rmancinasandClaude Opus 5
d645ba51d3
A.N.A. is the Rosarito office's tourist auto book and the second carrier
the policy OCR pipeline reads. It ships two unrelated faces, and the split
is different from GMX's: GMX ships two documents about one policy, A.N.A.
ships two products.
AUTOMOBILE (SPECIAL POLICY FOR TOURISTS) insures a car; vehicle table,
9 numbered sections, one
LIMIT OF LIABILITY column
DRIVER'S POLICY (the office: "licencia") insures up to 5 named drivers;
no vehicle at all, 6 unnumbered
sections in a different order,
SUM INSURED + PREMIUM columns
The four automobile products the office sells (amplia / responsabilidad
civil, annual / by-the-day) are the same layout with different numbers, so
they get one parser rather than four.
These are born-digital portal PDFs, so pdftotext -layout returns exact
columns and the driver's-policy parser uses that: its two value columns
print the same shape (100,000.00 usd. / 18.70 usd.) with no per-row label,
so horizontal position is the only thing separating them. The split comes
from the header's own offsets, not a constant, because they shift between
products; when it can't be read every amount is reported as a sum insured
and the reviewer is told, rather than half the premiums being filed as
coverage limits.
Three things the layout will punish a naive read for:
- Each PDF prints its face two or three times (ORIGINAL, AGENT COPY, then
a receipt and three travel cards) and the pipeline concatenates every
page before parsing. The coverage walk is bounded to the first copy and
the driver list to the first POLICY HOLDER block. Unbounded, the licencia
returns the same person three times, which reads as a three-driver policy
rather than as a bug.
- The money row is read positionally off its header. An unused DISCOUNT
prints as a bare "-", so "find the six amounts" shifts every value one
column left on a discounted policy.
- Two five-digit numbers sit in the header band and only one is the agent
clave; the agent's street address is "BENITO JUAREZ 25 No.50 INT 38",
three lines above the No. cell holding the policy number.
Sections 6-8 print a PREMIUM where the others print a limit, so
ParsedCoverage gains an optional `premium` (GMX never fills it) and the
review table a column: $40 is what legal aid cost, not a $40 liability
limit. Exclusions follow the GMX rule and go in the risk label with a null
amount -- which matters more here, since a responsabilidad-civil policy
prints 0.00 for material damage and the two are identical on the page.
Also in this change:
- coveragePeriodDays is parsed and written. A.N.A. sells 3- and 4-day
policies; Policy.coveragePeriodDays defaults to 365, so a weekend policy
left at the default sits in the renewals window a year out. Derived from
the dates, cross-checked against the printed DAYS cell, disagreement
noted not resolved.
- Vehicles and named drivers are parsed, shown read-only in review, and
written as Vehicle / InsuredDriver rows on confirm, skipping any already
on the policy (VIN then plate; licence then name). The case that forces
the skip is confirming a renewal onto an existing policy. Nothing is ever
updated or deleted -- a changed plate lands as a second row for a human.
- Batch.provider is set from what the parsers actually claimed instead of
being hardcoded "GMX", so a mixed upload is labelled as mixed and the
header can never contradict its own documents. PolicyDocument.documentType
follows the same rule (was hardcoded GMX_POLICY).
- matchNote becomes TEXT. It was VARCHAR(191) and the note trail was sliced
to 190 chars, which cut the tail notes -- the "could not read X" ones.
- The policy detail page renders an array coveragesJson as a table. Both
shapes have always been possible there, but the object renderer was the
only one, so an OCR-confirmed policy showed a row per array index labelled
"0", "1", "2" with [object Object] as the value. ANA makes that routine.
GMX is untouched behaviourally; its two parsers now spread a shared empty
base instead of listing every null field. 29 new parser cases against
verbatim pdftotext output of three real ANA PDFs, 53 in the suite.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Jorge Cuadros & Asociados — Platform
Internal platform for a Baja California insurance brokerage and property-services
firm: a single expedient joining each client's properties/services,
insurance policies, account statement, and the firm's checkbook.
