Pagination

pagination is either page-based (page, pageSize) or offset-based (start, limit). Strapi picks the mode from whichever fields are present. pageCount on a request is ignored. withCount: false is passed through when you do not need the total.

The third tuple element is meta, or null when the endpoint returns none. meta.pagination is one of:

{ page: number; pageSize: number; pageCount: number; total: number }
{ start: number; limit: number; total: number }

Every page in memory

all: true fetches every page and concatenates the results. The client requests the first page to learn the total, then the rest in the same mode, up to concurrency at a time (default 5, set on the constructor).

const [err, all, meta] = await articles.findMany({
	params: { pagination: { pageSize: 100 } },
	all: true,
});

const [offsetErr, offsetAll] = await articles.findMany({
	params: { pagination: { start: 0, limit: 100 } },
	all: true,
});

This is the right tool for hundreds of documents and the wrong one for hundreds of thousands.

One page at a time

pages() yields each page as it arrives and only asks for the next when you do. Each iteration is the same [error, data, meta] tuple. Breaking out of the loop stops the requests.

for await (const [err, batch, meta] of articles.pages({
	params: { pagination: { pageSize: 100 } },
})) {
	if (err) throw new Error(err.message);
	await writeRows(batch);
}

An error ends the walk, because there is no cursor to continue from. An empty page ends it too, so a stale total cannot spin forever. Both pagination modes work. If the server answers an offset request with page-shaped meta, the next request stays in the mode you asked for.

pages() does not take all. Selection narrowing applies to each page the same way it does on findMany.