5m

is a near real time utility, 5mb or 5min

KLP

Installation | Use case | Annotated source | License

Installation

npm i 5m

Use case

I use this package with debug and aws-sdk to upload logs on S3.

Then there is a trigger that feeds every file created on S3 to an AWS Lambda which sends file content to a server that broadcast it using socket.io. Yes I know there are other cool tools like CloudWatch, Kinesis, etc. (but 5m is free as in speach and free as in beer too ;)

The cool part is that it is possible to trigger an AWS Lambda every time a new file is created, and for example load it on a database.

Here is an example code to achieve logging and upload on S3.

// File: log.js
//
// Prints to STDOUT if DEBUG environment variable is set properly.
// Uploads logs on some S3 bucket.
//
//     const log = require('./log')
//     log('Hello world')
//

const AWS = require('aws-sdk')
const debug = require('debug')
const fiveM = require('5m')

const s3 = new AWS.S3()

// Use package name as namespace for logging, just as an example.
const pkg = require('./package.json')
const namespace = pkg.name
const debug5m = debug(`${namespace}:5m`)
const debugPkg = debug(namespace)

const writeOnS3 = (data) => {
  const Bucket = 'my-bucket'

  const tstamp = new Date().getTime()
  const Key = `my/path/${tstamp}.log`

  s3.upload({ Bucket, Key }, error => {
    if (error) debug5m(error)
    else debug5m()
  })
}

function log (message) {
  const tstamp = new Date().getTime()
  const record = `${tstamp} ${namespace} ${message}`

  fiveM(namespace, writeOnS3)(record)
  debugPkg(message)
}

module.exports = log

Annotated source

Store data in a namespaced bucket, as well as its function to flush data.

var bucket = {}
var flush = {}

Considering 1 char should be 1 byte, and dates are expressed in milliseconds, the following constants express 5 MB and 5 minutes. Aproximation is ok, since we want to achieve a near real time. The timeout used for flushing data can be set, for testing or other purpouse via the FIVEM_TIMEOUT_MILLISECONDS environment variable.

const fiveMb = 1024 * 1024 * 5
const fiveMin = 300 * 1000
const flushTimeout = process.env.FIVEM_TIMEOUT_MILLISECONDS || fiveMin

Make sure no data is lost on exit.

process.on('exit', () => {
  for (var namespace in bucket) flush[namespace]()
})

Create the 5m function.

Since 5m is allowed as npm package name, but not as JavaScript identifier, using a roman number like Vm can be confusing, so maybe naming the function as fiveM is a good idea.

/**
 * **@param** `{String}` namespace
 * **@param** `{Function}` write
 *
 * **@returns** `{Function}` logger
 */
function fiveM (namespace, write) {

Create the namespaced flush function: write data and clean up.

  flush[namespace] = () => {
    if (bucket[namespace]) {
      write(bucket[namespace])

      delete bucket[namespace]
    }
  }

Create the logger function.

  /**
   * @param {*} data
   */
  return function logger (data) {

If necessary, initialize data bucket and set timeout to flush it later.

    if (typeof bucket[namespace] === 'undefined') {
      bucket[namespace] = ''

      setTimeout(flush[namespace], flushTimeout)
    }

Append data to named bucket.

    bucket[namespace] += data

Check if data is bigger than 5 MB.

    const exceededSpace = bucket[namespace] && (bucket[namespace].length > 0) && (bucket[namespace].length > fiveMb)

If yes, flush it!

    if (exceededSpace) flush[namespace]()
  }
}

Export 5m function.

module.exports = fiveM

License

MIT