Find emoji byte index in a String in Rust

main.rs

extern crate unicode_segmentation;
use unicode_segmentation::UnicodeSegmentation;

use regex;

struct EmojiFinder {
    re: regex::Regex,
}

impl EmojiFinder {
    pub fn new() -> Self {
        Self {
            re: regex::Regex::new(r"\p{Emoji}|\p{Emoji_Presentation}|\p{Emoji_Modifier}|\p{Emoji_Modifier_Base}|\p{Emoji_Component}").unwrap(),
        }
    }

    /// Return byte indices of emojis in the text.
    pub fn find(&self, s: &str) -> Vec<usize> {
        let mut indices = vec![];
        let mut index = 0;

        for grapheme in s.graphemes(true) {
            if self.re.is_match(grapheme) {
                indices.push(index);
            }
            index += grapheme.bytes().len();
        }

        indices
    }
}

fn main() {
    let finder = EmojiFinder::new();

    dbg!(finder.find("Hello😀😇")); // [5, 9]

    // ZWJ (3 byte)
    dbg!(finder.find("👩😇")); // [0, 4]
    dbg!(finder.find("💻😇")); // [0, 4]
    dbg!(finder.find("👩‍💻😇")); // [0, 11]

    // Family (4byte char + ZWJ for each)
    dbg!(finder.find("👨😇")); // [0,  4]
    dbg!(finder.find("👨‍👦😇")); // [0, 11]
    dbg!(finder.find("👨‍👩‍👦😇")); // [0, 18]
    dbg!(finder.find("👩‍👩‍👦‍👦😇")); // [0, 25]

    // Variation (4 byte)
    dbg!(finder.find("👍😇")); // [0, 4]
    dbg!(finder.find("👍🏽😇")); // [0, 8]
}