Polish pass on the survey popover:
- Entrance: slide-up + fade on mount (300ms ease-out, respects
prefers-reduced-motion via motion-reduce:transition-none).
- Step progress: "1 / N" indicator on multi-question surveys so users know
how much is left.
- Esc closes the card (counts as a dismissal); listener attached once with a
dismissRef so it captures current answers, not a stale first-render closure.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Design review of the live survey popover surfaced three issues:
1. NPS 0–10 wrapped to two rows (flex-wrap + flex-1 made the 8/9/10 row
stretch absurdly wide) — now a single equal-width grid row (grid-cols-N
by scale; static map so Tailwind keeps the classes).
2. Single-question ratings forced a redundant "Devam" tap — tapping a score
now auto-advances (450ms debounce to allow changing the pick) and the
submit button is hidden on rating questions.
3. On mobile the card crowded the Chatwoot launcher — bumped to bottom-28
(sm:bottom-24 keeps desktop unchanged).
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Remove the EAN line from the matched-article cards and stop fetching/
shipping eans in /p/oem (its only consumer was that line). Snapshot
still holds the data; the endpoint just no longer queries it.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
OEM detail "slow load" was two things, not the query (DB 135ms / API
13-120ms):
1. The parts-panel link opened a NEW TAB → full SPA cold boot every
click. Switch to in-app client navigation on plain click (real href
kept, so ctrl/cmd/middle-click still opens a new tab).
2. The /p/oem response shipped each article's oeNumbers + compatible
lists (up to 200 each × 60 articles) that the UI never renders — 96%
of a 370 KB payload. Ship lean articles; aggregates already carry
the cross-refs. 60-article code: 370 KB → ~12 KB.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
trial_urgency_banner_viewed fires ~1s into the page load and raced the lazy
surveys extension — events captured before its hook exists can never activate
an event-triggered survey (affects real users, not just e2e). capture() now
awaits surveysReceiverReady() (onSurveysLoaded, 3s safety timeout) and runs the
eligibility check after the event has gone through.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Long cross-reference lists (some parts have 90+ aftermarket numbers,
160+ OE codes) were hard to scan. Add a filter input that matches
across all three sections (matched articles, aftermarket equivalents,
OE cross-refs) by brand (substring) or code (normalised, so "1j0 973"
finds "1J0 973 702"). Clear button + "no results" state.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
posthog-js silently drops every capture when userAgentData.brands contains
HeadlessChrome (UA string/webdriver masking doesn't help), which also starves
event-triggered surveys in Playwright runs against dev.sase.tr. Production
keeps the bot filter.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Formbricks CE turned out to gate person-based targeting (setUserId/attributes
→ 403 enterprise) — the exact flexibility surveys need. PostHog already holds
the person properties, so surveys move back there in API mode: posthog-js
evaluates eligibility (event triggers, targeting flags like
subscription_status=active, wait periods, per-distinct_id dedup) via
getActiveMatchingSurveys, and we render the popover ourselves — zero PostHog
branding, sase.tr dark-theme styling.
- lib/surveys.ts: display manager + capture payload builders that mirror
posthog-js's own "survey shown/dismissed/sent" shapes exactly
($survey_response_<qid>, $set $survey_dismissed/<id>[/iter], seenSurvey_*,
lastSeenSurveyDate) so the PostHog Surveys results UI works unchanged
- components/survey-popover.tsx: single_choice (+Diğer), open text, rating/NPS
- posthog.ts: capture/pageview/identify now schedule survey checks; register
deploy_env (dev.sase.tr ships the key now → staging traffic is filterable)
- remove @formbricks/js, its CSP entries and build args
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Every OEM code was a link → unmatched codes (the majority — fasteners,
clips, body parts TecDoc doesn't carry) opened an empty "no
equivalents" page. Now the schema page batch-checks its codes against
the snapshot (POST /p/matched, one indexed query) and links ONLY the
matched ones; unmatched codes render as plain text. Copy stays for all.
