Stop reading JSON files with require

April 14, 2016 • 5 minutes read • javascript, node.js

There you are, coding at 11:00 PM, trying to get your unit tests back to a green state and no matter how hard you try, you just can’t find why. Not even a dozen of console.log can help you.

Well, that was me last night.

At this stage I should probably make a tattoo stop reading JSON files with node’s require function or even put it up somewhere on the wall. Ok, that’s too much, but after losing 1 hour troubleshooting it deserves at least a blog post.

Reading JSON files with Node.js

JavaScript is beautiful when it comes to JSON. There is no need for a third-party JSON Parser libraries, (such as Json.NET or Gson) and it’s pretty straightforward to parse a JSON file into a variable, just use require(), like this:

settings.json

{
  "name": "My Application Name",
  "tags": [
    "nodejs",
    "javascript"
  ]
}

index.js

var settings = require('./settings.json')
console.log(settings.tags) // ['nodejs', 'javascript']

Easy and readable, right?

But that function has an important side effect that can be both good and bad, that will depend on the context of your application.

In my case, I was using it in a unit test and it was causing more harm than good.

Ok, let’s get straight to the point.

require() will always cache the content of the loaded module (our JSON file, in this case). The next time require() is called for the same module, it will restore it from the cache instead of reading it again. That’s awesome! But…

I was reading the file once for each test case. The first test changed a value on that JSON, and because the content was cached by require(), the changed value was available on the second test case too. I was expecting it to be the original value, but it wasn’t. The result? A broken test.

For those who like to see code, here is a practical hypothetical example.

var expect = require('chai').expect
var fs = require('fs')

describe("Require", () => {
  it("should be able to change settings values", () => {
    var settings = require('./settings.json')
    settings.tags.push('v8')
    expect(settings.tags).to.be.deep.equal(["nodejs", "javascript","v8"])
  })

  it("should reload settings file", () => {
    var settings = require('./settings.json')
    expect(settings.tags).to.be.deep.equal(["nodejs", "javascript"])
  })
})

You might expect that both tests cases will pass. But they won’t. The second will fail, because settings.tags now contains "v8".

There is also another problem with this approach. require() is synchronous. As a good node.js developer, you know that blocking I/O is dangerous. It’s ok to use it to load native or NPM modules, because we need it to be synchronous and cached, but JSON files are a separate story.

The solution is quite easy, just use the built-in fs module when reading JSON files. I’m now using the following helper function.

var readJson = (path, cb) => {
  fs.readFile(require.resolve(path), (err, data) => {
    if (err)
      cb(err)
    else
      cb(null, JSON.parse(data))
  })
}

And our example would be rewritten to:

describe("File System", () => {
  it("should be able to change settings values", (done) => {
    readJson('./settings.json', (err, settings) => {
      settings.tags.push('v8')
      expect(settings.tags).to.be.deep.equal(["nodejs", "javascript","v8"])
      done(err)
    })
  })

  it("should reload settings file", (done) => {
    readJson('./settings.json', (err, settings) => {
      expect(settings.tags).to.be.deep.equal(["nodejs", "javascript" ])
      done(err)
    })
  })
})

Non-cached, non-blocking IO, tests are passing and code is still very readable :-)

Cheers!

Discuss on Twitter Edit on GitHub