Data Probes

Data probes are read-only observers of the data channel. They compute a value from the input data and return it. Probes do not modify the data or the context directly.

In pipelines, probe results are typically written into the context by the node that wraps the probe, using a context_key. The probe itself remains completely context-agnostic.

What is a data probe?

A data probe is a subclass of semantiva.data_processors.data_processors.DataProbe. It:

  • Declares a single input_data_type class method.

  • Implements _process_logic(self, data, **params) -> Any.

  • Does not declare an output_data_type (SVA231 warns against it).

Example: mean of a float collection

from semantiva.data_processors.data_processors import DataProbe
from semantiva.examples.test_utils import FloatDataCollection, FloatDataType

class MeanProbe(DataProbe):
    """Compute the mean of a :class:`FloatDataCollection`."""

    @classmethod
    def input_data_type(cls):
        return FloatDataCollection

    def _process_logic(self, data: FloatDataCollection) -> float:
         values = [item.data for item in data]
         return sum(values) / max(len(values), 1)

probe = MeanProbe()
collection = FloatDataCollection(
    [FloatDataType(1.0), FloatDataType(2.0), FloatDataType(3.0), FloatDataType(4.0)]
)
mean_value = probe(collection)
print("mean:", mean_value)

mean: 2.5

Using probes in pipelines

Inside a pipeline, a node wraps a probe and takes care of writing its result into the context when context_key is provided. The probe does not see the context object at all.

For example, in YAML you would configure a node like:

pipeline:
  nodes:
    - processor: my.probes.MeanProbe
      context_key: "mean_value"

The node injects the probe result under "stats.mean" using a context observer. The probe remains a pure function of its input data and parameters.

(The full pipeline wiring and CLI usage is covered later in Pipelines in Semantiva.)

Design guidelines

  • Keep probes small and focused on single metrics or summaries.

  • Prefer returning plain Python values (floats, dicts, small dataclasses).

  • Never read or write the context directly from a probe.

  • Use semantiva dev lint to ensure the DataProbe contracts (SVA230–SVA232) are satisfied.

Next steps