Compiled parser combinators and regular expressions in pure julia
77 Stars
Updated Last
1 Year Ago
Started In
April 2020

CombinedParsers in pure Julia

Dev Build Status Codecov A package for combining parsers and transforming strings into julia types.

Compose parsers parsimoneously within a functional parser combinator paradigm, utilize Julia's type inference for transformations, log conveniently for debugging, and let Julia compile your parser for performance.

The CombinedParsers design

  • is fast due to Julia parametric types, and compiler optimizations with generated functions,
  • parsing result transformations infer the domain data types,
  • is composable and optimizable with Julia method dispatch,
  • provides flexible public API for parsing, matching, iteration
  • can be defined with PCRE and EBNF syntax.

Getting started

CombinedParsers.jl is a registered package. Install with

] add CombinedParsers

Example: rational numbers arithmetics

This example demonstrates reading of arithmetical terms for rational numbers. Reflecting operator precedence, term are subterms, interleaved by */, and subterms are Either integer numbers

@syntax subterm = Either{Rational{Int}}([NumericParser(Int)]; convert=true)

or a subterm can also be an additive term in parentheses:

@syntax for parentheses in subterm
    mult = evaluate |> join(subterm, CharIn("*/"), infix=:prefix )
    @syntax term = evaluate |> join(mult,    CharIn("+-"), infix=:prefix )

This CombinedParser definition in 5,5 lines registers a @term_string macro for parsing and evaluating rational arithmetics:

julia> term"4*10+2"

Is every rational answer ultimately the inverse of a universal question in life?

Details in the full documentation example.


This package leverages Julia's compiler and superior type system to parsing.

I am thankful for contributions and inspiration from many great packages:


A bunch of fast text parsing tools, used in CSV.jl

CombinedParsers composes with fast TextParse.jl both ways because CombinedParser <: TextParse.AbstractToken and by providing a method for TextParse.tryparsenext, (leveraging the supreme Julia compiler, type and package architecture).

  • If you seek support with a CSV example, please contact me (e.g. address text field parsing).


  • The work was strongly inspired by the great Scala fastparse package, and also the elm parser.
  • Parsers.jl, a collection of parsers for date and primitive types, inspired the parse methods.
  • Automa.jl, a Julia package for text validation, parsing, and tokenizing based on state machine compiler. The package compiles deterministic finite automata. (Currently there is no inter-operation possible, because in Automa processing of parsed tokens is done with actions).
  • ParserCombinator.jl was a great inspiration. Yet I decided for a new design with a focus on transformations and type inference with parametric types, instead of basing this work off ParserCombinator, written before 2016 (and fixed for Julia 1.0 in 2018). CombinedParsers integrates into the Julia 1.0 Iteration API, small Union{Nothing,T} where T types instead of using Nullables, compiler optimizations and generated functions. I want to provide benchmarks comparisons with ParserCombinator.jl.

Next Steps

  • Syntax freeze -- your comments are appreciated!
  • decide for a error tracing strategy, backtracking. If you want to collaborate on stepping & debugging, please reach out to me.
  • Performance optimizations
  • streaming
  • test coverage underestimated (PCRE tests are not included in travis)
  • Code Style: Blue

Contributing and Questions

Contributions and feedback are very welcome, especially regarding brief syntax and constructor dispatch. Please open an issue if you encounter any problems or would just like to ask a question, or contact me at

Used By Packages

No packages found.