New test

Metadata saved to the .json sidecar next to the test file

Test id — becomes part of the filename and is the permanent key for history (e.g. LOGIN-001) Language — Python runs on the bundled harness; TypeScript runs on the TS engine's @playwright/test (/opt/pw-ts) Name Description

Tags (comma separated) Target version tag Timeout seconds (0 = default) Video recording

Turn video off for tests that mostly wait. The video file grows with wall-clock time whenever the page repaints — a 30-minute wait on a page with a clock or progress bar costs about 80 MB and shows almost nothing. The activity replay still captures those runs, and stays small.

enabled

Groups

Freight extras Freight in TypeScript Freight platform Freight regression Freight smoke Python and TypeScript, side by side Smoke

Weekly schedules are managed on the Schedules page.

Playwright code the file must define `run(page, ctx)`

ctx.base_url is the target URL from config.json · ctx.log("...") writes to the run log · ctx.screenshot("name") saves a screenshot · with ctx.timed("label"): times a phase · raise AssertionError to fail.
Editor: type page. or ctx. (or press Ctrl-Space) for the API list with docs · Ctrl-F search · Ctrl-/ comment.

“Record in a browser tab” opens the Remote Recorder: the hub drives its own bundled Chromium and you interact with a live picture of it — so it works on a deployed hub (AWS/office) and from a locked-down workstation browser. Everything finishes IN that tab: stopping offers “create a new test”, which brings you back here with the code filled in. If your browser blocks the pop-up (Safari), use the “Record new test” button on the Tests page instead — it opens in this same tab.

"""A password reset, end to end -- through the email.

The flow people test by hand because "it goes through email": ask for a
reset, WAIT for the email (a mail queue is not instant), open it, follow the
link, choose a new password, sign in with it -- and then check the link is
dead, because a reset link that works twice turns every old email into a key
to the account.

The demo catches its outgoing mail in a Mailbox page (as MailHog or Mailpit
do in a real test environment), and the test correlates by the request's
reference number: other resets may be in the mailbox too. The reset link
carries a secret token, so the test CLICKS it in the email rather than
navigating to it -- a navigation would write the token into the run's log.
The new password is random per run and never logged.
"""
import secrets

from _lib.freight import Freight

ACCOUNT = "driver"                               # tester/manager are shared: never reset
ADDRESS = "driver@acme-freight.example"
RESET_LINK = "#mail-body a[href*='/login/reset/']"


def run(page, ctx):
    site = Freight(page, ctx)
    site.open("login/")
    page.click("#forgot-link")
    page.fill("#forgot-login", ACCOUNT)
    page.click("#forgot-btn")
    ref = page.inner_text("#forgot-ref").strip()
    ctx.log(f"reset requested, reference {ref}")

    with ctx.timed("wait for the email"):
        site.open(f"mailbox/?to={ADDRESS}")
        # the mailbox checks for new mail every two seconds and reloads itself
        subject = f"#mail-list a.mail-subject:has-text('ref {ref}')"
        page.wait_for_selector(subject, timeout=60_000)
    page.click(subject)
    page.wait_for_selector(RESET_LINK)
    message = page.url
    ctx.screenshot("the reset email")

    new_password = "Pw-" + secrets.token_urlsafe(12)
    with ctx.timed("choose a new password"):
        page.click(RESET_LINK)
        page.fill("#new-password", new_password)
        page.fill("#confirm-password", new_password)
        page.click("#reset-btn")
        page.wait_for_selector("#login-btn")
    with ctx.timed("sign in with the new password"):
        page.fill("#username", ACCOUNT)
        page.fill("#password", new_password)
        page.click("#login-btn")
        page.wait_for_selector("#signed-in-user")
    ctx.log("signed in with the new password")
    page.click("#sign-out")
    page.wait_for_selector("#login-btn")

    with ctx.timed("follow the same link again"):
        page.goto(message)
        page.click(RESET_LINK)
        page.wait_for_selector("#reset-invalid, #reset-form")
    ctx.screenshot("the same link, a second time")
    assert page.locator("#reset-form").count() == 0, (
        "the reset link worked a SECOND time: any old reset email is still a key to the "
        "account -- a reset link must stop working once it has been used")
    ctx.log(page.inner_text("#reset-invalid-why"))

Saving writes the test file (.py or .spec.ts) and its .json sidecar — the folder stays the source of truth.

New test

Metadata saved to the .json sidecar next to the test file

Groups

Playwright code the file must define run(page, ctx)

Playwright code the file must define `run(page, ctx)`