Breaking starbounder.org's fucking nasty ass shitty captcha

--What is the first digit of the number five hundred seventy-five million three hundred seventy-eight thousand eight hundred sixty-one?
--I DON'T KNOW !
--This is easier to do for a computer than a human
--This is the worst captcha i've ever seen
--please, don't use that kind of captcha

units = {}
units.one = 1
units.two = 2
units.three = 3
units.four = 4
units.five = 5
units.six = 6
units.seven = 7
units.eight = 8
units.nine = 9

deci = {}
deci.twenty  = 2
deci.thirty  = 3
deci.forty   = 4
deci.fifty   = 5
deci.sixty   = 6
deci.seventy = 7
deci.eighty  = 8
deci.ninety  = 9

special = {}
special.ten       = {1, 0}
special.eleven    = {1, 1}
special.twelve    = {1, 2}
special.thirteen  = {1, 3}
special.fourteen  = {1, 4}
special.fifteen   = {1, 5}
special.sixteen   = {1, 6}
special.seventeen = {1, 7}
special.eighteen  = {1, 8}
special.nineteen  = {1, 9}

function nullret()
    return "_","_","_","_","_","_","_","_","_"
end

function breakDown9d(tn)
    --tn: text number
    print(tn)

    --in case the whole captcha was inputted and not just the text number
    if string.find(tn, "of the number") then
        tn = string.match(tn, "of the number (.*)\?")
    elseif string.match(tn, "^ ") then
        tn = string.match(tn, "^ (.*)")
    end
    if string.find(tn, "\?") then
        tn = string.match(tn, "(.*)\?")
    end
    --now tn looks like:
    --five hundred seventy-five million three hundred seventy-eight thousand eight hundred sixty-one

    local a,b,c,d,e,f,g,h,i, rest
    local millions,thousands,hundreds

    millions, rest = string.match(tn, "(.*) million (.*)")
    if not rest then return nullret() end
    thousands, rest = string.match(rest, "(.*) thousand (.*)")
    hundreds = rest
    --split "five hundred seventy-five million three hundred seventy-eight thousand eight hundred sixty-one"
        --millions  = five hundred seventy-five
        --thousands = three hundred seventy-eight
        --hundreds  = eight hundred sixty-one

    a,b,c = breakDown3d(millions)
    d,e,f = breakDown3d(thousands)
    g,h,i = breakDown3d(hundreds)

    return a,b,c,d,e,f,g,h,i
end
function breakDown3d(tn)
    --tn looks like something like
    --"five hundred seventy-five"

    local a,b,c
    --split "one hundred twenty-six" -> "one", "twenty-six"
    --the ? is for the 9/100 chance of just getting "x hundred". Makes the space optional.
    a, rest = string.match(tn, "(.*) hundred ?(.*)")
    a = units[a] or "_"
    --because units["one"] is 1
    --if units[a] is undefined, or sets the value to "_"

    --if not rest then
    --  --"one hundred"
    --  a = 0
    --  rest = tn
    --end

    --parsing the last two digits. Lots of cases.
    if string.find(rest, "-") then
    --ex case: "one hundred [twenty-six]"
        b, c = string.match(rest, "(.*)\-(.*)")
        b = deci[b]
        c = units[c]
    elseif special[rest] then
    --ex case: "one hundred [twelve,thirteen...]"
        b, c = unpack(special[rest])
    elseif units[rest] then
    --ex case: "one hundred [four,five,six...]"
        b = 0
        c = units[rest]
    elseif deci[rest] then
    --ex case: "one hundred [twenty,thirty...]"
        b = deci[rest]
        c = 0
    elseif rest == "" then
    --ex case: "one hundred[]"
        b = 0
        c = 0
    end

    return a or "_", b or "_", c or "_"
end

--example of use
local captcha = "What is the fifth digit of the number six hundred seventy-six million seven hundred eighty-two thousand five hundred seventeen? "
local n = {breakDown9d(captcha)}
print(table.concat(n, " "))
print("1 2 3 4 5 6 7 8 9")