Overview

I prototyped a TEI/XML file creation app using Google Cloud Vision API and GakuNin RDM, so this is a memo of that work.

Background

I needed an environment for creating TEI/XML files that reflect OCR results using Google Cloud Vision API. So I prototyped an environment that uses GakuNin RDM as the backend to manage files per user and execute OCR.

How to Use

Creating a Folder

Access the following.

https://ge-manager.vercel.app/

Log in using GakuNin RDM from the upper right of the screen.

A project list like the following is displayed.

Navigate to the appropriate level and press the create folder button.

Here, we create a folder named “sample.”

Then, press the “GE Manager” link.

You will be navigated to a page like the following.

Executing the Process

This time, we will use “fragm1a” from “e-codices - Virtual Manuscript Library of Switzerland.”

https://www.e-codices.unifr.ch/loris/gau/gau-Fragment/gau-Fragment_frag001a.jp2/full/full/0/default/jpg

Enter the image URL and press the upload button. Once uploaded, the screen changes to the following.

Next, press the “Execute OCR” button. If completed successfully, it displays as follows.

Next, press the “Create TEI/XML” button. If completed successfully, it is displayed along with the TEI/XML as follows.

Here is an example of displaying the downloaded file in Oxygen XML Editor. You can verify the OCR results from Google Cloud Vision API.

GakuNin RDM Files

The various files created through the above process are saved as files in the GakuNin RDM folder.

Reference: Preparing Image Files Accessible via URL

Using mdx.jp object storage to prepare image files accessible via URL.

This time, we create a bucket named ge-editor and prepare the following file.

{}""]VSet{}rast"""}""ieSEP,ARomifr"cenedfiDts"n"enDio:t:ccNou"ti"nr":""p:"c2g:a:e0[el["0-"""[:8eA:*"-dl"s"1il{]3g0to:e-owL-1r"ie7",sd",ti,Btuocrk"et","s3:GetObject"],

Then, by executing the following, files uploaded to the above bucket become downloadable.

ss33m:d/x/g%e-se3dcimtdor/-:noP-oclhieccyk-ucpedrattiefdicatesetpolicyconfig.jsons3://ge-editor

This is based on the following reference.

https://docs.mdx.jp/ja/index.html#bucket全体をまとめて公開する方法を教えてください。

Then, connect GakuNin RDM with the object storage by referring to the following.

This allows files uploaded via GakuNin RDM to be downloaded at URLs like the following.

https://s3ds.mdx.jp/ge-editor/files/sample/sample.jpg

Summary

I hope this serves as a helpful reference for creating draft text using OCR in the creation of TEI/XML files.