Awesome

A Graph Database for Julia, built on top of SQLite.jl.

<br><br>

Definitions

SQLiteGraph.jl uses the Property Graph Model of the Cypher Query Language (PDF).

• A Node describes a discrete object in a domain.
• Nodes can have 0+ labels that classify what kind of node they are.
• An Edge describes a directional relationship between nodes.
• An edge must have a type that classifies the relationship.
• Both edges and nodes can have additional key-value properties that provide further information.

<br><br>

Edges and Nodes

• Nodes and Edges have a simple representation:

struct Node
    id::Int
    labels::Vector{String}
    props::EasyConfig.Config
end

struct Edge
    source::Int
    target::Int
    type::String
    props::EasyConfig.Config
end

• With simple constructors:

Node(id, labels...; props...)

Edge(source_id, target_id, type; props...)

<br><br>

Adding Elements to the Graph

using SQLiteGraph

db = DB()

insert!(db, Node(1, "Person", "Actor"; name="Tom Hanks"))

insert!(db, Node(2, "Movie"; title="Forest Gump"))

insert!(db, Edge(1, 2, "Acts In"; awards=["Best Actor in a Leading Role"]))

<br><br>

Editing Elements

insert! will not replace an existing node or edge. Instead, use replace!.

replace!(db, Node(2, "Movie"; title="Forest Gump", genre="Drama"))

<br><br>

Simple Queries

• Use getindex to access elements.
• If : is used as an index, an iterator is returned.
• If s::String is used as an index, an iterator of nodes where s is one of the labels is returned.

db[1]  # Node(1, "Person", "Actor"; name="Tom Hanks")

for node in db[:]
    println(node)
end

only(db["Movie"])  # Node(2, "Movie"; title="Forest Gump", genre="Drama")

# (Pretend the graph is populated with many more items.  The following return iterators.)

db[1, :, "Acts In"]  # All movies that Tom Hanks acts in

db[:, 2, "Acts In"]  # All actors in "Forest Gump"

db[1, 2, :]  # All relationships between "Tom Hanks" and "Forest Gump"

db[:, :, :]  # All edges

<br><br>

✨ Attribution ✨

SQLiteGraph is STRONGLY influenced by https://github.com/dpapathanasiou/simple-graph.

<br><br>

Under the Hood Details

• Nodes and edges are saved in the nodes and edges tables, respectively.
• nodes
  • id (INTEGER): unique identifier of a node
  • labels (TEXT): stored as ;-delimited (thus ; cannot be used in a label)
  • props (TEXT): stored as JSON3.write(props)
• edges
  • source (INTEGER): id of "from" node (nodes(id) is a foreign key)
  • target (INTEGER): id of "to" node (nodes(id) is a foreign key)
  • type (TEXT): the "class" of the edge/relationship
  • props (TEXT)