require "uchar"

local function commatize(s, start = 1, step = 3, sep = ",")
    local function add_seps(n, dp)
        local lz = ""
        if !dp and n:startswith("0") and n != "0" then
            local k = n:lstrip("0")
            if k == "" then k = "0" end
            lz = "0":rep(#n - #k)
            n = k
        end
        if dp then n = n:reverse() end  -- invert if after decimal point
        local i = #n - step + 1
        while i >= 2 do
            n = n:sub(1, i - 1) .. sep .. n:sub(i)
            i -= step
        end
        if dp then n = n:reverse() end  -- invert back
        return lz .. n
    end

    local function is_exponent(c) return c in "eEdDpP^∙x↑*⁰¹²³⁴⁵⁶⁷⁸⁹" end

    local t = uchar.of(s)
    local acc = start == 1 ? "" : t:sub(1, start - 1)
    local n = ""
    local dp = false
    for j = start, t:len() do
        local c = t:get(j)
        if c >= "0" and c <= "9" then
            n ..= t:get(j)
            if j == t:len() do
                if acc != "" and is_exponent(acc[-1]) then
                    acc = s
                else
                    acc ..= add_seps(n, dp)
                end
            end
        elseif n != "" then
            if acc != "" and is_exponent(acc[-1]) then
                acc = s
                break
            elseif t:get(j) != "." then
                acc ..= add_seps(n, dp) .. t:sub(j)
                break
            else
                acc ..= add_seps(n, dp) .. t:get(j)
                dp = true
                n = ""
            end
        else
            acc ..= t:get(j)
        end
    end
    print(s);
    print(acc);
    print()
end

commatize("123456789.123456789", 1, 2, "*")
commatize(".123456789", 1, 3, "-")
commatize("57256.1D-4", 1, 4, "__")
commatize("pi=3.14159265358979323846264338327950288419716939937510582097494459231", 1, 5, " ")
commatize("The author has two Z$100000000000000 Zimbabwe notes (100 trillion).", 1, 3, ".")

local defaults = {
    "\"-in Aus$+1411.8millions\"",
    "===US$0017440 millions=== (in 2000 dollars)",
    "123.e8000 is pretty big.",
    "The land area of the earth is 57268900(29% of the surface) square miles.",
    "Ain't no numbers in this here words, nohow, no way, Jose.",
    "James was never known as 0000000007",
    "Arthur Eddington wrote: I believe there are 15747724136275002577605653961181555468044717914527116709366231425076185631031296 protons in the universe.",
    "   $-140000±100 millions.",
    "6/9/1946 was a good year for some."
}

for defaults as d do commatize(d) end