如何使用转换库管理python fsm中的状态转换

时间:2017-02-16 21:09:40

标签: python transitions fsm

我正在尝试使用transitions模块实现状态机。 Python版本2.7.13和过渡版本是0.4.4版本。

在项目文档中,示例都通过在命令提示符下键入函数调用来进行状态。从转换文档中的第一个示例获取片段,wake_up的状态是通过调用命名函数work_out>>> batman.wake_up() >>> batman.work_out() >>> batman.state 'hungry'

来实现的
next_state

我希望状态机能够自动前进通过模型数据的状态。下面的玩具示例是我想要做的但是依赖于使用虚拟函数作为设置next_state的指针。

有没有办法做到这一点,不涉及创建from transitions import Machine class AModel(object): def __init__(self): self.sv = 0 # state variable of the model def on_enter_sA(self): print "Entered sA" self.next_state = self.to_sB def on_enter_sB(self): print "Entered sB" if self.sv < 3: self.next_state = self.to_sB else: self.next_state = self.to_sC def on_enter_sC(self): print "Entered sC" if self.sv == 6: self.next_state = self.to_sD def on_enter_sD(self): print "Entered sD" self.next_state = self.to_sA def next_state(self): pass #setup model and state machine model = AModel() #init transitions model list_of_states = ['sA','sB','sC','sD'] transitions = [ {'trigger':'to_sA','source':'sD','dest':'sA'}, {'trigger':'to_sB','source':'sA','dest':'sB'}, {'trigger':'to_sC','source':'sB','dest':'sC'}, {'trigger':'to_sD','source':'sC','dest':'sD'} ] machine = Machine(model=model, states=list_of_states, initial='sA', transitions=transitions) model.next_state = model.to_sB #init next state pointer #begin main for i in range(0,8): print 'iter is: ' + str(i) + " -model state is:" + model.state model.sv = i #update model state variable, local state logic #will determine what next_state points to model.next_state() 函数并像指针一样使用它?转换文档具有Ordered转换和条件转换但我真正想要的是具有条件有序转换。

是否可以在不使用函数指针的情况下重写下面的代码?

{{1}}

谢谢!

2 个答案:

答案 0 :(得分:2)

之前已请求此功能(请参阅this issue)。正如您所看到的,还有人正在研究它。他可能会在不久的将来提出拉动请求。我没有回顾他的变化,但一旦发生这种变化肯定会做。

现在,您可以让模型处理条件检查并将其与有序转换相结合,以省去频繁更新next_state函数指针的需要。由于您只是检查索引,因此可能如下所示:

from transitions import Machine


class AModel(object):
    def __init__(self):
        self.sv = 0  # state variable of the model
        self.conditions = {  # each state 
            'sA': 0,
            'sB': 3,
            'sC': 6,
            'sD': 0,
        }

    def poll(self):
        if self.sv >= self.conditions[self.state]:
            self.next_state()


# setup model and state machine
model = AModel()

# init transitions model
list_of_states = ['sA', 'sB', 'sC', 'sD']
machine = Machine(model=model, states=list_of_states, initial='sA', ordered_transitions=True)

# begin main
for i in range(0, 10):
    print('iter is: ' + str(i) + " -model state is:" +  model.state)
    model.sv = i
    model.poll()

我假设每次索引增加时都会轮询模型。如果是这种情况,self.sv == 6self.sv >= 6执行相同的操作(适用于sCsD)。 但是,如果故意选择运算符,则可以更改模型条件检查以使用运算符值元组:

from transitions import Machine
import operator


class AModel(object):
    def __init__(self):
        self.sv = 0  # state variable of the model
        self.conditions = {  # each state
            'sA': (operator.ne, None),
            'sB': (operator.ge, 3),
            'sC': (operator.eq, 6),
            'sD': (operator.ne, None),
        }

    def poll(self):
        op, value = self.conditions[self.state]
        if op(self.sv, value):
            self.next_state()


# setup model and state machine
model = AModel()

# init transitions model
list_of_states = ['sA', 'sB', 'sC', 'sD']
machine = Machine(model=model, states=list_of_states, initial='sA', ordered_transitions=True)

