Normative: Allow annotations after YYYY-MM and MM-DD

In keeping with the IXDTF format which extends the grammar of RFC 3339 with any number of annotations, we should allow annotations (time zone, calendar, and unknown after #2397 lands) after the short YYYY-MM PlainYearMonth and MM-DD PlainMonthDay forms. If we were to allow UTC offsets ±UU[:UU] alongside the time zone annotation, that would be ambiguous in one case: YYYY-MM-UU would be ambiguous with YYYY-MM-DD. This PR makes the following choices: - UTC offset is allowed after MM-DD - UTC offset is disallowed after YYYY-MM (even if it is positive, or contains a minute component, which would not be ambiguous) Other choices are certainly possible. Closes: #2379
tc39 · Sep 1, 2022 · 03bff3f · 03bff3f
1 parent b4d9abc
commit 03bff3f
Show file tree

Hide file tree

Showing 4 changed files with 64 additions and 23 deletions.
diff --git a/polyfill/lib/ecmascript.mjs b/polyfill/lib/ecmascript.mjs
@@ -18,7 +18,6 @@ const ObjectDefineProperty = Object.defineProperty;
 const ObjectGetOwnPropertyDescriptor = Object.getOwnPropertyDescriptor;
 const ObjectIs = Object.is;
 const ObjectEntries = Object.entries;
-const StringPrototypeSlice = String.prototype.slice;
 
 import bigInt from 'big-integer';
 import Call from 'es-abstract/2020/Call.js';
@@ -388,18 +387,12 @@ export const ES = ObjectAssign({}, ES2020, {
       return { hour, minute, second, millisecond, microsecond, nanosecond, calendar };
     }
     // Reject strings that are ambiguous with PlainMonthDay or PlainYearMonth.
-    // The calendar suffix is `[u-ca=${calendar}]`, i.e. calendar plus 7 characters,
-    // and must be stripped so presence of a calendar doesn't result in interpretation
-    // of otherwise ambiguous input as a time.
-    const isoStringWithoutCalendar = calendar
-      ? ES.Call(StringPrototypeSlice, isoString, [0, -(calendar.length + 7)])
-      : isoString;
     try {
-      const { month, day } = ES.ParseTemporalMonthDayString(isoStringWithoutCalendar);
+      const { month, day } = ES.ParseTemporalMonthDayString(isoString);
       ES.RejectISODate(1972, month, day);
     } catch {
       try {
-        const { year, month } = ES.ParseTemporalYearMonthString(isoStringWithoutCalendar);
+        const { year, month } = ES.ParseTemporalYearMonthString(isoString);
         ES.RejectISODate(year, month, 1);
       } catch {
         return { hour, minute, second, millisecond, microsecond, nanosecond, calendar };
@@ -417,6 +410,9 @@ export const ES = ObjectAssign({}, ES2020, {
       year = ES.ToInteger(yearString);
       month = ES.ToInteger(match[2]);
       calendar = match[3];
+      if (calendar !== undefined && calendar !== 'iso8601') {
+        throw new RangeError('YYYY-MM format is only valid with iso8601 calendar');
+      }
     } else {
       let z;
       ({ year, month, calendar, day: referenceISODay, z } = ES.ParseISODateTime(isoString));
@@ -430,6 +426,10 @@ export const ES = ObjectAssign({}, ES2020, {
     if (match) {
       month = ES.ToInteger(match[1]);
       day = ES.ToInteger(match[2]);
+      calendar = match[10];
+      if (calendar !== undefined && calendar != 'iso8601') {
+        throw new RangeError('MM-DD format is only valid with iso8601 calendar');
+      }
     } else {
       let z;
       ({ month, day, calendar, year: referenceISOYear, z } = ES.ParseISODateTime(isoString));
@@ -1628,7 +1628,11 @@ export const ES = ObjectAssign({}, ES2020, {
     try {
       ({ calendar } = ES.ParseISODateTime(identifier));
     } catch {
-      throw new RangeError(`Invalid calendar: ${identifier}`);
+      try {
+        ({ calendar } = ES.ParseTemporalYearMonthString(identifier));
+      } catch {
+        ({ calendar } = ES.ParseTemporalMonthDayString(identifier));
+      }
     }
     if (!calendar) calendar = 'iso8601';
     return new TemporalCalendar(calendar);

diff --git a/polyfill/lib/regex.mjs b/polyfill/lib/regex.mjs
@@ -36,14 +36,21 @@ export const zoneddatetime = new RegExp(
 
 export const time = new RegExp(`^T?${timesplit.source}(?:${zonesplit.source})?(?:${calendar.source})?$`, 'i');
 
