API Overview

This document describes the HTTP API exposed by the application using Next.js Route Handlers in the App Router (app/api/*/route.js).

All routes below are served under the /api prefix.

Frontend developers: For practical usage examples and the small apiClient helper layer, read docs/frontend-api-usage.md. That document is the frontend-oriented single source of truth.

1. Configuration Dependencies

The API expects a valid server configuration.

Required environment variables:

MONGODB_URI — database connection string (used by lib/db.js).
SESSION_SECRET — JWT signing secret for session cookies.
NAS_ROOT_PATH — NAS mount root for storage operations.

Optional environment variables:

SESSION_COOKIE_SECURE — override for the cookie Secure flag (true/false).

1.1 Search provider configuration (RHL-016)

The Search API can run with different provider backends.

SEARCH_PROVIDER (optional)
- Allowed values: fs | qsirch
- Default: fs

Notes:

fs is a local/test fallback that traverses the NAS-like folder structure directly.
qsirch is the intended production provider (indexed search on QNAP).

If SEARCH_PROVIDER=qsirch, these variables are required:

QSIRCH_BASE_URL — base URL of the Qsirch service (must be reachable from inside the app container).
- Example: http://192.168.0.22:8080
QSIRCH_ACCOUNT — QTS/Qsirch account used for server-to-server search.
QSIRCH_PASSWORD — password for the account.
QSIRCH_PATH_PREFIX — path prefix that contains the branch folders.
- Example: /Niederlassungen

Optional Qsirch tuning:

QSIRCH_DATE_FIELD
- Allowed values: modified | created (case-insensitive)
- Default: modified
QSIRCH_MODE
- Allowed values: sync | async | auto (case-insensitive)
- Default: sync

Notes:

The current implementation is sync-first.
auto currently behaves like sync (placeholder for a later async implementation).

Normalization rule (A.1)

lib/config/validateEnv.js accepts these values case-insensitively.
Runtime normalizes Qsirch values (trim + lowercase for QSIRCH_DATE_FIELD / QSIRCH_MODE, and trim for QSIRCH_PATH_PREFIX) so behavior matches validation.

The environment can be validated via:

lib/config/validateEnv.js
scripts/validate-env.mjs

In Docker/production-like runs, execute node scripts/validate-env.mjs before starting the server to fail fast.

Note: You may see a Node warning like MODULE_TYPELESS_PACKAGE_JSON when running env validation. The validation still succeeds; the warning is harmless.

2. Authentication & Authorization

2.1 Sessions

Authentication uses a signed JWT stored in an HTTP-only cookie (auth_session).

To access protected endpoints:

POST /api/auth/login to obtain the cookie.
Send subsequent requests with that cookie.

Notes:

In production-like setups, cookies should be Secure and the app should run behind HTTPS.
For local HTTP testing (http://localhost:3000), you may set SESSION_COOKIE_SECURE=false in your local docker env file.

2.2 RBAC: Branch access (RHL-021 / RHL-041)

RBAC is enforced on filesystem-related endpoints.

Role model:

branch — restricted to own branch
admin — access all branches
superadmin — access all branches
dev — access all branches

Response semantics:

401 Unauthorized: no valid session
403 Forbidden: session exists but branch access is not allowed

2.3 User management authorization (RHL-041 prerequisite for RHL-012)

User management is a separate capability from branch access.

Rules:

Allowed: superadmin, dev
Forbidden: admin, branch

When an endpoint is guarded by this capability, it must return:

403 AUTH_FORBIDDEN_USER_MANAGEMENT

{
	"error": {
		"message": "Forbidden",
		"code": "AUTH_FORBIDDEN_USER_MANAGEMENT"
	}
}

Notes:

RHL-041 does not introduce user management endpoints.
RHL-012 will add the actual user management endpoints and must enforce this capability consistently.

3. Error Handling & Conventions

3.1 Standard error response format

Most endpoints return JSON.

Success responses keep their existing shapes (unchanged).

Error responses always use this standardized shape:

{
	"error": {
		"message": "Human readable message",
		"code": "SOME_MACHINE_CODE",
		"details": {}
	}
}

Notes:

error.message is intended for humans (UI, logs).
error.code is a stable machine-readable identifier (frontend handling, tests, monitoring).
error.details is optional. When present, it must be a JSON object (e.g. validation info).

Binary endpoints

Some endpoints may return a non-JSON body on the 200 happy path (for example application/pdf).

Rules for such endpoints:

On success: return the documented binary payload.
On non-200 errors: return the standardized JSON error payload above.

