Resolve Gregorian weekday/month names with CLDR inheritance

CHANGELOG.md

@@ -26,6 +26,7 @@ The format is based on [Keep a Changelog](https://keepachangelog.com/en/1.0.0/),
 - Security in case of vulnerabilities.
 
 ## [Unreleased]
+- Fix `Worldwide::Calendar::Gregorian.weekday_names` and `.month_names` returning partial results for 104 locales, including `en-CA`, `en-GB`, `en-IN` and every other `en-001` descendant. CLDR inheritance is item by item, but `I18n` returns the first node it finds in the fallback chain, so a locale that overrides a single entry (`en-CA` overrides only `Sept`) hid the rest of the entries it inherits: `month_names(locale: "en-CA", width: :abbreviated)` returned `["Sept"]` instead of all twelve names. Both methods now resolve names with CLDR's inheritance rules and raise `MissingCalendarDataError` instead of silently returning a degraded key. [#582](https://github.com/Shopify/worldwide/pull/582)
 
 ---
 

lib/worldwide/calendar/gregorian.rb

@@ -3,27 +3,51 @@
 module Worldwide
   module Calendar
     class Gregorian
+      class MissingCalendarDataError < StandardError; end
+
       class << self
         # We intentionally don't support `short` weekday names, as we haven't found anyone asking for them.
         # People usually want the full weekday name, or the abbreviated version.
         VALID_WEEKDAY_WIDTHS = [:abbreviated, :narrow, :wide].freeze
+        WEEKDAYS = [:sun, :mon, :tue, :wed, :thu, :fri, :sat].freeze
         def weekday_names(width: :wide, locale: I18n.locale)
           raise ArgumentError, "Invalid width: #{width}" unless VALID_WEEKDAY_WIDTHS.include?(width)
 
-          Worldwide::Cldr.t("calendars.gregorian.days.stand_alone.#{width}", locale: locale)
+          complete_names("calendars.gregorian.days.stand_alone.#{width}", WEEKDAYS, locale)
         end
 
         VALID_MONTH_WIDTHS = [:abbreviated, :narrow, :wide].freeze
+        MONTHS = (1..12).to_a.freeze
         def month_names(width: :wide, locale: I18n.locale)
           raise ArgumentError, "Invalid width: #{width}" unless VALID_MONTH_WIDTHS.include?(width)
 
-          Worldwide::Cldr.t("calendars.gregorian.months.stand_alone.#{width}", locale: locale).values
+          complete_names("calendars.gregorian.months.stand_alone.#{width}", MONTHS, locale).values
         end
 
         def quarter(date, locale: I18n.locale)
           format_string = Worldwide::Cldr.t("calendars.gregorian.additional_formats.yQQQ", locale: locale)
           Worldwide::Cldr::DateFormatPattern.format(date, format_string, locale: locale)
         end
+
+        private
+
+        # `stand_alone` is an alias to `format` in CLDR for most locales, and many locales
+        # override only a handful of entries, so these names have to be assembled with
+        # CLDR's inheritance rules rather than read out of a single locale's data.
+        #
+        # Anything still missing afterwards is a data bug. Raise rather than let it through:
+        # `Worldwide::Cldr.t` degrades a missing key into a humanized version of its last
+        # segment, so the alternative is silently handing back the string "wide".
+        def complete_names(key, expected_keys, locale)
+          names = Worldwide::Cldr.resolved_hash(key, locale: locale)
+          missing = expected_keys - names.keys
+
+          unless missing.empty?
+            raise MissingCalendarDataError, "Missing CLDR data for '#{key}' in locale '#{locale}': #{missing.join(", ")}"
+          end
+
+          expected_keys.to_h { |name_key| [name_key, names.fetch(name_key)] }
+        end
       end
     end
   end

