feat: add separate static method for just parsing urls

npm · Oct 26, 2022 · 1754842 · 1754842
1 parent 7024fa4
commit 1754842
Show file tree

Hide file tree

Showing 5 changed files with 108 additions and 86 deletions.
diff --git a/lib/from-url.js b/lib/from-url.js
@@ -1,44 +1,6 @@
 'use strict'
 
-const url = require('url')
-
-const safeUrl = (u) => {
-  try {
-    return new url.URL(u)
-  } catch {
-    // this fn should never throw
-  }
-}
-
-const lastIndexOfBefore = (str, char, beforeChar) => {
-  const startPosition = str.indexOf(beforeChar)
-  return str.lastIndexOf(char, startPosition > -1 ? startPosition : Infinity)
-}
-
-// accepts input like git:github.com:user/repo and inserts the // after the first :
-const correctProtocol = (arg, protocols) => {
-  const firstColon = arg.indexOf(':')
-  const proto = arg.slice(0, firstColon + 1)
-  if (Object.prototype.hasOwnProperty.call(protocols, proto)) {
-    return arg
-  }
-
-  const firstAt = arg.indexOf('@')
-  if (firstAt > -1) {
-    if (firstAt > firstColon) {
-      return `git+ssh://${arg}`
-    } else {
-      return arg
-    }
-  }
-
-  const doubleSlash = arg.indexOf('//')
-  if (doubleSlash === firstColon + 1) {
-    return arg
-  }
-
-  return `${arg.slice(0, firstColon + 1)}//${arg.slice(firstColon + 1)}`
-}
+const correctUrl = require('./parse-url')
 
 // look for github shorthand inputs, such as npm/cli
 const isGitHubShorthand = (arg) => {
@@ -71,49 +33,13 @@ const isGitHubShorthand = (arg) => {
     secondSlashOnlyAfterHash
 }
 
-// attempt to correct an scp style url so that it will parse with `new URL()`
-const correctUrl = (giturl) => {
-  // ignore @ that come after the first hash since the denotes the start
-  // of a committish which can contain @ characters
-  const firstAt = lastIndexOfBefore(giturl, '@', '#')
-  // ignore colons that come after the hash since that could include colons such as:
-  // git@github.com:user/package-2#semver:^1.0.0
-  const lastColonBeforeHash = lastIndexOfBefore(giturl, ':', '#')
-
-  if (lastColonBeforeHash > firstAt) {
-    // the last : comes after the first @ (or there is no @)
-    // like it would in:
-    // proto://hostname.com:user/repo
-    // username@hostname.com:user/repo
-    // :password@hostname.com:user/repo
-    // username:password@hostname.com:user/repo
-    // proto://username@hostname.com:user/repo
-    // proto://:password@hostname.com:user/repo
-    // proto://username:password@hostname.com:user/repo
-    // then we replace the last : with a / to create a valid path
-    giturl = giturl.slice(0, lastColonBeforeHash) + '/' + giturl.slice(lastColonBeforeHash + 1)
-  }
-
-  if (lastIndexOfBefore(giturl, ':', '#') === -1 && giturl.indexOf('//') === -1) {
-    // we have no : at all
-    // as it would be in:
-    // username@hostname.com/user/repo
-    // then we prepend a protocol
-    giturl = `git+ssh://${giturl}`
-  }
-
-  return giturl
-}
-
 module.exports = (giturl, opts, { gitHosts, protocols }) => {
   if (!giturl) {
     return
   }
 
-  const correctedUrl = isGitHubShorthand(giturl)
-    ? `github:${giturl}`
-    : correctProtocol(giturl, protocols)
-  const parsed = safeUrl(correctedUrl) || safeUrl(correctUrl(correctedUrl))
+  const correctedUrl = isGitHubShorthand(giturl) ? `github:${giturl}` : giturl
+  const parsed = correctUrl(correctedUrl, protocols)
   if (!parsed) {
     return
   }

diff --git a/lib/index.js b/lib/index.js
@@ -3,6 +3,8 @@
 const LRU = require('lru-cache')
 const hosts = require('./hosts.js')
 const fromUrl = require('./from-url.js')
+const parseUrl = require('./parse-url.js')
+const getProtocols = require('./protocols.js')
 
 const cache = new LRU({ max: 1000 })
 
