proc-log

proc-log

Emits events on the process object which a listener can consume and print to the terminal or log file.

This is used by various modules within the npm CLI stack in order to send log events that can be consumed by a listener on the process object.

Currently emits log, output, input, and time events.

API

const { log, output, input, time } = require('proc-log')

output

• output.standard(...args) calls process.emit('output', 'standard', ...args)

  This is for general standard output. Consumers will typically show this on stdout (after optionally formatting or filtering it).

• output.error(...args) calls process.emit('output', 'error', ...args)

  This is for general error output. Consumers will typically show this on stderr (after optionally formatting or filtering it).

• output.buffer(...args) calls process.emit('output', 'buffer', ...args)

  This is for buffered output. Consumers will typically buffer this until they are ready to display.

• output.flush(...args) calls process.emit('output', 'flush', ...args)

  This is to indicate that the output buffer should be flushed.

• output.LEVELS an array of strings of all output method names

log

• log.error(...args) calls process.emit('log', 'error', ...args)

  The highest log level. For printing extremely serious errors that indicate something went wrong.

• log.warn(...args) calls process.emit('log', 'warn', ...args)

  A fairly high log level. Things that the user needs to be aware of, but which won't necessarily cause improper functioning of the system.

• log.notice(...args) calls process.emit('log', 'notice', ...args)

  Notices which are important, but not necessarily dangerous or a cause for excess concern.

• log.info(...args) calls process.emit('log', 'info', ...args)

  Informative messages that may benefit the user, but aren't particularly important.

• log.verbose(...args) calls process.emit('log', 'verbose', ...args)

  Noisy output that is more detail that most users will care about.

• log.silly(...args) calls process.emit('log', 'silly', ...args)

  Extremely noisy excessive logging messages that are typically only useful for debugging.

• log.http(...args) calls process.emit('log', 'http', ...args)

  Information about HTTP requests made and/or completed.

• log.timing(...args) calls process.emit('log', 'timing', ...args)

  Timing information.

• log.pause() calls process.emit('log', 'pause')

  Used to tell the consumer to stop printing messages.

• log.resume() calls process.emit('log', 'resume')

  Used to tell the consumer that it is ok to print messages again.

• log.LEVELS an array of strings of all log method names

input

• input.start(fn?) calls process.emit('input', 'start')

  Used to tell the consumer that the terminal is going to begin reading user input. Returns a function that will call input.end() for convenience.

  This also takes an optional callback which will run input.end() on its completion. If the callback returns a Promise then input.end() will be run during finally().

• input.end() calls process.emit('input', 'end')

  Used to tell the consumer that the terminal has stopped reading user input.

• input.read(...args): Promise calls process.emit('input', 'read', resolve, reject, ...args)

  Used to tell the consumer that the terminal is reading user input and returns a Promise that the producer can await until the consumer has finished its async action.

  This emits resolve and reject functions (in addition to all passed in arguments) which the consumer must use to resolve the returned Promise.

time

• time.start(timerName, fn?) calls process.emit('time', 'start', 'timerName')

  Used to start a timer with the specified name. Returns a function that will call time.end() for convenience.

  This also takes an optional callback which will run time.end() on its completion. If the callback returns a Promise then time.end() will be run during finally().

• time.end(timerName) calls process.emit('time', 'end', timeName)

  Used to tell the consumer to stop a timer with the specified name.

Examples

log

Every log method calls process.emit('log', level, ...otherArgs) internally. So in order to consume those events you need to do process.on('log', fn).

Colorize based on level

Here's an example of how to consume proc-log log events and colorize them based on level:

const chalk = require('chalk')

process.on('log', (level, ...args) => {
  if (level === 'error') {
    console.log(chalk.red(level), ...args)
  } else {
    console.log(chalk.blue(level), ...args)
  }
})

Pause and resume

log.pause and log.resume are included so you have the ability to tell your consumer that you want to pause or resume your display of logs. In the npm CLI we use this to buffer all logs on init until we know the correct loglevel to display. But we also setup a second handler that writes everything to a file even if paused.

let paused = true
const buffer = []

// this handler will buffer and replay logs only after `procLog.resume()` is called
process.on('log', (level, ...args) => {
  if (level === 'resume') {
    buffer.forEach((item) => console.log(...item))
    paused = false
    return
  } 

  if (paused) {
    buffer.push([level, ...args])
  } else {
    console.log(level, ...args)
  }
})

// this handler will write everything to a file
process.on('log', (...args) => {
  fs.appendFileSync('debug.log', args.join(' '))
})

input

start and end

producer.js

const { output, input } = require('proc-log')
const { readFromUserInput } = require('./my-read')

// Using callback passed to `start`
try {
  const res = await input.start(
    readFromUserInput({ prompt: 'OK?', default: 'y' })
  )
  output.standard(`User said ${res}`)
} catch (err) {
  output.error(`User cancelled: ${err}`)
}

// Manually calling `start` and `end`
try {
  input.start()
  const res = await readFromUserInput({ prompt: 'OK?', default: 'y' })
  output.standard(`User said ${res}`)
} catch (err) {
  output.error(`User cancelled: ${err}`)
} finally {
  input.end()
}

consumer.js

const { read } = require('read')

process.on('input', (level) => {
  if (level === 'start') {
    // Hide UI to make room for user input being read
  } else if (level === 'end') {
    // Restore UI now that reading is ended
  }
})

Using read to call read()

producer.js

const { output, input } = require('proc-log')

try {
  const res = await input.read({ prompt: 'OK?', default: 'y' })
  output.standard(`User said ${res}`)
} catch (err) {
  output.error(`User cancelled: ${err}`)
}

consumer.js

const { read } = require('read')

process.on('input', (level, ...args) => {
  if (level === 'read') {
    const [res, rej, opts] = args
    read(opts).then(res).catch(rej)
  }
})

Changelog

5.0.0 (2024-09-03)

⚠️ BREAKING CHANGES

• proc-log now supports node ^18.17.0 || >=20.5.0

Bug Fixes

• aa3a1f6 #101 align to npm 10 node engine range (@hashtagchris)

Chores

• 2697148 #101 run template-oss-apply (@hashtagchris)
• 4e025d2 #99 bump @npmcli/eslint-config from 4.0.5 to 5.0.0 (@dependabot[bot])
• 3d6ab24 #88 bump @npmcli/template-oss to 4.22.0 (@lukekarrys)
• e12dd15 #83 chore: chore: postinstall for dependabot template-oss PR (@lukekarrys)
• 3cfe857 #100 postinstall for dependabot template-oss PR (@hashtagchris)
• 2b4320c #100 bump @npmcli/template-oss from 4.23.1 to 4.23.3 (@dependabot[bot])