timescale_loader

timescale_loader ¤

TimescaleDBDataAccess ¤

TimescaleDBDataAccess(
    start: str,
    end: str,
    uuids: list[str],
    db_config: dict[str, Any],
)

Load timeseries data for given UUIDs and a time range from TimescaleDB.

Needs the postgres extra (pip install ts-shape[postgres]: SQLAlchemy and the psycopg2 driver). Creating the loader does not connect; the first fetch does. repr(loader) masks the password.

Initialize the TimescaleDB loader and create the database engine.

Parameters:

Name	Type	Description	Default
`start`	`str`	Start of the time range (inclusive) to query.	required
`end`	`str`	End of the time range (inclusive) to query.	required
`uuids`	`list[str]`	List of UUIDs to retrieve data for.	required
`db_config`	`dict[str, Any]`	Connection settings with the keys `db_user`, `db_pass`, `db_host` and `db_name`, and optionally `db_port` (default 5432) and `db_driver` (the DB-API driver, default `"psycopg2"`). The password is used as is; it needs no URL escaping.	required

Raises:

Type	Description
`ImportError`	SQLAlchemy or the database driver is not installed.

fetch_data_as_parquet ¤

fetch_data_as_parquet(output_dir: str) -> None

Retrieve timeseries data from TimescaleDB and save it as Parquet files.

Rows are grouped by the hour of systime and written to one file per UUID and hour at <output_dir>/year/month/day/hour/<uuid>.parquet.

Parameters:

Name	Type	Description	Default
`output_dir`	`str`	Base directory to save the Parquet files.	required

fetch_data_as_dataframe ¤

fetch_data_as_dataframe() -> pd.DataFrame

Retrieve timeseries data from TimescaleDB as a single DataFrame.

Returns:

Type	Description
`DataFrame`	A combined DataFrame with data for all specified UUIDs within the time
`DataFrame`	range.