Merge feature/264-order-management into develop @W-23254713 - #3914

Add returnOmsOrder mutation

@W-22918057 Extract standalone OrderTracking component

Pure refactor: move the inline tracking render from renderShippingMethod in
order-detail.jsx into a standalone, testable app/components/order-tracking
component, wired at both call sites (delivery/OMS-fallback and OMS-multi). The
OMS-multi call site no longer needs a wrapping React.Fragment (key moves onto
OrderTracking), and the now-unused ChakraLink import is removed from
order-detail.jsx.

DOM output is identical and i18n message IDs are unchanged (shipping-status
descriptors are kept as inline formatMessage literals so babel-plugin-formatjs
still statically extracts them). The existing orders.test.js suite plus new
isolated component tests pass (96/96); lint clean.

First of the Order Tracking epic work items.

@W-22918455 Render estimated delivery date on order tracking

Render expectedDeliveryDate and actualDeliveryDate on the OrderTracking
component as locale-formatted "Expected delivery: <date>" / "Delivered: <date>"
lines, shown when present and omitted when absent (or when the date is missing
or malformed — formatTrackingDate returns null on an unparseable value so it
never renders "Invalid Date" or throws). Wired at both order-detail call sites;
these fields are OMS-only (no ECOM fallback — they exist only on
omsData.shipments[]).

Adds two i18n keys (extracted + compiled). Unit and page tests cover the
present, absent, and malformed cases across single, multi-shipment, and
ECOM-fallback paths.

* @W-22918457 Guard null delivery dates in OrderTracking

QA on W-22918455 found that new Date(null) returns the epoch (1970-01-01), not Invalid Date, so a null expectedDeliveryDate/actualDeliveryDate slipped past the isNaN guard and rendered '31 Dec 1969' to the shopper. Add a falsy guard (catches null, undefined, '') to formatTrackingDate, with a regression test for the explicit-null case. Defensive regardless of whether OMS serializes null on the wire.

* @W-22918457 Re-trigger CI (flaky e2e SLAS login)

Consume useOrder's isError and render a full-card error (OrderNotFoundCard-style: square corners, title + description + Back to Order History) instead of hanging on the loading skeleton forever (the AC6 bug). A successful order with no omsData is NOT an error — it renders normally via the ECOM fallback. Adds the OrderLoadError component, i18n keys, and unit tests for the error, back-link, and no-omsData paths. Partial (products-only) failure keeps graceful degradation per the team's decision.

… can ingest(#3875)

@W-22952388@ Sync payment-instrument amount before createOrder so OMS can ingest

…3878)

* @W-22821836@ Invalidate customer orders cache after createOrder

After a registered shopper places an order, useCustomerOrders kept
serving the stale (pre-order) result and only refreshed on a fresh
login, because createOrder's cache-update matrix only invalidated
getCustomerBaskets. Add getCustomerOrders to the same invalidate list
so the account order-history page reflects the new order immediately.
Guest checkouts (no customerId) remain a no-op.

* @W-22806925@ Integrate SCAPI Cancel Order API with UI

- Replace dummy cancel handler with real useShopperOrdersMutation(CancelOmsOrder)
- Fetch reason codes from useOmsMetaData API when modal opens
- Update cancel eligibility: omsData must exist + quantityAvailableToCancel === quantityOrdered for all items
- Remove oms.enabled config flag — cancel visibility driven by order.omsData presence
- Remove hardcoded CANCELLATION_REASONS — dropdown populated from API response
- Hide dropdown when no reason codes available (allows cancel without reason)
- Pass isSubmitting from mutation state to disable modal buttons during request
- Show skeleton while reason codes are loading
- Remove extracted translation messages for hardcoded reasons
- Add mock for useShopperOrdersMutation/useOmsMetaData in order detail tests

* Fix: keep modal open during API call, close only after response

* Fix CI: prettier formatting and add changelog entry

* Revert changelog entry — will add when merging to develop

* Fix prettier: correct indentation for map, revert formatMessage to multiline

* Pre-fetch OMS metadata on page load instead of on modal open

* Fix prettier: inline useOmsMetaData and formatMessage calls

* Fix prettier conflict: extract formatMessage to variables (under 100 char width)

* Fix prettier: match showCancelSuccess format with showCancelError

* Address Jie's review: initialize default reason and add no-reason test

- Pre-select the default reason code when modal opens (from API response)
- Add test: passes empty string when no reason codes provided and confirm clicked
- Update existing tests to expect default reason pre-selection

* Fix lint and tighten canCancel check

- Add item.omsData != null guard to prevent undefined === undefined passing
- Inline formatMessage in showCancelSuccess (satisfies generated-project prettier)
- Inline test props (satisfies generated-project prettier)

* Fix prettier: revert showCancelSuccess formatMessage to multiline (matches showCancelError)

feat: surface OMS return eligibility on order detail

…sages (#3884)

* @W-22930993@ Handle cancel order API error scenarios with specific messages

- 404: "Unable to cancel order" / "We could not find this order. Please refresh and try again."
- 409: "Unable to cancel order" / "This order is already being processed and cannot be canceled. Please reach out to the Merchant."
- Everything else (400, 401, 500, network): "Something went wrong" / "We couldn't process your cancellation right now. Please wait a moment and try again."

