uni.rb

require 'fileutils'

UNICODE_VERSION = "15.0.0"
BASE_DIR = "#{__dir__}/#{UNICODE_VERSION}"
UCD_DIR = "#{BASE_DIR}/ucd"
BASE_URL = "https://www.unicode.org/Public/#{UNICODE_VERSION}/ucd"

def download_file(file)
  dir = "#{BASE_DIR}/#{File.dirname(file)}"
  FileUtils.mkdir_p(dir)
  Dir.chdir(dir) do
    `curl -LO "#{BASE_URL}/#{file}"` unless File.exist?("#{BASE_DIR}/#{file}")
  end
end

def load_file(file)
  download_file(file)
  File.read("#{BASE_DIR}/#{file}").lines(chomp: true)
end

def load_binary_property()
  data = []
  in_binary_property = false
  load_file("PropertyAliases.txt").each do |line|
    in_binary_property ||= !!(line =~ /# Binary Properties/)
    next unless in_binary_property
    m = line.match(/^([^;]+);([^;]+)(?:;([^;]+))?$/)
    next unless m
    short, long, other = m[1].strip, m[2].strip, m[3]&.strip
    data << [long, [short, other].compact]
  end
  data
end

def load_general_category_value()
  data = []
  load_file("PropertyValueAliases.txt").each do |line|
    m = line.match(/^gc ; ([^;#]+);([^;#]+)(?:;([^;#]+))?/)
    next unless m
    short, long, other = m[1].strip, m[2].strip, m[3]&.strip
    data << [long, [short, other].compact]
  end
  data
end

def load_script_value()
  data = []
  load_file("PropertyValueAliases.txt").each do |line|
    m = line.match(/^sc ; ([^;#]+);([^;#]+)(?:;([^;#]+))?/)
    next unless m
    short, long, other = m[1].strip, m[2].strip, m[3]&.strip
    data << [long, [short, other].compact]
  end
  data
end

def load_age_value()
  data = []
  load_file("PropertyValueAliases.txt").each do |line|
    m = line.match(/^age; (\d+\.\d+)\s*;([^;#]+)(?:;([^;#]+))?/)
    next unless m
    short, long, other = m[1].strip, m[2].strip, m[3]&.strip
    data << [long, [short, other].compact]
  end
  data
end

def load_block_value()
  data = []
  load_file("PropertyValueAliases.txt").each do |line|
    m = line.match(/^blk; ([^;#]+);([^;#]+)(?:;([^;#]+))?/)
    next unless m
    short, long, other = m[1].strip, m[2].strip, m[3]&.strip
    data << [long, [short, other].compact]
  end
  data
end

binary_properties = load_binary_property()
general_category_values = load_general_category_value()
script_values = load_script_value()
age_values = load_age_value()
block_values = load_block_value()

puts "val BinaryPropertyNames: Map[String, String] = Map("
binary_properties.each do |(name, aliases)|
  [name, *aliases].uniq.each do |a|
    puts "  #{a.inspect} -> #{name.inspect},"
  end
end
puts ")"
puts

puts "val GeneralCategoryValues: Map[String, String] = Map("
general_category_values.each do |(name, aliases)|
  [name, *aliases].uniq.each do |a|
    puts "  #{a.inspect} -> #{name.inspect},"
  end
end
puts ")"
puts

puts "val ScriptValues: Map[String, String] = Map("
script_values.each do |(name, aliases)|
  [name, *aliases].uniq.each do |a|
    puts "  #{a.inspect} -> #{name.inspect},"
  end
end
puts ")"
puts

puts "val AgeValues: Set[String] = Set("
age_values.each do |(_, (name))|
  puts "  #{name.inspect},"
end
puts ")"
puts

puts "val BlockValues: Set[String] = Set("
block_values.each do |(name, _)|
  puts "  #{name.inspect},"
end
puts ")"
puts