fix(arborist): clean-up orphaned .store/ entries on linked install

Author: manzoorwanijk

workspaces/arborist/lib/arborist/reify.js

@@ -10,7 +10,7 @@ const { callLimit: promiseCallLimit } = require('promise-call-limit')
 const { depth: dfwalk } = require('treeverse')
 const { dirname, resolve, relative, join } = require('node:path')
 const { log, time } = require('proc-log')
-const { lstat, mkdir, rm, symlink } = require('node:fs/promises')
+const { lstat, mkdir, readdir, rm, symlink } = require('node:fs/promises')
 const { moveFile } = require('@npmcli/fs')
 const { subset, intersects } = require('semver')
 const { walkUp } = require('walk-up-path')
@@ -120,6 +120,7 @@ module.exports = cls => class Reifier extends cls {
     await this[_diffTrees]()
     await this.#reifyPackages()
     if (linked) {
+      await this.#cleanOrphanedStoreEntries()
       // swap back in the idealTree
       // so that the lockfile is preserved
       this.idealTree = oldTree
@@ -1247,6 +1248,44 @@ module.exports = cls => class Reifier extends cls {
     timeEnd()
   }
 
+  // After a linked install, scan node_modules/.store/ and remove any
+  // directories that are not referenced by the current ideal tree.
+  // Store entries become orphaned when dependencies are updated or
+  // removed, because the diff never sees the old store keys.
+  async #cleanOrphanedStoreEntries () {
+    const storeDir = resolve(this.path, 'node_modules', '.store')
+    let entries
+    try {
+      entries = await readdir(storeDir)
+    } catch {
+      return
+    }
+
+    // Collect valid store keys from the isolated ideal tree.
+    // Store entries have location: node_modules/.store/{key}/node_modules/{pkg}
+    const validKeys = new Set()
+    for (const child of this.idealTree.children) {
+      if (child.isInStore) {
+        const key = child.location.split('/')[2]
+        validKeys.add(key)
+      }
+    }
+
+    const orphaned = entries.filter(e => !validKeys.has(e))
+    if (!orphaned.length) {
+      return
+    }
+
+    log.silly('reify', 'cleaning orphaned store entries', orphaned)
+    await promiseAllRejectLate(
+      orphaned.map(e =>
+        rm(resolve(storeDir, e), { recursive: true, force: true })
+          .catch(/* istanbul ignore next - rm with force rarely fails */
+            er => log.warn('cleanup', `Failed to remove orphaned store entry ${e}`, er))
+      )
+    )
+  }
+
   // last but not least, we save the ideal tree metadata to the package-lock
   // or shrinkwrap file, and any additions or removals to package.json
   async [_saveIdealTree] (options) {

workspaces/arborist/test/isolated-mode.js

@@ -1637,6 +1637,85 @@ tap.test('bins are installed', async t => {
   t.ok(binFromBarToWhich)
 })
 
+tap.test('orphaned store entries are cleaned up on dependency update', async t => {
+  const graph = {
+    registry: [
+      { name: 'which', version: '1.0.0', dependencies: { isexe: '^1.0.0' } },
+      { name: 'which', version: '2.0.0', dependencies: { isexe: '^1.0.0' } },
+      { name: 'isexe', version: '1.0.0' },
+    ],
+    root: {
+      name: 'myproject',
+      version: '1.0.0',
+      dependencies: { which: '1.0.0' },
+    },
+  }
+  const { dir, registry } = await getRepo(graph)
+  const cache = fs.mkdtempSync(`${getTempDir()}/test-`)
+  const storeDir = path.join(dir, 'node_modules', '.store')
+
+  // First install — which@1.0.0
+  const arb1 = new Arborist({ path: dir, registry, packumentCache: new Map(), cache })
+  await arb1.reify({ installStrategy: 'linked' })
+
+  const entriesAfterV1 = fs.readdirSync(storeDir)
+  t.ok(entriesAfterV1.some(e => e.startsWith('which@1.0.0-')),
+    'store has which@1.0.0 entry after first install')
+
+  // Update package.json to depend on which@2.0.0
+  const pkgPath = path.join(dir, 'package.json')
+  const pkg = JSON.parse(fs.readFileSync(pkgPath, 'utf8'))
+  pkg.dependencies.which = '2.0.0'
+  fs.writeFileSync(pkgPath, JSON.stringify(pkg))
+
+  // Second install — which@2.0.0
+  const arb2 = new Arborist({ path: dir, registry, packumentCache: new Map(), cache })
+  await arb2.reify({ installStrategy: 'linked' })
+
+  const entriesAfterV2 = fs.readdirSync(storeDir)
+  t.ok(entriesAfterV2.some(e => e.startsWith('which@2.0.0-')),
+    'store has which@2.0.0 entry after update')
+  t.notOk(entriesAfterV2.some(e => e.startsWith('which@1.0.0-')),
+    'old which@1.0.0 store entry is removed after update')
+})
+
+tap.test('orphaned store entries are cleaned up on dependency removal', async t => {
+  const graph = {
+    registry: [
+      { name: 'which', version: '1.0.0', dependencies: { isexe: '^1.0.0' } },
+      { name: 'isexe', version: '1.0.0' },
+    ],
+    root: {
+      name: 'myproject',
+      version: '1.0.0',
+      dependencies: { which: '1.0.0' },
+    },
+  }
+  const { dir, registry } = await getRepo(graph)
+  const cache = fs.mkdtempSync(`${getTempDir()}/test-`)
+  const storeDir = path.join(dir, 'node_modules', '.store')
+
+  // First install
+  const arb1 = new Arborist({ path: dir, registry, packumentCache: new Map(), cache })
+  await arb1.reify({ installStrategy: 'linked' })
+
+  t.ok(fs.readdirSync(storeDir).length > 0, 'store has entries after install')
+
+  // Remove the dependency
+  const pkgPath = path.join(dir, 'package.json')
+  const pkg = JSON.parse(fs.readFileSync(pkgPath, 'utf8'))
+  delete pkg.dependencies
+  fs.writeFileSync(pkgPath, JSON.stringify(pkg))
+
+  // Reinstall
+  const arb2 = new Arborist({ path: dir, registry, packumentCache: new Map(), cache })
+  await arb2.reify({ installStrategy: 'linked' })
+
+  const entriesAfterRemoval = fs.readdirSync(storeDir)
+  t.equal(entriesAfterRemoval.length, 0,
+    'all store entries are removed when dependencies are removed')
+})
+
 function setupRequire (cwd) {
   return function requireChain (...chain) {
     return chain.reduce((path, name) => {