Error status is read from error.response.status (commerce-sdk-isomorphic ResponseError shape).
Cancel button remains enabled on error to allow retry.

* Fix prettier: inline defaultMessage values

* Address review: disable button on terminal errors, fix copy

- Disable cancel button after 404/409 (terminal — retrying won't help)
- Keep button enabled for transient errors (500/network) to allow retry
- Fix spelling: "canceled" → "cancelled", "Merchant" → "merchant"
- Reword 409 copy: "Please contact the merchant for assistance."

* Address review: revert copy change, fix spelling only, add error tests

- Revert message rewording — keep original copy, only fix "canceled" → "cancelled"
  and "Merchant" → "merchant"
- Add unit tests for cancel error scenarios (404, 409, 500, network failure)
- Tests verify correct title and description for each error code

* Fix lint: split long defaultMessage lines, multiline isDisabled, remove unused act

…s unavailable (#3883)

* @W-22930989@ Hide cancellation reason dropdown when OMS metadata API fails

The dropdown is already hidden when reasonCodes is undefined (metadata
fetch failed). Add retry: 1 to the useOmsMetaData query so React Query
automatically retries once before giving up. If both attempts fail,
the modal renders without the dropdown and the shopper can still
confirm cancellation — the server applies the default reason.

* Add unit tests for metadata API failure scenarios

* Address review: remove retry override, conditional modal text

- Remove retry: 1 — let TanStack default (3) handle resilience
- Show "Confirm cancellation below." when dropdown is hidden (no reason codes)
- Show "Select a reason and confirm cancellation." when dropdown is visible
- Update test for conditional text

feat: return-item-selection modal on order detail

…3895)

feat: add return review step + returnOmsOrder submission

* @W-22806929@ Add comprehensive cancel order integration tests

12 tests covering:
- Eligibility: OMS order shows cancel, non-OMS hides it, partially shipped
  disables, already-cancelled disables, wrong customer disables
- Happy path: modal opens, API called with correct payload, success feedback
- Reason selection: submits selected reason to API
- Error states: button disabled after 409 (terminal), enabled after 500 (retry)
- Modal UX: keep order closes without API call, no-reason-codes shows
  alternative text

* Address review: fix vacuous tests and lint errors

- Fix "cancel submits with selected reason": rewrite as "cancel submits
  empty body when no reason codes available" — tests the actual contract
  (cancelReasonCodes: [] means no dropdown, empty body sent to API)
- Fix "does not show cancel button for different customer": rename to
  "disables cancel button", assert unconditionally (button IS rendered
  but disabled for OMS orders with wrong customerId)
- Rename "quantityAvailableToCancel < quantityOrdered" to "not fully
  cancellable" — clearer wording matching the strict-equality contract
- Fix prettier: multiline mock resolved values
- Remove conditional expects (jest/no-conditional-expect lint rule)

merge develop into feature/264-order-management

…3900)

 order-level status matrix util from item omsData

item-level return error states

…r badges (#3902)

* feat: wire getOrderDisplayStatus into order badges @W-23093717

Use item-level status derivation to show Cancelled badge (red with ×
icon) on both order-detail and order-history pages when all items are
cancelled. Non-cancelled orders retain the existing raw status display.

* Address review: hoist isCancelled out of IIFEs, add negative tests

- Hoist `isCancelled` derivation above the JSX return in order-detail
  and to the top of the .map() callback in order-history, removing the
  inline IIFEs.
- Add negative tests asserting a partially-cancelled order does NOT
  render the "Cancelled" badge (regression guard).

* Address review: memoize isCancelled, add OMS-only scope comment

- Wrap isCancelled in useMemo on order-detail to avoid recomputing on
  every render (matches canCancel/returnableItems pattern).
- Add comment clarifying the derivation is OMS-only; pure ECOM
  cancellations fall through to the raw status string in the badge.
- Fix broken import from merge (return-error-utils missing `import {`).

* Extract shared OrderStatusBadge component

De-duplicate the cancelled badge JSX from order-detail and order-history
into a reusable component with memoized status derivation. Unifies the
message ID (order_status_badge.label.cancelled) so translators only
localize the string once. Prepares a clean extension point for follow-up
status WIs.

