Angular search-results page renders <flight-details-body-actions> →
<flight-actions> with NO overrides inside every expanded flight body —
share/buy/register/status all surface there. A prior refactor confused
this with the dedicated /schedule/details page, where Angular's
flight-schedule-details DOES set [share]=false [buy]=false [print]=false
[details]=false [register]=false because that page-level summary owns
those affordances. The strip was removed from both contexts, leaving
the search results page (e.g. /ru-ru/schedule/route/AER-LED-…) without
any buy button when a flight is expanded.
ScheduleFlightBody now accepts an opt-in showActions flag and renders
the existing <FlightActions> at the bottom (Angular-parity gating via
canBuyTicket / canViewFlightStatus). DayGroupedFlightList opts in;
ScheduleDetailsPage stays opted out so its page-level summary remains
the single owner of share/buy on the details page.
Note on e2e: tests/e2e/schedule-route-buy-button.spec.ts asserts the
button surfaces after expanding the first card, but the local dev
server's curl-based API proxy is currently being blocked by the
upstream WAF ("Доступ к сайту временно ограничен"), so the spec runs
green only against environments that reach /api. CI + deployed
verification suites cover that path. Behaviour is also locked in by:
- ScheduleFlightBody.test.tsx — strip renders iff showActions=true
- DayGroupedFlightList.test.tsx — passes showActions=true through
Inline export const loader from page.tsx didn't run — _ROUTER_DATA
showed loaderData[(lang)/onlineboard/page] = null and useLoaderData()
threw 'Cannot read properties of null'. Modern.js conventional routes
require the loader in a co-located data.ts file.
useLoaderData() now defensively handles null (defaults to undefined,
component falls back to useRef(new Date())). Worst case if loader still
doesn't fire: same hydration drift as before, no crash.
Same pattern as OnlineBoard: route loader supplies todayYyyymmdd() once
on the server; FlightsMapStartPage threads it through useMemo dep arrays
for searchParams + calendarParams so SSR and client hydration agree on
the same dateFrom/dateTo values.
Removes the local todayYyyymmdd() copy in favour of the shared util.
Route loader at src/routes/[lang]/onlineboard/page.tsx computes today's
yyyyMMdd once on the server. Result rides _ROUTER_DATA into the client
bundle, so the first hydration render sees the same value the SSR render
saw — no diverging new Date() calls during render.
OnlineBoardFilter accepts an optional today prop; getBoardMinDate /
getBoardMaxDate take a base Date instead of calling new Date()
themselves; the four todayIso() callsites read the precomputed
todayIsoStr. Existing tests omit the prop and use a fresh new Date()
fallback (captured once via useRef) — back-compat preserved.
Adds three pure helpers to src/shared/utils/datetime: todayYyyymmdd(),
yyyymmddToDate(), yyyymmddToIso().
Triage doc: docs/superpowers/specs/2026-04-27-ssr-hydration-fix.md
(Step 1, OnlineBoard. FlightsMap to follow in next commit.)
Pulls in 13 modified + 4 new source files that were uncommitted on main
when this branch forked. Without them, ScheduleStartPage.test.tsx fails
4 tests against the committed main state, which would mask real
regressions during the CI/CD pipeline rollout.
Source files only — no test infra or pipeline code. The user's main
checkout still owns these changes; this commit will dedupe naturally
once the branches reconcile.
- ScheduleDetailsPage: drop shiftYmd helper and selectedYmd local —
both were left over from the removed day-sibling search path.
- ScheduleFlightBody.test: drop fireEvent import + FUTURE_340D /
FUTURE_1H / YESTERDAY / todayUtc constants; they belonged to the
Buy/Status button tests that moved to the summary-header layer.
- flight-details + error-handling integration tests: mock
useCityName / useStationDisplayName so OnlineBoardDetailsPage can
render without an ApiClientProvider wrapping — the station lookup
hooks now transitively depend on useApiClient via the cached
useDictionaries fetcher introduced in 7deb46a.
On a typical page the console showed 25-30 duplicate 'Failed to load
resource' errors because every consumer hook fired its own copy of
the same network request:
- useDictionaries: once per `useCityName`/`useStationDisplayName`
call (6-10x per render across StationDisplay, PopularRequestItem,
mini-list rows, etc.) — now a module-level WeakMap<ApiClient>
single-flight cache returns the same in-flight Promise.
- usePopularRequests: same pattern across start-page and search-
history dropdowns — cached via the same mechanism.
- useAppSettings: 7+ callers — cached.
Dropped console error count on /ru-ru/ from 29 to 5 (the remaining 5
are WAF 403 infra issues from the dev:full proxy cookie, not code).
Also updates e2e specs:
- schedule-details-mini-list-scoped: asserts the new single-card
rail behaviour (was still checking for the old 3-row flat list).
- smoke /xx/smoke: targets `[data-testid=error-page-404]` instead
of `text=404` — the latter matches both the <title> tag (hidden
by user-agent CSS) and multiple DOM nodes, tripping strict-mode.
