Parser Toolkit

PTK - (c) Jérôme Laheurte 2015

What is PTK ?

PTK is a LR(1) parser "generator" for Python. It is not actually a "generator" in the sense that it doesn't output source code, using Python's dynamic nature to build everything it needs at runtime instead. Also, it supports asynchronous parsing; see the API documentation for details.

This code is licensed under the GNU LGPL version 3 or, at your option, any later version.

Why another one of those ?

There are a number of parser generators for Python out there. Most of them only support LL(1) or PEG. The other ones are either

  • Unmaintained
  • Straight translations from Yacc/Bison, and thus use an ugly syntax
  • All of the above

The main goals of PTK are

  • Clean, compact, Python-friendly syntax
  • Support for asynchronous input streams: why would you need the whole input string to start working when the underlying system is actually an automaton ?
  • Play nice in 'special' cases, like when the underlying 'filesystem' is a PyZipFile archive.
  • Don't use hacks like module-level introspection to compensate for an ugly design (I'm looking at you PLY). Those tend to produce subtle and headache-inducing bugs when running from compiled code.

Supported platforms

All unit tests pass on the following platforms/Python version:

  Linux OSX Windows
2.7 X X X
3.2 X   X
3.3     X
3.4     X
3.5   X  

See the Buildbot for details.

Although Python 2.6 is not officially supported, it does work with a few minor code modifications, namely replacing functools.total_ordering with the backport to 2.6.


Using pip:

$ pip install -U ptk

From source:

$ wget https://pypi.python.org/packages/source/p/ptk/ptk-1.3.2.tar.gz
$ tar xjf ptk-1.3.2.tar.bz2; cd ptk-1.3.2
$ sudo python ./setup.py install

Sample usage

Four-operations integer calculator:

from ptk.parser import LRParser, production, leftAssoc
from ptk.lexer import ReLexer, token
import operator, six

@leftAssoc('+', '-')
@rightAssoc('*', '/')
class Parser(LRParser, ReLexer):
    def number(self, tok):
        tok.value = int(tok.value)
    @production('E -> number<n>')
    def litteral(self, n):
        return n
    @production('E -> "-" E<val>', priority='*')
    def minus(self, val):
        return -val
    @production('E -> "(" E<val> ")"')
    def paren(self, val):
        return val
    @production('E -> E<left> "+"<op> E<right>')
    @production('E -> E<left> "-"<op> E<right>')
    @production('E -> E<left> "*"<op> E<right>')
    @production('E -> E<left> "/"<op> E<right>')
    def binaryop(self, left, op, right):
        return {
            '+': operator.add,
            '-': operator.sub,
            '*': operator.mul,
            '/': operator.floordiv
            }[op](left, right)

parser = Parser()
while True:
    expr = six.moves.input('> ')
    print parser.parse(expr)

Code samples

The samples subdirectory in the source tree contains the aforementioned calculator and a script that generates a skeleton Python file from a Yacc or Bison grammar file.

API documentation

The full documentation is hosted here.


Version 1.3.2:

  • Fix Python regular expression based lexer (use match instead of search)

Version 1.3.1:

  • Fix version number in README.rst

Version 1.3.0:

  • Added deferred_lexer and deferred_parser (asynchronous parsing using Twisted Deferred objects)
  • Asynchronous classes cannot be imported from 'regular' modules anymore, import them explicitely from 'ptk.async_lexer' and 'ptk.async_parser'.

Version 1.2.0:

  • Production methods cannot have the same name any more. This was idiotic to begin with. Inheritance thus works as expected.
  • Add AsyncLexer and AsyncLRParser for asynchronous parsing.

Version 1.1.0:

  • Added repeat operators ('*', '+', '?') in production syntax.
  • Support for more yacc/bison declarations in yacc2py sample (most are actually ignored)