testivai (Python)

Local-first visual regression for playwright-python — pixel + DOM comparison with tunable tolerances, a self-contained HTML report, and PR-comment baseline approvals. Shares the exact on-disk contract with the JS/TS TestivAI adapters, so mixed-language repos use one set of baselines, one report, one approval flow.

pip install testivai[playwright]

With pytest (pytest-playwright)

def test_homepage(page, testivai_witness):
    page.goto("http://localhost:3000")
    testivai_witness(page, "homepage")

The bundled pytest plugin runs testivai report at session end: captures are diffed against .testivai/baselines/, tolerances from .testivai/config.json apply, and visual-report/index.html + results.json are written. Approve baselines with npx testivai approve --all (or a /testivai approve PR comment via the GitHub Action).

Without pytest

from testivai import witness, run_report

witness(page, "homepage", ignore_selectors=[".live-feed"], variant="firefox")
run_report()

What captures do (identical to the JS adapters)

• Stabilization by default: animations/transitions complete instantly at their final state, caret hidden, web fonts awaited — the top causes of flaky visual diffs, neutralized.
• ignore_selectors: hidden from the pixels and excluded from the DOM snapshot (dynamic content never counts as a change).
• variant: folds into the snapshot name (homepage__firefox) so parallel browsers/viewports never overwrite each other's baselines.

The compare/report half runs via the @testivai/witness CLI — Node.js is already present wherever playwright-python is (Playwright for Python ships a Node driver). Resolution order: TESTIVAI_CLI env → testivai on PATH → npx --yes @testivai/witness.

Docs: https://github.com/testivai/testivai-oss/blob/main/docs/frameworks/python.md

MIT.