Feature Definitions - Attributes

FeatureDefinitionAttributes(perfdb)

Class used for handling feature definition attributes. Can be accessed via perfdb.features.definitions.attributes.

Parameters:

• perfdb

  (PerfDB) –

  Top level object carrying all functionality and the connection handler.

Source code in echo_postgres/perfdb_root.py

def __init__(self, perfdb: e_pg.PerfDB) -> None:
    """Base class that all subclasses should inherit from.

    Parameters
    ----------
    perfdb : PerfDB
        Top level object carrying all functionality and the connection handler.

    """
    self._perfdb: e_pg.PerfDB = perfdb

get(feature_names=None, object_names=None, object_models=None, data_source_types=None, attribute_names=None, attribute_values=None, filter_type='and', output_type='dict', values_only=False)

Method to get the attributes of the given features.

The most useful keys/columns returned are:

• data_source_type_name
• attribute_name
• attribute_value
• data_type_name

Parameters:

• feature_names

  (list[str] | None, default: None ) –

  Names of the features to get the attributes from. If set to None will get from all. By default None

• object_names

  (list[str] | None, default: None ) –

  List of object names to filter the results. If set, will add the models of these objects to the object_models filter. By default None

• object_models

  (list[str] | None, default: None ) –

  List of object model names to filter the results. By default None

• data_source_types

  (list[str] | None, default: None ) –

  List of data source type names to filter the results. By default None

• attribute_names

  (list[str] | None, default: None ) –

  List of attribute names to filter the results. If set to None will get all. By default None

• attribute_values

  (dict[str, Any] | None, default: None ) –

  Dictionary of attribute names and values to filter the results. It must be in the format {attribute_name: attribute_value, ...}. By default None

• filter_type

  (Literal['and', 'or'], default: 'and' ) –

  How to treat multiple filters. Can be one of ["and", "or"]. By default "and"

• output_type

  (Literal['dict', 'DataFrame'], default: 'dict' ) –

  Output type of the data. Can be one of ["dict", "DataFrame"] By default "dict"

• values_only

  (bool, default: False ) –

  If set to True, will only return the values of the attributes, skipping display_name, id, etc.

Returns:

• dict[str, dict[str, dict[str, Any | dict[str, Any]]]] –

  In case output_type is "dict", returns a dictionary in the format {object_model_name: {feature_name: {attribute_name: {attribute: value, ...}, ...}, ...} ...} If values_only is True, the innermost dictionary will be {attribute_name: value, ...}

• DataFrame –

  In case output_type is "DataFrame", returns a DataFrame with the following format: index = MultiIndex[object_model_name, feature_name, attribute_name], columns = [attribute, ...] If values_only is True, the columns will be ["attribute_value"]

Source code in echo_postgres/feature_definition_attributes.py

@validate_call
def get(
    self,
    feature_names: list[str] | None = None,
    object_names: list[str] | None = None,
    object_models: list[str] | None = None,
    data_source_types: list[str] | None = None,
    attribute_names: list[str] | None = None,
    attribute_values: dict[str, Any] | None = None,
    filter_type: Literal["and", "or"] = "and",
    output_type: Literal["dict", "DataFrame"] = "dict",
    values_only: bool = False,
) -> dict[str, dict[str, dict[str, Any | dict[str, Any]]]] | DataFrame:
    """Method to get the attributes of the  given features.

    The most useful keys/columns returned are:

    - data_source_type_name
    - attribute_name
    - attribute_value
    - data_type_name

    Parameters
    ----------
    feature_names : list[str] | None, optional
        Names of the features to get the attributes from. If set to None will get from all. By default None
    object_names : list[str] | None, optional
        List of object names to filter the results. If set, will add the models of these objects to the object_models filter.
        By default None
    object_models : list[str] | None, optional
        List of object model names to filter the results.
        By default None
    data_source_types : list[str] | None, optional
        List of data source type names to filter the results.
        By default None
    attribute_names : list[str] | None, optional
        List of attribute names to filter the results. If set to None will get all. By default None
    attribute_values : dict[str, Any] | None, optional
        Dictionary of attribute names and values to filter the results. It must be in the format {attribute_name: attribute_value, ...}.
        By default None
    filter_type : Literal["and", "or"], optional
        How to treat multiple filters. Can be one of ["and", "or"].
        By default "and"
    output_type : Literal["dict", "DataFrame"], optional
        Output type of the data. Can be one of ["dict", "DataFrame"]
        By default "dict"
    values_only : bool, optional
        If set to True, will only return the values of the attributes, skipping display_name, id, etc.

