pytimbr-sqla

Timbr Python SQLAlchemy connector


Keywords
timbr, timbr-python, timbr-connector, python-connector, PyTimbr, py-timbr, pytimbr_sqla, pytimbrsqlalchemy, Py-TimbrSQLAlchemy, Py-Timbr-SQLAlchemy, knowledge-graph, python3, sqlalchemy
License
MIT
Install
pip install pytimbr-sqla==1.0.2

Documentation

Timbr logo

timbr Python connector sample file

This is a sample repository for how to connect to timbr using SQLAlchemy and Python.

Dependencies

  • Access to a timbr-server
  • Python from 3.7.13 or newer
  • Support SQLAlchemy 1.4.36 or newer but not version 2.x yet.

Installation

  • Install as clone repository:

  • Install using pip and git:

    • pip install git+https://github.com/WPSemantix/timbr_python_SQLAlchemy
  • Install using pip:

    • pip install pytimbr-sqla

Known issues

If you encounter a problem installing PyHive with sasl dependencies on windows, install the following wheel (for 64bit Windows) by running:

pip install https://download.lfd.uci.edu/pythonlibs/archived/cp37/sasl-0.3.1-cp37-cp37m-win_amd64.whl

For Python 3.9:

pip install https://download.lfd.uci.edu/pythonlibs/archived/sasl-0.3.1-cp39-cp39-win_amd64.whl

Sample usage

  • For an example of how to use the Python SQLAlchemy connector for timbr, follow this example file
  • For an example of how to use the Python SQLAlchemy connector with 'PyHive' as async query for timbr, follow this example file
  • For an example of how to use the Python SQLAlchemy connector with 'PyHive' as sync query for timbr, follow this example file

Connection parameters

General example

  hostname = '<TIMBR_IP/HOST>'
  port = '<TIMBR_PORT>'
  ontology = '<ONTOLOGY_NAME>'
  protocol = '<http/https>'
  username = '<TIMBR_USER/token>'
  password = '<TIMBR_PASSWORD/TOKEN_VALUE>'

  # hostname - The IP / Hostname of the Timbr server (not necessarily the hostname of the Timbr platform).
  # port - The port to connect to in the Timbr server. Timbr's default port with enabled_ssl is 443 without SSL is 11000.
  # ontology = The name of the ontology (knowledge graph) to connect.
  # protocol - Connection protocol can be 'http' or 'https'.
  # username - Use 'token' as the username when connecting using a Timbr token, otherwise use the user name.
  # password - If using a token as a username then the pass is the token value, otherwise its the user's password.

HTTP example with dummy data

Username and password

  hostname = 'mytimbrenv.com'
  port = '11000'
  ontology = 'my_ontology'
  protocol = 'http'
  username = 'timbr'
  password = 'StrongPassword'

Timbr token

  hostname = 'mytimbrenv.com'
  port = '11000'
  ontology = 'my_ontology'
  protocol = 'http'
  username = 'token'
  password = '<TOKEN_VALUE>'

HTTPS example with dummy data

Username and password

  hostname = 'mytimbrenv.com'
  port = '443'
  ontology = 'my_ontology'
  protocol = 'https'
  username = 'timbr'
  password = 'StrongPassword'

Timbr token

  hostname = 'mytimbrenv.com'
  port = '443'
  ontology = 'my_ontology'
  protocol = 'https'
  username = 'token'
  password = '<TOKEN_VALUE>'

Connect options

Connect using 'pytimbr_sqla' and 'SQLAlchemy' packages

  from sqlalchemy import create_engine

  # Declare the connection variables
  # General example
  hostname = '<TIMBR_IP/HOST>'
  port = '<TIMBR_PORT>'
  ontology = '<ONTOLOGY_NAME>'
  protocol = '<http/https>'
  username = '<TIMBR_USER/token>'
  password = '<TIMBR_PASSWORD/TOKEN_VALUE>'

  # hostname - The IP / Hostname of the Timbr server (not necessarily the hostname of the Timbr platform).
  # port - The port to connect to in the Timbr server. Timbr's default port with enabled_ssl is 443 without SSL is 11000.
  # ontology = The name of the ontology (knowledge graph) to connect.
  # protocol - Connection protocol can be 'http' or 'https'.
  # username - Use 'token' as the username when connecting using a Timbr token, otherwise use the user name.
  # password - If using a token as a username then the pass is the token value, otherwise its the user's password.

