Getting Neo4j cypher queries to work in java GraphRepository

[MichaelJaniak]

I'm using a Neo4j database with various node classes and would like to search through the parameter strings of one particular node class. For example, if I have three nodes in class.MyNode, one with String name = "foo", another with name = "bar" and a third with name = "car", if I search for "ar", I would like to return a list containing nodes "bar" and "car".

The following cypher query works perfectly well in Neo4j's data browser :

START n=node:__types__(className='com.myapp.MyNode')
WHERE n.name=~ /(?i).*xyz.*/
RETURN n

However, when I add it to my repository, it returns all the nodes of class.MyNode and not the subset I need. Here is the code I've been using for the repository:

public interface MyNodeRepository extends GraphRepository<MyNode>,
	RelationshipOperationsRepository<MyNode>,
	NamedIndexRepository<MyNode>{
	
	@Query(	"START n=node:__types__(className='com.myapp.MyNode')"+ 
			"WHERE n.name=~ /(?i).*{0}.*/"+
			"RETURN n")
        Set<MyNode> getNodes(String search);	
}

And here is the node:

@NodeEntity
public class MyNode {
	@Indexed(indexType=IndexType.FULLTEXT, indexName="name") 
	@Fetch private String name;

        etc
}

I'm passing parameters just like the bible suggests and I've been pulling my hair out trying to figure out why this doesn't work.

Any thoughts on how to fix this would be much appreciated!!!

[lassewesth]

Michael,

I'd suggest taking advantage of SDN's auto-repo-foo: http://static.springsource.org/sprin.../html/#d5e1308

@NodeEntity
class MyNode {
    @GraphId
    private Long id;

    @Indexed(indexType = IndexType.FULLTEXT, indexName = "name")
    private String name;

    MyNode() {
    }

    public MyNode(String name) {
        this.name = name;
    }
}

interface MyNodeRepository extends GraphRepository<MyNode> {
    Page<MyNode> findByNameLike(String name, Pageable page);
}

@RunWith(SpringJUnit4ClassRunner.class)
@ContextConfiguration
public class FullTextSearchTests {
    @Autowired
    private MyNodeRepository myNodeRepository;

    @Configuration
    @EnableNeo4jRepositories
    static class Config extends Neo4jConfiguration {
        @Bean
        public GraphDatabaseService graphDatabaseService() {
            return new ImpermanentGraphDatabase();
        }
    }

    @Test
    public void shouldFindBySubstring() throws Exception {
        myNodeRepository.save(new MyNode("foo"));
        myNodeRepository.save(new MyNode("bar"));
        myNodeRepository.save(new MyNode("car"));

        assertThat(myNodeRepository.findByNameLike("*ar*", new PageRequest(0, 20)).getNumberOfElements(), is(2));
        assertThat(myNodeRepository.findByNameLike("*o", new PageRequest(0, 20)).getNumberOfElements(), is(1));
        assertThat(myNodeRepository.findByNameLike("ca?", new PageRequest(0, 20)).getNumberOfElements(), is(1));
    }
}

[MichaelJaniak]

Thank you for the prompt reply! That is working for me now.

[MichaelJaniak]

Sorry but I have to get back to this problem. Although the auto-repo is good for simple queries, I still need to use the regular expression for more complicated queries.

For example:

public interface MyRepository extends GraphRepository<User>,
	RelationshipOperationsRepository<User>,
	NamedIndexRepository<User>{
	
	@Query(	"START user=node({0}) "+
	"MATCH user-[:FRIENDS*1..2]-friends "+
	"WITH user, friends "+
	"MATCH friends-[:REL*1]->relatives "+
	"WHERE relatives.name=~ /(?i).*{1}.*/ "+
	"RETURN relatives")	
	Set<User> findRelativesOfFriends(long userId, String search);
}

This works in the data browser, it returns a subset of relatives, but in java regardless of what search parameter I pass, it always returns the whole set.

Am I doing something wrong?

[lassewesth]

public interface MyRepository extends GraphRepository<User>,
	RelationshipOperationsRepository<User>,
	NamedIndexRepository<User>{
	
	@Query(	"START user=node({0}) "+
	"MATCH user-[:FRIENDS*1..2]-friends "+
	"WITH user, friends "+
	"MATCH friends-[:REL*1]->relatives "+
	"WHERE relatives.name=~ /(?i).*{1}.*/ "+
	"RETURN relatives")	
	Set<User> findRelativesOfFriends(long userId, String search);
}

This works in the data browser, it returns a subset of relatives, but in java regardless of what search parameter I pass, it always returns the whole set.

Well, it actually does what you ask it; so it interprets that regex as

match .* one time, then match .* a bunch of times

- the {1} is not substituted, it is taken literally

Instead, do

@Query(	"START user=node({0}) "+
	"MATCH user-[:FRIENDS*1..2]-friends "+
	"WITH user, friends "+
	"MATCH friends-[:REL*1]->relatives "+
	"WHERE relatives.name=~ {1} "+
	"RETURN relatives")	
	Set<User> findRelativesOfFriends(long userId, String regex);

...

myRepo.findRelativesOfFriends(42, "(?i).*my search string.*");

Let me know if it works.

Regards,

Lasse

[MichaelJaniak]

That's perfect, works exactly like that, thank you for your help!