ArangoDB v3.5 reached End of Life (EOL) and is no longer supported.

This documentation is outdated. Please see the most recent version here: Latest Docs

Import Graph data with Stream Transactions

Importing graph data is possible using bulk import functions for every vertex and edge collection of the graph. Nevertheless sometimes one could have the need to preserve certain domain specific invariants (consistency) while inserting the data. For example while populating a movie database, it would be desirable to insert together with each movie, all its actors and the related outgoing edges reaching them.

This can be achieved using a Stream Transaction which inserts data into all the related vertex and edge collections using batch insert. On the other side this would not be possible using bulk import functions, since they do not support Stream Transactions.

Example for every batch of movies:

List<Movie> movies = ...;           // movies of this batch
List<Actor> actors = ...;           // actors who acted in such movies 
List<HasActor> hasActorEdges = ...; // movie --> actor edges

StreamTransactionEntity tx = database.beginStreamTransaction(new StreamTransactionOptions()
                .writeCollections(MOVIES, ACTORS, HAS_ACTOR));

DocumentCreateOptions options = new DocumentCreateOptions().streamTransactionId(tx.getId());

try {
    database.collection(MOVIES).insertDocuments(c1Vertices, options);
    database.collection(ACTORS).insertDocuments(toVertices, options);
    database.collection(HAS_ACTOR).insertDocuments(edges, options.overwrite(true));
    database.commitStreamTransaction(tx.getId());
} catch (Exception e) {
    database.abortStreamTransaction(tx.getId());
    // handle e
}

If the batch is too large, you may incur intermediate commits (see transactions limitations).