Accept Etc/GMT* as a valid time zone id.

Etc/GMT* time zones are  listed in the INAN time zone database and
they should be accepted as valid.

This CL will be followed by a CL for moving time zone name checks to C++
that will accept all the time zone names (e.g. EST5EDT, Hongkong, ROK, Zulu).

Bug: chromium:364374
Test: intl/date-format/timezone.js
Test: mjsunit/regress/regress-crbug-364374
Cq-Include-Trybots: luci.v8.try:v8_linux_noi18n_rel_ng
Change-Id: If0e5327d7e980504a9cb3d2b641e907ebce61180
Reviewed-on: https://chromium-review.googlesource.com/1159546
Commit-Queue: Jungshik Shin <jshin@chromium.org>
Reviewed-by: Sathya Gunasekaran <gsathya@chromium.org>
Cr-Commit-Position: refs/heads/master@{#54879}

src/js/intl.js

@@ -194,6 +194,7 @@ function GetServiceRE() {
  * Matches valid IANA time zone names.
  */
 var TIMEZONE_NAME_CHECK_RE = UNDEFINED;
+var GMT_OFFSET_TIMEZONE_NAME_CHECK_RE = UNDEFINED;
 
 function GetTimezoneNameCheckRE() {
   if (IS_UNDEFINED(TIMEZONE_NAME_CHECK_RE)) {
@@ -203,6 +204,14 @@ function GetTimezoneNameCheckRE() {
   return TIMEZONE_NAME_CHECK_RE;
 }
 
+function GetGMTOffsetTimezoneNameCheckRE() {
+  if (IS_UNDEFINED(GMT_OFFSET_TIMEZONE_NAME_CHECK_RE)) {
+     GMT_OFFSET_TIMEZONE_NAME_CHECK_RE = new GlobalRegExp(
+         '^(?:ETC/GMT)(?<offset>0|[+-](?:[0-9]|1[0-4]))$');
+  }
+  return GMT_OFFSET_TIMEZONE_NAME_CHECK_RE;
+}
+
 /**
  * Matches valid location parts of IANA time zone names.
  */
@@ -1510,7 +1519,7 @@ function CreateDateTimeFormat(locales, options) {
     {__proto__: null, skeleton: ldmlString, timeZone: tz}, resolved);
 
   if (resolved.timeZone === "Etc/Unknown") {
-    throw %make_range_error(kUnsupportedTimeZone, tz);
+    throw %make_range_error(kInvalidTimeZone, tz);
   }
 
   %MarkAsInitializedIntlObjectOfType(dateFormat, DATE_TIME_FORMAT_TYPE);
@@ -1644,18 +1653,28 @@ function canonicalizeTimeZoneID(tzID) {
     return 'UTC';
   }
 
-  // TODO(jshin): Add support for Etc/GMT[+-]([1-9]|1[0-2])
-
   // We expect only _, '-' and / beside ASCII letters.
-  // All inputs should conform to Area/Location(/Location)* from now on.
-  var match = %regexp_internal_match(GetTimezoneNameCheckRE(), tzID);
-  if (IS_NULL(match)) throw %make_range_error(kExpectedTimezoneID, tzID);
+  // All inputs should conform to Area/Location(/Location)*, or Etc/GMT* .
+  // TODO(jshin): 1. Support 'GB-Eire", 'EST5EDT", "ROK', 'US/*', 'NZ' and many
+  // other aliases/linked names when moving timezone validation code to C++.
+  // See crbug.com/364374 and crbug.com/v8/8007 .
+  // 2. Resolve the difference betwee CLDR/ICU and IANA time zone db.
+  // See http://unicode.org/cldr/trac/ticket/9892 and crbug.com/645807 .
+  let match = %regexp_internal_match(GetTimezoneNameCheckRE(), tzID);
+  if (IS_NULL(match)) {
+    let match =
+      %regexp_internal_match(GetGMTOffsetTimezoneNameCheckRE(), upperID);
+     if (!IS_NULL(match) && match.length == 2)
+       return "Etc/GMT" + match.groups.offset;
+     else
+       throw %make_range_error(kInvalidTimeZone, tzID);
+  }
 