---------

Signed-off-by: sf-madhuri-uppu <madhuri.uppu@salesforce.com>

feat: show return status in order status badge

…#3898)

* fix(order-tracking): make carrier tracking links external

A carrier tracking URL can arrive without a scheme (e.g. `www.carrier.com/t`).
Rendered directly in an `href`, the browser treats it as a relative path and
resolves it against the current page, so the tracking-number link navigated
inside the app (`/account/orders/www.carrier.com/t`) instead of going to the
carrier.

Add `ensureExternalUrl` (app/utils/url.js): prepend `https://` to a scheme-less
URL, keep already-absolute http(s) URLs, and reject unsafe or non-web values
(`javascript:`, `data:`, `mailto:`, app-internal paths, …) by returning
`undefined` so the link renders inactive rather than executable or wrong. Apply
it to the OrderTracking tracking-number link.

* chore: trigger CI re-run

* harden ensureExternalUrl per review feedback

Address review findings on the carrier-tracking-link normalizer. The function
trusted the parsed scheme but never validated the resulting host, so several
inputs produced an href that read like the carrier yet navigated elsewhere.

- Reject userinfo: `https://www.ups.com@evil.com` parses to host `evil.com`
  with `www.ups.com` as the username — now rejected on both parse branches.
- Reject backslash forms (`\` is treated as `/` by the URL parser) and run the
  relative-path guard on the raw input so a control char between slashes can't
  collapse `/x/host` into a protocol-relative `//host`.
- Reject non-string input (was throwing on a non-string and crashing the render).
- Reject bare filenames / empty-label hosts (e.g. `data.html`, `a..b`) that
  aren't real external hosts.
- Drop the unreachable http branch and the dead IPv4 check.

Centralize the checks in `isSafeExternalUrl` (http(s) only, no userinfo,
plausible host), applied to both the absolute and prepended candidates. Add
regression tests asserting the resolved host for each case.

* harden ensureExternalUrl against dotted-protocol host confusion

A scheme-less value shaped `label.tld://host` parses to a dotted protocol, so it skipped the dot-less validation and fell through to be re-prepended into `https://label.tld//host` — a link that reads like one host but navigates to another (same class as the userinfo spoof). Reject a dotted-protocol parse that carries an authority; a genuine `host:port` has an empty host on that parse and still resolves. Review feedback.

…#3906)

* feat: add Track Shipment button to order detail @W-23091033

Render a "Track Shipment" order action per OMS shipment that has a
tracking URL — each button opens its own shipment's carrier URL in a new
tab (same destination as the tracking-number link), with no numeric
suffix. When no shipment has a tracking URL, a single disabled button is
shown. The Order Actions row is now full-width stacked on mobile and
inline from the sm breakpoint up, so additional buttons no longer
overflow on narrow screens.

* feat: per-shipment tracking cards + single track action @W-23091033

Align the order-detail tracking UI with storefront-next and the order
tracking epic's "flat list, no address association" scope:

- Render tracking as a flat list of per-shipment cards (one card per
  omsData.shipments[] entry, ECOM order.shipments[] fallback), each with
  carrier, status, tracking-number link, and delivery dates. No address
  on the cards and no positional OMS-to-ECOM index join.
- Consolidate the Track Shipment order action to a single button that
  links to the first shipment with a carrier URL (disabled when none),
  matching storefront-next getTrackShipmentHref.
- Move the shipping address to a single order-level block shown only for
  single-shipment delivery orders; multi-shipment orders no longer
  associate an address per shipment.

Updates the order detail tests and extracts the new Tracking heading
message.

* feat: per-shipment item boxes on order detail @W-23091033

Restructure the Items Ordered area to mirror storefront-next: render one
bordered box per ECOM delivery shipment, each with a "Shipment N" header,
that shipment's status, its items (grouped strictly by shipmentId; all items
when there's a single delivery shipment), and its own native shipping address
and method. Pickup-only orders (no delivery shipment) fall back to a flat item
list.

The top summary card no longer carries a standalone shipping-address block —
the address now lives inside each shipment box. The Tracking cards stay in a
separate flat section (OMS-preferred, ECOM fallback) and carry no address:
there is no reliable join key between an OMS shipment and a specific ECOM
shipment, so a per-shipment box never index-joins OMS data (the deferred TD).

Reconcile the multi-shipment / BOPIS tests to the new per-box behavior: they
previously asserted addresses were hidden and only passed because the products
endpoint was unmocked (which suppressed the box render); they now mock products
and assert each shipment's own address renders in its box, plus strict
item-to-box placement by shipmentId.

