Tortoise-orm信號(hào)實(shí)現(xiàn)及使用場景源碼詳解
場景
在使用Tortoise操作數(shù)據(jù)庫的時(shí)候發(fā)現(xiàn),通過對(duì)操作數(shù)據(jù)庫模型加以裝飾器,如@pre_save(Model),可以實(shí)現(xiàn)對(duì)這個(gè)模型在savue時(shí),自動(dòng)調(diào)用被裝飾的方法,從而實(shí)現(xiàn)對(duì)模型的一些操作。
在此先從官方文檔入手,看一下官方的對(duì)于模型信號(hào)的Example
# -*- coding: utf-8 -*-
"""
This example demonstrates model signals usage
"""
from typing import List, Optional, Type
from tortoise import BaseDBAsyncClient, Tortoise, fields, run_async
from tortoise.models import Model
from tortoise.signals import post_delete, post_save, pre_delete, pre_save
class Signal(Model):
id = fields.IntField(pk=True)
name = fields.TextField()
class Meta:
table = "signal"
def __str__(self):
return self.name
@pre_save(Signal)
async def signal_pre_save(
sender: "Type[Signal]", instance: Signal, using_db, update_fields
) -> None:
print('signal_pre_save', sender, instance, using_db, update_fields)
@post_save(Signal)
async def signal_post_save(
sender: "Type[Signal]",
instance: Signal,
created: bool,
using_db: "Optional[BaseDBAsyncClient]",
update_fields: List[str],
) -> None:
print('post_save', sender, instance, using_db, created, update_fields)
@pre_delete(Signal)
async def signal_pre_delete(
sender: "Type[Signal]", instance: Signal, using_db: "Optional[BaseDBAsyncClient]"
) -> None:
print('pre_delete', sender, instance, using_db)
@post_delete(Signal)
async def signal_post_delete(
sender: "Type[Signal]", instance: Signal, using_db: "Optional[BaseDBAsyncClient]"
) -> None:
print('post_delete', sender, instance, using_db)
async def run():
await Tortoise.init(db_url="sqlite://:memory:", modules={"models": ["__main__"]})
await Tortoise.generate_schemas()
# pre_save,post_save will be send
signal = await Signal.create(name="Signal")
signal.name = "Signal_Save"
# pre_save,post_save will be send
await signal.save(update_fields=["name"])
# pre_delete,post_delete will be send
await signal.delete()
if __name__ == "__main__":
run_async(run())
以上代碼可直接復(fù)制后運(yùn)行,運(yùn)行后的結(jié)果:
signal_pre_save <class '__main__.Signal'> Signal <tortoise.backends.sqlite.client.SqliteClient object at 0x7f8518319400> None
post_save <class '__main__.Signal'> Signal <tortoise.backends.sqlite.client.SqliteClient object at 0x7f8518319400> True None
signal_pre_save <class '__main__.Signal'> Signal_Save <tortoise.backends.sqlite.client.SqliteClient object at 0x7f8518319400> ['name']
post_save <class '__main__.Signal'> Signal_Save <tortoise.backends.sqlite.client.SqliteClient object at 0x7f8518319400> False ['name']
pre_delete <class '__main__.Signal'> Signal_Save <tortoise.backends.sqlite.client.SqliteClient object at 0x7f8518319400>
post_delete <class '__main__.Signal'> Signal_Save <tortoise.backends.sqlite.client.SqliteClient object at 0x7f8518319400>
可以發(fā)現(xiàn),對(duì)模型進(jìn)行保存和刪除時(shí)候,都會(huì)調(diào)用對(duì)應(yīng)的信號(hào)方法。
源碼
從導(dǎo)包可以得知,tortoise的所有信號(hào)方法都在tortoise.signals中。
from enum import Enum
from typing import Callable
Signals = Enum("Signals", ["pre_save", "post_save", "pre_delete", "post_delete"])
def post_save(*senders) -> Callable:
"""
Register given models post_save signal.
:param senders: Model class
"""
def decorator(f):
for sender in senders:
sender.register_listener(Signals.post_save, f)
return f
return decorator
def pre_save(*senders) -> Callable:
...
def pre_delete(*senders) -> Callable:
...
def post_delete(*senders) -> Callable:
...