Left rail previously rendered the open flight PLUS its day-±1
siblings from a route search. For a connecting itinerary the three
rows were visually identical (Moscow → Murmansk on the same times),
so users read them as duplicates. Angular's schedule-flights-mini-list
only shows a multi-day accordion when schedule.length > 1 and falls
back to a single-card view otherwise; mirror that by always passing
an empty flights[] to ScheduleFlightsMiniList — it shows only the
synthesized open flight.
Header-left drops the Онлайн-Табло / Расписание / Карта полетов
tab strip; Angular's schedule-flight-details-view only slots
<details-back>, so the top 'Вернуться к Расписанию' link is the
single navigation affordance on the details page.
FlightsMiniListItem now joins _childFlightIds as 'SU 6188, SU 6341'
for connecting itineraries — Angular's flights-details-list-flight
surfaces every leg's number in the rail label, not just the primary.
Removes the day-sibling useScheduleSearch call + the miniListFlights
filter memo + PageTabs import + pageTabs JSX — all unused now.
React templated the booking URL as '${locale}-${locale}', which
produced 'sb/app/ru-ru-ru-ru' for a BCP-47 'ru-ru' prop (our router
emits locales in BCP-47 form). The resulting link 404'd on the
Aeroflot booking tool.
Angular's BuyTicketLogic.getLink hardcodes 'sb/app/ru-ru' regardless
of the current UI language; do the same. The 'locale' prop is kept
optional on BuyTicketButton for backward-compat with existing
callers but is no longer consumed inside the URL builder.
Each flight card and the Пересадка strip are now sibling elements
inside .schedule-details — each flight in its own <section class="frame">,
the TransferBar standalone between them. The shared outer frame
wrapper is gone, so the dark page background shows through the
between-block gaps instead of one continuous white surface.
That produces the 'three separate white cards on dark bg'
visual Angular uses for a connecting itinerary (flight 1 frame |
Пересадка | flight 2 frame) — 40px white margins that previously
bled into the surrounding frame disappeared because the flex gap
now renders against the actual page background.
The Пересадка wrapper bumps to $space-xxl (40px) margin: at 15px
the strip blended with the regular between-card gap from
`.schedule-details { gap: $space-l }` and read as a sibling card
rather than a separator. 40/40 mirrors Angular's breathing room.
FlightActions gains a `forceBuy` prop that bypasses the
canBuyTicket() status/window gate. The schedule summary passes it
because the Buy pill there is a generic 'open the Aeroflot booking
tool for this route' affordance — the user picks any date in the
booking flow, so hiding the button on a specific day's 'Cancelled'
status (as the Onlineboard detail page does) loses a useful entry
point. Board detail pages still pass the default (status-gated).
Between two flight cards on a connecting itinerary the TransferBar
used to sit flush against the preceding FlightSchedule block and the
next flight's header — no breathing room, no edge, read as a shared
card rather than a separator. Now the strip is wrapped in a
'__transfer' div that adds '$space-l' top/bottom margin and gives
the inner .transfer-bar the same 'border-radius' Angular uses for its
'.transfer-bar--separated' card variant.
Connecting itineraries now render details-header-badge with the small
round airline icon (36×36) from Angular's `[round]="isConnecting"`
path and drop the 'Авиакомпания' caption, so the SU 6188 + SU 6341
row sits compactly next to the share/buy/last-update cluster instead
of stretching two wide wordmarks across the summary.
Share + Buy buttons removed from ScheduleFlightBody — Angular's
`flight-schedule-details` wires `[share]=false [buy]=false
[print]=false [details]=false [register]=false` into its inner
flight-actions, so a per-leg action strip was never meant to exist.
The page-level summary header now owns those affordances.
OperatorLogo.scss: override the 180×46 rule inside .details-header-badge
when the logo carries .operator-logo--round so the connecting-summary
badge doesn't force a wide wordmark.
BoardDetailsHeader.scss is imported from DetailsHeaderBadge.tsx so
consumers (schedule details summary) that use the badge without the
full BoardDetailsHeader wrapper still pick up flex/gap/typography.
The schedule details page now renders Angular's <schedule-details-header>
summary block (badges per flight + share/last-update + full-route
timeline) between the day-tabs strip and the per-leg cards, so a
connecting itinerary like SU 6188 + SU 6341 surfaces both flight
numbers and the combined Moscow→Murmansk timeline up top instead of
jumping straight from the date tabs to the first-leg detail card.
Mini-list duplicate fix: when the sibling search returned 0 matches
the fallback path used to leak the URL-parsed per-leg breakdown into
the rail, producing a first-leg-only row stacked next to the
synthesized combined row. Now the fallback is empty — the mini-list
just shows the (synthesized) current flight on its own.
FullRouteTimeline now uses the API's pre-formatted .localTime instead
of the full ISO .local, so 00:30 / 02:00 shows up instead of
2026-04-26T00:30:00+03:00.
useAppSettings.buyTicketMaxHours: parse <n>d as well as <n>h (Angular
ships 330d for buyPeriod.max). Without this the Buy button hides for
any flight more than ~3 days out.
Plumbed sortMode/onSortChange/hideColumnHeaders through DayGroupedFlightList
so the sticky ScheduleColumnHeaders and the inner list stay in sync
(removes 2 TS errors in ScheduleSearchPage).
