Configuration store with a cli prompt and JSON Schema validation. Currently supports strings, booleans and numbers. Extends configstore, which synchronously saves the configuration to ~/.config/configstore/[module-name].json
on every change.
The basic API is the same as configstore, but the constructor takes a name (optional) and schema (see api below).
const config = require('config-prompt')({
a: { type: 'string', required: true },
b: { type: 'boolean', default: true },
c: { type: 'boolean', required: true }
})
config.set('a', 'beep')
let a = config.get('a') // beep
let all = config.all // { a: 'beep', b: true }
console.log(a, all)
Prompt the user for entries that are required and missing, which is only c
continuing the example above:
config.prompt(err => {
console.log(config.get('c'))
})
To prompt for all entries, not just the missing:
config.prompt({ all: true }, callback)
Put these tasks in a config.js
file.
const gulp = require('gulp')
const config = require('config-prompt')({
myString: { type: 'string', required: true },
myFlag: { type: 'boolean', default: true }
})
// Show all config entries
gulp.task('config:print', done => config.print(done))
// Move the config file to trash. Asks for confirmation.
gulp.task('config:trash', done => config.trash(done))
// Prompt for missing config entries. Run this before any other task.
gulp.task('config:prompt', done => config.prompt(done))
// Allow other tasks to consume config
module.exports = config
Say some task needs the configuration. Simply require()
your config.js
and add config:prompt
as a task dependency.
const gulp = require('gulp')
const config = require('./config')
gulp.task('build', ['config:prompt'], (done) => {
if (config.get('myFlag') === true) {
// something
}
done()
})
Then, when a user runs gulp build
, the user will be asked to provide any missing configuration entries.
config = configPrompt([name][, schema])
The name defaults to the name of the module (in package.json
) that required config-prompt
- i.e. the module parent. If no schema is provided, it will guess based on current configuration entries. E.g. a string value is added to the in-memory schema as { type: 'string', required: true }
.
config.set(key, value)
Set a value. Same as configstore.set(), but the key must be present in the schema. Note it does not validate on set.
config.get(key)
Get a value. Same as configstore.get().
config.all
Getter and setter for all entries. Same as configstore.all.
config.validate()
Validate current entries according to the schema. Returns an array of errors in the form of { field: 'data.myKey', message: 'is required' }
config.envify([override])
Returns an object with all entries as CONSTANT_CASE
environment variables. Add extra entries in the override argument (not saved to the config). There's no enforced format for the keys. Snake, param, camel or pascal case, it's all good.
// Returns { MY_KEY: 'foo', MY_NUM: 23, SOMETHING_ELSE: 'else' }
config.set('my-key', 'foo')
config.envify({ myNum: 23, something_else: 'else' })
config.print([options,] callback)
Prints configuration to console.
options
env
(bool, default false): print entries as CONSTANT_CASE
environment variables, like config.envify()
config.prompt([options,] callback)
Prompts to overwrite or fill missing entries. Runs validation after all questions have been answered, then prompts again for invalid entries. Will be silent if there are no questions to ask.
options
all
(bool, default false): prompt for all entries, not just those that are required and missing.nodeEnv
(bool, default true): prompt to set process.env.NODE_ENV
to either "development" or "production", if not set.(!) If you happened to notice the additional options in the source: those are likely to be refactored out and should not be relied upon.
config.trash(callback)
Moves the configuration file to trash after confirmation.
You can manage multiple config stores with the included CLI. Install config-prompt
globally and run config-prompt
without any arguments to get started.
With npm do:
npm install config-prompt