Truck Arrival died with "[@mantine/core] Duplicate options are not supported"
when a booking carried the same container number on two container lines (the
unit uniqueness is per line, not per booking):
- bookingContainerWeights now GROUPs BY container number (MAX weight)
- the weighing modal dedupes both the containers MultiSelect and the
assigned-trucks Select defensively, so bad data degrades instead of
crashing the whole page
Import GRN is now issued automatically when the arrived train is unloaded:
autoUnloadArrivedBookings stamps GRN-IMPORT-<date>-<booking> on both the
create and update paths (never overwrites an existing GRN). The per-row GRN
button in the import queue lights up with no extra clicks.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The marshalling manifest cut off at one page ("... 10 more row(s) not
shown") because the fallback drew rows only until it hit the bottom band.
Now the table flows across as many pages as it needs:
- page 1 keeps the full header + summary tiles; continuation pages get a slim
"(continued — page N)" header and a re-drawn table header
- the verification notice and signature lines stay pinned to the final page,
moving to a fresh page when rows run too deep for the bottom band
- the copy watermark repeats on every page of its copy
- a 12-page safety cap keeps the old truncation note as a last resort
Verified by standalone render: 50-row manifest -> 2 pages, all 50 rows, no
truncation; 5-row doc stays 1 page; two-copy freight order still renders 2
watermarked pages.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
buildTabularFallbackPdf parsed the whole HTML at once, so a two-copy document
(freight order: Port Operations copy + Gate Security & Carrier copy) came out
as ONE page with every tile duplicated and no watermarks.
- documents wrapped in <section class="copy"> now render one page per copy,
each parsed independently (no more merged/duplicated tiles)
- each page carries its copy label as a large rotated light-gray watermark,
drawn beneath the content (new watermarkOp, 30-degree text matrix)
- new assemblePdf() multi-page assembler; assembleSinglePagePdf untouched for
its existing callers
- the meta label is parsed from the document ("Booking") instead of the
hardcoded "TRAIN / SCHEDULE"
Verified by compiling the util standalone and rendering a two-copy freight
order: 2 pages, both watermarks present, tiles once per page, valid xref/EOF;
copy-less documents (marshalling) still render a single unwatermarked page.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- Added status and cargo filters to the ShipmentRequestsPage.
- Implemented date range filtering for preferred dates.
- Introduced sorting options for shipment requests based on submission date and reference.
- Enhanced the display of shipment request details, including status badges and customer information.
- Updated the UI to include a search input with clear functionality and improved layout for filters.
feat: add equipment return option in new shipment form
- Introduced a toggle for equipment return in the NewShipmentPage.
- Updated form schema to include field for container contracts.
- Enhanced user experience with visual feedback on the equipment return selection.
fix: update booking DTO to include equipment return option
- Added field to CreateBookingUnderContractDto for per-shipment override.
- Updated related types and schemas to accommodate the new field for better contract handling.
"Auto Load Ready Items" now opens a train picker (pre-dispatch trains with
these bookings assigned, via the existing loadable-trains flow). No train
available -> no auto-loading, with a clear notice. Loading goes through the
existing per-wagon load path, so items without an allocated wagon are skipped
with a reason.
The train association is stored on the existing warehouse_loadings table
(no new table needed): new train_schedule_id column + a note recording train
number, origin -> destination, and departure time; wagon_id becomes nullable.
The trainless load-passed-export endpoint, its frontend wiring, and the unused
useLoadPassedExport hook are removed.
Migration 2100000000000 (idempotent) also applied to the dev database.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The customer-truck freight order went through ContractPdfService, whose only
fallback is a crude text dump - so on servers without Chromium the document
rendered as unstyled text.
- render via the shared PdfRenderService with the styled tabular fallback
(same path as marshalling / release documents)
- rebuild the freight-order HTML in the standard document shape: brand header,
booking summary tiles, one trucks table (plate / driver / type / containers /
arrival), gate notice, signature lines - which is also what the fallback
parser reads
- keep the two watermarked copies (Port Operations, Gate Security & Carrier)
for the Chromium path; drop the now-unused ContractPdfService injection
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- remove Incoterms, HS codes, and Item code from the truck-entrance form:
nothing in the booking captures them, so they were always hand-typed noise
- remove the hand-typed "Warehouse code and location" input: the backend now
stamps it from the warehouse/yard/zone the operator actually selected
(WH / YARD / ZONE codes), so the GRN and notes always match reality
- Declaration number and Item description widen to full rows
Backend DTO keeps the optional fields for compatibility.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
The receive form now arrives filled with everything the booking already knows,
instead of the operator re-typing it:
- container numbers come from the per-unit records (booking_container_units)
entered at booking time, falling back to the line-level aggregate
- customs seal number prefilled from the units' seal numbers
- net weight prefilled from the booking's declared cargo weight (tonnes)
- driver signatory defaults to the arriving driver's name
Existing prefills (owner, TIN, phone, booking ref, cargo, packaging, unit
count, first-mile / customer truck + driver) unchanged; all fields stay
editable where they were editable before.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Approve delivery now tracks the handover lifecycle exactly: the button appears
(detail page + portal dashboard) the moment a handover is generated and
disappears when the customer signs. The bookings list attaches the
handoverAwaitingSignature flag via one batched query per page; status
heuristics (COMPLETED / TRUCK_ASSIGNED+arrived) are gone.
- generate the arrival handover for ANY self-haul truck: portal-assigned OR
walk-in registered at the gate (isSelfHaulBooking: assigned_at set, or no
EDR last-mile leg). Same rule now guards the exit paper.
- remind every 5 minutes (in-app + SMS + email) until the handover is signed
(@Cron in HandoverService; one reminder per booking per tick)
- Truck Leaving no longer opens blank: the import queue mapper now carries
inv.notes, so the saved arrival renders read-only with only gate-out time
and gross weight editable
- container bookings get "Weigh truck? Yes/No": No skips tare/gross and the
container weight match (weighingSkipped on ReleaseOrderDto, decision made at
arrival sticks for the exit via the Weighing: SKIPPED note). Bulk always
weighs, unchanged.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
READY_FOR_PICKUP allows STORED (park an import item back into storage), but
STORED allowed only RESERVED / READY_FOR_LOADING - so readyForPickup() hit
assertTransition(STORED, READY_FOR_PICKUP) and threw. The item could never
return to pickup, and getNextInventoryAction returned null for an import STORED
item, leaving the row with no action at all.
- allow STORED -> READY_FOR_PICKUP
- an inspected import STORED item now advances to ready-for-pickup
readyForPickup() still rejects non-IMPORT inventory, so the new edge cannot be
reached from the export flow.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
AddGrnNumberToWarehouseInventory1828000000000 is recorded in the migrations
table but the column is absent - it was added, then dropped out-of-band. Because
TypeORM has the original recorded it will never re-run, so every GRN read/write
fails with "column grn_number does not exist":
- bulkReceive() INSERT names grn_number (receive to warehouse)
- importQueueByStatuses() Unloaded + Dispatch queues
- exportInventoryByStatus() Received / Ready-To-Load / Loaded tabs
- grnDocument() GRN PDF
Re-adds the column, backfills from the "GRN Number:" receive note, recreates the
partial index. Idempotent, and down() is a deliberate no-op so reverting the
repair cannot re-introduce the outage.
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
- Added default capacities for container and bulk wagons.
- Updated wagonsFor method to consider weight and length for wagon calculations.
- Improved handling of overweight bookings with appropriate warnings.
- Adjusted tests to reflect changes in wagon capacity and validation logic.
