Files
pn-new-crm/src/lib/services/client-merge.service.ts

520 lines
19 KiB
TypeScript
Raw Normal View History

feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
/**
* Client merge service - atomically combines two client records.
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
*
* Used by:
* - /admin/duplicates review queue (when an admin confirms a merge)
* - the at-create suggestion path ("use existing client") - though
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
* that path uses the lighter `attachInterestToClient` and never
* actually merges two pre-existing clients
* - the migration script's `--apply` (eventually)
*
* NOT reversible: a merge is permanent. Every merge writes a
* `client_merge_log` row containing a snapshot of the loser's pre-merge
* state, but this is a forensic/audit record only there is NO
* `unmergeClients` implementation, and the child rows reattached to the
* winner are not restorable from the snapshot. Operators must treat
* merge as a destructive, one-way operation. (The original design called
* for a 7-day `dedup_undo_window_days` reversibility window; that undo
* pathway was never built, so the setting has no effect and the snapshot
* is retained purely as an audit trail.)
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
*
* Design reference: docs/superpowers/specs/2026-05-03-dedup-and-migration-design.md §6.
*/
import { and, eq, sql } from 'drizzle-orm';
import { db } from '@/lib/db';
import {
clients,
clientContacts,
clientAddresses,
clientNotes,
clientTags,
clientRelationships,
clientMergeLog,
clientMergeCandidates,
} from '@/lib/db/schema/clients';
import { interests } from '@/lib/db/schema/interests';
feat(tenancies-p2): rename berth_reservations → berth_tenancies (schema + perms + UI) 73-file atomic rename per docs/tenancies-design.md: - Migration 0085: rename table + indexes + FK constraints; rename documents.reservation_id → tenancy_id; migrate jsonb permission maps (reservations resource → tenancies; collapse create+activate → manage); rewrite historical audit_logs.entity_type='berth_reservation' → 'berth_tenancy'. FK renames wrapped in DO blocks so dev DBs that pre-date the FK additions don't abort. - Schema: berthReservations → berthTenancies; BerthReservation type → BerthTenancy; indexes idx_br_* / idx_brr_* → idx_bt_*. - RolePermissions: resource { view, create, activate, cancel } collapses to { view, manage, cancel }; all 8 default seed bundles + role-form + matrix updated. - Service: berth-reservations.service.ts → berth-tenancies.service.ts; endReservation → endTenancy; listReservations → listTenancies. - API: /api/v1/berth-reservations → /api/v1/tenancies (+ nested [id]); /api/v1/berths/[id]/reservations → /api/v1/berths/[id]/tenancies. - Validators: reservations.ts → tenancies.ts; RESERVATION_STATUSES → TENANCY_STATUSES; endReservationSchema → endTenancySchema. - Routes: /{portSlug}/berth-reservations → /{portSlug}/tenancies; /portal/my-reservations → /portal/my-tenancies. - Components: src/components/reservations/* → src/components/tenancies/*; BerthReservationsTab → BerthTenanciesTab; ClientReservationsTab → ClientTenanciesTab; ReservationList → TenancyList. - Socket events: berth_reservation:* → berth_tenancy:*; payload reservationId → tenancyId. - Webhook events: berth_reservation.* → berth_tenancy.*. - Portal: getPortalUserReservations → getPortalUserTenancies; PortalReservation → PortalTenancy; PortalDashboard.counts.activeReservations → activeTenancies; PortalNav label "Reservations" → "Tenancies". - Dossier: DossierReservation → DossierTenancy; reservationDecisions → tenancyDecisions across smart-archive-dialog + bulk-archive routes. - Documents schema: documents.reservationId → documents.tenancyId (TS + DB column + index + FK constraint). - Activity feed label berth_reservation → berth_tenancy (matched against migrated historical audit rows). KEPT (separate concepts): - Reservation Agreement document type (the contract sent to clients). - "Reservation" pipeline stage name. - {{reservation.*}} merge tokens in template authoring. - interest.reservationStatus / reservationDocStatus / dateReservationSent fields (track agreement signing on the deal). - reservation-agreement-context.ts service (builds merge context for the Reservation Agreement doc; only its DB imports were renamed). Verified: tsc clean, 1480/1480 vitest passing, migration applied. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-25 15:09:35 +02:00
import { berthTenancies } from '@/lib/db/schema/tenancies';
import { payments } from '@/lib/db/schema/pipeline';
import { companyMemberships } from '@/lib/db/schema/companies';
import { yachts } from '@/lib/db/schema/yachts';
import { invoices } from '@/lib/db/schema/financial';
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
import { auditLogs } from '@/lib/db/schema/system';
import { ConflictError, NotFoundError, ValidationError } from '@/lib/errors';
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
// ─── Public API ─────────────────────────────────────────────────────────────
export interface MergeFieldChoices {
/** Per-field overrides - `winner` keeps the surviving client's value;
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
* `loser` copies the loser's value over. Fields not listed default
* to `winner` (no change). */
fullName?: 'winner' | 'loser';
nationalityIso?: 'winner' | 'loser';
preferredContactMethod?: 'winner' | 'loser';
preferredLanguage?: 'winner' | 'loser';
timezone?: 'winner' | 'loser';
source?: 'winner' | 'loser';
sourceDetails?: 'winner' | 'loser';
}
export interface MergeOptions {
winnerId: string;
loserId: string;
/** ID of the user performing the merge (for audit + clientMergeLog.mergedBy). */
mergedBy: string;
/**
* Caller's port - defends against any future caller forgetting to
* pre-validate. Today the sole route caller pre-checks via ctx.portId,
* so this is defense-in-depth: a future bulk-import or CLI caller
* that omits the check still cannot trigger a cross-tenant merge.
* Optional for backwards compat; mergeClients enforces same-port
* regardless, but when set the assertion is "winner.portId === callerPortId"
* (and by transitive same-port rule, loser too).
