Plural Formula gets ignored for Singular Case for languages where Singular Form is not found in the first index

The logic change was introduced in #532512-38: Plural string storage is broken, editing UI is missing, likely a simple faulty assumption was made.

It seems to me the following simplification can be made as a step forward:

$translated_array = explode(PoItem::DELIMITER, $this->translatedString);
// No need to use plural formula if the translation only contains the singular case.
$index = count($translated_array) === 1 ? 0 : $this->getPluralIndex();
// Nth plural form, fallback to second plural form.
$return = $translated_array[$index] ?? $translated_array[1];

From if ($this->count == 1 || $index == 0 || count($translated_array) == 1) { I kept only the last condition. As this issue states, the presumption $this->count == 1 indicates $index = 0 is faulty, so that needs to be removed and the $index == 0 case is superfluous anyways because both the if and the else case is $return = $translated_array[0] for zero index.

I think the fallback only exists when somehow locale_get_plural doesn't exist or a translation is not provided for all cases of the plural formula and so while it's not quite correct to use the presumption that [1] is somehow appropriate it is also quite irrelevant. I think so but I am not sure. Perhaps the last translation should be used instead of the second as that's usually more of an "open" case?

Ps1.: please continue not crediting me on core issues.

Ps2: speaking of faulty presumptions in this area, #1273968-2: remove eval from locale.module presumed precomputing 0-199 is enough but quite a few languages including French, Italian, Portuguese and Spanish contain a special case for 1000000. And Cornish is a nightmare. Another issue should be filed to fix this . It'll be fun. https://php-gettext.github.io/Languages/

Don't see an MR. but it should have test coverage when opened please

Yeah, tests fail... here's a slightly better version, with comments:

    $translated_array = explode(PoItem::DELIMITER, $this->translatedString);
    // No need to check plural formulas if only one translation is provided.
    $index = count($translated_array) === 1 ? 0 : $this->getPluralIndex();
    // If the translation is not found, it's guesswork, and the best guess
    // is the most common case where the first plural form is for n = 1 and
    // the second is for everything else. It's not always correct but
    // ¯\_(ツ)_/¯
    $return = $translated_array[$index] ?? $translated_array[$this->count != 1];

The fallback was changed from fixed [1] to [$this->count != 1] (perhaps amend the comment to emphasize strict comparison doesn't work here -- it can be float for example). There are multiple cases when it's incorrect, like "0 is a special case" which what this issue is about but also say Czech has n == 1 ? 0 : n >= 2 && n <= 4 ? 1 : 2 which means for all n > 4 the guess will use the second translation but the third would be correct. But: since either the plural formula is not accessible or there are no translations for this count , it's just impossible to know for sure which is the correct one. So it tries to do ... something. By this point we know there are two translations and the most common plural formula is n!=1 and so the highest chance of being correct is to just use the second translation. It's ... not possible to do much better because , really, if you have three Slovenian translations which does n % 100 == 1 ? 0 : n % 100 == 2 ? 1 : (n % 100 == 3 || n % 100 == 4) ? 2 : 3 well good luck knowing which one is missing.

Sure, it might be too much I was trying to optimize but it was just an idea.