Group Index Numbering

[Lucretiel]

I'm working with regex-syntax on some stuff, and doing some transformations on the parsed syntax. Do I correctly understand that the Capture indices are assigned essentially lexically, from left to right? If I'm removing certain capture groups from the tree but keeping others, I can just renumber them in the order that they're encountered from left-to-right, starting at 1? I'm planning on using the search_slots API and want to make sure I'm doing everything correctly.

[BurntSushi]

Yes, it's done lexically. Each capture group index can be determined completely by just assigning an index (starting at 1) to each opening parenthesis in the order in which it appears in the concrete syntax. So (?<a>(?<b>(?<c>))|(?<d>)) has a=1, b=2, c=3 and d=4. You can confirm this with regex-cli:

$ regex-cli debug hir '(?<a>(?<b>(?<c>))|(?<d>))'
    parse time:  24.225µs
translate time:  7.544µs

Capture(
    Capture {
        index: 1,
        name: Some(
            "a",
        ),
        sub: Alternation(
            [
                Capture(
                    Capture {
                        index: 2,
                        name: Some(
                            "b",
                        ),
                        sub: Capture(
                            Capture {
                                index: 3,
                                name: Some(
                                    "c",
                                ),
                                sub: Empty,
                            },
                        ),
                    },
                ),
                Capture(
                    Capture {
                        index: 4,
                        name: Some(
                            "d",
                        ),
                        sub: Empty,
                    },
                ),
            ],
        ),
    },
)