Commit Graph

13 Commits

Author SHA1 Message Date
julian b4f1418858 fix(entry): enforce the camera press-gate + duplicate-ticket defenses
Field report (park-buzi): a BLINKING entry button still printed — the lamp
encoded blink-vs-solid (radar-only vs radar+camera) but #suppressReason only
checked the radar, so a radar false-positive (rain, pedestrian) minted a real
signed ticket. Three layered fixes:

1. CAMERA gate on the physical press: with an entry camera configured, a press
   is live only in the lamp's SOLID state (LaneStatus.entry busy, mirrored into
   EntryFlow via onLaneStatus). Suppress-only — the camera stays advisory (never
   opens, never traps). Camera-less sites keep the radar-only gate; a faulty
   camera is dropped via the existing bypassPresenceCamera admin toggle.

2. Cooldown as a REAL backstop behind presence: the presence branch returned
   early, so entryCooldownSec was dead wherever a loop was wired. Now it bounds
   the stationary-car double-ticket (a motion radar drops a motionless car →
   spurious loop-clear re-arms one-car-one-ticket → same car reprints).

3. Post-hoc duplicate-plate anomaly (entry-side twin of plateSwapSuspected):
   when entry ANPR recognizes a plate already OPEN under another session entered
   within ENTRY_DUP_PLATE_WINDOW_MIN (default 15 min), sign ONE
   entry.duplicatePlate anomaly naming both tickets for the operator to void.
   ANPR stays non-blocking (rides the post-open snapshot as before).

REJECTED: camera-vetoed re-arm (defer re-arm until the lane flips free). The
camera has no leave events — "free" is a ~30s silence timeout that never lapses
inside a queue, so every queued car after the first would be suppressed until
an operator intervened. Blocking legit entry at peak beats nothing; the proper
preventive fix is a pass-through sensor (passedInput) — recorded as open in
wiki/concepts/entry-double-press.md.

Also: setup.relayTest reason was missing from both web catalogs (parity is only
enforced sq<->en, so the build passed) — added.

Tests: entry-press-gate.test.ts (blink suppresses / solid prints / camera-less
unaffected / bypass honored / cooldown catches the dropout re-press / residual
risk documented / still-present re-press stays suppressed) +
entry-duplicate-plate.test.ts (flags open dup, ignores closed/stale/self/other
plates). Suite 258 green.

Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
2026-07-04 18:41:06 +02:00
julian 6505a4a73b feat(entry): admin bypass of the presence gate for faulty radar/camera
Build desktop / desktop (push) Successful in 4m12s
Build & push images / images (push) Successful in 2m50s
CI / check (push) Successful in 41s
The entry button (physical press AND the operator-issued mint) requires
radar/loop presence + camera detection to confirm a real vehicle. When one
of those devices is faulty, the gate blocks legitimate transient entry. Let
the ADMIN drop a specific signal as a requirement until support fixes the
hardware — the admin is not the adversary, but weakening an anti-fraud gate
stays attributed and auditable:

- Granular: bypass radar and camera independently (Setup → controller
  section). A dead camera drops only the camera check; a dead radar only
  radar. Both off = normal gate; both on = press-to-print.
- Signed: a DEDICATED endpoint (PUT /api/site-config/presence-bypass,
  site:update) appends a signed config_change {setting, value, prev,
  operator} per actually-changed signal — new ledger type. No-op toggles
  sign nothing; disabling signs too. Kept out of the generic site PUT.
- Flagged: every vehicle_entry issued (and every refusal anomaly) while
  bypassed carries presenceBypassed:[...] in its signed payload.
- Persists until turned off; amber warning in Setup while active. The
  booth entry light treats a bypassed signal as satisfied (server
  re-checks authoritatively). Physical-button path falls through to the
  cooldown backstop when radar is bypassed.
- Migration 0020: two boolean site_config columns (default off).

Fixes a latent bug surfaced by the tests: firstRelayByDirection returned no
presenceInput, so issueForOperator's radar gate always read "presence loop
unavailable" — operator-issue never actually gated on radar. The resolver
now attaches the presence input serving the relay (mirrors relayForButton).