  # Create new sqlalchemy connection
  engine = create_engine(f"timbr+{protocol}://{username}@{ontology}:{password}@{hostname}:{port}")

  # Connect to the created engine
  conn = engine.connect()

  # Execute a query
  query = "SHOW CONCEPTS"
  concepts = conn.execute(query).fetchall()

  # Display the results of the execution
  for concept in concepts:
    print(concept)

Attention:

timbr works only as async when running a query, if you want to use standard PyHive you have two options

Connect using 'PyHive' and 'SQLAlchemy' packages

Connect using PyHive Async Query

  from sqlalchemy import create_engine
  from TCLIService.ttypes import TOperationState

  # Declare the connection variables
  # General example
  hostname = '<TIMBR_IP/HOST>'
  port = '<TIMBR_PORT>'
  ontology = '<ONTOLOGY_NAME>'
  protocol = '<http/https>'
  username = '<TIMBR_USER/token>'
  password = '<TIMBR_PASSWORD/TOKEN_VALUE>'
  connect_args = {
    'configuration': {
      'set:hiveconf:hiveMetadata': 'true',
      'set:hiveconf:active_datasource': '<datasource_name>',
      'set:hiveconf:queryTimeout': '<TIMEOUT_IN_SECONDS>',
    },
  }

  # hostname - The IP / Hostname of the Timbr server (not necessarily the hostname of the Timbr platform).
  # port - The port to connect to in the Timbr server. Timbr's default port with enabled_ssl is 443 without SSL is 11000.
  # ontology = The name of the ontology (knowledge graph) to connect.
  # protocol - Connection protocol can be 'http' or 'https'.
  # username - Use 'token' as the username when connecting using a Timbr token, otherwise use the user name.
  # password - If using a token as a username then the pass is the token value, otherwise its the user's password.
  # connect_args - The connection special arguments for extra customization. The only argument you must have is the first one (set:hiveconf:hiveMetadata) the others are optional.

  # Create new sqlalchemy connection
  engine = create_engine(f"hive+{protocol}://{username}@{ontology}:{password}@{hostname}:{port}", connect_args = connect_args)

  # Connect to the created engine
  conn = engine.connect()
  dbapi_conn = engine.raw_connection()
  cursor = dbapi_conn.cursor()

  # Execute a query
  query = "SHOW CONCEPTS"
  cursor.execute(query)

  # Check the status of this execution
  status = cursor.poll().operationState
  while status in (TOperationState.INITIALIZED_STATE, TOperationState.RUNNING_STATE):
    status = cursor.poll().operationState

  # Display the results of the execution
  results = cursor.fetchall()
  print(results)

Connect using PyHive Sync Query

  from sqlalchemy import create_engine
  from TCLIService.ttypes import TOperationState

  # Declare the connection variables
  # General example
  hostname = '<TIMBR_IP/HOST>'
  port = '<TIMBR_PORT>'
  ontology = '<ONTOLOGY_NAME>'
  protocol = '<http/https>'
  username = '<TIMBR_USER/token>'
  password = '<TIMBR_PASSWORD/TOKEN_VALUE>'
  connect_args = {
    'configuration': {
      'set:hiveconf:async': 'false',
      'set:hiveconf:hiveMetadata': 'true',
      'set:hiveconf:active_datasource': '<datasource_name>',
      'set:hiveconf:queryTimeout': '<TIMEOUT_IN_SECONDS>',
    },
  }

  # hostname - The IP / Hostname of the Timbr server (not necessarily the hostname of the Timbr platform).
  # port - The port to connect to in the Timbr server. Timbr's default port with enabled_ssl is 443 without SSL is 11000.
  # ontology = The name of the ontology (knowledge graph) to connect.
  # protocol - Connection protocol can be 'http' or 'https'.
  # username - Use 'token' as the username when connecting using a Timbr token, otherwise use the user name.
  # password - If using a token as a username then the pass is the token value, otherwise its the user's password.
  # connect_args - The connection special arguments for extra customization. The only 2 arguments you must have are the first and the second one (set:hiveconf:async, set:hiveconf:hiveMetadata) the others are optional.

  # Create new sqlalchemy connection
  engine = create_engine(f"hive+{protocol}://{username}@{ontology}:{password}@{hostname}:{port}", connect_args = connect_args)

  # Connect to the created engine
  conn = engine.connect()

  # Use the connection to execute a query
  query = "SHOW CONCEPTS"
  results = conn.execute(query).fetchall()

  # Display the results of the execution
  print(results)