Errors report incorrect column number when JSON includes multibyte characters

[fpdotmonkey]

If you have a JSON packet that has multibyte, i.e. non-ASCII, characters, then errors reported related to that packet such as the data error I trigger below will report a higher column number than you would see in a text editor, likely a byte offset.

This renders the reported column number useless in instances of non-English data or emojis and is magnified if the data is a large single-line packet, which isn't uncommon.

use serde::{Deserialize, Serialize};

fn main() {
    // `'not a number'` is at line 1 column 66
    let json = "{\"my_ascii_chars\": \"asdf\", \"my_utf8_chars\": \"/ˈa.ʎo/\", \"my_int\": \"not a number\"}";
    let test: Option<ColumnTest> = match serde_json::from_str::<ColumnTest>(json) {
        Ok(_) => panic!("no error generated"),
        Err(error) => {
            println!("{error}");
            assert_eq!(error.column(), 66);
            None
        }
    };
}

#[derive(Debug, Serialize, Deserialize)]
struct ColumnTest {
    my_ascii_chars: String,
    my_utf8_chars: String,
    my_int: u32,
}

Output:

invalid type: string "not a number", expected u32 at line 1 column 81
thread 'main' panicked at 'assertion failed: `(left == right)`
  left: `81`,
 right: `66`', src/main.rs:10:13
note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace