tanat.store.sequence.builder.type package

Submodules

tanat.store.sequence.builder.type.event module

Event sequence store builder.

class tanat.store.sequence.builder.type.event.EventSequenceStoreBuilder

Bases: SequenceStoreBuilder

Fluent builder for Event sequence stores.

Exposes time_column explicitly on every add_* call.

add_csv(path: str | Path, *, id_column: str, features: str | list[str], time_column: str | None = None, is_static: bool = False, **reader_kwargs) → EventSequenceStoreBuilder

Register a CSV file.

add_dataframe(data: DataFrame | LazyFrame, *, id_column: str, features: str | list[str], time_column: str | None = None, is_static: bool = False, **_kw) → EventSequenceStoreBuilder

Register an in-memory Polars / Pandas DataFrame.

add_parquet(path: str | Path, *, id_column: str, features: str | list[str], time_column: str | None = None, is_static: bool = False, **reader_kwargs) → EventSequenceStoreBuilder

Register a Parquet file (glob patterns supported).

add_sql(connection: str, query: str, *, id_column: str, features: str | list[str], time_column: str | None = None, is_static: bool = False, **sql_kwargs) → EventSequenceStoreBuilder

Register a SQL query (requires connectorx).

tanat.store.sequence.builder.type.interval module

Interval sequence store builder.

class tanat.store.sequence.builder.type.interval.IntervalSequenceStoreBuilder(*, sort_anchor: str = 'start')

Bases: SequenceStoreBuilder

Fluent builder for Interval sequence stores.

Exposes start_column and end_column explicitly on every add_* call. Pass sort_anchor at construction time to control the intra-sequence sort column: "start" (default), "end", or "middle" (midpoint of each interval).

__init__(*, sort_anchor: str = 'start') → None

add_csv(path: str | Path, *, id_column: str, features: str | list[str], start_column: str | None = None, end_column: str | None = None, is_static: bool = False, **reader_kwargs) → IntervalSequenceStoreBuilder

Register a CSV file.

add_dataframe(data: DataFrame | LazyFrame, *, id_column: str, features: str | list[str], start_column: str | None = None, end_column: str | None = None, is_static: bool = False, **_kw) → IntervalSequenceStoreBuilder

Register an in-memory Polars / Pandas DataFrame.

add_parquet(path: str | Path, *, id_column: str, features: str | list[str], start_column: str | None = None, end_column: str | None = None, is_static: bool = False, **reader_kwargs) → IntervalSequenceStoreBuilder

Register a Parquet file (glob patterns supported).

add_sql(connection: str, query: str, *, id_column: str, features: str | list[str], start_column: str | None = None, end_column: str | None = None, is_static: bool = False, **sql_kwargs) → IntervalSequenceStoreBuilder

Register a SQL query (requires connectorx).

tanat.store.sequence.builder.type.state module

State sequence store builder.

class tanat.store.sequence.builder.type.state.StateSequenceStoreBuilder(*, end_value: datetime | int | float | None = None, validate_continuity: bool = True)

Bases: SequenceStoreBuilder

Fluent builder for State sequence stores.

States are contiguous and non-overlapping intervals defined by start_column and an optional end_column.

When end_column is omitted at registration time, T_END is auto-computed as the next T_START within each sequence. Pass end_value at construction time to set the sentinel for the last state of every sequence (None → leaves the last T_END as null).

When end_column is provided by the user, validate_continuity (default: True) checks that states are truly contiguous (T_END[i] == T_START[i+1] within each sequence) before writing. Set it to False to skip this check on large datasets where the cost of a full collect() is unacceptable.

__init__(*, end_value: datetime | int | float | None = None, validate_continuity: bool = True) → None

add_csv(path: str | Path, *, id_column: str, features: str | list[str], start_column: str | None = None, end_column: str | None = None, is_static: bool = False, **reader_kwargs) → StateSequenceStoreBuilder

Register a CSV file.

add_dataframe(data: DataFrame | LazyFrame, *, id_column: str, features: str | list[str], start_column: str | None = None, end_column: str | None = None, is_static: bool = False, **_kw) → StateSequenceStoreBuilder

Register an in-memory Polars / Pandas DataFrame.

add_parquet(path: str | Path, *, id_column: str, features: str | list[str], start_column: str | None = None, end_column: str | None = None, is_static: bool = False, **reader_kwargs) → StateSequenceStoreBuilder

Register a Parquet file (glob patterns supported).

add_sql(connection: str, query: str, *, id_column: str, features: str | list[str], start_column: str | None = None, end_column: str | None = None, is_static: bool = False, **sql_kwargs) → StateSequenceStoreBuilder

Register a SQL query (requires connectorx).

