peggyjs / peggy-tag Goto Github PK

I just discovered peggyjs. Really nice work!

This peggy-tag library plays a similar role to my pegjs-template library for PEG.js. I'd love to contribute the features from that library to this one. In particular, I'd be interested in contributing:

1. The ability to interpolate in functions and use them as semantic actions. This allows you to get syntax highlighting and type checking for semantic actions, to use TypeScript in them, to capture JavaScript variables, etc etc. This isn't just interpolating the functions' source as a string; that would prevent things like capturing variables. Instead, under the hood, the functions are passed in to the parser via the grammar context.
2. The ability to define "partial" grammars that you can interpolate into other grammars, allowing you to share common definitions across multiple grammars. I find this pretty useful for lexical stuff in particular. Again, this isn't just interpolating the shared definitions as a string, because we want to take functions along for the ride.

I've found these features quite useful in my own work. What do you think; do these feel like a good fit for peggy-tag?

With the PR about to drop, line numbers will be correct from this code:

import peggy from "peggy-tag";

const parse123 = peggy`foo = $("f" "o"+)`;
console.log(parse123("foooo")); // "foooo"

but column numbers for errors thrown from the last line will be off by three. Find a way to determine where "foooo" actually starts.

• Use a test runner. Probably ava.
• Update CI matrix

@sanket143 can you take a look at this please, to see if it's more-or-less what you wanted, and to see if there's anything that can be improved?

It would be nice if peggy.GrammarError.format supported line and column offsets so that errors could be formatted with the CallSite of the place where the template originated. This will work in node, and possibly in v8-based browsers with the trick from stack-trace.

This probably means ts+terser+rollup.