintable(info), m_blockCodeFence); |
229 | 0 | } break; |
230 | 0 | case MD_BLOCK_H: { |
231 | 0 | MD_BLOCK_H_DETAIL *detail = static_cast<MD_BLOCK_H_DETAIL *>(det); |
232 | 0 | QTextBlockFormat blockFmt; |
233 | 0 | QTextCharFormat charFmt; |
234 | 0 | int sizeAdjustment = 4 - int(detail->level); // H1 to H6: +3 to -2 |
235 | 0 | charFmt.setProperty(QTextFormat::FontSizeAdjustment, sizeAdjustment); |
236 | 0 | charFmt.setFontWeight(QFont::Bold); |
237 | 0 | blockFmt.setHeadingLevel(int(detail->level)); |
238 | 0 | m_needsInsertBlock = false; |
239 | 0 | if (m_cursor.document()->isEmpty()) { |
240 | 0 | m_cursor.setBlockFormat(blockFmt); |
241 | 0 | m_cursor.setCharFormat(charFmt); |
242 | 0 | } else { |
243 | 0 | m_cursor.insertBlock(blockFmt, charFmt); |
244 | 0 | } |
245 | 0 | qCDebug(lcMD, "H%d", detail->level); |
246 | 0 | } break; |
247 | 0 | case MD_BLOCK_LI: { |
248 | 0 | m_needsInsertBlock = true; |
249 | 0 | m_listItem = true; |
250 | 0 | MD_BLOCK_LI_DETAIL *detail = static_cast<MD_BLOCK_LI_DETAIL *>(det); |
251 | 0 | m_markerType = detail->is_task ? |
252 | 0 | (detail->task_mark == ' ' ? QTextBlockFormat::MarkerType::Unchecked : QTextBlockFormat::MarkerType::Checked) : |
253 | 0 | QTextBlockFormat::MarkerType::NoMarker; |
254 | 0 | qCDebug(lcMD) << "LI"; |
255 | 0 | } break; |
256 | 0 | case MD_BLOCK_UL: { |
257 | 0 | if (m_needsInsertList) // list nested in an empty list |
258 | 0 | m_listStack.push(m_cursor.insertList(m_listFormat)); |
259 | 0 | else |
260 | 0 | m_needsInsertList = true; |
261 | 0 | MD_BLOCK_UL_DETAIL *detail = static_cast<MD_BLOCK_UL_DETAIL *>(det); |
262 | 0 | m_listFormat = QTextListFormat(); |
263 | 0 | m_listFormat.setIndent(m_listStack.size() + 1); |
264 | 0 | switch (detail->mark) { |
265 | 0 | case '*': |
266 | 0 | m_listFormat.setStyle(QTextListFormat::ListCircle); |
267 | 0 | break; |
268 | 0 | case '+': |
269 | 0 | m_listFormat.setStyle(QTextListFormat::ListSquare); |
270 | 0 | break; |
271 | 0 | default: // including '-' |
272 | 0 | m_listFormat.setStyle(QTextListFormat::ListDisc); |
273 | 0 | break; |
274 | 0 | } |
275 | 0 | qCDebug(lcMD, "UL %c level %d", detail->mark, int(m_listStack.size()) + 1); |
276 | 0 | } break; |
277 | 0 | case MD_BLOCK_OL: { |
278 | 0 | if (m_needsInsertList) // list nested in an empty list |
279 | 0 | m_listStack.push(m_cursor.insertList(m_listFormat)); |
280 | 0 | else |
281 | 0 | m_needsInsertList = true; |
282 | 0 | MD_BLOCK_OL_DETAIL *detail = static_cast<MD_BLOCK_OL_DETAIL *>(det); |
283 | 0 | m_listFormat = QTextListFormat(); |
284 | 0 | m_listFormat.setIndent(m_listStack.size() + 1); |
285 | 0 | m_listFormat.setNumberSuffix(QChar::fromLatin1(detail->mark_delimiter)); |
286 | 0 | m_listFormat.setStyle(QTextListFormat::ListDecimal); |