No more dead-end empty pages from the parts panel.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- cover_file uploads in Directus land in the public blog-assets MinIO bucket;
API resolves them to storage.sase.tr URLs (Directus itself is Tailscale-only)
- POST /blog/cache/purge (automation token) lets a Directus Flow drop the
30-min list cache the moment a post changes in the CMS
- blog list + detail pages now render purely from the API; the 4 hand-authored
posts were migrated to Directus earlier and the JSX copies are removed
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
@formbricks/js is a loader shim that injects the real SDK as an external
script from anket.sase.tr — connect-src alone wasn't enough; the browser
blocked /js/formbricks.umd.cjs and no survey could render.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Decode persist looked brands up with a case-sensitive eq(), so an
uppercase decode string ("FORD") missed canonical "Ford" → brand_id
NULL + raw uppercase stored as brand_name, splitting one brand across
casing variants in analytics/catalog. Now matches brands
case-insensitively and stores the canonical name. Migration
0013_fix_brand_casing backfills existing rows (60 on prod, 1 on dev).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
"P karşılığı bulunamadı" / "P kapsamında" read as nonsense — "P" is
the internal name, not a user-facing word. Use plain Turkish: "Bu OEM
kodu için uyumlu parça bulunamadı" and describe the catalog gap
without a brand or single-letter label.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Finishes 3b14431 (which only captured the file renames): apply the
identifier/endpoint/env/UI changes so the code matches the new paths
— PModule/PController/PSourceDbService, @Controller("p"), /p/oem,
config key `p`, P_DB_ENABLED/P_DB_URL, "P kataloğundan…" copy.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Drop the third-party brand name from the product surface and code.
The OEM detail feature is now "P" (short for parts) everywhere:
endpoint /p/oem, PModule/PController/PSourceDbService, config key
`p`, env P_DB_ENABLED/P_DB_URL, UI copy ("P kataloğundan…"), and the
unused tab label. Physical snapshot DB stays `td` (server-only).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Blog posts now live in the shared Directus instance (Coolify / Süper Panel
project) instead of the per-env blog_posts table, so prod and staging serve
identical content. API response shape is unchanged; Redis list cache and the
n8n automation endpoint keep working as before.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
The snapshot stores scrape-local '/_debug/...' image paths that 404
off-host. Filter the OEM lookup to http(s) image URLs and render the
thumbnail only when one exists, so the detail page shows clean
code-focused cards instead of a grid of broken-image placeholders.
Real CDN URLs will surface automatically once present.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Full marketing-copy audit. Removes price-comparison claims (pricing
step, hero variant B, blog tip) that aren't built; normalises VIN→şase
across TR copy (plan features, e-commerce section, FAQ, blog, about) —
keeping VIN only in the educational "VIN nedir" post and the EN file;
de-consumerises framing (FAQ "aracınız", about "bireysel kullanıcılar").
Fabricated metrics/testimonials left untouched pending review.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Resolve a catalog OEM code to its TecDoc equivalents on a new
/dashboard/oem/$code page: the aftermarket parts that carry it
(brand + article number + image + EAN), buyable supplier
substitutes, and OE cross-references (same part under other makes).
- API: TecdocModule (read-only postgres-js client to the imported
`td` snapshot), GET /tecdoc/oem?code=. Normalisation-based match
(TecDoc stores `1J0 973 702`, catalog gives `1J0973702`); exact
match recovers ~1/10 vs normalised ~5/10 on real codes. Self-
disables without TECDOC_DB_* env → { matched: false }.
- Web: OEM code in the parts panel is now a link (new tab) to the
detail page; "N/A" stays plain text.
- Mirrors CatalogSourceDbModule (raw queries, no Drizzle modelling).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The landing OEM card, the manual-vs-Sase comparison table, and a blog
post advertised "price comparison" / "order history & part tracking" —
features that aren't built. Replaced with real capabilities
(exploded-diagram matching, interactive part diagrams, brand catalogs)
in both tr + en. Mirrors the same accuracy fix already shipped to the
Novu lifecycle emails.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Only ~1.6% of trials reach checkout: the 30-day full-access trial shows the
(urgency) banner solely in its final 7 days, leaving ~3 weeks with no in-app
push, and the value moment (first few queries) goes unconverted.
