Skip to content

Latest commit

 

History

History
160 lines (137 loc) · 4.48 KB

README.md

File metadata and controls

160 lines (137 loc) · 4.48 KB
Raw

ndjson-apply

A CLI tool to transform a stream of newline-delimited JSON by applying a JS function to each JSON object.

Features:

  • take the JS function to apply from a file
  • the function may return async results
  • preview the transformation results with the --diff option

NPM License Node

Summary

Install

npm i -g ndjson-apply

How To

Basic

cat some_data.ndjson | ndjson-apply some_transform_function.js > some_data_transformed.ndjson
# Which can also be written
ndjson-apply some_transform_function.js < cat some_data.ndjson > some_data_transformed.ndjson

where some_transform_function.js just needs to export a JS function. This should work both with the ESM export syntax

// some_transform_function.js
export default function (doc) {
  doc.total = doc.a + doc.b
  if (doc.total % 2 === 0) {
    return doc
  } else {
    // returning null or undefined drops the entry
  }
}

or with the CommonJS export syntax

// some_transform_function.js
module.exports = function (doc) {
  doc.total = doc.a + doc.b
  if (doc.total % 2 === 0) {
    return doc
  } else {
    // returning null or undefined drops the entry
  }
}

Async

That function can also be async:

import { getSomeExtraData } from './path/to/get_some_extra_data.js'

// some_async_transform_function.js
export default async function (doc) {
  doc.total = doc.a + doc.b
  if (doc.total % 2 === 0) {
    doc.extraData = await getSomeExtraData(doc)
    return doc
  } else {
    // returning null or undefined drops the entry
  }
}

Diff mode

As a way to preview the results of your transformation, you can use the diff mode

cat some_data.ndjson | ndjson-apply some_transform_function.js --diff

which will display a colored diff of each line before and after transformation.

For more readability, each line diff output is indented and on several lines.

Filter mode

Use the js function only to filter lines: lines returning true will be let through. No transformation will be applied.

cat some_data.ndjson | ndjson-apply some_transform_function.js --filter

Use sub-function

Given a function_collection.js file like:

// function_collection.js
export function foo (obj) {
  obj.timestamp = Date.now()
  return obj
}

export function bar (obj) {
  obj.count += obj.count
  return obj
}

You can use those subfunction by passing their key as an additional argument

cat some_data.ndjson | ndjson-apply ./function_collection.js foo
cat some_data.ndjson | ndjson-apply ./function_collection.js bar

This should also work with the CommonJS syntax:

// function_collection.cjs
module.exports = {
  foo: (obj) => {
    obj.timestamp = Date.now()
    return obj
  },
  bar: (obj) => {
    obj.count += obj.count
    return obj
  }
}

Pass additional arguments

Any remaining argument will be passed to the function

# Pass '123' as argument to the exported function
cat some_data.ndjson | ndjson-apply ./function.js 123
# Pass '123' as argument to the exported sub-function foo
cat some_data.ndjson | ndjson-apply ./function_collection.js foo 123

Typescript support

To use ndjson-apply with .ts files, you can execute it with tsx as follow:

# Get a tsx executable
npm install --global tsx
# Use ndjson-apply-ts just like you would use ndjson-apply
ndjson-apply-ts ./some_transform_function.ts < ./tests/assets/sample.ndjson

See also