=begin
    mo.rb - A simple class for operating GNU MO file.

    Copyright (C) 2003-2006  Masao Mutoh
    Copyright (C) 2002  Masahiro Sakai, Masao Mutoh
    Copyright (C) 2001  Masahiro Sakai

        Masahiro Sakai                  <s01397ms@sfc.keio.ac.jp>
        Masao Mutoh                     <mutoh@highway.ne.jp>

    You can redistribute this file and/or modify it under the same term
    of Ruby.  License of Ruby is included with Ruby distribution in
    the file "README".

    $Id: mo.rb,v 1.4 2006/04/29 17:17:15 mutoh Exp $
=end

require 'gettext/iconv'

class MOFile < Hash #:nodoc:
  class InvalidFormat < RuntimeError; end;

  Header = Struct.new(:magic,
                      :revision,
                      :nstrings,
                      :orig_table_offset,
                      :translated_table_offset,
                      :hash_table_size,
                      :hash_table_offset)

  MAGIC_BIG_ENDIAN    = "\x95\x04\x12\xde"
  MAGIC_LITTLE_ENDIAN = "\xde\x12\x04\x95"

  def self.open(arg = nil, output_charset = nil)
    result = self.new(output_charset)
    result.load(arg)
  end

  def initialize(output_charset = nil)
    @filename = nil
    @last_modified = Time.now
    @little_endian = true
    @output_charset = output_charset
    super()
  end

  def update!
    if FileTest.exist?(@filename)
      load (@filename) unless (@last_modified == File.ctime(@filename))
    else
      puts "#{@filename} was lost." if $DEBUG
      clear
    end
    self
  end

  def load(arg)
    case arg
    when String
      begin
	@last_modified = File.ctime(arg)
      rescue Exception
      end
      load_from_file(arg)
    when IO
      load_from_stream(arg)
    end
    @filename = arg
    self
  end

  def load_from_stream(io)
    magic = io.read(4)
    case magic
    when MAGIC_BIG_ENDIAN
      @little_endian = false
    when MAGIC_LITTLE_ENDIAN
      @little_endian = true
    else
      raise InvalidFormat.new(sprintf("Unknown signature %s", magic.dump))
    end

    header = Header.new(magic, *(io.read(4 * 6).unpack(@little_endian ? 'V6' : 'N6')))
    raise InvalidFormat.new(sprintf("file format revision %d isn't supported", header.revision)) if header.revision > 0

    io.pos = header.orig_table_offset
    orig_table_data = io.read((4 * 2) * header.nstrings).unpack(@little_endian ? 'V*' : 'N*')

    io.pos = header.translated_table_offset
    trans_table_data = io.read((4 * 2) * header.nstrings).unpack(@little_endian ? 'V*' : 'N*')

    original_strings = Array.new(header.nstrings)
    for i in 0...header.nstrings
      io.pos = orig_table_data[i * 2 + 1]
      original_strings[i] = io.read(orig_table_data[i * 2 + 0])
    end

    clear
    for i in 0...header.nstrings
      io.pos = trans_table_data[i * 2 + 1]
      str = io.read(trans_table_data[i * 2 + 0])

      if original_strings[i] == ""
        if str
          @charset = nil
          @nplurals = nil
          @plural = nil
          str.each_line{|line|
            if /^Content-Type:/i =~ line and /charset=((?:\w|-)+)/i =~ line
              @charset = $1
            elsif /^Plural-Forms:\s*nplurals\s*\=\s*(\d*);\s*plural\s*\=\s*([^;]*)\n?/ =~ line
              @nplurals = $1
              @plural = $2
            end
            break if @charset and @nplurals
          }
          @nplurals = "1" unless @nplurals
          @plural = "0" unless @plural
        end
      else
        if @output_charset
          begin
            str = Iconv.iconv(@output_charset, @charset, str).join if @charset
          rescue Iconv::Failure
            if $DEBUG
              $stderr.print "@charset = ", @charset, "\n"
              $stderr.print "@output_charset = ", @output_charset, "\n"
              $stderr.print "msgid = ", original_strings[i], "\n"
              $stderr.print "msgstr = ", str, "\n"
            end
          end
        end
      end
      self[original_strings[i]] = str
    end
    self
  end

