emoji-datasource 14.0.0 → 14.0.2

checksums.yaml

@@ -1,7 +1,7 @@
 ---
 SHA256:
-  metadata.gz: b3935f8318904fee0e0c3e662d843c7720e6ec2e7df4dafe98ffe7365d3f6236
-  data.tar.gz: a7c9b43189f238363cf275d68037970d5f3d5f03c379e6415bab44bab25ff938
+  metadata.gz: 705920a1ffdf8ca2028830eeb6420d94d23955b1c189613011e666c6028b9a45
+  data.tar.gz: bb399eaa99c6448fdda1826fae0921c86ad1f261aa2ec892e4de4815e70e197d
 SHA512:
-  metadata.gz: 2049e3e19cb85aa80dc0fe47604d3f715b02977d50cdd9cd345aecdd949f7dbaca7cc15c7a0e26078bc04803fe3c3cc4c1fd63e6f2822e3cf00cf2ed161a3aaa
-  data.tar.gz: ee10a8cb1a4cb899a659700363956aa97ab3994c74ce563fa7cd461a18e2ab92de669163e5c25df20d2c780e332084b4ef402127798d4bd78e32ecbbd130fce8
+  metadata.gz: 6518b1ac91acc735b23a63bc7405dab7ed4d1c8928a0f850656b9278cd1e32a007d1ded4568b06a68849067faad2e8b7ee52131e1cc1912498ea555ab7aece8e
+  data.tar.gz: fa032904d287a9bd0d8a9e1614c6e86dc2f6e90b1b162bcdc357b328e6b187ba2ff0f3069d3bc4836453eb21acaa6688b6bb58db314ed767f4db2019807c8348

data/.rubocop.yml

@@ -40,6 +40,9 @@ Metrics/BlockLength:
     - '*.gemspec'
     - 'spec/**/*_spec.rb'
 
+Metrics/AbcSize:
+  Max: 18
+
 Lint/MissingSuper:
   Enabled: false
 

data/README.md

@@ -22,6 +22,21 @@ Find emoji by short name
 
 ```ruby
 EmojiDatasource.find_by_short_name('+1')
+#=> EmojiDatasource::Emoji: :+1: (👍)
+```
+
+Find emoji by raw string:
+
+```ruby
+EmojiDatasource.find_by_char('👍🏾')
+#=> EmojiDatasource::Emoji: :+1::skin-tone-5: (👍🏾)
+```
+
+Find emoji by unicode hex character code:
+
+```ruby
+EmojiDatasource.find_by_unified("1F469-200D-2764-FE0F-200D-1F468")
+#=> EmojiDatasource::Emoji: :woman-heart-man: (👩‍❤️‍👨)
 ```
 
 Convert emoji short name to character
@@ -33,9 +48,17 @@ EmojiDatasource.short_name_to_char('+1') # => 👍
 this also supports skin variations
 
 ```ruby
-EmojiDatasource.short_name_to_char('+1::skin-tone-2') # => 👍
+EmojiDatasource.short_name_to_char('+1::skin-tone-2') # => 👍🏻
 ```
 
+Get base emoji for skin variation:
+
+```ruby
+emoji = EmojiDatasource.find_by_short_name(':+1::skin-tone-5:')
+#=> EmojiDatasource::Emoji: :+1::skin-tone-5: (👍🏾)
+emoji.base
+#=> EmojiDatasource::Emoji: :+1: (👍)
+```
 
 ## Supported Ruby Versions
 

data/lib/emoji_datasource/emoji.rb

@@ -2,38 +2,86 @@
 
 module EmojiDatasource
   class Emoji
-    attr_reader :data
+    attr_reader :data, :unified
 
-    def initialize(data)
+    def initialize(data, variation: nil)
       @data = data
+      @variation = variation
+      @unified = variation ? @data.dig('skin_variations', variation, 'unified') : @data['unified']
     end
 
+    # Raw emoji string
     def to_char
-      EmojiDatasource.unified_to_char(data[:unified])
+      EmojiDatasource.unified_to_char(unified)
     end
 
-    def skin_variations
-      return unless @data[:skin_variations]
+    # Official emoji name (pretty long)
+    def name
+      return @data['name'] unless @variation
 
