They hold a TIN but no business licence, so there is no eTrade record to look
their registration up in. A checkbox on the first wizard step marks them, and
everything that assumed a trade licence bends around it:
- The company step replaces the eTrade lookup with typed registration details
— name, region, zone, woreda, kebele, house number — required exactly because
they are now on screen. applyEtradeSourcedFields skips the lookup rather than
failing it, so what the customer sends is what is stored.
- No freight-forwarder role. Forwarding is licensed work, so the option is not
offered, and the API refuses it at start-onboarding and at every later
role-add rather than letting approval fail on a document they cannot produce.
- No per-role business-licence upload, client-side or in the completion gate.
- Their own document set (company_onboarding_documents_cooperative) merges on
top of the nationality one, admin-managed like every other set. Nationality
wins a fileKey collision so no slot renders twice, and the DARS paper is not
injected into it — the set it merges onto already carries one.
- The owner is typed in full; with no eTrade manager on file the licence
comparison reports "nothing to compare against", which backoffice now
explains rather than leaving as a bare dash.
Stored as an attributes flag, not a column: everything it changes is
behavioural, and nothing queries or joins on it.
SourcedField rendered read-only whenever a value existed, so the input a
customer had just typed into turned into a badge as soon as the step saved and
they navigated back — and dropped out of requiredKeys at the same time. It now
locks on ownership instead: a Fayda verification owns what its claims filled,
everything else stays an editable, prefilled input.
The representation step follows the same rule and asks in the right order:
- The power-of-attorney question collapses to its answer once given, with a
button back to it (none for a freight forwarder, whose answer is forced).
- A foreign company picks how to prove the person outright — Fayda or a
passport — rather than being shown both at once.
- The representative's own fields appear only once the person is established,
and only for what the verification did not supply; what it did supply is
already on the panel above and is no longer repeated beneath it.
- Dropped the freight-forwarder lecture and the DARS blurb; the badge and the
upload field's own help text already say both.
VAT numbers accept any non-blank value. A foreign tax authority's carries
letters and dashes and a co-operative's follows neither pattern, so the 10-11
digit rule only ever rejected numbers we had no business judging.
fayda-identity covers the two subjects, the passport alternative and the
eTrade owner comparison in place of ~40 general-manager assertions.
poa-delegation drops the poaSameAsOwner waiver case and asserts the opposite:
a forwarder is refused without a delegation letter however it represents
itself.
Data rescue, not just schema tidying — the order matters:
1. Backfill companies.email from generalManagerEmail before the key is
stripped, or companies whose only address lived there stop receiving mail.
2. Backfill ownerName/Email/Phone (now required) from owner, GM, contact
person, then the columns; seed etradeManagerName so existing rows read as
matching rather than as a false mismatch.
3. Convert poaSameAsOwner: non-forwarders become poaDeclared='no' with their
poa* details cleared; forwarders become 'yes' and keep them, so they now
owe a DARS letter they were previously waived.
4. Derive poaDeclared for everyone else.
5. Move drafts off the deleted 'personnel' wizard step.
6. Drop the general_manager_* columns and strip gm*/poaSameAsOwner.
down() restores the column shape only; the stripped values are gone by design.
Replaces the owner/general-manager/PoA trio with a single identity whose
subject is the PoA when the company declares one and the owner otherwise.
- Drop the general manager everywhere: entity columns, DTOs, required-field
list, self-service attributes, gm* identity handling.
- New explicit poaDeclared answer ("yes"/"no") replaces poaSameAsOwner. A
DARS delegation letter is required iff it is "yes"; a freight forwarder is
forced to "yes" server-side and gets no waiver.
- Owner name/email/phone become typeable and required, prefilled from the
eTrade lookup, never falling back to the authenticated account.
- Store eTrade's manager separately (etradeManagerName/Phone) and expose
ownerMatchesEtrade so backoffice compares the asserted owner against the
licence instead of against itself.
- Foreign companies satisfy the identity with Fayda or a passport number, on
whichever subject is verifying (poaPassportNumber added).
- Write companies.email/phone from the owner unconditionally, so a company
without a Fayda-verified owner still has a notification address.
LAST_MILE_REQUESTS URL constants were missing the /api prefix every other
endpoint in URLS.ts carries — all five calls (get, submit, contract
view/document/sign) 404'd against the deployed API, so the departure
notification's confirm link never loaded for the customer.
Also widen GET /last-mile-requests/:id from @BookingStaff to @MixedAudience
with the same ownership check submit()/sign() already use — the confirm
page calls this as its first request, before the customer has done
anything else, so it can't be staff-only.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
End-to-end testing of the direct truck-to-train flow surfaced two defects
in the carriage acceptance sheet's no-wagon fallback. Container numbers
were read from freight.containers, which only gains rows at allocation,
so a direct booking's declared containers never appeared; they are read
from booking_container_units now, seal numbers included. The weight used
bulkTotalWeightTons alone, which is null for CONTAINER and PER_TON bulk
bookings, printing 0.00 — cargoTotalWeightVgm is the real weight there
and only holds an item count for PER_ITEM break-bulk.
Separately, warehouse cargo can be loaded from the Load-to-Train queue or
from the schedule, but loading from the schedule never advanced
warehouse_inventory. The booking went IN_TRANSIT while its cargo still
read as sitting in the warehouse — the same inconsistency that produced
the spurious dispatch block. loadBooking now moves the inventory to
LOADED in the same transaction, and no-ops for direct bookings.
Export cargo reaches a train two ways, but the platform only modelled
one. Direct truck-to-train cargo loads straight onto the wagon, never
enters a warehouse and so never has a GRN — yet assertExportReceivedWithGrn
required one before the carriage acceptance sheet could be issued or the
booking loaded from inside its schedule.
