A Property Graph library for python that supports reading and writing property graphs in a textual format, reading an writing via Cypher, and a simple schema language.
- generate documentation for your property graph structure
- load, parse, and manipulate PG Schema documents
- load saved graphs from a YAML format, schema embedded
- generate cypher from saved graphs to load property graphs
You can install the package via:
pip install pypropgraphThe module can be invoked directly and provides a set of basic commands that allow parsing, inspection, cypher statement generation, and loading ontologies.
The invocation is:
python -m propgraph {operation} {file ...}?where operation is one of:
validate- parse and validate the graphcypher- generate cypher create/merge statementsload- load the ontology into a property graph databaseschema.check- check the syntax of a schemaschema.doc- generate Markdown documentation for the schema
If the file is omitted, the command will read from stdin. Otherwise, each file specified will be read and operated on in the order they are specified.
The module currently supports loading ontologies directly into RedisGraph.
The following options can be specified for connecting to the database:
--host {name}|{ip}- the host of the database, defaults to 0.0.0.0--port {port}- the port, defaults to 6379--password {password}- the database password, default is no password--graph {key}- the graph key, defaults to "test"--infer- infer identity and labels from @id and @type, respectively
Adding the --show-query option will allow you to see the Cypher statements as
they are executed.
The YAML-based format is a simple dictionary of nodes and edges.
At the top-level, a graph is a dictionary whose keys define the nodes, schema, and edges. The keys can either be:
~schema- the schema definition for the property graph~edges- a set of fully qualified edges- {label} - a node label
A:
~label: Component
id: 'A'
name: 'Component A'
use: 12
~edges:
- ~to: B
~label: imports
- ~to: C
~label: imports
B:
~label: Component
id: 'B'
name: 'Component B'
use: 6
C:
~label: Component
id: 'C'
name: 'Component C'
use: 7
~edges:
e1:
~from: C
~to: B
~label: importsWhen a schema is specified via ~schema, the properties that establish
the node's identity can be specified.
Alternatively, we can use inferencing and the edge label format:
A:
@type: Component
@id: 'A'
name: 'Component A'
use: 12
:imports:
- ~to: B
- ~to: C
B:
@type: Component
@id: 'B'
name: 'Component B'
use: 6
C:
@type: Component
@id: 'C'
name: 'Component C'
use: 7
:imports:
- ~from: C
~to: BA node is a simple dictionary whose key/value pairs define properties all except for two special labels:
~labels- the set of Node labels~edges- the edges connected to the node- {label} - a property
A property can either be a simple key/value pair where the key will be the
property name. It can also be defined with the name: and value: keys for
property name values that are harder to encode as a key:
Funky:
name: 'Town'
p1:
name: "Meaning of life"
value: 42Node can specify an edge via a label and enumerate the target nodes and edge properties:
A:
id: 'A'
:child:
- ~to: B
use: 1209
- ~to: C
use: 432
B:
id: 'B'
:child:
e1:
~to: C
use: 128
C:
id: 'C'Nodes can also specify a set of edges that originate at the node via the ~edges
key. The edges are specified as a list or key labeled set:
A:
id: 'A'
~edges:
- ~to: B
~label: child
use: 1209
- ~to: C
~label: child
use: 432
B:
id: 'B'
~edges:
e1:
~to: C
~label: child
use: 128
C:
id: 'C'Edges can also be specified at the graph level instead of in the node. At the top-level, a single ~edges key is allowed that can specify edges from and to nodes. The
~from key must also be specified:
A:
id: 'A'
B:
id: 'B'
C:
id: 'C'
~edges:
- ~from: A
~to: B
~label: child
use: 1209
- ~from: A
~to: C
~label: child
use: 432
- ~from: B
~to: C
~label: child
use: 128Alternatively, you can specify a label for the edge and enumerate the edges associated with that label:
A:
id: 'A'
B:
id: 'B'
C:
id: 'C'
:child:
- ~from: A
~to: B
use: 1209
- ~from: A
~to: C
use: 432
- ~from: B
~to: C
use: 128Reusing the '@type' and '@id' properties from JSON-LD, if these properties are specified on a node, then the following inferences can be optionally applied:
- The value of @type will turn into a node label
- The value of @id will be used as the key for matching the node
Otherwise, labels and identity must be accomplished via the schema.