-      @data[:skin_variations].transform_keys(&:to_s)
+      "#{@data['name']} (#{variation_emojis.map(&:name).join(', ')})"
+    end
+
+    # Short code that often can be used to find emoji in pickers and chat apps
+    def short_name
+      return @data['short_name'] unless @variation
+
+      "#{@data['short_name']}::#{variation_emojis.map(&:short_name).join('::')}"
+    end
+
+    # All known short names (base +short_name+ and maybe some aliases)
+    def short_names
+      return @data['short_names'] unless @variation
+
+      @short_names ||= @data['short_names'].flat_map do |short_name|
+        full_short_name = "#{short_name}::#{variation_emojis.map(&:short_name).join('::')}"
+
+        # Allow to find complex emojis with multiple equal skin tone codes by single skin tone
+        # For some emojis (like :women_holding_hands:) there are special single skin tone emojis.
+        # but for others (like :people_holding_hands:) there are no such versions.
+        # However, some implementations in the wild (e.g. Slack and EmojiMart) understand codes
+        # containing only single skin tone (e.g. `:people_holding_hands::skin-tone-N:`)
+        if variation_emojis.size > 1 && variation_emojis.uniq.size == 1
+          abbr_short_name = "#{short_name}::#{variation_emojis.first.short_name}"
+        end
+
+        [full_short_name, abbr_short_name].compact
+      end
+    end
+
+    # All known skin tone variations
+    def variations
+      return @variations = [] if @variation
+
+      @variations ||= @data.fetch('skin_variations', {}).each_key.map do |key|
+        self.class.new(data, variation: key)
+      end
+    end
+
+    # Base emoji, without variations applied.
+    # E.g. for `:+1::skin-tone-2:` base will be just `:+1:`
+    def base
+      return self unless @variation
+
+      EmojiDatasource.find_by_unified(@data['unified'])
     end
 
     def method_missing(method_name, *arguments, &block)
-      if @data.key?(method_name)
-        @data[method_name]
+      if @data.key?(method_name.to_s)
+        @data[method_name.to_s]
       else
         super
       end
     end
 
     def respond_to_missing?(method_name, include_private = false)
-      return true if @data.key?(method_name)
+      return true if @data.key?(method_name.to_s)
 
       super
     end
 
     def inspect
-      "#{self.class.name}:#{short_name}"
+      "#{self.class.name}: :#{short_name}: (#{to_char})"
     end
 
     def to_s
@@ -47,5 +95,15 @@ module EmojiDatasource
     def to_json(**args)
       @data.to_json(**args)
     end
+
+    private
+
+    def variation_emojis
+      return [] unless @variation
+
+      @variation_emojis ||= @variation.split('-').map do |unified|
+        EmojiDatasource.find_by_unified(unified)
+      end
+    end
   end
 end

data/lib/emoji_datasource/version.rb

@@ -1,5 +1,5 @@
 # frozen_string_literal: true
 
 module EmojiDatasource
-  VERSION = '14.0.0'
+  VERSION = '14.0.2'
 end

data/lib/emoji_datasource.rb

@@ -4,23 +4,50 @@ require 'json'
 
 require_relative 'emoji_datasource/version'
 require_relative 'emoji_datasource/emoji'
-require_relative 'emoji_datasource/short_name_to_char'
 
 module EmojiDatasource
   class Error < StandardError; end
 
   EMOJI_DATA_PATH = File.join(__dir__, '..', 'vendor', 'emoji-datasource', 'emoji.json')
 
+  # @example
+  #   short_name_to_char('+1') #=> "👍"
   def self.short_name_to_char(name)
-    EmojiDatasource::ShortNameToChar.generate(name)
+    find_by_short_name(name)&.to_char
   end
 
+  # Finds emoji by short code (e.g. `+1`, `:+1:`, `:+1::skin-tone-5:`)
+  # @param name [String] short code with or without wrapping colons.
+  # @return [EmojiDatasource::Emoji] if there is an emoji matching +name+
+  # @return [nil] if there are no emojis matching +name+
+  # @example
+  #   find_by_short_name('+1') #=> EmojiDatasource::Emoji: :+1: (👍)
   def self.find_by_short_name(name)
     return unless name
 
-    data.detect do |emoji|
-      emoji.short_name == name || emoji.short_names.include?(name)
-    end
+    name = name.delete_prefix(':').delete_suffix(':') # Allow to find `+1` by `:+1:`
+    name.delete_suffix!('::skin-tone-1') # special case for default skin tone
+    short_name_lookup_map[name]
+  end
+
+  # Finds emoji by unified hex representation of Unicode codepoints
+  # @param unified [String] Dash separated hexadecimal Unicode character codes for a single emoji
+  # @return [EmojiDatasource::Emoji] if given Unicode codepoints is a known emoji
+  # @return [nil] if there are no emojis with given codepoints in the dataset
+  # @example
+  #   short_name_to_char('1F44D') #=> EmojiDatasource::Emoji: :+1: (👍)
+  def self.find_by_unified(unified)
+    unified_lookup_map[unified.upcase]
+  end
+
+  # Find emoji object by raw Unicode string with a single emoji in it
+  # @param raw_emoji [String] Single emoji
+  # @return [EmojiDatasource::Emoji] if given string contains a single known emoji
+  # @return [nil] if there are no such emoji in the dataset
+  # @example
+  #   find_by_char('👍') #=> EmojiDatasource::Emoji: :+1: (👍)
+  def self.find_by_char(raw_emoji)
+    find_by_unified(char_to_unified(raw_emoji))
   end
 