287 | 0 | m_listFormat.setStart(detail->start); |
288 | 0 | qCDebug(lcMD, "OL xx%d level %d start %d", detail->mark_delimiter, int(m_listStack.size()) + 1, detail->start); |
289 | 0 | } break; |
290 | 0 | case MD_BLOCK_TD: { |
291 | 0 | MD_BLOCK_TD_DETAIL *detail = static_cast<MD_BLOCK_TD_DETAIL *>(det); |
292 | 0 | ++m_tableCol; |
293 | | // absolute movement (and storage of m_tableCol) shouldn't be necessary, but |
294 | | // movePosition(QTextCursor::NextCell) doesn't work |
295 | 0 | QTextTableCell cell = m_currentTable->cellAt(m_tableRowCount - 1, m_tableCol); |
296 | 0 | if (!cell.isValid()) { |
297 | 0 | qWarning("malformed table in Markdown input"); |
298 | 0 | return 1; |
299 | 0 | } |
300 | 0 | m_cursor = cell.firstCursorPosition(); |
301 | 0 | QTextBlockFormat blockFmt = m_cursor.blockFormat(); |
302 | 0 | blockFmt.setAlignment(MdAlignment(detail->align)); |
303 | 0 | m_cursor.setBlockFormat(blockFmt); |
304 | 0 | qCDebug(lcMD) << "TD; align" << detail->align << MdAlignment(detail->align) << "col" << m_tableCol; |
305 | 0 | } break; |
306 | 0 | case MD_BLOCK_TH: { |
307 | 0 | ++m_tableColumnCount; |
308 | 0 | ++m_tableCol; |
309 | 0 | if (m_currentTable->columns() < m_tableColumnCount) |
310 | 0 | m_currentTable->appendColumns(1); |
311 | 0 | auto cell = m_currentTable->cellAt(m_tableRowCount - 1, m_tableCol); |
312 | 0 | if (!cell.isValid()) { |
313 | 0 | qWarning("malformed table in Markdown input"); |
314 | 0 | return 1; |
315 | 0 | } |
316 | 0 | auto fmt = cell.format(); |
317 | 0 | fmt.setFontWeight(QFont::Bold); |
318 | 0 | cell.setFormat(fmt); |
319 | 0 | } break; |
320 | 0 | case MD_BLOCK_TR: { |
321 | 0 | ++m_tableRowCount; |
322 | 0 | m_nonEmptyTableCells.clear(); |
323 | 0 | if (m_currentTable->rows() < m_tableRowCount) |
324 | 0 | m_currentTable->appendRows(1); |
325 | 0 | m_tableCol = -1; |
326 | 0 | qCDebug(lcMD) << "TR" << m_currentTable->rows(); |
327 | 0 | } break; |
328 | 0 | case MD_BLOCK_TABLE: |
329 | 0 | m_tableColumnCount = 0; |
330 | 0 | m_tableRowCount = 0; |
331 | 0 | m_currentTable = m_cursor.insertTable(1, 1); // we don't know the dimensions yet |
332 | 0 | break; |
333 | 0 | case MD_BLOCK_HR: { |
334 | 0 | qCDebug(lcMD, "HR"); |
335 | 0 | QTextBlockFormat blockFmt; |
336 | 0 | blockFmt.setProperty(QTextFormat::BlockTrailingHorizontalRulerWidth, 1); |
337 | 0 | m_cursor.insertBlock(blockFmt, QTextCharFormat()); |
338 | 0 | } break; |
339 | 0 | default: |
340 | 0 | break; // nothing to do for now |
341 | 0 | } |
342 | 0 | return 0; // no error |
343 | 0 | } |
344 | | |
345 | | int QTextMarkdownImporter::cbLeaveBlock(int blockType, void *detail) |
346 | 0 | { |
347 | 0 | Q_UNUSED(detail); |
348 | 0 | switch (blockType) { |
349 | 0 | case MD_BLOCK_P: |