It replaces a legacy PHP/Access app (see RESUME.md and PLAN.md for the full
history and rebuild rationale, and docs/BACKLOG.md for
everything still outstanding).
The UI is Spanish-first; the codebase and this document are in English.
Stack
| Layer | Tech | Port |
|---|---|---|
| Web | Next.js 14 (App Router, React 18) | 3000 |
| API | NestJS 10 · Passport local + express-session · Argon2 |
3001 |
| Database | MySQL 8 via Prisma 5 (@jorgecuadros/database workspace pkg) |
3306 |
| Migration | Python 3 pipeline (legacy Access → staging → transforms) | — |
Monorepo managed with pnpm workspaces. Node >= 20.
Repository layout
apps/
web/ Next.js frontend (@jorgecuadros/web)
api/ NestJS backend (@jorgecuadros/api)
scripts/seed-user.mjs idempotent admin seeder
packages/
database/ Prisma schema + generated client (@jorgecuadros/database)
prisma/schema.prisma
migration/ One-off Python ETL from the legacy Access DB (run_all.py)
docker/ Dockerfiles for api + web
docker-compose.yml mysql + api + web
.env.example copy to .env
API feature modules: auth, users, customers, policies, properties,
billing, bank, reports, notifications, renewals, mail, statements,
policy-ocr, ocr, storage, settings, ops.
Web routes: /inicio, /clientes, /polizas (+ /polizas/captura, policy
PDF OCR capture), /servicios, /estado-cuenta, /banco (chequera),
/recibos (utility statement OCR capture), /notificaciones (mass email +
renewal avisos; /renovaciones is an alias onto its Pólizas tab), /reportes,
/catalogos, /operaciones (DB ingest/backup, ADMIN), /usuarios, /login.
Two OCR intakes share one OcrProvider seam (src/ocr/, Tesseract today):
utility statements → ledger rows (docs/STATEMENT_OCR.md)
and carrier policy PDFs → Policy rows (docs/POLICY_OCR.md).
Both need tesseract-ocr, tesseract-ocr-data-spa, poppler-utils and object
storage; each reports its own availability and disables only itself if either
is missing.
Prerequisites
- Node.js >= 20 and pnpm (
npm i -g pnpm) - Docker (for MySQL, or bring your own MySQL 8)
- Python 3 — only if you run the legacy data migration
Run it locally (development)
1. Install
pnpm install
pnpm blocks postinstall build scripts by default; the trusted ones
(argon2, prisma, @prisma/client, @prisma/engines, @nestjs/core) are
allowlisted in pnpm-workspace.yaml, so the native builds run automatically.
2. Configure environment
cp .env.example .env
Then edit .env. For the Docker MySQL below the defaults already line up;
just set a real SESSION_SECRET:
DATABASE_URL=mysql://jorgecuadros:jorgecuadros@localhost:3306/jorgecuadros
SESSION_SECRET=<any long random string>
WEB_ORIGIN=http://localhost:3000
NEXT_PUBLIC_API_ORIGIN=http://localhost:3001
The API loads DATABASE_URL, SESSION_SECRET, WEB_ORIGIN, and optional
PORT (default 3001). WEB_ORIGIN is comma-separated — list every origin the
app is reached under, or credentialed fetches from the missing ones fail CORS.
The web app needs no API URL of its own: the browser derives it from the page it
loaded (same host on port 3001 over plain HTTP, or the same-origin /api path
behind a TLS proxy). Set NEXT_PUBLIC_API_ORIGIN (dev) or API_ORIGIN (deploy,
read at request time) only to override that — for instance when running the API
on a non-default port.
3. Start MySQL
docker compose up -d mysql
(Or point DATABASE_URL at an existing MySQL 8 instance.)
