Java Fedora Client

Java client for managing interactions with the PASS data in Fedora. Includes the data model represented as POJOs, annotated with Jackson for easy conversion to JSON.

PASS POJOs

The model is kept up to date with the pass-data-model project, but may change as new model requirements are identified.

PASS Client

The interfaces in pass-client-api can be used to access both Fedora and Elasticsearch

Note: this client does not currently perform any validation such as duplicate checking, or verifying required fields, it assumes these kinds of checks take place outside of the client. It also does not yet support batch transactions, or respond appropriately to various HTTP statuses that come back from Fedora. These will need to be added as needed once the client can be developed fully.

CRUD functions

The CRUD calls for Fedora perform basic read, write, update and delete functions using the model objects in the pass-model module. The Java doc provide guidance on how to use the various functions. For example, to create a record, simply pass a populated model object into the client's createResource function and you will receive a URI that can be used to retrieve the object:

Grant grant = new Grant();
//populate Grant
...
PassClient client = PassClientFactory.getPassClient();
URI uri = client.createResource(grant);

Note that to update an object, it is important to first read it from the database, make your changes to the retrieved object, then pass the same object to the updateResource method. Information is stored within the object that is vital to perform a proper update. In a future iteration, this will throw an exception if the object changed between the read and update. Here is an example:

PassClient client = PassClientFactory.getPassClient();
Grant grant = (Grant) client.readResource(uri);
//modify the grant object.
grant.setAwardNumber("abc123");
client.updateResource(grant);

findBy functions

The findBy functions allow you to look up records by a specific field, for example, searching for Grant by localAwardId might look like this:

String awardNumber = "AB123456";
PassClient client = PassClientFactory.getPassClient();
URI grantUri = client.findByAttribute(Grant.class, "awardNumber", awardNumber);

The Java docs provide more information about this functionality.

Crawling/iterating the repository.

Simple walking of PASS entities is achieved by providing a Consumer<URI>, which is invoked for each matching PASS entity. The API provides a mechanism for crawling all PassEntities (under an implicit base URI, FedoraConfig.getBaseUrl()):

PassClient client = PassClientFactory.getPassClient();

// Process all PASS entities
client.client.processAllEntities(myConsumer);

// Process only submissions
client.client.processAllEntities(myConsumer, Submission.class);

For finer grained control, use org.dataconservancy.pass.client.fedora.RepositoryCrawler. Resources can be ignored (not sent to the consumer, but their children still recursed), or skipped entirely (thus preventing recursion to their children). RepositoryCrawler.Ignore and RepositoryCrawler.Skip classes have useful predicates for skipping or ignoring resources.

For example, to iterate ONLY submission resources (and not binary files that might be children of them), use recursion depth of 1, and ignore the parent submissions/ container, as it is not a Submission:

import static org.dataconservancy.pass.client.fedora.RepositoryCrawler.Ignore.IGNORE_ROOT;
import static org.dataconservancy.pass.client.fedora.RepositoryCrawler.Skip.depth;
...
 
RepositoryCrawler crawler = new RepositoryCrawler();
int numVisited = crawler.visit("http://localhost:8080/fcrepo/rest/submissions/", myConsumer, IGNORE_ROOT, depth(1));

As another example, to iterate ALL pass entities, ignoring ACLs, ignoring parent containers, and ignoring children of PASS entities (like binary files POSTed to submissions), start with the repository root, to a depth of 2, and ignore containers:

import static org.dataconservancy.pass.client.fedora.RepositoryCrawler.Ignore.IGNORE_CONTAINERS;
import static org.dataconservancy.pass.client.fedora.RepositoryCrawler.Skip.SKIP_ACLS;
...

RepositoryCrawler crawler = new RepositoryCrawler();
int numVisited = crawler.visit(URI.create(FedoraConfig.getBaseUrl()), myConsumer, IGNORE_CONTAINERS,
            depth(2).or(SKIP_ACLS));

Configuration

Configuration may be provided via system properties, or environment variables. System properties are case-sensitive and separated by periods, as per Java conventions. Environment variables should be uppercase and separated by underscores, as per OS conventions. For example, the fedora user may be provided by a system property -Dpass.fedora.user=myUser or as an environment variable PASS_FEDORA_USER=myUser. System properties override environment variables.

• pass.fedora.baseurl (default=http://localhost:8080/fcrepo/rest)
• pass.fedora.user (default=fedoraAdmin)
• pass.fedora.password (default=moo)
• pass.elasticsearch.url (defaults = http://localhost:9200)
• pass.elasticsearch.indices (default = pass)
• pass.elasticsearch.limit (defaults = 200) you can also override the default by using the findBy functions that accept a limit and offset value

A note on pass.elasticsearch.indices: a value of "" will cause all indices on the host to be searched, as should a target value of _all or *.

Integration tests with Fedora and Elasticsearch

The integration test module pass-client-integration uses Docker to spin up an instance of Fedora and Elasticsearch for testing the client against.

To run the integration testing environment manually, from within pass-client-integration, do

mvn docker:run -Pstandard

This will run Fedora at standard port (8080) and Elasticsearch at port 9200. This mode is very useful for testing/debugging/developing against the databases from within the IDE. Repository content is stored in target, so if it is run after integration tests, the repository will still retain all data deposited during.