quicktype generates types and serializers from JSON, JSON Schema, TypeScript, and GraphQL queries, to help you use JSON type-safely in many languages: C#, Go, Rust, Crystal, C++, Objective-C, Java, TypeScript, JavaScript, JavaScript PropTypes, Flow, Swift, Kotlin, Elm, JSON Schema, Ruby, Dart, Python, Pike, and our latest language, Haskell.
quicktype has some remarkable features that differentiate it from other JSON converters:
Type inference: quicktype infers optionals, dates, UUIDs, enums, integers, and unions. It also infers maps (versus objects) using a Markov chain.
Careful naming: quicktype’s naming system creates nice, legal, unique names for types and properties, handling reserved words and tricky corner cases (e.g. `{ “”: “this is legal”, “null”: “so is this”, "1;DROP TABLE users”: “and this" }`).
Unions for heterogeneous data: JSON data is often heterogenous. quicktype infers this, and creates union types in languages that support them, or synthetic union types in languages that don’t (e.g. try quicktyping `[0, “zero”]` as Swift and Go).
Type unification. This works across multiple samples, so you can quicktype a directory of API data, for example, and unify types across all responses (e.g. you’ll get just one `Customer` type, even if customer data occurs in many samples). You can also provide multiple samples for the same type for better coverage.
Marshalling code: In addition to types, quicktype generates functions for marshalling your types to and from JSON.
Supports dynamic languages: quicktype can add dynamic typechecks for JavaScript, TypeScript, Flow, Python, and Ruby.
Convenient CLI: Run `quicktype https://blockchain.info/latestblock -o LatestBlock.ts` to quicktype a Bitcoin API in TypeScript.
Private hosted app: https://app.quicktype.io/ runs quicktype on the client, so servers never see your data (most JSON converters send your JSON to their server)
Typed input: Feed quicktype TypeScript or JSON Schema instead of JSON for better control over generated types.
Code quality: quicktype emits clean code
It's really incredible to take a master type-definition in TS/JSON Schema and generate language bindings in every language. Especially from a JS/TS Node script as part of a build pipeline.
Or to start new type definitions by using API output/JSON files.
I do the same workflow, where I generate types during build. I often generate them directly from the datasource (e.g. directly from the API). Then I get compile-time errors if the API changes.
