fix: preserve registry path when replacing a host

owlstronaut · owlstronaut · commit 0886e7abced0 · 2025-04-10T13:52:58.000-07:00
diff --git a/workspaces/arborist/lib/arborist/reify.js b/workspaces/arborist/lib/arborist/reify.js
@@ -895,9 +895,19 @@ module.exports = cls => class Reifier extends cls {
       if ((this.options.replaceRegistryHost === resolvedURL.hostname) ||
            this.options.replaceRegistryHost === 'always') {
         const registryURL = new URL(this.registry)
+
         // Replace the host with the registry host while keeping the path intact
         resolvedURL.hostname = registryURL.hostname
         resolvedURL.port = registryURL.port
+
+        // Make sure we don't double-include the path if it's already there
+        const registryPath = registryURL.pathname.replace(/\/$/, '')
+
+        if (registryPath && registryPath !== '/' && !resolvedURL.pathname.startsWith(registryPath)) {
+          // Since hostname is changed, we need to ensure the registry path is included
+          resolvedURL.pathname = registryPath + resolvedURL.pathname
+        }
+
         return resolvedURL.toString()
       }
       return resolved
diff --git a/workspaces/arborist/test/arborist/reify.js b/workspaces/arborist/test/arborist/reify.js
@@ -3506,6 +3506,61 @@ t.test('should preserve exact ranges, missing actual tree', async (t) => {
     })
     await arb.reify()
   })
+
+  t.test('registry path prepending', async t => {
+    // A registry path is prepended to resolved URLs that don't already have it
+    const abbrevPackument4 = JSON.stringify({
+      _id: 'abbrev',
+      _rev: 'lkjadflkjasdf',
+      name: 'abbrev',
+      'dist-tags': { latest: '1.1.1' },
+      versions: {
+        '1.1.1': {
+          name: 'abbrev',
+          version: '1.1.1',
+          dist: {
+            // Note: This URL has no path component that matches our registry path
+            tarball: 'https://external-registry.example.com/abbrev-1.1.1.tgz',
+          },
+        },
+      },
+    })
+
+    const testdir = t.testdir({
+      project: {
+        'package.json': JSON.stringify({
+          name: 'myproject',
+          version: '1.0.0',
+          dependencies: {
+            abbrev: '1.1.1',
+          },
+        }),
+      },
+    })
+
+    // Set up the registry with a deep path
+    const registryHost = 'https://registry.example.com'
+    const registryPath = '/custom/deep/path/registry'
+    const registry = `${registryHost}${registryPath}`
+
+    tnock(t, registryHost)
+      .get(`${registryPath}/abbrev`)
+      .reply(200, abbrevPackument4)
+
+    // This is the critical test - the tarball URL in the packument doesn't have our registry path, but when replaceRegistryHost is 'always', we should get a request to this URL which includes the registry path
+    tnock(t, registryHost)
+      .get(`${registryPath}/abbrev-1.1.1.tgz`)
+      .reply(200, abbrevTGZ)
+
+    const arb = new Arborist({
+      path: resolve(testdir, 'project'),
+      registry,
+      cache: resolve(testdir, 'cache'),
+      replaceRegistryHost: 'always',
+    })
+
+    await t.resolves(arb.reify(), 'reify should complete successfully')
+  })
 })
 
 t.test('install stategy linked', async (t) => {
