#!/usr/bin/ruby

# Usage:
#   $ wget http://www.unicode.org/Public/UNIDATA/CaseFolding.txt
#   $ ruby CaseFolding.rb CaseFolding.txt > ../enc/unicode/casefold.h

class CaseFolding
  module Util
    module_function

    def hex_seq(v)
      v.map {|i| "0x%04x" % i}.join(", ")
    end

    def print_table_1(dest, data)
      for k, v in data.sort
        sk = (Array === k and k.length > 1) ? "{#{hex_seq(k)}}" : ("0x%04x" % k)
        dest.print("  {#{sk}, {#{v.length}, {#{hex_seq(v)}}}},\n")
      end
    end

    def print_table(dest, type, data)
      dest.print("static const #{type}_Type #{type}_Table[] = {\n")
      i = 0
      data.each do |n, d|
        dest.print("#define #{n} (*(#{type}_Type (*)[#{d.size}])(#{type}_Table+#{i}))\n")
        i += d.size
        print_table_1(dest, d)
      end
      dest.print("};\n\n")
    end
  end

  include Util

  attr_reader :fold, :fold_locale, :unfold, :unfold_locale

  def load(filename)
    pattern = /([0-9A-F]{4,6}); ([CFT]); ([0-9A-F]{4,6})(?: ([0-9A-F]{4,6}))?(?: ([0-9A-F]{4,6}))?;/

    @fold = fold = {}
    @unfold = unfold = [{}, {}, {}]
    turkic = []

    IO.foreach(filename) do |line|
      next unless res = pattern.match(line)
      ch_from = res[1].to_i(16)

      if res[2] == 'T'
        # Turkic case folding
        turkic << ch_from
        next
      end

      # store folding data
      ch_to = res[3..6].inject([]) do |a, i|
        break a unless i
        a << i.to_i(16)
      end
      fold[ch_from] = ch_to

      # store unfolding data
      i = ch_to.length - 1
      (unfold[i][ch_to] ||= []) << ch_from
    end

    # move locale dependent data to (un)fold_locale
    @fold_locale = fold_locale = {}
    @unfold_locale = unfold_locale = [{}, {}]
    for ch_from in turkic
      key = fold[ch_from]
      i = key.length - 1
      unfold_locale[i][i == 0 ? key[0] : key] = unfold[i].delete(key)
      fold_locale[ch_from] = fold.delete(ch_from)
    end
    self
  end

  def display(dest)
    # print the header
    dest.print("/* DO NOT EDIT THIS FILE. */\n")
    dest.print("/* Generated by enc/unicode/case-folding.rb */\n\n")

    # print folding data

    # CaseFold + CaseFold_Locale
    name = "CaseFold_11"
    print_table(dest, name, "CaseFold"=>fold, "CaseFold_Locale"=>fold_locale)

    # print unfolding data

    # CaseUnfold_11 + CaseUnfold_11_Locale
    name = "CaseUnfold_11"
    print_table(dest, name, name=>unfold[0], "#{name}_Locale"=>unfold_locale[0])

    # CaseUnfold_12 + CaseUnfold_12_Locale
    name = "CaseUnfold_12"
    print_table(dest, name, name=>unfold[1], "#{name}_Locale"=>unfold_locale[1])

    # CaseUnfold_13
    name = "CaseUnfold_13"
    print_table(dest, name, name=>unfold[2])

    # table sizes
    fold_table_size = fold.size + fold_locale.size
    dest.printf("#define FOLD_TABLE_SIZE\t\t%d\n", (fold_table_size * 1.2))
    unfold1_table_size = unfold[0].size + unfold_locale[0].size
    dest.printf("#define UNFOLD1_TABLE_SIZE\t%d\n", (unfold1_table_size * 1.2))
    unfold2_table_size = unfold[1].size + unfold_locale[1].size
    dest.printf("#define UNFOLD2_TABLE_SIZE\t%d\n", (unfold2_table_size * 1.5))
    unfold3_table_size = unfold[2].size
    dest.printf("#define UNFOLD3_TABLE_SIZE\t%d\n", (unfold3_table_size * 1.7))
  end

  def self.load(*args)
    new.load(*args)
  end
end

if $0 == __FILE__
  require 'optparse'
  dest = nil
  fold_1 = false
  ARGV.options do |opt|
    opt.banner << " [INPUT]"
    opt.on("--output-file=FILE", "-o", "output to the FILE instead of STDOUT") {|output|
      dest = (output unless output == '-')
    }
    opt.parse!
    abort(opt.to_s) if ARGV.size > 1
  end
  filename = ARGV[0] || 'CaseFolding.txt'
  data = CaseFolding.load(filename)
  if dest
    open(dest, "wb") do |f|
      data.display(f)
    end
  else
    data.display(STDOUT)
  end
end