    Returns
    -------
    dict[str, dict[str, dict[str, Any | dict[str, Any]]]]
        In case output_type is "dict", returns a dictionary in the format {object_model_name: {feature_name: {attribute_name: {attribute: value, ...}, ...}, ...} ...}
        If values_only is True, the innermost dictionary will be {attribute_name: value, ...}
    DataFrame
        In case output_type is "DataFrame", returns a DataFrame with the following format: index = MultiIndex[object_model_name, feature_name, attribute_name], columns = [attribute, ...]
        If values_only is True, the columns will be ["attribute_value"]
    """
    # getting models of the objects
    if object_names:
        obj_models = (
            self._perfdb.objects.instances.get(object_names=object_names, output_type="DataFrame")["object_model_name"]
            .unique()
            .tolist()
        )
        object_models = list(set(object_models + obj_models)) if object_models else obj_models

    # checking if all attribute names are valid and also getting its ids
    if attribute_names or attribute_values:
        wanted_attrs = set()
        if attribute_names:
            wanted_attrs = set(attribute_names)
        if attribute_values:
            wanted_attrs = wanted_attrs.union(set(attribute_values.keys()))

        wanted_attrs = list(wanted_attrs)
        existing_attributes = self._perfdb.attributes.get_ids(attribute_names=wanted_attrs)
        if missing_attributes := set(wanted_attrs) - set(
            existing_attributes.keys(),
        ):
            raise ValueError(f"The following attribute names do not exist: {missing_attributes}")

    # building the WHERE clause
    where = []
    if feature_names:
        where.append(
            sql.SQL("feature_name IN ({names})").format(names=sql.SQL(", ").join(map(sql.Literal, feature_names))),
        )
    if object_models:
        where.append(sql.SQL("object_model_name IN ({names})").format(names=sql.SQL(", ").join(map(sql.Literal, object_models))))
    if data_source_types:
        where.append(
            sql.SQL("data_source_type_name IN ({names})").format(names=sql.SQL(", ").join(map(sql.Literal, data_source_types))),
        )
    if attribute_names:
        where.append(
            sql.SQL(" attribute_id IN ({attribute_ids}) ").format(
                attribute_ids=sql.SQL(",").join(sql.Literal(aid) for aid in existing_attributes.values()),
            ),
        )
    if attribute_values:
        attr_vals_query = [
            sql.SQL(
                " (attribute_id = {attribute_id} AND attribute_value = {attribute_value}::TEXT::JSONB) ",
            ).format(
                attribute_id=sql.Literal(existing_attributes[an]),
                attribute_value=sql.Literal(f'"{av}"') if isinstance(av, str) else sql.Literal(av),
            )
            for an, av in attribute_values.items()
        ]
        attr_vals_query = sql.SQL(f" {filter_type.upper()} ").join(attr_vals_query)
        where.append(attr_vals_query)
    if where:
        where = sql.SQL(f" {filter_type.upper()} ").join(where)

    # checking if all attribute names are valid and also getting its ids
    if attribute_names or attribute_values:
        wanted_attrs = set()
        if attribute_names:
            wanted_attrs = set(attribute_names)
        if attribute_values:
            wanted_attrs = wanted_attrs.union(set(attribute_values.keys()))

        wanted_attrs = list(wanted_attrs)
        existing_attributes = self._perfdb.attributes.get_ids(attribute_names=wanted_attrs)
        if missing_attributes := set(wanted_attrs) - set(
            existing_attributes.keys(),
        ):
            raise ValueError(f"The following attribute names do not exist: {missing_attributes}")

    # building the query
    query = [
        sql.SQL(
            "SELECT {values} FROM performance.v_feature_attributes",
        ).format(
            values=sql.SQL(
                "object_model_name, feature_name, attribute_name, attribute_value, data_type_name",
            )
            if values_only
            else sql.SQL("*"),
        ),
    ]
    if where:
        query.append(sql.SQL(" WHERE "))
        query.append(where)
    query.append(sql.SQL(" ORDER BY object_model_name, feature_name, attribute_name"))
    query = sql.Composed(query)

    # executing the query
    with self._perfdb.conn.reconnect() as conn:
        # setting attribute_value as object to avoid casting json column as string
        df = conn.read_to_pandas(query, post_convert="pyarrow")

    # casting the attribute values
    df = cast_attributes(df=df, index_cols=["object_model_name", "feature_name"])

    df = df.set_index(["object_model_name", "feature_name", "attribute_name"])

    # returning the result
    if output_type == "dict":
        # dropping unwanted columns
        if values_only:
            df = df["attribute_value"]
            temp = df.to_dict()
        else:
            temp = df[["attribute_id", "attribute_value", "data_type_id", "data_type_name", "modified_date"]].to_dict(orient="index")
        # converting from format {(object_model, feature_name, attribute_name): value} to {object_model: {feature_name: {attribute_name: value, ...}, ...}, ...}
        output = {}
        for (object_model, feature_name, attribute_name), value in temp.items():
            if object_model not in output:
                output[object_model] = {}
            if feature_name not in output[object_model]:
                output[object_model][feature_name] = {}
            output[object_model][feature_name][attribute_name] = value
        return output
    if output_type == "DataFrame" and values_only:
        df = df[["attribute_value"]].copy()

    return df