How can I access the pytest session or CLI arguments inside the `pytest_runtest_logreport` hook?

General Discussions
1

I am building a pytest plugin that uses the pytest_runtest_logreport hook to execute some code whenever a test fails. I want to gate this plugin using a CLI argument added via pytest_addoption.

However, I cannot figure out how to access the pytest session state or the CLI arguments inside pytest_runtest_logreport.

Is there a proper way to access session or arguments from within this hook?

2

One thing to keep in mind is that every report object passed to pytest_runtest_logreport has a reference to the pytest config. You can use that to fetch CLI options directly:

def pytest_runtest_logreport(report):
    config = report.config
    if config.getoption("--my-flag") and report.failed:
        print(f"Test failed: {report.nodeid}")
````Preformatted text`


This avoids global variables entirely. The key is remembering that report.config gives you access to all CLI arguments and session state without extra hooks.
3

Another approach that worked well for me is to capture CLI flags in pytest_configure and attach them to the config object:

def pytest_addoption(parser):
    parser.addoption("--enable-feature", action="store_true")

def pytest_configure(config):
    config.enable_feature = config.getoption("--enable-feature")

def pytest_runtest_logreport(report):
    if getattr(report.config, "enable_feature", False) and report.failed:
        print(f"Custom action for {report.nodeid}")

This pattern is nice because you can safely attach any session-level data you need and reference it later in your hooks.

4

Sometimes pytest_runtest_logreport feels a bit limited. I found using pytest_runtest_makereport easier because it always gives you the item along with the call and report, so accessing CLI options is straightforward:

def pytest_runtest_makereport(item, call):
    report = pytest.TestReport.from_item_and_call(item, call)
    if item.config.getoption("--my-flag") and report.failed:
        print(f"Failed test: {item.nodeid}")
    return report

This way, you have full access to session, item, and CLI args, making it easier to gate custom logic.

If you want, I can also draft three more variations that explain how to combine session-scoped fixtures with CLI options so your plugin behaves consistently across all tests.