  # From gettext-0.12.1/gettext-tools/lib/hash.c
  def prime?(candidate)
    divn = 3
    sq = divn * divn

    while (sq < candidate && candidate % divn != 0)
      divn += 1
      sq += 4 * divn
      divn += 1
    end
    candidate % divn != 0
  end

  # From gettext-0.12.1/gettext-tools/lib/hash.c
  def next_prime(seed)
    seed |= 1
    while (! prime?(seed))
      seed += 2
    end
    seed
  end

  # From gettext-0.12.1/gettext-runtime/intl/hash-string.h
  # Defines the so called `hashpjw' function by P.J. Weinberger
  # [see Aho/Sethi/Ullman, COMPILERS: Principles, Techniques and Tools,
  # 1986, 1987 Bell Telephone Laboratories, Inc.] 
  HASHWORDBITS = 32
  def hash_string(str)
    hval = 0
    i = 0
    str.each_byte do |b|
      break if b == '\0'
      hval <<= 4
      hval += b.to_i
      g = hval & (0xf << (HASHWORDBITS - 4))
      if (g != 0)
        hval ^= g >> (HASHWORDBITS - 8)
        hval ^= g
      end
    end
    hval
  end

  def save_to_stream(io)
    #Save data as little endian format.
    header_size = 4 * 7
    table_size  = 4 * 2 * size

    hash_table_size = next_prime((size * 4) / 3) 
    hash_table_size = 3 if hash_table_size <= 2
    header = Header.new(
      MAGIC_LITTLE_ENDIAN,          # magic
      0,                            # revision
      size,                         # nstrings
      header_size,                  # orig_table_offset
      header_size + table_size,     # translated_table_offset
      hash_table_size,              # hash_table_size
      header_size + table_size * 2  # hash_table_offset
    )
    io.write(header.to_a.pack('a4V*'))

    ary = to_a
    ary.sort!{|a, b| a[0] <=> b[0]} # sort by original string

    pos = header.hash_table_size * 4 + header.hash_table_offset

    orig_table_data = Array.new()
    ary.each{|item, _|
      orig_table_data.push(item.size)
      orig_table_data.push(pos)
      pos += item.size + 1 # +1 is <NUL>
    }
    io.write(orig_table_data.pack('V*'))

    trans_table_data = Array.new()
    ary.each{|_, item|
      trans_table_data.push(item.size)
      trans_table_data.push(pos)
      pos += item.size + 1 # +1 is <NUL>
    }
    io.write(trans_table_data.pack('V*'))

    hash_tab = Array.new(hash_table_size)
    j = 0
    ary[0...size].each {|key, _|
      hash_val = hash_string(key)
      idx = hash_val % hash_table_size
      if hash_tab[idx] != nil
        incr = 1 + (hash_val % (hash_table_size - 2))
        begin
          if (idx >= hash_table_size - incr)
            idx -= hash_table_size - incr
          else
            idx += incr
          end
        end until (hash_tab[idx] == nil)
      end
      hash_tab[idx] = j + 1
      j += 1
    }
    hash_tab.collect!{|i| i ? i : 0}

    io.write(hash_tab.pack('V*'))

    ary.each{|item, _| io.write(item); io.write("\0") }
    ary.each{|_, item| io.write(item); io.write("\0") }

    self
  end

  def load_from_file(filename)
    File.open(filename, 'rb'){|f| load_from_stream(f)}
  end

  def save_to_file(filename)
    File.open(filename, 'wb'){|f| save_to_stream(f)}
  end

  def set_comment(msgid_or_sym, comment)
    #Do nothing
  end


  attr_accessor :little_endian, :path, :last_modified
  attr_reader :charset, :nplurals, :plural
end


# Test

if $0 == __FILE__
  if (ARGV.include? "-h") or (ARGV.include? "--help")
    STDERR.puts("mo.rb [filename.mo ...]")
    exit
  end

  ARGV.each{ |item|
    mo = MOFile.open(item)
    puts "------------------------------------------------------------------"
    puts "charset  = \"#{mo.charset}\""
    puts "nplurals = \"#{mo.nplurals}\""
    puts "plural   = \"#{mo.plural}\""
    puts "------------------------------------------------------------------"
    mo.each do |key, value|
      puts "original message = \"#{key}\""
      puts "translated message = \"#{value}\""
      puts "--------------------------------------------------------------------"
    end
  }
end