-// The short forms of YearMonth and MonthDay are only for the ISO calendar.
+// The short forms of YearMonth and MonthDay are only for the ISO calendar, but
+// a calendar annotation is still allowed and will throw if not ISO.
 // Non-ISO calendar YearMonth and MonthDay have to parse as a Temporal.PlainDate,
 // with the reference fields.
 // YYYYMM forbidden by ISO 8601 because ambiguous with YYMMDD, but allowed by
 // RFC 3339 and we don't allow 2-digit years, so we allow it.
 // Not ambiguous with HHMMSS because that requires a 'T' prefix
-export const yearmonth = new RegExp(`^(${yearpart.source})-?(${monthpart.source})$`);
-export const monthday = new RegExp(`^(?:--)?(${monthpart.source})-?(${daypart.source})$`);
+// In YearMonth, a time zone annotation is allowed but no UTC offset, because
+// YYYY-MM-UU is ambiguous with YYYY-MM-DD
+export const yearmonth = new RegExp(
+  `^(${yearpart.source})-?(${monthpart.source})(?:\\[${timeZoneID.source}\\])?(?:${calendar.source})?$`
+);
+export const monthday = new RegExp(
+  `^(?:--)?(${monthpart.source})-?(${daypart.source})(?:${zonesplit.source})?(?:${calendar.source})?$`
+);
 
 const fraction = /(\d+)(?:[.,](\d{1,9}))?/;
 

diff --git a/polyfill/test/validStrings.mjs b/polyfill/test/validStrings.mjs
@@ -281,10 +281,10 @@ const timeSpec = seq(
   choice([':', timeMinute, [':', timeSecond, [timeFraction]]], seq(timeMinute, [timeSecond, [timeFraction]]))
 );
 const timeSpecWithOptionalTimeZoneNotAmbiguous = withSyntaxConstraints(seq(timeSpec, [timeZone]), (result) => {
-  if (/^(?:(?!02-?30)(?:0[1-9]|1[012])-?(?:0[1-9]|[12][0-9]|30)|(?:0[13578]|10|12)-?31)$/.test(result)) {
+  if (/^(?:(?!02-?30)(?:0[1-9]|1[012])-?(?:0[1-9]|[12][0-9]|30)|(?:0[13578]|10|12)-?31)/.test(result)) {
     throw new SyntaxError('valid PlainMonthDay');
   }
-  if (/^(?![−-]000000)(?:[0-9]{4}|[+−-][0-9]{6})-?(?:0[1-9]|1[012])$/.test(result)) {
+  if (/^(?![−-]000000)(?:[0-9]{4}|[+−-][0-9]{6})-?(?:0[1-9]|1[012])/.test(result)) {
     throw new SyntaxError('valid PlainYearMonth');
   }
 });
@@ -306,6 +306,19 @@ const calendarTime = choice(
   seq(timeDesignator, timeSpec, [timeZone], [calendar]),
   seq(timeSpecWithOptionalTimeZoneNotAmbiguous, [calendar])
 );
+const annotatedYearMonth = withSyntaxConstraints(
+  seq(dateSpecYearMonth, [timeZoneBracketedAnnotation], [calendar]),
+  (result, data) => {
+    if (data.calendar !== undefined && data.calendar !== 'iso8601') {
+      throw new SyntaxError('retry if YYYY-MM with non-ISO calendar');
+    }
+  }
+);
+const annotatedMonthDay = withSyntaxConstraints(seq(dateSpecMonthDay, [timeZone], [calendar]), (result, data) => {
+  if (data.calendar !== undefined && data.calendar !== 'iso8601') {
+    throw new SyntaxError('retry if MM-DD with non-ISO calendar');
+  }
+});
 
 const durationFractionalPart = withCode(between(1, 9, digit()), (data, result) => {
   const fraction = result.padEnd(9, '0');
@@ -368,10 +381,10 @@ const goals = {
   Date: calendarDateTime,
   DateTime: calendarDateTime,
   Duration: duration,
-  MonthDay: choice(dateSpecMonthDay, calendarDateTime),
+  MonthDay: choice(annotatedMonthDay, calendarDateTime),
   Time: choice(calendarTime, calendarDateTimeTimeRequired),
   TimeZone: choice(timeZoneIdentifier, seq(date, [timeSpecSeparator], timeZone, [calendar])),
-  YearMonth: choice(dateSpecYearMonth, calendarDateTime),
+  YearMonth: choice(annotatedYearMonth, calendarDateTime),
   ZonedDateTime: zonedDateTime
 };
 

diff --git a/spec/abstractops.html b/spec/abstractops.html
@@ -1096,8 +1096,14 @@ <h1>ISO 8601 grammar</h1>
           TimeHour `:` TimeMinute `:` TimeSecond TimeFraction?
           TimeHour TimeMinute TimeSecond TimeFraction?
 
