QTextHtmlImporter: don't forget to appendBlock after block tag closed

If we see a closing tag that really demands a new block after it,
like </ul>, that needs to be done even if some ignorable whitespace
and "inline" tags come after it. Don't get distracted by those.

Also add a comment in QTextDocument::setHtml() to remind the reader that
HTML parsing is a two-pass algorithm.

Pick-to: 6.6 6.7
Fixes: QTBUG-81662
Change-Id: If723c9d3c211a684725055a06bcf87be4e38923a
Reviewed-by: Qt CI Bot <[email protected]>
Reviewed-by: Eskil Abrahamsen Blomfeldt <[email protected]>

3 files changed, 27 insertions, 1 deletions

diff --git a/src/gui/text/qtextdocument.cpp b/src/gui/text/qtextdocument.cpp
index fc172c5057f..b5dd8c2b830 100644
--- a/src/gui/text/qtextdocument.cpp
+++ b/src/gui/text/qtextdocument.cpp
@@ -1301,6 +1301,8 @@ void QTextDocument::setHtml(const QString &html)
     d->enableUndoRedo(false);
     d->beginEditBlock();
     d->clear();
+    // ctor calls parse() to build up QTextHtmlParser::nodes list
+    // then import() populates the QTextDocument from those
     QTextHtmlImporter(this, html, QTextHtmlImporter::ImportToDocument).import();
     d->endEditBlock();
     d->enableUndoRedo(previousState);
diff --git a/src/gui/text/qtextdocumentfragment.cpp b/src/gui/text/qtextdocumentfragment.cpp
index 03d2afc0d1b..c663ac884c3 100644
--- a/src/gui/text/qtextdocumentfragment.cpp
+++ b/src/gui/text/qtextdocumentfragment.cpp
@@ -488,7 +488,8 @@ void QTextHtmlImporter::import()
          *      means there was a tag closing in the input html
          */
         if (currentNodeIdx > 0 && (currentNode->parent != currentNodeIdx - 1)) {
-            blockTagClosed = closeTag();
+            const bool lastBlockTagClosed = closeTag();
+            blockTagClosed = blockTagClosed || lastBlockTagClosed;
             // visually collapse subsequent block tags, but if the element after the closed block tag
             // is for example an inline element (!isBlock) we have to make sure we start a new paragraph by setting
             // hasBlock to false.
@@ -540,6 +541,7 @@ void QTextHtmlImporter::import()
 
             appendBlock(block, currentNode->charFormat);
 
+            blockTagClosed = false;
             hasBlock = true;
         }
 
diff --git a/tests/auto/gui/text/qtextdocumentfragment/tst_qtextdocumentfragment.cpp b/tests/auto/gui/text/qtextdocumentfragment/tst_qtextdocumentfragment.cpp
index 67d4594ec5c..f2c3b36dcdc 100644
--- a/tests/auto/gui/text/qtextdocumentfragment/tst_qtextdocumentfragment.cpp
+++ b/tests/auto/gui/text/qtextdocumentfragment/tst_qtextdocumentfragment.cpp
@@ -16,6 +16,8 @@
 
 #include <qtextcursor.h>
 
+using namespace Qt::StringLiterals;
+
 QT_FORWARD_DECLARE_CLASS(QTextDocument)
 
 class tst_QTextDocumentFragment : public QObject
@@ -245,6 +247,7 @@ private slots:
     void html_fromFirefox();
     void html_emptyInlineInsideBlock();
     void css_fontAndWordSpacing();
+    void html_brWithWhitespaceAfterList();
 
 private:
     inline void setHtml(const QString &html)
@@ -4320,5 +4323,24 @@ void tst_QTextDocumentFragment::css_fontAndWordSpacing()
     }
 }
 
+void tst_QTextDocumentFragment::html_brWithWhitespaceAfterList() // QTBUG-81662
+{
+    setHtml(QString::fromLatin1("<ul><li>one</li><li>two</li></ul>\n <br/>\nhello"));
+
+    QCOMPARE(doc->blockCount(), 3);
+
+    QTextBlock block = doc->begin();
+    QVERIFY(block.textList());
+
+    block = block.next();
+    QVERIFY(block.textList());
+
+    block = block.next();
+    QCOMPARE(block.text(), u"\u2028hello"_s);
+
+    block = block.next();
+    QVERIFY(block.text().isEmpty());
+}
+
 QTEST_MAIN(tst_QTextDocumentFragment)
 #include "tst_qtextdocumentfragment.moc"