Python collectors

To use Buildkite Test Engine with your Python projects, use the buildkite-test-collector package with pytest.

The recommended way to set up a new test suite is to add the Tests Buildkite plugin to the step that runs your tests. It works with every runner that bktec supports, and the entire setup is configuration-only — you can get a test suite running through changes to pipeline.yml alone, with no modifications to your application code.

Use the language-specific test collector documented on this page when you want deeper framework integration — such as custom execution tags, span annotations, or richer per-framework data. Language-specific collectors still pair well with the Tests Buildkite plugin, but adding one requires changes to your application code.

You can also upload test results by importing JSON or JUnit XML.

Tests Buildkite plugin example for pytest

The following step uses the Tests Buildkite plugin to run a pytest suite through bktec. The plugin downloads bktec, requests an OIDC token, ensures the test suite exists, and exports the environment variables that bktec expects, so the step's command only needs to invoke bktec run:

steps:
  - label: "pytest"
    command: bktec run
    plugins:
      - tests#v1.0.0:
          test-runner: pytest
    parallelism: 4

See the Tests Buildkite plugin page for the full plugin reference, including all supported options and dynamic parallelism with bktec plan.

pytest collector

pytest is a testing framework for Python. If you're already using pytest, then you can install buildkite-test-collector to collect test results into your Test Engine dashboard.

Before you start, make sure pytest runs with access to CI environment variables.

To get started with buildkite-test-collector:

In your CI environment, set the BUILDKITE_ANALYTICS_TOKEN environment variable securely to your Buildkite Test Engine API token.
Add buildkite-test-collector to your list of dependencies. Some examples:
- If you're using a requirements.txt file, add buildkite-test-collector on a new line.
- If you're using a setup.py file, add buildkite-test-collector to the extras_require argument, like this:
```
extras_require={"dev": ["pytest", "buildkite-test-collector"]}
```
- If you're using Pipenv, run pipenv install --dev buildkite-test-collector.
If you're using another tool, see your dependency management system's documentation for help.

Commit and push your changes:

$ git add .
$ git commit -m "Install and set up Buildkite Test Engine"
$ git push

Once you're done, in your Test Engine dashboard, you'll see analytics of test executions on all branches that include this code.

If you don't see branch names, build numbers, or commit hashes in Test Engine, then read CI environments to learn more about exporting your environment to the collector.

Upload custom tags for test executions

You can group test executions using custom tags to compare metrics across different dimensions, such as:

Language versions
Cloud providers
Instance types
Team ownership
and more

We offer a tagging solution based on pytest custom markers.

Upload-level tags

In your conftest.py file, you can use pytest global hook to tag all your test executions in a centralized way.

import pytest
import sys

def pytest_itemcollected(item):
  # add execution tag to all tests
  item.add_marker(pytest.mark.execution_tag("test.framework.name", "pytest"))
  item.add_marker(pytest.mark.execution_tag("test.framework.version", pytest.__version__))
  item.add_marker(pytest.mark.execution_tag("cloud.provider", "aws"))
  item.add_marker(pytest.mark.execution_tag("language.version", sys.version))

Execution-level tags

For more granular control, you can programmatically or statically add tags to target individual tests.

To do it statically, targeting a single test or module:

import pytest

 @pytest.mark.execution_tag("team", "frontend")
 def test_add():
     assert 1 + 1 == 2

To do it programmatically, for example:

import pytest
import sys

def pytest_itemcollected(item):
  # You can use the rich data provided by pytest to selectively add execution tag to tests
  if "e2e" in item.location[0]:
    item.add_marker(pytest.mark.execution_tag("type", "browser"))