unicode

brili.ts

@@ -440,7 +440,7 @@ function evalInstr(instr: bril.Instruction, state: State): Action {
       else
         value = BigInt(Math.floor(instr.value))
     } else if (typeof instr.value === "string") {
-      if(instr.value.length > 1) throw error(`char must have one character`);
+      if([...instr.value].length !== 1) throw error(`char must have one character`);
       value = instr.value;
     } else {
       value = instr.value;
@@ -754,18 +754,18 @@ function evalInstr(instr: bril.Instruction, state: State): Action {
   }
 
   case "char2int": {
-    let code = getChar(instr, state.env, 0).charCodeAt(0);
-    let val = BigInt.asIntN(64, BigInt(code));
+    let code = getChar(instr, state.env, 0).codePointAt(0);
+    let val = BigInt.asIntN(64, BigInt(code as number));
     state.env.set(instr.dest, val);
     return NEXT;
   }
 
   case "int2char": {
-    let i = getInt(instr, state.env, 0)
-    if(i > 65535 || i < 0 || (55295 < i && i < 57344)) {
+    let i = getInt(instr, state.env, 0);
+    if(i > 1114111 || i < 0) {
       throw error(`value ${i} cannot be converted to char`);
-    }
-    let val = String.fromCharCode(Number(i));
+    } 
+    let val = String.fromCodePoint(Number(i));
     state.env.set(instr.dest, val);
     return NEXT;
   }
@@ -854,7 +854,7 @@ function evalFunc(func: bril.Function, state: State): Value | null {
 
 function parseChar(s: string): string {
   let c = s;
-  if (c.length == 1) {
+  if ([...c].length === 1) {
     return c;
   } else {
     throw error(`char argument to main must have one character; got ${s}`);

docs/lang/char.md

@@ -8,10 +8,10 @@ The character extension adds one new base type:
 
     "char"
 
-Characters are [UTF-16][] code units.
+Characters are a singular [Unicode character][].
 
 
-[UTF-16]: https://en.wikipedia.org/wiki/UTF-16
+[Unicode character]: https://www.unicode.org/glossary/#character
 
 Operations
 ----------
@@ -26,8 +26,8 @@ Comparison operators, which take two `char` values and produce a `bool`:
 
 Conversion operators:
 
-- `char2int`: One argument: a variable of type `char`. Returns an integer representing the UTF-16 code unit of the given value.
-- `int2char`: One argument: a variable of type `int`. Returns the corresponding UTF-16 code unit. Throws if the value does not correspond to a valid character.
+- `char2int`: One argument: a variable of type `char`. Returns an integer representing the Unicode code point of the given value.
+- `int2char`: One argument: a variable of type `int`. Returns the corresponding Unicode character. Throws if the value does not correspond to a valid Unicode code point.
 
 Printing
 --------

test/interp-error/char-error/badconversion.bril

@@ -1,7 +1,7 @@
 @main {
-  i: int = const 56193;
+  i: int = const 1114112;
 
   c: char = int2char i;
 
-  print i;
+  print c;
 }

test/interp-error/char-error/badconversion.err

@@ -1 +1 @@
-error: value 56193 cannot be converted to char
+error: value 1114112 cannot be converted to char

test/interp/char/char.bril

@@ -2,5 +2,6 @@
   c1: char = const 'h';
   c2: char = const 'e';
   c3: char = const 'y';
-  print c1 c2 c3;
+  c4: char = const '🐶';
+  print c1 c2 c3 c4;
 }

test/interp/char/char.out

@@ -1 +1 @@
-h e y
+h e y 🐶

test/interp/char/char_args.bril

@@ -1,4 +1,4 @@
-# ARGS: h e y
-@main(char_one : char, char_two : char, char_three : char) {
-  print char_one char_two char_three;
+# ARGS: h e y 🐱
+@main(c1: char, c2: char, c3: char, c4: char) {
+  print c1 c2 c3 c4;
 }

test/interp/char/char_args.out

@@ -1 +1 @@
-h e y
+h e y 🐱

test/interp/char/conversions.bril

@@ -1,9 +1,13 @@
 @main() {
   c1: char = const 'A';
+  c2: char = const '𝛀'; 
   i1: int = const 65;
-
-  c2: char = int2char i1;
-  i2: int = char2int c1;
+  i2: int = const 120512;
 
-  print c2 i2;
+  c3: char = int2char i1;
+  i3: int = char2int c1;
+  c4: char = int2char i2;
+  i4: int = char2int c2;
+
+  print c3 i3 c4 i4;
 }

test/interp/char/conversions.out

@@ -1 +1 @@
-A 65
+A 65 𝛀 120512