其內(nèi)部實(shí)現(xiàn)的四個(gè)信號(hào)方法分別是模型的保存后,保存前,刪除前,刪除后。
其內(nèi)部裝飾器代碼也十分簡單,就是對(duì)裝飾器中的參數(shù)(也就是模型),注冊一個(gè)監(jiān)聽者,而這個(gè)監(jiān)聽者,其實(shí)就是被裝飾的方法。
如上面的官方示例中:
# 給模型Signal注冊一個(gè)監(jiān)聽者,它是方法signal_pre_save
@pre_save(Signal)
async def signal_pre_save(
sender: "Type[Signal]", instance: Signal, using_db, update_fields
) -> None:
print('signal_pre_save', sender, instance, using_db, update_fields)
而到了Model類中,自然就有一個(gè)register_listener方法,定睛一看,上面示例Signal中并沒有register_listener方法,所以自然就想到了,這個(gè)方法必定在父類Model中。
class Model:
...
@classmethod
def register_listener(cls, signal: Signals, listener: Callable):
...
if not callable(listener):
raise ConfigurationError("Signal listener must be callable!")
# 檢測是否已經(jīng)注冊過
cls_listeners = cls._listeners.get(signal).setdefault(cls, []) # type:ignore
if listener not in cls_listeners:
# 注冊監(jiān)聽者
cls_listeners.append(listener)
接下來注冊后,這個(gè)listeners就會(huì)一直跟著這個(gè)Signal類。只需要在需要操作關(guān)鍵代碼的地方,進(jìn)行調(diào)用即可。
看看在模型save的時(shí)候,都干了什么?
async def save(
self,
using_db: Optional[BaseDBAsyncClient] = None,
update_fields: Optional[Iterable[str]] = None,
force_create: bool = False,
force_update: bool = False,
) -> None:
...
# 執(zhí)行保存前的信號(hào)
await self._pre_save(db, update_fields)
if force_create:
await executor.execute_insert(self)
created = True
elif force_update:
rows = await executor.execute_update(self, update_fields)
if rows == 0:
raise IntegrityError(f"Can't update object that doesn't exist. PK: {self.pk}")
created = False
else:
if self._saved_in_db or update_fields:
if self.pk is None:
await executor.execute_insert(self)
created = True
else:
await executor.execute_update(self, update_fields)
created = False
else:
# TODO: Do a merge/upsert operation here instead. Let the executor determine an optimal strategy for each DB engine.
await executor.execute_insert(self)
created = True
self._saved_in_db = True
# 執(zhí)行保存后的信號(hào)
await self._post_save(db, created, update_fields)
拋開其他代碼,可以看到,在模型save的時(shí)候,其實(shí)是先執(zhí)行保存前的信號(hào),然后執(zhí)行保存后的信號(hào)。
自己實(shí)現(xiàn)一個(gè)信號(hào)
有了以上的經(jīng)驗(yàn),可以自己實(shí)現(xiàn)一個(gè)信號(hào),比如我打算做個(gè)數(shù)據(jù)處理器的類,我想在這個(gè)處理器工作中,監(jiān)聽處理前/后的信號(hào)。
# -*- coding: utf-8 -*-
from enum import Enum
from typing import Callable, Dict
# 聲明枚舉信號(hào)量
Signals = Enum("Signals", ["before_process", "after_process"])
# 處理前的裝飾器
def before_process(*senders):
def decorator(f):
for sender in senders:
sender.register_listener(Signals.before_process, f)
return f
return decorator
# 處理后的裝飾器
def after_process(*senders):
def decorator(f):
for sender in senders:
sender.register_listener(Signals.after_process, f)
return f
return decorator
class Model(object):
_listeners: Dict = {
Signals.before_process: {},
Signals.after_process: {}
}
@classmethod
def register_listener(cls, signal: Signals, listener: Callable):
