gemoji 4.0.0.rc2

https://emojipedia.org/emoji-13.0/

Rakefile

@@ -20,5 +20,5 @@ namespace :db do
 end
 
 file 'vendor/unicode-emoji-test.txt' do |t|
-  system 'curl', '-fsSL', 'http://unicode.org/Public/emoji/12.0/emoji-test.txt', '-o', t.name
+  system 'curl', '-fsSL', 'http://unicode.org/Public/emoji/13.0/emoji-test.txt', '-o', t.name
 end

db/dump.rb

@@ -12,34 +12,32 @@ seen_existing = {}
 for category in categories
   for sub_category in category[:emoji]
     for emoji_item in sub_category[:emoji]
-      unicodes = emoji_item[:sequences].sort_by(&:bytesize)
-      existing_emoji = nil
-      unicodes.detect do |raw|
-        existing_emoji = Emoji.find_by_unicode(raw)
+      raw = emoji_item[:sequences][0]
+      existing_emoji = Emoji.find_by_unicode(raw) || Emoji.find_by_unicode("#{raw}\u{fe0f}")
+      if seen_existing.key?(existing_emoji)
+        existing_emoji = nil
+      else
+        seen_existing[existing_emoji] = true
       end
-      existing_emoji = nil if seen_existing.key?(existing_emoji)
+      description = emoji_item[:description].sub(/^E\d+(\.\d+)? /, '')
       output_item = {
-        emoji: unicodes[0],
-        description: emoji_item[:description],
+        emoji: raw,
+        description: description,
         category: category[:name],
       }
       if existing_emoji
-        eu = existing_emoji.unicode_aliases
-        preferred_raw = eu.size == 2 && eu[0] == "#{eu[1]}\u{fe0f}" ? eu[1] : eu[0]
         output_item.update(
-          emoji: preferred_raw,
           aliases: existing_emoji.aliases,
           tags: existing_emoji.tags,
           unicode_version: existing_emoji.unicode_version,
           ios_version: existing_emoji.ios_version,
         )
-        seen_existing[existing_emoji] = true
       else
         output_item.update(
-          aliases: [emoji_item[:description].gsub(/\W+/, '_').downcase],
+          aliases: [description.gsub(/\W+/, '_').downcase],
           tags: [],
-          unicode_version: "12.0",
-          ios_version: "13.0",
+          unicode_version: "13.0",
+          ios_version: "14.0",
         )
       end
       output_item[:skin_tones] = true if emoji_item[:skin_tones]

db/emoji-test-parser.rb

@@ -9,12 +9,8 @@ module EmojiTestParser
     "\u{1F3FE}", # medium-dark skin tone
     "\u{1F3FF}", # dark skin tone
   ]
-  HAIR_MODIFIERS = [
-    "\u{1F9B0}", # red-haired
-    "\u{1F9B1}", # curly-haired
-    "\u{1F9B2}", # bald
-    "\u{1F9B3}", # white-haired
-  ]
+  SKIN_TONES_RE = /(#{SKIN_TONES.join("|")})/o
+  SKIP_TYPES = ["unqualified", "component"]
 
   module_function
 
@@ -52,12 +48,12 @@ module EmojiTestParser
         else
           row, desc = line.split("#", 2)
           desc = desc.strip.split(" ", 2)[1]
-          codepoints, _ = row.split(";", 2)
+          codepoints, qualification = row.split(";", 2)
+          next if SKIP_TYPES.include?(qualification.strip)
           emoji_raw = codepoints.strip.split.map { |c| c.hex }.pack("U*")
-          next if HAIR_MODIFIERS.include?(emoji_raw)
           emoji_normalized = emoji_raw
             .gsub(VARIATION_SELECTOR_16, "")
-            .gsub(/(#{SKIN_TONES.join("|")})/o, "")
+            .gsub(SKIN_TONES_RE, "")
           emoji_item = emoji_map[emoji_normalized]
           if SKIN_TONES.any? { |s| emoji_raw.include?(s) }
             emoji_item[:skin_tones] = true if emoji_item
@@ -91,7 +87,7 @@ if $0 == __FILE__
     html_output = true
   end
 
-  _, categories = EmojiTestParser.parse
+  _, categories = EmojiTestParser.parse(File.expand_path("../../vendor/unicode-emoji-test.txt", __FILE__))
 
   trap(:PIPE) { abort }
 

gemoji.gemspec

@@ -1,6 +1,6 @@
 Gem::Specification.new do |s|
   s.name    = "gemoji"
-  s.version = "4.0.0.pre0"
+  s.version = "4.0.0.rc2"
   s.summary = "Unicode emoji library"
   s.description = "Character information and metadata for Unicode emoji."
 

