Folder Analysis and Anonymization

Analyze and anonymize all documents in a Nextcloud folder as a single batch. Entities detected across multiple files are consolidated, so an entity recognized in one file is treated as the same entity in all other files.

API Endpoint

Start folder analysis

POST /api/anonymization/batch/folder

Request body — exactly one of folderId or folderPath is required:

Providing neither, or providing both, results in HTTP 400.

By folder path (human-readable, existing usage):

{
  "folderPath": "/Documents/WOB-2024"
}

By folder ID (rename-proof, ideal for integrations that already hold a Nextcloud node ID — e.g. the FilePicker from @nextcloud/dialogs, Files-app context actions, or other Conduction apps):

{
  "folderId": 12345
}

When folderId resolves to multiple mounts within the user's tree (the same file ID surfacing through personal storage + a share + a group folder), a mount with write permission is preferred so anonymized copies can be written back into the source folder. If no writable mount exists, the first readable node is used — extraction-only flows still work, but the subsequent anonymization step will fail to write back to a read-only location.

Response — always includes both identifiers regardless of which input was used:

{
  "batchId": "a1b2c3d4-...",
  "folderId": 12345,
  "folderPath": "/Documents/WOB-2024",
  "fileCount": 5,
  "files": [
    { "fileId": 101, "fileName": "report.pdf", "status": "uploaded" },
    { "fileId": 102, "fileName": "letter.docx", "status": "uploaded" }
  ]
}

The endpoint creates a batch from all files in the specified folder (flat scan, direct children only — subdirectories are skipped). A background extraction job is queued automatically. Path-based callers receive a free upgrade path: capture folderId from the response and use it on reruns to stay rename-proof.

Example: start analysis from a Nextcloud FilePicker result

The Nextcloud @nextcloud/dialogs FilePicker returns Node objects with a native fileid. Pass that directly — no path derivation required:

import { getFilePickerBuilder, FilePickerType } from '@nextcloud/dialogs'
import axios from '@nextcloud/axios'
import { generateUrl } from '@nextcloud/router'

const picker = getFilePickerBuilder(t('docudesk', 'Select folder to analyze'))
  .setMultiSelect(false)
  .setType(FilePickerType.Choose)
  .allowDirectories(true)
  .build()

const [folder] = await picker.pick()

const { data } = await axios.post(
  generateUrl('/apps/docudesk/api/anonymization/batch/folder'),
  { folderId: folder.fileid }
)

console.log(data.batchId, data.folderPath, data.fileCount)

Error responses

Status	Condition
400	Neither folderId nor folderPath provided, both provided, path/ID is not a folder, folder is empty, folder exceeds max batch size
401	Not authenticated
404	Folder not found (ID not accessible by the current user, or path does not exist)

Progressive Polling

Extraction runs as a background job. Poll for progress:

Batch status

GET /api/anonymization/batch/{batchId}/status

Returns overall progress, per-file status, and entity count.

Entity consolidation (progressive)

GET /api/anonymization/batch/{batchId}/entities

Available during extraction (not only after completion). Response includes:

{
  "entities": [...],
  "entityCount": 12,
  "complete": false,
  "filesProcessed": 3
}

• complete: false — extraction still in progress, partial results
• complete: true — all files extracted, full entity list
• filesProcessed — number of files analyzed so far

Entities are deduplicated across files using exact case-insensitive matching. The fileCount field shows how many files contain each entity.

Review and Anonymize

After extraction completes (batchStatus: "review"), review the consolidated entity list and anonymize:

POST /api/anonymization/batch/{batchId}/anonymize

Request body:

{
  "entities": [
    { "type": "PERSON", "value": "Jan Jansen" },
    { "type": "EMAIL", "value": "jan@example.com" }
  ]
}

Anonymized Output

Anonymized files are saved in a dedicated subfolder under the source folder with their original filename (no _anonymized suffix):

/Documents/WOB-2024/
  report.pdf              (original, unchanged)
  letter.docx             (original, unchanged)
  anonymised/
    report.pdf            (anonymized copy)
    letter.docx           (anonymized copy)

The subfolder name defaults to anonymised and is tenant-configurable via the admin settings (config key docudesk.anonymisation.output_subfolder_name). See the batch anonymization docs for shared configuration details.

Original files are never modified.

Source-discovery filter

Files whose base name ends with _anonymized (legacy outputs from pre-layout runs) are excluded from folder source discovery. This prevents re-anonymising legacy redacted files in a re-run. Legacy files are left as-is; operators may delete or rename them manually.

Move-failure fallback

When the post-anonymisation move to the subfolder fails (permissions, disk error), the anonymised file is preserved at its legacy path (<source>/<base>_anonymized.<ext>) and the per-file response includes a warning field with code: "MOVE_FAILED". The anonymisation itself is still counted as successful.

Background job behaviour

The background FolderExtractionJob applies the same source-discovery filter: _anonymized-suffixed files in the source folder are skipped during the extraction phase. Per-job retry is idempotent — if a file is already extracted, the retry skips it without re-processing.

Batch State

Batch state is stored in Nextcloud's distributed cache with a 2-hour TTL. The TTL resets on every status or entity poll (keep-alive pattern), so the batch remains active as long as it is being used.

Configuration

Setting	Key	Default
Maximum files per batch	docudesk_batch_max_files	100

Configurable by admins via IAppConfig.