[flask advanced] analyze the thread isolation mechanism in flask in combination with the source code

Personal profile

• Author's brief introduction ： Hello everyone , I'm Daniel , New star creator of the whole stack .
• Blogger's personal website ： A Niu's blog house
• Stand by me ： give the thumbs-up + Collection ️+ Leaving a message.
• Series column ：flask Framework quick start
• Maxim ： To be light , Because there are people who are afraid of the dark ！
  

---

I found a good AI learning website a few days ago , Easy to understand , Humor and wit humor , I can't help but share it with you .

Click to jump to the website ： AI learning

Preface

We wrote in the previous introductory series request Object mentioned this request In fact, it is a multi-threaded mechanism , As a global variable , How does it achieve that each thread in a multithread does not affect each other , Thread safe , This article will combine source code analysis flask Thread isolation mechanism in ！

Question elicitation

First, let's explain thread isolation , He means that the threads in a process do not affect each other at runtime , Like the following code ：

import threading

class B:
    a = 0

b = B()
def func():
    b.a += 1
    print(" Threads 1 : a = " + str(b.a) + "\n")

new_t = threading.Thread(target=func)
new_t.start()

b.a += 1
print(" The main thread  : a = " + str(b.a))

You can see ,a Global variable , Main thread and thread pair variables a The influence of is mutual , Not isolated , Corresponding to flask in ,request Is a global variable , It is possible to accept many network requests at the same time , To start multiple threads . In order to make request The correct point is its instantiation Request object , Ensure that requests do not interfere with each other , Thread isolation mechanism is very important ！ Let's take a look flask Thread isolation mechanism in ！

flask Thread isolation implementation in

As we can think , We can use python Dictionary in this data structure , Like this ：

request = {
    
 	thread_id1:Request1;
 	thread_id2:Request2;
 	...
}

So it's actually flask The interior is similar , It also uses the basic data type of dictionary to realize thread isolation , Let's go and have a look ！

Dictionaries can only store data , And we also need to manipulate data , Therefore, it needs to be encapsulated in an object , We introduced flask One of the cores of werkzeug, The implementation of thread isolation we want to see happens to be werkzeug Inside local.py In file ：

You can see such a Local object , In fact, it is a package of dictionary principles ！

So we use this Local Object to test thread isolation ：

import threading,time
from werkzeug.local import Local

b = Local()
b.a = 0

def func():
    b.a = 1
    print(" Threads 1 : a = " + str(b.a))

new_t = threading.Thread(target=func,name='thread_01')
new_t.start()

time.sleep(1)
b.a += 2
print(" The main thread  : a = " + str(b.a))

You can see , We use it Local The instantiated object is thread isolated ！

Next, let's continue to look at the previous picture

Before written _request_ctx_stack and _app_ctx_stack These variables actually point to LocalStack This stack , namely LocalStack Instantiated object ！

Now let's finish reading Local after , We also guess this from the name LocalStack Not simple , It is a thread isolated stack ！
We can see LocalStack Source code ：

You can see LocalStack Isolate objects from threads Local It was packaged , And it provides push,pop, And get the top element of the stack top Method ！

Seeing this, we can conclude that ：Local Using dictionary to realize thread isolation ,LocalStack Yes Local It is encapsulated to form a thread isolated stack ！

Still use LocalStack Write a case to see ：

import threading,time
from werkzeug.local import LocalStack

myStack = LocalStack()
myStack.push(1)
print(" In the main thread push after , The value is ：" + str(myStack.top))

def fun():
    #  Thread one 
    myStack.push(2)
    print(" In thread one push after , The value is ：" + str(myStack.top))
    

new_t = threading.Thread(target=fun,name='thread_01')
new_t.start()
time.sleep(1)

# The main thread 
print(" Finally, the value of the main thread is  ：" + str(myStack.top))

You can see that threads do not affect each other ！

So flask in AppContext and RequestContext It's also thread isolated ( Because these two contexts are pushed LocalStack), Essentially, when used request when , Thread isolation makes request Can correctly find its instantiation Request!

summary

The significance of using thread isolation lies in ： Enable the current thread to correctly reference the object it has created , Instead of referring to objects created by other threads !

From this to push ：

Not only request Is thread isolated ,session and g Object is also （ Be careful :flask Core objects of app Not thread isolated , There is only one global , That is to say current_app Thread isolation is meaningless ！ and request Instantiate the object Request There are many ！）

Conclusion

If you think the blogger's writing is good , You can pay attention to the current column , Bloggers will finish this series ！ You are also welcome to subscribe to other good columns of bloggers .

Series column
Soft grinding css
Hard bubble javascript
The front end is practical and small demo