atomscale.similarity.provider#

Provider classes for similarity trajectory data.

Note on unix-time units: the similarity API emits unix_times in seconds (raw DB values), unlike the metrology / optical / ellipsometry endpoints which ship unix_timestamp_ms in milliseconds. The column is renamed to UNIX Timestamp (unit-less display name); the magnitude-based fallback in align._infer_absolute_time() correctly detects seconds for these values. In typed Python surfaces the canonical type for individual unix_times values is decimal.Decimal; pandas columns stay numeric for performance.

Classes

SimilarityTrajectoryProvider()

class atomscale.similarity.provider.SimilarityTrajectoryProvider[source]

Bases: TimeseriesProvider[SimilarityTrajectoryResult]

fetch_raw(client: BaseClient, data_id: str, **kwargs: Any) → Any[source]

Fetch similarity trajectory data from the API.

Parameters:

  • client (BaseClient) – The API client.

  • data_id (str) – The source ID for the similarity query.

  • **kwargs (Any) – Must include ‘workflow’ (required). Optional parameters: window_span, reference_ids, softmax_mode, reference_n_values.

Return type:

Any

Returns:

Raw API response payload.

Raises:

KeyError – If ‘workflow’ is not provided in kwargs.

to_dataframe(raw: Any) → DataFrame[source]

Convert raw payload to a tidy DataFrame with domain-specific renames/index.

Parameters:

raw (Any)

Return type:

DataFrame

build_result(client: BaseClient, data_id: str, data_type: str, ts_df: DataFrame, *, workflow: str = '', window_span: float = 0.0, source_data_ids: Sequence[UUID | str] | None = None) → SimilarityTrajectoryResult[source]

Build time series result object

Parameters:

  • client (BaseClient)

  • data_id (str)

  • data_type (str)

  • ts_df (DataFrame)

  • workflow (str)

  • window_span (float)

  • source_data_ids (Sequence[UUID | str] | None)

Return type:

SimilarityTrajectoryResult