In reply to .
Couldn’t you iterate over all articles only once, and store (to a JSON file or so) any links (both source and target) you encounter, and work from there? When subsequently generating the actual pages, you’d merely have to fetch and include the relevant links from the existing array.
