theanalyst / rply

An attempt to port David Beazley's PLY to RPython, and give it a cooler API.

RPLY

Welcome to RPLY! A pure python parser generator, that also works with RPython. It is a more-or-less direct port of David Beazley's awesome PLY, with a new public API, and RPython support.

You can find the documentation online.

Basic API:

from rply import ParserGenerator, LexerGenerator
from rply.token import BaseBox

lg = LexerGenerator()
# Add takes a rule name, and a regular expression that defines the rule.
lg.add("PLUS", r"\+")
lg.add("MINUS", r"-")
lg.add("NUMBER", r"\d+")

lg.ignore(r"\s+")

# This is a list of the token names. precedence is an optional list of
# tuples which specifies order of operation for avoiding ambiguity.
# precedence must be one of "left", "right", "nonassoc".
# cache_id is an optional string which specifies an ID to use for
# caching. It should *always* be safe to use caching,
# RPly will automatically detect when your grammar is
# changed and refresh the cache for you.
pg = ParserGenerator(["NUMBER", "PLUS", "MINUS"],
        precedence=[("left", ['PLUS', 'MINUS'])], cache_id="myparser")

@pg.production("main : expr")
def main(p):
    # p is a list, of each of the pieces on the right hand side of the
    # grammar rule
    return p[0]

@pg.production("expr : expr PLUS expr")
@pg.production("expr : expr MINUS expr")
def expr_op(p):
    lhs = p[0].getint()
    rhs = p[2].getint()
    if p[1].gettokentype() == "PLUS":
        return BoxInt(lhs + rhs)
    elif p[1].gettokentype() == "MINUS":
        return BoxInt(lhs - rhs)
    else:
        raise AssertionError("This is impossible, abort the time machine!")

@pg.production("expr : NUMBER")
def expr_num(p):
    return BoxInt(int(p[0].getstr()))

lexer = lg.build()
parser = pg.build()

class BoxInt(BaseBox):
    def __init__(self, value):
        self.value = value

    def getint(self):
        return self.value

Then you can do:

parser.parse(lexer.lex("1 + 3 - 2+12-32"))

You can also substitute your own lexer. A lexer is an object with a next() method that returns either the next token in sequence, or None if the token stream has been exhausted.

Why do we have the boxes?

In RPython, like other statically typed languages, a variable must have a specific type, we take advantage of polymorphism to keep values in a box so that everything is statically typed. You can write whatever boxes you need for your project.

If you don't intend to use your parser from RPython, and just want a cool pure Python parser you can ignore all the box stuff and just return whatever you like from each production method.

Error handling

By default, when a parsing error is encountered, an rply.ParsingError is raised, it has a method getsourcepos(), which returns an rply.token.SourcePosition object.

You may also provide an error handler, which, at the moment, must raise an exception. It receives the Token object that the parser errored on.

pg = ParserGenerator(...)

@pg.error
def error_handler(token):
    raise ValueError("Ran into a %s where it wasn't expected" % token.gettokentype())

Python compatibility

RPly is tested and known to work under Python 2.6, 2.7, 3.1, and 3.2. It is also valid RPython for PyPy checkouts from 6c642ae7a0ea onwards.

Links

• Source code and issue tracker
• PyPI releases
• Talk at PyCon US 2013: So you want to write an interpreter?