Files
bookorbit/server
Gary J. Espitia S. 2498e09083 fix(scanner): invalidate dir scan state on book deletion (#1226)
* fix(scanner): invalidate dir scan state on book deletion

Deleting a book row left its library_dir_scan_state entries intact, so the incremental scanner kept skipping the unchanged directory and never re-created the book on rescan. Invalidate the deleted books' folder paths and ancestors after deleteByIds so the next scan re-reads them and re-discovers the records.

* fix(scanner): make scan-state invalidation atomic

Prevent deleted books and concurrent scans from restoring stale directory state by versioning each library folder and coordinating writes under a database transaction.

* test(server): repair full-suite failures

Update the Book Dock expectation to the unit-record processor contract and bound the watchdog cutoff against timestamps captured around the sweep.

---------

Co-authored-by: neon <18559658+neonsolstice@users.noreply.github.com>
2026-08-31 15:23:57 -06:00
..

BookOrbit - Server

NestJS 11 backend running on Fastify.

Running

From the repo root: pnpm dev (starts both server and client).

Server only: pnpm start:dev (from this directory).

The server loads environment variables from .env in this directory. See .env.example for all available options.

API docs

OpenAPI docs are disabled by default. Enable them with SWAGGER_ENABLED=true:

SWAGGER_ENABLED=true pnpm start:dev

Then open:

Dev watch uses a Nest config without the Swagger metadata generator so the server does not restart immediately after startup. Full build output still uses the standard Nest config and generates src/metadata.ts for enriched docs.

Module structure

Each feature is a self-contained NestJS module under src/modules/:

src/modules/
├── auth/           Authentication (JWT, OIDC, password reset)
├── book/           Book CRUD, cover images, file serving
├── library/        Library management
├── scanner/        File system scanning, metadata extraction
├── collection/     User book collections
├── smart-scope/           Saved search/filter views
├── user/           User management
├── role/           Roles and permissions (RBAC)
├── metadata/       Book metadata providers
├── kobo/           Kobo device sync
├── annotation/     Highlights and annotations
├── bookmark/       Reading position bookmarks
├── reader-preferences/  Per-user reader settings
├── cbz/            CBZ archive handling
├── path/           File path utilities
├── seed/           Database seeding
└── app-settings/   Global application settings

Database

Schema files live in src/db/schema/, one per domain. Migrations are generated by Drizzle Kit.

pnpm db:generate add_some_column   # generate migration from schema diff
pnpm db:migrate                     # apply pending migrations
pnpm db:studio                      # open Drizzle Studio (DB browser)

Testing

pnpm test           # unit tests
pnpm test:watch     # unit tests in watch mode
pnpm e2e:run        # app-smoke end-to-end test from server package
pnpm test:cov       # unit tests with coverage

For suite-based e2e runs, run from the repository root:

pnpm run e2e:list
pnpm run e2e:all
pnpm run e2e:run -- scanner-scenarios

Quality

pnpm type-check      # TypeScript checks (no emit)
pnpm lint:check      # Non-mutating lint checks
pnpm lint:fix        # Auto-fix lint issues
pnpm format:check    # Prettier check
pnpm format          # Prettier write

Key conventions

  • Constructor injection with @Inject(DB) for the Drizzle instance.
  • Config via @nestjs/config named configs (appConfig, dbConfig, authConfig, storageConfig, mailerConfig). Never read process.env directly in services.
  • DTOs use class-validator decorators. Global ValidationPipe rejects unknown fields.
  • Throw NestJS HttpException subclasses (NotFoundException, ForbiddenException, etc.) - never raw Error.
  • All user-owned data is scoped by userId. Service methods check ownership.