Download the PHP package acdh-oeaw/arche-ingest without Composer

A collection of ARCHE ingestion script templates

The REST API provided by the ARCHE is quite a low-level from the point of view of real-world data ingestions. To make ingestions simpler, the arche-lib-ingest library has been developed. While it provides a convenient high-level data ingestion API, it's still only a library which requires you to write your own ingestion script.

This repository is aimed at closing this gap - it provides a set of data ingestion scripts (built on top of the the arche-lib-ingest) which can be used by people with almost no programming skills.

Scripts provided

There are two script variants provided:

• Console scripts variant where where parameters are passed trough the command line.
  The benefit of this variant is easiness of use, especially in CI/CD workflows.
  • bin/arche-import-metadata imports metadata from an RDF file
  • bin/arche-import-binary (re)ingests a single resource's binary content (to be used when file name and/or location changed)
  • bin/arche-delete-resource removes a given repository resource (allows recursion, etc.)
  • bin/arche-delete-triples removes metadata triples specified in the ttl file (but doesn't remove repository resources)
  • bin/arche-update-redmine updates a Redmine issue describing the data curation/ingestion process (see a dedicated section at the bottom of the README)
• Template variant where you adjust execution parameters and/or the way the script works by editign its content.
  The benefit of this variant is that it allows to treat the adjusted script as a documentation of the ingestion process and/or adjust it to your particular needs.
  • add_metadata_sample.php adds metadata triples specified in the ttl file preserving all existing metadata of repository resources
  • delete_metadata_sample.php removes metadata triples specified in the ttl file (but doesn't remove repository resources)
  • delete_resource_sample.php removes a given repository resource (allows recursion, etc.)
  • import_binary_sample.php imports binary data from the disk
  • import_metadata_sample.php imports metadata from an RDF file
  • reimport_single_binary.php reingests a single resource's binary content (to be used when file name and/or location changed)

Installation & Usage

Runtime environment

You need PHP and Composer.

You can also use the acdhch/arche-ingest Docker image (the {pathToDirectoryWithFilesToIngest} will be available at the /data location inside the Docker container):

Console script variant

• Install with:

• Update regularly with:

• Run with:

  e.g.

  • To get the list of available parameters run

  e.g.

Running inside GitHub Actions

Do not store your ARCHE credentials in the workflow configuration file. Use repository secrets instead (see example below).

A fragment of your workflow's yaml config may look like that:

Running on ACDH Cluster

First, get the arche-ingestion workload console as described here

Then:

• Run screen -S mySessionName
• Go to your ingestion directory

• Run scripts using {scriptName}, e.g.

• If the script will take long to run, you may safely quit the console with CTRL+a + d followed by exit.
  • To get back to the script log again into repo-ingestion@hephaistos and run

Template variant

• Clone this repository.

• Run

• Adjust the script of your choice.

  • Available parameters are provided at the beginning of the script.
  • Don't adjust anything below the

  line until you consider yourself a programmer and would like to change the way a script works.

• Run the script with

  • You can consider reading input from a file and/or saving output to a log file, e.g. with:

  (see the section below for hints on the input file format)

Long runs

If you are performing time consuming operations, e.g. a large data ingestion, you may consider running scripts in a way they won't stop when you turn your computer off.

You can use nohup or screen for that, e.g.:

• nohup - run with:

  • If you want to run template script variants that way, you have to prepare the input data file.
    It should look as follows:

  e.g.

• screen

  • start a screen session with

  • Then run your commands as usual
  • Hit CTRL+a followed by a d to leave the screen session.
  • You can get back to the screen session with

Reporting errors

Create a subtask of the Redmine issue #17641.

• Provide information on the exact location of the ingestion script location (including the script file itself) and any other information which may be required to replicated the problem.
• Assign Mateusz and Norbert as watchers.

Using arche-update-redmine in a GitHub workflow

The basic idea is to execute data processing steps in a following way:

• note down the step name so we can read it instead of a failure
• perform the step
• call the arche-update-redmine

and have a separate on-failure job step which makes an arche-update-redmine call noting the faillure.

Remarks:

• As a good practice we should include the GitHub job URL in the Redmine issue note. For that we set up a dedicated environment variable.
• It goes without saying Redmine access credentials are stored as a repository secret.
• The way you store the main Redmine issue ID doesn't matter as it's not secret. Do it a way you want (here we just hardcode it in the workflow using an environment variable)