Summary

Parsley is a parsing library for people who find parsers scary or annoying. I wrote it because I wanted to parse a programming language, and tools like PLY or ANTLR or Bison were very hard to understand and integrate into my Python code. Most parser generators are based on LL or LR parsing algorithms that compile to big state machine tables. It was like I had to wake up a different section of my brain to understand or work on grammar rules.

Parsley, like pyparsing and ZestyParser, uses the PEG algorithm, so each expression in the grammar rules works like a Python expression. In particular, alternatives are evaluated in order, unlike table-driven parsers such as yacc, bison or PLY.

Parsley is an implementation of OMeta, an object-oriented pattern-matching language developed by Alessandro Warth at http://tinlizzie.org/ometa/ . For further reading, see Warth’s PhD thesis, which provides a detailed description of OMeta: http://www.vpri.org/pdf/tr2008003_experimenting.pdf

How It Works

Parsley compiles a grammar to a Python class, with the rules as methods. The rules specify parsing expressions, which consume input and return values if they succeed in matching.

Interface

The starting point for defining a new grammar is parsley.makeGrammar(grammarSource, bindings), which takes a grammar definition and a dict of variable bindings for its embedded expressions and produces a Python class. Grammars can be subclassed as usual, and makeGrammar can be called on these classes to override rules and provide new ones. Grammar rules are exposed as methods.

Example Usage

from parsley import makeGrammar
exampleGrammar = """
ones = '1' '1' -> 1
twos = '2' '2' -> 2
stuff = (ones | twos)+
"""
Example = makeGrammar(exampleGrammar, {})
g = Example("11221111")
result = g.stuff()
print result

→ [1, 2, 1, 1]