Documents — Take Confirmed Bundles

Type: documents-take-classified  ·  Package: Documents Activities v1.4.5  ·  Output: List<object>

Since: v1.3.0

Picks up the bundles whose split a person has already confirmed in the classification station. It downloads each one’s file, takes it out of the inbox (so it isn’t processed twice) and returns them with their local path and their confirmed segments. The playbook iterates and extracts each segment.

Documents — Take Confirmed Bundles   → output: bundles

For Each   items = bundles   itemVariable = bundle
   └── For Each   items = bundle["segments"]   itemVariable = segment
          └── IDP — Extract   path     = bundle["path"]
                              document = bundle["document"]
                              project  = segment["type"]
                              pages    = segment["pageStart"] + "-" + segment["pageEnd"]

How it works

1. Asks Zoan Cloud for the confirmed bundles (up to limit).
2. For each one, downloads the file to outputDir as <id>-<original name> (creating the folder if it does not exist) and, if the bundle has an OCR cache, downloads it alongside (….layout.json).
3. Marks it as taken, so it is not returned again.
4. Returns the list. The log says how many bundles it took.

With no confirmed bundles, it returns an empty list. It needs the Zoan Cloud runtime context.

Required parameters

None.

Optional parameters

Parameter	Type	Default	Description
limit	int	50	Maximum bundles to take at once.
outputDir	string	the system temp folder	Where to download the files. Created if it does not exist.

Output

A List<object> of bundles; each one:

Field	Type	What it is
id	string	The bundle’s id in Zoan Cloud.
reference	string	The key it was uploaded with.
path	string	The local path of the downloaded file.
segments	List	The confirmed segments: [{ pageStart, pageEnd, type, recognized }].
document	Dictionary	The full reading stored as an OCR cache, or null if the bundle has none.

The reading comes back with the bundle

If the bundle was uploaded with its OCR cache (the document from Documents — Send Bundle to Classification), each returned bundle carries document: the full reading. Pass it to IDP — Extract together with each segment’s range (pages) and the extraction slices that reading instead of re-reading — the same paper is not paid for twice. If there is no cache (or it could not be downloaded, which leaves a warning in the log), document comes empty and extraction goes by path as usual.

Idempotency

Taking a bundle marks it as taken, so it is not returned again. If extracting a segment fails, IDP — Extract’s idempotency (by reference) avoids paying again for the segments already extracted on a retry.

Run it on a schedule

Since human confirmation is asynchronous, the usual approach is to run this playbook with a scheduled trigger that keeps draining what has been confirmed.

Example

Extract each confirmed segment with its own reference, grouped by the source bundle:

Documents — Take Confirmed Bundles   limit = 20   outputDir = workFolder
                                     → output: bundles
For Each   items = bundles   itemVariable = bundle
  For Each   items = bundle["segments"]   itemVariable = segment
    IDP — Extract   path          = bundle["path"]
                    document      = bundle["document"]
                    project       = segment["type"]
                    pages         = segment["pageStart"] + "-" + segment["pageEnd"]
                    reference     = bundle["reference"] + "#" + segment["pageStart"]
                    sourceGroupId = bundle["reference"]
                    → output: res
    Log   message = res["reference"] + " to review: " + toString(res["needsReview"])

Common errors

Message	Cause and fix
Document activities need a Zoan Cloud runtime context, which the Runner provides when it launches the playbook. Missing: …. …	The run has no Zoan Cloud. Run it from Zoan Cloud or from the Designer with a session.
Zoan Cloud rejected the request (…): …	Zoan Cloud did not accept the query, the download or the taken mark. The detail carries the code; retry.
Layout cache unavailable (…); the extractor will re-read the file. (warning)	The OCR cache could not be downloaded. It does not fail: IDP — Extract will read the file.

Related activities

• Documents — Send Bundle to Classification — sends the uncertain bundle to the station.
• IDP — Extract — extracts each confirmed segment.