Source code accompanying the presentation which you can find here. This presentation was originally given at 2021-03-30 during the final lecture of the University of Groningen's C++ course.
Run make in either of the two folders (simple or optimized). The source code requires C++20 (at least the optimized version does, compiler bounds for the simple version are probably more lenient).
Each of the two implementations currently contain all code in a single file. This is not very clean, but hopefully makes tinkering easier.
Code is heavily documented. Enjoy :-).
~Wiebe-Marten Wijnja
If you want to read more about this subject, here are a couple resources:
- Parsing: a timeline: A semi-complete (the last ~11 years are missing, and the author might be slightly biased) history of grammars and parsing techniques. Contains a lot more information about different kinds of parsing, and how they were introduced/used/superseded during the last 100 years.
- C++'s Boost.Spirit v2 guide, documentation and getting-started guide for the current LTS release of the Boost.Spirit combinator-based parser/lexer/prettyprinter (resp. named 'Spirit.Qi', 'Spirit.Lex', and 'Spirit.Karma').
- C++'s Boost.Spirit X3 guide, documentation and getting-started guide for the next release of Boost.Spirit, which is optimized in many ways because it only targets C++14 and beyond. However, it's not as feature-complete as v2 at this time (or at least the guides are not).
- C++'s PEGTL is an alternative C++ parser combinator library, whose syntax uses structs inheriting from templates, rather than using operator overloading. Also have very good documentation.
- C++'s Lexy another C++-based parser combinator library with slightly different design choices, and also a nice 'getting started' tutorial.
- Rust's Nom library page. A 'zero-copy' parser combinator focused mainly on parsing binary formats. Might give a nice insight in how parser combinators might be written/used in other systems languages.
- Haskell's Parsec was the first, original, implementation based on the theoretical papers, with top-notch documentation. Haskell nowadays has many 'spin-off'-libraries that to the user work exactly like Parsec but behind the scenes try to optimize for different use-cases. Parsec's source-code itself remains the most readable of the bunch. A lot of the code I wrote for the presentation is based on Parsec's examples and source code.
- Monadic Parser Combinators A long technical report about the Monadic-style (the original approach which is more flexible but less optimizable) of Combinator-based parsing.
- Generalized Parser Combinators A paper on writing Parser Combinators outputting a GLL-parser (that can thus accept ambiguous grammars and deal with left recursion).
- Invertible Syntax Descriptions: Unifying Parsing and Pretty Printing talks about using the same EDSL to create a parser-prettyprinter at the same time. This technique is already used widely in some Haskell web-frameworks to allow type-safe URL generation/recognition from a single URL description.