"""注冊監(jiān)聽者"""
# 判斷是否已經(jīng)存在監(jiān)聽者
cls_listeners = cls._listeners.get(signal).setdefault(cls, [])
if listener not in cls_listeners:
# 如果不存在,則添加監(jiān)聽者
cls_listeners.append(listener)
def _before_process(self):
# 取出before_process監(jiān)聽者
cls_listeners = self._listeners.get(Signals.before_process, {}).get(self.__class__, [])
for listener in cls_listeners:
# 調(diào)用監(jiān)聽者
listener(self.__class__, self)
def _after_process(self):
# 取出after_process監(jiān)聽者
cls_listeners = self._listeners.get(Signals.after_process, {}).get(self.__class__, [])
for listener in cls_listeners:
# 調(diào)用監(jiān)聽者
listener(self.__class__, self)
class SignalModel(Model):
def process(self):
"""真正的調(diào)用端"""
self._before_process()
print("Processing")
self._after_process()
# 注冊before_process信號(hào)
@before_process(SignalModel)
def before_process_listener(*args, **kwargs):
print("before_process_listener1", args, kwargs)
# 注冊before_process信號(hào)
@before_process(SignalModel)
def before_process_listener(*args, **kwargs):
print("before_process_listener2", args, kwargs)
# 注冊after_process信號(hào)
@after_process(SignalModel)
def before_process_listener(*args, **kwargs):
print("after_process_listener", args, kwargs)
if __name__ == '__main__':
sm = SignalModel()
sm.process()
輸出結(jié)果:
before_process_listener1 (<class '__main__.SignalModel'>, <__main__.SignalModel object at 0x7ff700116e50>) {}
before_process_listener2 (<class '__main__.SignalModel'>, <__main__.SignalModel object at 0x7ff700116e50>) {}
Processing
after_process_listener (<class '__main__.SignalModel'>, <__main__.SignalModel object at 0x7ff700116e50>) {}
總結(jié)
筆者通過對(duì)`tortoise-orm`源碼的學(xué)習(xí),抽絲剝繭,提取了信號(hào)實(shí)現(xiàn)的方式。其核心就是通過一個(gè)字典存儲(chǔ)調(diào)用方自定義的process方法,然后分別在真正的調(diào)用端的前/后觸發(fā)這些自定義方法即可。
以上就是Tortoise-orm信號(hào)實(shí)現(xiàn)及使用場景源碼詳解的詳細(xì)內(nèi)容,更多關(guān)于Tortoise orm信號(hào)場景的資料請(qǐng)關(guān)注腳本之家其它相關(guān)文章!
相關(guān)文章
PyQt5+Caffe+Opencv搭建人臉識(shí)別登錄界面
這篇文章主要為大家詳細(xì)介紹了PyQt5+Caffe+Opencv搭建人臉識(shí)別登錄界面,文中示例代碼介紹的非常詳細(xì),具有一定的參考價(jià)值,感興趣的小伙伴們可以參考一下2019-08-08
Django 拆分model和view的實(shí)現(xiàn)方法
今天小編就為大家分享一篇Django 拆分model和view的實(shí)現(xiàn)方法,具有很好的參考價(jià)值,希望對(duì)大家有所幫助。一起跟隨小編過來看看吧2019-08-08
利用Python?Matplotlib繪圖并輸出圖像到文件中的方式
這篇文章主要介紹了利用Python?Matplotlib繪圖并輸出圖像到文件中的方式,具有很好的參考價(jià)值,希望對(duì)大家有所幫助,如有錯(cuò)誤或未考慮完全的地方,望不吝賜教2023-09-09
Python利用watchdog模塊監(jiān)控文件變化
這篇文章主要為大家介紹一個(gè)Python中的模塊:watchdog模塊,它可以實(shí)現(xiàn)監(jiān)控文件的變化。文中通過示例詳細(xì)介紹了watchdog模塊的使用,需要的可以參考一下2022-06-06
解決Pyinstaller打包為可執(zhí)行文件編碼錯(cuò)誤的問題
這篇文章主要介紹了解決Pyinstaller打包為可執(zhí)行文件編碼錯(cuò)誤的問題,具有很好的參考價(jià)值,希望對(duì)大家有所幫助。一起跟隨小編過來看看吧2021-03-03
Flask框架運(yùn)用Ajax實(shí)現(xiàn)數(shù)據(jù)交互的示例代碼
使用Ajax技術(shù)網(wǎng)頁應(yīng)用能夠快速地將增量更新呈現(xiàn)在用戶界面上,而不需要重載刷新整個(gè)頁面,這使得程序能夠更快地回應(yīng)用戶的操作,本文將簡單介紹使用AJAX如何實(shí)現(xiàn)前后端數(shù)據(jù)通信2022-11-11