lib/emoji.rb

@@ -68,82 +68,6 @@ module Emoji
       "\u{00ae}",  # registered
       "\u{2122}",  # trade mark
       "\u{3030}",  # wavy dash
-      "\u{263a}",  # smiling face
-      "\u{261D}",  # index pointing up
-      "\u{270C}",  # victory hand
-      "\u{270D}",  # writing hand
-      "\u{2764}",  # red heart
-      "\u{2763}",  # heavy heart exclamation
-      "\u{2668}",  # hot springs
-      "\u{2708}",  # airplane
-      "\u{2600}",  # sun
-      "\u{2601}",  # cloud
-      "\u{2602}",  # umbrella
-      "\u{2744}",  # snowflake
-      "\u{2603}",  # snowman
-      "\u{2660}",  # spade suit
-      "\u{2665}",  # heart suit
-      "\u{2666}",  # diamond suit
-      "\u{2663}",  # club suit
-      "\u{260e}",  # telephone
-      "\u{2709}",  # envelope
-      "\u{270F}",  # pencil
-      "\u{2712}",  # black nib
-      "\u{2702}",  # scissors
-      "\u{26a0}",  # warning
-      "\u{2B06}",  # up arrow
-      "\u{2197}",  # up-right arrow
-      "\u{27A1}",  # right arrow
-      "\u{2198}",  # down-right arrow
-      "\u{2B07}",  # down arrow
-      "\u{2199}",  # down-left arrow
-      "\u{2B05}",  # left arrow
-      "\u{2196}",  # up-left arrow
-      "\u{2195}",  # up-down arrow
-      "\u{2194}",  # left-right arrow
-      "\u{21A9}",  # right arrow curving left
-      "\u{21AA}",  # left arrow curving right
-      "\u{2934}",  # right arrow curving up
-      "\u{2935}",  # right arrow curving down
-      "\u{2721}",  # star of David
-      "\u{262F}",  # yin yang
-      "\u{271D}",  # latin cross
-      "\u{25B6}",  # play button
-      "\u{25C0}",  # reverse button
-      "\u{23CF}",  # eject button
-      "\u{2640}",  # female sign
-      "\u{2642}",  # male sign
-      "\u{267B}",  # recycling symbol
-      "\u{2611}",  # ballot box with check
-      "\u{2714}",  # heavy check mark
-      "\u{2716}",  # heavy multiplication x
-      "\u{303D}",  # part alternation mark
-      "\u{2733}",  # eight-spoked asterisk
-      "\u{2734}",  # eight-pointed star
-      "\u{2747}",  # sparkle
-      "\u{203C}",  # double exclamation mark
-      "\u{2049}",  # exclamation question mark
-      "\u{23}\u{20E3}",  # keycap: #
-      "\u{2A}\u{20E3}",  # keycap: *
-      "\u{30}\u{20E3}",  # keycap: 0
-      "\u{31}\u{20E3}",  # keycap: 1
-      "\u{32}\u{20E3}",  # keycap: 2
-      "\u{33}\u{20E3}",  # keycap: 3
-      "\u{34}\u{20E3}",  # keycap: 4
-      "\u{35}\u{20E3}",  # keycap: 5
-      "\u{36}\u{20E3}",  # keycap: 6
-      "\u{37}\u{20E3}",  # keycap: 7
-      "\u{38}\u{20E3}",  # keycap: 8
-      "\u{39}\u{20E3}",  # keycap: 9
-      "\u{2139}",  # information
-      "\u{24C2}",  # circled M
-      "\u{1F17F}", # P button
-      "\u{3297}",  # Japanese “congratulations” button
-      "\u{3299}",  # Japanese “secret” button
-      "\u{25AA}",  # black small square
-      "\u{25AB}",  # white small square
-      "\u{25FB}",  # white medium square
-      "\u{25FC}",  # black medium square
     ].freeze
 
     private_constant :VARIATION_SELECTOR_16, :TEXT_GLYPHS

test/emoji_test.rb

@@ -50,7 +50,6 @@ class EmojiTest < TestCase
   GENDER_EXCEPTIONS = [
     "man_with_gua_pi_mao",
     "woman_with_headscarf",
-    "man_in_tuxedo",
     "pregnant_woman",
     "isle_of_man",
     "blonde_woman",
@@ -91,18 +90,14 @@ class EmojiTest < TestCase
   test "missing or incorrect unicodes" do
     emoji_map, _ = EmojiTestParser.parse(File.expand_path("../../vendor/unicode-emoji-test.txt", __FILE__))
     source_unicode_emoji = emoji_map.values
+    supported_sequences = Emoji.all.flat_map(&:unicode_aliases)
     text_glyphs = Emoji.const_get(:TEXT_GLYPHS)
 
     missing = 0
     message = "Missing or incorrect unicodes:\n"
     source_unicode_emoji.each do |emoji|
       emoji[:sequences].each do |raw|
-        found = Emoji.find_by_unicode(raw)
-        if text_glyphs.include?(raw)
-          assert_nil found, Emoji::Character.hex_inspect(raw)
-          next
-        end
-        next if found
+        next if text_glyphs.include?(raw) || Emoji.find_by_unicode(raw)
         message << "%s (%s)" % [Emoji::Character.hex_inspect(raw), emoji[:description]]
         if found = Emoji.find_by_unicode(raw.gsub("\u{fe0f}", ""))
           message << " - could be %s (:%s:)" % [found.hex_inspect, found.name]
@@ -115,13 +110,6 @@ class EmojiTest < TestCase
     assert_equal 0, missing, message
   end
 
-  test "raw representation does not include VARIATION SELECTOR 16 unless necessary" do
-    emoji = Emoji.all.select do |emoji|
-      !emoji.custom? && emoji.raw.end_with?("\u{fe0f}") && emoji.unicode_aliases.size == 2
-    end
-    assert_equal [], emoji
-  end
-
   test "emoji have category" do
     missing = Emoji.all.select { |e| e.category.to_s.empty? }
     assert_equal [], missing.map(&:name), "some emoji don't have a category"