Custom SPARQL functions¶

You can extend the KB with custom Python functions callable directly from SPARQL queries using the fn: prefix.

How it works¶

1. Place Python files in your bringup package's functions/ directory
2. Use the @kb_function decorator to register functions
3. At configure time, TriplestarKBNode discovers and registers all functions
4. They become available as fn:functionName(...) in SPARQL

Example¶

from triplestar_core.functions import kb_function

@kb_function("hello")
def hello(name: str) -> str:
    return f"Hello, {name}!"


@kb_function("add")
def add(a: float, b: float) -> float:
    return a + b

Called in SPARQL:

PREFIX fn: <http://triplestar.local/functions/>
SELECT ?greeting ?sum WHERE {
  BIND(fn:hello("World") AS ?greeting)
  BIND(fn:add(1, 2) AS ?sum)
}

Type conversions¶

The function registry handles type conversion automatically:

Return values are converted back to RDF literals using the same ROS → RDF conversion table.

The @kb_function decorator¶

from triplestar_core.functions import kb_function

# Registered with the function's name
@kb_function()
def my_function(...): ...

# Or with an explicit name
@kb_function("customName")
def my_function(...): ...

Functions are registered into a global FunctionRegistry instance (triplestar_core.functions.registry). The node iterates this registry during configuration and adds each function to the KB via kb.add_kb_function().