Pagination

Every collection answers { data, page }:

{
  "data": [],
  "page": {
    "number": 1,
    "size": 50,
    "total_items": 128,
    "total_pages": 3,
    "has_more": true
  }
}

Pass ?page= to move. Pages start at 1. Anything outside 1 to 1000 answers 400 invalid_request rather than an empty page, so a loop that lost count stops instead of spinning.

The catalog is the one exception to the size: it pages 24 at a time and caps at page 100, because it reads a search index rather than a table.

curl "https://uloggd.com/api/v1/library?page=2" \
  -H "Authorization: Bearer ulg_live_..."

Copies walk by cursor

/api/v1/library/copies is the one collection that does not walk by page number. It answers { data, page: { size, has_more }, next_cursor }, and the next page is ?cursor= with whatever came back in next_cursor.

curl "https://uloggd.com/api/v1/library/copies?limit=24&cursor=WyIyMDI2…" \
  -H "Authorization: Bearer ulg_live_..."

The reason is that a shelf changes while you walk it. Recording a copy while somebody is on page 2 pushes everything one place along, and an offset would hand back the row page 1 already showed. The cursor carries the sort key and the row's id together, so it points at a place rather than at a count.

Changing sort starts again: a cursor belongs to the order it was made in.