lib/worldwide/cldr.rb

@@ -68,8 +68,64 @@ def config
         CONFIG
       end
 
+      # Look up a structural (hash-valued) CLDR key with CLDR's own inheritance rules.
+      #
+      # CLDR resolves locale data item by item: the effective data for a locale is the
+      # union of its own data with each of its ancestors', the more specific locale
+      # winning per item, and `<alias>` elements resolved against the locale originally
+      # requested (UTS #35 §4.1 "Multiple Inheritance" and §4.4 "Alias Elements").
+      #
+      # `I18n`'s fallback backend instead returns the first non-nil node it finds in the
+      # chain, which is correct for leaves but wrong for whole nodes: a locale that
+      # overrides a single entry hides every sibling entry it should inherit. `en-CA`
+      # only overrides September ("Sept" rather than "Sep"), so a plain `t` of the
+      # abbreviated month names returns a one-entry hash instead of all twelve.
+      #
+      # Leaf lookups should keep using `t`; only whole-node lookups need this.
+      def resolved_hash(key, locale: I18n.locale)
+        with_cldr do
+          merge_ancestors(key.to_s, locale.to_sym, Set.new)
+        end
+      end
+
       private
 
+      # Walks the CLDR fallback chain from the least specific ancestor to the most
+      # specific, so that descendants overwrite what they inherit.
+      def merge_ancestors(key, locale, seen)
+        # Guards against a cycle between two aliases pointing at each other.
+        return {} unless seen.add?([key, locale])
+
+        fallbacks[locale].reverse_each.with_object({}) do |ancestor, merged|
+          case (node = unresolved_node(key, ancestor))
+          when ::Symbol # A CLDR `<alias>`, resolved against the requested locale rather than the ancestor.
+            deep_merge!(merged, merge_ancestors(node.to_s, locale, seen))
+          when ::Hash
+            deep_merge!(merged, node)
+          end
+        end
+      end
+
+      # A single locale's own contribution: no fallbacks, no alias resolution, and no
+      # exception handler (`default: nil` makes a miss return nil instead of degrading).
+      def unresolved_node(key, locale)
+        I18n.t(key, locale: locale, default: nil, fallback: false, resolve: false)
+      rescue ::I18n::InvalidLocale
+        nil
+      end
+
+      def deep_merge!(target, source)
+        source.each do |key, value|
+          target[key] = if target[key].is_a?(::Hash) && value.is_a?(::Hash)
+            deep_merge!(target[key], value)
+          else
+            value
+          end
+        end
+
+        target
+      end
+
       def respond_to_missing?(method_name, include_private = false)
         I18n.respond_to?(method_name, include_private)
       end

test/worldwide/calendar/gregorian_test.rb

@@ -7,6 +7,8 @@ module Calendar
     class GregorianTest < ActiveSupport::TestCase
       include PluralizationHelper
 
+      WIDTHS = [:abbreviated, :narrow, :wide].freeze
+
       setup do
         @calendar = Worldwide::Calendar::Gregorian
       end
@@ -64,10 +66,65 @@ class GregorianTest < ActiveSupport::TestCase
         assert_equal "Q4 2016", Worldwide::Calendar::Gregorian.quarter(Date.new(2016, 12, 1))
       end
 