*/
callerPortId?: string;
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
/** Per-field choice overrides. Multi-value fields (contacts, addresses,
* notes, tags) are always preserved from both sides; this only
* affects single-value scalar fields on the `clients` row. */
fieldChoices?: MergeFieldChoices;
}
export interface MergeResult {
mergeLogId: string;
movedRows: {
interests: number;
contacts: number;
addresses: number;
notes: number;
tags: number;
relationships: number;
fix(tenancies-audit): resolve findings from 7-agent system-wide rename audit MUST-FIX: - src/app/api/v1/admin/users/[id]/permission-overrides/route.ts:70 — the PUT allowlist still gated `reservations: {view,create,activate,cancel}`. Stale: would reject valid `tenancies.{view,manage,cancel}` writes and silently accept ghost `reservations.*` writes that never land. Replaced. - src/lib/services/alert-rules.ts:68 — `reservation.no_agreement` alert emitted `entityType: 'reservation'`. Every other tenancy-related audit/socket/dashboard label is `'berth_tenancy'`. Inconsistent dedupe + activity-feed label miss. - tests/e2e/exhaustive/08-portal.spec.ts:6 — hardcoded /portal/my-reservations navigates to a 404 every run. - tests/e2e/exhaustive/03-reservations.spec.ts — entire spec renamed to 03-tenancies.spec.ts; tab + button locators updated to match renamed UI. SHOULD-FIX (consistency): - src/components/clients/client-detail.tsx — useRealtimeInvalidation only caught 3 of the 4 berth_tenancy:* events; added the `:created` listener. - src/lib/services/client-merge.service.ts — MergeResult.movedRows.reservations + snapshot.reservations + local loserReservations / movedReservations renamed to tenancies / loserTenancies / movedTenancies. No external consumers grep-confirmed. - src/lib/services/gdpr-bundle-builder.ts — GdprBundle.reservations field renamed to .tenancies; user-facing HTML section "Reservations" → "Tenancies"; local reservationRows → tenancyRows. - 6 UI copy strings: gdpr-export-button, bulk-archive-wizard, bulk-hard-delete-dialog, hard-delete-dialog, admin-sections-browser ×2, admin/import/page, won-status-panel — all "reservations" prose updated to "tenancies" (occupancy-record sense). - tests/integration/api/tenancies.test.ts — handler import aliases `createReservationHandler` etc renamed to `createTenancyHandler` etc. - tests/unit/services/berth-tenancies.test.ts — local helper makeReservation → makeTenancyLocal (avoids shadow of the renamed factory). - scripts/audit-permissions.ts — stale allowlist entry for /berth-reservations/[id]/route.ts removed (path no longer exists). - docs/runbooks/permission-audit.md — stale row for same path removed. - docs/tenancies-design.md — fixed factual error ("tenancies.service.ts" → "berth-tenancies.service.ts"). Verified: tsc clean, 1493/1493 vitest. Dev-server note: the running `next dev` process started before P2 and shows Turbopack cached compile errors against the renamed schema files. Source is correct (./tenancies); restart `next dev` to clear the cache. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-25 16:03:14 +02:00
tenancies: number;
payments: number;
companyMemberships: number;
yachts: number;
invoices: number;
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
};
}
/**
* Atomically merge `loserId` into `winnerId`. Throws if:
* - either id doesn't exist or belongs to a different port
* - the loser has already been merged (mergedIntoClientId set)
* - the winner is itself archived
*/
export async function mergeClients(opts: MergeOptions): Promise<MergeResult> {
if (opts.winnerId === opts.loserId) {
throw new ValidationError('Cannot merge a client into itself');
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
}
return await db.transaction(async (tx) => {
// ── Lock both rows for the duration. The first FOR UPDATE that
// arrives wins; a concurrent second merge of the same loser
// will see `mergedIntoClientId` set and bail. ──────────────────────
const [winnerRow] = await tx
.select()
.from(clients)
.where(eq(clients.id, opts.winnerId))
.for('update');
const [loserRow] = await tx
.select()
.from(clients)
.where(eq(clients.id, opts.loserId))
.for('update');
if (!winnerRow) throw new NotFoundError('client');
if (!loserRow) throw new NotFoundError('client');
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
if (winnerRow.portId !== loserRow.portId) {
throw new ValidationError('Cannot merge clients across different ports');
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
}
if (opts.callerPortId && winnerRow.portId !== opts.callerPortId) {
// Defense-in-depth: even though the route already pre-checks, a
// future caller that wires this service from a CLI / bulk-import
// path would otherwise be able to merge any two clients sharing a
// port (or, if the same-port check above were ever weakened, two
// arbitrary clients). Refuse upfront.
throw new ValidationError('Cannot merge clients in another port');
}
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
if (loserRow.mergedIntoClientId) {
throw new ConflictError('That client has already been merged into another record.');
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
}
if (winnerRow.archivedAt) {
throw new ConflictError('Cannot merge into an archived client');
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
}
// ── Snapshot the loser's full state before any mutation. Written to
// `client_merge_log.mergeDetails` as a forensic/audit record only;
// NOT used to restore — merge is one-way (no `unmergeClients`). ─────
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
const loserContacts = await tx
.select()
.from(clientContacts)
.where(eq(clientContacts.clientId, opts.loserId));
const loserAddresses = await tx
.select()
.from(clientAddresses)
.where(eq(clientAddresses.clientId, opts.loserId));
const loserNotes = await tx
.select()
.from(clientNotes)
.where(eq(clientNotes.clientId, opts.loserId));
const loserTags = await tx
.select()
.from(clientTags)
.where(eq(clientTags.clientId, opts.loserId));
const loserInterests = await tx
.select({ id: interests.id })
.from(interests)
.where(eq(interests.clientId, opts.loserId));
