Translating languages is hard. You need to parse the original (with all the strange syntax and warts it allows above and beyond what the documentation says), determine what names mean, discover the semantics, and translate to the target language without losing the subtle details.
I'd be surprised if you found a solution for this off-the-shelf. Part of the problem is there are a vast number N of source languages (made worse by dialects), and a vast number of targets M, requiring a library of NxM translators to be lying around. And if you wait a few months, N and M both move. One hope is to translate N langauges to a common universal core language, and translate that to M targets, now only requiring N+M... but nobody has found a really universal language, and if you wait a few months, N and M move anyway.
One can consider building (or getting someone else to build) a specific translator for the task; this usually is uneconomical, at least if you want it to be reliable, because so much of the translator machinery (parsing, name resolution, pattern matching/translating, ...) tends to get rebuilt from scratch.
One can amortize the cost of the translator infrastructure if you are careful, by building (a lot of) shared machinery. Once you do that, building a translator is easier (still not easy) and the economics make more sense for individual cases. You can read more about this approach at: What kinds of patterns could I enforce on the code to make it easier to translate to another programming language?