A skeleton for creating an OpenRefine reconciliation service via a docker container (originally based on OpenRefine/reconciliation_service_skeleton and also drawing inspiration from OpenRefine Style Reconciliation Containers

The services uses a Flask app running by default on port 5000.

To run the container using a CSV based reconciliation file `/path/to/myfile.csv', with column name Searchval for the column to fuzzily match against and Idval as the id value:

docker run --name reconpy -p 5001:5000 -d -v /path/to:/tmp/import -e RECONFILE=myfile.csv -e SEARCHCOL=Searchval -e IDCOL=Idval psychemedia/recon-py

For example:

docker run --name reconpy -p 5001:5000 -d -v pwd:/tmp/import -e RECONFILE=UK-MPs.csv -e SEARCHCOL=DisplayAs -e IDCOL=Member_Id psychemedia/recon-py

and try to reconcile on a name such as Diana Abbot.

The docker-compose file creates a network between three containers:

• OpenRefine
• Jupyter notebook
• reconciler

A demo notebook can be found in the running notebook server work directory. It shows how to connect (ports inside the docker-compose network are the internal ports, not port mapped ports).

The Jupyter notebook can see the reconcile service but OpenRefine appears not to when I try to add it as a reconciliation service.

See OpenRefine/reconciliation_service_skeleton for additional ideas about how to further use the service, including constructing your own reconciliation service containers.