fix(tenancies-audit): resolve findings from 7-agent system-wide rename audit MUST-FIX: - src/app/api/v1/admin/users/[id]/permission-overrides/route.ts:70 — the PUT allowlist still gated `reservations: {view,create,activate,cancel}`. Stale: would reject valid `tenancies.{view,manage,cancel}` writes and silently accept ghost `reservations.*` writes that never land. Replaced. - src/lib/services/alert-rules.ts:68 — `reservation.no_agreement` alert emitted `entityType: 'reservation'`. Every other tenancy-related audit/socket/dashboard label is `'berth_tenancy'`. Inconsistent dedupe + activity-feed label miss. - tests/e2e/exhaustive/08-portal.spec.ts:6 — hardcoded /portal/my-reservations navigates to a 404 every run. - tests/e2e/exhaustive/03-reservations.spec.ts — entire spec renamed to 03-tenancies.spec.ts; tab + button locators updated to match renamed UI. SHOULD-FIX (consistency): - src/components/clients/client-detail.tsx — useRealtimeInvalidation only caught 3 of the 4 berth_tenancy:* events; added the `:created` listener. - src/lib/services/client-merge.service.ts — MergeResult.movedRows.reservations + snapshot.reservations + local loserReservations / movedReservations renamed to tenancies / loserTenancies / movedTenancies. No external consumers grep-confirmed. - src/lib/services/gdpr-bundle-builder.ts — GdprBundle.reservations field renamed to .tenancies; user-facing HTML section "Reservations" → "Tenancies"; local reservationRows → tenancyRows. - 6 UI copy strings: gdpr-export-button, bulk-archive-wizard, bulk-hard-delete-dialog, hard-delete-dialog, admin-sections-browser ×2, admin/import/page, won-status-panel — all "reservations" prose updated to "tenancies" (occupancy-record sense). - tests/integration/api/tenancies.test.ts — handler import aliases `createReservationHandler` etc renamed to `createTenancyHandler` etc. - tests/unit/services/berth-tenancies.test.ts — local helper makeReservation → makeTenancyLocal (avoids shadow of the renamed factory). - scripts/audit-permissions.ts — stale allowlist entry for /berth-reservations/[id]/route.ts removed (path no longer exists). - docs/runbooks/permission-audit.md — stale row for same path removed. - docs/tenancies-design.md — fixed factual error ("tenancies.service.ts" → "berth-tenancies.service.ts"). Verified: tsc clean, 1493/1493 vitest. Dev-server note: the running `next dev` process started before P2 and shows Turbopack cached compile errors against the renamed schema files. Source is correct (./tenancies); restart `next dev` to clear the cache. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-25 16:03:14 +02:00
const loserTenancies = await tx
feat(tenancies-p2): rename berth_reservations → berth_tenancies (schema + perms + UI) 73-file atomic rename per docs/tenancies-design.md: - Migration 0085: rename table + indexes + FK constraints; rename documents.reservation_id → tenancy_id; migrate jsonb permission maps (reservations resource → tenancies; collapse create+activate → manage); rewrite historical audit_logs.entity_type='berth_reservation' → 'berth_tenancy'. FK renames wrapped in DO blocks so dev DBs that pre-date the FK additions don't abort. - Schema: berthReservations → berthTenancies; BerthReservation type → BerthTenancy; indexes idx_br_* / idx_brr_* → idx_bt_*. - RolePermissions: resource { view, create, activate, cancel } collapses to { view, manage, cancel }; all 8 default seed bundles + role-form + matrix updated. - Service: berth-reservations.service.ts → berth-tenancies.service.ts; endReservation → endTenancy; listReservations → listTenancies. - API: /api/v1/berth-reservations → /api/v1/tenancies (+ nested [id]); /api/v1/berths/[id]/reservations → /api/v1/berths/[id]/tenancies. - Validators: reservations.ts → tenancies.ts; RESERVATION_STATUSES → TENANCY_STATUSES; endReservationSchema → endTenancySchema. - Routes: /{portSlug}/berth-reservations → /{portSlug}/tenancies; /portal/my-reservations → /portal/my-tenancies. - Components: src/components/reservations/* → src/components/tenancies/*; BerthReservationsTab → BerthTenanciesTab; ClientReservationsTab → ClientTenanciesTab; ReservationList → TenancyList. - Socket events: berth_reservation:* → berth_tenancy:*; payload reservationId → tenancyId. - Webhook events: berth_reservation.* → berth_tenancy.*. - Portal: getPortalUserReservations → getPortalUserTenancies; PortalReservation → PortalTenancy; PortalDashboard.counts.activeReservations → activeTenancies; PortalNav label "Reservations" → "Tenancies". - Dossier: DossierReservation → DossierTenancy; reservationDecisions → tenancyDecisions across smart-archive-dialog + bulk-archive routes. - Documents schema: documents.reservationId → documents.tenancyId (TS + DB column + index + FK constraint). - Activity feed label berth_reservation → berth_tenancy (matched against migrated historical audit rows). KEPT (separate concepts): - Reservation Agreement document type (the contract sent to clients). - "Reservation" pipeline stage name. - {{reservation.*}} merge tokens in template authoring. - interest.reservationStatus / reservationDocStatus / dateReservationSent fields (track agreement signing on the deal). - reservation-agreement-context.ts service (builds merge context for the Reservation Agreement doc; only its DB imports were renamed). Verified: tsc clean, 1480/1480 vitest passing, migration applied. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-25 15:09:35 +02:00
.select({ id: berthTenancies.id })
.from(berthTenancies)
.where(eq(berthTenancies.clientId, opts.loserId));
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
const loserRelationshipsAsA = await tx
.select()
.from(clientRelationships)
.where(eq(clientRelationships.clientAId, opts.loserId));
const loserRelationshipsAsB = await tx
.select()
.from(clientRelationships)
.where(eq(clientRelationships.clientBId, opts.loserId));
const snapshot = {
loser: loserRow,
contacts: loserContacts,
addresses: loserAddresses,
notes: loserNotes,
tags: loserTags,
interests: loserInterests.map((r) => r.id),