Per the Суббота/Воскресенье/Понедельник headers added an extra
click and zero information — every FlightsMiniListItem already
carries its own date. Replace the per-day accordion wrapper with a
straight chronological column. Always merge the open flight into
the rendered list (the open flight loads via a separate details
endpoint and may not appear in the [-1, +1] sibling search). Strip
the now-orphan day-header / day-body SCSS rules and rewrite the
unit tests to assert the flat-list behaviour.
Mirrors Angular's CurrentScheduleService.getScheduleType +
compareFlightsByPId: when the [-1, +1] route search returns the
open flight (matched by carrier+number signature, including each
leg of a connecting itinerary), keep only those instances; when
no match exists, fall back to a 1-item list with just the open
flight (Angular's 'default-schedule' branch). Old behaviour
returned the full route search and dumped every unrelated MOW-MMK
option into the rail.
Add e2e regression that loads the SU 6188 + SU 6341 itinerary and
asserts the rail shows only SU 6188 — not SU 6190 / SU 6699 (the
other Sunday MOW-MMK options that used to appear).
Probed the live Angular page and the schedule-details Борт row uses
`<svg><use xlink:href="/assets/img/sprite.svg#company">` — a
stroked side-view jet silhouette — NOT the simpler top-down plane
glyph from toolkit/icons/plane (that one only appears in the row
indicator next to flight times). Port the seven #company paths
from sprite.svg verbatim into the leg-details panel so the icon
matches the legacy app exactly.
The plane and food row icons in ScheduleLegDetails were custom React
SVGs (a cargo-airliner side-view and a knife+fork glyph) that didn't
match the legacy app. Copy the actual paths verbatim from
ClientApp/src/app/toolkit/icons/plane and .../dining so the
React panel shows the same stylised top-view plane and the cloche-
on-base 'Питание на борту' icon Angular ships. Use currentColor on
both fills/strokes so the existing $blue link colour still applies.
Reserve a 36px slot in __icon so the rows line up despite the
plane (19×19) vs dining (34×26) intrinsic-size mismatch.
Angular's flight-details-meal.component.html renders each
Эконом / Комфорт / Бизнес sub-icon under *ngIf=hasEconomyMeal etc —
flights with no meal data show just the cutlery icon and caption
with no class pills. React was hardcoding all three regardless of
data, so SU 6188 (whose API returns meal=[]) showed three meaningless
icons; SU 6341 (meal=[Comfort, Economy, Business]) showed the right
ones by accident.
Read leg.equipment.meal, build a Set<MealType>, render each pill
only when its type is in the set. Add a unit test covering empty,
partial, and full meal data and an e2e regression on the live
MOW→LED→MMK itinerary (test asserts SU 6188 has none, SU 6341 has
all three). The e2e depends on backend data and can flake when the
dev proxy WAF cookie has expired.
API returns daysOfWeek.flight as a string of ISO weekday digits
("1"=Mon.."7"=Sun) where each character is the number of one
operating day, NOT a 7-char position bitmask. E.g. SU 6188's
flight value is "156" → Mon + Fri + Sat operating; SU 6341's is
"1234567" → daily. The old reader treated it as bitmask and only
checked position[i]==='1', so SU 6188 highlighted only Mon and SU
6341 highlighted only Mon — the highlighting looked random
relative to Angular which uses the digit-list semantics.
Walk the input character-by-character, build a Set<weekdayNumber>,
mark badge active when its day-number is in the set. Defends
against non-digit characters and out-of-range digits. Rewrite the
unit tests to match the real wire format and add a regression
case for the SU 6188 "156" pattern.
handleFlightClick previously emitted only the first flight ID even
when the row was a connecting itinerary, so the details page only
showed leg 1 (e.g. SU 6188 Moscow→St-Petersburg) and dropped leg 2
(SU 6341 St-Petersburg→Murmansk). Walk `_childFlightIds` instead,
interleaving each leg's airport codes around its segment so the
output URL is /schedule/{depAir}/{flight1}-{date}/{midAir}/
{flight2}-{date}/{arrAir}?request=… — the splat route already
parses any number of segments and the details page already maps
over flights[], so both cards + the Пересадка transfer bar render
correctly.
Add an e2e that clicks a connecting row, asserts the multi-segment
URL pattern, the two .schedule-details__flight cards, and the
Пересадка bar. The test depends on live backend data so it can be
flaky in environments where the dev proxy cookie has expired.
The Schedule row had been switched to navigate-on-click in commit
a26adad as a forward-looking implementation of TIRREDESIGN-4. Angular
on the live test stand still uses inline-expand (verified 2026-04-23
on flights.test.aeroflot.ru — clicking a row toggles
.flight-list-item.selected and renders schedule-search-result-flight-
body / connecting-flight-body inline). React must not lead Angular —
restore the inline-expand wiring so both stays in lockstep.
Drops the schedule-specific branch in FlightList that disabled
expandable and wired onClick to navigate. The expand-via-onFlightClick-
or-renderExpandedBody rule applies uniformly to Board and Schedule
rows again, exactly like before commit a26adad.
gnezim