350 | 0 | m_listItem = false; |
351 | 0 | break; |
352 | 0 | case MD_BLOCK_UL: |
353 | 0 | case MD_BLOCK_OL: |
354 | 0 | if (Q_UNLIKELY(m_needsInsertList)) |
355 | 0 | m_listStack.push(m_cursor.createList(m_listFormat)); |
356 | 0 | if (Q_UNLIKELY(m_listStack.isEmpty())) { |
357 | 0 | qCWarning(lcMD, "list ended unexpectedly"); |
358 | 0 | } else { |
359 | 0 | qCDebug(lcMD, "list at level %d ended", int(m_listStack.size())); |
360 | 0 | m_listStack.pop(); |
361 | 0 | } |
362 | 0 | break; |
363 | 0 | case MD_BLOCK_TR: { |
364 | | // https://github.com/mity/md4c/issues/29 |
365 | | // MD4C doesn't tell us explicitly which cells are merged, so merge empty cells |
366 | | // with previous non-empty ones |
367 | 0 | int mergeEnd = -1; |
368 | 0 | int mergeBegin = -1; |
369 | 0 | for (int col = m_tableCol; col >= 0; --col) { |
370 | 0 | if (m_nonEmptyTableCells.contains(col)) { |
371 | 0 | if (mergeEnd >= 0 && mergeBegin >= 0) { |
372 | 0 | qCDebug(lcMD) << "merging cells" << mergeBegin << "to" << mergeEnd << "inclusive, on row" << m_currentTable->rows() - 1; |
373 | 0 | m_currentTable->mergeCells(m_currentTable->rows() - 1, mergeBegin - 1, 1, mergeEnd - mergeBegin + 2); |
374 | 0 | } |
375 | 0 | mergeEnd = -1; |
376 | 0 | mergeBegin = -1; |
377 | 0 | } else { |
378 | 0 | if (mergeEnd < 0) |
379 | 0 | mergeEnd = col; |
380 | 0 | else |
381 | 0 | mergeBegin = col; |
382 | 0 | } |
383 | 0 | } |
384 | 0 | } break; |
385 | 0 | case MD_BLOCK_QUOTE: { |
386 | 0 | qCDebug(lcMD, "QUOTE level %d ended", m_blockQuoteDepth); |
387 | 0 | --m_blockQuoteDepth; |
388 | 0 | m_needsInsertBlock = true; |
389 | 0 | } break; |
390 | 0 | case MD_BLOCK_TABLE: |
391 | 0 | qCDebug(lcMD) << "table ended with" << m_currentTable->columns() << "cols and" << m_currentTable->rows() << "rows"; |
392 | 0 | m_currentTable = nullptr; |
393 | 0 | m_cursor.movePosition(QTextCursor::End); |
394 | 0 | break; |
395 | 0 | case MD_BLOCK_LI: |
396 | 0 | qCDebug(lcMD, "LI at level %d ended", int(m_listStack.size())); |
397 | 0 | m_listItem = false; |
398 | 0 | break; |
399 | 0 | case MD_BLOCK_CODE: { |
400 | 0 | m_codeBlock = false; |
401 | 0 | m_blockCodeLanguage.clear(); |
402 | 0 | m_blockCodeFence = 0; |
403 | 0 | if (m_blockQuoteDepth) |
404 | 0 | qCDebug(lcMD, "CODE ended inside QUOTE %d", m_blockQuoteDepth); |
405 | 0 | else |
406 | 0 | qCDebug(lcMD, "CODE ended"); |
407 | 0 | m_needsInsertBlock = true; |
408 | 0 | } break; |
409 | 0 | case MD_BLOCK_H: |
410 | 0 | m_cursor.setCharFormat(QTextCharFormat()); |
411 | 0 | break; |
412 | 0 | default: |
413 | 0 | break; |
414 | 0 | } |
415 | 0 | return 0; // no error |
416 | 0 | } |
417 | | |
418 | | int QTextMarkdownImporter::cbEnterSpan(int spanType, void *det) |
419 | 0 | { |
