Editorial: Improve ParseMonthCode with small month code grammar

Co-Authored-By: Richard Gibson <[email protected]>

Author: ptomato

polyfill/lib/monthcode.mjs

@@ -2,31 +2,24 @@ import {
   String as StringCtor,
   RangeError as RangeErrorCtor,
   TypeError as TypeErrorCtor,
-  StringPrototypeIndexOf,
   StringPrototypePadStart,
-  StringPrototypeSlice
+  RegExpPrototypeExec
 } from './primordials.mjs';
 
 import Call from 'es-abstract/2024/Call.js';
 import ToPrimitive from 'es-abstract/2024/ToPrimitive.js';
 
+import { monthCode as MONTH_CODE_REGEX } from './regex.mjs';
+
 export function ParseMonthCode(argument) {
   const value = ToPrimitive(argument, StringCtor);
   if (typeof value !== 'string') throw new TypeErrorCtor('month code must be a string');
-  if (
-    value.length < 3 ||
-    value.length > 4 ||
-    value[0] !== 'M' ||
-    Call(StringPrototypeIndexOf, '0123456789', [value[1]]) === -1 ||
-    Call(StringPrototypeIndexOf, '0123456789', [value[2]]) === -1 ||
-    (value[1] + value[2] === '00' && value[3] !== 'L') ||
-    (value[3] !== 'L' && value[3] !== undefined)
-  ) {
-    throw new RangeErrorCtor(`bad month code ${value}; must match M01-M99 or M00L-M99L`);
-  }
-  const isLeapMonth = value.length === 4;
-  const monthNumber = +Call(StringPrototypeSlice, value, [1, 3]);
-  return { monthNumber, isLeapMonth };
+  const match = Call(RegExpPrototypeExec, MONTH_CODE_REGEX, [value]);
+  if (!match) throw new RangeErrorCtor(`bad month code ${value}; must match M01-M99 or M00L-M99L`);
+  return {
+    monthNumber: +(match[1] ?? match[3] ?? match[5]),
+    isLeapMonth: (match[2] ?? match[4] ?? match[6]) === 'L'
+  };
 }
 
 export function CreateMonthCode(monthNumber, isLeapMonth) {

polyfill/lib/regex.mjs

@@ -69,3 +69,5 @@ const fraction = /(\d+)(?:[.,](\d{1,9}))?/;
 const durationDate = /(?:(\d+)Y)?(?:(\d+)M)?(?:(\d+)W)?(?:(\d+)D)?/;
 const durationTime = new RegExpCtor(`(?:${fraction.source}H)?(?:${fraction.source}M)?(?:${fraction.source}S)?`);
 export const duration = new RegExpCtor(`^([+-])?P${durationDate.source}(?:T(?!$)${durationTime.source})?$`, 'i');
+
+export const monthCode = /^M(?:(00)(L)|(0[1-9])(L)?|([1-9][0-9])(L)?)$/;

spec/calendar.html

@@ -82,19 +82,22 @@ <h1>
       <emu-alg>
         1. Let _monthCode_ be ? ToPrimitive(_argument_, ~string~).
         1. If _monthCode_ is not a String, throw a *TypeError* exception.
-        1. If the length of _monthCode_ is not 3 or 4, throw a *RangeError* exception.
-        1. If the first code unit of _monthCode_ is not 0x004D (LATIN CAPITAL LETTER M), throw a *RangeError* exception.
-        1. If the second code unit of _monthCode_ is not in the inclusive interval from 0x0030 (DIGIT ZERO) to 0x0039 (DIGIT NINE), throw a *RangeError* exception.
-        1. If the third code unit of _monthCode_ is not in the inclusive interval from 0x0030 (DIGIT ZERO) to 0x0039 (DIGIT NINE), throw a *RangeError* exception.
+        1. If ParseText(StringToCodePoints(_monthCode_), |MonthCode|) is a List of errors, throw a *RangeError* exception.
         1. Let _isLeapMonth_ be *false*.
         1. If the length of _monthCode_ is 4, then
-          1. If the fourth code unit of _monthCode_ is not 0x004C (LATIN CAPITAL LETTER L), throw a *RangeError* exception.
+          1. Assert: The fourth code unit of _monthCode_ is 0x004C (LATIN CAPITAL LETTER L).
           1. Set _isLeapMonth_ to *true*.
         1. Let _monthCodeDigits_ be the substring of _monthCode_ from 1 to 3.
         1. Let _monthNumber_ be ℝ(StringToNumber(_monthCodeDigits_)).
         1. If _monthNumber_ is 0 and _isLeapMonth_ is *false*, throw a *RangeError* exception.
         1. Return the Record { [[MonthNumber]]: _monthNumber_, [[IsLeapMonth]]: _isLeapMonth_ }.
       </emu-alg>
+      <emu-grammar type="definition">
+        MonthCode :::
+          `M00L`
+          `M0` NonZeroDigit `L`?
+          `M` NonZeroDigit DecimalDigit `L`?
+      </emu-grammar>
     </emu-clause>
 
     <emu-clause id="sec-temporal-createmonthcode" type="abstract operation">