Add gettext support

[dutradda]

First translation was written for portuguese, according to https://cucumber.io/docs/gherkin/reference/#spoken-languages.

The PR also enables the language reset during runtime.

[dutradda]

@youtux, @olegpidsadnyi and @sliwinski-milosz

Hello guys,
can any of you take a look on my PR?

I have to change the tox/travis settings because the pytest 4.5 and after are crashing the tests.

[coveralls]

Coverage increased (+0.07%) to 93.615% when pulling 4a52482 on dutradda:gettext-support into 72a8764 on pytest-dev:master.

[sliwinski-milosz]

Travis has been fixed on master so you can revert your travis related changes

[dutradda]

@sliwinski-milosz
code updated

[olegpidsadnyi]

To be honest I don't really understand why do we need gettext for 11 keywords. This seems like an overkill. Another point is the current language. I think this should be a pytest option that the plugin introduces.

[olegpidsadnyi]

The whole point is to be compatible with the cucumber? To let people use pytetst instead of cucumber? Then the language support is like this:
https://cucumber.io/docs/gherkin/reference/#spoken-languages

#language header, and only keywords. There's no need in gettext. This is just a simple dict lookup once the header is set on the feature file. Otherwise english.

See https://github.com/cucumber/gherkin-python/blob/master/gherkin/gherkin-languages.json

[youtux]

Maybe we should instead import and use gherkin and delegate the parsing of feature files to this library, provided by cucumber.io.
If we use that, we will have automatic support for all the languages, and issues like #311 would probably not exist, or they would have to be fixed by the upstream.

[dutradda]

@olegpidsadnyi and @youtux
How we can proceed?

I really want to see multiple languages support on pytest-bdd. We already written the feature files in portuguese here. And the Gherkin language recommends that.

I can change my PR to fit the needs of the project.
But I need to know the next steps.

[youtux]

@dutradda, if you are willing to work on this, I would propose to continue in this way:

1. add the dependency gherkin-official to this project
2. parse feature files with that library, instead of having our own parsing logic

I am not sure how much work it would require, but I think it is a viable option.

[dutradda]

@youtux
I'll do it and keep you informed

[jsa34]

@youtux is this still relevant/needed?

[youtux]

@youtux is this still relevant/needed?

Now that we use the official parser it should come out of the box. #698

maybe we should add a test though to make sure that everything work fine when using a feature file in a different language

[jsa34]

Good point. I've got a few other issues that just need asserting and tests to close them. I'll add this to them