In general, if derived formats including ePUB and MOBI from HTML, also HTML from txt, also unwrapping txt from wrapped txt, are to work “correctly” then there needs to be *some* degree of expectation on the formatting of the incoming texts. Otherwise these tasks cannot be successfully automated.

Going the other way, the automated wrapping of txt is has built-in support by most (all?) modern text tools, including web browsers, e-book readers, text editors, etc.