Skip to content

Context

A Context is the payload your flow operates on. It carries params (immutable inputs) plus any mutable or computed state the flow needs as it progresses.

The two halves

Params

Params are the inputs handed to the flow. They are frozen dataclasses extending BaseParams.

from dataclasses import dataclass
from pyuow import BaseParams


@dataclass(frozen=True)
class OrderParams(BaseParams):
    customer_id: str
    sku: str
    quantity: int

Context

BaseContext[PARAMS] is a Protocol with a single attribute: params: PARAMS. Any class with a params field that matches structurally satisfies it.

Two convenience bases are provided:

  • BaseMutableContext — a mutable dataclass with one safety net: re-assigning an attribute that already exists raises AttributeCannotBeOverriddenError. Good for flows that compute and store intermediate state.
  • BaseImmutableContext — a frozen=True dataclass. Good for pure pipelines.
from dataclasses import dataclass
from pyuow import BaseContext, BaseMutableContext, BaseImmutableContext


@dataclass
class OrderContext(BaseContext[OrderParams]):
    params: OrderParams
    in_stock: bool = False


@dataclass
class MutableCtx(BaseMutableContext[OrderParams]):
    pricing: Pricing | None = None


@dataclass(frozen=True)
class ImmutableCtx(BaseImmutableContext[OrderParams]):
    fingerprint: str

Mutable safety net

BaseMutableContext allows new attributes to be set but blocks overwrites:

ctx = MutableCtx(params=OrderParams(...))
ctx.pricing = compute_pricing(ctx)   # OK - first assignment
ctx.pricing = something_else          # raises AttributeCannotBeOverriddenError

This catches an entire class of bugs where two units overwrite each other's contributions.

Specialised contexts

PyUoW ships with two context flavours layered on top of BaseContext:

  • BaseDomainContext — adds a Batch field, used with DomainTransactionalWorkManager to flush domain changes after a flow runs.
  • InMemoryDataPointContext — a context with a built-in DataPoint producer/consumer registry. Pair it with ProducesDataPoints and ConsumesDataPoints units.

You can compose them — BaseDomainContext and InMemoryDataPointContext are designed to be mixed into your own context class.

Reference

  • pyuow.context — BaseContext, BaseParams, mutable / immutable bases