Multiple MATCH clauses?

[Johnny Weng Luu]

It seems not possible to have more than one MATCH clause.

            START ref = node(*)

            MATCH ref-[:HAS_DEPARTMENTS]->departments<-[:IS_DEPARTMENT]-department

            MATCH ref-[:HAS_CATEGORIES]->categories

            WHERE ref.name = "App" and department.name = "Multimedia"

            CREATE category = {name: "Keyboard"}

            CREATE department-[:HAS_CATEGORY]->category

            CREATE categories<-[:IS_CATEGORY]-category

However, I need the second MATCH clause to fetch "categories" for the last CREATE clause.

Is there a way around this?

Johnny

[Johnny Weng Luu]

This worked pretty great!

            START ref = node(*)

            MATCH ref-[:HAS_DEPARTMENTS]->departments<-[:IS_DEPARTMENT]-department

            WHERE ref.name = "App" and department.name = "Multimedia"

            WITH ref, department

            MATCH ref-[:HAS_CATEGORIES]->categories

            CREATE category = {name: "Keyboard"}

            CREATE department-[:HAS_CATEGORY]->category

            CREATE categories<-[:IS_CATEGORY]-category

Worked great when I added a WITH clause.

1. Is this the preferred way?

2. In my query you can see that I am fetch all nodes (ref = node(*)) and getting the reference node with ref.name = "App". Is this the recommended way to get a specific node since you cannot pick one from the START clause directly?

[Alistair Jones]

Hi Johnny,

The MATCH clause can accept multiple patterns, separated by commas.  So you could write:

MATCH ref-[:HAS_DEPARTMENTS]->departments<-[:IS_DEPARTMENT]-department, ref-[:HAS_CATEGORIES]->categories

When you use the WITH clause, you're effectively starting a new chained query, passing the ref and department variables along.

On 20 May 2012 00:06, Johnny Weng Luu <johnny....@gmail.com> wrote:

1. Is this the preferred way?

I think it would be preferable to just use a single MATCH clause, with multiple patterns. Only use WITH if you can't achieve what you want with a simpler query.

2. In my query you can see that I am fetch all nodes (ref = node(*)) and getting the reference node with ref.name = "App". Is this the recommended way to get a specific node since you cannot pick one from the START clause directly?

I don't know your full graph, but it doesn't look like you need to start from node(*) (all nodes in the graph).  Can you share some more details about how you are modelling your domain?  What does "ref" represent?

cheers,

-Alistair

[Michael Hunger]

yes this is the preferred way, WITH allows you to chain queries.

each of the queries can be a read-or a modifying query.

with two read-queries you can create constructs like having.

#2 

- you would normally index start-nodes with the properties that you want to look up, then you can query them directly in the start-clause

- other ways would be category nodes that e.g. an apps-category linked to your "app-nodes" from which you then iterate the attached nodes and find the app-node

- if you need this specific app-node very often it might be sensible to store it in your app and use it directly in app=node({appNode})

Please try to use parameters {param} wherever possible so that you can enjoy cypher's query caching and you create reusable queries anyway.

Michael