From 1b7c82a741ef6d5672200ee01a152011ca446397 Mon Sep 17 00:00:00 2001
From: Andrei Neculau <andrei.neculau@gmail.com>
Date: Sat, 18 Mar 2023 20:15:16 +0100
Subject: [PATCH] handle git+ssh with semver

---
 resolving/git-resolver/src/parsePref.ts | 67 ++++++++++++++-----------
 1 file changed, 39 insertions(+), 28 deletions(-)

diff --git a/resolving/git-resolver/src/parsePref.ts b/resolving/git-resolver/src/parsePref.ts
index 8f66058a74b..30a979a8258 100644
--- a/resolving/git-resolver/src/parsePref.ts
+++ b/resolving/git-resolver/src/parsePref.ts
@@ -38,15 +38,8 @@ export async function parsePref (pref: string): Promise<HostedPackageSpec | null
   if (colonsPos === -1) return null
   const protocol = pref.slice(0, colonsPos)
   if (protocol && gitProtocols.has(protocol.toLocaleLowerCase())) {
-    const urlparse = new URL(escapeColon(pref))
+    const urlparse = new URL(correctUrl(pref))
     if (!urlparse?.protocol) return null
-    const match = urlparse.protocol === 'git+ssh:' && matchGitScp(pref)
-    if (match) {
-      return {
-        ...match,
-        normalizedPref: pref,
-      }
-    }
 
     const committish = (urlparse.hash?.length > 1) ? decodeURIComponent(urlparse.hash.slice(1)) : null
     return {
@@ -58,13 +51,6 @@ export async function parsePref (pref: string): Promise<HostedPackageSpec | null
   return null
 }
 
-function escapeColon (url: string) {
-  if (!url.includes('@')) return url
-  const [front, ...backs] = url.split('@')
-  const escapedBacks = backs.map(e => e.replace(/:([^/\d]|\d+[^:/\d])/, ':/$1'))
-  return [front, ...escapedBacks].join('@')
-}
-
 function urlToFetchSpec (urlparse: URL) {
   urlparse.hash = ''
   const fetchSpec = url.format(urlparse)
@@ -151,18 +137,43 @@ function setGitCommittish (committish: string | null) {
   return { gitCommittish: committish }
 }
 
-function matchGitScp (spec: string) {
-  // git ssh specifiers are overloaded to also use scp-style git
-  // specifiers, so we have to parse those out and treat them special.
-  // They are NOT true URIs, so we can't hand them to `url.parse`.
-  //
-  // This regex looks for things that look like:
-  // git+ssh://git@my.custom.git.com:username/project.git#deadbeef
-  //
-  // ...and various combinations. The username in the beginning is *required*.
-  const matched = spec.match(/^git\+ssh:\/\/([^:]+:[^#]+(?:\.git)?)(?:#(.*))$/i)
-  return (matched != null) && (matched[1].match(/:[0-9]+\/?.*$/i) == null) && {
-    fetchSpec: matched[1],
-    gitCommittish: matched[2],
+// see https://github.com/npm/hosted-git-info/blob/f03bfbd3022c8f6283a991ff879ed97704ac35fa/lib/parse-url.js#L3
+function lastIndexOfBefore (str: string, char: string, beforeChar: string) {
+  const startPosition = str.indexOf(beforeChar)
+  return str.lastIndexOf(char, startPosition > -1 ? startPosition : Infinity)
+}
+
+// see https://github.com/npm/hosted-git-info/blob/f03bfbd3022c8f6283a991ff879ed97704ac35fa/lib/parse-url.js#L41
+// attempt to correct an scp style url so that it will parse with `new URL()`
+function correctUrl (giturl: string) {
+  // 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
 }