TabularLoader

loader.tabular.TabularLoader(unit=None, number_type=None, **kwargs)

Vectorized base class for loading tabular data with column/field specs.

TabularLoader provides a ZERO ROW ITERATION framework for loading CSV, TSV, and other delimited formats into EventData. All operations use vectorized numpy / pandas / pyarrow operations.

Configuration Attributes

delimiter: Field delimiter character (default: ,). header_row: Row index containing column headers (default: 0). id_column: Source column name for event IDs (auto-generated if None). name_column: Source column name for event names (optional). start_column: Source column name for start coordinate (required). end_column: Source column name for end coordinate (None for instant events). duration_column: Source column name for duration (alternative to end_column). event_type_column: Source column name for event type (optional). default_event_type: Default event type if no column provides one. column_specs: Mapping or sequence describing the per-column translation from source to :class:DataField. See :mod:timetoalign.loader.tabular.field_parsers. field_specs: Optional sequence/mapping of blueprint-mode :class:SemanticField instances that promote raw fields to paired semantic fields. coordinate_unit: TimeUnit for coordinate values. coordinate_type: NumberType for coordinate parsing.

Examples

>>> from timetoalign.core import IntField
>>> class MyLoader(TabularLoader):
...     delimiter = "\t"
...     start_column = "onset"
...     end_column = "offset"
...     coordinate_unit = TimeUnit.seconds
...     column_specs = {"velocity": IntField(name="velocity"), "channel": int}

Attributes

Name	Description
table	The original source data as a faithful PyArrow table.