-  var result = toTitleCaseTimezoneLocation(match[1]) + '/' +
+  let result = toTitleCaseTimezoneLocation(match[1]) + '/' +
                toTitleCaseTimezoneLocation(match[2]);
 
   if (!IS_UNDEFINED(match[3]) && 3 < match.length) {
-    var locations = %StringSplit(match[3], '/', kMaxUint32);
+    let locations = %StringSplit(match[3], '/', kMaxUint32);
     // The 1st element is empty. Starts with i=1.
     for (var i = 1; i < locations.length; i++) {
       result = result + '/' + toTitleCaseTimezoneLocation(locations[i]);

src/messages.h

@@ -533,8 +533,6 @@ class ErrorUtils : public AllStatic {
   T(BigIntNegativeExponent, "Exponent must be positive")                       \
   T(BigIntTooBig, "Maximum BigInt size exceeded")                              \
   T(DateRange, "Provided date is not in valid range.")                         \
-  T(ExpectedTimezoneID,                                                        \
-    "Expected Area/Location(/Location)* for time zone, got %")                 \
   T(ExpectedLocation,                                                          \
     "Expected letters optionally connected with underscores or hyphens for "   \
     "a location, got %")                                                       \
@@ -556,6 +554,7 @@ class ErrorUtils : public AllStatic {
   T(InvalidWeakSetValue, "Invalid value used in weak set")                     \
   T(InvalidStringLength, "Invalid string length")                              \
   T(InvalidTimeValue, "Invalid time value")                                    \
+  T(InvalidTimeZone, "Invalid time zone specified: %")                         \
   T(InvalidTypedArrayAlignment, "% of % should be a multiple of %")            \
   T(InvalidTypedArrayIndex, "Invalid typed array index")                       \
   T(InvalidTypedArrayLength, "Invalid typed array length: %")                  \
@@ -576,7 +575,6 @@ class ErrorUtils : public AllStatic {
   T(ToRadixFormatRange, "toString() radix argument must be between 2 and 36")  \
   T(TypedArraySetOffsetOutOfBounds, "offset is out of bounds")                 \
   T(TypedArraySetSourceTooLarge, "Source is too large")                        \
-  T(UnsupportedTimeZone, "Unsupported time zone specified %")                  \
   T(ValueOutOfRange, "Value % out of range for % options property %")          \
   /* SyntaxError */                                                            \
   T(AmbiguousExport,                                                           \

src/objects/intl-objects.cc

@@ -225,11 +225,10 @@ void SetResolvedDateSettings(Isolate* isolate, const icu::Locale& icu_locale,
   if (U_SUCCESS(status)) {
     // In CLDR (http://unicode.org/cldr/trac/ticket/9943), Etc/UTC is made
     // a separate timezone ID from Etc/GMT even though they're still the same
-    // timezone. We'd not have "Etc/GMT" here because we canonicalize it and
-    // other GMT-variants to "UTC" in intl.js and "UTC" is turned to "Etc/UTC"
-    // by ICU before getting here.
-    // TODO(jshin): Figure out the cause of crbug.com/719609 and re-enable
-    //  DCHECK(canonical_time_zone != UNICODE_STRING_SIMPLE("Etc/GMT")) .
+    // timezone. We have Etc/UTC because 'UTC', 'Etc/Universal',
+    // 'Etc/Zulu' and others are turned to 'Etc/UTC' by ICU. Etc/GMT comes
+    // from Etc/GMT0, Etc/GMT+0, Etc/GMT-0, Etc/Greenwich.
+    // ecma402##sec-canonicalizetimezonename step 3
     if (canonical_time_zone == UNICODE_STRING_SIMPLE("Etc/UTC") ||
         canonical_time_zone == UNICODE_STRING_SIMPLE("Etc/GMT")) {
       JSObject::SetProperty(

test/intl/date-format/timezone.js

@@ -31,11 +31,15 @@
 // var df = Intl.DateTimeFormat();
 // assertEquals(getDefaultTimeZone(), df.resolvedOptions().timeZone);
 
