Latest version always returns less specific language

[dlebech]

It looks like the library returns the least-specific version of a language after the discussion in #23 and the latest update. Is this expected behavior?

> const acceptLanguage = require('accept-language')
> acceptLanguage.languages(['en', 'en-US'])
> acceptLanguage.get('en-US')
'en'

I was expecting a match on en-US here. Reversing the order of 'en' and 'en-US' produces the correct result. Is this expected behavior and do we thus need to be careful with the ordering of the specific languages?

[tinganho]

It does the checking:

1. en-US == en ? no
2. en == en ? yes
3. en-US == en-US ? yes

And then it adds matched ones to a list. And it takes the first one in the matched list. But I do agree it should match based on the most specific as possible here.

[tinganho]

Just a question, what do you expect this should return?

> const acceptLanguage = require('accept-language')
> acceptLanguage.languages(['en', 'en-US'])
> acceptLanguage.get('en, en-US')
?

[tinganho]

@micky2be maybe have some inputs as well?

[dlebech]

@tinganho Thanks for your explanation. For your example, I would expect en because it's the first one on the accept list. I don't know why a browser would report en,en-US in that order though (my own browser reports en-US,en;q=0.8,da;q=0.6,sv;q=0.4), but I also admit that I don't know enough about how the accept languages work in general, nor have I fully read the RFC for this :-)

Basically, my assumption was that if there's a specific match early in the accept string, like en-US in my case, then that would always match first, no matter the order of the languages given to languages([...]).

[micky2be]

From the first issue it should return ’en-US’ obviously. It should match the higher choice first. Don't remember the code but my guess is that it matches both and return the first in the list.

If you have ’en’ and ’en-US’ in both accepted language and requested language, the priority is used. If equal priority, the order in the list takes over (closer to index 0 wins)

[tinganho]

This issue should be fixed now.