Official Testinel plugin for pytest

Testinel’s pytest plugin captures structured Selenium and Playwright test execution data directly from pytest and sends it to Testinel, where your test results become searchable, comparable, and actually useful. No log scraping. No brittle CI hacks. Just deterministic test analytics.

📦 Getting Started

Prerequisites

You need a Testinel account and project.

Python 3.11 through 3.16 is required.

Installation

Getting Testinel into your project is straightforward. Just run this command in your terminal:

pip install --upgrade pytest-testinel

Or with uv:

uv add pytest-testinel

Configuration

Set Testinel reporter DSN environment variable TESTINEL_DSN. Get DSN for your project at https://testinel.dev/projects/.

Examples:

# Report to Testinel (HTTPS)
export TESTINEL_DSN="https://your.testinel.endpoint/ingest/xxxxxxxxxxxx/"

# Report to a local file (JSON)
export TESTINEL_DSN="file:///tmp/testinel-results.json"

# Or use a direct file path
export TESTINEL_DSN="./testinel-results.json"

Logging

Testinel uses pytest's standard logging controls.

Each reported event has a UUID event_id in its JSON body, including events written to a local file. Each HTTP event submission and upload-link request gets a fresh UUID request_id, sent in both the X-Request-ID header and the JSON body. Sending the same event again preserves its event_id and generates a new request_id. The IDs are logged at DEBUG before sending the request. Signed binary uploads retain the headers supplied by the storage provider.

If an event submission returns HTTP 404, Testinel disables reporting for the rest of the run and emits one terminal warning and one WARNING log record. Check TESTINEL_DSN and your project at https://testinel.dev/projects/. Tests continue normally with their usual exit status. No further events or new attachments are sent; queued uploads are discarded, while an upload already in progress may finish. A fresh pytest process tries reporting again. Other request failures, including attachment-link or storage-upload 404s, keep their existing best-effort behavior.

To set the logging level to INFO for all pytest loggers and show the messages live, run:

pytest --log-level=INFO --log-cli-level=INFO

To change the level for pytest-testinel only, set its parent logger in your project's conftest.py:

import logging


def pytest_configure() -> None:
    logging.getLogger("pytest_testinel").setLevel(logging.INFO)

To also show these messages live, enable pytest's live logging:

pytest -o log_cli=true

Other loggers keep their existing levels. Replace INFO with DEBUG, WARNING, or another standard Python logging level as needed.

Use pytest's --log-file, --log-format, and related options to control log storage and formatting.

Recommended pytest flags

For better debugging and richer failure context, it is highly recommended to run pytest with:

--showlocals --tb=long -vv

Why:

• --showlocals: includes local variable values in tracebacks, which makes root-cause analysis much faster.
• --tb=long: shows full, non-truncated tracebacks so you can see complete failure paths.
• -vv: increases verbosity, showing more detailed test identifiers and execution output.

Example:

pytest --showlocals --tb=long -vv

Recommended Playwright flags

If you run browser tests with pytest-playwright, these flags provide better artifacts for debugging:

--tracing=retain-on-failure --video=retain-on-failure --screenshot=only-on-failure --output=test-results

Why:

• --tracing=retain-on-failure: captures a full Playwright trace for failed tests only.
• --video=retain-on-failure: keeps video recordings only for failed tests.
• --screenshot=only-on-failure: saves screenshots at failure time.
• --output=test-results: stores artifacts in a predictable directory.

Example:

pytest --tracing=retain-on-failure --video=retain-on-failure --screenshot=only-on-failure --output=test-results

Development

Run the test matrix with tox:

uv run tox

The matrix covers Python 3.11 through 3.16 against the latest available release in each pytest major version: 7, 8, and 9. Each environment installs the package as a wheel. Missing Python interpreters fail the run; install them before running the full matrix.

Run one environment, or pass arguments to pytest:

uv run tox -e py311-pytest7
uv run tox -e py315-pytest9 -- tests/test_testinel.py -v

Use uv run tox run -r to recreate environments and resolve the latest pytest releases after new versions are published.