-df = Intl.DateTimeFormat(undefined, {timeZone: 'UtC'});
-assertEquals('UTC', df.resolvedOptions().timeZone);
+[
+  'UtC', 'gmt', 'Etc/UTC', 'Etc/GMT', 'Etc/GMT0', 'Etc/GMT+0',
+  'etc/gmt-0', 'etc/zulu', 'Etc/universal', 'etc/greenwich'
+].forEach((timezone) => {
+  const df = Intl.DateTimeFormat(undefined, {timeZone: timezone});
+  assertEquals('UTC', df.resolvedOptions().timeZone);
+})
 
-df = Intl.DateTimeFormat(undefined, {timeZone: 'gmt'});
-assertEquals('UTC', df.resolvedOptions().timeZone);
+// See test/mjsunit/regress/regress-crbug-364374.js for additional/ tests.
 
 df = Intl.DateTimeFormat(undefined, {timeZone: 'America/Los_Angeles'});
 assertEquals('America/Los_Angeles', df.resolvedOptions().timeZone);
@@ -43,22 +47,29 @@ assertEquals('America/Los_Angeles', df.resolvedOptions().timeZone);
 df = Intl.DateTimeFormat(undefined, {timeZone: 'Europe/Belgrade'});
 assertEquals('Europe/Belgrade', df.resolvedOptions().timeZone);
 
-// Check Etc/XXX variants. They should work too.
-df = Intl.DateTimeFormat(undefined, {timeZone: 'Etc/UTC'});
-assertEquals('UTC', df.resolvedOptions().timeZone);
-
-df = Intl.DateTimeFormat(undefined, {timeZone: 'Etc/GMT'});
-assertEquals('UTC', df.resolvedOptions().timeZone);
-
 df = Intl.DateTimeFormat(undefined, {timeZone: 'euRope/beLGRade'});
 assertEquals('Europe/Belgrade', df.resolvedOptions().timeZone);
 
+// Etc/GMT-14 to Etc/GMT+12 are valid.
+df = Intl.DateTimeFormat(undefined, {timeZone: 'etc/gmt+12'});
+assertEquals('Etc/GMT+12', df.resolvedOptions().timeZone);
+
+df = Intl.DateTimeFormat(undefined, {timeZone: 'etc/gmt+9'});
+assertEquals('Etc/GMT+9', df.resolvedOptions().timeZone);
+
+df = Intl.DateTimeFormat(undefined, {timeZone: 'etc/gmt-9'});
+assertEquals('Etc/GMT-9', df.resolvedOptions().timeZone);
+
+df = Intl.DateTimeFormat(undefined, {timeZone: 'etc/gmt-14'});
+assertEquals('Etc/GMT-14', df.resolvedOptions().timeZone);
+
+assertThrows('Intl.DateTimeFormat(undefined, {timeZone: \'Etc/GMT+13\'})');
+
 // : + - are not allowed, only / _ are.
 assertThrows('Intl.DateTimeFormat(undefined, {timeZone: \'GMT+07:00\'})');
 assertThrows('Intl.DateTimeFormat(undefined, {timeZone: \'GMT+0700\'})');
 assertThrows('Intl.DateTimeFormat(undefined, {timeZone: \'GMT-05:00\'})');
 assertThrows('Intl.DateTimeFormat(undefined, {timeZone: \'GMT-0500\'})');
-assertThrows('Intl.DateTimeFormat(undefined, {timeZone: \'Etc/GMT+0\'})');
 assertThrows('Intl.DateTimeFormat(undefined, ' +
     '{timeZone: \'America/Los-Angeles\'})');