/v1
Queue an extraction for one video, or up to 200, against one schema.
Request
One call submits anything from a single video to two hundred. videos is always a list, so growing from one to many changes the length of an array rather than the endpoint you call.
batch_too_large and nothing is queued.name and description, and may set type — one of NuExtract 3.0's type names: verbatim-string, string, number, integer, boolean, date or enum (which also takes options). Without a type, the first column is verbatim-string and the rest are string. Up to 12 columns, and the same schema runs against every video in the call.schema.columns, with a schema.descriptions object giving one description per field. Nested objects and nested lists are refused. See /docs/guides/writing-a-schema.multiple, keeps at most this many rows per video and drops the rest. 1 to 500.schema_version; the latest is used otherwise.Send an Idempotency-Key header to make a retry of this call safe.
Response
Extraction runs on a queue, so this answers with an id rather than with rows. Read the job to collect them, or register a webhook_url and be told when there is something to collect.
Errors
400 for a malformed body or unusable schema, and for more than 200 videos — batch_too_large, with the message 200 video limit per batch exceeded. 401 for a bad key, 402 when the workspace is out of credits, 409 for an Idempotency-Key reused with a different body.
no_captions — costing nothing, while the rest of the job carries on.