Fix load/cast of QString to avoid issues with surrogate pairs.

ModOrganizer2 · Dec 2, 2023 · aac872c · aac872c
1 parent f4ee4da
commit aac872c
Show file tree

Hide file tree

Showing 3 changed files with 43 additions and 20 deletions.
diff --git a/src/pybind11-qt/pybind11_qt_basic.cpp b/src/pybind11-qt/pybind11_qt_basic.cpp
@@ -35,30 +35,36 @@ namespace pybind11::detail {
      */
     bool type_caster<QString>::load(handle src, bool)
     {
-
         PyObject* objPtr = src.ptr();
 
-        if (!PyBytes_Check(objPtr) && !PyUnicode_Check(objPtr)) {
-            return false;
+        if (PyBytes_Check(objPtr)) {
+            value = QString::fromUtf8(PyBytes_AsString(objPtr));
+            return true;
         }
+        else if (PyUnicode_Check(objPtr)) {
+            switch (PyUnicode_KIND(objPtr)) {
+            case PyUnicode_1BYTE_KIND:
+                value = QString::fromUtf8(PyUnicode_AsUTF8(objPtr));
+                break;
+            case PyUnicode_2BYTE_KIND:
+                value = QString::fromUtf16(
+                    reinterpret_cast<char16_t*>(PyUnicode_2BYTE_DATA(objPtr)),
+                    PyUnicode_GET_LENGTH(objPtr));
+                break;
+            case PyUnicode_4BYTE_KIND:
+                value = QString::fromUcs4(
+                    reinterpret_cast<char32_t*>(PyUnicode_4BYTE_DATA(objPtr)),
+                    PyUnicode_GET_LENGTH(objPtr));
+                break;
+            default:
+                return false;
+            }
 
-        // Ensure the string uses 8-bit characters
-        PyObject* strPtr =
-            PyUnicode_Check(objPtr) ? PyUnicode_AsUTF8String(objPtr) : objPtr;
-
-        if (!strPtr) {
-            return false;
+            return true;
         }
-
-        // Extract the character data from the python string
-        value = QString::fromUtf8(PyBytes_AsString(strPtr));
-
-        // Deallocate local copy if one was made
-        if (strPtr != objPtr) {
-            Py_DecRef(strPtr);
+        else {
+            return false;
         }
-
-        return true;
     }
 
     /**
@@ -71,8 +77,8 @@ namespace pybind11::detail {
     handle type_caster<QString>::cast(QString src, return_value_policy /* policy */,
                                       handle /* parent */)
     {
-        static_assert(sizeof(QChar) == 2);
-        return PyUnicode_FromString(src.toStdString().c_str());
+        return PyUnicode_DecodeUTF16(reinterpret_cast<const char*>(src.utf16()),
+                                     2 * src.length(), nullptr, 0);
     }
 
     bool type_caster<QVariant>::load(handle src, bool)

diff --git a/tests/python/test_qt.cpp b/tests/python/test_qt.cpp
@@ -11,6 +11,14 @@ PYBIND11_MODULE(qt, m)
 {
     // QString
 
+    m.def("create_qstring_with_emoji", []() {
+        return QString::fromUtf16(u"\U0001F600");
+    });
+
+    m.def("consume_qstring_with_emoji", [](QString const& qstring) {
+        return qstring.length();
+    });
+
     m.def("qstring_to_stdstring", [](QString const& qstring) {
         return qstring.toStdString();
     });

diff --git a/tests/python/test_qt.py b/tests/python/test_qt.py
@@ -10,10 +10,19 @@ def test_qstring():
 
     assert m.qstring_to_stdstring("éàüö") == "éàüö"
     assert m.stdstring_to_qstring("éàüö") == "éàüö"
+    assert m.qstring_to_stdstring("خالد") == "خالد"
+    assert m.qstring_to_stdstring("🌎") == "🌎"
 
     assert m.qstring_to_int("2") == 2
     assert m.int_to_qstring(2) == "2"
 
+    emoji = m.create_qstring_with_emoji()
+
+    assert emoji.encode("utf-16be", "surrogatepass") == b"\xd8\x3d\xde\x00"
+    assert m.consume_qstring_with_emoji(emoji) == 2
+
+    assert m.consume_qstring_with_emoji("🌎") == 2
+
 
 def test_qstringlist():
     assert m.qstringlist_join([""], "--") == ""