The Data Model
To store information about our users and their activities on or our social blogging application, let’s think about how we’ll want to model this data. For this example, we’ll have :User, :Post, and :Tag labels for our nodes. Conceptually, we want to capture:
- users publishing posts, or
(:User)-[:PUBLISHED]->(:Post) - tags tagging posts, or
(:Tag)-[:TAGGED]->(:Post) - users liking posts, or
(:User)-[:LIKED]->(:Post)
Before beginning, we’ll want to create uniqueness constraints for the appopriate label and property pairs. This will also create indexes on these label and property pairs. We don’t want any duplicate :User, :Post, or :Tag nodes, so blog/__init__.py contains the following and is run when the app is started:
from .views import app
from .models import graph
def create_uniqueness_constraint(label, property):
query = "CREATE CONSTRAINT ON (n:{label}) ASSERT n.{property} IS UNIQUE"
query = query.format(label=label, property=property)
graph.cypher.execute(query)
create_uniqueness_constraint("User", "username")
create_uniqueness_constraint("Tag", "name")
create_uniqueness_constraint("Post", "id")
With a uniqueness constraint on the :User label by the username property, Neo4j ensures there are never :User labeled nodes that share the same username. Any transactions that violate this constraint will be rejected. The same applies to the uniquess constraints on :Tag by name and :Post by id.
Next, we’ll go through each view defined in views.py and discuss the corresponding logic in models.py.