Skip to content

MorvanZhou/whenact

Repository files navigation

WhenAct

WhenAct is a module that defines a decision flow. A 中文文档 is available to.

The executing flow looks like:

decision1: [when0] > [action0]
decision2: [when1] > [action10 > action11]
decision3: [when20 > when21] > [action2]
decision4: [when31 > when32] > [action3]

When all when in one decision is satisfied, than it runs to it's following action.

This decision process has two modes,

  1. auto_break=True, when runs into the first when=True, this flow will finish after it's following act.
  2. auto_break=False, the flow will continue flow even when when=False.

For example:

# auto_break = True
decision1: [when0=True] > [action0]
decision2: X
decision3: X
decision4: X
# auto_break = False
decision1: [when0=True] > [action0]
decision2: [when1=False] > X
decision3: [when20=True > when21=True] > [action2]
decision4: [when31=False > X] > X

Install

pip install whenact

Create WhenAct decision flow

Simpling use whenact.add() to add new decision to the flow process.

import whenact

def w1(ctx):
    return True

def a1(ctx):
    return "done"


whenact.add(when=w1, act=a1)

whenact.print_flow()
# p0: [w1] > [a1]

hist = whenact.run()
assert hist.last_output == "done"
assert hist.outputs == ["done"]

More complex pipeline can include more than one decision. Using whenact.add() to accumulate decisions.

import whenact


def w_false(ctx):
    return False


def w_true(ctx):
    return True


def a1(ctx):
    return 1


def a2(ctx):
    return 2


def a3(ctx):
    return 3


whenact.add(when=w_false, act=a1)
whenact.add(when=w_true, act=[a2, a3])

whenact.print_flow()
# p0: [w1] > [a1]
# p1: [w2] > [a2 > a3]

hist = whenact.run()
print(hist.summary)
# [p1: w1, p2: w2 > a2 > a3]

assert hist.first_output == 2
assert hist.last_output == 3
assert hist.outputs == [2, 3]

The context(ctx) in each when and act function passes context information from outside. You can store external information in context, then pass it to the flow. Moreover, values can be set to the context when inside those functions, then be carried out once the flow is finished.

import whenact


def w_false(ctx):
    return False


def w_true(ctx):
    return True


def a1(ctx):
    ctx["action"] = "a1 action"


def a2(ctx):
    ctx["action"] = "a2 action"


def a3(ctx):
    ctx["action"] += " with a3"


whenact.add(when=w_false, act=a1)
whenact.add(when=w_true, act=[a2, a3])

whenact.print_flow()


# p0: [w1] > [a1]
# p1: [w2] > [a2 > a3]

class TestContext(whenact.BaseContext):
    pass


ctx = TestContext()

hist = whenact.run(ctx)
print(hist.summary)
# [p1: w1, p2: w2 > a2 > a3]

assert hist.last_output is None
assert hist.outputs == [None, None]
assert ctx["action"] == "a2 action with a3"

There is another way to set a decision flow.

import whenact


def w_false(ctx):
    return False


def w_true(ctx):
    return True


def a1(ctx):
    return 1


def a2(ctx):
    return 2


flow = whenact.DecisionFlow([
    whenact.Decision(when=[w_false], act=[a1], name="D1"),
    whenact.Decision(when=[w_true], act=[a2], name="D2"),
]
)
print(flow)
# D1: [w_false] > [a1]
# D2: [w_true] > [a2]

hist = flow.run()
assert hist.first_output == 2

More examples

More examples can be found in tests