Tomlin: a TOML parser in Janet

[pyrmont]

I've recently completed Tomlin, a TOML parser in Janet. Tomlin converts a TOML-formatted string into a Janet data structure. Details of the conversion are in Tomlin's api.md.

One of the interesting things was seeing how relatively easy it was to convert the ABNF grammar for TOML to a parser expression grammar for Tomlin. So these rules:

toml = expression *( newline expression )

expression =  ws [ comment ]
expression =/ ws keyval ws [ comment ]
expression =/ ws table ws [ comment ]

became these rules:

:main (* :expr (any (* :nl :expr)) -1)
:expr (+ :keyval-expr :table-expr :cmnt-expr)

:cmnt-expr   (* :ws (? :comment))
:keyval-expr (* :ws :keyval :ws (? :comment))
:table-expr  (* :ws :table :ws (? :comment))

TOML is an intentionally simple language but I nevertheless think it's a good example of the power of PEGs.

A lot of the initial development time was actually spent writing conformance tests. The most popular repository of tests (iarna/toml-spec-tests) only includes output tests in YAML. Since (a) Janet doesn't have a YAML parser and (b) the in-progress official TOML conformance repository (toml-lang/compliance) proposes to use JSON, I wrote output tests encoded in the proposed encoding. These specs are not included in Tomlin but are available in toml-specs.