-
Authorize and login (request access token)
-
Create multiple documents (single create document calls, since you need to pair our id with your document)
-
Store Id/Ids from response
-
In batches of 50 check (please see the rule of thumb after this list) for documents state (collection filtered by ids, type and states [ready, not_ready, parent, data_duplicate, thrash, repairing]). You can check doc IDS from multiple accesible agendas.
-
Save data from response for documents in ready/not_ready state
-
In case returned document has state parent make another request to get his children (document detail endpoint for parent will return list of children)
-
After you iterate over your doc ids in processing state repeat steps 4,5,6 for the leftover array of document ids which are still unprocessed
Rule of thumb is to check for changes every 5 minutes during the first hour and after that every 60 minutes. In case you hit the request rate limit please implement graceful handling of 429 response code.
In case of split documents it can take multiple hours/days when some manual extraction is performed. Automatically processed documents will arrive first.
If you want to check if the parent processing if finished or in progress you can check the status of the parent document via detail endpoint. Finalized parent document detail will return multiple documents.