class RDoc::Parser

def self.alias_extension(old_ext, new_ext)
  old_ext = old_ext.sub(/^\.(.*)/, '\1')
  new_ext = new_ext.sub(/^\.(.*)/, '\1')
  parser = can_parse_by_name "xxx.#{old_ext}"
  return false unless parser
  RDoc::Parser.parsers.unshift [/\.#{new_ext}$/, parser]
  true
end

def self.binary?(file)
  return false if file =~ /\.(rdoc|txt)$/
  s = File.read(file, 1024) or return false
  return true if s[0, 2] == Marshal.dump('')[0, 2] or s.index("\x00")
  mode = 'r:utf-8' # default source encoding has been changed to utf-8
  s.sub!(/\A#!.*\n/, '')     # assume shebang line isn't longer than 1024.
  encoding = s[/^\s*\#\s*(?:-\*-\s*)?(?:en)?coding:\s*([^\s;]+?)(?:-\*-|[\s;])/, 1]
  mode = "rb:#{encoding}" if encoding
  s = File.open(file, mode) {|f| f.gets(nil, 1024)}
  not s.valid_encoding?
end

def self.can_parse file_name
  parser = can_parse_by_name file_name
  # HACK Selenium hides a jar file using a .txt extension
  return if parser == RDoc::Parser::Simple and zip? file_name
  parser
end

def self.can_parse_by_name file_name
  _, parser = RDoc::Parser.parsers.find { |regexp,| regexp =~ file_name }
  # The default parser must not parse binary files
  ext_name = File.extname file_name
  return parser if ext_name.empty?
  if parser == RDoc::Parser::Simple and ext_name !~ /txt|rdoc/ then
    case mode = check_modeline(file_name)
    when nil, 'rdoc' then # continue
    else
      RDoc::Parser.parsers.find { |_, p| return p if mode.casecmp?(p.name[/\w+\z/]) }
      return nil
    end
  end
  parser
rescue Errno::EACCES
end

def self.check_modeline file_name
  line = File.open file_name do |io|
    io.gets
  end
  /-\*-\s*(.*?\S)\s*-\*-/ =~ line
  return nil unless type = $1
  if /;/ =~ type then
    return nil unless /(?:\s|\A)mode:\s*([^\s;]+)/i =~ type
    type = $1
  end
  return nil if /coding:/i =~ type
  type.downcase
rescue ArgumentError
rescue Encoding::InvalidByteSequenceError # invalid byte sequence
end

def self.for top_level, content, options, stats
  file_name = top_level.absolute_name
  return if binary? file_name
  parser = use_markup content
  unless parser then
    parse_name = file_name
    # If no extension, look for shebang
    if file_name !~ /\.\w+$/ && content =~ %r{\A#!(.+)} then
      shebang = $1
      case shebang
      when %r{env\s+ruby}, %r{/ruby}
        parse_name = 'dummy.rb'
      end
    end
    parser = can_parse parse_name
  end
  return unless parser
  content = remove_modeline content
  parser.new top_level, file_name, content, options, stats
rescue SystemCallError
  nil
end

def self.parse_files_matching(regexp)
  RDoc::Parser.parsers.unshift [regexp, self]
end

def self.remove_modeline content
  content.sub(/\A.*-\*-\s*(.*?\S)\s*-\*-.*\r?\n/, '')
end

def self.use_markup content
  markup = content.lines.first(3).grep(/markup:\s+(\w+)/) { $1 }.first
  return unless markup
  # TODO Ruby should be returned only when the filename is correct
  return RDoc::Parser::Ruby if %w[tomdoc markdown].include? markup
  markup = Regexp.escape markup
  _, selected = RDoc::Parser.parsers.find do |_, parser|
    /^#{markup}$/i =~ parser.name.sub(/.*:/, '')
  end
  selected
end

def self.zip? file
  zip_signature = File.read file, 4
  zip_signature == "PK\x03\x04" or
    zip_signature == "PK\x05\x06" or
    zip_signature == "PK\x07\x08"
rescue
  false
end

def handle_tab_width(body)
  if /\t/ =~ body
    tab_width = @options.tab_width
    body.split(/\n/).map do |line|
      1 while line.gsub!(/\t+/) do
        b, e = $~.offset(0)
        ' ' * (tab_width * (e-b) - b % tab_width)
      end
      line
    end.join "\n"
  else
    body
  end
end

def initialize top_level, file_name, content, options, stats
  @top_level = top_level
  @top_level.parser = self.class
  @store = @top_level.store
  @file_name = file_name
  @content = content
  @options = options
  @stats = stats
  @preprocess = RDoc::Markup::PreProcess.new @file_name, @options.rdoc_include
  @preprocess.options = @options
end