1 files changed, 30 insertions, 0 deletions
diff --git a/test/util/UtilTest.mc b/test/util/UtilTest.mc
index ea689a6..c19d6d0 100644
--- a/test/util/UtilTest.mc
+++ b/test/util/UtilTest.mc
@@ -64,3 +64,33 @@ function jobsRunToCompletion(logger as Test.Logger) as Boolean {
     Jobs.runToCompletion(job, 3);
     return job.calls == 4 && job.done == 12;
 }
+
+(:test)
+function bytesUtf8Validation(logger as Test.Logger) as Boolean {
+    var valid = [[0x41]b, [0xC3, 0xA9]b, [0xE2, 0x82, 0xAC]b, [0xF0, 0x9F, 0x98, 0x80]b, [0xF4, 0x8F, 0xBF, 0xBF]b];
+    var invalid = [
+        [0x00]b,                   // NUL
+        [0xC0, 0x80]b,             // overlong NUL
+        [0xE0, 0x80, 0x80]b,       // overlong
+        [0xED, 0xA0, 0x80]b,       // surrogate
+        [0xF4, 0x90, 0x80, 0x80]b, // above U+10FFFF
+        [0xF5, 0x80, 0x80, 0x80]b, // invalid lead byte
+        [0xE2, 0x82]b,             // truncated
+        [0x80]b,                   // lone continuation
+        [0xC3, 0x28]b              // bad continuation
+    ];
+    for (var i = 0; i < valid.size(); i++) {
+        var text = Bytes.toUtf8(valid[i] as ByteArray);
+        if (text == null || !Bytes.equal(Bytes.fromUtf8(text), valid[i] as ByteArray)) {
+            logger.error("valid UTF-8 case " + i + " rejected or changed");
+            return false;
+        }
+    }
+    for (var i = 0; i < invalid.size(); i++) {
+        if (Bytes.toUtf8(invalid[i] as ByteArray) != null) {
+            logger.error("invalid UTF-8 case " + i + " accepted");
+            return false;
+        }
+    }
+    return true;
+}