QR code upload¶
With the QR code upload, files can be uploaded directly from a smartphone to an archival object – without having to log in on the mobile device.
Use cases¶
- On site in the archive: quickly take photographs of documents or objects and assign them directly to the right object
- Teamwork: share the QR code so that colleagues can upload files to an object
- External staff: temporary upload access without an Anton account
Setting up the QR code upload¶
The upload-qr-lods setting contains the levels of description for which the QR upload button is displayed, for example files and items [5,6]. To deactivate the QR upload, an empty array [] is saved, or the content of the field is deleted.
The button is shown to accounts with the role Editor or Admin, as for uploading through the form. An upload code applies to one record and expires after upload-qr-token-minutes (60 minutes unless set otherwise).
Activating the upload¶
1. Open the object¶
Navigate to the desired unit of description (a file or an item, for example).
2. Generate the QR code¶
- Click the QR Upload button (📱 icon) in the toolbar
- A modal opens with:
- a QR code to scan
- an upload URL to copy or share
- a token (validity configurable, default: 1 hour)
3. Choose processing options (optional)¶
Image processing options can be activated in the modal:
| Option | Description | Tool on the server |
|---|---|---|
| HEIC → JPEG | Convert iPhone images (HEIC/HEIF) to JPEG automatically; the JPEG becomes the master, the HEIC stays only with «Keep the delivered file» | ImageMagick (magick, with libheif) |
| Images → PDF | Combine several uploaded images losslessly into one PDF | img2pdf |
| OCR | Perform text recognition (makes PDFs searchable); scans become PDF/A-2b, a PDF with text stays PDF; the PDF with its text layer becomes the master, the delivered PDF stays only with «Keep the delivered file» | ocrmypdf + Tesseract (deu+eng) |
Below them is Keep the delivered file, off by default. HEIC → JPEG and OCR are part of taking the file in (preingest, see digital archive): their result is the master, and the delivered file is replaced. The event on the medium records the conversion with its tool and the checksums of both files. Ticked, the file as delivered stays as a kept original of the medium. With «images → PDF» the delivered images are always kept.
These settings are saved and apply to all uploads of the archive. Only accounts holding the capability Change digital preservation can change them (see User administration); everyone else sees the switches without being able to change them.
After saving, the page reports that the files have been received. They are processed and stored in the background and appear on the object once that is done, after a few minutes for many scans with OCR. An OCR run stops after 600 seconds (OCRMYPDF_TIMEOUT); the file is then stored without a text layer.
Pipeline details (for admins)¶
The pipeline runs in this order:
0. IN THE REQUEST
check token and files → job ProcessTokenUpload
1. BEFORE IMPORT (queue, ProcessTokenUpload)
HEIC → magick → JPG (HEIC kept as a version only with «Keep the
delivered file»)
JPGs → img2pdf → PDF
PDF → ocrmypdf → PDF/A-2b for scans, otherwise PDF, with OCR text layer (master;
the delivered PDF kept only with «Keep the delivered file»)
2. IMPORT
addAntonMedium() → event MediumAdded
3. AFTER IMPORT (async, queue)
MediumIdentifyAndConvert
├─ PRONOM identification
├─ cloud sync (if configured)
├─ thumbnail generation
└─ RefreshFulltext
└─ pdftotext → media_texts (full-text index)
The pipeline is controlled centrally via the image-upload-processing setting
(array). Possible values: heic2jpg, images2pdf, ocr.
Example with all three steps activated:
Setting::setValue('image-upload-processing', ['heic2jpg', 'images2pdf', 'ocr']);
Prerequisites on the server:
- ImageMagick with HEIF support (libheif)
img2pdf(Python tool)ocrmypdfwithtesseract-ocr-deuandtesseract-ocr-engpdftotext(packagepoppler-utils) — also needed for full-text extraction
Anton does not fail silently if a tool is missing — the pipeline stage concerned is skipped and documented in the log.