feat(mf2): Apply NFC normalization to names and keys (unicode-org/message-format-wg#885)

Author: XM5jDcsHTyGJtQqlCi

mf2/messageformat/src/cst/names.ts

@@ -54,9 +54,9 @@ export function parseNameValue(
   if (!isNameStartCode(cc)) return null;
   cc = src.charCodeAt(++pos);
   while (isNameCharCode(cc)) cc = src.charCodeAt(++pos);
-  const name = src.substring(nameStart, pos);
+  const value = src.substring(nameStart, pos).normalize();
   if (bidiChars.has(cc)) pos += 1;
-  return { value: name, end: pos };
+  return { value, end: pos };
 }
 
 export function isValidUnquotedLiteral(str: string): boolean {

mf2/messageformat/src/cst/parse-cst.ts

@@ -142,10 +142,13 @@ function parseVariant(ctx: ParseContext, start: number): CST.Variant {
 
     if (pos > start && !ws.hasWS) ctx.onError('missing-syntax', pos, "' '");
 
-    const key =
-      ch === '*'
-        ? ({ type: '*', start: pos, end: pos + 1 } satisfies CST.CatchallKey)
-        : parseLiteral(ctx, pos, true);
+    let key: CST.CatchallKey | CST.Literal;
+    if (ch === '*') {
+      key = { type: '*', start: pos, end: pos + 1 };
+    } else {
+      key = parseLiteral(ctx, pos, true);
+      key.value = key.value.normalize();
+    }
     if (key.end === pos) break; // error; reported in pattern.errors
     keys.push(key);
     pos = key.end;

mf2/messageformat/src/data-model/parse.ts

@@ -91,7 +91,9 @@ function variant(): Model.Variant {
       keys.push({ type: '*' });
       pos += 1;
     } else {
-      keys.push(literal(true));
+      const key = literal(true);
+      key.value = key.value.normalize();
+      keys.push(key);
     }
   }
   return { keys, value: pattern(true) };

mf2/messageformat/src/data-model/resolve-variable.ts

@@ -42,6 +42,10 @@ function getValue(scope: unknown, name: string): unknown {
         return getValue(scope[head], tail);
       }
     }
+
+    for (const [key, value] of Object.entries(scope)) {
+      if (key.normalize() === name) return value;
+    }
   }
 
   return undefined;

mf2/messageformat/src/functions/string.ts

@@ -33,12 +33,13 @@ export function string(
   input?: unknown
 ): MessageString {
   const str = input === undefined ? '' : String(input);
+  const selStr = str.normalize();
   const [locale] = mergeLocales(locales, input, options);
   return {
     type: 'string',
     source,
     locale,
-    selectKey: keys => (keys.has(str) ? str : null),
+    selectKey: keys => (keys.has(selStr) ? selStr : null),
     toParts: () => [{ type: 'string', source, locale, value: str }],
     toString: () => str,
     valueOf: () => str