infer

infer ¤

Infer :class:~ts_shape.signals.spec.SignalSpec from data.

Inference is a starting point, not an oracle: every inferred field records why it was chosen in spec.notes, and anything the data cannot tell (units, a counter that never reset yet, the real outage threshold of a change-encoded signal) should be set with :meth:SignalSpecs.override.

parse_times ¤

parse_times(values: Series) -> pd.Series

A time column as datetime64; unparseable values become NaT.

Text with different UTC offsets (+01:00 next to +02:00) is converted to UTC with a :class:~ts_shape.errors.DataQualityWarning.

Parameters:

Name	Type	Description	Default
`values`	`Series`	The time column.	required

Returns:

Type	Description
`Series`	The parsed column.

signal_groups ¤

signal_groups(
    dataframe: DataFrame, uuid_column: str, time_column: str
) -> list[tuple[str, pd.DataFrame]]

Rows per signal, keyed by the uuid as text, sorted by key.

Nullable, pyarrow and string columns are converted to NumPy dtypes first (the input is not changed). Rows without a uuid are dropped and uuids that differ only in type (1 and "1") are one signal; both cases emit a :class:~ts_shape.errors.DataQualityWarning.

Parameters:

Name	Type	Description	Default
`dataframe`	`DataFrame`	Long-format frame.	required
`uuid_column`	`str`	Signal identifier column.	required
`time_column`	`str`	Timestamp column.	required

Returns:

Type	Description
`list[tuple[str, DataFrame]]`	`(uuid, rows)` pairs.

Raises:

Type	Description
`ColumnNotFoundError`	When `uuid_column` or `time_column` is missing.

infer_spec ¤

infer_spec(
    group: DataFrame,
    uuid: str,
    time_column: str = "systime",
) -> SignalSpec

Infer the spec of one signal from its rows.

Parameters:

Name	Type	Description	Default
`group`	`DataFrame`	The rows of one signal.	required
`uuid`	`str`	The signal's uuid, written to the spec.	required
`time_column`	`str`	Timestamp column.	`'systime'`

Returns:

Type	Description
`SignalSpec`	The inferred :class:`SignalSpec`; `notes` explain each field.

Raises:

Type	Description
`ColumnNotFoundError`	When `time_column` is missing.

infer_specs ¤

infer_specs(
    dataframe: DataFrame,
    overrides: (
        Mapping[str, Mapping[str, Any]] | None
    ) = None,
    *,
    uuid_column: str = "uuid",
    time_column: str = "systime"
) -> SignalSpecs

Infer a :class:SignalSpec for every uuid in a long-format frame.

Parameters:

Name	Type	Description	Default
`dataframe`	`DataFrame`	Long-format data (`systime`, `uuid`, `value_*`).	required
`overrides`	`Mapping[str, Mapping[str, Any]] \| None`	`{uuid: {field: value}}` applied after inference -- the place for what the data cannot tell (units, outage limits).	`None`
`uuid_column`	`str`	Column identifying the signal.	`'uuid'`
`time_column`	`str`	Timestamp column.	`'systime'`

Returns:

Type	Description
`SignalSpecs`	class:`SignalSpecs` keyed by uuid (as text); each spec's `notes`
`SignalSpecs`	explain the inferred fields.

Raises:

Type	Description
`ColumnNotFoundError`	When `uuid_column` or `time_column` is missing.