Mistake search_dates for ru language

Hello everyone!

I have an issue with the following example:

search_dates('через 2 часа', languages = ['ru'])    # через 2 часа = in 2 hours
output:  [('2 часа', datetime.datetime(2020, 6, 11, 8, 58, 34, 823410))]
should be: [('через 2 часа', datetime.datetime(2020, 6, 11, 12, 58, 34, 823410))]

The answer is supposed to be "2 hours from now (later)", but it returns "2 hours before now (earlier)".
The thing is, the sentance is exactly following the structure defined in dateparser/dateparser/data/date_translation_data/ru.py , so I'm confused why the error takes place.
I try to figure out the logic of the parser, and so far I understood that for languages other than english it's supposed to return 2 results: one for translated to english sentance, and the other - for the original sentance. Did I get it correctly?

How can it be fixed?
Thanks!

Start by reproducing the search_dates('через 2 часа', languages=['ru']) example and inspect dateparser/dateparser/data/date_translation_data/ru.py, which the report identifies as relevant. Trace how the Russian phrase is interpreted and add or update coverage for the expected future datetime; done means the full phrase returns the correct two-hours-later result.