Add TrialValueUpsell — a calm, value-framed banner that fires on PROVEN VALUE
(user has decoded >= 3 customer vehicles, read server-truth via /vehicles/history)
and only OUTSIDE the urgency window (days > 7), so it fills the gap without ever
overlapping or competing with the urgency banner. B2B copy: frames around
"müşteri araçlarını sınırsız sorgulama", never personal-vehicle wording.
Tracks viewed/cta_clicked/dismissed in PostHog; dismissal persists per trial.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- initSentry() is now idempotent via a cached promise, and the degradation
reporter / feedback opener await it — so an empty/error state that renders
before the SDK's dynamic import resolves no longer drops the event (not missing
the failure is the whole point of the auto-capture).
- environment now resolves hostname-first (sase.tr → production, dev.sase.tr →
staging), so the compose-baked VITE_SENTRY_ENVIRONMENT default ("production")
no longer makes dev browser events indistinguishable from prod.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
handleCheckoutCompleted marked the payment 'completed' BEFORE activating the
subscription, but the idempotency guard keys off payment.status === 'completed'.
So if activateSubscription threw after the payment was marked completed, the
Stripe webhook retry would hit the guard, skip activation, and leave a
paid-but-never-activated subscription.
Fix: activate FIRST, then mark completed — a mid-activation failure leaves the
payment 'pending' so the retry re-runs cleanly. Guard activateSubscription with
an idempotency check (return early when already active) so a retry after a
successful activate but failed mark-completed cannot re-fire revenue events
(PostHog subscription_activated + Meta CAPI Purchase), re-consume referral
credit, or double-insert Full-plan brands. Admin manual-activate already
pre-rejects 'active', so legitimate first activations are unaffected.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
EFT/Havale was retired; Stripe is now the sole payment method. Remove the
EFT code surface (shared PaymentMethod "eft" + EftPaymentInput + eftReceiptUrl,
EFT_RECEIPT_REQUIRED error code, billing UI receipt/filter/label paths,
payments.service eft read paths). DB columns (eft_receipt_url, bank_account_id,
bank_accounts) are kept and marked @deprecated to preserve historical records
and avoid a destructive migration — same pattern as the retired iyzico column.
Faz 3 conversion lever: set locale "tr" on the Stripe Checkout session. The
audience is Turkish B2B and ~60% of sessions reached the foreign-language
hosted page but never started a payment intent (pure abandonment, not decline).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The two things only the browser can add on top of the backend catalog-degradation
reporting:
- Replay-on-failure: when the catalog UI renders empty-tree (decoded vehicle, no
categories) or a drill loadError, capture a browser Sentry warning and flush the
Session Replay → you can WATCH the user hit the dead-end (serkan's session,
reproducible). Per-session deduped (one replay/session covers the whole journey).
- In-app feedback: a "Çalışmadı mı? Bildir" button on the empty-parts, empty-tree
and loadError states opens the Sentry feedback dialog pre-tagged with the
vehicle/category (+ session replay) — turns a parts shop's complaint into a
structured, triageable report instead of an email.
Browser events are fingerprinted source="browser" so they form their own
"what users actually saw" issues (carrying replays) next to the server-side
detections. tsc + biome + web build clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Paid ads pull many end-users who trial but never pay (~1% trial→paid in the DB
funnel). Add a soft qualifier before the register form: the visitor picks their
segment (parts dealer / wholesaler / e-commerce / service-fleet / vehicle-owner).
Business segments proceed; "vehicle owner" gets a gentle B2B-only note with a
"continue anyway" escape hatch (soft, no hard gate — protects trial volume).
The choice is captured (signup_segment_selected event + b2b_segment person
property) so qualified-trial ratio is measurable and channel→segment→paid becomes
queryable; later it can seed a Meta audience / CAPI qualified signal. Copy follows
the B2B framing rule (no "kendi aracınız"; "sınırsız şase sorgulama"). Persisted in
localStorage so a reload / OAuth round-trip doesn't re-ask.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
dev.sase.tr and sase.tr both run NODE_ENV=production (and ship the same
production Vite build), so every Sentry event — api, worker, and browser — was
tagged environment=production, making dev errors indistinguishable from prod.