10 new tests: 5 gate combinations (each bypass drops only its signal +
records it), 5 route tests (RBAC, signed transitions, no-op, validation).

Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
2026-07-04 16:52:34 +02:00
julian 33c4ea1e91 feat(entry): operator-issued entry + exit plate-swap reconciliation
Build desktop / desktop (push) Successful in 4m29s
Build & push images / images (push) Successful in 2m51s
CI / check (push) Successful in 37s
Two halves of one anti-fraud design.

(A) Operator-issued entry — when the physical entry button is broken, an
operator can issue an entry ticket so a real car isn't blocked out of the lot.
This hands the operator-adversary a mint, so it is:
  - PRESENCE-GATED like the physical button: a real car must be present (radar/
    loop AND camera busy). Enforced BOTH sides — the server re-checks current
    presence so a direct POST can't bypass a disabled button; no presence loop
    => feature unavailable; a no-presence attempt signs an anomaly.
  - FLAGGED: vehicle_entry source=manual + operatorInitiated + operator, PLUS a
    companion entry.operatorIssued anomaly (the adversary path always leaves a
    red-flag row).
  - capacity-OVERRIDE allowed but stamped lotFull (a broken button mustn't trap
    a legit car).
  New session:create permission (migration 0019 -> operator role, admin-
  revocable), POST /api/entry/issue (open-shift gated), EntryFlow.
  issueForOperator; the fraud-critical print->sign->open->snapshot sequence is
  factored into one shared #issueTicket (button + operator). UI: the entry
  BarrierLight becomes a clickable issue-control when presence+permission+shift
  meet (confirm -> issue).

(B) Exit plate-swap reconciliation — defends the ticket-swap fraud the mint
enables (paid car let out on a fresh $0 ticket, original ticket lingers
"inside", occupancy drifts up by phantom cars). The plate is the invariant:
ExitFlow.#reconcilePlateAtExit compares the exiting plate against all OPEN
sessions' entry plates, EXACT + HIGH-CONFIDENCE only (>=0.85; a fuzzy read never
gates — ANPR is advisory). On a match under a DIFFERENT ticket:
  - BOOTH path: returns swap_suspected + signs exit.plateSwapSuspected; the
    pay/exit modal shows a red warning + "Override & release" (override signs an
    attributed exit.plateSwapOverride). Flag+override, never a silent hard block
    (exit fails-open; a plate is never the sole gate).
  - READER path (no operator): log-only anomaly + fail-open.
  Extended BoothExitResult + /api/exit (override); boothExit client returns a
  structured swap result.

Verified: full monorepo build/lint/test green (229 server tests incl. 4 new:
hold-on-swap, override-releases-with-attribution, low-confidence-no-warning,
own-plate-no-warning). New wiki: operator-issued-entry.md +
plate-reconciliation.md; cross-linked from entry-exit-points, capacity-
occupancy, index. Preserves "a plate never OPENS a barrier alone — and now never
TRAPS a car alone either."

Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
2026-07-01 12:17:52 +02:00
julian ecaaefd899 refactor(vision): ANPR rides the entry/exit snapshot, drop polling reader
Rework the ANPR trigger to the real design: when a transient presses the button or a
subscriber passes QR/RFID, the entry/exit fires and takes its evidence snapshot — that
is the moment to recognize. snapshotAsync now takes the VisionClient and, after storing
each snapshot from an opt-in (config.anpr) camera, runs ANPR on the SAME image and
records the plate against the SAME session identity (device_events kind:"read" with
plate/confidence/region/snapshotId/source:"entry-exit-snapshot"). One image serves both
evidence and plate extraction; recognition fires only on a real entry/exit — no polling.

The entry/exit/subscription flows take an optional VisionClient and pass it through;
server.ts wires it. Removed the polling VisionReader and VISION_POLL_MS/VISION_DEDUPE_MS.

Advisory + fire-and-forget: a low-confidence/no-plate result records nothing, a vision
failure never delays or changes the open, and the plate does not feed the access
decision. Verified e2e: a simulated entry snapshot on an anpr camera (live fast_alpr)
stored the snapshot for the session and recorded {identity, plate:AA558EE, 0.999,
region:Albania, snapshotId}. Build + lint green.

Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
2026-06-19 16:53:43 +02:00
julian 30e7fe85de feat(booth): refusal snapshots, subscriber access medium, one-car-one-ticket entry
Three booth-integrity improvements that share the entry/exit flows and activity log.

Refusal snapshots: previously only an accepted open captured a camera image; now
every refusal/hold anomaly fires the directional camera too (a turned-away car is
exactly the evidence wanted) — entry refused-full/held, exit refused
closed/no-session/unpaid/grace-expired (booth + reader paths), refused subscription.
A refused entry has no ticket id, so a synthetic REFUSED- ref keys the anomaly + photo
together. Same fire-and-forget contract; failed captures still show as tiles.

Subscriber access medium: the subscription flow already signed `via`
(qr|card|plate) into entry/exit payloads; surface it as a typed LedgerPayload.via, a
cyan chip in the ticker, and an "Entry medium" modal row (sq+en). Display-only.

One car = one ticket: the entry button could be mashed to mint many tickets per car
(corrupting occupancy + enabling ticket-shopping at exit) — the old #inFlight guard
only blocked overlapping presses. Add a per-relay guard configured on the relay spec:
PRESENCE mode (presenceInput ties ticketing to a vehicle loop on a Dingtian input —
one ticket per car, re-armed when the loop clears) or COOLDOWN fallback
(entryCooldownSec) when there's no barrier feedback. A suppressed press is unsigned
device_events telemetry, not a signed anomaly. SetupWizard exposes both fields.
Fail-closed entry and barrier-is-not-a-door invariants untouched; guard state is
in-memory/rebuildable, starts armed after restart.

Wiki: new entry-double-press; updated entry-exit-points, booth-console, index.

Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
2026-06-19 12:54:54 +02:00
julian bbf61c48df fix(ticket): 11-digit IDs — fix KP-300H barcode line-overflow
The Cashino KP-300H printed entry tickets as raster garbage (solid black
bars / banding) while the Rongta printed the same byte stream fine. Root
cause: the barcode overflowed the print line, not data corruption.

A 13-digit Code128 at module width 3 is ~534 dots. The KP-300H prints 72mm
(512 usable dots at 203 dpi), so the symbol overran the line and the firmware
rendered the overflow as raster noise. The Rongta runs 80mm (576 dots) and had
just enough room — which is why only the Cashino failed. Confirmed on hardware:
plain text printed clean, the barcode was the trigger, and an 11-digit code at
width 3 (~468 dots) both fits and scans the full value at the exit reader.

- Ticket IDs reduced 13 → 11 digits (10 random + Luhn). Length is driven by
  guess-resistance (10^10 space, ~1-in-10^7 to hit a live OPEN ticket even with
  thousands parked — the booth-operator threat model), not volume.
- validateTicketCode is now length-agnostic (\d{10,14} + Luhn) so legacy
  13-digit tickets still in circulation keep validating; the id stays opaque.

Also: sendRaw now closes the print socket GRACEFULLY (end()+FIN, wait for
close) instead of write-then-destroy, which could RST mid-stream and truncate a
job. A separate latent bug found while diagnosing, fixed here.

Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
2026-06-19 11:35:13 +02:00
julian f31e57b4ae feat: explainable activity log — reasons, subscriber names, snapshot gaps
The live activity feed flagged anomalies with no explanation and showed
opaque session keys. Make events self-describing and clickable.

- Clickable feed rows → read-only event-detail modal: humanized fields,
  entry/exit snapshots, and signed-chain provenance collapsed behind an
  audit disclosure (operator sees the story, auditor expands for crypto).
- Localized reason codes (backend i18n): the signed ledger now carries a
  stable REASON_CODE + params (+ English fallback) instead of free-text
  English. The UI translates via reason.<code> catalogs in sq/en, so an
  Albanian operator reads Albanian — from the same immutable event. Adding
  a language is a catalog change, no re-signing. (@parking/shared
  REASON_CODES, reasonPayload; entry/exit/subscription flows emit codes.)
- Subscriber-name resolution: a SUBSESS-… occurrence now shows the
  subscription holder's name (fallback "Abonent"/"Subscriber"). Resolved
  read-time server-side (events API + WS push) as a non-signed
  subscriberLabel; cached with invalidation on subscription edit/delete.
- Failed-snapshot visibility: a camera that was attempted but unreachable
  now shows a "⚠ camera unreachable" tile instead of a silent gap. The
  snapshots API returns failures[] from telemetry, filtered so a recovered
  capture shows no stale warning.

Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
2026-06-19 10:57:17 +02:00
julian cf1ff5676d feat(tariff): V2 — legacy-parity pricing (time-of-day, category, seasonal, flat)
Bring the legacy ParkSQL2017 pricing BREADTH onto our engine while keeping
integer-minor-unit money + immutable signed versions (rejecting legacy's
float money / mutable rows). TariffStructure becomes a discriminated union:
V1 = the original bare ladder (UNCHANGED, verbatim algorithm, golden-
regression-tested against the live version); V2 = {version:2, tz, shared
knobs, defaultCard, windowedCards[]} where each card is flat OR a block
ladder and may be scoped by wall-clock hour window / day-of-week / date
range / vehicle category.

computeFeeV2 prices by stepping one increment at a time, advancing the
ladder by ELAPSED minutes (continuous) while selecting the active card by
WALL-CLOCK time in the version's FROZEN tz. Decisions: tz is a per-site
setting (site_config.timezone, default Europe/Tirane) stamped server-side
into each version on publish — never the host clock (reproducibility);
default-card cap governs a mixed day; precedence = specificity
(date>dow>hour) -> priority -> name (total, order-independent), validation
rejects ambiguous ties; category = a card FIELD, frozen in the signed
vehicle_entry payload (site_config.default_vehicle_category default), read
at both pricing call-sites.

Composer: default card front-and-centre (flat/ladder toggle), tiers under
an "Advanced" disclosure; emits BARE V1 when no tiers (back-compat). DB:
migrations 0005 (timezone) + 0006 (default_vehicle_category). Stood up
vitest in @parking/shared (was zero tests on the ledger-feeding fee fn);
36 tests incl. golden V1 regression, happy-hour/overnight/dow/flat/category/
cap edges, precedence shuffle-invariance, Europe/Tirane DST determinism,
validation matrix — all green. No event-chain change.

Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
2026-06-18 20:00:13 +02:00
julian 5697137c52 feat(subscription): rename permit→subscription + monthly pricing
The "permit/lejet" feature is really a subscription. Full rename of the
mutable master data, plus a recurring monthly price.

- DB (migration 0004, data-preserving ALTER RENAME): permits→subscriptions,
  permit_credentials/_plates→subscription_*, sessions.permit_id→subscription_id.
- Pricing: per-subscription priceMinor + period(monthly) + currency, with a
  site default (site_config.subscription_monthly_price_minor) pre-filling the form.
- Server: subscription-flow.ts (SubscriptionFlow), routes/subscriptions.ts
  (/api/subscriptions). Web: SubscriptionManager, route, i18n (sq Abonimet/en).
- The signed ledger `permitId` payload is intentionally kept — immutable
  hash-chained history; renaming it would break verification of past events.

Deferred (wiki notes): fee collection into the ledger/shift (a shift-attributed
payment), LPR/ANPR plate source, time-of-day access windows (overnight subscriber).

Also carries the device-footer UI surface (api DeviceStatus, router mount,
i18n devices) due to shared-file overlap with the preceding footer commit.

Verified end-to-end on a fresh DB and migration on a live-DB copy (sessions
preserved). Live DB migrated. Full monorepo builds clean.