fix(tenancies-audit): resolve findings from 7-agent system-wide rename audit MUST-FIX: - src/app/api/v1/admin/users/[id]/permission-overrides/route.ts:70 — the PUT allowlist still gated `reservations: {view,create,activate,cancel}`. Stale: would reject valid `tenancies.{view,manage,cancel}` writes and silently accept ghost `reservations.*` writes that never land. Replaced. - src/lib/services/alert-rules.ts:68 — `reservation.no_agreement` alert emitted `entityType: 'reservation'`. Every other tenancy-related audit/socket/dashboard label is `'berth_tenancy'`. Inconsistent dedupe + activity-feed label miss. - tests/e2e/exhaustive/08-portal.spec.ts:6 — hardcoded /portal/my-reservations navigates to a 404 every run. - tests/e2e/exhaustive/03-reservations.spec.ts — entire spec renamed to 03-tenancies.spec.ts; tab + button locators updated to match renamed UI. SHOULD-FIX (consistency): - src/components/clients/client-detail.tsx — useRealtimeInvalidation only caught 3 of the 4 berth_tenancy:* events; added the `:created` listener. - src/lib/services/client-merge.service.ts — MergeResult.movedRows.reservations + snapshot.reservations + local loserReservations / movedReservations renamed to tenancies / loserTenancies / movedTenancies. No external consumers grep-confirmed. - src/lib/services/gdpr-bundle-builder.ts — GdprBundle.reservations field renamed to .tenancies; user-facing HTML section "Reservations" → "Tenancies"; local reservationRows → tenancyRows. - 6 UI copy strings: gdpr-export-button, bulk-archive-wizard, bulk-hard-delete-dialog, hard-delete-dialog, admin-sections-browser ×2, admin/import/page, won-status-panel — all "reservations" prose updated to "tenancies" (occupancy-record sense). - tests/integration/api/tenancies.test.ts — handler import aliases `createReservationHandler` etc renamed to `createTenancyHandler` etc. - tests/unit/services/berth-tenancies.test.ts — local helper makeReservation → makeTenancyLocal (avoids shadow of the renamed factory). - scripts/audit-permissions.ts — stale allowlist entry for /berth-reservations/[id]/route.ts removed (path no longer exists). - docs/runbooks/permission-audit.md — stale row for same path removed. - docs/tenancies-design.md — fixed factual error ("tenancies.service.ts" → "berth-tenancies.service.ts"). Verified: tsc clean, 1493/1493 vitest. Dev-server note: the running `next dev` process started before P2 and shows Turbopack cached compile errors against the renamed schema files. Source is correct (./tenancies); restart `next dev` to clear the cache. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-25 16:03:14 +02:00
tenancies: loserTenancies.map((r) => r.id),
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
relationshipsAsA: loserRelationshipsAsA,
relationshipsAsB: loserRelationshipsAsB,
fieldChoices: opts.fieldChoices ?? {},
mergedAt: new Date().toISOString(),
};
// ── Apply field choices on the winner. We only touch fields the
// caller explicitly asked to copy from the loser; everything
// else stays as-is. ────────────────────────────────────────────────
const fieldUpdates: Partial<typeof winnerRow> = {};
if (opts.fieldChoices?.fullName === 'loser') fieldUpdates.fullName = loserRow.fullName;
if (opts.fieldChoices?.nationalityIso === 'loser')
fieldUpdates.nationalityIso = loserRow.nationalityIso;
if (opts.fieldChoices?.preferredContactMethod === 'loser')
fieldUpdates.preferredContactMethod = loserRow.preferredContactMethod;
if (opts.fieldChoices?.preferredLanguage === 'loser')
fieldUpdates.preferredLanguage = loserRow.preferredLanguage;
if (opts.fieldChoices?.timezone === 'loser') fieldUpdates.timezone = loserRow.timezone;
if (opts.fieldChoices?.source === 'loser') fieldUpdates.source = loserRow.source;
if (opts.fieldChoices?.sourceDetails === 'loser')
fieldUpdates.sourceDetails = loserRow.sourceDetails;
if (Object.keys(fieldUpdates).length > 0) {
await tx
.update(clients)
.set({ ...fieldUpdates, updatedAt: new Date() })
.where(eq(clients.id, opts.winnerId));
}
// ── Reattach. Each table that points at the loser via clientId
// gets pointed at the winner instead. ─────────────────────────────
const movedInterests = (
await tx
.update(interests)
.set({ clientId: opts.winnerId, updatedAt: new Date() })
.where(eq(interests.clientId, opts.loserId))
.returning({ id: interests.id })
).length;
fix(tenancies-audit): resolve findings from 7-agent system-wide rename audit MUST-FIX: - src/app/api/v1/admin/users/[id]/permission-overrides/route.ts:70 — the PUT allowlist still gated `reservations: {view,create,activate,cancel}`. Stale: would reject valid `tenancies.{view,manage,cancel}` writes and silently accept ghost `reservations.*` writes that never land. Replaced. - src/lib/services/alert-rules.ts:68 — `reservation.no_agreement` alert emitted `entityType: 'reservation'`. Every other tenancy-related audit/socket/dashboard label is `'berth_tenancy'`. Inconsistent dedupe + activity-feed label miss. - tests/e2e/exhaustive/08-portal.spec.ts:6 — hardcoded /portal/my-reservations navigates to a 404 every run. - tests/e2e/exhaustive/03-reservations.spec.ts — entire spec renamed to 03-tenancies.spec.ts; tab + button locators updated to match renamed UI. SHOULD-FIX (consistency): - src/components/clients/client-detail.tsx — useRealtimeInvalidation only caught 3 of the 4 berth_tenancy:* events; added the `:created` listener. - src/lib/services/client-merge.service.ts — MergeResult.movedRows.reservations + snapshot.reservations + local loserReservations / movedReservations renamed to tenancies / loserTenancies / movedTenancies. No external consumers grep-confirmed. - src/lib/services/gdpr-bundle-builder.ts — GdprBundle.reservations field renamed to .tenancies; user-facing HTML section "Reservations" → "Tenancies"; local reservationRows → tenancyRows. - 6 UI copy strings: gdpr-export-button, bulk-archive-wizard, bulk-hard-delete-dialog, hard-delete-dialog, admin-sections-browser ×2, admin/import/page, won-status-panel — all "reservations" prose updated to "tenancies" (occupancy-record sense). - tests/integration/api/tenancies.test.ts — handler import aliases `createReservationHandler` etc renamed to `createTenancyHandler` etc. - tests/unit/services/berth-tenancies.test.ts — local helper makeReservation → makeTenancyLocal (avoids shadow of the renamed factory). - scripts/audit-permissions.ts — stale allowlist entry for /berth-reservations/[id]/route.ts removed (path no longer exists). - docs/runbooks/permission-audit.md — stale row for same path removed. - docs/tenancies-design.md — fixed factual error ("tenancies.service.ts" → "berth-tenancies.service.ts"). Verified: tsc clean, 1493/1493 vitest. Dev-server note: the running `next dev` process started before P2 and shows Turbopack cached compile errors against the renamed schema files. Source is correct (./tenancies); restart `next dev` to clear the cache. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-25 16:03:14 +02:00
const movedTenancies = (
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
await tx
