Config

relationalai.config.config

Represents a validated PyRel configuration.

Use create_config to load your raiconfig.yaml and optionally override specific settings in code.

Instances are typically created by create_config, either from a supported config file or programmatically. The config validates that at least one connection exists, can auto-select default_connection when only one connection is defined, and provides helpers such as Config.get_connection and Config.get_session.

Examples

Load a config and override a specific setting:

from relationalai.config import create_config

cfg = create_config(jobs={"print_progress": False})

Notes

For Snowflake connections, the authenticator setting defaults to "username_password" when omitted.

Attributes

.active_profile

Config.active_profile: (str, optional)

Name of the profile currently applied.

.profiles

Config.profiles: dict

Profile overlays keyed by name (YAML key profile).

.connections

Config.connections: dict

Named connection definitions.

.default_connection

Config.default_connection: (str, optional)

Connection name used when no explicit name is provided.

.execution

Config.execution: ExecutionConfig

Execution middleware configuration. See ExecutionConfig for details.

.data

Config.data: DataConfig

Data loading and streaming configuration. See DataConfig for details.

.compiler

Config.compiler: CompilerConfig

Compiler configuration. See CompilerConfig for details.

.model

Config.model: ModelConfig

Model configuration. See ModelConfig for details.

.reasoners

Config.reasoners: ReasonersConfig

Reasoners configuration. See ReasonersConfig for details.

.debug

Config.debug: DebugConfig

Debug configuration. See DebugConfig for details.

.jobs

Config.jobs: JobsConfig

Job execution configuration. See JobsConfig for details.

.outputs

Config.outputs: OutputsConfig

Settings for the tables and relations your model produces. Use this to customize how specific outputs are computed — for example, to send a particular recursive relation to the logic reasoner. See OutputsConfig for details.

.tables

Config.tables: (TablesConfig, optional)

Table type defaults and per-table overrides. Controls whether model.Table() output tables are created as iceberg or native Snowflake tables, and allows per-table FQN remapping, external volume configuration, and schema prefixing. See TablesConfig for details.

Methods

.get_default_connection()

Config.get_default_connection() -> ConnectionConfig

Return the connection named by Config.default_connection.

Returns:

• ConnectionConfig - The selected default connection.

Raises:

• ValueError - Raised in the following cases:
  • Multiple connections exist but default_connection is not set.
  • default_connection names a connection that is not present.

Notes:

If Config.default_connection is unset and exactly one connection is defined, that connection is returned.

Referenced By:

RelationalAI Documentation
└──  Build With RelationalAI
    └──  Understand how PyRel works > Build a semantic model
        └──  Create a model instance
            └──  Access configuration values

.get_connection()

Config.get_connection(connection_type: type[T], name: str | None = None) -> T

Return a configured connection, optionally by name.

If name is omitted, this returns the default connection (see Config.get_default_connection). The selected connection is validated to be an instance of connection_type.

Parameters:

• connection_type

  (type[T]) - Expected connection class.

• name

  (str, default: None) - Connection name to look up in Config.connections. Defaults to Config.default_connection.

Returns:

• T - The selected connection.

Raises:

• ValueError - Raised in the following cases:
  • name is provided but no such connection exists.
  • No default connection can be determined.
  • The selected connection is not an instance of connection_type.

.get_session()

Config.get_session(
    connection_type: (
        type[SnowflakeConnection]
        | type[DuckDBConnection]
        | type[LocalConnection]
        | None
    ) = None,
) -> snowflake.snowpark.Session | duckdb.DuckDBPyConnection | requests.Session

Return a session for a configured connection.

When connection_type is omitted, this uses the default connection (see Config.get_default_connection). When provided, it resolves a connection of that type via Config.get_connection.

Parameters:

• connection_type

  (type[SnowflakeConnection] | type[DuckDBConnection] | type[LocalConnection], default: None) - Connection class to select when you have multiple connections configured.

Returns:

• snowflake.snowpark.Session | duckdb.DuckDBPyConnection | requests.Session - The underlying session object for the selected connection.

Inheritance Hierarchy

Subclassed By

 config > config
├──  ConfigFromDBT
├──  ConfigFromRAIConfigToml
├──  ConfigFromSnowflake
└──  RAIConfig

Used By

.
├──  semantics > frontend > base
│   └──  Model
└──  services > reasoners > client
    └──  ReasonersClient

Returned By

 config > config
└──  create_config

Referenced By

RelationalAI Documentation
└──  Build With RelationalAI
    └──  Understand how PyRel works
        ├──  Configure PyRel
        │   ├──  Overview
        │   │   └──  How validation works
        │   ├──  Load configuration from files
        │   │   ├──  How file discovery works
        │   │   └──  Create a raiconfig.yaml file
        │   └──  Configure Snowflake authentication
        │       └──  Clear the session cache
        └──  Build a semantic model
            └──  Create a model instance
                └──  Access configuration values