OpenCloud
Replies: 3 comments 2 replies
|
OpenCloud uses Tika to index data for search. Depending on the Tika image used, it also supports OCR of PDFs and other formats such as images. You need to use the tika-full image in docker (compose). You might want to test if that fits your needs. |
|
That sounds really interesting; I’ll give it a go and let you know how it goes. |
|
I asked an AI to explain to me what Tika-Full will do. This is how I understand it now: It will perform OCR on a PDF. However, the result will not be saved in the PDF. The OpenCloud full-text search will still work, though, as the OpenCloud database receives information about the document. MAC OS and Windows will not benefit from this. I would therefore need to have OCRmyPDF implemented as well, so that these operating systems can also benefit from the result. This would mean I would no longer need Tika-Full, provided I can get the text incorporated into the PDF. |
|
Feedback: I’ve currently got OpenCloud (for testing) running successfully in Komodo via Docker Compose. It might also work in Dockhand with the next version (1.0.39). My request for multi-Compose support is currently being implemented there. I’ve just tested the full-text search with Tika Full. I particularly like the full-text search in the web interface and in the app, even for images. Above all, it’s great that the full-text search works immediately after uploading the file. I’ll be trying out a workaround for my scanned documents over the next few days. I’m now convinced that OpenCloud is much better suited to my use cases than Nextcloud. My planned workflow looks like this: Via Docker Compose stack: In general, however, I think it would be a great feature if newly uploaded PDFs without OCR recognition were automatically processed with OCR. That would simplify the workflow considerably and would certainly be a major added value for many users. Particularly when searching on Windows and Mac. |
My Workaround: opencloud-ocrmyscanI have built a temporary solution for my scanning workflow and have now developed it to a point where I feel comfortable recommending it to others as well. I have also tried to document the setup and configuration as thoroughly as possible: The Docker Compose setup watches scan folders, processes incoming PDFs with OCRmyPDF, embeds the recognized text into the PDF, and uploads the resulting searchable PDF/A files to OpenCloud via WebDAV. The workflow also includes options for scan processing such as rotating and deskewing pages. For my previous Google Drive workflow, I used ABBYY FineReader on Windows to prepare scanned documents before uploading them. This solution now covers that controlled scanning workflow in an automated and self-hosted way as a Docker Compose stack running on my UGREEN NAS. On other platforms such as Nextcloud, OCR processing can already be integrated into the upload workflow. But personally, I really like the responsiveness of OpenCloud and the ability to use full-text search in the iOS app. That is why OpenCloud is still my preferred solution. I would therefore still love to see OCR text embedding become a native part of OpenCloud at some point. This workaround only covers documents that go through the defined scanning workflow. A PDF without an OCR text layer that is uploaded directly through the web interface or an app will not be processed by it. The same applies to such documents synced to Windows or macOS, where local search still cannot make use of text that was never embedded into the PDF. So this is mainly a bridge for my scanning workflow rather than a replacement for native OCR support in OpenCloud. Hopefully it can be useful to others with a similar setup as well. 🙂 
|
Uh oh!
There was an error while loading. Please reload this page.
EN:
I have looked at various file-syncing solutions, including Nextcloud, Seafile, pCloud, Google Drive, OneDrive and, most recently, the OpenCloud demo.
There is one important feature currently missing from OpenCloud: the ability to automatically apply OCR to PDF files that do not already have OCR recognition.
Background
A modern file storage system should make it possible to find all documents quickly and easily. That is why I consider the following features to be particularly important:
Full-text search on a Mac, for example, is no problem for synchronised documents in Google Drive. However, this requires the PDF files to already have OCR text recognition.
In Nextcloud, I was able to meet almost all of these requirements. The only limitation is the lack of full-text search in the iPhone app, which has not been available there for years.
That is why I became aware of OpenCloud. Based on my current impression, OpenCloud could meet almost all of the requirements mentioned. However, newly uploaded or scanned PDF documents are not currently processed automatically via OCR.
If OpenCloud were also to support integration with OCRmyPDF, I believe the solution would be a decisive step ahead of most self-hosted drive and file-syncing solutions.
DE:
Ich habe mir verschiedene File-Sync-Lösungen angesehen, darunter Nextcloud, Seafile, pCloud, Google Drive, OneDrive und zuletzt auch die Demo von OpenCloud.
Bei OpenCloud fehlt mir aktuell noch eine wichtige Funktion: die Möglichkeit, PDF-Dateien ohne vorhandene OCR-Erkennung automatisch mit OCR anzureichern.
Hintergrund
Eine moderne Dateiablage sollte es ermöglichen, alle Dokumente schnell und unkompliziert wiederzufinden. Deshalb halte ich folgende Funktionen für besonders wichtig:
Die Volltextsuche auf dem Mac ist bei synchronisierten Dokumenten in Google Drive beispielsweise kein Problem. Voraussetzung ist allerdings, dass die PDF-Dateien bereits über OCR-Texterkennung verfügen.
In Nextcloud konnte ich nahezu alle Anforderungen umsetzen. Die einzige Einschränkung ist die fehlende Volltextsuche in der iPhone-App, die dort bereits seit Jahren nicht verfügbar ist.
Deshalb bin ich auf OpenCloud aufmerksam geworden. Nach meinem aktuellen Eindruck könnte OpenCloud nahezu alle genannten Anforderungen erfüllen. Allerdings werden neu hochgeladene oder eingescannte PDF-Dokumente derzeit nicht automatisch per OCR verarbeitet.
Wenn OpenCloud zusätzlich eine Integration von OCRmyPDF unterstützen würde, wäre die Lösung aus meiner Sicht den meisten selbst gehosteten Drive- und Filesync-Lösungen einen entscheidenden Schritt voraus.
All reactions