feat(tenancies-p2): rename berth_reservations → berth_tenancies (schema + perms + UI) 73-file atomic rename per docs/tenancies-design.md: - Migration 0085: rename table + indexes + FK constraints; rename documents.reservation_id → tenancy_id; migrate jsonb permission maps (reservations resource → tenancies; collapse create+activate → manage); rewrite historical audit_logs.entity_type='berth_reservation' → 'berth_tenancy'. FK renames wrapped in DO blocks so dev DBs that pre-date the FK additions don't abort. - Schema: berthReservations → berthTenancies; BerthReservation type → BerthTenancy; indexes idx_br_* / idx_brr_* → idx_bt_*. - RolePermissions: resource { view, create, activate, cancel } collapses to { view, manage, cancel }; all 8 default seed bundles + role-form + matrix updated. - Service: berth-reservations.service.ts → berth-tenancies.service.ts; endReservation → endTenancy; listReservations → listTenancies. - API: /api/v1/berth-reservations → /api/v1/tenancies (+ nested [id]); /api/v1/berths/[id]/reservations → /api/v1/berths/[id]/tenancies. - Validators: reservations.ts → tenancies.ts; RESERVATION_STATUSES → TENANCY_STATUSES; endReservationSchema → endTenancySchema. - Routes: /{portSlug}/berth-reservations → /{portSlug}/tenancies; /portal/my-reservations → /portal/my-tenancies. - Components: src/components/reservations/* → src/components/tenancies/*; BerthReservationsTab → BerthTenanciesTab; ClientReservationsTab → ClientTenanciesTab; ReservationList → TenancyList. - Socket events: berth_reservation:* → berth_tenancy:*; payload reservationId → tenancyId. - Webhook events: berth_reservation.* → berth_tenancy.*. - Portal: getPortalUserReservations → getPortalUserTenancies; PortalReservation → PortalTenancy; PortalDashboard.counts.activeReservations → activeTenancies; PortalNav label "Reservations" → "Tenancies". - Dossier: DossierReservation → DossierTenancy; reservationDecisions → tenancyDecisions across smart-archive-dialog + bulk-archive routes. - Documents schema: documents.reservationId → documents.tenancyId (TS + DB column + index + FK constraint). - Activity feed label berth_reservation → berth_tenancy (matched against migrated historical audit rows). KEPT (separate concepts): - Reservation Agreement document type (the contract sent to clients). - "Reservation" pipeline stage name. - {{reservation.*}} merge tokens in template authoring. - interest.reservationStatus / reservationDocStatus / dateReservationSent fields (track agreement signing on the deal). - reservation-agreement-context.ts service (builds merge context for the Reservation Agreement doc; only its DB imports were renamed). Verified: tsc clean, 1480/1480 vitest passing, migration applied. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-25 15:09:35 +02:00
.update(berthTenancies)
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
.set({ clientId: opts.winnerId, updatedAt: new Date() })
feat(tenancies-p2): rename berth_reservations → berth_tenancies (schema + perms + UI) 73-file atomic rename per docs/tenancies-design.md: - Migration 0085: rename table + indexes + FK constraints; rename documents.reservation_id → tenancy_id; migrate jsonb permission maps (reservations resource → tenancies; collapse create+activate → manage); rewrite historical audit_logs.entity_type='berth_reservation' → 'berth_tenancy'. FK renames wrapped in DO blocks so dev DBs that pre-date the FK additions don't abort. - Schema: berthReservations → berthTenancies; BerthReservation type → BerthTenancy; indexes idx_br_* / idx_brr_* → idx_bt_*. - RolePermissions: resource { view, create, activate, cancel } collapses to { view, manage, cancel }; all 8 default seed bundles + role-form + matrix updated. - Service: berth-reservations.service.ts → berth-tenancies.service.ts; endReservation → endTenancy; listReservations → listTenancies. - API: /api/v1/berth-reservations → /api/v1/tenancies (+ nested [id]); /api/v1/berths/[id]/reservations → /api/v1/berths/[id]/tenancies. - Validators: reservations.ts → tenancies.ts; RESERVATION_STATUSES → TENANCY_STATUSES; endReservationSchema → endTenancySchema. - Routes: /{portSlug}/berth-reservations → /{portSlug}/tenancies; /portal/my-reservations → /portal/my-tenancies. - Components: src/components/reservations/* → src/components/tenancies/*; BerthReservationsTab → BerthTenanciesTab; ClientReservationsTab → ClientTenanciesTab; ReservationList → TenancyList. - Socket events: berth_reservation:* → berth_tenancy:*; payload reservationId → tenancyId. - Webhook events: berth_reservation.* → berth_tenancy.*. - Portal: getPortalUserReservations → getPortalUserTenancies; PortalReservation → PortalTenancy; PortalDashboard.counts.activeReservations → activeTenancies; PortalNav label "Reservations" → "Tenancies". - Dossier: DossierReservation → DossierTenancy; reservationDecisions → tenancyDecisions across smart-archive-dialog + bulk-archive routes. - Documents schema: documents.reservationId → documents.tenancyId (TS + DB column + index + FK constraint). - Activity feed label berth_reservation → berth_tenancy (matched against migrated historical audit rows). KEPT (separate concepts): - Reservation Agreement document type (the contract sent to clients). - "Reservation" pipeline stage name. - {{reservation.*}} merge tokens in template authoring. - interest.reservationStatus / reservationDocStatus / dateReservationSent fields (track agreement signing on the deal). - reservation-agreement-context.ts service (builds merge context for the Reservation Agreement doc; only its DB imports were renamed). Verified: tsc clean, 1480/1480 vitest passing, migration applied. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-25 15:09:35 +02:00
.where(eq(berthTenancies.clientId, opts.loserId))
.returning({ id: berthTenancies.id })
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
).length;
// Contacts: move loser's contacts to winner, but DON'T duplicate any
// already-present (channel, value) pair. Loser-only ones get
// demoted to non-primary so the winner's primary stays intact.
const winnerContacts = await tx
.select({ channel: clientContacts.channel, value: clientContacts.value })
.from(clientContacts)
.where(eq(clientContacts.clientId, opts.winnerId));
const winnerContactKeys = new Set(
winnerContacts.map((c) => `${c.channel}::${c.value.toLowerCase()}`),
);
let movedContacts = 0;
for (const c of loserContacts) {
const key = `${c.channel}::${c.value.toLowerCase()}`;
if (winnerContactKeys.has(key)) {
// Winner already has this contact - drop loser's row (cascade
// will clean up when loser is archived). The snapshot records it
// for audit, but this drop is not reversible (merge is one-way).
