imprecise hint

[mircealungu]

This should say: you need to change one letter rather than add one. Or?

[tfnribeiro]

The library doesn't support danish letters (æ -> ae, ø -> oe and å -> aa)

[tfnribeiro]

I looked at another library called leven, which seems to not have this issue:

https://www.npmjs.com/package/leven

So we could migrate to this one instead?

[mircealungu]

yes, let's!

[tfnribeiro]

So turns out it might actually be due to another library we use: import removeAccents from "remove-accents";

The library does transform Æ -> AE and then that results in the problem we have now. I don't know what is best to do here, should it be to remove this but people might then struggle to type à á ã and so on or we try to avoid cases where it's only the ø æ å ?

[mircealungu]

write our own "library" for remove accents?

[mircealungu]

e.g. cf. Claude:

Source: https://claude.ai/share/7e1f36bf-fa2c-43a7-bdeb-d07aafc5e019

[tfnribeiro]

Seems fine on a first look - we could give it a try.

[tfnribeiro]

So I tested that and it doesn't really work for Å:

[tfnribeiro]

I asked DeepSeek to see what it comes up with and this was the solution:

function removeAccents(str) {
  // Temporarily replace these with unique placeholders
  let placeholderCounter = 0;
  const placeholderMap = {};
  let tempStr = str.replace(/[ÅåÆæØø]/g, (match) => {
    const placeholder = `__SCANDI_${placeholderCounter++}__`;
    placeholderMap[placeholder] = match;
    return placeholder;
  });
  // Now normalize and remove diacritics
  tempStr = tempStr.normalize("NFD").replace(/[\u0300-\u036f]/g, "");

  // Restore the Scandinavian characters
  return tempStr.replace(
    /__SCANDI_(\d+)__/g,
    (_, num) => placeholderMap[`__SCANDI_${num}__`],
  );
}

I think this could be slightly simplified since the only character that doesn't work with normalization is the Åå. I guess because it is seen as a Áá essentially.

[tfnribeiro]

Sidenote, the function:

function removeQuotes(x) {
  return x.replace(/[^a-zA-Z ]/g, "");
}

does more than just remove quotes. It also removes "-" for example. Is this the intended behavior? For example, umbrella in Portuguese is "guarda-chuva", so removing the dash here would make it technically incorrect same with "It's" and "Its", they are different words.

[mircealungu]

guarda-chuva is a good name :)

[tfnribeiro]

guarda-chuva is a good name :)

Literally meaning: "Guard Rain" 😆