I18n_export.rb

# frozen_string_literal: true

module I18nTaskYamlExt
  UNMASKED_EMOJI = /
    (?:
      (?:\p{Emoji_Presentation}|\p{Emoji}\uFE0F)   # base emoji
      (?:\u200D(?:\p{Emoji_Presentation}|\p{Emoji}\uFE0F))* # + ZWJ parts
    )
  /ux

  def dump(tree, options)
    builder = Psych::Visitors::YAMLTree.create
    builder << tree
    ast = builder.tree
    _process_node(ast)
    strip_trailing_spaces(restore_emojis(ast.to_yaml(nil, options || {})))
  end

  private

  def _process_node(node)
    case node
    when Psych::Nodes::Scalar
      node.plain = false
      node.quoted = true
      node.style = node.value.include?("\n") ? Psych::Nodes::Scalar::LITERAL : Psych::Nodes::Scalar::DOUBLE_QUOTED # <== THE ENTIRE PURPOSE OF THIS MONKEYPATCH 🫠
      node.value = _mask_emoji(node.value) if node.style == Psych::Nodes::Scalar::LITERAL # pre-mask emoji because otherwise libyaml will double-quote when we want literal style
    when Psych::Nodes::Mapping
      # only process the values, not the keys
      node.children.each_slice(2) { |_key, value| _process_node(value) }
    when Psych::Nodes::Stream, Psych::Nodes::Document, Psych::Nodes::Sequence
      node.children.each { |node| _process_node(node) }
    else
      raise "not handling #{node.inspect}"
    end
  end

  # libyaml will do this, but we want to do it first so that libyaml doesn't _also_
  # mark the node as unprintable and thus prevent it from being in a literal
  # https://github.com/yaml/libyaml/issues/279
  # "Hello 👋 world 🌍!" => "Hello \\u0001F44B world \\u0001F30D!"
  def _mask_emoji(string)
    string.gsub(UNMASKED_EMOJI) do |emoji|
      emoji.codepoints
        .map { |cp| format('\\u%08X', cp) }
        .join('_')
    end
  end
end

I18n::Tasks::Data::Adapter::YamlAdapter.singleton_class.prepend I18nTaskYamlExt