420 | 0 | QTextCharFormat charFmt; |
421 | 0 | if (!m_spanFormatStack.isEmpty()) |
422 | 0 | charFmt = m_spanFormatStack.top(); |
423 | 0 | switch (spanType) { |
424 | 0 | case MD_SPAN_EM: |
425 | 0 | charFmt.setFontItalic(true); |
426 | 0 | break; |
427 | 0 | case MD_SPAN_STRONG: |
428 | 0 | charFmt.setFontWeight(QFont::Bold); |
429 | 0 | break; |
430 | 0 | case MD_SPAN_U: |
431 | 0 | charFmt.setFontUnderline(true); |
432 | 0 | break; |
433 | 0 | case MD_SPAN_A: { |
434 | 0 | MD_SPAN_A_DETAIL *detail = static_cast<MD_SPAN_A_DETAIL *>(det); |
435 | 0 | QString url = QString::fromUtf8(detail->href.text, int(detail->href.size)); |
436 | 0 | QString title = QString::fromUtf8(detail->title.text, int(detail->title.size)); |
437 | 0 | charFmt.setAnchor(true); |
438 | 0 | charFmt.setAnchorHref(url); |
439 | 0 | if (!title.isEmpty()) |
440 | 0 | charFmt.setToolTip(title); |
441 | 0 | charFmt.setForeground(m_palette.link()); |
442 | 0 | qCDebug(lcMD) << "anchor" << url << title; |
443 | 0 | } break; |
444 | 0 | case MD_SPAN_IMG: { |
445 | 0 | m_imageSpan = true; |
446 | 0 | m_imageFormat = QTextImageFormat(); |
447 | 0 | MD_SPAN_IMG_DETAIL *detail = static_cast<MD_SPAN_IMG_DETAIL *>(det); |
448 | 0 | m_imageFormat.setName(QString::fromUtf8(detail->src.text, int(detail->src.size))); |
449 | 0 | m_imageFormat.setProperty(QTextFormat::ImageTitle, QString::fromUtf8(detail->title.text, int(detail->title.size))); |
450 | 0 | break; |
451 | 0 | } |
452 | 0 | case MD_SPAN_CODE: |
453 | 0 | charFmt.setFont(m_monoFont); |
454 | 0 | charFmt.setFontFixedPitch(true); |
455 | 0 | break; |
456 | 0 | case MD_SPAN_DEL: |
457 | 0 | charFmt.setFontStrikeOut(true); |
458 | 0 | break; |
459 | 0 | } |
460 | 0 | m_spanFormatStack.push(charFmt); |
461 | 0 | qCDebug(lcMD) << spanType << "setCharFormat" << charFmt.font().families().constFirst() |
462 | 0 | << charFmt.fontWeight() << (charFmt.fontItalic() ? "italic" : "") |
463 | 0 | << charFmt.foreground().color().name(); |
464 | 0 | m_cursor.setCharFormat(charFmt); |
465 | 0 | return 0; // no error |
466 | 0 | } |
467 | | |
468 | | int QTextMarkdownImporter::cbLeaveSpan(int spanType, void *detail) |
469 | 0 | { |
470 | 0 | Q_UNUSED(detail); |
471 | 0 | QTextCharFormat charFmt; |
472 | 0 | if (!m_spanFormatStack.isEmpty()) { |
473 | 0 | m_spanFormatStack.pop(); |
474 | 0 | if (!m_spanFormatStack.isEmpty()) |
475 | 0 | charFmt = m_spanFormatStack.top(); |
476 | 0 | } |
477 | 0 | m_cursor.setCharFormat(charFmt); |
478 | 0 | qCDebug(lcMD) << spanType << "setCharFormat" << charFmt.font().families().constFirst() |
479 | 0 | << charFmt.fontWeight() << (charFmt.fontItalic() ? "italic" : "") |
480 | 0 | << charFmt.foreground().color().name(); |
481 | 0 | if (spanType == int(MD_SPAN_IMG)) |
