Import & export

The whole app as newline-delimited JSON — settings, segments, templates, dynamic content, users, subscriptions, in-app messages and their engagement, messages, test devices, imports and deliveries — one object per line, in REST field names, streamed rather than buffered. Free, on every tier, at any time.

By default the stream holds one transaction open so every entity in the file is the same instant; the last line is a terminal record carrying per-entity counts, and the importer refuses a file without it, because a truncated stream otherwise looks exactly like a complete one.

What is not in it: journeys and journey history, custom events, and the daily usage rollups have no export entity. Nor do secrets and routing state — API keys, platform credentials, device credentials, Live Activity and push-to-start tokens, and media metadata are all deliberately withheld and named in the manifest.

One entity of the export rendered as CSV and streamed as an attachment. Most entities contain the same records as the NDJSON archive. subscriptions joins every device to all portable fields on its user profile, because a standalone CSV cannot express the archive's two related streams. test_subscriptions adds the test-device name and timestamp; importing it restores the subscription, user profile and named test-device marker. An unrecognised entity is a 404 that names the full list.

Every cell, including the header row, is escaped against spreadsheet formula injection: a value beginning with =, +, -, @, a tab or a carriage return comes back with a leading apostrophe. The CSV import route strips that back off; if you parse these files yourself, strip it too.

Imports subscriptions from a CSV uploaded as a multipart file field. Unlike the NDJSON restore this is an upsert on the push token, so it can be run against an app that already has data.

Small files are processed inline and answer 200 with the summary. Anything larger is spooled and handed to the job queue, answering 202 with an import_id to poll on GET /v1/apps/{app_id}/imports/{import_id}. Uploads over the server's CSV size ceiling are refused with 413 and the spool file is removed.

Reads an OpenPush NDJSON archive back in, uploaded as a multipart file field. This is a restore into an EMPTY app, not a merge — a target that already has subscriptions is refused with 409, which says how many it found. (409 rather than 400 on purpose: the file is fine, the target is wrong.)

Rows commit in batches as they are read, so a refusal partway through leaves the rows already written in place and says how many. There is deliberately no cleanup sweep — the empty-app precondition is what makes the remedy trivial: delete the app, make a new one, re-run the import.

The 25 most recent import jobs for one app, newest first, with row and byte progress, status, summary and any error. Older jobs are not reachable through this route.

Progress for one import job — status, rows done against rows total, the summary, and the error if it failed. This is the route to poll after a 202 from the CSV import.