Add matcher for checking missing interpolation keys

[creasty]

Wrote matcher for checking missing pairs of interpolation keys like:
'I am %{name}' and 'Soy %{nam3}'
'I am %{name}' and 'Soy name'

[aprescott]

This is a really nice addition. I think it would be good to actually include this as part of the functionality of the be_a_subset_of and be_a_complete_translation_of matchers. That way, not only are you asserting that translations aren't missing, you're also asserting that the interpolation keys didn't get altered or missed. What do you think?

[mixmix]

I built something similar a while ago : #15

From experience I recommend a test case which passes as long as the keys are all present, regardless of order

e.g expect
english: %{name} is now an admin of %{group}
bad_french: %{group} a un nouvel admin %{name}
to match

[creasty]

@aprescott

Well, actually I did in that way at the first place.
But I've changed it, after I thought translations didn't always have to fully contain interpolation keys across all languages.
And as I18.t, it simply ignores binding for missing keys, instead of raising an error.

So it's okay if be_a_complete_translation_of to take second argument to enable this feature.
How about that?

[aprescott]

@creasty interesting. I'm trying to imagine a case where you wouldn't want an interpolation key in one language even though it's used in another. I suppose if that's a legitimate use case, then yeah, an option argument seems fine to me! (Not a maintainer or owner of this repo, I just would really like to have the extra guarantee as a user of i18n-spec. :))

[creasty]

@s01us
It does, literally
https://github.com/tigrish/i18n-spec/raw/92c929732895270cd55ca552fa9fb99d3cc4dc9d/lib/i18n-spec/matchers/be_a_complete_interpolation_key_of_matcher.rb#L31

[creasty]

@aprescott aight, I'm gonna change it later

[creasty]

@s01us Ah, sorry I might have misunderstood your advise.
What you meant to say was to write fixtures having different order of keys so that spec could guarantee of its behavior, isn't it.

[creasty]

Added check_interpolation_key as the second argument to be_a_subset_of and be_a_complete_translation_of. And did a bit of refactoring.

[aprescott]

I don't know what the minimum version of Ruby is that is supported by this library, but it'd be great to use a keyword argument here of check_interpolation_key: false. I also have a feeling that a default value of true would be more useful.

[creasty]

Using keyword argument seems to me fine tho it's, in general, difficult problem to take Ruby version up to least 2.x unnecessarily.
And about the default value, I just want this feature to fail existing specs.

[deivid-rodriguez]

I just dropped support for EOL'd rubies, so we can actually apply this suggestion now 👍

[creasty]

Quite uncertain whether @tigrish is alive...

[creasty]

I'm tired of waiting.
I've extracted this feature into my own gem without depending on i18n-spec:
https://github.com/creasty/i18n_interpolation_spec

[deivid-rodriguez]

@creasty Are you still interested in getting this merged in? I think it would be a valuable addition!