Resolve the environment from the canonical prod host instead: COOLIFY_FQDN /
BETTER_AUTH_URL on the server (the same signal isCatalogBackfillEnabled uses),
window.location.hostname on the web. So dev now tags as "staging" and stays
filterable. Explicit SENTRY_ENVIRONMENT / VITE_SENTRY_ENVIRONMENT still win.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The existing CAPI only sent CompleteRegistration (signup). For a B2B funnel where
trials are cheap (~25 TRY) but paid is rare (~1% of trials), the highest-value
signal Meta can optimize on is the realized-revenue Purchase. Add
MetaCapiService.sendPurchase and fire it from activateSubscription — the shared
chokepoint for BOTH Stripe (webhook) and EFT/manual activation — so all paid
revenue is sent regardless of method. Hashed-email Advanced Matching (no browser
fbp/fbc in the webhook); event_id = purchase_<subscriptionId> dedupes a browser
Purchase. Awaited so it ships before the short request returns; fail-open.
This is the "teach Meta to find payers, not end-users" lever from adsOpt.md Phase 0.
Still gated on activating CAPI in prod (merge + META_CAPI_PIXEL_ID/ACCESS_TOKEN env).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The subscription-expiry cron filtered status="active" only, so trials
never transitioned out of "trial" after end_date. Every access gate
keys off status, so trials past end_date kept full product access for
free (revenue leak) and inflated the active-trial count. On prod: 105
stuck trials, 3528 brand grants still live.
- expiry processor now sweeps status IN (active, trial) past end_date
(lt() still skips NULL end_date, so perpetual subs are untouched)
- add "trial" to SubscriptionStatus union — it was used in the DB and
code but missing from the type (both subscription.ts and user.ts)
Proven read-only on prod: old WHERE caught 0, fixed catches 105.
Revocation uses the existing set-expired + delete-userBrands path.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Mock @sentry/nestjs and assert that a getCategoryWithParts loadError drives the
redis dedup key + Sentry.captureMessage("…drill-load-error…"), so the silent-
failure reporting can't regress unnoticed.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The catalog failures that hurt UX — a drill/parts fetch that fails into an empty
"couldn't load" panel, or a decoded vehicle whose category tree comes back empty
("model var ama parça yok") — all return HTTP 200 with a degraded body. Nothing
throws, so the global Sentry exception filter never sees them and they go
unnoticed (serkan's complaint was exactly this class). Report them explicitly.
- new common/catalog-degradation.ts: reportCatalogDegradation(kind, ctx),
fingerprinted by kind+source+brand so each failure mode collapses into one
countable Sentry issue (e.g. "drill-load-error · pl24/Ford — N events, M users").
- categories.service: capture on getCategoryWithParts loadError and on an empty
getCategoryTree, Redis-deduped to <=1 event/hour per category/vehicle so a
broken catalog can't flood the stream; telemetry never throws into the request.
tsc + biome clean, categories suite 10/10.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
payment_success / payment_failed / subscription_activated are captured inside
the Stripe webhook handler and activateSubscription — short requests that return
immediately. posthog-node's fire-and-forget flush was abandoned before the send
completed, so these events were written to the DB but never reached PostHog
(DB had 4 completed Stripe payments in 30d; PostHog had 1 payment_success and 0
payment_failed). payment_initiated, fired in a normal user request, landed fine —
which is what isolated the cause to the webhook/short-request context.
Add PostHogService.flush() and await it at the end of handleWebhook and after the
subscription_activated capture in activateSubscription (the shared Stripe+EFT
chokepoint). Restores server-side paid-conversion visibility so trial→paid ROI is
measurable in PostHog instead of only the DB.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
isLeafLinkPath flagged EVERY pcat node as a leaf (any pcat: linkPath), so the
backfill worker fetched a mid-group folder's "parts" — a 1-level drill that
only revealed its sub-groups — and never recursed. Deep pcat trees (whose top
level is ~100% folders) were left one level shallow: no parts seeded, so the
new tree part-counts stayed 0 and the cross-tree catalog search (DB-only) found
nothing on a freshly decoded vehicle. Exactly why serkan's Ford Mondeo looked
like an empty catalog.
