Plural rules definition about gettext

[oscarotero]

That's great. No problem, I can wait.

[mlocati]

Done. See https://github.com/mlocati/cldr-to-gettext-plural-rules/

[oscarotero]

Nice!
I think this data should be saved in an external php file, so it can be re-generated in any time without modify the Locales class. What do you think about implementing something like this?:

php convert.php phparray locales-data.php

So, in Locales class, we can create a method to retrieve this information:

public static function getData()
{
    if (!isset(self::$data)) {
        self::$data = include __DIR__.'/locales-data.php';
    }

    return self::$data;
}

[mlocati]

Nice idea. Done 😉

[oscarotero]

Ok, I've changed the Locales class to work with the new source. There are some changes, for example in the russian language that now has 4 plurals instead 3, I hope this is right (I have no idea).
I don't know whether the territories data is useful, but I've keept it anyway.

[mlocati]

The russian language that now has 4 plurals instead 3, I hope this is right (I have no idea)

Previously we had 3 plurals with this formula:
(n%10==1 && n%100!=11 ? 0 : n%10>=2 && n%10<=4 && (n%100<10 || n%100>=20) ? 1 : 2)

But the CLDR data says this:

"ru": {
    "pluralRule-count-one": "v = 0 and i % 10 = 1 and i % 100 != 11 @integer 1, 21, 31, 41, 51, 61, 71, 81, 101, 1001, …",
    "pluralRule-count-few": "v = 0 and i % 10 = 2..4 and i % 100 != 12..14 @integer 2~4, 22~24, 32~34, 42~44, 52~54, 62, 102, 1002, …",
    "pluralRule-count-many": "v = 0 and i % 10 = 0 or v = 0 and i % 10 = 5..9 or v = 0 and i % 100 = 11..14 @integer 0, 5~19, 100, 1000, 10000, 100000, 1000000, …",
    "pluralRule-count-other": "   @decimal 0.0~1.5, 10.0, 100.0, 1000.0, 10000.0, 100000.0, 1000000.0, …"
},

As you can see, there should be 4 plurals.
And the new formula is

(n % 10 == 1 && n % 100 != 11) ? 0
:
((n % 10 >= 2 && n % 10 <= 4 && (n % 100 < 12 || n % 100 > 14)) ? 1
:
((n % 10 == 0 || n % 10 >= 5 && n % 10 <= 9 || n % 100 >= 11 && n % 100 <= 14) ?
2
:
3
))

Which seems congruent with the CLDR data (for gettext we have that n == i and v == 0)

BUT the fourth case never occurs for integers... So, yes: it seems there's still some optimizations to be done...

[mlocati]

There are 4 cases that I have to investigate.

You have eagle eyes ;) Thanks!

[mlocati]

@oscarotero Ok, I added a procedure that removes the superfluous cases; the languages cleaned have been the 4 problematic ones: be- Belarusian,pl- Polish,ru- Russian anduk` - Ukrainian.

[mlocati]

Just a side note: Daiki Ueno (head developer of the gettext tools) seems interested in my cldr-to-gettext-plural-rules 😉

[oscarotero]

That's fantastic!!
I've pushed some changes with the new data exported by your tool and everything works fine.

You have eagle eyes ;) Thanks!

The phpunit test has eagle eyes 👓

[mlocati]

@oscarotero Just a notice: https://packagist.org/packages/gettext/languages

[oscarotero]

Ok.
Do you think it's better to include this package as a dependency and generate the data using the composer post-install event?

[mlocati]

If you include my package, you can also use the Language class without using pre-generated php code...