recently the npm registry was updated to restrict available pkg names based on the presence of a pkg with a similar name whose diff is only punctuation. this is an awesome tool that helps people find pkg names, i'd like to make a PR to help account for this new registry restriction. if welcome, PR incoming.

for more details on the change, see this blog post: http://blog.npmjs.org/post/168978377570/new-package-moniker-rules

Fixes https://github.com/sindresorhus/npm-name/issues/15

This solution should account for punctuation using is-name-taken, which caches the bulk of npm package names so that only the last few need to be fetched.

One thing to note is that npmName.many is kinda slow as it causes multiple requests to sync the list of packages. Instead, the list of names should really all be sent off to is-name-taken together. If anyone wants, I can add this as well.

In some cases (e.g. in np - see sindresorhus/np#350 and sindresorhus/np#317) you want to resolve a package name for a given registry that wasn't configured to be the user's default registry. I think that npm-name should still use the registry URL returned by registry-url as the default one, but also accept an optional registry URL as the second argument, i.e. npmName(name, [registryUrl]) and npmName.many(names, [registryUrl]).

// cc @weekens @jdziat

NPM has organization which work along with the scoped package features. Therefore, I think it would be handy to do:

npm-name @elemental

The @ to detect an org instead of a package name.

Reproduced with node v0.10.35 and v0.12.2

➜  ~  npm i -g npm-name
/usr/local/bin/npm-name -> /usr/local/lib/node_modules/npm-name/cli.js
[email protected] /usr/local/lib/node_modules/npm-name
├── [email protected] ([email protected])
├── [email protected] ([email protected])
└── [email protected] ([email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected], [email protected])

➜  ~  npm-name aaabbbccc

events.js:72
        throw er; // Unhandled 'error' event
              ^
GotError: https://registry.npmjs.org/aaabbbccc response code is 404 (Not Found)
    at /usr/local/lib/node_modules/npm-name/node_modules/got/index.js:120:12
    at BufferStream.<anonymous> (/usr/local/lib/node_modules/npm-name/node_modules/got/node_modules/read-all-stream/index.js:52:3)
    at BufferStream.emit (events.js:117:20)
    at finishMaybe (/usr/local/lib/node_modules/npm-name/node_modules/got/node_modules/read-all-stream/node_modules/readable-stream/lib/_stream_writable.js:460:14)
    at endWritable (/usr/local/lib/node_modules/npm-name/node_modules/got/node_modules/read-all-stream/node_modules/readable-stream/lib/_stream_writable.js:469:3)
    at BufferStream.Writable.end (/usr/local/lib/node_modules/npm-name/node_modules/got/node_modules/read-all-stream/node_modules/readable-stream/lib/_stream_writable.js:436:5)
    at IncomingMessage.onend (_stream_readable.js:502:10)
    at IncomingMessage.g (events.js:180:16)
    at IncomingMessage.emit (events.js:117:20)
    at _stream_readable.js:944:16

We cannot do replacements in the requested name as the npm API does not support the shortened names so the only way to check if a name would be rejected for typosquatting is to try every combination of characters in every position which seems impractical and would perform really bad.

So until npm exposes an API to check names, we're just checking the name as-is.

Fixes #25

pinto - at time of writing - isn't take on npm. The results for this specific word are super flaky though, and I don't know why. It fails about 80% of the time, returning false when it should obviously return true since it's available. There don't seem to be any errors.

You can run the below code in this RunKit notebook.

const isAvailable = require('npm-name')
const partition = require('lodash.partition')

const fetch = word => {
  const array = new Array(100).fill(word)
  return Promise.all(array.map(isAvailable))
    .then(names => partition(names, Boolean))
}

const test = async word => {
  const [avail, not] = await fetch(word)
  console.log(`${word}: ${avail.length} / ${not.length}`)
}

await test('pinto')

Seems like you don't handle case when there is no response from "got". But I can't figure out why does it happen inside got. Npm registry url is available. Help pls.

npm-name: 6.0.1 System: Linux Node: 12.18

It is thrown in this part of code:

try {
if (isOrganization) {
	await got.head(npmOrganizationUrl + name.toLowerCase(), {timeout: 10000});
} else {
	await got.head(registryUrl + name.toLowerCase(), {timeout: 10000, headers});
}

return false;
} catch (error) {
const {statusCode} = error.response;

if (statusCode === 404) {
	return true;
}

if (isScopedPackage && statusCode === 401) {
	return true;
}

throw error;
}

I was just searching names for my project when I saw that this package does not support scoped packages.

Check https://runkit.com/embed/mzsivagq3l35 for a demo.

Ran into an interesting issue using this module in a nextjs project.

This dependency was not found:  
* electron in ./node_modules/npm-name/node_modules/got/index.js  
To install it, you can run: npm install --save electron

I don't want to introduce electron to the project obviously, and looking through the source of got, I'm not sure why it's being required.

Any insight into this?

Hi, I work on a government company that uses proxies, so this npm-name module was preventing me to use the generator-node from the yeoman team because of the lack of proxy support. As this is a utility for npm name discovery I believe that it makes sense to depend on the configurations from the npmrc file. Firstly it looks for the npmrc file, then for a https-proxy entry and then passes the url of the proxy to got. I had to use proxy-agent because tunnel was giving me a "Parse Error" error. Hope it helps :-)

Alex