Use the captured hasSubgroups flag: a pcat node is a parts leaf only when it is
NOT a known parent group. Folders now queue a children job and the existing
recursion (queueCategoryJob → processChildren → getChildren) drills to full
depth (MAX_DEPTH=5), seeding parts at every leaf. The rolling backfill rescan
warms existing shallow vehicles wave by wave; new decodes warm deep via the
reactive schedulePrefetch. Rate limiter + PCAT_PACE + backlog guard keep it
bounded (all env-tunable to ramp).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
pcat/emex vehicles decode into a tree whose top level is ~100% folders with
zero parts (6746/9458 root nodes are parents, 0 direct leaves across 477
vehicles). Users — serkan filazi among them — saw an opaque wall of identical
folder cards, assumed the parts feature was missing, and bounced without ever
drilling (his session: 3 view-mode toggles, 0 category clicks; his data was
2-3 clicks deep and fully reachable).
Two changes, no upstream/proxy cost:
- getCategoryTree now rolls up a subtree partCount per node (one grouped parts
read + JS post-order sum). The grid already renders "N parça", so folders
now read "Fren sistemi · 142 parça" and steer drilling toward the branches
that actually contain parts; empty branches stay quiet. Counts reflect
drilled data and grow as branches are visited.
- A one-line drill hint on a "cold" tree (nothing drilled yet → no counts) so
first-visit users know parts live inside the subcategories instead of reading
the folder wall as a broken/empty catalog. Drops away once any branch is hit.
Verified: subtree roll-up unit test + existing suite (10/10), tsc + biome clean
across api/web/shared.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The Meta Pixel (fbevents 2.9.x) dispatches ALL events from a Web Worker created
off a blob: URL. With no worker-src directive, the browser fell back to
script-src (which lacks blob:), so the worker was CSP-blocked and NO pixel
events fired — PageView, CompleteRegistration, everything were 0 in a live
headless test (even a manual fbq('track','PageView') emitted nothing).
connect-src/img-src already allowed facebook.com, so this was the sole
remaining client-side blocker. Adds worker-src 'self' blob: (also unblocks
Remotion's blob worker on the landing demos).
Note: CAPI (server-side) already covers the signup conversion regardless; this
restores the browser pixel for PageView + retargeting/audience events.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
VIN-decoded Peugeot/Citroën scope categories ("mekanik", "kaporta", …)
carry json-vin-main-groups.action / json-vin-illustrations.action link
paths, but getCategoryWithPartsInner's isPsaParent gate only recognised
the catalog-browse paths (psa:: / json-illustrations.action). So every
VIN PSA parent fell through to the leaf path, fetched no parts, and
rendered an empty panel — the largest current "0 parça" cluster
(~20-30 empty parts_panel_viewed/day across all Peugeot/Citroën models;
serkan filazi's complaint among them).
The drill already works end-to-end (getChildren → fetchSubGroupsByPath →
fetchVinMainGroups/fetchVinIllustrations → vin-image-board parts); only
the gate was out of sync with PL24Service.isPsaVin*Path. Verified live on
prod: scope "mekanik" → 6 main-groups → "motor" → 10 illustrations →
12 OEM parts + schema + 15 hotspots.
Also surface a retryable loadError (matching the pl24/emex group-node
branches) when a PSA parent drill comes back empty, instead of a
misleading empty grid. +2 regression tests pin the dispatch to the
PL24Service.isPsaVin* helpers so it can't silently drift again.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The browser pixel under-counts signups badly: ~96% of paid traffic is mobile
in-app browsers where iOS ITP / ad-blockers drop client events, and the OAuth
path never fired it reliably. Meta recorded ~0 registrations for a 7.5K-spend
campaign while PostHog saw 98 facebook signups — so Meta could neither optimize
toward nor attribute signups, which is the main driver of the low signup rate.
