module Haml::Util

def self.escape_html(html)
  CGI.escapeHTML(html.to_s)
end

def self.escape_html_safe(html)
  html = html.to_s
  (html.respond_to?(:html_safe?) && html.html_safe?) ? html : escape_html(html)
end

def balance(scanner, start, finish, count = 0)
  str = ''.dup
  scanner = StringScanner.new(scanner) unless scanner.is_a? StringScanner
  regexp = Regexp.new("(.*?)[\\#{start.chr}\\#{finish.chr}]", Regexp::MULTILINE)
  while scanner.scan(regexp)
    str << scanner.matched
    count += 1 if scanner.matched[-1] == start
    count -= 1 if scanner.matched[-1] == finish
    return [str.strip, scanner.rest] if count == 0
  end
end

def check_encoding(str)
  if str.valid_encoding?
    # Get rid of the Unicode BOM if possible
    # Shortcut for UTF-8 which might be the majority case
    if str.encoding == Encoding::UTF_8
      return str.gsub(/\A\uFEFF/, '')
    elsif str.encoding.name =~ /^UTF-(16|32)(BE|LE)?$/
      return str.gsub(Regexp.new("\\A\uFEFF".encode(str.encoding)), '')
    else
      return str
    end
  end
  encoding = str.encoding
  newlines = Regexp.new("\r\n|\r|\n".encode(encoding).force_encoding(Encoding::ASCII_8BIT))
  str.force_encoding(Encoding::ASCII_8BIT).split(newlines).each_with_index do |line, i|
    begin
      line.encode(encoding)
    rescue Encoding::UndefinedConversionError => e
      yield <<MSG.rstrip, i + 1
lid #{encoding.name} character #{e.error_char.dump}
    end
  end
  return str
end

def check_haml_encoding(str, &block)
  str = str.dup if str.frozen?
  bom, encoding = parse_haml_magic_comment(str)
  if encoding; str.force_encoding(encoding)
  elsif bom; str.force_encoding(Encoding::UTF_8)
  end
  return check_encoding(str, &block)
end

def contains_interpolation?(str)
  /#[\{$@]/ === str
end

def handle_interpolation(str)
  scan = StringScanner.new(str)
  yield scan while scan.scan(/(.*?)(\\*)#([\{@$])/)
  scan.rest
end

def human_indentation(indentation)
  if !indentation.include?(?\t)
    noun = 'space'
  elsif !indentation.include?(?\s)
    noun = 'tab'
  else
    return indentation.inspect
  end
  singular = indentation.length == 1
  "#{indentation.length} #{noun}#{'s' unless singular}"
end

def inspect_obj(obj)
  case obj
  when String
    %Q!"#{obj.gsub(/[\x00-\x7F]+/) {|s| s.dump[1...-1]}}"!
  when Symbol
    ":#{inspect_obj(obj.to_s)}"
  else
    obj.inspect
  end
end

def parse_haml_magic_comment(str)
  scanner = StringScanner.new(str.dup.force_encoding(Encoding::ASCII_8BIT))
  bom = scanner.scan(/\xEF\xBB\xBF/n)
  return bom unless scanner.scan(/-\s*#\s*/n)
  if (coding = try_parse_haml_emacs_magic_comment(scanner))
    return bom, coding
  end
  return bom unless scanner.scan(/.*?coding[=:]\s*([\w-]+)/in)
  return bom, scanner[1]
end

def rails_xss_safe?
  false
end

def silence_warnings
  the_real_stderr, $stderr = $stderr, StringIO.new
  yield
ensure
  $stderr = the_real_stderr
end

def try_parse_haml_emacs_magic_comment(scanner)
  pos = scanner.pos
  return unless scanner.scan(/.*?-\*-\s*/n)
  # From Ruby's parse.y
  return unless scanner.scan(/([^\s'":;]+)\s*:\s*("(?:\\.|[^"])*"|[^"\s;]+?)[\s;]*-\*-/n)
  name, val = scanner[1], scanner[2]
  return unless name =~ /(en)?coding/in
  val = $1 if val =~ /^"(.*)"$/n
  return val
ensure
  scanner.pos = pos
end

def unescape_interpolation(str, escape_html = nil)
  res = ''.dup
  rest = Haml::Util.handle_interpolation str.dump do |scan|
    escapes = (scan[2].size - 1) / 2
    char = scan[3] # '{', '@' or '$'
    res << scan.matched[0...-3 - escapes]
    if escapes % 2 == 1
      res << "\##{char}"
    else
      interpolated = if char == '{'
        balance(scan, ?{, ?}, 1)[0][0...-1]
      else
        scan.scan(/\w+/)
      end
      content = eval("\"#{interpolated}\"")
      content = "#{char}#{content}" if char == '@' || char == '$'
      content = "Haml::Util.escape_html_safe((#{content}).to_s)" if escape_html
      res << "\#{#{content}}"
    end
  end
  res + rest
end