stellar/laboratory-backend

Dernière activité source enregistrée: 2 avril 2026 à 18:14

Catalogue SkillsMP mis à jour: 29 octobre 2026 à 18:30

skills collectés: 1

Étoiles GitHub: 1
Forks GitHub: 4

Installer avec un assistant IA

Copiez ce prompt dans l’assistant IA que vous utilisez.

Follow https://skillsmp.com/skill-install/prompt.md to install Agent Skills from https://github.com/stellar/laboratory-backend.

README du dépôt

Fichier source du README : README.md

Lire le README source

Afficher le README complet

# Contract Data API A Node.js REST API for managing contract data using Express.js with PostgreSQL. - [Contract Data API](#contract-data-api) - [API Design Doc](#api-design-doc) - [Prerequisites](#prerequisites) - [Technology Stack](#technology-stack) - [Setup](#setup) - [1. Install dependencies](#1-install-dependencies) - [2. Environment configuration](#2-environment-configuration) - [3. Database setup](#3-database-setup) - [4. Start the API](#4-start-the-api) - [Makefile](#makefile) - [Available Scripts](#available-scripts) - [API Endpoints](#api-endpoints) - [Project Structure](#project-structure) - [Environment Variables](#environment-variables) - [Appendix: Prisma Utilities](#appendix-prisma-utilities) ## API Design Doc [API Design Doc](https://github.com/stellar/platform-design-docs/blob/contract-data-api/contract-data-api/design.md#api-design) by Data Team. ## Maintenance Doc [Maintenance Doc](https://docs.google.com/document/d/1GQIJnFkJK5KKwUhKxQXd-lqLh4vB2x3DU1-K-RAO7a4/edit?tab=t.0) ## Prerequisites - Node.js (v22.23.4 or higher) - pnpm package manager - PostgreSQL database ## Technology Stack - Framework: Express.js 5.2.1 - Database: PostgreSQL with Prisma 6.19.0 - Language: TypeScript - Package Manager: pnpm ## Setup ### 1. Install dependencies ```bash pnpm install ``` ### 2. Environment configuration #### Connection modes There are two primary ways the API connects to PostgreSQL, plus a third variant that combines a local proxy with the direct mode. **Mode A: Direct database URL** Used when `DATABASE_URL` is set. The app connects directly to whatever database URL you provide. No proxy is required if you already have direct access to a reachable PostgreSQL instance. **Mode B: Cloud SQL Connector with IAM** Used when `DATABASE_URL` is not set. Required env vars: `DB_NAME`, `POSTGRES_CONNECTION_NAME`, `POSTGRES_IAM_USER`, `GOOGLE_APPLICATION_CREDENTIALS`. The app uses Google Cloud credentials to authenticate via IAM and opens a Unix socket locally to reach Cloud SQL. **Mode C: Local Cloud SQL Proxy + direct URL** You can run `cloud-sql-proxy` locally, then set `DATABASE_URL` to point at that proxy. This uses the direct connection code path, but the network hop is through the proxy you started. ##### Mode A: Direct database URL Use this when you can connect directly to a reachable PostgreSQL instance without any proxy. 1. Set: - `DATABASE_URL` - PostgreSQL connection string Result: the app connects directly using the provided URL. ##### Mode B: Cloud SQL Connector with IAM Use this when you want the app to authenticate to Cloud SQL using IAM credentials and a local Unix socket. 1. Obtain Google Cloud service account credentials (`creds.json`) from the DATA team. - Place the file in the project root as `creds.json`. - It is already in `.gitignore`. 2. Set these environment variables: - `DB_NAME` - PostgreSQL database name - `POSTGRES_CONNECTION_NAME` - Cloud SQL instance connection name - `POSTGRES_IAM_USER` - IAM database user email - `GOOGLE_APPLICATION_CREDENTIALS` - Path to `creds.json` (e.g., `./creds.json`) - `GOOGLE_CLOUD_SQL_IP_TYPE` - (optional) IP type for the connector: `PUBLIC`, `PRIVATE`, or `PSC`. Default: `PRIVATE`. 3. Ensure `DATABASE_URL` is NOT set. Result: the app uses IAM auth and opens a local Unix socket to Cloud SQL. ##### Mode C: Local Cloud SQL Proxy + direct URL Use this when you want to run the proxy yourself, but still use the direct connection code path. 1. Start the proxy: ```bash ./cloud-sql-proxy --auto-iam-authn <INSTANCE_CONNECTION_NAME> ``` 2. Set: - `DATABASE_URL` - PostgreSQL connection string pointing at the proxy Example: ```bash DATABASE_URL="postgresql://<YOUR_IAM_EMAIL>@localhost:5432/<DATABASE_NAME>?host=/cloudsql/<INSTANCE_CONNECTION_NAME>" ``` When running via `docker compose`, use `host.docker.internal` in the host portion of `DATABASE_URL` to reach the proxy on your machine, for example: ```bash DATABASE_URL="postgresql://<YOUR_IAM_EMAIL>@host.docker.internal:5432/<DATABASE_NAME>?host=/cloudsql/<INSTANCE_CONNECTION_NAME>" ``` #### How connections