feat: add HostedGitInfo.fromManifest (#288)

ljharb · web-flow · commit ef0865cc5c28 · 2025-04-14T10:35:15.000-07:00
This encapsulates the logic used in `npm repo`

Per Slack discussion. Once this is published and updated in the cli,
I'll happily adapt the `npm repo` code to use this method.

The method will either return a HGI instance, a string URL, or `null`.
diff --git a/README.md b/README.md
@@ -19,7 +19,7 @@ const info = hostedGitInfo.fromUrl("git@github.com:npm/hosted-git-info.git", opt
 */
 ```
 
-If the URL can't be matched with a git host, `null` will be returned.  We
+If the URL can't be matched with a git host, `null` will be returned. We
 can match git, ssh and https urls.  Additionally, we can match ssh connect
 strings (`git@github.com:npm/hosted-git-info`) and shortcuts (eg,
 `github:npm/hosted-git-info`).  GitHub specifically, is detected in the case
@@ -59,6 +59,11 @@ Implications:
   * *noCommittish* — If true then committishes won't be included in generated URLs.
   * *noGitPlus* — If true then `git+` won't be prefixed on URLs.
 
+### const infoOrURL = hostedGitInfo.fromManifest(manifest[, options])
+
+* *manifest* is a package manifest, such as that returned by [`pacote.manifest()`](https://npmjs.com/pacote)
+* *options* is an optional object. It can have the same properties as `fromUrl` above.
+
 ## Methods
 
 All of the methods take the same options as the `fromUrl` factory.  Options
diff --git a/lib/index.js b/lib/index.js
@@ -7,6 +7,26 @@ const parseUrl = require('./parse-url.js')
 
 const cache = new LRUCache({ max: 1000 })
 
+function unknownHostedUrl (url) {
+  try {
+    const {
+      protocol,
+      hostname,
+      pathname,
+    } = new URL(url)
+
+    if (!hostname) {
+      return null
+    }
+
+    const proto = /(?:git\+)http:$/.test(protocol) ? 'http:' : 'https:'
+    const path = pathname.replace(/\.git$/, '')
+    return `${proto}//${hostname}${path}`
+  } catch {
+    return null
+  }
+}
+
 class GitHost {
   constructor (type, user, auth, project, committish, defaultRepresentation, opts = {}) {
     Object.assign(this, GitHost.#gitHosts[type], {
@@ -56,6 +76,34 @@ class GitHost {
     return cache.get(key)
   }
 
+  static fromManifest (manifest, opts = {}) {
+    if (!manifest || typeof manifest !== 'object') {
+      return
+    }
+
+    const r = manifest.repository
+    // TODO: look into also checking the `bugs`/`homepage` URLs
+
+    const rurl = r && (
+      typeof r === 'string'
+        ? r
+        : typeof r === 'object' && typeof r.url === 'string'
+          ? r.url
+          : null
+    )
+
+    if (!rurl) {
+      throw new Error('no repository')
+    }
+
+    const info = (rurl && GitHost.fromUrl(rurl.replace(/^git\+/, ''), opts)) || null
+    if (info) {
+      return info
+    }
+    const unk = unknownHostedUrl(rurl)
+    return GitHost.fromUrl(unk, opts) || unk
+  }
+
   static parseUrl (url) {
     return parseUrl(url)
   }
diff --git a/test/file.js b/test/file.js
@@ -0,0 +1,14 @@
+const HostedGit = require('..')
+const t = require('tap')
+
+t.test('file:// URLs', t => {
+  const fileRepo = {
+    name: 'foo',
+    repository: {
+      url: 'file:///path/dot.git',
+    },
+  }
+  t.equal(HostedGit.fromManifest(fileRepo), null)
+
+  t.end()
+})
diff --git a/test/github.js b/test/github.js
@@ -270,3 +270,80 @@ t.test('string methods populate correctly', t => {
 
   t.end()
 })
+
+t.test('from manifest', t => {
+  t.equal(HostedGit.fromManifest(), undefined, 'no manifest returns undefined')
+  t.equal(HostedGit.fromManifest(), undefined, 'no manifest returns undefined')
+  t.equal(HostedGit.fromManifest(false), undefined, 'false manifest returns undefined')
+  t.equal(HostedGit.fromManifest(() => {}), undefined, 'function manifest returns undefined')
+
+  const unknownHostRepo = {
+    name: 'foo',
+    repository: {
