我有5個(gè)類，如下所示。執(zhí)行從類 5 開(kāi)始。在這里，依賴注入被用作偵聽(tīng)器對(duì)象傳遞給類的構(gòu)造函數(shù)，然后該對(duì)象調(diào)用 listen（） 方法。這部分很好。Bot麻煩從現(xiàn)在開(kāi)始。從方法中，我調(diào)用另一個(gè)調(diào)用的方法，該方法位于類（ 文件中）中。然后，該方法從 2 個(gè)不同的類和 中調(diào)用另外兩個(gè)方法。listen()process()Processorprocessor.pyprocess()TokenizerCore如您所見(jiàn)，所有這些都是耦合的，此處不使用依賴關(guān)系注入。我不知道該怎么做。我已經(jīng)嘗試了python包和包的示例，但據(jù)我所知，這些示例與單級(jí)依賴關(guān)系有關(guān)，而不是這樣的情況。dependency-injectorpinject我曾想過(guò)在一個(gè)模塊中創(chuàng)建所有必需的對(duì)象，并在需要時(shí)將其作為參數(shù)傳遞給所有類。但這似乎不是一個(gè)好習(xí)慣。我能做些什么來(lái)解耦上面的代碼？我應(yīng)該如何進(jìn)行模塊化？# core.pyclass Core: def decide(self, data): """ Makes decisions based on input data """ # blah blah-# tokenizer.pyimport nltkfrom nltk.corpus import stopwordsfrom nltk.tokenize import word_tokenizestoplist = set(stopwords.words('english'))class Tokenizer: def tokenize(self, query): """ Tokenize the input query string """ tokenized_query = word_tokenize(query) clean_query = [w for w in tokenized_query if not w in stoplist] return clean_query-# processor.pyimport tokenizerimport coreclass Processor: def process(self, query): """ Send the input query to tokenizer tokenized input is send to the Core module for decision making """ tokenized_input = tokenizer.Tokenizer().tokenize(query) core.Core().decide(tokenized_input)-# listener.pyimport processorclass Listener: def listen(self): """ Continuosly listens to user input """ while True: query=input() processor.Processor().process(query)-# bot.pyimport listenerclass Bot: def __init__(self, listener): listener.listen()Bot(listener.Listener())
2 回答
ABOUTYOU
TA貢獻(xiàn)1812條經(jīng)驗(yàn) 獲得超5個(gè)贊
我認(rèn)為你可以重新考慮這個(gè)設(shè)計(jì)，記住兩個(gè)概念：
Python 多重繼承來(lái)設(shè)計(jì)你的類。
使用 super（） 和 Python 的方法解析順序 （MRO） 在測(cè)試代碼中注入類依賴項(xiàng)的模擬。
關(guān)于第一點(diǎn)，您的類將如下所示：
# core.py
class Core:
def decide(self, data):
""" Makes decisions based on input data """
# blah blah
# tokenizer.py
class Tokenizer:
def tokenize(self, query):
""" Tokenize the input query string """
return query
# processor.py
# from tokenizer import Tokenizer
# from core import Core
class Processor(Core, Tokenizer):
def process(self, query):
"""
Send the input query to tokenizer
tokenized input is send to the Core module for decision making
"""
tokenized_input = super().tokenize(query)
super().decide(tokenized_input)
# listener.py
# from processor import Processor
class Listener(Processor):
def listen(self):
""" Continuosly listens to user input """
while True:
query=input()
super().process(query)
# bot.py
#from listener import Listener
class Bot(Listener):
def start_listener(self):
super().listen()
Bot().start_listener()
通過(guò)這種OO設(shè)計(jì)和super（）的使用，我們可以利用MRO在類的依賴項(xiàng)中注入模擬，我將展示如何在SUT（被測(cè)主題）中為其依賴項(xiàng)注入模擬。
機(jī)器人和處理器的示例：
class MockCore(Core):
def decide(self, data):
""" Here you can implement the behavior of the mock """
class MockTokenizer(Tokenizer):
def tokenize(self, query):
""" Here you can implement the behavior of the mock """
return query
class ProcessorSut(Processor, MockCore, MockTokenizer):
'Here we are injecting mocks for Processor dependencies'
class Bot(Listener):
def start_listener(self):
super().listen()
class MockListener(Listener):
def listen(self):
""" Here you can implement the behavior of the mock """
return
class BotSut(Bot, MockListener):
'Here we are injecting a mock for the Listener dependency of Bot'
看到我們的SUT類的MRO，我們可以理解為什么多重繼承和super（）的使用允許我們以這種方式注入模擬。
機(jī)器人和處理器解決方案的結(jié)果 MRO：
Help on class BotSut in module __main__:
class BotSut(Bot, MockListener)
| Here we are injecting a mock for the Listener dependency of Bot
|
| Method resolution order:
| BotSut
| Bot
| MockListener
| Listener
| Processor
| Core
| Tokenizer
| builtins.object
...
...
Help on class ProcessorSut in module __main__:
class ProcessorSut(Processor, MockCore, MockTokenizer)
| Here we are injecting mocks for Processor dependencies
|
| Method resolution order:
| ProcessorSut
| Processor
| MockCore
| Core
| MockTokenizer
| Tokenizer
| builtins.object
楊__羊羊
TA貢獻(xiàn)1943條經(jīng)驗(yàn) 獲得超7個(gè)贊
您可以將實(shí)用程序類設(shè)置為類屬性：
class Core:
def decide(self, data):
return False
class Tokenizer:
def tokenize(self, query):
return [] # ...
class Processor:
tokenizer_class = Tokenizer
core_class = Core
def process(self, query):
tokenized_input = self.tokenizer_class().tokenize(query)
return self.core_class().decide(tokenized_input)
class Listener:
processor_class = Processor
def listen(self):
while True:
query = input()
self.processor_class().process(query)
然后，您可以使用測(cè)試框架的模擬/補(bǔ)丁功能，例如pytest的：monkeypatch
def test_foo(monkeypatch):
monkeypatch.setattr(Processor, 'tokenizer_class', FooTokenizer)
# etc...
添加回答
舉報(bào)
0/150
提交
取消