LEAF Bagger

Supports a preservation workflow through the creation of Archival Information Packages (AIP) from a LEAF environment.

:warning: These command-line scripts are only compatible with CWRC Repository v2.0 based on Linked Academic Editing Framework (LEAF). This replaces preservation workflow: cwrc_preservation used with the CWRC Repository v1.0 as v1.0 reached end-of-life Jan 5th, 2025.

Overview

The LEAF Bagger preservation toolkit contains scripts supporting a preservation workflow for a LEAF environment. The primary objective is to manage the flow of content from the CWRC repository into an OpenStack Swift repository for preservation (the destination may be extended for partner projects). Also, the repository provides an application to audit the contents of the source and preserved objects. The scripts are deployable within an OCI container to align with the deployment of CWRC Repository v2.0 and other LEAF installations.

Overview: Preservation workflow

• Poll Drupal for items to preserve

  • leaf-bagger.py
  • Drupal views to indicate new/changed items since a given date (paginated) - used by https://github.com/cwrc/leaf-isle-bagger to determine which items to add to the preservation workflow
  • views.view.preservation_show_media_timestamps.yml
    • https://${SITE_URL}/view/views/preservation_show_node_timestamps?page=${page}&_format=json
  • views.view.preservation_show_node_timestamps.yml
    • https://${SITE_URL}/views/preservation_show_media_timestamps?page=${page}&_format=json

• Create AIP (archival information packages) from the Drupal item (metadata and media)

  • leaf-bagger.py
  • Drupal module drupal/getjwtonlogin used by https://github.com/cwrc/leaf-isle-bagger and https://github.com/cwrc/islandora_bagger (forked from mjordan) to acquire a JWT token within the login response with the JWT token used to pull media/content from the Drupal site for preservation
  • https://github.com/cwrc/islandora_bagger (forked from mjordan) pulls metadata/media/content from the Drupal site for preservation
  • Drupal module Islandora Bagger Integration (forked from mjordan) provides many features but currently using:
  • REST endpoint to allow registering archival information package (AIP) creation events with Drupal
  • Drupal view of the resulting information (views.view.preservation_show_registered_events.yml):
  • https://${SITE_URL}/views/preservation/show_registered_events?page=${page}&_format=json

• Audit preserved content

  • leaf-bagger-audit.py

What is CWRC

CWRC (Canadian Writing Research Collaboratory) is an “online infrastructure for literary research in and about Canada designed to meet the challenges and embrace the opportunities of the digital turn.” In other words, CWRC is a living repository (i.e., contains content that may be updated, for example, as facts and assertions about a person, place, or event are discovered or changed). CWRC, as of August 2023, contains ~410,000 objects accounting for 1TB+ in storage. The content comes from multiple research projects created by researchers located in many areas of Canada.

CWRC infrastructure is hosted with the Digital Research Alliance of Canada on the Arbutus Cloud hosted at the University of Victoria with data backups hosted in London ON and preservation with UofA Library (via OLRC).

Requirements

The Dockerfile in this repository describes the requirements and setup. An overview of requirements includes:

• Assumes a Drupal instance with the following required Drupal Views REST endpoints
  • views/preservation_show_node_timestamps?page={page}&changed={date_filter}
  • views/preservation_show_media_timestamps?page={page}&changed={date_filter}
• An instance of Islandora-Bagger CLI via isle-bagger or an equivalent setup
  • directory of to store generated AIPs (long-term for auditing)
  • jwtonlogin (to add the user's JWT token to the response of a successful login)
  • See isle-bagger for details
• A shell with OpenStack Swift environment variables
• PHP 8.3+ for Islandora-Bagger
• Python 3.10
• leaf-isle-bagger/requirements.txt`

Workflow: Archival Package Creation

The preservation workflow acts on a polling model where a script runs at regular intervals asking the repository for a list of new/changed items within a given window of time. Any new/changed item has an archival information package generated (AIP) and added to the preservation endpoint.

leaf-bagger.py

• REST request to the site for a list of nodes filtered by change date (or media change date, i.e., if the attached Drupal Media changes the Drupal Node change date will not change)
• For each item in the list, build an archival information package (AIP) via the islandora-bagger CLI tool and store the generated AIP locally
  • More information on the tool:
  • islandora-bagger
  • isle-bagger
• Upload AIP to an OpenStack Swift instance
• Run a post-upload validation

Result: a report of items added to the preservation endpoint.