+      url: 'https://nope.com',
+    },
+  }
+  t.same(HostedGit.fromManifest(unknownHostRepo), 'https://nope.com/')
+
+  const insecureUnknownHostRepo = {
+    name: 'foo',
+    repository: {
+      url: 'http://nope.com',
+    },
+  }
+  t.same(HostedGit.fromManifest(insecureUnknownHostRepo), 'https://nope.com/')
+
+  const insecureGitUnknownHostRepo = {
+    name: 'foo',
+    repository: {
+      url: 'git+http://nope.com',
+    },
+  }
+  t.same(HostedGit.fromManifest(insecureGitUnknownHostRepo), 'http://nope.com')
+
+  const badRepo = {
+    name: 'foo',
+    repository: {
+      url: '#',
+    },
+  }
+  t.equal(HostedGit.fromManifest(badRepo), null)
+
+  const manifest = {
+    name: 'foo',
+    repository: {
+      type: 'git',
+      url: 'git+ssh://github.com/foo/bar.git',
+    },
+  }
+
+  const parsed = HostedGit.fromManifest(manifest)
+  t.same(parsed.browse(), 'https://github.com/foo/bar')
+
+  const monorepo = {
+    name: 'clowncar',
+    repository: {
+      type: 'git',
+      url: 'git+ssh://github.com/foo/bar.git',
+      directory: 'packages/foo',
+    },
+  }
+
+  const honk = HostedGit.fromManifest(monorepo)
+  t.same(honk.browse(monorepo.repository.directory), 'https://github.com/foo/bar/tree/HEAD/packages/foo')
+
+  const stringRepo = {
+    name: 'foo',
+    repository: 'git+ssh://github.com/foo/bar.git',
+  }
+  const stringRepoParsed = HostedGit.fromManifest(stringRepo)
+  t.same(stringRepoParsed.browse(), 'https://github.com/foo/bar')
+
+  const nonStringRepo = {
+    name: 'foo',
+    repository: 42,
+  }
+  t.throws(() => HostedGit.fromManifest(nonStringRepo))
+
+  t.end()
+})
diff --git a/test/gitlab.js b/test/gitlab.js
@@ -321,3 +321,80 @@ t.test('string methods populate correctly', t => {
 
   t.end()
 })
+
+t.test('from manifest', t => {
+  t.equal(HostedGit.fromManifest(), undefined, 'no manifest returns undefined')
+  t.equal(HostedGit.fromManifest(), undefined, 'no manifest returns undefined')
+  t.equal(HostedGit.fromManifest(false), undefined, 'false manifest returns undefined')
+  t.equal(HostedGit.fromManifest(() => {}), undefined, 'function manifest returns undefined')
+
+  const unknownHostRepo = {
+    name: 'foo',
+    repository: {
+      url: 'https://nope.com',
+    },
+  }
+  t.same(HostedGit.fromManifest(unknownHostRepo), 'https://nope.com/')
+
+  const insecureUnknownHostRepo = {
+    name: 'foo',
+    repository: {
+      url: 'http://nope.com',
+    },
+  }
+  t.same(HostedGit.fromManifest(insecureUnknownHostRepo), 'https://nope.com/')
+
+  const insecureGitUnknownHostRepo = {
+    name: 'foo',
+    repository: {
+      url: 'git+http://nope.com',
+    },
+  }
+  t.same(HostedGit.fromManifest(insecureGitUnknownHostRepo), 'http://nope.com')
+
+  const badRepo = {
+    name: 'foo',
+    repository: {
+      url: '#',
+    },
+  }
+  t.equal(HostedGit.fromManifest(badRepo), null)
+
+  const manifest = {
+    name: 'foo',
+    repository: {
+      type: 'git',
+      url: 'git+ssh://gitlab.com/foo/bar.git',
+    },
+  }
+
+  const parsed = HostedGit.fromManifest(manifest)
+  t.same(parsed.browse(), 'https://gitlab.com/foo/bar')
+
+  const monorepo = {
+    name: 'clowncar',
+    repository: {
+      type: 'git',
+      url: 'git+ssh://gitlab.com/foo/bar.git',
+      directory: 'packages/foo',
+    },
+  }
+
+  const honk = HostedGit.fromManifest(monorepo)
+  t.same(honk.browse(monorepo.repository.directory), 'https://gitlab.com/foo/bar/tree/HEAD/packages/foo')
+
+  const stringRepo = {
+    name: 'foo',
+    repository: 'git+ssh://gitlab.com/foo/bar.git',
+  }
+  const stringRepoParsed = HostedGit.fromManifest(stringRepo)
+  t.same(stringRepoParsed.browse(), 'https://gitlab.com/foo/bar')
+
+  const nonStringRepo = {
+    name: 'foo',
+    repository: 42,
+  }
+  t.throws(() => HostedGit.fromManifest(nonStringRepo))
+
+  t.end()
+})