+      AmbiguousMonthDay :
+          ValidMonthDay TimeZone?
+
+      AmbiguousYearMonth :
+          DateSpecYearMonth TimeZoneBracketedAnnotation?
+
       TimeSpecWithOptionalTimeZoneNotAmbiguous :
-          TimeSpec TimeZone? but not one of ValidMonthDay or DateSpecYearMonth
+          TimeSpec TimeZone? but not one of AmbiguousMonthDay or AmbiguousYearMonth
 
       TimeSpecSeparator :
           DateTimeSeparator TimeSpec
@@ -1115,6 +1121,12 @@ <h1>ISO 8601 grammar</h1>
       CalendarDateTimeTimeRequired :
           Date TimeSpecSeparator TimeZone? Calendar?
 
+      AnnotatedYearMonth:
+          DateSpecYearMonth TimeZoneBracketedAnnotation? Calendar?
+
+      AnnotatedMonthDay:
+          DateSpecMonthDay TimeZone? Calendar?
+
       DurationWholeSeconds :
           DecimalDigits[~Sep]
 
@@ -1195,15 +1207,15 @@ <h1>ISO 8601 grammar</h1>
           Duration
 
       TemporalMonthDayString :
-          DateSpecMonthDay
+          AnnotatedMonthDay
           CalendarDateTime
 
       TemporalTimeString :
           CalendarTime
           CalendarDateTimeTimeRequired
 
       TemporalYearMonthString :
-          DateSpecYearMonth
+          AnnotatedYearMonth
           CalendarDateTime
 
       TemporalZonedDateTimeString :
@@ -1214,8 +1226,8 @@ <h1>ISO 8601 grammar</h1>
           TemporalInstantString
           CalendarDateTime
           CalendarTime
-          DateSpecYearMonth
-          DateSpecMonthDay
+          AnnotatedYearMonth
+          AnnotatedMonthDay
     </emu-grammar>
     <ul>
       <li>
@@ -1237,8 +1249,13 @@ <h1>
     <emu-note>The value of ! ToIntegerOrInfinity(*undefined*) is 0.</emu-note>
     <emu-alg>
       1. Let _parseResult_ be ~empty~.
-      1. For each nonterminal _goal_ of &laquo; |TemporalDateTimeString|, |TemporalInstantString|, |TemporalMonthDayString|, |TemporalTimeString|, |TemporalYearMonthString|, |TemporalZonedDateTimeString| &raquo;, do
+      1. For each nonterminal _goal_ of &laquo; |TemporalDateTimeString|, |TemporalInstantString|, |TemporalTimeString|, |TemporalZonedDateTimeString| &raquo;, do
         1. If _parseResult_ is not a Parse Node, set _parseResult_ to ParseText(StringToCodePoints(_isoString_), _goal_).
+      1. For each nonterminal _goal_ of &laquo; |TemporalMonthDayString|, |TemporalYearMonthString| &raquo;, do
+        1. If _parseResult_ is not a Parse Node, then
+          1. Set _parseResult_ to ParseText(StringToCodePoints(_isoString_), _goal_).
+          1. If _parseResult_ contains a |CalendarName| Parse Node _c_, and CodePointsToString(_c_) is not *"iso8601"*, then
+            1. Throw a *RangeError* exception.
       1. If _parseResult_ is not a Parse Node, throw a *RangeError* exception.
       1. Let each of _year_, _month_, _day_, _hour_, _minute_, _second_, _fSeconds_, and _calendar_ be the source text matched by the respective |DateYear|, |DateMonth|, |DateDay|, |TimeHour|, |TimeMinute|, |TimeSecond|, |TimeFraction|, and |CalendarName| Parse Node contained within _parseResult_, or an empty sequence of code points if not present.
       1. If the first code point of _year_ is U+2212 (MINUS SIGN), replace the first code point with U+002D (HYPHEN-MINUS).