I'm using docutils to transform and extract information from rst/sphinx
documents. I've made good progress, but I'm now looking for a way to
extract inline content from the documents and re-serialize it to rST.
The extraction should work (just recurse through non-Text non-Inline
content and grab Inline and Text stuff) but I don't really see a way to
re-serialize content to rST (such as an rST writer, which should be
useful to apply mechanical transformations to rST documents anyway),
is there none built-in or available through a third-party?
 to move from a terrible home-grown translation system to Sphinx's builtin one