How to recover from isolated failures

ToDo: what if a small percentage of items in a preservation run fail?

• option 1.: enhancement: script CLI argument specifying a list of item IDs to preserve thus overriding the REST request to the repository
  • May 2024: force_single_node option added
• option 2.: enhancement: only generate AIP and upload if the preservation endpoint is missing the item or is stale/outdated
• option 3.: run the islandora-bagger CLI outside the scripting (downside: loss of the report)
  • from within the container: cd ${BAGGER_APP_DIR} && ./bin/console app:islandora_bagger:create_bag -vvv --settings=var/sample_per_bag_config.yaml --node=${drupal_node_id}

Workflow: Archival Package Audit

The preservation workflow includes an audit step checking, in a basic way, that what exists in the repository is preserved in the preservation endpoint. This step assumes the AIP creation script will fail in unexpected ways and tries to act as a second set of eyes to identify and report failures.

leaf-bagger-audit.py

• REST request to the site for a list of all nodes (and their change date, either from the node or attached media whichever is more recent, i.e., if the attached Drupal Media changes the Drupal Node change date will not change)
• Test against the locally stored AIP for presence and timestamp
• Test against the Swift collection for presence, timestamps, and checksums (locally stored AIP and Swift objects)

Result: an audit report indicating the status of all nodes in the repository and their preservation status in a CSV file.

Tests & linting

The Nox Python automation tooling helps automate testing and linting. The tool is integrated as part of the CI/CD. The noxfile.py contains the configuration.

Install as per your OS, e.g., apt install nox

To run tests and linting:

nox

To run only tests

nox -s test

To run only linting

nox -s lint

To run tests outside nox

• setup virtual environment

python3 -m venv ./rootfs/leaf-isle-bagger/venv
./rootfs/leaf-isle-bagger/venv/bin/python3 -m pip install -r rootfs/leaf-isle-bagger/requirements.txt
./rootfs/leaf-isle-bagger/venv/bin/python3 -m pip install -r rootfs/leaf-isle-bagger/requirements_test.txt

• run tests within the virtual environment

./rootfs/leaf-isle-bagger/venv/bin/pytest rootfs/leaf-isle-bagger/tests/

CI/CD

• Run tests and code linting on each code push
• Build an OCI container to host the preservation workflow tooling and automate execution
• Future: automate deployment

Deployment

The scripts are meant to be executed within a containerized environment. For alternate approaches, review the Dockerfile layers for installation and docker-compose.yml for environment variable settings.

How to run from within a container

• docker compose exec bagger with-contenv bash
• su -s /bin/bash nginx -c "./venv/bin/python3 leaf-bagger.py --server ${BAGGER_DRUPAL_URL} --output /tmp/z.csv --force_single_node 1 --container cwrc-test"
• see rootfs/etc/s6-overlay/scripts/bagger-setup.sh for an example of both leaf-bagger.py and leaf-bagger-audit.py

Dependencies

The OCI container image is based on the isle-bagger image and isle-buildkit. Access to a Drupal site is also required with the container running within a [leaf-base-i8] container deployment or independently (i.e., in a separate deployment).

Settings

Local settings: see isle-bagger and parent containers for more settings (e.g., islandora-bagger tool settings). .env.sample contains a sample .env for docker-compose

Two docker-compose secrets are also used

Docker-compose env vars

Updates

How to update the base?

For an isle-buildkit update (gist, follow dependencies in the Dockerfile layers)

• update isle-bagger (Dockerfile or bake file) with the new isle-buildkit TAG
  • create a new tag release
• update leaf-isle-bagger (Dockerfile or bake file) with the isle-bagger TAG
  • create a new tag release
• update the new leaf-isle-bagger OCI image TAG in the production deployment

Note: if wanting to test leaf-isle-bagger and isle-bagger locally

• update the isle-bagger .env with the ISLE_TAG and docker compose build
  • this creates a locally stored image with the TAG set in the .env
• docker compose build leaf-isle-bagger using the same TAG for isle-bagger as in the previous step
  • this creates a locally stored image with the TAG set in the .env
• docker compose up -d to run the container
• docker compose exec bagger with-contenv bash to shell into the container

See the following as an alternative to specifying an OCI image registry and tag in the Dockerfile: https://docs.docker.com/build/bake/reference/. As an example, see isle-buildkit docker-bake.hcl.