haive.core.persistence.postgres_saver_override

PostgreSQL persistence utilities with Pydantic support.

This module provides utilities for handling Pydantic models in PostgreSQL persistence. The main functionality is the JSON encoder configuration that ensures Pydantic models are properly serialized to JSONB columns.

The override classes are kept for backward compatibility and as a fallback when using connection strings directly. However, the preferred approach is to configure the connection pool with the configure parameter.

Classes

AsyncPostgresSaverNoPreparedStatements	Async PostgresSaver with proper configuration.
PostgresSaverNoPreparedStatements	PostgresSaver that disables prepared statements and handles Pydantic models.

Functions

configure_postgres_json(connection)	Configure a PostgreSQL connection to handle Pydantic JSON serialization.
pydantic_aware_json_dumps(obj)	JSON encoder that handles Pydantic models.

Module Contents

class haive.core.persistence.postgres_saver_override.AsyncPostgresSaverNoPreparedStatements(conn, pipe=None, serde=None)

Bases: langgraph.checkpoint.postgres.aio.AsyncPostgresSaver

Async PostgresSaver with proper configuration.

This class is kept for backward compatibility and for cases where you need to use from_conn_string directly. The preferred approach is to configure the connection pool with the configure parameter in postgres_config.py.

Parameters:

• conn (langgraph.checkpoint.postgres._ainternal.Conn)

• pipe (psycopg.AsyncPipeline | None)

• serde (langgraph.checkpoint.serde.base.SerializerProtocol | None)

classmethod from_conn_string(conn_string)

Async:

Parameters:

conn_string (str)

Return type:

AsyncPostgresSaverNoPreparedStatements

Create an AsyncPostgresSaver with proper configuration.

class haive.core.persistence.postgres_saver_override.PostgresSaverNoPreparedStatements(conn, pipe=None, serde=None)

Bases: langgraph.checkpoint.postgres.PostgresSaver

PostgresSaver that disables prepared statements and handles Pydantic models.

This class is kept for backward compatibility and for cases where you need to use from_conn_string directly. The preferred approach is to configure the connection pool with the configure parameter in postgres_config.py.

Parameters:

• conn (_internal.Conn)

• pipe (psycopg.Pipeline | None)

• serde (langgraph.checkpoint.serde.base.SerializerProtocol | None)

classmethod from_conn_string(conn_string)

Create a PostgresSaver with proper configuration.

Parameters:

conn_string (str)

Return type:

PostgresSaverNoPreparedStatements

haive.core.persistence.postgres_saver_override.configure_postgres_json(connection)

Configure a PostgreSQL connection to handle Pydantic JSON serialization.

Parameters:

connection – A psycopg connection object

haive.core.persistence.postgres_saver_override.pydantic_aware_json_dumps(obj)

JSON encoder that handles Pydantic models.

This encoder ensures that Pydantic models are properly serialized when stored in PostgreSQL JSONB columns.