Update TinyCBOR with bugfixes

Updated to https://github.com/thiagomacieira/tinycbor commit
ac9369d8ec8511bc7516266ae6b07f7da860c954.

Fabrice Fontaine (1):
      fix undefined encode_half in json2cbor

Pedro Oliveira (1):
      Fixed minor error in the example code.

Svyatoslav Phirsov (2):
      Typo fixed in stdlib fread(...) usage
      typo in dumprecursive return type

Thiago Macieira (9):
      Install the tinycbor-version.h header.
      Update version number for a possible but unlikely 0.5.3 release
      Fix #137: off-by-one error in UTF-8 decoding
      Update Travis CI to Ubuntu Xenial
      Pretty: fix use of uninitialised variable
      Validation: fix out-of-bounds access when content ends in a string
      Parser: apply the same memory-check update
      Tests: remove useless comment
      Tests: Catch an earlier QCOMPARE failure in compare()

phirsov (6):
      Fix off-by-one causing buffer overflow in open_memstream
      Protect macro argument expansion using parentheses
      eliminating misleading messages in case .config file not yet created
      Run check silently in Travis
      Make AppVeyor test suit run silent as in Travis
      enhancement #149 implemented: access half-precision floating point data as single float

Change-Id: I9e3d261ad9bf41cfb2b6fffd159088f1cc9b3b02
Reviewed-by: Edward Welbourne <edward.welbourne@qt.io>

2 files changed, 155 insertions, 6 deletions

diff --git a/src/3rdparty/tinycbor/tests/encoder/data.cpp b/src/3rdparty/tinycbor/tests/encoder/data.cpp
index c33fb605aa..8b00cfec1f 100644
--- a/src/3rdparty/tinycbor/tests/encoder/data.cpp
+++ b/src/3rdparty/tinycbor/tests/encoder/data.cpp
@@ -123,6 +123,42 @@ QVariant make_ilmap(const std::initializer_list<QPair<QVariant, QVariant>> &list
     return QVariant::fromValue(IndeterminateLengthMap(list));
 }
 
+void addHalfFloat()
+{
+    QTest::addColumn<QByteArray>("output");
+    QTest::addColumn<unsigned>("rawInput");
+    QTest::addColumn<double>("floatInput");
+
+    QTest::newRow("+0") << raw("\x00\x00") << 0U << 0.0;
+    QTest::newRow("-0") << raw("\x80\x00") << 0x8000U << 0.0;
+
+    QTest::newRow("min.denorm") << raw("\x00\x01") << 1U << ldexp(1.0, -14) * ldexp(1.0, -10);
+    QTest::newRow("-min.denorm") << raw("\x80\x01") << 0x8001U << ldexp(-1.0, -14) * ldexp(1.0, -10);
+
+    QTest::newRow("max.denorm") << raw("\x03\xff") << 0x03ffU << ldexp(1.0, -14) * (1.0 - ldexp(1.0, -10));
+    QTest::newRow("-max.denorm") << raw("\x83\xff") << 0x83ffU << ldexp(-1.0, -14) * (1.0 - ldexp(1.0, -10));
+
+    QTest::newRow("min.norm") << raw("\x04\x00") << 0x0400U << ldexp(1.0, -14);
+    QTest::newRow("-min.norm") << raw("\x84\x00") << 0x8400U << ldexp(-1.0, -14);
+
+    QTest::newRow("1.0") << raw("\x3c\x00") << 0x3c00U << 1.0;
+    QTest::newRow("-1.0") << raw("\xbc\x00") << 0xbc00U << -1.0;
+
+    QTest::newRow("1.5") << raw("\x3e\x00") << 0x3e00U << 1.5;
+    QTest::newRow("-1.5") << raw("\xbe\x00") << 0xbe00U << -1.5;
+
+    QTest::newRow("max") << raw("\x7b\xff") << 0x7bffU << ldexp(1.0, 15) * (2.0 - ldexp(1.0, -10));
+    QTest::newRow("-max") << raw("\xfb\xff") << 0xfbffU << ldexp(-1.0, 15) * (2.0 - ldexp(1.0, -10));
+
+    QTest::newRow("inf") << raw("\x7c\x00") << 0x7c00U << myInf();
+    QTest::newRow("-inf") << raw("\xfc\x00") << 0xfc00U << myNInf();
+
+    QTest::newRow("nan1") << raw("\x7c\x01") << 0x7c01U << myNaN();
+    QTest::newRow("nan2") << raw("\xfc\x01") << 0xfc01U << myNaN();
+    QTest::newRow("nan3") << raw("\x7e\x00") << 0x7e00U << myNaN();
+    QTest::newRow("nan4") << raw("\xfe\x00") << 0xfe00U << myNaN();
+}
+
 void addColumns()
 {
     QTest::addColumn<QByteArray>("output");
diff --git a/src/3rdparty/tinycbor/tests/encoder/tst_encoder.cpp b/src/3rdparty/tinycbor/tests/encoder/tst_encoder.cpp
index f30c522601..458f55eb10 100644
--- a/src/3rdparty/tinycbor/tests/encoder/tst_encoder.cpp
+++ b/src/3rdparty/tinycbor/tests/encoder/tst_encoder.cpp
@@ -41,6 +41,13 @@ class tst_Encoder : public QObject
 {
     Q_OBJECT
 private slots:
+    void floatAsHalfFloat_data();
+    void floatAsHalfFloat();
+    void halfFloat_data();
+    void halfFloat();
+    void floatAsHalfFloatCloseToZero_data();
+    void floatAsHalfFloatCloseToZero();
+    void floatAsHalfFloatNaN();
     void fixed_data();
     void fixed();
     void strings_data();
@@ -178,21 +185,127 @@ CborError encodeVariant(CborEncoder *encoder, const QVariant &v)
     return CborErrorUnknownType;
 }
 
