Document Editor & Pagination Engine

The existing package rebuilt pagination in a way that broke down under load — the timing and frequency of rebuilds, combined with unsafe DOM measurement, produced both wrong page splits and input lag on longer documents.

My first attempt was an incremental rebuild — recomputing only the region a change affected instead of the whole document. It worked for smaller documents, but broke down again as page and word count grew: the same freezing returned, just at a higher threshold.

I replaced it with three changes instead: throttling how often rebuilds ran, deferring non-urgent rebuild work to idle time, and making DOM measurement safer so it stopped interleaving badly with layout changes. That combination held up at the document lengths that actually mattered.

I built a page-tracking hook that locates paginated page nodes in the document, computes which page is in view on scroll, debounces the updates, and writes the current page into the editor's state store — shown in the editor footer as the current page of the total.

Creating a file from an upload needed its page count upfront. I added page-count extraction that reads PDF uploads via pdfjs-dist and Final Draft (FDX) uploads via the product's existing FDX parser, then passes that count through the create-file flow to be stored on the file record.

I extended the Express file-record helpers so the current page could persist, and extended the editor's client-side state and CSS so the pagination layout held together end-to-end.