Add JS/Python profiler parity tests

[ahmdkaml]

What changed

• Added a Node.js harness (scripts/profile-js.js) to execute the JavaScript profiler from CI.
• Added parity tests covering multiple fixture datasets.
• Fixed JS profiler behavior to match the Python implementation:
  • Preserve "none" as a categorical value instead of treating it as missing.
  • Emit small_group reliability flags.
• Updated the profiler CI workflow to provision Node.js so the parity tests run automatically.

This automates the Python↔JavaScript parity check described in SPEC.md and helps prevent future implementation drift.
Closes #110

[vercel]

@ahmdkaml is attempting to deploy a commit to the yashkewlani2020-gmailcom's projects Team on Vercel.

A member of the Team first needs to authorize it.

[yakew7]

Looks fine Merging the PR

[Copilot]

Pull request overview

Adds automated Python↔JavaScript profiler parity checks (per faircode/SPEC.md / issue #110) by introducing a Node.js harness, fixture-backed pytest parity tests, and CI provisioning for Node so parity runs on every PR.

Changes:

• Add scripts/profile-js.js Node harness to execute the JS profiler engine and emit JSON.
• Add pytest parity coverage across multiple fixture CSVs (including new small.csv and credit_customers.csv).
• Update JS engine behavior to better align with the Python profiler ("none" handling and small_group reliability flags), and update CI (audits.yml) to install Node 20 for profiler tests.

Reviewed changes

Copilot reviewed 5 out of 9 changed files in this pull request and generated 1 comment.

Show a summary per file

File	Description
tests/test_js_parity.py	Adds Python↔JS parity test that runs Node harness and compares profiler JSON results.
tests/fixtures/small.csv	New minimal fixture dataset for parity testing.
tests/fixtures/credit_customers.csv	New larger categorical-heavy fixture dataset for parity testing.
scripts/profile-js.js	New Node.js harness to run the JS profiler engine from CI/tests and print JSON.
assets/profiler-engine.js	Adjusts missing-token handling and adds small_group flag messages in JS to match Python behavior.
.github/workflows/audits.yml	Provisions Node.js in the profiler job so parity tests can run in CI.

Suppressed comments (2)

tests/test_js_parity.py:36

• subprocess.run(["node", "scripts/profile-js.js", ...]) will fail if the test is executed from a non-repo-root working directory. Using the resolved harness path (and the discovered NODE executable) makes this robust and matches how the CSV path is handled.

    completed = subprocess.run(
        ["node", "scripts/profile-js.js", str(csv)],
        capture_output=True,
        text=True,
        encoding="utf-8",

tests/test_js_parity.py:46

• The parity assertion currently drops flags entirely. flags are part of the profiler contract (and are tested on the Python side in tests/test_profiler.py), so excluding them means JS flagging drift could slip through even if the structured metrics still match. If exact string formatting is too brittle across languages, consider comparing a normalized form of the flags (e.g., stripping numeric literals) in addition to the full structured JSON.

    # Flags are human-readable messages. They duplicate information already
    # present in the structured output and may differ because Python and
    # JavaScript format floating-point values differently (e.g. 6.25 -> 6.2
    # vs 6.3). Compare the structured data instead.
    python_result = dict(python_result)

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.