https://www.qt.io/blog/security-advisory-qtextmarkdownimporter-hbo https://codereview.qt-project.org/c/qt/qtbase/+/635699 --- a/src/gui/text/qtextmarkdownimporter.cpp +++ b/src/gui/text/qtextmarkdownimporter.cpp @@ -28,5 +28,6 @@ static const QChar qtmi_Space = u' '; -static constexpr auto markerString() noexcept { return "---"_L1; } +static constexpr auto lfMarkerString() noexcept { return "---\n"_L1; } +static constexpr auto crlfMarkerString() noexcept { return "---r\n"_L1; } // TODO maybe eliminate the margins after all views recognize BlockQuoteLevel, CSS can format it, etc. @@ -120,4 +121,45 @@ } +/*! \internal + Split any Front Matter from the Markdown document \a md. + Returns a pair of QStringViews: if \a md begins with qualifying Front Matter + (according to the specification at https://jekyllrb.com/docs/front-matter/ ), + put it into the \c frontMatter view, omitting both markers; and put the remaining + Markdown into \c rest. If no Front Matter is found, return all of \a md in \c rest. +*/ +static auto splitFrontMatter(QStringView md) +{ + struct R { + QStringView frontMatter, rest; + explicit operator bool() const noexcept { return !frontMatter.isEmpty(); } + }; + + const auto NotFound = R{{}, md}; + + /* Front Matter must start with '---\n' or '---\r\n' on the very first line, + and Front Matter must end with another such line. + If that is not the case, we return NotFound: then the whole document is + to be passed on to the Markdown parser, in which '---\n' is interpreted + as a "thematic break" (like
in HTML). */ + QLatin1StringView marker; + if (md.startsWith(lfMarkerString())) + marker = lfMarkerString(); + else if (md.startsWith(crlfMarkerString())) + marker = crlfMarkerString(); + else + return NotFound; + + const auto frontMatterStart = marker.size(); + const auto endMarkerPos = md.indexOf(marker, frontMatterStart); + + if (endMarkerPos < 0 || md[endMarkerPos - 1] != QChar::LineFeed) + return NotFound; + + Q_ASSERT(frontMatterStart < md.size()); + Q_ASSERT(endMarkerPos < md.size()); + const auto frontMatter = md.sliced(frontMatterStart, endMarkerPos - frontMatterStart); + return R{frontMatter, md.sliced(endMarkerPos + marker.size())}; +} + void QTextMarkdownImporter::import(const QString &markdown) { @@ -144,19 +186,12 @@ QStringView md = markdown; - if (m_features.testFlag(QTextMarkdownImporter::FeatureFrontMatter) && md.startsWith(markerString())) { - qsizetype endMarkerPos = md.indexOf(markerString(), markerString().size() + 1); - if (endMarkerPos > 4) { - qsizetype firstLinePos = 4; // first line of yaml - while (md.at(firstLinePos) == '\n'_L1 || md.at(firstLinePos) == '\r'_L1) - ++firstLinePos; - auto frontMatter = md.sliced(firstLinePos, endMarkerPos - firstLinePos); - firstLinePos = endMarkerPos + 4; // first line of markdown after yaml - while (md.size() > firstLinePos && (md.at(firstLinePos) == '\n'_L1 || md.at(firstLinePos) == '\r'_L1)) - ++firstLinePos; - md = md.sliced(firstLinePos); - doc->setMetaInformation(QTextDocument::FrontMatter, frontMatter.toString()); - qCDebug(lcMD) << "extracted FrontMatter: size" << frontMatter.size(); + if (m_features.testFlag(QTextMarkdownImporter::FeatureFrontMatter)) { + if (const auto split = splitFrontMatter(md)) { + doc->setMetaInformation(QTextDocument::FrontMatter, split.frontMatter.toString()); + qCDebug(lcMD) << "extracted FrontMatter: