raabro

A very dumb PEG parser library.

Son to aabro, grandson to neg, grand-grandson to parslet. There is also a javascript version jaabro.

a sample parser/rewriter

You use raabro by providing the parsing rules, then some rewrite rules.

The parsing rules make use of the raabro basic parsers seq, alt, str, rex, eseq, …

The rewrite rules match names passed as first argument to the basic parsers to rewrite the resulting parse trees.

require 'raabro'


module Fun include Raabro

  # parse
  #
  # Last function is the root, "i" stands for "input".

  def pa(i); rex(nil, i, /\(\s*/); end
  def pz(i); rex(nil, i, /\)\s*/); end
  def com(i); rex(nil, i, /,\s*/); end

  def num(i); rex(:num, i, /-?[0-9]+\s*/); end

  def args(i); eseq(nil, i, :pa, :exp, :com, :pz); end
  def funame(i); rex(nil, i, /[a-z][a-z0-9]*/); end
  def fun(i); seq(:fun, i, :funame, :args); end

  def exp(i); alt(nil, i, :fun, :num); end

  # rewrite
  #
  # Names above (:num, :fun, ...) get a rewrite_xxx function.
  # "t" stands for "tree".
  #
  # The trees with a nil name are handled by rewrite_(tree) a default
  # rewrite function

  def rewrite_num(t); t.string.to_i; end

  def rewrite_fun(t)
    [ t.children[0].string ] +
    t.children[1].odd_children.collect { |a| rewrite(a) }
  end
end


p Fun.parse('mul(1, 2)')
  # => ["mul", 1, 2]

p Fun.parse('mul(1, add(-2, 3))')
  # => ["mul", 1, ["add", -2, 3]]

p Fun.parse('mul (1, 2)')
  # => nil (doesn't accept a space after the function name)

This sample is available at: doc/readme0.rb.

custom rewrite()

By default, a parser gets a rewrite(t) that looks at the parse tree node names and calls the corresponding rewrite_{node_name}().

It’s OK to provide a custom rewrite(t) function.

module Hello include Raabro

  def hello(i); str(:hello, i, 'hello'); end

  def rewrite(t)
    [ :ok, t.string ]
  end
end

basic parsers

One makes a parser by composing basic parsers, for example:

  def args(i); eseq(:args, i, :pa, :exp, :com, :pz); end
  def funame(i); rex(:funame, i, /[a-z][a-z0-9]*/); end
  def fun(i); seq(:fun, i, :funame, :args); end

where the fun parser is a sequence combining the funame parser then the args one. :fun (the first argument to the basic parser seq) will be the name of the resulting (local) parse tree.

Below is a list of the basic parsers provided by Raabro.

The first parameter to the basic parser is the name used by rewrite rules.
The second parameter is a Raabro::Input instance, mostly a wrapped string.

def str(name, input, string)
  # matching a string

def rex(name, input, regex_or_string)
  # matching a regexp
  # no need for ^ or \A, checks the match occurs at current offset

def seq(name, input, *parsers)
  # a sequence of parsers

def alt(name, input, *parsers)
  # tries the parsers returns as soon as one succeeds

def altg(name, input, *parsers)
  # tries all the parsers, returns with the longest match

def rep(name, input, parser, min, max=0)
  # repeats the the wrapped parser

def ren(name, input, parser)
  # renames the output of the wrapped parser

def jseq(name, input, eltpa, seppa)
  # seq(name, input, eltpa, seppa, eltpa, seppa, eltpaa, seppa, ...)

def eseq(name, input, startpa, eltpa, seppa, endpa)
  # seq(name, input, startpa, eltpa, seppa, eltpa, seppa, ..., endpa)

LICENSE

MIT, see LICENSE.txt