A schema can be specified at the top-level via the ~schema key. The schema itself is either embedded directly as text or has a single source key specifying the file location.
For example, in the imports graph example, the id property can be specified as
property that identifies the node. This can be helpful for generating merge or match queries.
The schema format is described separately and allows you to define nodes, labels, properties, and their descriptions.
The schema can be embedded as text:
~schema: |
(:Component {id})
.id = 'the component identifier'
.name = 'the component descriptive name'
.use = int 'a count of usage'
-[:imports]->(:Component) = 'an imported component'
A:
~label: Component
id: 'A'
name: 'Component A'
use: 12
~edges:
- ~to: B
~label: imports
- ~to: C
~label: imports
B:
~label: Component
id: 'B'
name: 'Component B'
use: 6
C:
~label: Component
id: 'C'
name: 'Component C'
use: 7
~edges:
- ~from: C
~to: B
~label: importsor via reference:
~schema:
source: schema.pgs
A:
~label: Component
id: 'A'
name: 'Component A'
use: 12
~edges:
- ~to: B
~label: imports
- ~to: C
~label: imports
B:
~label: Component
id: 'B'
name: 'Component B'
use: 6
C:
~label: Component
id: 'C'
name: 'Component C'
use: 7
~edges:
- ~from: C
~to: B
~label: importsThe graph source is just raw YAML and should be loaded directly using the yaml package:
import yaml
with open('graph.yaml','r') as input:
graph_data = yaml.load(input,Loader=yaml.Loader)Once you have loaded the graph YAML, you can read the graph into a sequence of item (NodeItem or EdgeRelationItem):
import yaml
from propgraph import read_graph
with open('graph.yaml','r') as input:
graph_data = yaml.load(input,Loader=yaml.Loader)
for item in read_graph(graph_data):
print(item)These items can be turned into cypher merge or create statements:
import yaml
from propgraph import read_graph, graph_to_cypher
with open('graph.yaml','r') as input:
graph_data = yaml.load(input,Loader=yaml.Loader)
for query in graph_to_cypher(read_graph(graph_data)):
print(query,end=';\n')Finally, the graph can easily be loaded into RedisGraph:
import yaml
from propgraph import read_graph, graph_to_cypher
import redis
from redisgraph import Graph
r = redis.Redis(host='localhost',port=6379,password='...')
rg = Graph('test',r)
with open('graph.yaml','r') as input:
graph_data = yaml.load(input,Loader=yaml.Loader)
for query in graph_to_cypher(read_graph(graph_data)):
rg.query(query)A schema can be loaded from a file:
from propgraph import SchemaParser
parser = SchemaParser()
with open('schema.pgs','r') as input:
schema = parser.parse(input)or a string:
from propgraph import SchemaParser
source = '''
(:Component {id})
.id = 'the component identifier'
.name = 'the component descriptive name'
.use = int 'a count of usage'
-[:imports]->(:Component) = 'an imported component'
'''
parser = SchemaParser()
schema = parser.parse(input)Documentation in Markdown format can be generate from the schema object:
import sys
from propgraph import SchemaParser
source = '''
(:Component {id})
.id = 'the component identifier'
.name = 'the component descriptive name'
.use = int 'a count of usage'
-[:imports]->(:Component) = 'an imported component'
'''
parser = SchemaParser()
schema = parser.parse(input)
schema.documentation(sys.stdout)Note: incomplete ...
read_graph(source,location=None,schema=None)
Reads a graph into a sequence of items
graph_to_cypher(stream,merge=True)
Transforms a sequence of items into a sequence of cypher statements
cypher_for_node(item,merge=True)
Returns a cypher statement to create a node from a node item.
cypher_for_edge_relation(item,merge=True)
Returns a cypher statement to create an edge from a edge relation item.