482 | 0 | m_imageSpan = false; |
483 | 0 | return 0; // no error |
484 | 0 | } |
485 | | |
486 | | int QTextMarkdownImporter::cbText(int textType, const char *text, unsigned size) |
487 | 0 | { |
488 | 0 | if (m_needsInsertBlock) |
489 | 0 | insertBlock(); |
490 | 0 | #if QT_CONFIG(regularexpression) |
491 | 0 | static const QRegularExpression openingBracket(QStringLiteral("<[a-zA-Z]")); |
492 | 0 | static const QRegularExpression closingBracket(QStringLiteral("(/>|</)")); |
493 | 0 | #endif |
494 | 0 | QString s = QString::fromUtf8(text, int(size)); |
495 | |
|
496 | 0 | switch (textType) { |
497 | 0 | case MD_TEXT_NORMAL: |
498 | 0 | #if QT_CONFIG(regularexpression) |
499 | 0 | if (m_htmlTagDepth) { |
500 | 0 | m_htmlAccumulator += s; |
501 | 0 | s = QString(); |
502 | 0 | } |
503 | 0 | #endif |
504 | 0 | break; |
505 | 0 | case MD_TEXT_NULLCHAR: |
506 | 0 | s = QString(QChar(u'\xFFFD')); // CommonMark-required replacement for null |
507 | 0 | break; |
508 | 0 | case MD_TEXT_BR: |
509 | 0 | s = QString(qtmi_Newline); |
510 | 0 | break; |
511 | 0 | case MD_TEXT_SOFTBR: |
512 | 0 | s = QString(qtmi_Space); |
513 | 0 | break; |
514 | 0 | case MD_TEXT_CODE: |
515 | | // We'll see MD_SPAN_CODE too, which will set the char format, and that's enough. |
516 | 0 | break; |
517 | 0 | #if QT_CONFIG(texthtmlparser) |
518 | 0 | case MD_TEXT_ENTITY: |
519 | 0 | if (m_htmlTagDepth) |
520 | 0 | m_htmlAccumulator += s; |
521 | 0 | else |
522 | 0 | m_cursor.insertHtml(s); |
523 | 0 | s = QString(); |
524 | 0 | break; |
525 | 0 | #endif |
526 | 0 | case MD_TEXT_HTML: |
527 | | // count how many tags are opened and how many are closed |
528 | 0 | #if QT_CONFIG(regularexpression) && QT_CONFIG(texthtmlparser) |
529 | 0 | { |
530 | 0 | int startIdx = 0; |
531 | 0 | while ((startIdx = s.indexOf(openingBracket, startIdx)) >= 0) { |
532 | 0 | ++m_htmlTagDepth; |
533 | 0 | startIdx += 2; |
534 | 0 | } |
535 | 0 | startIdx = 0; |
536 | 0 | while ((startIdx = s.indexOf(closingBracket, startIdx)) >= 0) { |
537 | 0 | --m_htmlTagDepth; |
538 | 0 | startIdx += 2; |
539 | 0 | } |
540 | 0 | } |
541 | 0 | m_htmlAccumulator += s; |
542 | 0 | if (!m_htmlTagDepth) { // all open tags are now closed |
543 | 0 | qCDebug(lcMD) << "HTML" << m_htmlAccumulator; |
544 | 0 | m_cursor.insertHtml(m_htmlAccumulator); |
545 | 0 | if (m_spanFormatStack.isEmpty()) |
546 | 0 | m_cursor.setCharFormat(QTextCharFormat()); |
547 | 0 | else |
548 | 0 | m_cursor.setCharFormat(m_spanFormatStack.top()); |
549 | 0 | m_htmlAccumulator = QString(); |
550 | 0 | } |
551 | 0 | #endif |
552 | 0 | s = QString(); |
553 | 0 | break; |
554 | 0 | } |
555 | | |
556 | 0 | switch (m_blockType) { |
557 | 0 | case MD_BLOCK_TD: |
558 | 0 | m_nonEmptyTableCells.append(m_tableCol); |