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
continue;
}
await tx
.update(clientContacts)
.set({ clientId: opts.winnerId, isPrimary: false, updatedAt: new Date() })
.where(eq(clientContacts.id, c.id));
movedContacts += 1;
}
// Addresses: same shape as contacts, but uniqueness is harder to
// detect cleanly (free-text street). Just move them all and let the
// user dedupe in the UI later.
const movedAddresses = (
await tx
.update(clientAddresses)
.set({ clientId: opts.winnerId, isPrimary: false, updatedAt: new Date() })
.where(eq(clientAddresses.clientId, opts.loserId))
.returning({ id: clientAddresses.id })
).length;
const movedNotes = (
await tx
.update(clientNotes)
.set({ clientId: opts.winnerId, updatedAt: new Date() })
.where(eq(clientNotes.clientId, opts.loserId))
.returning({ id: clientNotes.id })
).length;
// Tags: copy any loser-only tag to the winner; drop overlap.
const winnerTags = await tx
.select({ tagId: clientTags.tagId })
.from(clientTags)
.where(eq(clientTags.clientId, opts.winnerId));
const winnerTagSet = new Set(winnerTags.map((t) => t.tagId));
let movedTags = 0;
for (const t of loserTags) {
if (!winnerTagSet.has(t.tagId)) {
await tx.insert(clientTags).values({ clientId: opts.winnerId, tagId: t.tagId });
movedTags += 1;
}
}
await tx.delete(clientTags).where(eq(clientTags.clientId, opts.loserId));
// Relationships: rewrite each FK side to point at the winner. Keep
// both sides regardless - even if A and B both end up as the same
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
// person, the row is preserved for audit; the UI hides self-loops.
const movedRelationships =
(
await tx
.update(clientRelationships)
.set({ clientAId: opts.winnerId })
.where(eq(clientRelationships.clientAId, opts.loserId))
.returning({ id: clientRelationships.id })
).length +
(
await tx
.update(clientRelationships)
.set({ clientBId: opts.winnerId })
.where(eq(clientRelationships.clientBId, opts.loserId))
.returning({ id: clientRelationships.id })
).length;
// Payments: re-point every payment row from the loser to the winner.
// Critical: payments.clientId is notNull onDelete:'cascade', so leaving
// them on the archived loser would let a later hard-delete cascade away
// the winner's financial history. Scoped to the port defensively.
const movedPayments = (
await tx
.update(payments)
.set({ clientId: opts.winnerId })
.where(and(eq(payments.clientId, opts.loserId), eq(payments.portId, winnerRow.portId)))
.returning({ id: payments.id })
).length;
// Company memberships: re-point, but dedup against the unique constraint
// `unique_cm_exact` (companyId, clientId, role, startDate). If the winner
// already has an equivalent membership, drop the loser's row instead of
// re-pointing so we don't violate the unique index.
const winnerMemberships = await tx
.select({
companyId: companyMemberships.companyId,
role: companyMemberships.role,
startDate: companyMemberships.startDate,
})
.from(companyMemberships)
.where(eq(companyMemberships.clientId, opts.winnerId));
const winnerMembershipKeys = new Set(
winnerMemberships.map(
(m) => `${m.companyId}::${m.role}::${m.startDate ? m.startDate.toISOString() : ''}`,
),
);
const loserMemberships = await tx
.select({
id: companyMemberships.id,
companyId: companyMemberships.companyId,
role: companyMemberships.role,
startDate: companyMemberships.startDate,
})
.from(companyMemberships)
.where(eq(companyMemberships.clientId, opts.loserId));
let movedCompanyMemberships = 0;
for (const m of loserMemberships) {
const key = `${m.companyId}::${m.role}::${m.startDate ? m.startDate.toISOString() : ''}`;
if (winnerMembershipKeys.has(key)) {
// Winner already has an equivalent membership - drop the loser's row.
await tx.delete(companyMemberships).where(eq(companyMemberships.id, m.id));
continue;
}
await tx
.update(companyMemberships)
.set({ clientId: opts.winnerId, updatedAt: new Date() })
.where(eq(companyMemberships.id, m.id));
movedCompanyMemberships += 1;
}
// Yachts: re-point polymorphic ownership on the denormalized owner
// columns only. yacht_ownership_history is handled separately.
const movedYachts = (
await tx
.update(yachts)
.set({ currentOwnerId: opts.winnerId, updatedAt: new Date() })
.where(
and(
eq(yachts.currentOwnerType, 'client'),
eq(yachts.currentOwnerId, opts.loserId),
eq(yachts.portId, winnerRow.portId),
),
)
.returning({ id: yachts.id })
).length;
// Invoices: re-point polymorphic billing entity from loser to winner.
