Metadata Support for the OAR Open Data Platform (via NERDm)

This package provides support for metadata across the different components of the OAR Open Data Platform. At the core of this support is the NERD model; this package includes its JSON Schema and Linked Data definitions, validators, translators, and scripts for loading. The package also provides support for ancillary metadata used by the RMM and the SDP, including the NIST theme taxonomy and NERDm field documentation.

Contents

docker/    --> Docker containers for running tests
java/      --> Java source code supporting NERDm metadata
jq/        --> Conversion libraries for the jq, used for translating POD to NERDm
model/     --> Directory contain JSON Schemas, Context maps, and field 
                 documentation supporting the NERDm metadata framework
 examples/ --> Sample NERDm instance documents
python/    --> Python source code supporting NERDm metadata
scripts/   --> Tools for creating and loading metadata, installing this package,
                 and running all tests

Prerequisites

Record translation, validation, and loading require a few third-party tools:

• Python 2.7.X
• Python library: jsonschema 2.5.x or later
• Python library: jsonspec 0.9.16
• Python library: requests
• jq (build with libonig2)
• Python library: pymongo 3.4.X

Pymongo is used to load metadata into a MongoDB database; thus, loading also requires a running instance of MongoDB. Loading unit tests require that the environment variable MONGO_TESTDB_URL be set to an accessible MongoDB database; it's value has the form, 'mongodb://HOST[:PORT]/TESTDB'.

As an alternative to explicitly installing the prerequisites to run the tests, the docker directory contains scripts for building a Docker container with these installed. Running the docker/run.sh script will build the containers (caching them locally), start the container, and put the user in a bash shell in the container. From there, one can run the tests or use the jq and validate tools to interact with metadata files.

Converting a POD record

If relying on the Docker container for the prerequisite tools (see above), start the container via the run.sh to start a bash shell. Inside this shell, the curent directory will be oar-pdr/metadata (i.e. the directory that contains this README).

To convert a single POD format Dataset document into a NERDm Resource document, run the jq command with the following pattern:

jq -L jq --arg id ID -f jq/podds2nerdres.jq POD-FILE > NERDM-FIlE

where POD-FILE is the input POD Dataset filename, NERDM-FILE is the output NERDm Resource document filename, and ID is the identifier to assign to the output record. The jq/tests/data a sample POD Dataset document, janaf_pod.json; to convert it, then, to NERDm, type:

jq -L jq --arg id ark:ID -f jq/podds2nerdres.jq jq/tests/data/janaf_pod.json > janaf_nerdm.json

The test data directory also contains a copy of the NIST PDL Catalog; it can be converted to an array of NERDm Resource records with the following:

jq -L jq --arg id ark:ID -f jq/podcat2nerdres.jq jq/tests/data/nist-pdl-oct2016.json > nist-resources.json

Converting a POD Catalog

An entire POD Catalog document can be converted to a set of NERDm Resource files (i.e. each output file containing one Resource record) using the pdl2resource.py script. Here's an example running the script on the example PDL file that is in the jq/tests/data directory:

scripts/pdl2resources.py -d tmp jq/tests/data/nist-pdl-oct2016.json

The -d option sets the directory where the output files are stored. Other command-line options allow one to convert only a portion of the datasets found in the file; run with the --help option to see the details.

With this script each output resource docuemnt is assigned an ARK identifier.

Validating a NERDm record

This module includes a schema documents that can be used to validate NERDm record. The validate command accomplishes this. For example, to validate the janaf_nerdm.json file, type:

validate -L model janaf_nerdm.json

Loading data into MongoDB

NERDm resource records, like those created by pdl2resources.py, can be loaded into a MongoDB database via the script, ingest-nerdm-res.py. To load the records created from the the above example running pdl2resources.py, where the NERDm records were written to a directory called tmp, simply type:

scripts/ingest-nerdm-res.py -M mongodb://localhost/testdb tmp

This assumes there is a MongoDB instance running on the local machine where the user has write access to the testdb database. The script will validate each NERDm file against the NERDm schemas before loading it into the database.

Note that NERDm resource records must have unique identifier (stored in its @id property. Thus loading a new record will overwrite any previous record with the same identifier.

Loading NERDm field documentation

This package also provides support for loading NERDm field documentation into the database as well, via the ingest-field-info.py script. This documentation is used by the Science Data Portal (or any other client) to get information about available NERDm properties that are available and searchable.

This data is stored in a JSON format, and a default version exists as model/nerdm-fields-help.json. To load this data into the database, simply type:

  scripts/ingest-field-info.py -M mongodb://localhost/testdb model/nerdm-fields-help.json

Each field record has a unique name, so loading a new record will overwrite any previous record with the same name.

Running Tests

To run all the tests associated with this metadata component (assuming all prerequisites are installed), type:

  scripts/testall.py

If prerequisites are not installed (e.g. they're not easily installable), you can test the metadat component via docker as described above. To run the tests within docker, type:

  docker/run.sh testall