A very dumb PEG parser library.
Son to aabro, grandson to neg, grand-grandson to parslet. There is also a javascript version jaabro.
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.
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
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, eltpa, seppa, ...)
#
# a sequence of `eltpa` parsers separated (joined) by `seppa` parsers
def eseq(name, input, startpa, eltpa, seppa, endpa)
#
# seq(name, input, startpa, eltpa, seppa, eltpa, seppa, ..., endpa)
#
# a sequence of `eltpa` parsers separated (joined) by `seppa` parsers
# preceded by a `startpa` parser and followed by a `endpa` parser
seq
is special, it understands "quantifiers": '?'
, '+'
or '*'
. They make behave seq
a bit like a classical regex.
module CartParser include Raabro
def fruit(i)
rex(:fruit, i, /(tomato|apple|orange)/)
end
def vegetable(i)
rex(:vegetable, i, /(potato|cabbage|carrot)/)
end
def cart(i)
seq(:cart, i, :fruit, '*', :vegetable, '*')
end
# zero or more fruits followed by zero or more vegetables
end
(Yes, this sample parser parses string like "appletomatocabbage", it's not very useful, but I hope you get the point about .seq
)
An instance of Raabro::Tree
is passed to rewrite()
and rewrite_{name}()
functions.
The most useful methods of this class are:
class Raabro::Tree
# Look for the first child or sub-child with the given name.
# If the given name is nil, looks for the first child with a name (not nil).
#
def sublookup(name=nil)
# Gathers all the children or sub-children with the given name.
# If the given name is nil, gathers all the children with a name (not nil).
# When a child matches, does not pursue gathering from the children of the
# matching child.
#
def subgather(name=nil)
end
I'm using "child or sub-child" instead of "descendant" because once a child or sub-child matches, those methods do not consider the children or sub-children of that matching entity.
Here is a closeup on the rewrite functions of the sample parser at doc/readme1.rb (extracted from an early version of floraison/dense):
require 'raabro'
module PathParser include Raabro
# (...)
def rewrite_name(t); t.string; end
def rewrite_off(t); t.string.to_i; end
def rewrite_index(t); rewrite(t.sublookup); end
def rewrite_path(t); t.subgather(:index).collect { |tt| rewrite(tt) }; end
end
Where rewrite_index(t)
returns the result of the rewrite of the first of its children that has a name and rewrite_path(t)
collects the result of the rewrite of all of its children that have the "index" name.
By default, a parser will return nil when it cannot successfully parse the input.
For example, given the above Fun
parser, parsing some truncated input would yield nil
:
tree = Sample::Fun.parse('f(a, b')
# yields `nil`...
One can reparse with error: true
and receive an error array with the parse error details:
err = Sample::Fun.parse('f(a, b', error: true)
# yields:
# [ line, column, offest, error_message, error_visual ]
[ 1, 4, 3, 'parsing failed .../:exp/:fun/:arg', "f(a, b\n ^---" ]
The last string in the error array looks like when printed out:
f(a, b
^---
Consider the following toy parser:
module ToPlus include Raabro
# parse
def to_plus(input); rep(:tos, input, :to, 1); end
# rewrite
def rewrite(t); [ :ok, t.string ]; end
end
Sample::ToPlus.parse('totota')
# yields nil since all the input was not parsed, "ta" is remaining
Sample::ToPlus.parse('totota', all: false)
# yields
[ :ok, "toto" ]
# and doesn't care about the remaining input "ta"
Sample::ToPlus.parse('totota', error: true)
# yields
[ 1, 5, 4, "parsing failed, not all input was consumed", "totota\n ^---" ]
The last string in the error array looks like when printed out:
totota
^---
MIT, see LICENSE.txt