masked method work incorrectly with unmasked results

[le0pard]

Describe the bug

All inputs have activated number mask with local pt-BR. Last input calculate sum for all other inputs and have mask too:

const mask = new Mask({ number: { locale: 'pt-BR', fraction: 2 } })
const totalAmount = Array.rom(document.querySelectorAll('.inputs')).reduce((agg, target) => {
  const value = target.value ? parseFloat(mask.unmasked(target.value.toString())) : 0

  return agg + (isNaN(value) ? 0 : value)
}, 0)
document.querySelector('.totalInput').value = totalAmount.toString()
document.querySelector('.totalInput').dispatchEvent(new Event('input', { bubbles: true })) // trigger mask.js

This lead that number incorrectly masked in total field. Instead of 10.34 it get 1034. So this mean:

mask.masked(mask.unmasked('10.234,54')) !== '10.234,54'

Steps to reproduce

const { Mask } = Maska;

const mask = new Mask({ number: { locale: 'pt-BR', fraction: 2 } })
console.log('correct', mask.unmasked('10.234,54')) // correct 10234.54
console.log('incorrect', mask.masked('10234.54')) // incorrect 1.023.454 (should be "10.234,54")
console.log('incorrect', mask.masked(mask.unmasked('10.234,54')), '!==', '10.234,54')

Reproduction link

Example

[rozsazoltan]

https://github.com/beholdr/maska/issues/225 - If I understand correctly, in both cases, the masked handles the float attribute value as an integer: what is incorrectly or not expected.

[le0pard]

question here: should mask.masked(mask.unmasked(a)) === a or we should expect non deterministic results?

@rozsazoltan - it is number (float), not integers. and it lost fraction

[le0pard]

I think problem, that for mask.masked it should not expect value in opts.locale, but in locale, which it did mask.unmasked, which looks like en

[beholdr]

Yes, it's a known issue with some locales. You can pass initially formatted value as a workaround:

mask.masked(Intl.NumberFormat({ locale: 'pt-BR' }).format(10234.54)) // = 10.234,54

[le0pard]

Yes, but why mask.masked(mask.unmasked(a)) === a non deterministic? If you return unmasked in locale, which you not expect to handle in masked, this mean logic is incorrect.

[le0pard]

const mask = new Mask({ number: { locale: 'pt-BR', fraction: 2 } })

Intl.NumberFormat('pt-BR').format(10234.54) // 10.234,54
mask.masked('10.234,54') // 10.234,54 - same, just waste cpu time

as I understand, expectation:

const number = 10223.34
const numberAsStr = number.toString() // '10223.34'
mask.masked('10223.34') // should return '10.223,34', but not working
mask.unmasked('10.223,34') // should return '10223.34', so parseFloat('10223.34') => 10223.34

and this is not happening now

[beholdr]

Yeah, it’s currently not working with locales that have dot as separator. 10223.34 is not the same as 10.234,54 that’s why I suggest to pass value pre-formatted with Intl.NumberFormat.format()

If you have a universal solution for parsing all locales, PR are welcome.

[le0pard]

@beholdr I think stuff like "parsing" and "formating" need to be split, so parsing can work with normal numbers (not only based on locale), while "formating" do format based requested locale

[beholdr]

@le0pard I agree with that, but it’s more complicated. Problem is not to parse a correctly formatted number like 10.234,56. Problem is to parse a partially formatted number, like you have 10.234,56 and then press a backspace and get 10.234,5 which you should parse to correct float. I just couldn't solve this case at the time.

[arthurgousset]

Hey @beholdr I saw you ran a quick root cause analysis. Would you be open to a contribution with a possible fix?

[beholdr]

@arthurgousset yes, of course. Unfortunately, I don't have time for deep analysis right now. But if you can solve this issue, it will be cool.

[arthurgousset]

@beholdr Unfortunately it looks like this is a larger design question. Our tool (workback.ai) works best with self contained, well defined bugs. Are there any such bugs you'd welcome a contribution for?

[le0pard]

works best with self contained, well defined bugs

in this case for now it useless, because such task can do even beginner developer :)

[arthurgousset]

such task can do even beginner developer

That's a great point @le0pard, would you say all self-contained bugs are successfully resolved by junior developers in repos you work in?

[le0pard]

yes

[arthurgousset]

That's awesome, sounds like you are on top of your game in the repos you work in!