-void compare(const QVariant &input, const QByteArray &output)
+template <typename Input, typename FnUnderTest>
+void encodeOne(Input input, FnUnderTest fn_under_test, QByteArray &buffer, CborError &error)
 {
-    QByteArray buffer(output.length(), Qt::Uninitialized);
     uint8_t *bufptr = reinterpret_cast<quint8 *>(buffer.data());
     CborEncoder encoder;
     cbor_encoder_init(&encoder, bufptr, buffer.length(), 0);
 
-    QCOMPARE(encodeVariant(&encoder, input), CborNoError);
-    QCOMPARE(encoder.remaining, size_t(1));
-    QCOMPARE(cbor_encoder_get_extra_bytes_needed(&encoder), size_t(0));
+    error = fn_under_test(&encoder, input);
+
+    if (error == CborNoError) {
+        QCOMPARE(encoder.remaining, size_t(1));
+        QCOMPARE(cbor_encoder_get_extra_bytes_needed(&encoder), size_t(0));
+
+        buffer.resize(int(cbor_encoder_get_buffer_size(&encoder, bufptr)));
+    }
+}
 
-    buffer.resize(int(cbor_encoder_get_buffer_size(&encoder, bufptr)));
+template <typename Input, typename FnUnderTest>
+void compare(Input input, FnUnderTest fn_under_test, const QByteArray &output)
+{
+    QByteArray buffer(output.length(), Qt::Uninitialized);
+    CborError error;
+
+    encodeOne(input, fn_under_test, buffer, error);
+    if (QTest::currentTestFailed())
+        return;
+
+    QCOMPARE(error, CborNoError);
     QCOMPARE(buffer, output);
 }
 
+void compare(const QVariant &input, const QByteArray &output)
+{
+    compare(input, encodeVariant, output);
+}
+
+void tst_Encoder::floatAsHalfFloat_data()
+{
+    addHalfFloat();
+}
+
+void tst_Encoder::floatAsHalfFloat()
+{
+    QFETCH(unsigned, rawInput);
+    QFETCH(double, floatInput);
+    QFETCH(QByteArray, output);
+
+    if (rawInput == 0U || rawInput == 0x8000U)
+        QSKIP("zero values are out of scope of this test case", QTest::SkipSingle);
+
+    if (qIsNaN(floatInput))
+        QSKIP("NaN values are out of scope of this test case", QTest::SkipSingle);
+
+    output.prepend('\xf9');
+
+    compare((float)floatInput, cbor_encode_float_as_half_float, output);
+}
+
+void tst_Encoder::halfFloat_data()
+{
+    addHalfFloat();
+}
+
+void tst_Encoder::halfFloat()
+{
+    QFETCH(unsigned, rawInput);
+    QFETCH(QByteArray, output);
+
+    uint16_t v = (uint16_t)rawInput;
+    output.prepend('\xf9');
+
+    compare(&v, cbor_encode_half_float, output);
+}
+
+void tst_Encoder::floatAsHalfFloatCloseToZero_data()
+{
+    QTest::addColumn<double>("floatInput");
+
+    QTest::newRow("+0") << 0.0;
+    QTest::newRow("-0") << -0.0;
+
+    QTest::newRow("below min.denorm") << ldexp(1.0, -14) * ldexp(1.0, -11);
+    QTest::newRow("above -min.denorm") << ldexp(-1.0, -14) * ldexp(1.0, -11);
+}
+
+void tst_Encoder::floatAsHalfFloatCloseToZero()
+{
+    QFETCH(double, floatInput);
+
+    QByteArray buffer(4, Qt::Uninitialized);
+    CborError error;
+
+    encodeOne((float)floatInput, cbor_encode_float_as_half_float, buffer, error);
+
+    QCOMPARE(error, CborNoError);
+
+    QVERIFY2(
+        buffer == raw("\xf9\x00\x00") || buffer == raw("\xf9\x80\x00"),
+        "Got value " + QByteArray::number(floatInput) + " encoded to: " + buffer);
+}
+
+void tst_Encoder::floatAsHalfFloatNaN()
+{
+    QByteArray buffer(4, Qt::Uninitialized);
+    CborError error;
+
+    encodeOne(myNaNf(), cbor_encode_float_as_half_float, buffer, error);
+
+    QCOMPARE(error, CborNoError);
+    QCOMPARE(buffer.size(), 3);
+
+    uint8_t ini_byte = (uint8_t)buffer[0],
+        exp = (uint8_t)buffer[1] & 0x7cU,
+        manth = (uint8_t)buffer[1] & 0x03U,
+        mantl = (uint8_t)buffer[2];
+
+    QCOMPARE((unsigned)ini_byte, 0xf9U);
+    QCOMPARE((unsigned)exp, 0x7cU);
+    QVERIFY((manth | mantl) != 0);
+}
+
 void tst_Encoder::fixed_data()
 {
     addColumns();