pythonnet · lostmsu · Jun 11, 2021 · Jun 8, 2021 · Jun 8, 2021 · Jun 8, 2021
diff --git a/AUTHORS.md b/AUTHORS.md
@@ -55,6 +55,7 @@
 -   Meinrad Recheis ([@henon](https://github.com/henon))
 -   Mohamed Koubaa ([@koubaa](https://github.com/koubaa))
 -   Patrick Stewart ([@patstew](https://github.com/patstew))
+-   Peter Kese ([@patstew](https://github.com/pkese))
 -   Raphael Nestler ([@rnestler](https://github.com/rnestler))
 -   Rickard Holmberg ([@rickardraysearch](https://github.com/rickardraysearch))
 -   Sam Winstanley ([@swinstanley](https://github.com/swinstanley))

diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -71,6 +71,7 @@ One must now either use enum members (e.g. `MyEnum.Option`), or use enum constru
 -   Exception stacktraces on `PythonException.StackTrace` are now properly formatted
 -   Providing an invalid type parameter to a generic type or method produces a helpful Python error
 -   Empty parameter names (as can be generated from F#) do not cause crashes
+-   Unicode strings with surrogates get truncated when converting from Python
 
 ### Removed
 

diff --git a/src/embed_tests/TestPyString.cs b/src/embed_tests/TestPyString.cs
@@ -94,5 +94,25 @@ public void TestUnicode()
             PyObject actual = new PyString(expected);
             Assert.AreEqual(expected, actual.ToString());
         }
+
+        [Test]
+        public void TestUnicodeSurrogateToString()
+        {
+            var expected = "foo\ud83d\udc3c";
+            var actual = PythonEngine.Eval("'foo\ud83d\udc3c'");
+            Assert.AreEqual(4, actual.Length());
+            Assert.AreEqual(expected, actual.ToString());
+        }
+
+        [Test]
+        [Ignore("Bug: Unicode conversion issue #1466")]
+        public void TestUnicodeSurrogate()
+        {
+            const string expected = "foo\ud83d\udc3c"; // "foo🐼"
+            PyObject actual = new PyString(expected);
+            // python treats "foo🐼" as 4 characters, dotnet as 5
+            Assert.AreEqual(4, actual.Length());
+            Assert.AreEqual(expected, actual.ToString());
+        }
     }
 }
diff --git a/src/runtime/runtime.cs b/src/runtime/runtime.cs
@@ -1646,11 +1646,12 @@ internal static string GetManagedString(IntPtr op)
             if (type == PyUnicodeType)
             {
                 using var p = PyUnicode_AsUTF16String(new BorrowedReference(op));
-                int length = (int)PyUnicode_GetSize(op);
-                char* codePoints = (char*)PyBytes_AsString(p.DangerousGetAddress());
+                var bytesPtr = p.DangerousGetAddress();
+                int bytesLength = (int)Runtime.PyBytes_Size(bytesPtr);
+                char* codePoints = (char*)PyBytes_AsString(bytesPtr);
                 return new string(codePoints,
                                   startIndex: 1, // skip BOM
-                                  length: length);
+                                  length: bytesLength/2-1); // utf16 - BOM
             }
 
             return null;