@@ -20,15 +22,7 @@ class GitHost {
   }
 
   static #gitHosts = { byShortcut: {}, byDomain: {} }
-  static #protocols = {
-    'git+ssh:': { name: 'sshurl' },
-    'ssh:': { name: 'sshurl' },
-    'git+https:': { name: 'https', auth: true },
-    'git:': { auth: true },
-    'http:': { auth: true },
-    'https:': { auth: true },
-    'git+http:': { auth: true },
-  }
+  static #protocols = getProtocols()
 
   static addHost (name, host) {
     GitHost.#gitHosts[name] = host
@@ -55,6 +49,10 @@ class GitHost {
     return cache.get(key)
   }
 
+  static parseUrl (url) {
+    return parseUrl(url)
+  }
+
   #fill (template, opts) {
     if (typeof template !== 'function') {
       return null

diff --git a/lib/parse-url.js b/lib/parse-url.js
@@ -0,0 +1,79 @@
+const url = require('url')
+const getProtocols = require('./protocols.js')
+
+const lastIndexOfBefore = (str, char, beforeChar) => {
+  const startPosition = str.indexOf(beforeChar)
+  return str.lastIndexOf(char, startPosition > -1 ? startPosition : Infinity)
+}
+
+const safeUrl = (u) => {
+  try {
+    return new url.URL(u)
+  } catch {
+    // this fn should never throw
+  }
+}
+
+// accepts input like git:github.com:user/repo and inserts the // after the first :
+const correctProtocol = (arg, protocols) => {
+  const firstColon = arg.indexOf(':')
+  const proto = arg.slice(0, firstColon + 1)
+  if (Object.prototype.hasOwnProperty.call(protocols, proto)) {
+    return arg
+  }
+
+  const firstAt = arg.indexOf('@')
+  if (firstAt > -1) {
+    if (firstAt > firstColon) {
+      return `git+ssh://${arg}`
+    } else {
+      return arg
+    }
+  }
+
+  const doubleSlash = arg.indexOf('//')
+  if (doubleSlash === firstColon + 1) {
+    return arg
+  }
+
+  return `${arg.slice(0, firstColon + 1)}//${arg.slice(firstColon + 1)}`
+}
+
+// attempt to correct an scp style url so that it will parse with `new URL()`
+const correctUrl = (giturl) => {
+  // ignore @ that come after the first hash since the denotes the start
+  // of a committish which can contain @ characters
+  const firstAt = lastIndexOfBefore(giturl, '@', '#')
+  // ignore colons that come after the hash since that could include colons such as:
+  // git@github.com:user/package-2#semver:^1.0.0
+  const lastColonBeforeHash = lastIndexOfBefore(giturl, ':', '#')
+
+  if (lastColonBeforeHash > firstAt) {
+    // the last : comes after the first @ (or there is no @)
+    // like it would in:
+    // proto://hostname.com:user/repo
+    // username@hostname.com:user/repo
+    // :password@hostname.com:user/repo
+    // username:password@hostname.com:user/repo
+    // proto://username@hostname.com:user/repo
+    // proto://:password@hostname.com:user/repo
+    // proto://username:password@hostname.com:user/repo
+    // then we replace the last : with a / to create a valid path
+    giturl = giturl.slice(0, lastColonBeforeHash) + '/' + giturl.slice(lastColonBeforeHash + 1)
+  }
+
+  if (lastIndexOfBefore(giturl, ':', '#') === -1 && giturl.indexOf('//') === -1) {
+    // we have no : at all
+    // as it would be in:
+    // username@hostname.com/user/repo
+    // then we prepend a protocol
+    giturl = `git+ssh://${giturl}`
+  }
+
+  return giturl
+}
+
+module.exports = (giturl, protocols = getProtocols()) => {
+  const withProtocol = correctProtocol(giturl, protocols)
+  return safeUrl(withProtocol) || safeUrl(correctUrl(withProtocol))
+}
diff --git a/lib/protocols.js b/lib/protocols.js
@@ -0,0 +1,9 @@
+module.exports = () => ({
+  'git+ssh:': { name: 'sshurl' },
+  'ssh:': { name: 'sshurl' },
+  'git+https:': { name: 'https', auth: true },
+  'git:': { auth: true },
+  'http:': { auth: true },
+  'https:': { auth: true },
+  'git+http:': { auth: true },
+})
diff --git a/test/parse-url.js b/test/parse-url.js
@@ -0,0 +1,10 @@
+const t = require('tap')
+const HostedGit = require('..')
+const parseUrl = require('../lib/parse-url.js')
+
+t.test('can parse git+ssh url by default', async t => {
+  // https://github.com/npm/cli/issues/5278
+  const u = 'git+ssh://git@abc:frontend/utils.git#6d45447e0c5eb6cd2e3edf05a8c5a9bb81950c79'
+  t.ok(parseUrl(u))
+  t.ok(HostedGit.parseUrl(u))
+})