LEXY_KEYWORD and whitespace

[AkaBlur]

Hi,

I'm still trying to wrap my head around lexy, for the most
part I could already create some simple parsers.

---

My issue is the behavior of LEXY_KEYWORD / dsl::keyword<Str>(Identifier)
As far as I understood, it tries to match the given Identifier on the supplied
input to create another input which must then match the supplied Str.

So basically Identifier must already be able to match Str.

The strange part (for my understanding) is the omitted whitespace at the end.
LEXY_KEYWORD expects a whitespace at the end of the input (which makes sense)
but doesn't consume it.

---

I've tried to create a simple keyword parser but I have to also parse the (actually expected)
whitespace after the expected keyword:

https://godbolt.org/z/s9z7b8EWT

It basically adds the usual whitespace class ws after the keyword.

constexpr auto ws = dsl::ascii::blank;
// ...
return LEXY_KEYWORD("int", id) >> ws + id;

The example for LEXY_KEYWORD didn't include this. Although it also didn't
supply a generated value.

struct production
{
    static constexpr auto rule = [] {
        // Define the general identifier syntax.
        auto head = dsl::ascii::alpha_underscore;
        auto tail = dsl::ascii::alpha_digit_underscore;
        auto id   = dsl::identifier(head, tail);

        // Parse a keyword.
        return LEXY_KEYWORD("int", id);
    }();
};

Would my approach be the "correct" way to implement LEXY_KEYWORD when trying
to match for specific keywords?

[foonathan]

If you want to skip whitespace, you need to enable whitespace skipping: https://godbolt.org/z/sTajc74sh

Read more here: https://lexy.foonathan.net/reference/dsl/whitespace/#doc

[AkaBlur]

I see, thanks.

So basically, LEXY_KEYWORD should always be used with automatic whitespace skipping?

[foonathan]

No, you can also use it with manual whitespace skipping like you did in your initial example. It just consumes characters until it no longer matches the identifier. This can be done when whitespace is reached or a punctuation character or anything else.