This adds a server-side CAPI CompleteRegistration:
- MetaCapiService + @Global module. Fail-open: no-ops unless META_CAPI_PIXEL_ID
+ META_CAPI_ACCESS_TOKEN are set; never throws (signup must not break).
SHA-256 hashed email + fbp/fbc/IP/UA.
- Fired from the better-auth user.create.after hook for ALL signups (reliable,
covers Google OAuth which the browser pixel missed entirely).
- A session-gated POST /analytics/meta/complete-registration endpoint adds
fbp/fbc/IP/UA (ad-click attribution) for the email path.
- The browser pixel now passes a shared event_id (signup_<userId>); the
premature Google client-pixel fire (fired on click, before completion) is
removed.
- All sources dedupe via event_id=signup_<userId>.
Activate by setting META_CAPI_PIXEL_ID + META_CAPI_ACCESS_TOKEN (Events Manager)
in the api env; META_CAPI_TEST_EVENT_CODE routes to Test Events for verification.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Meta Ads traffic converts to signup well below expectation. Same design, three
hero-copy variants (lossPill + title + subtitle) behind the exp-landing-copy
multivariate flag, measuring user_signed_up (experiment 83160):
- control: current "find the right part instantly" (feature / accuracy)
- variant_a: "end wrong-part returns" (ROI / business-outcome + risk reversal)
- variant_b: "find it in seconds, free, no card" (risk reversal / signup friction)
useFeatureFlag on the index hero, undefined -> control. B2B-safe copy
("customer vehicle", not "your car"). Stays draft until prod, then launch.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
New users land on an empty search box with only a tiny fill-only "try example"
link; activation (first vin_decode_success) sits at ~32%. This adds a prominent
one-click "decode a sample vehicle" card that autoDecodes the sample VIN ->
straight to the vehicle page (the aha-moment).
Gated behind the exp-activation-guided-decode multivariate flag (control/guided),
read ONLY for unactivated users (empty history) so power-user decodes don't
dilute the metric; undefined -> control (status quo). PostHog experiment 83153
measures vin_decode_success. B2B-safe copy (sample vehicle, not "your car").
The experiment stays in draft until this ships to prod (web flags need the prod
VITE_POSTHOG_KEY), then launch.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
So a non-prod env can hold the (public) project key and evaluate flags - kill
switches, rollout - WITHOUT shipping analytics events to the shared prod
PostHog project. The posthog-node client (which powers both capture and local
flag eval) now initializes whenever the project key is present; capture() is
gated on POSTHOG_CAPTURE_ENABLED (default on; set "false" on dev). Wired into
the api + worker compose blocks.
This lets dev.sase.tr exercise the kill switches end-to-end while keeping the
prod analytics project clean.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add server-side PostHog feature-flag evaluation to the API and wire three
high-leverage uses. The flags live in PostHog (project 127747), dormant and
fail-open, so this commit is a no-op until POSTHOG_PERSONAL_API_KEY is set and
a switch is flipped.
Phase 1 - upstream kill switches:
PostHogService gains local flag evaluation (onlyEvaluateLocally +
sendFeatureFlagEvents:false -> zero decode latency, no VIN leak) and
isSourceLive(). Guards at each source's decode entry (parts-catalogs, emex,
pl24 whole-source + per-brand via LEGACY_ARCH_SOURCE_TAG) let a flailing
upstream be disabled from the PostHog UI in ~5s instead of a
code-fix -> dev -> prod redeploy. Fail-open: any unresolved flag keeps the
source live, so a PostHog outage can never black out decoding.
Phase 2 - guarded rollout primitive:
isEnabled()/variant() with VIN bucketing, ready to ramp a new decode/parser
path 0->100% behind a decode-*-v2 flag (recipe in feature-flags-strategy.md).
Phase 4 - remote-config ops tuning:
prefetch-worker reads cfg-backfill-tuning to retune backfill
batchSize/maxBacklog/businessHoursOnly live; malformed/missing -> the
compiled-in constants.