const movedInvoices = (
await tx
.update(invoices)
.set({ billingEntityId: opts.winnerId, updatedAt: new Date() })
.where(
and(
eq(invoices.billingEntityType, 'client'),
eq(invoices.billingEntityId, opts.loserId),
eq(invoices.portId, winnerRow.portId),
),
)
.returning({ id: invoices.id })
).length;
// ── Archive the loser. The row stays in the DB (soft-archived, not
// deleted) so `mergedIntoClientId` can act as the redirect pointer
// for any stragglers (links / direct queries / saved views). This
// is a permanent redirect — the loser is never un-archived by a
// reverse-merge, as no unmerge pathway exists. ─────────────────────
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
await tx
.update(clients)
.set({
archivedAt: new Date(),
mergedIntoClientId: opts.winnerId,
updatedAt: new Date(),
})
.where(eq(clients.id, opts.loserId));
// ── Mark any open merge candidate row for this pair as resolved. ───
await tx
.update(clientMergeCandidates)
.set({
status: 'merged',
resolvedAt: new Date(),
resolvedBy: opts.mergedBy,
})
.where(
and(
eq(clientMergeCandidates.portId, winnerRow.portId),
// pair stored in canonical order - match either direction
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
sql`(
(${clientMergeCandidates.clientAId} = ${opts.winnerId}
AND ${clientMergeCandidates.clientBId} = ${opts.loserId})
OR
(${clientMergeCandidates.clientAId} = ${opts.loserId}
AND ${clientMergeCandidates.clientBId} = ${opts.winnerId})
)`,
),
);
// ── Write the merge log + audit log. ────────────────────────────────
const [logRow] = await tx
.insert(clientMergeLog)
.values({
portId: winnerRow.portId,
survivingClientId: opts.winnerId,
mergedClientId: opts.loserId,
mergedBy: opts.mergedBy,
mergeDetails: snapshot,
})
.returning({ id: clientMergeLog.id });
await tx.insert(auditLogs).values({
portId: winnerRow.portId,
userId: opts.mergedBy,
entityType: 'client',
entityId: opts.winnerId,
action: 'merge',
newValue: {
loserId: opts.loserId,
loserName: loserRow.fullName,
movedInterests,
fix(tenancies-audit): resolve findings from 7-agent system-wide rename audit MUST-FIX: - src/app/api/v1/admin/users/[id]/permission-overrides/route.ts:70 — the PUT allowlist still gated `reservations: {view,create,activate,cancel}`. Stale: would reject valid `tenancies.{view,manage,cancel}` writes and silently accept ghost `reservations.*` writes that never land. Replaced. - src/lib/services/alert-rules.ts:68 — `reservation.no_agreement` alert emitted `entityType: 'reservation'`. Every other tenancy-related audit/socket/dashboard label is `'berth_tenancy'`. Inconsistent dedupe + activity-feed label miss. - tests/e2e/exhaustive/08-portal.spec.ts:6 — hardcoded /portal/my-reservations navigates to a 404 every run. - tests/e2e/exhaustive/03-reservations.spec.ts — entire spec renamed to 03-tenancies.spec.ts; tab + button locators updated to match renamed UI. SHOULD-FIX (consistency): - src/components/clients/client-detail.tsx — useRealtimeInvalidation only caught 3 of the 4 berth_tenancy:* events; added the `:created` listener. - src/lib/services/client-merge.service.ts — MergeResult.movedRows.reservations + snapshot.reservations + local loserReservations / movedReservations renamed to tenancies / loserTenancies / movedTenancies. No external consumers grep-confirmed. - src/lib/services/gdpr-bundle-builder.ts — GdprBundle.reservations field renamed to .tenancies; user-facing HTML section "Reservations" → "Tenancies"; local reservationRows → tenancyRows. - 6 UI copy strings: gdpr-export-button, bulk-archive-wizard, bulk-hard-delete-dialog, hard-delete-dialog, admin-sections-browser ×2, admin/import/page, won-status-panel — all "reservations" prose updated to "tenancies" (occupancy-record sense). - tests/integration/api/tenancies.test.ts — handler import aliases `createReservationHandler` etc renamed to `createTenancyHandler` etc. - tests/unit/services/berth-tenancies.test.ts — local helper makeReservation → makeTenancyLocal (avoids shadow of the renamed factory). - scripts/audit-permissions.ts — stale allowlist entry for /berth-reservations/[id]/route.ts removed (path no longer exists). - docs/runbooks/permission-audit.md — stale row for same path removed. - docs/tenancies-design.md — fixed factual error ("tenancies.service.ts" → "berth-tenancies.service.ts"). Verified: tsc clean, 1493/1493 vitest. Dev-server note: the running `next dev` process started before P2 and shows Turbopack cached compile errors against the renamed schema files. Source is correct (./tenancies); restart `next dev` to clear the cache. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-25 16:03:14 +02:00
movedTenancies,
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
movedContacts,
movedAddresses,
movedPayments,
movedCompanyMemberships,
movedYachts,
movedInvoices,
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
},
});
return {
mergeLogId: logRow!.id,
movedRows: {
interests: movedInterests,
contacts: movedContacts,
addresses: movedAddresses,
notes: movedNotes,
tags: movedTags,
relationships: movedRelationships,
fix(tenancies-audit): resolve findings from 7-agent system-wide rename audit MUST-FIX: - src/app/api/v1/admin/users/[id]/permission-overrides/route.ts:70 — the PUT allowlist still gated `reservations: {view,create,activate,cancel}`. Stale: would reject valid `tenancies.{view,manage,cancel}` writes and silently accept ghost `reservations.*` writes that never land. Replaced. - src/lib/services/alert-rules.ts:68 — `reservation.no_agreement` alert emitted `entityType: 'reservation'`. Every other tenancy-related audit/socket/dashboard label is `'berth_tenancy'`. Inconsistent dedupe + activity-feed label miss. - tests/e2e/exhaustive/08-portal.spec.ts:6 — hardcoded /portal/my-reservations navigates to a 404 every run. - tests/e2e/exhaustive/03-reservations.spec.ts — entire spec renamed to 03-tenancies.spec.ts; tab + button locators updated to match renamed UI. SHOULD-FIX (consistency): - src/components/clients/client-detail.tsx — useRealtimeInvalidation only caught 3 of the 4 berth_tenancy:* events; added the `:created` listener. - src/lib/services/client-merge.service.ts — MergeResult.movedRows.reservations + snapshot.reservations + local loserReservations / movedReservations renamed to tenancies / loserTenancies / movedTenancies. No external consumers grep-confirmed. - src/lib/services/gdpr-bundle-builder.ts — GdprBundle.reservations field renamed to .tenancies; user-facing HTML section "Reservations" → "Tenancies"; local reservationRows → tenancyRows. - 6 UI copy strings: gdpr-export-button, bulk-archive-wizard, bulk-hard-delete-dialog, hard-delete-dialog, admin-sections-browser ×2, admin/import/page, won-status-panel — all "reservations" prose updated to "tenancies" (occupancy-record sense). - tests/integration/api/tenancies.test.ts — handler import aliases `createReservationHandler` etc renamed to `createTenancyHandler` etc. - tests/unit/services/berth-tenancies.test.ts — local helper makeReservation → makeTenancyLocal (avoids shadow of the renamed factory). - scripts/audit-permissions.ts — stale allowlist entry for /berth-reservations/[id]/route.ts removed (path no longer exists). - docs/runbooks/permission-audit.md — stale row for same path removed. - docs/tenancies-design.md — fixed factual error ("tenancies.service.ts" → "berth-tenancies.service.ts"). Verified: tsc clean, 1493/1493 vitest. Dev-server note: the running `next dev` process started before P2 and shows Turbopack cached compile errors against the renamed schema files. Source is correct (./tenancies); restart `next dev` to clear the cache. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-25 16:03:14 +02:00