Adds export_handover_mode to freight.bookings (null = WAREHOUSE, so
existing bookings are unaffected) and teaches the shared gate to skip
DIRECT_TO_TRAIN. Both call sites are fixed by that single early return.
For direct bookings the carriage acceptance sheet builds its lines from
the booking's own containers, falling back to the declared bulk tonnage,
and is issuable as soon as the mode is chosen. Direct bookings are also
removed from the warehouse receive queue, since that cargo is never
coming to the shed.
Staff choose the mode from the booking detail page via a new endpoint
reusing bookings:operations. Switching to direct is refused once
warehouse inventory exists, so the two flows cannot cross.
Warehouse-then-train keeps every gate it had.
The export-only assertAllocatedCargoLoaded guard threw a
BadRequestException whenever a booking on the schedule had warehouse
inventory in RECEIVED, STORED or READY_FOR_LOADING, making it
impossible to dispatch a train whose cargo had not been inspected and
loaded onto a wagon.
Leaving cargo behind is an operational decision, not an error state.
The dispatch confirm dialog already lists unassigned and unloaded
bookings and offers Dispatch anyway, so the readiness signal is
preserved — only the hard block is gone.
Wagon/locomotive conflicts and the Djibouti gatepass check still block
dispatch: those are physical and legal conflicts, not cargo readiness.
Import's customer_truck_assignments.arrived_at is set by a separate later
gate action (release()'s arrival branch — the truck returning to collect
already-warehoused goods). Export has no equivalent second step: the truck
delivering cargo to the warehouse arrives and is received in the same act,
so its arrival was never recorded anywhere.
Add markCustomerTruckArrived, mirroring release()'s existing self-haul
departure UPDATE (plate-matched, COALESCE(arrived_at, NOW())), and call it
from receive()/bulkReceive() for EXPORT bookings
BuyerDetails.Wereda had the same problem Region did: companies.woreda holds
names ("Yeka") MoR has no confirmed regex for, but every Wereda value MoR has
actually shown us (seller "12"/"13", the collection's "574") is 1-3 digits
like Region. Precautionary, not confirmed -- but the fix is identical either
way: resolve through EIMS_BUYER_WEREDA_CODES and refuse to file rather than
send a guessed code.
Generalises the Region resolver (resolveRegionCode -> resolveLocationCode) to
cover both fields instead of duplicating it.
No code was invented for "Yeka" -- EIMS_BUYER_WEREDA_CODES ships empty, so
this buyer now fails locally (new stop) instead of silently sending a name
that was never verified against MoR's schema.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Three live attempts turned six guesses into facts. Each fix below is the
gateway's own words, not a reading of the collection.
DocumentNumber and InvoiceCounter move differently, because MoR constrains
them differently. The counter must not skip -- "Invoice counter is not
correct. expected : 1" -- so a definitively refused document hands it back.
The document number must not repeat, so the attempt burns it. Both stay spent
after an ambiguous result, where MoR may have stored the document.
NatureOfSupplies is normalised to MoR's exact lowercase constant and rejected
outright if it is neither 'goods' nor 'service'; its schema branches on this
as a oneOf, so "Service" invalidated the whole ItemList.
Buyer region resolves through a name->code map and now FAILS locally when
unmapped. MoR validates Region against ^[0-9]{1,3}$ on both the seller and
buyer sides, so a name can never be sent and a guessed code on a tax document
is worse than refusing to file.
Seller phone, email, region and wereda are checked against MoR's own regexes
before anything is sent, so a placeholder like "_" fails locally instead of
costing a request and a counter.
EIMS_TAX_CODE stays required and unset in .env.example: the choice between
VAT0 (zero-rated) and VATEX (exempt) is a tax position awaiting finance, and
MoR's enum is recorded there for whoever decides.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
3300000000000 collided with BookingWagonCancellations after the rebase.
3320000000000 is also unavailable: BulkContractTemplates3320000000000 is
already recorded in freight.migrations on the shared dev database from a
branch not present in this checkout, so checking only src/migrations is not
sufficient.
3330000000000 is unique across src/migrations and greater than the current
maximum timestamp recorded in freight.migrations.
Rename the migration file and class. The migration has no explicit name field
and no other code references its previous identity.
Verify migration discovery through the actual runtime path:
scripts/migrate.js loads compiled dist/migrations/*.js migrations, while
application boot does not run migrations automatically. Confirm the renamed
migration is present in dist.
For controlled dev verification, remove its migration-history row and run
pnpm migration:run again. The migration is discovered and applied under
3330000000000; its idempotent DDL produces no schema changes where the EIMS
schema already exists.
Two gaps that only bite in production: nobody could see an invoice's filing
state, and a blocked chain was visible only in the logs.
A failed filing now notifies the staff who can act on it. An ambiguous result
is HIGH priority because it blocks every further invoice for the system number
until someone resolves it, and nothing else would surface that -- the sweep
just goes quiet. A deterministic rejection affects one invoice, so it is
normal priority. The alert never throws: it must not mask the filing outcome.
The backoffice invoice detail page gains an EIMS card showing status, IRN,
counter, submitted and acknowledged timestamps, and the gateway's own error
message, with actions gated on invoices:eims_register. FAILED offers "File
again" -- the reservation model already allows re-registering a rejected
invoice, so retry needed no new endpoint. UNKNOWN offers no re-file button at
all, since resubmitting risks a duplicate registration, and instead explains
that a supervisor must record the IRN or discard the attempt.
Also aligns the migration class name with its renamed file. The DDL is
idempotent, so re-applying under the new name is a no-op against the columns;
it leaves one superseded row in freight.migrations.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
