Unify is a tool for data harmonization. In simple, technical terms, it automates batch imports into Datomic and an object storage such as S3. Unify is built on foundational tooling derived from CANDEL, extracted and made more generic.
Unify's data harmonization tools are schema agnostic. That is, they can work on schemas for many different domains, provided some annotations are added to the schema as per Unify's metamodel.
Unify is under heavy development and should be considered early alpha.
Unify uses schema annotation in a Datomic database, called the "metamodel", to infer the details necessary for automating a batch import process. The combined schema and metamodel can defined in a Unify specified DSL, or it can be transacted more verbosely as raw, transactable Datomic edn.
Once the schema annotations are in place, an import can be defined declaratively in a config.edn file that specifies how to map data from files into the schema. Datomic handles reference resolution by resolving IDs. These can be defined a number of ways: as unique globally, as only unique in the context of a tree of data relationships, or as composite IDs composed of other fields.
End-to-end examples and tutorials are coming which will provide a more detailed
walkthrough of how to use Unify. Example data is available in this repo in the
test/resources/systems/ subdirectory.
For running the Unify CLI, install the latest version of Clojure.
Unify provides a simple local system setup for test use that only requires
docker-compose. You can inspect the local system and change the configuration,
or use it as a template. It can be found in the repo at ops/local/. To
get going with minimal fuss, with docker-compose
installed, run the following from
the project root:
util/start-local-system
You can terminate the system with:
util/stop-local-system
After standing up a local system, you can run:
./unify-local
with no arguments for command line help.
To stand up an example db with the CANDEL schema, you can run the following commands with these values set:
$DATABASE_NAME # a string, the name of the database
$SCHEMA_DIR # a directory containing a unify Schema
$WORKING_DIR # user defined output path, ./unify prepare will create this
$IMPORT_CONFIG_PATH # path to a config edn file describing an import
For this example, which you can run with this repo, they are set to:
$DATABASE_NAME=unify-example
$SCHEMA_DIR=test/resources/systems/candel/template-dataset/schema
$WORKING_DIR=~/scratch/candel-import-prepared
$IMPORT_CONFIG_PATH=test/resources/systems/candel/template-dataset/config.edn
The steps are (1) request a db.
./unify-local request-db --database unify-example --schema-directory test/resources/systems/candel/template-dataset/schema
Then (2) prepare the import. This reads the files linked in the config.edn file and turns them into Datomic transaction data as per the import specification in said file.
./unify-local prepare --working-directory $WORKING_DIR \
--schema-directory $SCHEMA_DIR \
--import-config $IMPORT_CONFIG_PATH
Once the data has been prepared, then for step (3) it can be transacted:
./unify-local transact --working-directory $WORKING_DIR \
--database $DATABASE_NAME
You can use a graphic browser to explore the schema in your example database at:
http://localhost:8899/schema/1.3.1/index.html
You can also query it by connecting with it from a normal Datomic peer process, or from any other language using the JSON query service that the local ops stand up with root URL of:
http://localhost:8889/query/$DB_NAME
There's an example Python query that can be run in isolation on the example CANDEL database with:
pip install requests # if needed
python ops/local-test-query.py
You can inspect this file for an example of working with the query service.
Stay tuned for more!
Apache 2.0, Copyright Vendekagon Labs, LLC.
Development on Unify is currently being supported by the Rare Cancer Research Foundation and Clojurists Together.