Implicit whitespace rules

[alexchandel]

Issue type

• Feature Request

Description

Manually inserting whitespace rules is somewhat cumbersome & error-prone, especially for larger grammars with 100s of rules. One of the nicest features of Perl's grammar-parsing package, Regexp::Grammars, is its implicit whitespace rule.

Essentially, Regexp::Grammars has two types of rules, "tokens" and "rules." In rule definitions, most sequences of whitespace are treated as implicit invocations of the <.ws> rule, which is automatically predefined to \s* & captures nothing (since matching whitespace is often useless). In token definitions, whitespace is ignored as in PEG.js.

It would be nice if PEG.js had an alternate rule type, like ::=, that implicitly matched whitespace.

[futagoza]

This has been bought up before (see #11), but before I tackle this, I want #256 to be implemented first. Would also prefer #311 and #523 to be done first, but they just ease the use of this feature request.

As for an alternative rule type invoked by syntax sugar like ::=, after #256 is done, it should be easier to implement (quite simply, it would just enable the rule annotation for ignoring whitespace, e.g @ignore(...ruleNames)).

[alexchandel]

I've made do with the _ rule for now. Since regex anchors like \b are missing from pegjs, I had to cook up a &[^A-Za-z0-9_] rule to terminate words with instead.

[futagoza]

PEG.js doesn't support using JavaScript native RegExps as expressions, which is why there is
no support for any anchors (see docs).