POSTHOG_PERSONAL_API_KEY wired into the api + worker compose blocks (empty ->
flags inert, no added latency). Tests updated for the new constructor params.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Both the catalog flow ($brandName/$modelId) and the VIN-decode flow
(vehicles/$id) render the same CategoryColumns. It was stuck inside a fixed
visual window: container min-height 320px, each column max-height 420px. With
20+ root categories the user saw an internal scrollbar inside an otherwise
short rectangle even though there was plenty of page room.
Make it flex naturally:
- Outer container: min-h-[24rem] keeps a solid floor for empty/cold state,
max-h-[calc(100dvh-220px)] keeps a 200-leaf catalog from running off the
bottom of the page. Within that band, height tracks the tallest column's
natural content (flex stretch — siblings share the height).
- Inner column panels: drop the 420px max-height. Each column flexes to the
shared height; overflow-y-auto only kicks in when the outer viewport cap
compresses the row.
Net result: small catalogs render a short, tight columns view; large
catalogs grow to fill the available height up to the viewport cap, then
the columns scroll individually. Same component fixes /catalog/.../$modelId,
/vehicles/$id, and the nested /categories/$categoryId drill.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
After the user picks the last meaningful dimension, the next step is determined
— there is no decision left to make. Yet the Ford and PSA selectors stopped at
that point and waited for a "Kataloga Git" / "Proceed" button click. For Ford
specifically this was the worst case: LEGACY_FORD's upstream returns no engines
or gearboxes, so picking a year/catCode IS the whole flow — the user had to
click twice for one decision.
Auto-fire onSelect when the last available dimension is filled in:
- **Ford**: when year picked and (!hasEngines && !hasGearboxes); when engine
picked and !hasGearboxes; when gearbox picked. Also: when config returns
empty (no variants at all), fire onSelect("_nor_", "_nor_", "_nor_") via a
one-shot useEffect so the selector skips itself rather than dead-ending on
a "no config" notice with no actionable button.
- **PSA**: when gearbox is picked. Body→engine→gearbox is always 3 steps,
picking gearbox is unambiguous intent. "Show All" buttons already auto-
advanced — this just removes the asymmetry on the explicit-pick path.
Drop the now-dead Proceed button + handleProceed handler in both. P5
restriction selector already auto-advances on `isFinal` and stays untouched.
i18n keys catalog.fordVariant.proceed / catalog.psaVariant.proceed are kept
in case we ever re-introduce a confirmation button.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
Two visible defects when a user picks a Ford model (variant selector step):
1. **N identical buttons.** Ford's `modelFamilyToModelList` gives every sub-catCode
the same caption — just the family name. parseFordCatCodesForFamily returned
`[{code:CBV,name:Kuga},{code:CBS,name:Kuga},{code:CTD,name:Kuga}]` for Kuga
2012-2020. The UI rendered three indistinguishable "Kuga" pills with no way
for the user to tell them apart. Same on Galaxy ("Galaxy"/"Galaxy"), every
multi-catCode family.
Fix: detect duplicate baseNames per family and graft a disambiguator —
year first (friendliest), catCode as a last resort. Result:
"Kuga (2013-2016)" / "Kuga (2016-2020)" / "Kuga (CTD)" — or, when year is
present even for unique entries, always include it for a uniform look.
2. **Section title was lying.** UI label was `catalog.fordVariant.modelYear`
= "Model Yılı" ("Model Year") but for Ford LEGACY these are catCodes
(generation identifiers), not years. Rename the key value to "Varyant" /
"Variant" — semantically correct for both Ford catCodes and Volvo years
(Volvo flows through the same selector and its codes ARE years; "Varyant"
covers both).
Cleared the affected Redis cache (`pl24:ford:config:*`) on dev so the next
hit fetches fresh disambiguated data; prod cache will roll over on its own
TTL after promote.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
The PL24 vehicle page rendered a Card titled `t("catalog.models")` ("Modeller")
whose content was the *selected vehicle's* specs (motor/kasa/vites/pazar). When
all four spec fields were null — common for sparse decodes — the user saw an
empty card with a bogus "Modeller" heading just below the model name.
