From b8db1233411893aaecb0b6a61b02b0ef3c1520e5 Mon Sep 17 00:00:00 2001 From: Lars Knoll Date: Thu, 30 Apr 2020 11:43:21 +0200 Subject: Add a method to determine the encoding for encoded HTML data This is a replacement for Qt::codecForHtml(). Change-Id: I31f03518fd9c70507cbd210a8bcf405b6a0106b1 Reviewed-by: Thiago Macieira --- .../text/qstringconverter/tst_qstringconverter.cpp | 89 ++++++++++++++++++++++ 1 file changed, 89 insertions(+) (limited to 'tests') diff --git a/tests/auto/corelib/text/qstringconverter/tst_qstringconverter.cpp b/tests/auto/corelib/text/qstringconverter/tst_qstringconverter.cpp index 3f4bbb413f..78595bc17b 100644 --- a/tests/auto/corelib/text/qstringconverter/tst_qstringconverter.cpp +++ b/tests/auto/corelib/text/qstringconverter/tst_qstringconverter.cpp @@ -68,6 +68,9 @@ private slots: void encodingForData_data(); void encodingForData(); + + void encodingForHtml_data(); + void encodingForHtml(); }; void tst_QStringConverter::constructByName() @@ -1722,6 +1725,92 @@ void tst_QStringConverter::encodingForData() QCOMPARE(e, encoding); } + +void tst_QStringConverter::encodingForHtml_data() +{ + QTest::addColumn("html"); + QTest::addColumn>("encoding"); + + QByteArray html = "blah"; + QTest::newRow("no charset") << html << std::optional(QStringConverter::Utf8); + + html = ""; + QTest::newRow("latin 15") << html << std::optional(); + + html = ""; + QTest::newRow("latin 1") << html << std::optional(QStringConverter::Latin1); + + html = "Test"; + QTest::newRow("latin 1 (#2)") << html << std::optional(QStringConverter::Latin1); + + html = "Test"; + QTest::newRow("UTF-8") << html << std::optional(QStringConverter::Utf8); + + html = "Test"; + QTest::newRow("UTF-8 (#2)") << html << std::optional(QStringConverter::Utf8); + + html = ""; + QTest::newRow("UTF-8, no quotes") << html << std::optional(QStringConverter::Utf8); + + html = ""; + QTest::newRow("UTF-8, single quotes") << html << std::optional(QStringConverter::Utf8); + + html = "Test"; + QTest::newRow("UTF-8, > terminator") << html << std::optional(QStringConverter::Utf8); + + html = "Test"; + QTest::newRow("UTF-8, > terminator with spaces") << html << std::optional(QStringConverter::Utf8); + + // Test invalid charsets. + html = "Test"; + QTest::newRow("utf/8") << html << std::optional(); + + html = ""; + QTest::newRow("invalid charset, no default") << html << std::optional(); + + html = "(QStringConverter::Utf8); + + + html = "(QStringConverter::Utf8); + + + html = "Test"; + QTest::newRow("invalid charset, early terminator (')") << html << std::optional(); + + const char src[] = { char(0xff), char(0xfe), char(0x7a), char(0x03), 0, 0 }; + html = src; + QTest::newRow("greek text UTF-16LE") << html << std::optional(QStringConverter::Utf16LE); + + + html = "ͻ\000"; + QTest::newRow("greek text UTF-8") << html << std::optional(QStringConverter::Utf8); + + html = "" + "

bla

"; // QTBUG-41998, ICU will return UTF-16. + QTest::newRow("legacy unicode UTF-8") << html << std::optional(QStringConverter::Utf8); +} + +void tst_QStringConverter::encodingForHtml() +{ + QFETCH(QByteArray, html); + QFETCH(std::optional, encoding); + + QCOMPARE(QStringConverter::encodingForHtml(html.constData(), html.size()), encoding); +} + class LoadAndConvert: public QRunnable { public: -- cgit v1.2.3