String element length

src/Neo/SmartContract/Native/StdLib.cs

@@ -16,6 +16,7 @@
 using System;
 using System.Globalization;
 using System.Numerics;
+using System.Text;
 
 namespace Neo.SmartContract.Native
 {
@@ -222,5 +223,22 @@ private static string[] StringSplit([MaxLength(MaxInputLength)] string str, stri
             StringSplitOptions options = removeEmptyEntries ? StringSplitOptions.RemoveEmptyEntries : StringSplitOptions.None;
             return str.Split(separator, options);
         }
+
+        [ContractMethod(CpuFee = 1 << 8)]
+        private static int StrLen([MaxLength(MaxInputLength)] string str)
+        {
+            // return the length of the string in elements
+            // it should return 1 for both  "🦆" and "ã"
+
+            TextElementEnumerator enumerator = StringInfo.GetTextElementEnumerator(str);
+            int count = 0;
+
+            while (enumerator.MoveNext())
+            {
+                count++;
+            }
+
+            return count;
+        }
     }
 }

tests/Neo.UnitTests/SmartContract/Native/UT_StdLib.cs

@@ -209,6 +209,47 @@ public void StringSplit()
             Assert.AreEqual("b", arr[1].GetString());
         }
 
+        [TestMethod]
+        public void StringElementLength()
+        {
+            var snapshot = TestBlockchain.GetTestSnapshot();
+
+            using var script = new ScriptBuilder();
+            script.EmitDynamicCall(NativeContract.StdLib.Hash, "strLen", "🦆");
+            script.EmitDynamicCall(NativeContract.StdLib.Hash, "strLen", "ã");
+            script.EmitDynamicCall(NativeContract.StdLib.Hash, "strLen", "a");
+
+            using var engine = ApplicationEngine.Create(TriggerType.Application, null, snapshot, settings: TestBlockchain.TheNeoSystem.Settings);
+            engine.LoadScript(script.ToArray());
+
+            Assert.AreEqual(engine.Execute(), VMState.HALT);
+            Assert.AreEqual(3, engine.ResultStack.Count);
+            Assert.AreEqual(1, engine.ResultStack.Pop().GetInteger());
+            Assert.AreEqual(1, engine.ResultStack.Pop().GetInteger());
+            Assert.AreEqual(1, engine.ResultStack.Pop().GetInteger());
+        }
+
+        [TestMethod]
+        public void TestInvalidUtf8Sequence()
+        {
+            // Simulating invalid UTF-8 byte (0xff) decoded as a UTF-16 char
+            const char badChar = (char)0xff;
+            var badStr = badChar.ToString();
+            var snapshot = TestBlockchain.GetTestSnapshot();
+
+            using var script = new ScriptBuilder();
+            script.EmitDynamicCall(NativeContract.StdLib.Hash, "strLen", badStr);
+            script.EmitDynamicCall(NativeContract.StdLib.Hash, "strLen", badStr + "ab");
+
+            using var engine = ApplicationEngine.Create(TriggerType.Application, null, snapshot, settings: TestBlockchain.TheNeoSystem.Settings);
+            engine.LoadScript(script.ToArray());
+
+            Assert.AreEqual(engine.Execute(), VMState.HALT);
+            Assert.AreEqual(2, engine.ResultStack.Count);
+            Assert.AreEqual(3, engine.ResultStack.Pop().GetInteger());
+            Assert.AreEqual(1, engine.ResultStack.Pop().GetInteger());
+        }
+
         [TestMethod]
         public void Json_Deserialize()
         {