Add ICustomMarshaler Utf8Marshaler

vmuriart · vmuriart · commit e487076d90ac · 2017-02-27T18:58:05.000-07:00
Refactor PyString_FromStringAndSize Link explains why `MarshalAs(UnmanagedType.LPWStr)` or `CharSet.Unicode` don't work http://stackoverflow.com/a/25128147/5208670
diff --git a/src/runtime/CustomMarshaler.cs b/src/runtime/CustomMarshaler.cs
@@ -123,4 +123,49 @@ public static ICustomMarshaler GetInstance(string cookie)
             return Instance;
         }
     }
+
+
+    /// <summary>
+    /// Custom Marshaler to deal with Managed String to Native
+    /// conversion on UTF-8. Use on functions that expect UTF-8 encoded
+    /// strings like `PyUnicode_FromStringAndSize`
+    /// </summary>
+    /// <remarks>
+    /// If instead we used `MarshalAs(UnmanagedType.LPWStr)` the output to
+    /// `foo` would be `f\x00o\x00o\x00`.
+    /// </remarks>
+    public class Utf8Marshaler : MarshalerBase
+    {
+        private static readonly MarshalerBase Instance = new Utf8Marshaler();
+        private static readonly Encoding PyEncoding = Encoding.UTF8;
+
+        public override IntPtr MarshalManagedToNative(object managedObj)
+        {
+            var s = managedObj as string;
+
+            if (s == null)
+            {
+                return IntPtr.Zero;
+            }
+
+            byte[] bStr = PyEncoding.GetBytes(s + "\0");
+            IntPtr mem = Marshal.AllocHGlobal(bStr.Length);
+            try
+            {
+                Marshal.Copy(bStr, 0, mem, bStr.Length);
+            }
+            catch (Exception)
+            {
+                Marshal.FreeHGlobal(mem);
+                throw;
+            }
+
+            return mem;
+        }
+
+        public static ICustomMarshaler GetInstance(string cookie)
+        {
+            return Instance;
+        }
+    }
 }
diff --git a/src/runtime/runtime.cs b/src/runtime/runtime.cs
@@ -1508,23 +1508,12 @@ internal static IntPtr PyBytes_AS_STRING(IntPtr ob)
             return ob + BytesOffset.ob_sval;
         }
 
-        internal static IntPtr PyString_FromStringAndSize(string value, int length)
-        {
-            // copy the string into an unmanaged UTF-8 buffer
-            int len = Encoding.UTF8.GetByteCount(value);
-            byte[] buffer = new byte[len + 1];
-            Encoding.UTF8.GetBytes(value, 0, value.Length, buffer, 0);
-            IntPtr nativeUtf8 = Marshal.AllocHGlobal(buffer.Length);
-            try
-            {
-                Marshal.Copy(buffer, 0, nativeUtf8, buffer.Length);
-                return PyUnicode_FromStringAndSize(nativeUtf8, length);
-            }
-            finally
-            {
-                Marshal.FreeHGlobal(nativeUtf8);
-            }
-        }
+        [DllImport(Runtime.dll, EntryPoint = "PyUnicode_FromStringAndSize")]
+        internal static extern IntPtr
+            PyString_FromStringAndSize(
+                [MarshalAs(UnmanagedType.CustomMarshaler, MarshalTypeRef = typeof(Utf8Marshaler))] string value,
+                int size
+            );
 
         [DllImport(Runtime.dll, CallingConvention = CallingConvention.Cdecl,
             ExactSpelling = true, CharSet = CharSet.Unicode)]
