Numeric Indexes and the Neo4J REST Server

Suppose your neo database contains suppliers for custom t-shirt printing. Some suppliers have a minimum order quantity, and you want to quickly lookup suppliers that would accept a given quantity. This is easy with the Embedded Neo4J server in Java:

1
2
3
4
5
6
7
8
9
10
11
12
13

  // Add to an index
  Index<Node> suppliers = graphDb.index().forNodes("suppliers");
  suppliers.add(someSupplier, "minimum-order", new ValueContext(5).indexNumeric());

  // Test query with a match
  QueryContext queryContext = QueryContext.numericRange("minimum-order", null, 8);
  IndexHits<Node> potentialSuppliers = suppliers.query(queryContext);
  assertEquals(1, potentialSuppliers.size());
  assertEquals(someSupplier, potentialSuppliers.getSingle());
  // Test query without a match
  queryContext = QueryContext.numericRange("minimum-order", null, 4);
  potentialSuppliers = suppliers.query(queryContext);
  assertEquals(0, potentialSuppliers.size());

However, those numeric ranges are not supported by Neo4J REST server. It’s probably best to write a plugin for this, but if you cannot use custom plugins (for example, I don’t think the Heroku Neo4J Add-On allows custom plugins) then the Neo4J Gremlin Plugin may be your best choice.

Here’s how you would do the same thing if you’re using Ruby’s Neography:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48

class NeoQuery
  GREMLIN_NUMERIC_INDEX_TEMPLATE = <<eos
    import org.neo4j.index.lucene.*;
    neo4j = g.getRawGraph();
    tx = neo4j.beginTx();
    idxManager = neo4j.index();
    cuIndex = idxManager.forNodes(index_name);
    node = neo4j.getNodeById(Long.parseLong(node_id));
    cuIndex.add(node, key_name, new ValueContext(value).indexNumeric());
    tx.success();
    tx.finish();
eos

  GREMLIN_NUMERIC_QUERY_TEMPLATE = <<eos
    import org.neo4j.index.lucene.*;
    neo4j = g.getRawGraph();
    idxManager = neo4j.index();
    cuIndex = idxManager.forNodes(index_name);
    cuIndex.query(QueryContext.numericRange(key_name, null, value, false, false));
eos

  @@neo = Neography::Rest.new(ENV["NEO4J_URL"] || "http://localhost:7474")

  def self.index_numeric(index_name, node, key_name, value)
    @@neo.execute_script(GREMLIN_NUMERIC_INDEX_TEMPLATE, {
      :index_name => index_name,
      :key_name => key_name,
      :node_id => node,
      :value => value})
  end

  def self.query_numeric(index_name, key_name, value)
    @@neo.execute_script(GREMLIN_NUMERIC_QUERY_TEMPLATE, {
      :index_name => index_name,
      :key_name => key_name,
      :value => value})
  end
end

# Add to index
NeoQuery.index_numeric('suppliers', your_node.neo_id, :minimum_order_size, 5)
# Test query with a match
potential_suppliers = NeoQuery.query_numeric('suppliers', :minimum_order_size, 8)
potential_suppliers.size.should == 1
Neography::Node.load(potential_suppliers[0]).neo_id.should == your_node.neo_id
# Test query without a match
potential_suppliers = NeoQuery.query_numeric('suppliers', :minimum_order_size, 4)
potential_suppliers.size.should == 0

Disclaimer: I have not tested behavior in a clustered environment and you should consider security before using execute_script.

Comments