How to Contribute

Contributing

This project has adopted the Microsoft Open Source Code of Conduct. For more information see the Code of Conduct FAQ or contact opencode@microsoft.com with any additional questions or comments.

Developing the PostgreSQL Tools Service

1. Ensure that Python 3.11 is installed - older versions of Python may work but are not supported
   • Note that when these instructions refer to 'python3' or 'pip3' you may need to use the commands 'python' or 'pip' instead if you are on Windows
2. Ensure that PostgreSQL is installed locally. This is required for the pg_config dependency of the PostgreSQL driver to be found.
3. Make sure you have MSVC 140 installed, Build Tools Installation
4. Run pip3 install -r requirements.txt from the root of the project
   • If you need to install pip, see pip Installation

Before Committing

We follow Python's PEP 8 style guide with a maximum line length of 160 characters. To ensure that your code has no style problems, run the following commands before committing (install them from pip first if needed):

1. Run autopep8 to format your code according to PEP 8
   • autopep8 --in-place --aggressive --aggressive --max-line-length 160 -r . (from the project's root directory)
2. Run flake8 to look for problems that need to be resolved manually
   • On Mac/Linux: ./scripts/flake8.sh
   • On Windows: .\scripts\flake8.ps1

Running Tests

The following directions can be used to run all tests: Run pip3 install -r requirements.txt if you haven't already

You can use scripts/test-all.sh or scripts\test-all.ps1 to run all tests. All of these scripts will accept any arguments that you give and pass them directly to nose2.

To run tests in VS Code editor, add the following in VS Code settings: "python.unitTest.nosetestsEnabled": true

Debugging Tests

Open vscode debugger's launch.json file. Replace the <full_test_name> with the full test name. { "name": "Python: Test Debugger", "type": "python", "request": "launch", "module": "nose2", "args": ["-v", "<full_test_name>"], "console": "integratedTerminal" },

Configuring Integration Tests

Integration tests require a local config file that contains the options for connecting to your test database. The template config file is located in tests/integration/config.json.txt. Copy this file to tests/integration/config.json and modify values as appropriate. The template lists the most common options, but any options that can be used when establishing a psycopg connection can also be set in the config file. You can put multiple server configurations in the file in order to run each integration test multiple times, once per server.

Creating Integration Tests

Integration tests can be inserted in line with our unit tests. The tests.integration module exports a integration_test decorator and a get_connection function that can be used in integration tests.

To declare that a test is an integration test, mark it with the imported @integration_test decorator. This will automatically patch psycopg.connect in your test to return the test database connection from your config file, and will let you use the get_connection function to retrieve that connection if you need it elsewhere. You can optionally pass min_version and max_version integer arguments to the @integration_test decorator to make the test only run when connected to the specified versions of Postgres. You can also use the create_extra_test_database function if your test needs additional databases.

Each integration test will run with its own database, which will be created before the test starts and dropped when the test ends.

You can find an integration test example in code in tests/query_execution/test_query_execution_service.py's test_query_execution_and_retrieval method.

Creating End-to-End JSON RPC Integration Tests

You can also create end-to-end JSON RPC integration tests, which test the entire service using JSON RPC inputs and outputs. An example of these tests is located in tests/json_rpc_tests/object_explorer_json_rpc_tests.py.

When creating one of these tests, mark the test with an integration test decorator as usual. Then, create tests.json_rpc_tests.RPCTestMessage objects for each message to send as part of the test, including any connection messages that are needed. Finally, initialize a tests.json_rpc_tests.JSONRPCTestCase object with the ordered list of messages to send, and call its .run() method to run the test.

When creating each tests.json_rpc_tests.RPCTestMessage, you should include a response verification callback that will be called with the response as a parameter, deserialized as a dictionary, when the server responds to the request. You should also include a list of notification verifiers. Each verifier is a tuple consisting of two functions. The first function is a filter function that will be called to determine which single notification the verifier applies to. If the verifier function returns true for multiple notifications, the first one will be used. The second function is a callback that will be called with the notification as a parameter, deserialized as a dictionary.

Remote Debugging with VS Code

It is possible to remotely attach a debugger to the PostgreSQL Tools Service through VS Code's Python extension. Just start the service with the command line argument --enable-remote-debugging and then, from the debug tab in VS Code, click 'Attach (Remote Debug)'.

If you want to debug startup, use the flag --enable-remote-debugging-wait instead, and the service will wait for you to attach the debugger before starting up.

By default, the remote debugging server runs on port 3000. If you need to use a different port, just pass it as the argument to that flag, e.g. --enable-remote-debugging=3001

Remote Debugging with Azure Data Studio

You can set configuration options in Azure Data Studio to let you attach the remote debugger to the PostgreSQL Tools Service running inside Azure Data Studio. Set pgsql.useDebugSource to true and set pgsql.debugSourcePath to the path to the root of your PostgreSQL Tools Service repo (i.e. the folder containing this readme file). If you want to debug startup, also set pgsql.enableStartupDebugging to true. Here are examples from a settings.json file:

    "pgsql.debugSourcePath": "/Users/mairvine/code/pgtoolsservice",
    "pgsql.useDebugSource": true,
    "pgsql.enableStartupDebugging": true

Building Executables

To build an executable, run the following commands starting from the main source code directory on the platform you want to build for. The output will be placed in a folder called build.

• On Mac and Linux: ./scripts/build.sh
• On Windows: .\scripts\build.ps1
  • Or, from cmd.exe: powershell.exe scripts\build.ps1