dataverse-migration-scripts

Scripts for migrating datasets from one Dataverse repository installation to another. These scripts were used to migrate datasets from version 4.20 (University of Alberta) to version 5.1 (Scholars Portal). For these scripts to work with other versions, updates may be needed. These scripts do not use the migration APIs that were added in version 5.6.

We are using Python 3 and pyDataverse.

DOIs

If the installation you are migrating datasets from is minting DOIs, those DOIs will need to be migrated from the Datacite account of the previous installation to the new one. If the target installation's Datacite account does not own the DOIs for the datasets your are migrating, you will not be able to publish the datasets in the new installation.

config.json

The config file contains information used in the scripts.

• base_url_origin: the base URL of the source installation.
• base_url_target: the base URL of the target installation.
• api_token_origin and api_token_target: the API tokens of superuser accounts of corresponding origin and target installations.
• dataverse_alias: an alias of the target dataverse datasets will be migrated into.
• dr: the directory of files of original dataverse.
• prefix_DOI: the DOI prefix for migrated datasets, for example "doi:10.5072/FK2".
• db_name: the name of the dataverse DB of target installation, for example "dvndb".
• db_user: the username of target DB, for example "dvnapp".
• db_password: the password of db_user.
• db_host: the hostname or ip address of target database.
• db_port: the port of target database.
• builtin_users_key: the BuiltinUsers.KEY from the settings table of the target installation.
• password: the initial password of new built in users that will be created in the target installation.
• directories: an array of all datasets in the original dataverse that you want to migrate, for example ['QBAYAF', 'KBHUA6'] would be entered for datasets with the persistentId's of doi:10.5072/FK2/QBAYAF and doi:10.5072/FK2/KBHUA.

Scripts

Each script for the migration is described below, and are included in the order that they should be run for the migration.

number_of_dt_dv.py

This is an optional script that can be used to fill the "directories" field in config.json. It lists all of the datasets in the root dataverse of the original installation recursively.

versions_files.py

This is main script for creating datasets with all the versions and files in the target installation. DOIs for published datasets will be preserved, and for unpublished datasets new persistent identifiers will be created.

At the end the script also creates correspondence between database ids of datasets from the old installation and the new installation. These ids are in created correspondence_old_new.json file. Correspondence between ids of DataFiles of the old installation and new installation will be saved in all_data_files.json. versions_files.log is also created by this script.

create_dataverses.py

This script creates dataverses in the target dataverse (with dataverse_alias) from the root dataverse of the original installation. It also creates dataverses.json file with all dataverses aliases and old ids of old installation.

create_correspondence.py

This script creates correspondence.json file that contains correspondence between dataverses and datasets that belong to them.

move_datasets.py

This script moves datasets created by versions_files.py to dataverses created by create_dataverses.py. It uses create_correspondence.py.

update_dates_dataverses.py

This script updates dates for migrated dataverses in the dvobject table of target installation. It uses psycopg2-binary that can be installed with pip install psycopg2-binary.

It needs the dataverse_dates.json file that can be generated from the original installation using the following query:

select id, createdate, modificationtime, publicationdate from dvobject where dtype='Dataverse'

...and convert it into JSON format.

update_dates.py

This script updates dates for migrated datasets. It updates the datasetversion table of the target installation. It uses correspondence_old_new.json that was created by versions_files.py.

update_dates_dvobject.py

This script updates dates in the dvobject table for migrated datasets. It uses correspondence_old_new.json that was created by versions_files.py.

update_dates_datafiles.py

This script updates dates for migrated datafiles in dvobject table of target installation. It uses all_data_files.json that was created by versions_files.py and datafile_dates.json that can be generated from original installation using the following query:

select id, createdate, modificationtime, publicationdate from dvobject where dtype='DataFile'

...and convert it into JSON format.

For dates to take effect in the target installation, solr re-indexing in place should be performed:

curl -X DELETE http://localhost:8080/api/admin/index/timestamps
curl http://localhost:8080/api/admin/index/continue

This re-indexing can be done at the end of migration process.

create_users.py

This script creates builtin users in the target installation. This script has customized affiliations set for importing users which can be modified or removed. It uses users.json that contains list of all the users from original installation. users.json can be created from the original installation using:

curl -H "X-Dataverse-key:$API_TOKEN" "$SERVER_URL/api/admin/list-users?itemsPerPage=1000"

explicit_groups.py

This script creates explicit groups and assign them to dataverses. It uses the dataverses.json file that was created by create_dataverses.py.

correspondence_explicit_groups.py

This script creates the correspondence_explicit_groups.json file that contains the correspondence between explicit groups of old and new installations.

role_dataverses.py

This script assigns roles for dataverses. It uses the dataverses.json file that was created by create_dataverses.py, and the correspondence_explicit_groups.json file that was created by correspondence_explicit_groups.py.

role_datasets.py

This script assigns roles for datasets. It uses the dataverses.json file that was created by create_dataverses.py, and the correspondence_explicit_groups.json file that was created by correspondence_explicit_groups.py.

update_filedownload_count.py

Optional: if one needs to preserve filedownload count, one will need contain of a guestbook, filedownloads and filemetadata tables in json format, as dataverse_guestbook.json, dataverse_filedownloads.json, dataverse_filemetadata.json. the script also uses correspondence_old_new.json, all_data_files.json and users.json

change_registration_DOI.py

This script updates DOI registration metadata with Datacite to ensure the updated dates are reflected.

Delete cached files

The last step of migration is deleting all .cached files of the datasets that were migrated in the filesystem. This is done so that the updated dates are reflected in the exported metadata files.