tenancies: movedTenancies,
payments: movedPayments,
companyMemberships: movedCompanyMemberships,
yachts: movedYachts,
invoices: movedInvoices,
feat(dedup): runtime surfaces — merge service, at-create suggestion, admin queue (P2) Adds the live dedup pipeline on top of the P1 library + P3 migration script. The new `client/interest` model now actively prevents duplicate client records at creation time and gives admins a queue to triage the borderline pairs the at-create check missed. Three layers, per design §7: Layer 1 — At-create suggestion ============================== `GET /api/v1/clients/match-candidates` Accepts free-text email / phone / name from the in-flight client form, normalizes them via the dedup library, and returns scored matches against the port's live client pool. Filters out low-confidence noise (the background scoring queue picks those up separately). Strict port scoping; never leaks across tenants. `<DedupSuggestionPanel>` (`src/components/clients/dedup-suggestion-panel.tsx`) Debounced React Query hook. Renders nothing for short inputs or no useful match. On a high-confidence match it interrupts visually with an amber-tinted card and a "Use this client" primary button. Medium confidence falls back to a softer "possible match — check before creating" treatment. `<ClientForm>` Renders the panel above the form (create path only — skipped on edit). New `onUseExistingClient` callback fires when the user picks the existing client; the form closes and the parent decides what to do (typically: navigate to that client's detail page or open the create-interest dialog pre-filled). Layer 2 — Merge service ======================= `mergeClients` (`src/lib/services/client-merge.service.ts`) The atomic merge primitive that everything else calls. Single transaction. Per §6 of the design: - Locks both rows (FOR UPDATE) so concurrent merges of the same loser fail with a clear error rather than racing. - Snapshots the full loser state (contacts / addresses / notes / tags / interest+reservation IDs / relationship rows) into the `client_merge_log.merge_details` JSONB column for the eventual undo flow. - Reattaches every loser-side row to the winner: interests, reservations, contacts (skipping duplicates by `(channel, value)`), addresses, notes, tags (deduped), relationships. - Optional `fieldChoices` — per-scalar overrides letting the user keep the loser's value for fullName / nationality / preferences / timezone / source. - Marks the loser archived with `mergedIntoClientId` set (a redirect pointer for stragglers; never hard-deleted within the undo window). - Resolves any matching `client_merge_candidates` row to status='merged'. - Writes audit log entry. Schema additions: - `clients.merged_into_client_id` (nullable text, indexed) — the redirect pointer set on archive. Tests: 6 cases against a real DB — happy path moves rows + writes log; self-merge / cross-port / already-merged refused; duplicate-contact deduped on reattach; fieldChoices copies loser values to winner. Layer 3 — Admin review queue ============================ `GET /api/v1/admin/duplicates` Pending merge candidates (status='pending') for the current port, with both client summaries hydrated for side-by-side rendering. Skips pairs where one side is already archived/merged. `POST /api/v1/admin/duplicates/[id]/merge` Confirms a candidate. Body picks the winner; the other side becomes the loser. Calls into `mergeClients` — the only path that writes `client_merge_log`. `POST /api/v1/admin/duplicates/[id]/dismiss` Marks the candidate dismissed. Future scoring runs skip the same pair until a score change recreates the row. `<DuplicatesReviewQueue>` (`/admin/duplicates`) Side-by-side card UI for each pending pair. Click a card to pick the winner; the other side is automatically the loser. Toolbar: "Merge into selected" + "Dismiss". No per-field merge editor in this PR — that's a future polish; the simple "pick the better row" flow handles ~80% of cases. Test coverage ============= 11 new integration tests (76 added in this branch total): - 6 mergeClients (atomicity, refusal cases, contact dedup, fieldChoices) - 5 match-candidates API (shape, port scoping, confidence tiers, Pattern F false-positive guard) Full vitest: 926/926 passing (was 858 before the dedup branch). Lint: clean. tsc: clean for new files (only pre-existing errors in unrelated `tests/integration/` files remain, same as before this PR). Out of scope, deferred ====================== - Background scoring cron that populates `client_merge_candidates` (the queue is empty until this lands; manual seeding works for now via the at-create flow). - Side-by-side per-field merge editor with checkboxes (the simple "pick the winner" UX shipped here covers ~80% of real cases). - Admin settings UI for tuning the dedup thresholds. Defaults from the design (90 / 50) are baked in for now. - `unmergeClients` (the snapshot is captured in client_merge_log; the undo endpoint just hasn't been wired yet). These are all natural follow-up PRs that don't block shipping the runtime UX. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
2026-05-03 14:59:04 +02:00
},
};
});
}
// ─── Convenience: list merge candidates for a port ──────────────────────────
export interface MergeCandidatePair {
id: string;
clientAId: string;
clientBId: string;
score: number;
reasons: string[];
status: string;
createdAt: Date;
}
/** Fetch pending merge candidate pairs for the admin review queue. */
export async function listPendingMergeCandidates(portId: string): Promise<MergeCandidatePair[]> {
const rows = await db
.select()
.from(clientMergeCandidates)
.where(
and(eq(clientMergeCandidates.portId, portId), eq(clientMergeCandidates.status, 'pending')),
)
.orderBy(sql`${clientMergeCandidates.score} DESC`);
return rows.map((r) => ({
id: r.id,
clientAId: r.clientAId,
clientBId: r.clientBId,
score: r.score,
reasons: Array.isArray(r.reasons) ? (r.reasons as string[]) : [],
status: r.status,
createdAt: r.createdAt,
}));
}