559 | 0 | break; |
560 | 0 | case MD_BLOCK_CODE: |
561 | 0 | if (s == qtmi_Newline) { |
562 | | // defer a blank line until we see something else in the code block, |
563 | | // to avoid ending every code block with a gratuitous blank line |
564 | 0 | m_needsInsertBlock = true; |
565 | 0 | s = QString(); |
566 | 0 | } |
567 | 0 | break; |
568 | 0 | default: |
569 | 0 | break; |
570 | 0 | } |
571 | | |
572 | 0 | if (m_imageSpan) { |
573 | | // TODO we don't yet support alt text with formatting, because of the cases where m_cursor |
574 | | // already inserted the text above. Rather need to accumulate it in case we need it here. |
575 | 0 | m_imageFormat.setProperty(QTextFormat::ImageAltText, s); |
576 | 0 | qCDebug(lcMD) << "image" << m_imageFormat.name() |
577 | 0 | << "title" << m_imageFormat.stringProperty(QTextFormat::ImageTitle) |
578 | 0 | << "alt" << s << "relative to" << m_cursor.document()->baseUrl(); |
579 | 0 | m_cursor.insertImage(m_imageFormat); |
580 | 0 | return 0; // no error |
581 | 0 | } |
582 | | |
583 | 0 | if (!s.isEmpty()) |
584 | 0 | m_cursor.insertText(s); |
585 | 0 | if (m_cursor.currentList()) { |
586 | | // The list item will indent the list item's text, so we don't need indentation on the block. |
587 | 0 | QTextBlockFormat bfmt = m_cursor.blockFormat(); |
588 | 0 | bfmt.setIndent(0); |
589 | 0 | m_cursor.setBlockFormat(bfmt); |
590 | 0 | } |
591 | 0 | if (lcMD().isEnabled(QtDebugMsg)) { |
592 | 0 | QTextBlockFormat bfmt = m_cursor.blockFormat(); |
593 | 0 | QString debugInfo; |
594 | 0 | if (m_cursor.currentList()) |
595 | 0 | debugInfo = "in list at depth "_L1 + QString::number(m_cursor.currentList()->format().indent()); |
596 | 0 | if (bfmt.hasProperty(QTextFormat::BlockQuoteLevel)) |
597 | 0 | debugInfo += "in blockquote at depth "_L1 + |
598 | 0 | QString::number(bfmt.intProperty(QTextFormat::BlockQuoteLevel)); |
599 | 0 | if (bfmt.hasProperty(QTextFormat::BlockCodeLanguage)) |
600 | 0 | debugInfo += "in a code block"_L1; |
601 | 0 | qCDebug(lcMD) << textType << "in block" << m_blockType << s << qPrintable(debugInfo) |
602 | 0 | << "bindent" << bfmt.indent() << "tindent" << bfmt.textIndent() |
603 | 0 | << "margins" << bfmt.leftMargin() << bfmt.topMargin() << bfmt.bottomMargin() << bfmt.rightMargin(); |
604 | 0 | } |
605 | 0 | return 0; // no error |
606 | 0 | } |
607 | | |
608 | | /*! |
609 | | Insert a new block based on stored state. |
610 | | |
611 | | m_cursor cannot store the state for the _next_ block ahead of time, because |
612 | | m_cursor.setBlockFormat() controls the format of the block that the cursor |
613 | | is already in; so cbLeaveBlock() cannot call setBlockFormat() without |
614 | | altering the block that was just added. Therefore cbLeaveBlock() and the |
