Separate finding the next document to parse and creating a job for it.

Cleans up the code somewhat more, imo. I'll stop now.