Module contents

Register SequenceStoreBuilder subtypes.

class tanat.store.sequence.builder.type.EventSequenceStoreBuilder

Bases: SequenceStoreBuilder

Fluent builder for Event sequence stores.

Exposes time_column explicitly on every add_* call.

add_csv(path: str | Path, *, id_column: str, features: str | list[str], time_column: str | None = None, is_static: bool = False, **reader_kwargs) → EventSequenceStoreBuilder

Register a CSV file.

add_dataframe(data: DataFrame | LazyFrame, *, id_column: str, features: str | list[str], time_column: str | None = None, is_static: bool = False, **_kw) → EventSequenceStoreBuilder

Register an in-memory Polars / Pandas DataFrame.

add_parquet(path: str | Path, *, id_column: str, features: str | list[str], time_column: str | None = None, is_static: bool = False, **reader_kwargs) → EventSequenceStoreBuilder

Register a Parquet file (glob patterns supported).

add_sql(connection: str, query: str, *, id_column: str, features: str | list[str], time_column: str | None = None, is_static: bool = False, **sql_kwargs) → EventSequenceStoreBuilder

Register a SQL query (requires connectorx).

class tanat.store.sequence.builder.type.IntervalSequenceStoreBuilder(*, sort_anchor: str = 'start')

Bases: SequenceStoreBuilder

Fluent builder for Interval sequence stores.

Exposes start_column and end_column explicitly on every add_* call. Pass sort_anchor at construction time to control the intra-sequence sort column: "start" (default), "end", or "middle" (midpoint of each interval).

__init__(*, sort_anchor: str = 'start') → None

add_csv(path: str | Path, *, id_column: str, features: str | list[str], start_column: str | None = None, end_column: str | None = None, is_static: bool = False, **reader_kwargs) → IntervalSequenceStoreBuilder

Register a CSV file.

add_dataframe(data: DataFrame | LazyFrame, *, id_column: str, features: str | list[str], start_column: str | None = None, end_column: str | None = None, is_static: bool = False, **_kw) → IntervalSequenceStoreBuilder

Register an in-memory Polars / Pandas DataFrame.

add_parquet(path: str | Path, *, id_column: str, features: str | list[str], start_column: str | None = None, end_column: str | None = None, is_static: bool = False, **reader_kwargs) → IntervalSequenceStoreBuilder

Register a Parquet file (glob patterns supported).

add_sql(connection: str, query: str, *, id_column: str, features: str | list[str], start_column: str | None = None, end_column: str | None = None, is_static: bool = False, **sql_kwargs) → IntervalSequenceStoreBuilder

Register a SQL query (requires connectorx).

class tanat.store.sequence.builder.type.StateSequenceStoreBuilder(*, end_value: datetime | int | float | None = None, validate_continuity: bool = True)

Bases: SequenceStoreBuilder

Fluent builder for State sequence stores.

States are contiguous and non-overlapping intervals defined by start_column and an optional end_column.

When end_column is omitted at registration time, T_END is auto-computed as the next T_START within each sequence. Pass end_value at construction time to set the sentinel for the last state of every sequence (None → leaves the last T_END as null).

When end_column is provided by the user, validate_continuity (default: True) checks that states are truly contiguous (T_END[i] == T_START[i+1] within each sequence) before writing. Set it to False to skip this check on large datasets where the cost of a full collect() is unacceptable.

__init__(*, end_value: datetime | int | float | None = None, validate_continuity: bool = True) → None

add_csv(path: str | Path, *, id_column: str, features: str | list[str], start_column: str | None = None, end_column: str | None = None, is_static: bool = False, **reader_kwargs) → StateSequenceStoreBuilder

Register a CSV file.

add_dataframe(data: DataFrame | LazyFrame, *, id_column: str, features: str | list[str], start_column: str | None = None, end_column: str | None = None, is_static: bool = False, **_kw) → StateSequenceStoreBuilder

Register an in-memory Polars / Pandas DataFrame.

add_parquet(path: str | Path, *, id_column: str, features: str | list[str], start_column: str | None = None, end_column: str | None = None, is_static: bool = False, **reader_kwargs) → StateSequenceStoreBuilder

Register a Parquet file (glob patterns supported).

add_sql(connection: str, query: str, *, id_column: str, features: str | list[str], start_column: str | None = None, end_column: str | None = None, is_static: bool = False, **sql_kwargs) → StateSequenceStoreBuilder

Register a SQL query (requires connectorx).