Cache package release for compression change in windows (#1281)

* bsd + zstd fallback implementation

* bsd + zstd fallback implementation

* Fix tar operations

* Add -v option for testing

* Fix order of args for tar

* Add GNUtar as default on windows

* Fix test

* Fix tar tests

* Fix lint issues

* Fix windows gnutar test case

* Temporarily remove thhe condition that prevents zstd usage on windows unless with GNUtar

* Address some comments and correct compression commands

* Add windows bsdtar test

* Fix windows test

* Fix test

* Separate args

* Fix old tests

* Add new tests

* Fix tests

* Fix lint test

* Refactor code

* Address review comments

* Fix test

* Fix tar test

* Add await to async function calls

* Fix test

* Update for beta release

* Fix audit issues

* Add fallback to gzip compression if cache not found

* Fix test

* Add test

* Address review comments

* Revert Address review comments

* Release 3.1.0-beta.2 cache package

* Fix issues

* Reconfigure catch block

* Add debug logging for gzip fall back

* Fix test

* Add end to end test for cache using bsd on windows
and address review comments

* Fix test

* Fix test

* Fix tests

* Add better comments

* Update packages/cache/src/internal/cacheHttpClient.ts

Co-authored-by: Bishal Prasad <bishal-pdmsft@github.com>

* Address review comments

* Update for new beta cache package release

* Address bugbash issues

* Fix tests

* Release new actions/cache minor version

Co-authored-by: Lovepreet Singh <pdotl@github.com>
Co-authored-by: Bishal Prasad <bishal-pdmsft@github.com>

packages/cache/src/internal/cacheHttpClient.ts

@@ -104,6 +104,7 @@ export async function getCacheEntry(
   const response = await retryTypedResponse('getCacheEntry', async () =>
     httpClient.getJson<ArtifactCacheEntry>(getCacheApiUrl(resource))
   )
+  // Cache not found
   if (response.statusCode === 204) {
     // List cache for primary key only if cache miss occurs
     if (core.isDebug()) {
@@ -118,6 +119,7 @@ export async function getCacheEntry(
   const cacheResult = response.result
   const cacheDownloadUrl = cacheResult?.archiveLocation
   if (!cacheDownloadUrl) {
+    // Cache achiveLocation not found. This should never happen, and hence bail out.
     throw new Error('Cache not found.')
   }
   core.setSecret(cacheDownloadUrl)

packages/cache/src/internal/cacheUtils.ts

@@ -7,7 +7,11 @@ import * as path from 'path'
 import * as semver from 'semver'
 import * as util from 'util'
 import {v4 as uuidV4} from 'uuid'
-import {CacheFilename, CompressionMethod} from './constants'
+import {
+  CacheFilename,
+  CompressionMethod,
+  GnuTarPathOnWindows
+} from './constants'
 
 // From https://github.com/actions/toolkit/blob/main/packages/tool-cache/src/tool-cache.ts#L23
 export async function createTempDirectory(): Promise<string> {
@@ -90,11 +94,6 @@ async function getVersion(app: string): Promise<string> {
 
 // Use zstandard if possible to maximize cache performance
 export async function getCompressionMethod(): Promise<CompressionMethod> {
-  if (process.platform === 'win32' && !(await isGnuTarInstalled())) {
-    // Disable zstd due to bug https://github.com/actions/cache/issues/301
-    return CompressionMethod.Gzip
-  }
-
   const versionOutput = await getVersion('zstd')
   const version = semver.clean(versionOutput)
 
@@ -116,9 +115,12 @@ export function getCacheFileName(compressionMethod: CompressionMethod): string {
     : CacheFilename.Zstd
 }
 
-export async function isGnuTarInstalled(): Promise<boolean> {
+export async function getGnuTarPathOnWindows(): Promise<string> {
+  if (fs.existsSync(GnuTarPathOnWindows)) {
+    return GnuTarPathOnWindows
+  }
   const versionOutput = await getVersion('tar')
-  return versionOutput.toLowerCase().includes('gnu tar')
+  return versionOutput.toLowerCase().includes('gnu tar') ? io.which('tar') : ''
 }
 
 export function assertDefined<T>(name: string, value?: T): T {

packages/cache/src/internal/constants.ts

@@ -11,6 +11,11 @@ export enum CompressionMethod {
   Zstd = 'zstd'
 }
 
+export enum ArchiveToolType {
+  GNU = 'gnu',
+  BSD = 'bsd'
+}
+
 // The default number of retry attempts.
 export const DefaultRetryAttempts = 2
 