3.2 Status code rules

The API uses the following status codes consistently:

400 — invalid/missing parameters, validation errors
401 — unauthenticated (missing/invalid session) or invalid credentials
403 — authenticated but not allowed (RBAC / capability mismatch)
404 — resource not found (branch/year/month/day/file does not exist)
500 — unexpected server errors (internal failures)

3.3 Common error codes

The API uses these machine-readable codes (non-exhaustive list):

Auth:
- AUTH_UNAUTHENTICATED
- AUTH_INVALID_CREDENTIALS
- AUTH_FORBIDDEN_BRANCH
- AUTH_FORBIDDEN_USER_MANAGEMENT (RHL-041; used by RHL-012)
Validation (generic):
- VALIDATION_MISSING_PARAM
- VALIDATION_MISSING_QUERY
- VALIDATION_INVALID_JSON
- VALIDATION_INVALID_BODY
- VALIDATION_MISSING_FIELD
Validation (password management):
- VALIDATION_WEAK_PASSWORD
Validation (filesystem route params):
- VALIDATION_BRANCH
- VALIDATION_YEAR
- VALIDATION_MONTH
- VALIDATION_DAY
- VALIDATION_FILENAME
- VALIDATION_FILE_EXTENSION
- VALIDATION_PATH_TRAVERSAL
Validation (search):
- VALIDATION_SEARCH_SCOPE
- VALIDATION_SEARCH_BRANCH
- VALIDATION_SEARCH_BRANCHES
- VALIDATION_SEARCH_DATE
- VALIDATION_SEARCH_RANGE
- VALIDATION_SEARCH_LIMIT
- VALIDATION_SEARCH_CURSOR
- VALIDATION_SEARCH_MISSING_FILTER
Storage:
- FS_NOT_FOUND
- FS_STORAGE_ERROR
Search (backend/provider):
- SEARCH_BACKEND_UNAVAILABLE (provider misconfiguration/unavailability)
Internal:
- INTERNAL_SERVER_ERROR

3.4 Implementation notes

Route handlers use shared helpers:

lib/api/errors.js (standard error payloads + withErrorHandling)
lib/api/storageErrors.js (maps filesystem errors like ENOENT to 404 vs 500)

3.5 Testing note

For realistic Secure cookie behavior, prefer HTTPS.
For local testing on http://localhost, many tools/browsers treat localhost as a special-case “secure context”. Behavior may vary between environments.

3.6 Caching & Freshness (RHL-006)

This project reads delivery-note PDFs from a NAS mount. New scans can appear at any time.

The caching strategy is designed to:

show new files predictably
reduce filesystem load where possible
avoid any accidental shared caching for auth-protected responses

3.6.1 HTTP caching policy

All JSON API responses explicitly disable HTTP caching:

Cache-Control: no-store

This is applied centrally by lib/api/errors.js (the json() / jsonError() helpers). The policy applies to both success and error responses.

For binary endpoints (e.g. PDF streaming), the route handler must explicitly set:

Cache-Control: no-store

3.6.2 Next.js route handler execution

All API route handlers are forced to execute dynamically at request time:

each route exports export const dynamic = "force-dynamic";

This avoids any static/ISR-like behavior for NAS-dependent endpoints and keeps auth/RBAC behavior safe.

3.6.3 Storage micro-cache (server-side TTL)

To reduce repeated fs.readdir() calls, lib/storage.js implements a small process-local TTL cache after RBAC is enforced (RBAC is checked in the route handlers).

TTL defaults:

listBranches() / listYears() → 60 seconds
listMonths() / listDays() / listFiles() → 15 seconds

Important:

TTL is a max staleness guarantee (new files may appear immediately; they must appear after TTL).
Cache is process-local (not shared across multiple app instances).

3.6.4 Frontend fetch guidelines

Frontend code should call these endpoints with explicit “fresh data” settings:

Use credentials:

fetch(url, { credentials: "include", cache: "no-store" });

Do not rely on next: { revalidate: ... } for these endpoints. Freshness is controlled via:
- Cache-Control: no-store (HTTP)
- server-side storage TTL micro-cache

4. Endpoints

4.1 `GET /api/health`

Purpose

Health check endpoint:

Verifies database connectivity (db.command({ ping: 1 })).
Verifies readability of NAS_ROOT_PATH.

Authentication: not required.

Response 200 (example)

{
	"db": "ok",
	"nas": {
		"path": "/mnt/niederlassungen",
		"entriesSample": ["@Recently-Snapshot", "NL01", "NL02"]
	}
}

4.2 `POST /api/auth/login`

Purpose

Authenticate a user and set the session cookie.

Authentication: not required.

Request body (JSON)

{ "username": "example.user", "password": "plain-text-password" }

Responses

200 { "ok": true }
400 (invalid JSON/body)
400 (missing username/password)
401 (invalid credentials)
500

4.3 `GET /api/auth/logout`

Purpose

Destroy the current session by clearing the cookie.

Authentication: recommended (but endpoint is idempotent).

Response

200 { "ok": true }

4.4 `GET /api/auth/me`

Purpose

Provide the current session identity for frontend consumers.

