Working with Graphs using Neo4j and Python: Cypher + Graphene + Neomodel = <3
Everything is ready to run with a single command after repository cloning. Thanks to docker-compose, we can spin up Jupyter Lab and Neo4j services in no time.
git clone https://github.com/Proteusiq/graphs.git
cd graphs
docker-compose upTo start only Neo4J service execute: docker-compose up neo4j
To shutdown, docker-compose down. We can add -v at the end of the command to remove also volumes, the datasets.
Docker-compose will log a link and token to Jupyter Lab. You can also get the link and token with docker container logs jupyter_lab. To connect to Neo4j, head to localhost:7474 and login in Neo4j browser. The connection url is bolt://localhost:7687. Login username and password is disabled for development, you can set them back on docker-compose file, default username:neo4j and password:test.
Hurrah, you made it. Try out the test.ipynb to get the joy of Python and Neo4j.
The docker-compose comes with default settings that can be overide. Example, the Jupyter Lab user is danpra. If you want to set your own user, edit the docker-compose.yml
Connection url in Jupyter Lab is bolt://neo4j:test@neo4j:7687 as neo4j is the name of our service in docker-compose.yml.
Cypher is the Python of query languages. It is beautiful, intuitive and simple. To show you it power, let's create a simple dataset and start asking it, yes our dataset, questions.
Create Graph Data Query
// Creating the graph: N.B: Enable multi statement query editor on your settings
//Create Person and Location
//Neo4j 3.X
MERGE (:Person {name:"Prayson",age:34, description:"The Matrix's Neo"})
MERGE (:Person {name:"Lea", age:32, description:"The awesome"})
MERGE (:Person {name:"Eloise",age:8, description:"The dancer"})
MERGE (:Person {name:"Nora",age:5, description:"The wise"})
MERGE (:Person {name:"Mario",age:3, description:"The Jumper"})
MERGE (:Location {address:"Gurrevej 48"});
// Neo4j 4.x
MERGE (:Person {name:"Prayson",age:34, description:"The Matrix's Neo"}),
(:Person {name:"Lea", age:32, description:"The awesome"}),
(:Person {name:"Eloise",age:8, description:"The dancer"}),
(:Person {name:"Nora",age:5, description:"The wise"}),
(:Person {name:"Mario",age:3, description:"The Jumper"}),
(:Location {address:"Gurrevej 48"});
//Add p person relationship to location l
MATCH (person:Person),(location:Location {address:"Gurrevej 48"})
WHERE person.name in ['Prayson','Mario', 'Nora','Eloise', 'Lea']
MERGE (person)-[:LIVES_IN]->(location);
// Add marriage
MATCH (p:Person {name:"Lea"}),(o:Person {name:"Prayson"})
MERGE (p)<-[:MARRIED_TO]-(o)<-[:MARRIED_TO]-(p);
// Add childrens
MATCH (p:Person), (o:Person)
WHERE p.name = "Prayson" and o.name IN ["Eloise","Nora","Mario"]
MERGE (p)<-[:CHILD_OF]-(o);
MATCH (p:Person), (o:Person)
WHERE p.name ="Lea" and o.name IN ["Eloise","Nora","Mario"]
MERGE (p)<-[:CHILD_OF]-(o);Show graph:
MATCH (n) RETURN n LIMIT 6
Who is Prayson?:
Query
MATCH (p:Person)
WHERE p.name = "Prayson"
RETURN p.description AS `Who is Prayson?`
// OR
MATCH (p:Person {name: "Prayson"})
RETURN p.description AS `Who is Prayson?`
What are the names and age of children leaving in Gurrevej 48 under 18 years old?:
Query
MATCH (under_age:Person),(location:Location)
WHERE under_age.age < 18 and location.address="Gurrevej 48"
RETURN under_age.name AS `Children`, under_age.age AS `Age`
Who lives in Gurrevej 48?:
Query
MATCH (p:Person)-[:LIVES_IN]->(l:Location)
WHERE l.address = "Gurrevej 48"
RETURN collect(p.name) AS `Who lives in Gurrevej 48?`
How many children does Prayson have?:
Query
MATCH (child:Person)-[:CHILD_OF]->(parent:Person)
WHERE parent.name = "Prayson"
RETURN count(child) AS `How many children does Prayson have?`
So far, we could have performed the queries above with SQL in tabular data. Let's gear up. We know that in real world, relationship are more richer and wider. Let's introduce more relationships.
Getting Large
//Add more members: My dad, brothers, nephew, and niece
//Neo4j 3.x
MERGE (:Person {name:"Wilfred", age:59, description:"Babu"})
MERGE (:Person {name:"Eric", age:35, description: "First blood"})
MERGE (:Person {name:"Jimmy", age:29, description: "Quantum MD"})
MERGE (:Person {name:"Trace", age:10})
MERGE (:Person {name:"Trisher", age:3});
//Neo4j 4.x
MERGE (:Person {name:"Wilfred", age:59, description:"Babu"}),
(:Person {name:"Eric", age:35, description: "First blood"}),
(:Person {name:"Jimmy", age:29, description: "Quantum MD"}),
(:Person {name:"Trace", age:10}),
(:Person {name:"Trisher", age:3});
//Add relationship
MATCH (parent:Person), (child:Person)
WHERE parent.name ="Wilfred" and child.name IN ["Eric","Prayson","Jimmy"]
MERGE (parent)<-[:CHILD_OF]-(child)
MATCH (parent:Person), (child:Person)
WHERE parent.name ="Eric" and child.name IN ["Trace","Trisher"]
MERGE (parent)<-[:CHILD_OF]-(child)
//Did not include my mother to keep things simple
The power of Cypher comes in query relationships. Imagine we wanted to know the names of Wilfred grandchildren. In SQL, you will have to perform a number of table joins to achieve these. The farther the link, the more joins you will have. With Cypher, two lines of code is all you need.
Two lines of Code
MATCH (grand_child:Person)-[:CHILD_OF*2]->(babu:Person {name: "Wilfred"})
RETURN collect(grand_child.name) AS `Wilfred's grand children are:`
There is more. We can ask: What is the shortest child-of path relationship between Mario and Jimmy?
Query
MATCH (mario:Person { name: 'Mario' }),(jimmy:Person { name: 'Jimmy' }),
path = shortestPath((mario)-[:CHILD_OF*]-(jimmy))
RETURN path
Remember: You need to rebuild the services with docker-compose up --build for any new changes in requirements.txt, yml or Dockerfile to take effect.
| NOTE | If using Neo4j 4.x, mount neomodel to notebook as the current neomodel does not support Neo4J 4.X. Until fixed, this is a temporary hack-solution. |
|---|
- [ ] Create graphene examples
- [ ] Add FastAPI to server GraphQL with graphene
- [ ] Add a second docker-compose with FastAPI image
I have pinned Neo4j to version 3.5.14 and downloaded Apoc 3.5 and Algo 3.5 See Apoc Github and Algo Github for more information.