extract Keyword break my grammar

[Tudyx]

While experimenting for a general tree-sitter grammar for log files, I've come to the point where I want to extract nodes whether there are within parentheses or not.
For instance, for the given log, I want to extract all the apples:

apple (apple)

A simple grammar that does exactly what I want is this one:

 rules: {
    log_file: $ => repeat(
      choice(
        $.apple,
        $.apple_parenthesis,
      )
    ),
    apple: $ => "apple",
    apple_parenthesis: $ => seq('(', $.apple, ')'),
 }

That's great but if add some word with "apple" in them, they also get selected. For instance, the "bad_apple" in this log get also selected:

apple (apple) bad_apple

If I understand well, the way to emulate "word_bound"/ \b using tree sitter is to use the keyword extraction feature. So I modify the grammar to take this in account

word: $ => $.word,

rules: {
    log_file: $ => repeat(
      choice(
        $.apple,
        $.apple_parenthesis,
        $.word,
      )
    ),
    apple: $ => "apple",
    apple_parenthesis: $ => seq('(', $.apple, `)`),
    word: $ => /\S+/,
 }

With this bad_apple doesn't get selected anymore but so do (apple) 😢 . (apple) get parsed as a word

(log_file [0, 0] - [1, 0]
  (apple [0, 0] - [0, 5])
  (word [0, 6] - [0, 13])
  (word [0, 14] - [0, 23]))

As word is the last rules of my grammar, I expect it to have the lowest priority. I've tried to play with pre() or conflicts but without success.

Obviously, something like

choice("apple", "(apple)"),

will select what I want but my goal is to access the inner apple for syntax highlighting .

Please could you help me to select only the good apples?

[WillLillis]

Hi! I spent some a little time working on this and I think I figured it out :). I think the issue was your use of \S for word, as that seems to grab the parentheses in (apple). I used a regex for word below to match any non-whitespace character besides '(' and ')', but you could also potentially use something like word: $ => /[a-zA-Z0-9_]+/.

module.exports = grammar({
    name: 'apple',

    word: $ => $.word,

    rules: {
        log_file: $ => repeat(
            choice(
                $.apple,
                $.apple_parens,
                $.word
            )
        ),

        apple: $ => "apple",
        apple_parens: $ => seq('(', $.apple, `)`),
        word: $ => /[^\s()]+/
    }
});

[Tudyx]

Thanks for your help! I've also ended up with the same solution on my side, see this grammar

[WillLillis]

Nice!