+      # `stand_alone` is an alias to `format` in CLDR for these locales, so the names only
+      # resolve if the alias in `root` is followed and then read back in the requested locale.
+      test "#weekday_names resolves the CLDR stand-alone alias" do
+        expected = { sun: "Sunday", mon: "Monday", tue: "Tuesday", wed: "Wednesday", thu: "Thursday", fri: "Friday", sat: "Saturday" }
+
+        assert_equal expected, @calendar.weekday_names(locale: :en)
+
+        expected = { sun: "søndag", mon: "mandag", tue: "tirsdag", wed: "onsdag", thu: "torsdag", fri: "fredag", sat: "lørdag" }
+
+        # `nb` holds no calendar data of its own; it inherits everything from `no`.
+        assert_equal expected, @calendar.weekday_names(locale: :nb)
+      end
+
+      test "#month_names resolves the CLDR stand-alone alias" do
+        assert_equal ["January", "February", "March", "April", "May", "June", "July", "August", "September", "October", "November", "December"], @calendar.month_names(locale: :en)
+        assert_equal ["januar", "februar", "mars", "april", "mai", "juni", "juli", "august", "september", "oktober", "november", "desember"], @calendar.month_names(locale: :nb)
+      end
+
+      # CLDR inheritance is item by item, so a locale that overrides one entry still
+      # inherits its siblings. `en-CA` only overrides September.
+      test "#month_names inherits the entries a locale does not override" do
+        expected = ["Jan", "Feb", "Mar", "Apr", "May", "Jun", "Jul", "Aug", "Sept", "Oct", "Nov", "Dec"]
+
+        assert_equal expected, @calendar.month_names(locale: :"en-CA", width: :abbreviated)
+        assert_equal expected, @calendar.month_names(locale: :"en-GB", width: :abbreviated)
+
+        # `ar-MA` overrides 7 of the 12 wide month names.
+        expected = ["يناير", "فبراير", "مارس", "أبريل", "ماي", "يونيو", "يوليوز", "غشت", "شتنبر", "أكتوبر", "نونبر", "دجنبر"]
+
+        assert_equal expected, @calendar.month_names(locale: :"ar-MA")
+      end
+
+      test "#weekday_names inherits the entries a locale does not override" do
+        # `se-FI` overrides 4 of the 7 wide weekday names.
+        expected = { sun: "sotnabeaivi", mon: "mánnodat", tue: "disdat", wed: "gaskavahkku", thu: "duorastat", fri: "bearjadat", sat: "lávvordat" }
+
+        assert_equal expected, @calendar.weekday_names(locale: :"se-FI")
+      end
+
       Worldwide::Locales.each do |locale|
         test "#quarter formatting doesn't fail (i.e., rely on date fields that have not been implemented) in #{locale}" do
           Worldwide::Calendar::Gregorian.quarter(Date.new(2016, 1, 1), locale: locale)
         end
+
+        # Guards the whole class of bug: a missing key degrades into a humanized copy of its
+        # last segment, so an unresolved lookup hands back the string "wide" rather than raising.
+        test "#weekday_names and #month_names return complete data in #{locale}" do
+          WIDTHS.each do |width|
+            weekdays = Worldwide::Calendar::Gregorian.weekday_names(width: width, locale: locale)
+
+            assert_equal [:sun, :mon, :tue, :wed, :thu, :fri, :sat], weekdays.keys, "#{locale} #{width} weekday names"
+            assert_empty weekdays.values.select { |name| name.nil? || name.empty? }, "#{locale} #{width} weekday names"
+
+            months = Worldwide::Calendar::Gregorian.month_names(width: width, locale: locale)
+
+            assert_equal 12, months.size, "#{locale} #{width} month names"
+            assert_empty months.select { |name| name.nil? || name.empty? }, "#{locale} #{width} month names"
+          end
+        end
       end
     end
   end

test/worldwide/cldr_test.rb

@@ -44,5 +44,33 @@ class CldrTest < ActiveSupport::TestCase
     ensure
       I18n.fallbacks = old_fallbacks
     end
+
+    test "#resolved_hash follows a CLDR alias and reads it back in the requested locale" do
+      # `en` has no `days.stand_alone.wide` of its own: `root` holds an alias to
+      # `days.format.wide`, which has to be resolved against `en`, not against `root`.
+      assert_nil Worldwide::Cldr.t("calendars.gregorian.days.stand_alone.wide", locale: :en, default: nil, fallback: false)
+      assert_equal "Sun", Worldwide::Cldr.t("calendars.gregorian.days.format.wide", locale: :root)[:sun]
+
+      assert_equal "Sunday", Worldwide::Cldr.resolved_hash("calendars.gregorian.days.stand_alone.wide", locale: :en)[:sun]
+    end
+
+    test "#resolved_hash merges every ancestor rather than stopping at the first match" do
+      # `en-CA` overrides only September, so `t` returns a single entry.
+      assert_equal({ 9 => "Sept" }, Worldwide::Cldr.t("calendars.gregorian.months.stand_alone.abbreviated", locale: :"en-CA"))
+
+      resolved = Worldwide::Cldr.resolved_hash("calendars.gregorian.months.stand_alone.abbreviated", locale: :"en-CA")
+
+      assert_equal (1..12).to_a, resolved.keys.sort
+      assert_equal "Sept", resolved[9]
+      assert_equal "Jan", resolved[1]
+    end
+
+    test "#resolved_hash returns an empty hash for a missing key instead of degrading it" do
+      # `t` humanizes the last segment of a key it cannot find, which is how a missing
+      # weekday lookup used to come back as the string "wide".
+      assert_equal "wide", Worldwide::Cldr.t("calendars.gregorian.__missing__.wide", locale: :en)
+
+      assert_empty Worldwide::Cldr.resolved_hash("calendars.gregorian.__missing__.wide", locale: :en)
+    end
   end
 end