2 releases

0.1.1 Jun 6, 2021
0.1.0 May 28, 2021

#236 in Parser tooling

Custom license

47KB
1.5K SLoC

ratatat

a parser combinator library for rust

goals

  • asymptotically fast parsers
  • practically fast for human-written files
  • easy to debug
  • syntax should look kinda like formal grammar
  • quick to experiment with

non-goals

  • fast for large data files
  • zero copy/allocation
  • general over input types (it's [u8] only)

concepts

steps for parsing a file:

  • read the file
  • convert to an Input
  • create a Context from the input
  • run a parser on the context
  • throw away the context and the input
  • use the results

Input

an input is a [u8] backed by a reference counted vec. this allows parsers to hold on to spans of the input, rather than copy them.

Context

a context has a reference to the input, and a cache of parsers. this allows parsers to precompute things for the entire input, memoize results, etc.

once parsing is complete, the context can be thrown away

Parser

a parser has a parse method that takes a context, a mutable position, and returns an optional result.

there are many built-in parsers, and parser combinators. for example:

  • str is a parser, which returns itself if the input contains it at the position
  • char behaves similarly
  • Map(parser,func) runs a parser, and on success, applies a function to its result
  • Filter(parser, func) runs a parser, then filters results by a function

Generator

a parser generator generates a parser for a type. this is the main way to make use of caching parsers in a Context

you can get a cached parser with context.parser::<Generator>() or parse immediately with context.parse::<Generator>(limit, &mut position)

there are also many built-in generators:

  • char generates a parser that parses any char
  • [Generator;N] generates a parser that parses a fixed-size array of values
  • Memo<Generator> generates a memoized parser from another generator

library structure

Take a look at the examples to see how it works. the json parser is a good place to start. usually, you'll want to implement Generator<YourType> for the root of your AST.

the core traits and types are in src/lib.rs

and the built-in parsers and combinators are in src/parsers/

Dependencies

~5–7.5MB
~139K SLoC