4. Create the schema + generate the Prisma client
The schema is managed with prisma db push (no migration history committed):
pnpm --filter @jorgecuadros/database exec prisma db push
pnpm --filter @jorgecuadros/database generate
5. Seed a sign-in user
node apps/api/scripts/seed-user.mjs
Idempotent (upsert by email). Defaults — override with SEED_EMAIL,
SEED_PASSWORD, SEED_NAME:
- email:
admin@jorgecuadros.local - password:
ChangeMe!2026 - role:
ADMIN
6. Run the apps (two terminals)
# API → http://localhost:3001
pnpm --filter @jorgecuadros/api start:dev
# Web → http://localhost:3000
pnpm --filter @jorgecuadros/web dev
Root shortcuts also exist: pnpm dev:api, pnpm dev:web.
7. Log in
Open http://localhost:3000, sign in with the seeded credentials. Sessions are cookie-based and last 8 hours.
Run it with Docker (full stack)
Builds MySQL + API + web from docker-compose.yml:
export SESSION_SECRET=$(openssl rand -hex 32)
docker compose up --build
Web on http://localhost:3000, API on http://localhost:3001. SESSION_SECRET
is required (compose fails without it). After first boot, push the schema and
seed a user against the container DB:
docker compose exec api node apps/api/scripts/seed-user.mjs
Common commands
| Task | Command |
|---|---|
| Install | pnpm install |
| Dev — API | pnpm dev:api |
| Dev — Web | pnpm dev:web |
| Build all | pnpm build |
| Generate Prisma client | pnpm prisma:generate |
| Push schema (dev) | pnpm --filter @jorgecuadros/database exec prisma db push |
| Prisma Studio | pnpm prisma:studio |
| API tests | pnpm --filter @jorgecuadros/api test |
| Lint (web / api) | pnpm --filter @jorgecuadros/web lint · ... /api lint |
| Seed admin user | node apps/api/scripts/seed-user.mjs |
Auth & roles
Session auth via Passport local strategy; passwords hashed with Argon2 (no
plaintext, unlike the legacy app). Roles gate the UI and API — e.g. managing
/catalogos and /usuarios requires the appropriate ability (ADMIN /
MANAGER). New users are created by an admin in /usuarios; the first admin
comes from the seed script above.
Legacy data migration (optional)
migration/ holds the one-off Python ETL that lifts data out of the old
Microsoft Access database into MySQL.
pip install -r migration/requirements.txt
python migration/run_all.py
⚠️
run_all.pytruncates and reloads all downstream tables. Never run an individualtransform_*.pyin isolation — it orphans dependent tables. Seemigration/RECONCILIATION.mdfor details.
Scheduled jobs
The API runs two automatic email sweeps. Neither cadence is in the source:
both are stored in app_settings and edited at /notificaciones →
"Programación de envíos" (ADMIN, setting:manage), taking effect immediately
without a restart. Shipped defaults:
| Job | Default | What it does |
|---|---|---|
| Pólizas | on, 06:00 daily (America/Tijuana) | Renewal avisos at 30/15 days before expiry and 7 days after. |
| Servicios | off | All four mass-email jobs in order, same as "Ejecutar todos". |
A scheduled run never uses the UI's send flags — in particular it ignores
debug, so a forgotten test toggle cannot silently stop customer mail. Full
detail in docs/MASS_EMAIL_NOTIFICATIONS.md.
Sending needs SES_* in the environment. Without it the API still boots and
logs mail to stdout in dev; in production every send fails loudly and is
recorded as FAILED rather than quietly going nowhere.
Production notes
- Use
pnpm --filter @jorgecuadros/database exec prisma migrate deployif/when a committed migration history is adopted; today dev usesdb push. - Set a strong
SESSION_SECRETand a locked-downDATABASE_URL. - The API expects
WEB_ORIGINto match the browser origin for session cookies. - Documents are stored in MinIO in the deployed environment (see
RESUME.md).