Claude-Session: https://claude.ai/code/session_01Xcm6ikLgGoCxxHrxtjkk5V
2026-06-18 13:15:04 +02:00
julian 727c62da90 ticket: site metadata header + scannable Albanian ticket; widen barcode
- site_config gains optional park identity (park_name, operator_name, nius,
  address, phone, email); additive Drizzle migration 0001. GET/PUT
  /api/site-config read/write the full config (PUT partial patch, admin only);
  SiteSettings + SetupWizard expose the fields.
- renderTicket() prints an Albanian header sourced from site_config, the
  all-numeric 13-digit ticket id (12 random + Luhn) as Code128, large digits,
  and a lost-ticket footer. CP852 codepage so ë/ç render.
- Widen the Code128 module width 2->3 and height 80->100 dots so the
  short-range "Simple" QR/barcode reader decodes reliably (was barely reading
  at module width 2 on the 80mm head).

See wiki/concepts/site-metadata.md and ticket-encoding.md.
2026-06-17 12:17:21 +02:00
julian 1efa77bf56 devices: pool-of-spaces model — drop lane, per-relay direction
A parking lot is one pool of spaces with a flexible set of entry/exit
points — no "lane". Direction is a property of each RELAY inside an access
controller; readers/cameras bind to a controller relay and inherit it.

Schema:
- drop `lane` from ledger_events, device_events, sessions
- rename lane_devices -> devices (no lane/direction columns)
- access config.relays=[{relay,direction,button?}]; reader/camera
  config.controllerId+relay binding
- fresh 0000_baseline migration (history reset; dev data was throwaway)

Signed ledger:
- remove `lane` from canonicalize(); bump signer keyId sw-hmac-v1 -> v2
  (v1 events won't verify under v2 — intentional, gated per-event by keyId)

Server:
- new device-resolve.ts (replaces lane-map.ts): relayForButton,
  relayForDevice, firstRelayByDirection, devicesByDirection
- entry-flow: button terminal -> its relay; exit/permit: reader's bound
  relay; dispatcher resolves the bound relay + inherited direction
- camera snapshots fire by direction site-wide, async, never block open
- DeviceConfig widened to nested JSON for relays[]

Web:
- wizard: no lane selector; add controllers (relay map + entry-button
  terminal) first, then bind readers/cameras/printers to a controller relay

Wiki: new entry-exit-points.md (replaces lane-direction); reworked
entry-exit-readers, parking-session, first-run-setup, device-registry,
append-only-event-chain, device-events; removed stale lane/LaneMap mentions.
2026-06-16 20:29:38 +02:00
julian e579fe5b6e server+web: capacity / FULL gate (occupancy fold + transient refuse)
Occupancy is a fold over the signed ledger (entries minus exits per identity);
getOccupancy returns {count, capacity, free, full}. Capacity is a single-row
site_config table (admin-set; null = uncapped; migration 0001, additive).

FULL gate lives in the transient entry flow: when full, refuse (no ticket, no
vehicle_entry, no open) and sign an anomaly. Permit entry is NOT gated --
subscribers are admitted past transient-full (their own maxConcurrent still
applies), so occupancy can read over capacity by design (reserve-for-permits).

Routes: GET /api/occupancy + GET /api/site-config (any role), PUT
/api/site-config (admin; non-negative int or null). Web SiteSettings: live
occupancy + FULL badge (everyone), capacity editor (admin).

Verified: fill to cap -> 3rd transient refused; permit admitted past full; exit
frees a slot; RBAC (operator can't set, -5 -> 400); verifyChain ok. Physical
FULL-sign relay output deferred.
2026-06-16 08:13:06 +02:00
julian 2696d281ce server: transient entry flow (button -> ticket -> signed entry -> open)
Closes the long-dangling thread from device-input-flow. On an access device's
rising input edge: print the ticket (failover), then sign vehicle_entry, then
pulseOpen, then cache the session projection.

Two invariants enforced:
- signed BEFORE open (an open with no signed event is the fraud signal);
- HOLD on print failure — no ticket means a transient can't pay on exit, so
  sign an anomaly and do NOT open, and do NOT write a vehicle_entry for a car
  that never got in.

Subscribes the same input bus as the device-telemetry writer (independent:
telemetry always records; entry acts only on an access device's on-edge,
debounced). Verified end to end against stubs: success path signs+opens+caches
and verifyChain ok; printer-down path emits only an anomaly with no open and
no entry; release edge ignored.
2026-06-15 18:32:40 +02:00