work The API uses a single connection entrypoint (`src/utils/connect.ts`): - If `DATABASE_URL` is set, it uses direct mode (Mode A or Mode C). - If `DATABASE_URL` is not set, it uses IAM mode (Mode B). #### Using Prisma CLI tools Prisma CLI tools require `DATABASE_URL` to be set, even when the API connects via IAM. Use Mode C (proxy + `DATABASE_URL`) or a direct URL from Mode A. #### Cursor signing key `CURSOR_SIGNING_KEY` is optional. When set, pagination cursors carry an HMAC-SHA256 signature and only cursors this server issued are accepted. Generate it as 32 random bytes, base64url-encoded (43 characters): ```bash node -e "console.log(require('crypto').randomBytes(32).toString('base64url'))" ``` - Startup fails if the key is shorter than 32 bytes. - Rotating the key rejects every outstanding cursor; clients restart from the first page. - Leave it empty to issue unsigned cursors. ### 3. Database setup #### Introspect your database ```bash pnpm prisma db pull ``` Reads the current database schema and updates `prisma/schema.prisma`. #### Generate Prisma client ```bash pnpm prisma:generate ``` Generates the Prisma client used by the app at runtime, based on the updated `prisma/schema.prisma`. #### Run database migrations ```bash pnpm migrate ``` Applies migrations and updates the database schema. ### 4. Start the API ```bash pnpm dev ``` Starts the API in development mode using `ts-node` and `nodemon`. ## Makefile A Makefile is included for convenience targets. Use the raw commands in this README, and refer to `Makefile` for the current target list. ## Available Scripts - `pnpm dev` - Run development server with nodemon - `pnpm dev:watch` - Run development server with file watching - `pnpm build` - Build TypeScript to JavaScript - `pnpm start` - Run production server - `pnpm typecheck` - Run TypeScript type checking - `pnpm prisma:generate` - Generate Prisma client - `pnpm migrate` - Run database migrations - `pnpm prisma:studio` - Open Prisma Studio - `pnpm test` - Run tests - `pnpm test:watch` - Run tests in watch mode - `pnpm lint` - Run linter - `pnpm lint:fix` - Run linter with auto-fix - `pnpm format` - Format code - `pnpm format:check` - Check formatting ## API Endpoints | Method | Endpoint | Description | | ------ | --------------------------- | ------------------------------------------- | | GET | `/api/contract/:id/storage` | Get contract data by ID | | GET | `/api/contract/:id/keys` | Get keys associated with data by ID | | GET | `/api/network_limits` | Get Soroban network limits and fees via RPC | | GET | `/health` | Health check | | GET | `/` | Redirects to `/health` | `curl http://localhost:3000/api/contract/{contract_id}/storage` - ?sort_by=durability&order=desc - Sort by durability descending - ?sort_by=ttl&order=asc - Sort by TTL ascending - ?sort_by=updated_at&order=desc - Sort by updated timestamp descending - ?filter_key=Balance - Filter results by key symbol (currently disabled; returns 400) `curl http://localhost:3000/api/contract/{contract_id}/keys` `curl "http://localhost:3000/api/network_limits?network=mainnet&rpc_url=https://mainnet.sorobanrpc.com"` - `network` is **required**: `mainnet`, `testnet`, or `futurenet` — the network the caller has selected (in the Laboratory UI, the network toggle). - `rpc_url` is **required**, and must be one of the vetted public RPC providers in `PUBLIC_RPC_URLS` (`src/utils/stellarNetworkConfig.ts`). - **The two must agree.** Each allowlisted URL belongs to exactly one network, so a Mainnet `rpc_url` sent with `network=testnet` is rejected with a `400` naming both networks and listing the usable Testnet URLs — rather than being answered with Mainnet's limits. Neither param has a default: with no `network`, a mismatch could not be detected; with no `rpc_url`, the response could silently describe a different network than intended. - Both come **from the request, not from `NETWORK_PASSPHRASE`**. Unlike the contract-data endpoints — whose network is implicit in which instance you hit — this endpoint reads live from the caller's RPC, so it behaves identically on every deployment regardless of how each instance's env is configured. - The response echoes the agreed network as `network_passphrase`, so a caller confirms which network the numbers describe instead of inferring it. - Results are cached per RPC URL for 5 minutes, with a stale fallback for up to 10 minutes if a refresh fails. Rate limited to 10 requests/minute per IP. ## Project Structure ``` src/ ├── config/ # Environment variable validation ├── controllers/ # Route handlers and business logic ├── helpers/ # Cursor encoding/decoding for pagination ├── middleware/ # Express middleware ├── pagination/ # Pagination link builders ├── query-builders/ # Raw SQL query construction (Prisma.sql) ├── routes/ # API route definitions with Zod validation ├── serializers/ # DB result → API response transformation ├── types/ # TypeScript types and enums ├── utils/ # Prisma client, logger, Stellar SDK service ├── index.ts # Main application entry └── instrument.ts # Sentry instrumentation tests/ # Jest tests (mirrors src/ structure) prisma/ ├── schema.prisma # Database schema └── migrations/ # Database migrations ``` ## Environment Variables | Variable | Required | Default | Description | | -------------------------------- | -------- | -------------------------------- | --------------------------------------------------------------------------------------------------------- | | `NODE_ENV` | No | - | Node/Express ecosystem convention is to set it to `production` when deploying the application. | | `ENVIRONMENT` | No | `development` | Deployment environment (e.g. `dev-testnet`, `prd-testnet`) | | `DEBUG` | No | - | Set to `true`, `1`, or `yes` to enable debug output (e.g. table listing) | | `PORT` | No | `3000` | HTTP server port (1-65535) | | `GIT_COMMIT` | No | - | Git commit SHA for release tracking (set at build/deploy time) | | `LOG_LEVEL` | No | `info` | Pino log level (`trace`, `debug`, `info`, `warn`, `error`, `fatal`) | | `TRUST_PROXY` | No | `loopback,linklocal,uniquelocal` | Comma-separated trusted proxy CIDRs or named tokens | | `CORS_ORIGINS` | No | All origins allowed | Comma-separated allowed CORS origins (strings and `/regex/` patterns) | | `PATH_PREFIX` | No | - | URL path prefix prepended to pagination `_links` (e.g. `/pubnet`, `/testnet`) | | `NETWORK_PASSPHRASE` | No | Testnet | Stellar network passphrase. Not used by `/api/network_limits`, which resolves the network from `rpc_url`. | | `HORIZON_URL` | No | - | Stellar Horizon API URL | | `RPC_URL` | No | - | Stellar Soroban RPC URL | | `DATABASE_URL` | Mode A/C | - | PostgreSQL connection string for direct connection | | `STATEMENT_TIMEOUT_MS` | No | `45000` | PostgreSQL `statement_timeout` in milliseconds (1-100000) | | `DB_NAME` | Mode B | - | PostgreSQL database name | | `POSTGRES_CONNECTION_NAME` | Mode B | - | Cloud SQL instance connection name | | `POSTGRES_IAM_USER` | Mode B | - | IAM database user email | | `GOOGLE_APPLICATION_CREDENTIALS` | Mode B | - | Path to service account credentials file | | `GOOGLE_CLOUD_SQL_IP_TYPE` | No | `PRIVATE` | Cloud SQL IP type: `PUBLIC`, `PRIVATE`, or `PSC` | | `SENTRY_DSN` | No | - | Sentry DSN for error monitoring (leave empty to disable) | | `CURSOR_SIGNING_KEY` | No | - | HMAC-SHA256 key for signing pagination cursors, see [Cursor signing key](#cursor-signing-key) | See [Environment configuration](#2-environment-configuration) for connection mode details. ## Deployment Merging to `main` runs the [Build](.github/workflows/build.yml) workflow: 1. Tests run, then the image is pushed to Docker Hub as `stellar/laboratory-backend:<sha>`. The **dev** environments in [stellar/kube](https://github.com/stellar/kube) pick it up automatically (ArgoCD Image Updater). 2. The `promote-to-prd` job pauses for approval of the `production` GitHub Environment. Once approved, it re-tags the same image (by digest) as `prd-<sha>`, which the **prd** environments pick up the same way. To release: open the workflow run for the merge commit, click **Review deployments**, approve `production`, then check that both prd applications are Synced and Healthy in ArgoCD. Notes: - The pause only exists while `production` has a **Required reviewers** rule (Settings → Environments). Without it the job runs straight through. - To promote the current `main` again without a new commit, run the Build workflow manually from `main`. It skips the build and promotes the existing `<sha>` image; other branches are ignored. ## Appendix: Prisma Utilities - Reset database: `npx prisma migrate reset` - Deploy migrations: `npx prisma migrate deploy` - Prisma Studio: `pnpm prisma:studio`

Chargement du README du dépôt…

Skills dans ce dépôt

Analystes en sécurité de l'information

1 catégories métier · 100% classifié

skill

métier

description

mis à jour

security-audit

Analystes en sécurité de l'information

Audit backend code for security vulnerabilities — DoS vectors, query performance, connection pool exhaustion, injection, missing rate limits, and resource exhaustion

Langue du texte source : anglais

2026-04-02

Affichage de 1 skills collectés sur 1.