how debug Azure Python function running in shell and debugging in Intellij IDE

Question

I'm new with Azure and I dont know how to configurate Azure run in Intellij project configuration.

At this moment I try to debug my first http-trigger function (in Python) that I run in the shell and put the debugger points.

The problem: executed function doesnt neither return its response string nor stop at debugger point

Here is the screenshot where you may see the localhost:7071 successfully started, but when I try to trigger my function with http://localhost:7071/api/first-http-trigger?name=sdaf it doesnt respond: Hello, {name}. This HTTP triggered function executed successfully. and doesnt stop the debugger

Also I've created azure config for Intellij but it doesnt run at all

UPDATE:

function.json

{
  "scriptFile": "__init__.py",
  "bindings": [
    {
      "authLevel": "anonymous",
      "type": "httpTrigger",
      "direction": "in",
      "name": "req",
      "methods": [
        "get",
        "post"
      ]
    },
    {
      "type": "http",
      "direction": "out",
      "name": "$return"
    }
  ]
}

Also I've fixed my problem with localhost api successfully http://localhost:7071/api/first-http-trigger?name=sdaf.

The problem was the func init was initiated in root directory but the func new was done in child one. So the configs were incorrect.

I've solved it just starting from the very start and did all commands in one directory (func init, func new, func start)

But there still a problem is left: the debugging points dont work and I cannot configure azure start in Idea configurations

Answer 1

But there still a problem is left: the debugging points don't work and I cannot configure azure start in Idea configurations

func init, func new, func start, this three things is essentially console commands. It can only make sure you can run the function app.

For Intellij IDEA and VS Code, the key is the extensions, or to say the plugins. First you need to get support for this language. And then you need a plugin to debug. For VS Code, it is 'azure function extension'. And for intellij IDEA, it is 'Azure Toolkit for Intellij'. 'azure function extension' of VS Code can support multiple language. But for 'Azure Toolkit for Intellij', unfortunately, it is focusing on Java(If you want to use it to debug the function app, you will find you can not select any module. This is because it didn't think your function app is a java project.)

It is recommended that you use the VS Code to develop function apps locally so that you can get debug support for function apps in all languages.(If you use VS Code, you just need two things, one is python plugin and another is azure function extension plugin, and then you can debug with no problem.)

Answer 2

Unfortunately, PyCharm takes breakpoints into account only when the code containing them is executed via the Run Configuration under the debugger. When started in the terminal - it is just some process IDE doesn't know much about.

It is not very convenient indeed as there are cases when you cannot execute the code in IDE directly, so PyCharm has a special feature called Python Debug Server.

I should admit I didn't try using it with Azure Python functions.

Debugger Architecture

For better understanding let's review how debugger works in PyCharm. It consists of two parts - Backend and Frontend:

• Backend is written in Python (fork of pydevd called pydevd-pycharm) and is responsible for interacting with Python process runtime (pausing execution, collecting variable values, etc.).
• Frontend is written in Java and Kotlin and it's responsibilities are displaying debugger-related buttons in PyCharm's UI and sending commands to Backend via sockets (e.g. you click "Pause" in debugger UI -> Frontend sends "Pause" command to Backend).

When you debug a simple Python script in PyCharm what is happening is:

• IDE starts Frontend - it waits for a connection from Backend
• IDE starts Backend and passes your script to it
• Backend makes necessary preparations to be able to interact with the runtime and starts your script
• Backend connects to Frontend and waits for commands

Solution Preview

As I said we cannot ask Backend to execute your code directly, but what we can do is

• Start Frontend in an infinite loop
• Insert Backend initialization logic inside the function itself
• Trigger Azure Python function
• Backend initialization logic will be executed launching the backend itself and making all the necessary preparations for it to be able to debug the function
• Backend will connect to Frontend and pause execution waiting for your commands
• You can interact with Frontend debugger as usual

Solution in Action

For this task we can use "Python Debug Server" Run configuration (Run top-menu -> Edit Configurations)

It will be used to start the debugger Frontend. Specify some random port and copy two lines with settrace call inside your function. These two lines are Backend initialization logic. You obviously need the Backend package itself (pydevd-pycharm), you can install it with pip on the interpreter used by the function. If you cannot use pip - Backend is stored inside PyCharm installation folder -> debug-eggs folder -> pydevd-pycharm.egg. You can make the Backend available in runtime by inserting

import sys
sys.path.append("path/to/pydevd-pycharm.egg")

... in your function before import pydevd_pycharm.

At this point

• Start the run configuration (Frontend will start running waiting for Backend)
• Trigger your function

# ... your func code ...

import sys
sys.path.append("path/to/pydevd-pycharm.egg")
import pydevd_pycharm
pydevd_pycharm.settrace('localhost', port=12345, stdoutToServer=True, stderrToServer=True)

# ... your func code ...

• Connection between Frontend and Backend should be established at this point allowing you to use debugger controls from PyCharm to step in/continue/pause/etc the function execution.

I have a screencast where I use this feature to debug generator3/__main__.py script started from the terminal, please feel free to check it out https://youtu.be/uYt3G9WEYvo

Answer 3

To debug a function you started using

func start

What you should do is attach the python debugger to the running process. You can do this using Run > Attach to Process... menu.

A pop up like this will appear:

selecting the right process can take several attempts, but if you hit the correct one you will see

Attaching to a process with PID=XXXXX
...
Connected to pydev debugger (build 222.3739.54)

Then trigger the function from your favorite HTTP client and your breakpoint should hit:

Hopefully in the future it will be like debugging for AZ Functions with Java (I think that one is more stream lined)