Using the Jupyter Notebook Interface for PGQL Property Graphs

16.4 Using the Jupyter Notebook Interface for PGQL Property Graphs

You can use the Jupyter notebook interface to create, load, and query PGQL property graphs through Python.

In addition, you can visualize PGQL property graphs using the graph visualization extension in both Jupyter Notebook and JupyterLab. See Getting Started with the Graph Visualization Extension in Jupyter Environments for more information.

The following steps show you how to create and visualize PGQL property graphs in a Jupyter environment (either Jupyter Notebook or JupyterLab).

Install the desired Jupyter Notebook interface (either Jupyter Notebook or JupyterLab).
Ensure that your Jupyter installation is added to the PATH environment variable.
Install the graph visualization extension in your Jupyter environment. See Installing the Graph Visualization Extension for Jupyter for more information.

Create a PGQL property graph using the OPG4Py Python API in a Jupyter notebook.

Establish a connection to the database using the Python oracledb driver.

The following code shows how to connect to your local database:

import oracledb
host = "<host>"
port = 1521
service_name = "<service_name>"
dsn = f"{host}:{port}/{service_name}"

conn = oracledb.connect(
    user="graphuser",
    password="<password_for_graphuser>",
    dsn=dsn
)
jdbc_url = f"jdbc:oracle:thin:@//{host}:{port}/{service_name}"
print("Connection established successfully.")

To connect to an Autonomous AI Database, download the database wallet and unzip it to a secure directory on the machine running your Jupyter notebook. Then, connect using a tns_alias from tnsnames.ora, as shown in the following code:

import oracledb

username = "adbgraphuser"
password = "<password_for_adbgraphuser>"
wallet_location = "<path_to_wallet_file>"
wallet_password = "<password_for_wallet>"
adb_tns_alias = "<tns_alias_in_tnsnames.ora>"

connection = oracledb.connect(
    user=username,
    password=password,
    dsn=adb_tns_alias,
    config_dir=wallet_location,
    wallet_location=wallet_location,
    wallet_password=wallet_password,
    )
jdbc_url = f"jdbc:oracle:thin:@{adb_tns_alias}?TNS_ADMIN={wallet_location}"
print("Connection established successfully.")

Create a PGQL property graph using the jdbc_url created in the previous step.


import opg4py

pgql_connection = opg4py.pgql.get_connection("<username>", "<password>", jdbc_url)
pgql_statement = pgql_connection.create_statement()

create_pgql_graph = '''
    CREATE PROPERTY GRAPH BANK_GRAPH_PGQL
    VERTEX TABLES (
            BANK_ACCOUNTS
            KEY (ID)
            PROPERTIES (ID, Name, Balance)
        )
        EDGE TABLES (
            BANK_TRANSFERS
            KEY (TXN_ID)
            SOURCE KEY (src_acct_id) REFERENCES BANK_ACCOUNTS(ID)
            DESTINATION KEY (dst_acct_id) REFERENCES BANK_ACCOUNTS(ID)
            PROPERTIES (src_acct_id, dst_acct_id, amount)
        )
    OPTIONS ( PG_PGQL )
'''

pgql_result_set = pgql_statement.execute(create_pgql_graph)

pgql_statement.close()
pgql_connection.close()

Run a PGQL query on the property graph and visualize the query results using PgqlGraphVisualization in oraclegraph Jupyter extension.

You can create a PgqlGraphVisualization instance as shown in the following code examples:

For Local Database

vq = PgqlGraphVisualization(
    "<db_username>",
    "<db_password>",
    connection={"dsn": f"{<host>}:{<port>}/{<service_name>}"},
)

For Autonomous AI Database (Wallet/tnsnames.ora)

vq = PgqlGraphVisualization(
    "<db_username>",
    "<db_password>",
    connection={
        "tns_alias": "<tns_alias_in_tnsnames.ora>",
        "wallet_location": "<wallet_location>",
    },
)

The following example shows how to visualize a PGQL query using PgqlGraphVisualization against the property graph created in step-4.

from pypgx import setloglevel
from oraclegraph import GraphVisualization, PgqlGraphVisualization

setloglevel('ROOT', 'WARN')

vq = PgqlGraphVisualization(
    "graphuser",
    "<password_for_graphuser>",
    connection={"dsn": f"<host_name:port>/<service_name>"},
)

query='''
SELECT * FROM GRAPH_TABLE (bank_graph_pgql
  MATCH (a) -[e]-> (b)
  WHERE a.ID=816
COLUMNS (a.ID AS src_ac, e.AMOUNT AS amount, b.ID AS dest_ac)
)
'''
graph_query = vq.visualize_query(query)

defaults_feature = {
  "interactionActive": True,
  "1stickyActive": True
}

base_styles = {
  "vertex": {
    "label": "${properties.ID}",
    "size": 20
  },
  "edge": {
    "label": "${properties.AMOUNT}"
  }
}

rule_based_styles = [{
    "stylingEnabled": True,
    "component": "vertex",
    "target": "vertex",
    "conditions": {
      "conditions": [{
        "property": "BALANCE",
        "operator": "<=",
        "value": 5000
      }],
    },
    "style": {
      "color": "green"
    },
    "legendTitle": "Balance Filter",
    "legendDisplayed": True
  }

]

settings = {
  "numberOfHops": 2,
  "showLegend": True,
  "defaults": defaults_feature,
  "baseStyles": base_styles,
  "ruleBasedStyles": rule_based_styles
}
bank_graph = GraphVisualization(data = graph_query, settings = settings)
bank_graph.height = 600
display(bank_graph)

The preceding code produces the following visualization result:

Figure 16-1 PGQl Property Graph Visualization

Description of "Figure 16-1 PGQl Property Graph Visualization"

Load and analyze the property graph in the graph server (PGX).
The following example shows loading the PGQL property graph into the graph server (PGX) and running graph algorithms for analysis:
```
base_url = "https://localhost:7007"
username = "graphuser"
password = "<password_for_graphuser>"

instance = graph_server.get_instance(base_url, username, password)
session = instance.create_session('jupyter')
graph = session.read_graph_by_name('BANK_GRAPH_PGQL', 'pg_pgql')
analyst = session.create_analyst()
analyst.pagerank(graph)
rs = graph.query_pgql("SELECT id(x), x.pagerank FROM MATCH (x) LIMIT 5")
rs.print()
```
Figure 16-2 Running Graph Algorithms in Jupyter Notebook

Description of "Figure 16-2 Running Graph Algorithms in Jupyter Notebook"

Parent topic: Getting Started with the Client Tools