Is there a tool (or a way) to parse a pg_dump'ed (structure only) dump file into smaller files (each function in its own file, its table in its own etc.) with ability to combine these files later into the proper dump file?
The main problem is dependencies. Sometimes functions are defined before types, sometimes after. The same for cross-schema references. Dump parsing is simple, but this process drops dependencies between objects and could not be reversed easily.