node-jq is a Node.js wrapper for jq - a lightweight and flexible command-line JSON processor

---

Installation

$ npm install node-jq --save
# or
$ yarn add node-jq
# or
$ pnpm install node-jq

Fast

You can use jq directly with npx:

npx node-jq '.foo' package.json

Advanced installation

By default, node-jq downloads jq during the installation process with a post-install script. Depending on your OS it downloads from [https://github.com/jqlang/jq/releases] into ./node_modules/node-jq/bin/jq to avoid colisions with any global installation. Check #161 #167 #171 for more information. You can safely rely on this location for your installed jq, we won't change this path without a major version upgrade.

Skipping binary installation and use your local jq

If you want to skip the installation step of jq, you can do it with different mechanisms. node-jq will assume jq is in $PATH when skipping the installation.

Choose the one that fits best your needs:

• Ignore the installation by not running it with: npm install node-jq --ignore-scripts

  npm install node-jq --ignore-scripts
  

• Set NODE_JQ_SKIP_INSTALL_BINARY environment variable to true.

  export NODE_JQ_SKIP_INSTALL_BINARY="true"
  npm install node-jq
  

• Set JQ_PATH environment variable

  export JQ_PATH="../../jq"
  npm install node-jq
  

• Set jq_path in .npmrc (npm config files https://docs.npmjs.com/cli/v8/configuring-npm/npmrc)

  # .npmrc
  jq_path=/usr/local/bin/jq
  

  npm install node-jq
  

Usage

jq example

Usually in your CLI using jq:

jq ".abilities[].moves" bulbasaur.json

and you get

{
  "name": "heartgold-soulsilver",
  "power": "10"
}
{
  "name": "platinum",
  "power": "50"
}
{
  "name": "diamond-pearl",
  "power": "99"
}

node-jq equivalent

With node-jq you could run it programmatically and interact with the output as a JavaScript Object:

NOTE: Take care of the filter that you are using with jq, mapping an array or any other iterative output isn't a valid JavaScript Object, that might fail at parse-time.

const jq = require('node-jq')

const filter = '.abilities[].moves'
const jsonPath = '/path/to/bulbasaur.json'
const options = {}

jq.run(filter, jsonPath, options)
  .then((output) => {
    console.log(output)
    /*
      {
        "name": "heartgold-soulsilver",
        "power": "10"
      },
      {
        "name": "platinum",
        "power": "50"
      },
      {
        "name": "diamond-pearl",
        "power": "99"
      }
    */
  })
  .catch((err) => {
    console.error(err)
    // Something went wrong...
  })

Options

path to jq binary

By default, the jq binary installed with the package is used. You can override this in several ways:

1. .npmrc configuration (recommended for project-level settings):

   jq-path=jq
   # or
   jq-path=/usr/local/bin/jq
   

2. Environment variable (for runtime overrides):

   export JQ_PATH=/path/to/jq
   

The priority order is: JQ_PATH environment variable → .npmrc configuration → downloaded binary.

input

input: 'file'

Run the jq query against a JSON file.

jq.run('.', '/path/to/file.json').then(console.log)
// { "foo": "bar" }

input: 'file' with multiple files

Run jq query against multiple JSON files.

jq.run('.', ['/path/to/file.json','path/to/other_file.json']).then(console.log)
// { "foo": "bar" }
// { "otherFoo": "andBar" }

input: 'json'

Run the jq query against an Object.

jq.run('.', { foo: 'bar' }, { input: 'json' }).then(console.log)
// { "foo": "bar" }

input: 'string'

Run the jq query against a String.

jq.run('.', '{ foo: "bar" }', { input: 'string' }).then(console.log)
// { "foo": "bar" }

---

output

output: 'pretty'

Return the output as a String.

jq.run('.', '/path/to/file.json', { output: 'string' }).then(console.log)
// {
//   "foo": "bar"
// }

output: 'json'

Return the output as an Object.

jq.run('.', '/path/to/file.json', { output: 'json' }).then(console.log)
// { foo: 'bar' }

output: 'compact'|'string'

Return the output as a String.

jq.run('.', '/path/to/file.json', { output: 'compact' }).then(console.log)
// {"foo":"bar"}
jq.run('.', '/path/to/file.json', { output: 'string' }).then(console.log)
// {"foo":"bar"}

---

slurp

slurp: true

Read input stream into array.

jq.run('.', ['/path/to/file.json','/path/to/other_file.json'], { output: 'json', slurp: true }).then(console.log)
// [
//   {
//     "foo": "bar"
//   },
//   {
//     "otherFoo": "andBar"
//   }
// ]

sort

sort: true

Sorts object keys alphabetically.

jq.run('.', ['/path/to/file.json'], { output: 'json', sort: true }).then(console.log)
// {
//   "a": 2,
//   "b": 1
// },

args

args: { myfruit: { hello: 'orange'}, myfruit2: "banana" }

Adds the --argjson myfruit "{ 'hello': 'orange' }" --arg myfruit2 orange arguments to the internal jq command

jq.run('{"fruit":$myfruit,"fruit2":$myfruit2}', ['/path/to/file.json'], { output: 'json', sort: true, args: { myfruit: { hello: 'orange' }, myfruit2: "banana" } }).then(console.log)
// {
//   fruit: {
//      hello: "orange"
//   },
//   fruit2: "banana"
// }

cwd

jq.run('.', ['file.json'], { output: 'json', sort: true }, '/path/to').then(console.log)
// {
//   "a": 2,
//   "b": 1
// },

detached

By default jq process will run 'attached' to the main process. That means that any interrupt signal main process receives will be propagated to jq process. For example, if main process receives SIGTERM, jq will also receive it and exit immediately.

However, in some cases you might not want jq to exit immediately and let it exit normally. For example, if you want to implement a graceful shutdown - main process receives SIGTERM, it finishes processing current json file and exits after processing is completed.

To achieve that run jq detached and NodeJS will not propagate SIGTERM to jq process allowing it to run until it completes.

jq.run('.', ['file.json'], { output: 'json', sort: true }, undefined, true).then(console.log)
// {
//   "a": 2,
//   "b": 1
// },

Projects using node-jq

• atom-jq: an Atom package for manipulating JSON
• json-splora: an Electron implementation for manipulating JSON
• Check more

Why?

Why would you want to manipulate JavaScript Objects with jq inside a nodejs app, when there are tools like ramda or lodash?

The idea was to port jq in node to be able to run it as-is. node-jq doesn't try to replace Array/Object filters, maps, transformations, and so on.

Our primary goal was to make jq syntax available inside an Atom extension: atom-jq.

Other than that, jq is an interesting CLI tool to quickly parse and manipulate the response of an API, such as:

curl 'https://jsonplaceholder.typicode.com/comments' | jq '.[].postId'

There are also people dealing with complex use-cases, and some of them want to port their bash scripts to node:

• ilya-sher.org/2016/05/11/most-jq-you-will-ever-need
• cloudadvantage.com.au/new-aws-command-line-tool-and-jq

Want to learn jq?

Seems hard to learn, but it really isn't.

jq is like sed for JSON. Slice, filter, map and transform structured data in a simple and powerful way.

Take a look at this great introduction or a jq lesson.

You can check out the official manual and fiddle around in the online playground jqplay.org.

License

License