   def self.unified_to_char(unified_name)
@@ -29,10 +56,40 @@ module EmojiDatasource
     unified_name.split('-').map(&:hex).pack('U*')
   end
 
+  def self.char_to_unified(raw_emoji)
+    return unless raw_emoji
+
+    raw_emoji.unpack('U*').map { |c| c.to_s(16).upcase }.join('-')
+  end
+
   def self.data
-    @data ||= JSON.parse(File.read(EMOJI_DATA_PATH), symbolize_names: true)
+    @data ||= JSON.parse(File.read(EMOJI_DATA_PATH))
       .map { |emoji_data| EmojiDatasource::Emoji.new(emoji_data) }
   end
+
+  # Utility hash map to search by emoji short code, including variants
+  # @api private
+  def self.short_name_lookup_map
+    @short_name_lookup_map ||= data.each_with_object({}) do |emoji, result|
+      emoji.short_names.each { |short_name| result[short_name] = emoji }
+      emoji.variations.each do |emoji_variant|
+        emoji_variant.short_names.each do |short_name|
+          result[short_name] = emoji_variant
+        end
+      end
+    end
+  end
+
+  # Utility hash map to search by unicode character sequence hex codes, including variants
+  # @api private
+  def self.unified_lookup_map
+    @unified_lookup_map ||= data.each_with_object({}) do |emoji, result|
+      result[emoji.unified] = emoji
+      emoji.variations.each do |emoji_variant|
+        result[emoji_variant.unified] = emoji_variant
+      end
+    end
+  end
 end
 
 # Preload emojies on startup

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: emoji-datasource
 version: !ruby/object:Gem::Version
-  version: 14.0.0
+  version: 14.0.2
 platform: ruby
 authors:
 - Justas Palumickas
 autorequire:
 bindir: exe
 cert_chain: []
-date: 2023-02-14 00:00:00.000000000 Z
+date: 2023-03-16 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: pry
@@ -104,7 +104,6 @@ files:
 - lib/emoji-datasource.rb
 - lib/emoji_datasource.rb
 - lib/emoji_datasource/emoji.rb
-- lib/emoji_datasource/short_name_to_char.rb
 - lib/emoji_datasource/version.rb
 - package.json
 - vendor/emoji-datasource/emoji.json
@@ -133,7 +132,7 @@ required_rubygems_version: !ruby/object:Gem::Requirement
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.2.32
+rubygems_version: 3.4.8
 signing_key:
 specification_version: 4
 summary: Emoji data from emoji-datasource npm package

data/lib/emoji_datasource/short_name_to_char.rb

@@ -1,49 +0,0 @@
-# frozen_string_literal: true
-
-module EmojiDatasource
-  class ShortNameToChar
-    def self.generate(name)
-      new(name).call
-    end
-
-    attr_reader :short_name
-
-    def initialize(short_name)
-      @short_name = short_name
-    end
-
-    def call
-      return unless short_name
-      return EmojiDatasource.find_by_short_name(short_name)&.to_char unless skin_tone_matches
-
-      emoji = EmojiDatasource.find_by_short_name(skin_tone_matches[1])
-      return unless emoji
-      return emoji.to_char if skin_tone_level == 1
-
-      char_with_skin_tone(emoji)
-    end
-
-    private
-
-    def char_with_skin_tone(emoji)
-      return unless skin_tone_emoji
-
-      skin_variation = emoji.skin_variations && emoji.skin_variations[skin_tone_emoji.unified]
-      return unless skin_variation
-
-      EmojiDatasource.unified_to_char(skin_variation[:unified])
-    end
-
-    def skin_tone_emoji
-      EmojiDatasource.find_by_short_name("skin-tone-#{skin_tone_level}")
-    end
-
-    def skin_tone_level
-      skin_tone_matches[2].to_i
-    end
-
-    def skin_tone_matches
-      short_name.match(/:?(.+)::skin-tone-(\d+):?/)
-    end
-  end
-end