RelationalAI Python API Reference#

This reference provides detailed information on the classes and functions available in the relationalai Python package.

Table Of Contents#

• Table Of Contents
• Installation
• Usage
• Core Classes
• Standard Library
  • Built-ins
  • Submodules
• Native App Management
• See Also

Installation#

You can install the relationalai Python package locally using pip:

#pip install relationalai

See Getting Started: Local Installation for complete instructions.

If you’re using a cloud notebook, like Snowflake, Google Colab, or Hex, refer to Getting Started: Cloud Notebooks for instructions on installing relationalai in those environments.

Usage#

The relationalai Python package is used to build and query models that represent the entities, rules, and relationships that define your organization, its data, and the logic that drives its operations and decisions.

RAI models require a connection to the RAI Native App to evaluate queries. This connection can be configured with a raiconfig.toml file created using the rai init CLI command or using the Config class in a Python script or notebook:

#import relationalai as rai


# Create a model using a raiconfig.toml file.
model = rai.Model("MyModel")

# Create a model using a Config object.
cfg = rai.Config({
    "user": "<SNOWFLAKE_USER>",
    "password": "<SNOWFLAKE_PASSWORD>",
    "account": "<SNOWFLAKE_ACCOUNT>",
    "role": "<SNOWFLAKE_ROLE>",
    "warehouse": "<SNOWFLAKE_WAREHOUSE>",
    "rai_app_name": "relationalai",
}
model = rai.Model("MyModel", config=cfg)

See the configuration guide for more information on configuring your model’s connection to the RAI Native App.

The relationalai package offers core classes and standard library modules for building models and analyzing data. Models are written using RAI’s declarative Python query-builder syntax. Check out the guides in Develop With RAI for a general introduction to building models.

Core Classes#

The following classes are used to define RAI models and write rules and queries:

Here’s an example of how these classes are used in a simple model:

#import relationalai as rai

model = rai.Model("MyModel")
#^^^^-- Model

Person = model.Type("Person")
Adult = model.Type("Adult")
#^^^^-- Type

Person.name.declare()
Person.age.declare()
#^^^^^^^^^-- Property


with model.rule():
#    ^^^^^^^^^^^^-- Context
    bob = Person.add(id=1)
    bob.set(name="Bob", age=30)
#   ^^^-- Instance (Producer)

with model.rule():
#   vvvvvv-- Instance (Producer)
    person = Person()
#   vvvvvvvvvv-- InstanceProperty (Producer)
    person.age >= 18
#   ^^^^^^^^^^^^^^^^-- Expression (Producer)
    person.set(Adult)
#              ^^^^^-- Type

#                  Context
#                 /
#    vvvvvvvvvvvvv    vvvvvv-- ContextSelect
with model.query() as select:
    adult = Adult()
    response = select(adult.name, adult.age)
#   ^^^^^^^^          ^^^^^^^^^^  ^^^^^^^^^
#           \                   \/
#            Context     InstanceProperty

See the Core Concepts guide for more details on how these classes fit together.

Standard Library#

The relationalai.std namespace contains tools for working with various kinds of data in a model. It also provides APIs for more advanced analytical operations, like graph algorithms.

Built-ins#

The following functions are available for import directly from relationalai.std:

Submodules#

For working with data, the following submodules are available:

Native App Management#

The relationalai package also provides access to common Native App management procedures, such as activating and deactivating the app and managing data streams, through an instance of the Provider class:

#import relationalai as rai

# Create a Provider instance and connect using the active profile in your
# raiconfig.toml file. Use the config parameter to connect using a Config
# object if you're not using a raiconfig.toml file.
app = rai.Provider()

# Activate the RAI Native App.
app.activate()

# Build a model, write rules, and run queries...

# Deactivate the RAI Native App.
app.deactivate()

Note that several methods of the Provider class require connecting to the native app using a Snowflake user that has been granted the app_admin application role. See Manage RAI for more information on managing the RAI Native App.

See Also#

• Core Concepts
• Basic Functionality
• Developer Guides
• Example Notebooks