TypeORM

TypeORM is an ORM that runs in Node.js, browsers, Cordova, Ionic, React Native, NativeScript, Expo, and Electron, usable with either TypeScript or modern JavaScript. It supports more databases than any other JS/TS ORM — MySQL/MariaDB, Postgres, CockroachDB, SQLite, Microsoft SQL Server, Oracle, SAP HANA, Google Spanner, and MongoDB — through a common driver architecture.

Unlike most JavaScript ORMs, TypeORM supports both the Active Record and Data Mapper patterns side by side, letting teams choose the style that fits their codebase. It ships with entities/decorators, an entity manager and repositories, a full migration system, connection pooling and replication, a fluent QueryBuilder, and hooks (listeners/subscribers), drawing heavy influence from Hibernate, Doctrine, and Entity Framework.

What You Get

• A common driver layer spanning MySQL/MariaDB, Postgres, CockroachDB, SQLite, SQL Server, Oracle, SAP HANA, Google Spanner, and MongoDB
• Both Active Record and Data Mapper patterns, selectable per project rather than locked in by the library
• Decorator-based entity/column/relation definitions plus an alternative entity-schema (non-decorator) API
• A CLI for generating and running database migrations, including automatic migration generation from entity diffs
• A fluent QueryBuilder for complex joins, pagination, and cross-database/cross-schema queries

Common Use Cases

• Building a Node.js/NestJS backend that needs a full-featured ORM with migrations and relations across a relational database
• Standardizing data access across a codebase that must support multiple database engines (e.g. Postgres in prod, SQLite in tests)
• Modeling complex domain relations (eager/lazy, self-referenced, closure-table) that simpler query builders don’t support natively
• Migrating an existing Active Record-style codebase to TypeORM without switching to a Data Mapper architecture (or vice versa)

Under The Hood

Architecture - the src/ tree separates concerns cleanly: driver/ implements the per-database adapter layer, query-builder/ and query-runner/ handle SQL generation and execution, metadata-builder//metadata-args//metadata/ build the entity/column/relation reflection model from decorators, schema-builder/ derives and applies schema/migrations, and entity-manager//repository/ expose the Data Mapper API while persistence/ backs the Active Record BaseEntity methods. Tech Stack - TypeScript targeting Node.js, browser, and mobile runtimes (Cordova/Ionic/React Native/NativeScript/Expo/Electron), built with gulp, tested via a pnpm workspace, and using Stryker for mutation testing (stryker.config.json) alongside SonarCloud for coverage reporting. Code Quality - the test/ directory contains roughly 2,700 TypeScript test files split into functional, unit, core, and github-issues regression suites, with a docker-compose.yml spinning up real database instances for integration testing rather than relying solely on mocks — a strong signal for an ORM where driver-specific SQL dialect bugs are easy to miss. API Design - decorator-based entities (@Entity, @Column, @OneToMany) keep model definitions declarative and close to the class they describe, while the dual Active Record/Data Mapper support means the public API intentionally offers two idiomatic entry points (entity.save() vs repository.save(entity)) rather than forcing one pattern on all consumers.