Decorator Reference

@ipy_register_func

from access_py_telemetry.decorators import ipy_register_func

Use this decorator for functions intended to be called primarily from inside a Jupyter notebook or IPython shell. It registers the function name with the service and configures ApiHandler, but telemetry is dispatched via IPython’s pre_run_cell event rather than from inside the wrapped function itself.

@ipy_register_func("mypackage_run")
def run(config):
    ...

Parameters

Parameter	Type	Default	Description
service	str	required	The service name (e.g. "intake_catalog"). Must match a key built from your config.yaml.
extra_fields	dict[str, Any] | None	None	Additional fields to attach to every telemetry record for this service.
pop_fields	Iterable[str] | None	None	Default fields to remove from the telemetry record.

With extra fields and pop_fields

@ipy_register_func(
    "mypackage_run",
    extra_fields={"config_version": "1.2"},
    pop_fields=["session_id"],
)
def run(config):
    ...

---

@register_func

from access_py_telemetry.decorators import register_func

Use this decorator for functions called outside Jupyter (e.g. in scripts or CLI tools). Telemetry is sent inside the function wrapper using send_in_loop, which detects whether an asyncio event loop is running and dispatches accordingly.

@register_func("mypackage_run")
def run(config):
    ...

Parameters

Same as @ipy_register_func.

---

Adding extra fields after decoration

Extra fields can be added at any point after decoration (e.g. once data is available at import time):

from access_py_telemetry.api import ApiHandler

ApiHandler().add_extra_fields("mypackage_run", {"cluster_name": cluster.name})

This is useful when the values you want to capture are not known at function definition time, but are available when the module is imported.

---

Removing default fields

To strip a field from the default telemetry payload:

from access_py_telemetry.api import ApiHandler

ApiHandler().remove_fields("mypackage_run", ["session_id"])

Default fields included in every record: name, function, args, kwargs, session_id, timestamp.

---

Checking what’s registered

from access_py_telemetry.registry import TelemetryRegister

registry = TelemetryRegister("intake_catalog")
print(registry)
# ["esm_datastore.search", "DfFileCatalog.search", "DfFileCatalog.__getitem__"]

---

Deregistering functions

registry = TelemetryRegister("mypackage_run")
registry.deregister("run")

# or multiple at once:
registry.deregister("run", "submit")