615 | | following cbEnterBlock() set variables to remember what formatting should |
616 | | come next, and insertBlock() is called just before the actual text |
617 | | insertion, to create a new block with the right formatting. |
618 | | */ |
619 | | void QTextMarkdownImporter::insertBlock() |
620 | 0 | { |
621 | 0 | QTextCharFormat charFormat; |
622 | 0 | if (!m_spanFormatStack.isEmpty()) |
623 | 0 | charFormat = m_spanFormatStack.top(); |
624 | 0 | QTextBlockFormat blockFormat; |
625 | 0 | if (!m_listStack.isEmpty() && !m_needsInsertList && m_listItem) { |
626 | 0 | QTextList *list = m_listStack.top(); |
627 | 0 | if (list) |
628 | 0 | blockFormat = list->item(list->count() - 1).blockFormat(); |
629 | 0 | else |
630 | 0 | qWarning() << "attempted to insert into a list that no longer exists"; |
631 | 0 | } |
632 | 0 | if (m_blockQuoteDepth) { |
633 | 0 | blockFormat.setProperty(QTextFormat::BlockQuoteLevel, m_blockQuoteDepth); |
634 | 0 | blockFormat.setLeftMargin(qtmi_BlockQuoteIndent * m_blockQuoteDepth); |
635 | 0 | blockFormat.setRightMargin(qtmi_BlockQuoteIndent); |
636 | 0 | } |
637 | 0 | if (m_codeBlock) { |
638 | 0 | blockFormat.setProperty(QTextFormat::BlockCodeLanguage, m_blockCodeLanguage); |
639 | 0 | if (m_blockCodeFence) { |
640 | 0 | blockFormat.setNonBreakableLines(true); |
641 | 0 | blockFormat.setProperty(QTextFormat::BlockCodeFence, QString(QLatin1Char(m_blockCodeFence))); |
642 | 0 | } |
643 | 0 | charFormat.setFont(m_monoFont); |
644 | 0 | } else { |
645 | 0 | blockFormat.setTopMargin(m_paragraphMargin); |
646 | 0 | blockFormat.setBottomMargin(m_paragraphMargin); |
647 | 0 | } |
648 | 0 | if (m_markerType == QTextBlockFormat::MarkerType::NoMarker) |
649 | 0 | blockFormat.clearProperty(QTextFormat::BlockMarker); |
650 | 0 | else |
651 | 0 | blockFormat.setMarker(m_markerType); |
652 | 0 | if (!m_listStack.isEmpty()) |
653 | 0 | blockFormat.setIndent(m_listStack.size()); |
654 | 0 | if (m_cursor.document()->isEmpty()) { |
655 | 0 | m_cursor.setBlockFormat(blockFormat); |
656 | 0 | m_cursor.setCharFormat(charFormat); |
657 | 0 | } else if (m_listItem) { |
658 | 0 | m_cursor.insertBlock(blockFormat, QTextCharFormat()); |
659 | 0 | m_cursor.setCharFormat(charFormat); |
660 | 0 | } else { |
661 | 0 | m_cursor.insertBlock(blockFormat, charFormat); |
662 | 0 | } |
663 | 0 | if (m_needsInsertList) { |
664 | 0 | m_listStack.push(m_cursor.createList(m_listFormat)); |
665 | 0 | } else if (!m_listStack.isEmpty() && m_listItem && m_listStack.top()) { |
666 | 0 | m_listStack.top()->add(m_cursor.block()); |
667 | 0 | } |
668 | 0 | m_needsInsertList = false; |
669 | 0 | m_needsInsertBlock = false; |
670 | 0 | } |
671 | | |
672 | | QT_END_NAMESPACE |