helixobs 0.5.0

Entity-centric observability client for scientific instrument pipelines

helixobs — Python client

Entity-centric observability for scientific instrument pipelines.

Track domain entities (raw observations, intermediate products, derived results) across disjoint asynchronous pipeline stages. The library attaches provenance links between entities so you can trace any scientific result back to the raw data that produced it.

Built on OpenTelemetry. Signals are standard OTLP — the HelixObs herald adds entity intelligence on top.

Installation

pip install helixobs

Requires Python 3.11+.

Quick start

from helixobs import setup
import logging

log = logging.getLogger("my.pipeline")

# One call wires both traces and logs with matching service name
tel = setup(
    "my-instrument.pipeline",
    instrument_id="MY_INSTRUMENT",
    endpoint="herald:4317",
    otlp=True,               # ship logs via OTLP to the collector (port 4319)
)

# Track an entity through a pipeline stage
with tel.create("detector", id=product_id, parents=[frame_id]) as token:
    result = process(item)
    log.info("item processed")           # helix_entity_id injected automatically
    if result.score > 0.95:
        token.add_event("helix.event.detection_confirmed", {"score": result.score})
# complete() called on exit; error() called on unhandled exception

Two OTLP endpoints:

Endpoint	Default port	Purpose
HelixObs Herald	4317	OTLP traces — entity spans
OTel Collector	4319	OTLP logs — structured log shipping

Set OTEL_EXPORTER_OTLP_ENDPOINT=http://<host>:4319 for log shipping, or pass log_endpoint to setup().

See USER_GUIDE.md for the full API reference, integration patterns, structured logging, and data model.

Development

pip install -e ".[dev]"
pytest