fix: std.char() rejects surrogate codepoints with an error

sjsonnet/src/sjsonnet/stdlib/StringModule.scala

@@ -210,13 +210,12 @@ object StringModule extends AbstractFunctionModule {
    */
   private object Char_ extends Val.Builtin1("char", "n") {
     def evalRhs(n: Eval, ev: EvalScope, pos: Position): Val = {
-      val c = n.value.asInt
-      if (!Character.isValidCodePoint(c)) {
-        Error.fail(s"Invalid unicode code point, got " + c)
+      val c0 = n.value.asInt
+      if (!Character.isValidCodePoint(c0) || (c0 >= 0xd800 && c0 <= 0xdfff)) {
+        Error.fail(s"Invalid unicode code point, got " + c0)
       }
-      val s = Character.toString(c)
-      // Single-codepoint result; ASCII printable except '"' and '\\' is JSON-safe.
-      if (c >= 0x20 && c < 0x7f && c != '"' && c != '\\') Val.Str.asciiSafe(pos, s)
+      val s = Character.toString(c0)
+      if (c0 >= 0x20 && c0 < 0x7f && c0 != '"' && c0 != '\\') Val.Str.asciiSafe(pos, s)
       else Val.Str(pos, s)
     }
   }

sjsonnet/test/resources/new_test_suite/char_surrogate_boundary.jsonnet

@@ -0,0 +1,9 @@
+// Test that std.char() works correctly for valid codepoints near the surrogate range
+// and rejects surrogate codepoints (0xD800-0xDFFF) with an error.
+[
+  std.codepoint(std.char(55295)),  // 0xD7FF - last valid before surrogates
+  std.codepoint(std.char(57344)),  // 0xE000 - first valid after surrogates
+  std.codepoint(std.char(65533)),  // 0xFFFD - replacement character itself
+  std.codepoint(std.char(0)),      // 0x0000 - null
+  std.codepoint(std.char(65)),     // 0x0041 - 'A'
+]

sjsonnet/test/resources/new_test_suite/char_surrogate_boundary.jsonnet.golden

@@ -0,0 +1,7 @@
+[
+   55295,
+   57344,
+   65533,
+   0,
+   65
+]

sjsonnet/test/resources/new_test_suite/error.char_surrogate.jsonnet

@@ -0,0 +1,3 @@
+// std.char() should reject surrogate codepoints (0xD800-0xDFFF)
+// High surrogate 0xD800 = 55296
+std.char(55296)

sjsonnet/test/resources/new_test_suite/error.char_surrogate.jsonnet.golden

@@ -0,0 +1 @@
+sjsonnet.Error: [std.char] Invalid unicode code point, got 55296

sjsonnet/test/src/sjsonnet/UnicodeHandlingTests.scala

@@ -163,23 +163,17 @@ object UnicodeHandlingTests extends TestSuite {
 
       assert(sjsonnet.Util.compareStringsByCodepoint(rawSurrogatePrefix, validSurrogatePair) < 0)
       assert(sjsonnet.Util.compareStringsByCodepoint(validSurrogatePair, rawSurrogatePrefix) > 0)
-
-      eval("(std.char(55296) + std.char(65535)) < (std.char(55296) + std.char(56320))") ==>
-      ujson.Bool(true)
-
-      eval(
-        "std.sort([std.char(55296) + std.char(56320), std.char(55296) + std.char(65535)])"
-      ) ==> ujson.Arr(rawSurrogatePrefix, validSurrogatePair)
     }
 
-    // Unpaired surrogate handling - sjsonnet-specific behavior
+    // Unpaired surrogate handling
     //
-    // Note: This is an intentional divergence from go-jsonnet and C++ jsonnet:
-    // - go/C++ reject unpaired surrogates in escape sequences at parse time
-    // - go-jsonnet's std.char() replaces surrogate codepoints with U+FFFD
-    // - sjsonnet was preserving unpaired surrogates throughout
+    // Note: The three reference implementations diverge on surrogate codepoints:
+    // - go-jsonnet replaces surrogates with U+FFFD
+    // - jsonnet-cpp preserves raw surrogates
+    // - jrsonnet rejects surrogates with an error
     //
-    // sjsonnet now reject these to align with go-jsonet/ c++ jsonnet
+    // sjsonnet rejects surrogates with an error (matching jrsonnet),
+    // since surrogates are not valid Unicode codepoints per the Unicode spec.
     //
 
     test("unpairedSurrogatesInEscapes") {
@@ -191,10 +185,10 @@ object UnicodeHandlingTests extends TestSuite {
       eval("\"\\uD83C\\uDF0D\"") ==> ujson.Str("🌍") // Earth emoji
     }
 
-    test("stdCharPreservesRawSurrogates") {
-      // sjsonnet preserves raw surrogate codepoints (go-jsonnet would replace with U+FFFD)
-      eval("std.codepoint(std.char(55296))") ==> ujson.Num(55296) // 0xD800 high surrogate
-      eval("std.codepoint(std.char(56320))") ==> ujson.Num(56320) // 0xDC00 low surrogate
+    test("stdCharRejectsSurrogates") {
+      evalErr("std.char(55296)").contains("Invalid unicode code point") // 0xD800
+      evalErr("std.char(56320)").contains("Invalid unicode code point") // 0xDC00
+      evalErr("std.char(57343)").contains("Invalid unicode code point") // 0xDFFF
     }
 
     test("invalidSurrogateHandling") {