bin/rake data:dump to parse emoji-test.txt and output JSON

Rakefile

@@ -8,4 +8,10 @@ Rake::TestTask.new do |t|
   t.verbose = true
 end
 
+namespace :data do
+  task :dump do
+    system "bin/parse-emoji-test", "vendor/Unicode-12.0/emoji-test.txt"
+  end
+end
+
 task default: :test

bin/parse-emoji-test

@@ -0,0 +1,13 @@
+#!/usr/bin/env ruby
+# frozen_string_literal: true
+
+# Parse https://unicode.org/Public/emoji/12.0/emoji-test.txt and output JSON.
+
+# Exit cleanly from an early interrupt
+Signal.trap("INT") { abort }
+
+require "json"
+require_relative "../data/emoji_test_parser"
+
+parsed = EmojiTestParser.parse(ARGV[0])
+puts JSON.pretty_generate(parsed)

data/emoji.rb

@@ -0,0 +1,28 @@
+# frozen_string_literal: true
+
+class Emoji
+  def initialize(data)
+    @codepoints = data[:codepoints].strip
+    @status = data[:status].strip
+    @unicode = data[:unicode]
+    @description = data[:description]
+  end
+
+  def to_h(group:, subgroup:)
+    {
+      unicode: unicode,
+      codepoints: codepoints,
+      description: description,
+      group: group,
+      subgroup: subgroup,
+      status: status,
+    }
+  end
+
+  def inspect
+    "#<Emoji #{@unicode}>"
+  end
+
+  private
+  attr_reader :codepoints, :status, :unicode, :description
+end

data/emoji_category.rb

@@ -0,0 +1,19 @@
+# frozen_string_literal: true
+
+class EmojiCategory
+  attr_reader :name, :type, :emojis
+
+  def initialize(name, type:, emojis: [])
+    @name = name
+    @type = type
+    @emojis = emojis
+  end
+
+  def add(emoji)
+    @emojis << emoji
+  end
+
+  def inspect
+    "#<EmojiCategory type: #{@type}, name: #{@name}, emojis: #{@emojis.size} #<Emoji> objects>"
+  end
+end

data/emoji_test_parser.rb

@@ -0,0 +1,91 @@
+# frozen_string_literal: true
+
+require_relative "emoji_category"
+require_relative "emoji"
+
+# Parse https://unicode.org/Public/emoji/12.0/emoji-test.txt into Ruby.
+class EmojiTestParser
+  def self.parse(file)
+    new(file).parse
+  end
+
+  def initialize(file)
+    @file = file
+  end
+
+  def parse
+    emoji_categories = []
+    group = nil
+    subgroup = nil
+
+    IO.readlines(file).each do |line|
+      case
+      when line.start_with?("# group: ")
+        group = parse_group(line)
+        emoji_categories << group
+      when line.start_with?("# subgroup: ")
+        subgroup = parse_subgroup(line)
+        group.add(subgroup)
+      when comment?(line) || newline?(line)
+        next
+      else
+        emoji = parse_emoji(line)
+        subgroup.add(emoji)
+      end
+    end
+
+    flatten(emoji_categories)
+  end
+
+  private
+  attr_reader :file
+
+  def parse_name(line)
+    _, name = line.split(":", 2)
+    name.strip
+  end
+
+  def parse_group(line)
+    group_name = parse_name(line)
+    EmojiCategory.new(group_name, type: "group")
+  end
+
+  def parse_subgroup(line)
+    group_name = parse_name(line)
+    EmojiCategory.new(group_name, type: "subgroup")
+  end
+
+  def comment?(line)
+    line.start_with?("#")
+  end
+
+  def newline?(line)
+    line.strip.empty?
+  end
+
+  EMOJI_LINE_REGEXP = /(?<codepoints>.+);(?<status>.+)# (?<unicode>[^[\s]]+)\s(?<description>.+)/
+  private_constant :EMOJI_LINE_REGEXP
+
+  def parse_emoji(line)
+    matched = line.match(EMOJI_LINE_REGEXP)
+    Emoji.new(matched)
+  end
+
+  def flatten(data)
+    emojis = []
+
+    data.each do |emoji_category|
+      emoji_category.emojis.each do |emoji_subcategory|
+        emoji_subcategory.emojis.each do |raw_emoji|
+          emoji = raw_emoji.to_h(
+            group: emoji_category.name,
+            subgroup: emoji_subcategory.name
+          )
+          emojis << emoji
+        end
+      end
+    end
+
+    emojis
+  end
+end