size" << split.frontMatter.size(); + md = split.rest; } } + const auto mdUtf8 = md.toUtf8(); m_cursor.beginEditBlock(); --- a/tests/auto/gui/text/qtextmarkdownimporter/data/front-marker-malformed1.md +++ b/tests/auto/gui/text/qtextmarkdownimporter/data/front-marker-malformed1.md @@ -0,0 +1,3 @@ +--- +name: "Pluto"--- +Pluto may not be a planet. And this document does not contain Front Matter. --- a/tests/auto/gui/text/qtextmarkdownimporter/data/front-marker-malformed2.md +++ b/tests/auto/gui/text/qtextmarkdownimporter/data/front-marker-malformed2.md @@ -0,0 +1,5 @@ +--- +name: "Sloppy" +--- +This document has trailing whitespace after its second Front Matter marker. +Therefore the marker does not qualify, and the document does not have Front Matter. --- a/tests/auto/gui/text/qtextmarkdownimporter/data/front-marker-malformed3.md +++ b/tests/auto/gui/text/qtextmarkdownimporter/data/front-marker-malformed3.md @@ -0,0 +1,4 @@ +--- +name: "Aborted YAML" +description: "The ending marker does not end with a newline, so it's invalid." +--- \ No newline at end of file --- a/tests/auto/gui/text/qtextmarkdownimporter/data/oss-fuzz-42533775.md +++ b/tests/auto/gui/text/qtextmarkdownimporter/data/oss-fuzz-42533775.md @@ -0,0 +1 @@ +--- --- \ No newline at end of file --- a/tests/auto/gui/text/qtextmarkdownimporter/data/yaml-crlf.md +++ b/tests/auto/gui/text/qtextmarkdownimporter/data/yaml-crlf.md @@ -0,0 +1,10 @@ +--- +name: "Venus" +discoverer: "Galileo Galilei" +title: "A description of the planet Venus" +keywords: + - planets + - solar system + - astronomy +--- +*Venus* is the second planet from the Sun, orbiting it every 224.7 Earth days. --- a/tests/auto/gui/text/qtextmarkdownimporter/tst_qtextmarkdownimporter.cpp +++ b/tests/auto/gui/text/qtextmarkdownimporter/tst_qtextmarkdownimporter.cpp @@ -549,4 +549,5 @@ QTest::newRow("fuzz20450") << "attempted to insert into a list that no longer exists"; QTest::newRow("fuzz20580") << ""; + QTest::newRow("oss-fuzz-42533775") << ""; // caused a heap-buffer-overflow } @@ -645,8 +646,13 @@ { QTest::addColumn("inputFile"); + QTest::addColumn("expectedFrontMatterSize"); QTest::addColumn("expectedBlockCount"); - QTest::newRow("yaml + markdown") << QFINDTESTDATA("data/yaml.md") << 1; - QTest::newRow("yaml only") << QFINDTESTDATA("data/yaml-only.md") << 0; + QTest::newRow("yaml + markdown") << QFINDTESTDATA("data/yaml.md") << 140 << 1; + QTest::newRow("yaml + markdown with CRLFs") << QFINDTESTDATA("data/yaml-crlf.md") << 140 << 1; + QTest::newRow("yaml only") << QFINDTESTDATA("data/yaml-only.md") << 59 << 0; + QTest::newRow("malformed 1") << QFINDTESTDATA("data/front-marker-malformed1.md") << 0 << 1; + QTest::newRow("malformed 2") << QFINDTESTDATA("data/front-marker-malformed2.md") << 0 << 2; + QTest::newRow("malformed 3") << QFINDTESTDATA("data/front-marker-malformed3.md") << 0 << 1; } @@ -654,4 +660,5 @@ { QFETCH(QString, inputFile); + QFETCH(int, expectedFrontMatterSize); QFETCH(int, expectedBlockCount); @@ -673,5 +680,7 @@ } QCOMPARE(blockCount, expectedBlockCount); // yaml is not part of the markdown text - QCOMPARE(doc.metaInformation(QTextDocument::FrontMatter), yaml); // without fences + if (expectedFrontMatterSize) + QCOMPARE(doc.metaInformation(QTextDocument::FrontMatter), yaml); // without fences + QCOMPARE(doc.metaInformation(QTextDocument::FrontMatter).size(), expectedFrontMatterSize); }