Semantics (frontend-friendly):

200 with { user: null } when unauthenticated
200 with { user: { userId, role, branchId, email } } when authenticated

Notes:

role is one of: branch | admin | superadmin | dev.
email is optional and may be null.

4.5 `POST /api/auth/change-password` (RHL-009)

Change password for the currently authenticated user.

4.6 `GET /api/branches`

Returns the list of branches (e.g. ['NL01', 'NL02']).

Authentication: required.

RBAC behavior

branch role → only own branch
admin / superadmin / dev → all branches

Response 200

{ "branches": ["NL01", "NL02"] }

4.7 `GET /api/branches/[branch]/years`

Example: /api/branches/NL01/years

Authentication: required.

4.8 `GET /api/branches/[branch]/[year]/months`

Example: /api/branches/NL01/2024/months

Authentication: required.

4.9 `GET /api/branches/[branch]/[year]/[month]/days`

Example: /api/branches/NL01/2024/10/days

Authentication: required.

4.10 `GET /api/files?branch=&year=&month=&day=`

Returns files for a given day.

Authentication: required.

4.11 `GET /api/files/:branch/:year/:month/:day/:filename`

Purpose

Stream (or download) a single PDF file from the NAS while enforcing authentication and branch-level RBAC.

4.12 `GET /api/search`

Purpose

Search delivery note content across PDFs.

Authentication: required.

RBAC behavior

branch role:
- results are limited to the user’s branchId
- attempting to query other branches returns 403 AUTH_FORBIDDEN_BRANCH
admin / superadmin / dev role:
- can search across branches using explicit scope parameters

Filter rule:

The backend requires at least one of: q, from, to.
If all three are missing, the API returns 400 VALIDATION_SEARCH_MISSING_FILTER.

5. API v1 freeze (RHL-008)

The endpoints and response shapes documented here (and in docs/frontend-api-usage.md) are considered API v1 for the first frontend implementation.

Rules:

Avoid breaking changes to existing endpoints, parameters, or response shapes.
Prefer additive changes (new endpoints, new optional fields).
If a breaking change becomes necessary, introduce a new endpoint rather than modifying the v1 contract.

6. Adding New Endpoints

When adding new endpoints:

Define URL + method.
Implement a route.js under app/api/....
Enforce auth:
- getSession() for protected endpoints
- return 401 AUTH_UNAUTHENTICATED when session is missing
Enforce branch RBAC (when applicable):
- canAccessBranch(session, branch)
- return 403 AUTH_FORBIDDEN_BRANCH when forbidden
Enforce user-management capability (only for user management APIs; RHL-012):
- requireUserManagement(session)
- return 403 AUTH_FORBIDDEN_USER_MANAGEMENT when forbidden
Use the standardized error contract:
- lib/api/errors.js (withErrorHandling, ApiError, helpers)
Add route tests (Vitest).
Update this document.

api.md 13 KB Historik Rå

API Overview

1. Configuration Dependencies

1.1 Search provider configuration (RHL-016)

2. Authentication & Authorization

2.1 Sessions

2.2 RBAC: Branch access (RHL-021 / RHL-041)

2.3 User management authorization (RHL-041 prerequisite for RHL-012)

3. Error Handling & Conventions

3.1 Standard error response format

3.2 Status code rules

3.3 Common error codes

3.4 Implementation notes

3.5 Testing note

3.6 Caching & Freshness (RHL-006)

3.6.1 HTTP caching policy

3.6.2 Next.js route handler execution

3.6.3 Storage micro-cache (server-side TTL)

3.6.4 Frontend fetch guidelines

4. Endpoints

4.1 GET /api/health

4.2 POST /api/auth/login

4.3 GET /api/auth/logout

4.4 GET /api/auth/me

4.5 POST /api/auth/change-password (RHL-009)

4.6 GET /api/branches

4.7 GET /api/branches/[branch]/years

4.8 GET /api/branches/[branch]/[year]/months

4.9 GET /api/branches/[branch]/[year]/[month]/days

4.10 GET /api/files?branch=&year=&month=&day=

4.11 GET /api/files/:branch/:year/:month/:day/:filename

4.12 GET /api/search

5. API v1 freeze (RHL-008)

6. Adding New Endpoints

api.md 13 KB

Historik Rå

4.1 `GET /api/health`

4.2 `POST /api/auth/login`

4.3 `GET /api/auth/logout`

4.4 `GET /api/auth/me`

4.5 `POST /api/auth/change-password` (RHL-009)

4.6 `GET /api/branches`

4.7 `GET /api/branches/[branch]/years`

4.8 `GET /api/branches/[branch]/[year]/months`

4.9 `GET /api/branches/[branch]/[year]/[month]/days`

4.10 `GET /api/files?branch=&year=&month=&day=`

4.11 `GET /api/files/:branch/:year/:month/:day/:filename`

4.12 `GET /api/search`