json-dts-generator

Generate TS declaration files from JSON files. Suitable for both small and large datasets.

Usage

Install it globally:

$ npm i -g json-dts-generator
$ json-dts-generator INPUT-DIR OUTPUT-DIR

Or use npx:

$ npx json-dts-generator INPUT-DIR OUTPUT-DIR

Purpose

Reads all JSON files inside INPUT-DIR (including those in subdirectories), parses each into a TS declaration file with matching name and places those into OUTPUT-DIR, matching the folder structure inside of INPUT-DIR.

Let's say we have the following folder structure:

data/
├── person.json
└── transaction.json

And file contents look like this:

person.json

{
  "name": "Jane",
  "age": 27
}

transaction.json

{
  "from": {
    "name": "Bob",
    "age": 30
  },
  "to": {
    "name": "Mark",
    "age": 34
  },
  "amount": 100
}

Running the command:

$ npx json-dts-generator data output

Will generate the following folder structure:

output/
├── _common.d.ts
├── person.d.ts
└── transaction.d.ts

The file person.d.ts will default export the type of person.json, while transaction.d.ts will do the same for transaction.json.

The _common.d.ts is a special file that holds all of the types generated from the input files. All of the other output files simply re-export the appropriate type from _common.d.ts.

The script processes the input folder (including subfolders) and generates a matching folder structure in the output.

Advantages

• Reuses same type declarations across different JSON files to avoid declaring the same type more than once. This makes it suitable for processing large datasets of thousands of JSON files.
• Type declarations are readable and Intellisense-friendly, using primitives instead of cryptic type aliases.
• Every generated type declaration inside _common.d.ts has additional context telling you where the type is used to make debugging easier if manual change is required.

Pitfalls

• Cannot infer proper types for empty arrays. The script will warn you if it runs into any of these so that you can fix them manually.