* feat: move tracking inside the first shipment box @W-23091033

Match the storefront-next design branch: the flat tracking list now renders
INSIDE the first shipment box, after its items and before its shipping address,
instead of as a standalone section above the items. It is still the whole flat
list (rendered in box 1 only, never split per shipment), so there is no
positional OMS↔ECOM index-join. Pickup-only orders (no delivery box) append the
flat tracking list after the flat item list so tracking is never lost.

Update the regression-lock test: it previously asserted tracking was a separate
section; it now asserts tracking sits inside the first shipment box with the
shipping address as a sibling block (no address inside a tracking card).

* feat: tracking as flat section below shipment boxes @W-23091033

OMS tracking (omsData.shipments[]) has no join key back to a specific ECOM
shipment (deferred TD-0326366), so the layout must not imply which tracking
belongs to which shipment. Move the flat tracking list out of the first
shipment box and render it as a single section BELOW all the boxes — a peer of
the boxes, not nested in one. Each shipment box still shows its own products
(grouped by ECOM shipmentId) and its own native address, both of which ARE
known; only the tracking↔shipment link is unknown, and the flat list is honest
about that. One code path covers single, multi, and pickup-only orders.

Update the regression-lock test: it now asserts tracking renders OUTSIDE (not
nested in) the shipment boxes, with no address/payment text in the section.

* feat: Track Shipment dropdown for multi-shipment orders @W-23091033

UX option A for multi-shipment tracking (one of two branches for review).

When an order has more than one shipment with a carrier tracking URL, the
Track Shipment order action becomes a dropdown (Popover — the shared UI barrel
exposes no Chakra Menu) listing each tracking number as an external link, so
the shopper can pick which carrier link to open. Single-shipment orders keep
the simple external-link button; orders with no tracking URL keep the single
disabled button. This is interim: we still cannot say which tracking maps to
which shipment (deferred TD-0326366), so options are labeled by tracking
number, not by shipment contents.

* chore(order-detail): self-contained rationale in tracking comments

Rewrite the order-detail and test comments to explain the multi-shipment
tracking behavior in project-internal terms (OMS and ECOM shipment arrays share
no join key, so a tracking entry can't be tied to a specific shipment) instead
of referencing tracking artifacts or another product. Customer-facing code
should not carry process/cross-product references; the rationale lives in the
spec and PR instead.

* fix(order-detail): address review findings on tracking layout @W-23091033

- Surface items that match no shipment box (untagged, or a shipmentId naming no
  delivery shipment) in an "Other items" box, so multi-shipment orders never
  silently drop purchased items while the item-count header still counts them.
- Localize the per-shipment box status pill via a shared formatter, so the box
  header and the tracking card render the same localized status instead of one
  showing a raw snake_case token in non-English locales.
- Make the disabled Track Shipment button use aria-disabled + a hidden
  disabled-reason hint (keyboard/SR focusable), matching the Cancel/Return
  buttons in the same row, instead of a native disabled button.

* fix(order-detail): meet AA contrast on shipment status pill @W-23091033

The "Shipment N" status pill rendered gray.700 text on gray.200 (4.04:1),
which fails WCAG AA (4.5:1) and tripped the registered happy-path a11y
snapshot (the color-contrast violation captured on the order detail page).
Darken the text to gray.800 on the same gray.200 background for 5.88:1.

…821845 (#3907)

order-management feature README + strip internal GUS refs

…eturn modal closes (#3910)

return focus to the trigger when the return modal closes

apply CX UI text guidelines to order & account UI

Replace the temporary 5.1.0-unstable-* dev build with stable 5.4.0, the
first release carrying the OMS Shopper Orders endpoints (oms-return-order,
oms-cancel-order, oms-meta-data). develop's 5.2.0 has no OMS endpoints, so
this is required for the OMS hooks to work rather than a plain bump.

Build + tsc clean; ShopperOrders hooks 33/33 pass against 5.4.0.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

The isomorphic 5.2.0 -> 5.4.0 bump adds endpoints with no React hooks
yet, so the 'all endpoints have hooks' guard tests failed. Add each new
endpoint (requestOtp/verifyOtp, setupCustomerPaymentMethodReference,
promoteTemporaryBasket, getComponent/resolveQualifiers, getProduct
Images/Prices/Promotions) to the expected unimplemented list with a TODO,
matching the test's documented convention. Implementing the hooks is out
of scope for the OMS integration.

Full commerce-sdk-react suite: 778/778 pass.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

The commerce-sdk-isomorphic 5.2.0 -> 5.4.0 upgrade pushes vendor.js to
397.07KB, just over the 397 kB budget. Bump to 398 kB (main.js unchanged).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>