# begin main
for i in range(0, 10):
    print('iter is: ' + str(i) + " -model state is:" + model.state)
    model.sv = i
    model.poll()

在这两种情况下,输出为:

iter is: 0 -model state is:sA
iter is: 1 -model state is:sB
iter is: 2 -model state is:sB
iter is: 3 -model state is:sB
iter is: 4 -model state is:sC
iter is: 5 -model state is:sC
iter is: 6 -model state is:sC
iter is: 7 -model state is:sD
iter is: 8 -model state is:sA
iter is: 9 -model state is:sB

但是,我再次假设可能出错的事情:我认为如果满足条件就改变状态就足够了。这就是条件的运作方式。但也许你每次轮询你的模型时都打算退出并输入状态。在这种情况下,您可以使用auto_transitions并使用getattr动态检索它们:

from transitions import Machine


class AModel(object):
    def __init__(self):
        self.sv = 0  # state variable of the model
        self.conditions = {  # each state
            'sA': 0,
            'sB': 3,
            'sC': 6,
            'sD': 0,
        }

    def poll(self):
        if self.sv >= self.conditions[self.state]:
            self.next_state()  # go to next state
        else:
            getattr(self, 'to_%s' % self.state)()  # enter current state again

    def on_enter(self):
        print('entered state %s' % self.state)

    def on_exit(self):
        print('exited state %s' % self.state)


# setup model and state machine
model = AModel()

# init transitions model
list_of_states = ['sA', 'sB', 'sC', 'sD']
machine = Machine(model=model, states=list_of_states, initial='sA',
                  ordered_transitions=True, before_state_change='on_exit',
                  after_state_change='on_enter')

# begin main
for i in range(0, 10):
    print('iter is: ' + str(i) + " -model state is:" +  model.state)
    model.sv = i
    model.poll()

为了简单起见,我添加了每次进入或退出状态时打印消息的功能。如果您使用记录器,则不需要这样做,因为transitions也会记录这些事件。

答案 1 :(得分:0)

我还没想出如何将它添加到过渡模块,但是我根据条件自动转换:

class StateMachine:
    def __init__(self, name, states, transitions, initialState):
        self.name=name
        self.states=set()
        self.transitions=set()
        self.state=None

        for s in states:
            _s=State(s)
            self.states.add(_s)
            if self.state==None:
                self.state=_s
            elif s==initialState:
                self.state=_s

        for t in transitions:
            # self.addTransition(t)

    def addTransition(self, transition):
        name=transition[0]
        for s in self.states:
            #fromState
            if s.name==transition[1]:
                fromState=s
            #toState
            elif s.name==transition[2]:
                toState=s
        #condition
        condition=getattr(self, transition[3])
        #action     
        if len(transition)==5:
            action = getattr(self, transition[4])()
        else:
            action=self.passAction
        t=Transition(name, fromState, toState, condition, action)
        fromState.transitions.add(t)

    def passAction(self):
        print('pass!!!')

    def run(self):
        self.state=self.state.testTransitions()

    def __repr__(self):
        return self.name+':'+str(self.states)

class State:
    def __init__(self, name='state'):
        self.name=name
        self.transitions=set()

    def __repr__(self):
        return self.name+':'+str(self.transitions)

    def testTransitions(self):
        state=self
        for t in self.transitions:
            if t.condition() is True:
                t.action()
                state=t.toState
        return state

class Transition:
    def __init__(self, name, fromState, toState, condition, action): 
        self.name=name
        self.fromState=fromState
        self.toState=toState
        self.condition=condition
        self.action=action

    def __repr__(self):
        return self.name

class TestSM(StateMachine):
    def __init__(self, name, states, transitions, initialState):
        StateMachine.__init__(self, name, states, transitions, initialState)

    def testCondition(self):
        print('testCondition!!!')
        return True



states=['a', 'b', 'c']
transitions=[
    ['aTOb', 'a', 'b', 'testCondition'],
    ['bTOc', 'b', 'c', 'testCondition'],
    ['cTOa', 'c', 'a', 'testCondition'],
    ]
sm=TestSM('testSM', states, transitions, 'a')

for s in sm.states:
    print(s.name)

print('fin')        

为了运行状态机,只需运行'run'函数,例如:

sm.run()