Replace the misleading card with a flat spec-chip row (icon + label : value),
hidden entirely when no specs exist. Use the shared CatalogHeader so brand →
vehicle has a proper breadcrumb back to "/catalog/$brandName" and uses the same
back-button affordance as the rest of the catalog surface. Use the shared
ViewModeToggle so view-mode buttons stop having hardcoded TR title attrs
("Izgara", "Agac", "Sutun") and inherit the i18n + a11y from the component.
Same cleanup for pcat/$catalogId_/$modelId and emex/$catalogCode_/$vehicleId:
both rolled their own header — both now use CatalogHeader with full crumbs and
i18n'd search/count/empty-state strings (catalog.pcat.*, catalog.emex.*).
Why now: catalog flow audit caught the empty-card bug ("model seçilince
yukarıda boş bir alan kalıyor"), plus drift between drill levels (some pages
used CatalogHeader, vehicle/pcat/emex did not). One unified pattern across
brands → models → vehicle → categories.
i18n: +catalog.categoryCount, +catalog.vehicleSpecs.{engine,body,transmission,
market}, +catalog.pcat.{searchPlaceholder,vehicleCount,schemaCount},
+catalog.emex.{searchPlaceholder,optionCount,variantCount,noResults,
noPartsTitle,noPartsHint,loadError}.
Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>
These three brands rendered parts with NO schema illustration (and no hotspots):
schema_pics=0 vs 56/29/19 parts on dev. Root cause: extractPsaImageTicketUrl only
read imageViewerParamsUrl from the id="jsinitparams" data-params attribute (where
Ford/PSA/Opel/Volvo put it). Hyundai/Kia/Nissan's jsinitparams carries only a
localization dict ({"commonTxt":{"ok":"TAMAM"}}); their imageViewerParamsUrl lives
in a separate <script> JSON blob → extractor returned null → image pipeline skipped.
Fix: when the jsinitparams-attribute parse yields no URL, fall back to a whole-HTML
scan for "imageViewerParamsUrl":"…" (decoding &). Brand-agnostic and additive;
Ford/PSA/Opel/Volvo unchanged (attribute path still wins). Proven live via the de
client: Hyundai 600x820/3 hotspots, Kia 600x820/11, Nissan 1024x560/13 — all download OK.
Also: drop a pre-existing noParameterAssign in fetchP4Page (derive `account` const
from `accountParam` instead of reassigning the param). +3 extractor unit tests.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Fiat catalog browse returned 0 models because the generic single-endpoint P5
model-list flow can't express Fiat's two-level hierarchy and the drill defaulted
to the tr account (Fiat is licensed only on de-708171).
- fetchVehicleList: dispatch fiatp_parts/fiatt_parts to new fetchFiatVehicleList,
which expands modelOverview (34 families) → models?modelFamily=N (model codes +
year ranges) into flat catalog vehicles whose catalogPath is the maingroups
endpoint. Verified live: 34 families → 123 models, end-to-end drill to parts+image.
- fetchMainGroups / fetchP5Restrictions: account-aware (resolveAccount → de + DE
proxy for Fiat; tr unchanged for every other P5 brand) so browse maingroups no
longer hit the tr demo/empty page.
- web: case-insensitive "/maingroup" gate so Fiat (lowercase /mdl/maingroups,
already a maingroups endpoint) skips the empty restriction selector and loads
categories directly.
- formatFiatYear: "(2016,2020)" → "2016-2020"; +unit tests.
Subgroups/parts/images already resolve account→de for Fiat (unchanged).
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Finding: /p5fiat/extern/vehicle/models IS the endpoint (de-licensed, demo:false) but
returns 'Cannot parse null string' regardless of manufacturer/make/brand param — needs
an undiscoverable param from the P5 SPA session (launchCatalog.do flow). Requires a real
PL24 web-UI network capture to resolve. Turkish Tofaş NM4/Egea is not in this catalog
anyway, so catalog-browse value is European-Fiat-only. Kept the generic P5 endpoint
self-discovery fallback. Fiat VIN-decode (European) is unaffected and still works.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