@@ -21,3 +26,13 @@ export const DefaultRetryDelay = 5000
 // over the socket during this period, the socket is destroyed and the download
 // is aborted.
 export const SocketTimeout = 5000
+
+// The default path of GNUtar on hosted Windows runners
+export const GnuTarPathOnWindows = `${process.env['PROGRAMFILES']}\\Git\\usr\\bin\\tar.exe`
+
+// The default path of BSDtar on hosted Windows runners
+export const SystemTarPathOnWindows = `${process.env['SYSTEMDRIVE']}\\Windows\\System32\\tar.exe`
+
+export const TarFilename = 'cache.tar'
+
+export const ManifestFilename = 'manifest.txt'

packages/cache/src/internal/contracts.d.ts

@@ -37,3 +37,8 @@ export interface InternalCacheOptions {
   compressionMethod?: CompressionMethod
   cacheSize?: number
 }
+
+export interface ArchiveTool {
+  path: string
+  type: string
+}

packages/cache/src/internal/tar.ts

@@ -3,25 +3,28 @@ import * as io from '@actions/io'
 import {existsSync, writeFileSync} from 'fs'
 import * as path from 'path'
 import * as utils from './cacheUtils'
-import {CompressionMethod} from './constants'
+import {ArchiveTool} from './contracts'
+import {
+  CompressionMethod,
+  SystemTarPathOnWindows,
+  ArchiveToolType,
+  TarFilename,
+  ManifestFilename
+} from './constants'
 
 const IS_WINDOWS = process.platform === 'win32'
 
-async function getTarPath(
-  args: string[],
-  compressionMethod: CompressionMethod
-): Promise<string> {
+// Returns tar path and type: BSD or GNU
+async function getTarPath(): Promise<ArchiveTool> {
   switch (process.platform) {
     case 'win32': {
-      const systemTar = `${process.env['windir']}\\System32\\tar.exe`
-      if (compressionMethod !== CompressionMethod.Gzip) {
-        // We only use zstandard compression on windows when gnu tar is installed due to
-        // a bug with compressing large files with bsdtar + zstd
-        args.push('--force-local')
+      const gnuTar = await utils.getGnuTarPathOnWindows()
+      const systemTar = SystemTarPathOnWindows
+      if (gnuTar) {
+        // Use GNUtar as default on windows
+        return <ArchiveTool>{path: gnuTar, type: ArchiveToolType.GNU}
       } else if (existsSync(systemTar)) {
-        return systemTar
-      } else if (await utils.isGnuTarInstalled()) {
-        args.push('--force-local')
+        return <ArchiveTool>{path: systemTar, type: ArchiveToolType.BSD}
       }
       break
     }
@@ -29,27 +32,133 @@ async function getTarPath(
       const gnuTar = await io.which('gtar', false)
       if (gnuTar) {
         // fix permission denied errors when extracting BSD tar archive with GNU tar - https://github.com/actions/cache/issues/527
-        args.push('--delay-directory-restore')
-        return gnuTar
+        return <ArchiveTool>{path: gnuTar, type: ArchiveToolType.GNU}
+      } else {
+        return <ArchiveTool>{
+          path: await io.which('tar', true),
+          type: ArchiveToolType.BSD
+        }
       }
-      break
     }
     default:
       break
   }
-  return await io.which('tar', true)
+  // Default assumption is GNU tar is present in path
+  return <ArchiveTool>{
+    path: await io.which('tar', true),
+    type: ArchiveToolType.GNU
+  }
 }
 
-async function execTar(
-  args: string[],
+// Return arguments for tar as per tarPath, compressionMethod, method type and os
+async function getTarArgs(
+  tarPath: ArchiveTool,
   compressionMethod: CompressionMethod,
-  cwd?: string
-): Promise<void> {
-  try {
-    await exec(`"${await getTarPath(args, compressionMethod)}"`, args, {cwd})
-  } catch (error) {
-    throw new Error(`Tar failed with error: ${error?.message}`)
+  type: string,
+  archivePath = ''
+): Promise<string[]> {
+  const args = [`"${tarPath.path}"`]
+  const cacheFileName = utils.getCacheFileName(compressionMethod)
+  const tarFile = 'cache.tar'
+  const workingDirectory = getWorkingDirectory()
+  // Speficic args for BSD tar on windows for workaround
+  const BSD_TAR_ZSTD =
+    tarPath.type === ArchiveToolType.BSD &&
+    compressionMethod !== CompressionMethod.Gzip &&
+    IS_WINDOWS
+
+  // Method specific args
+  switch (type) {
+    case 'create':
+      args.push(
+        '--posix',
+        '-cf',
+        BSD_TAR_ZSTD
+          ? tarFile
+          : cacheFileName.replace(new RegExp(`\\${path.sep}`, 'g'), '/'),
+        '--exclude',
+        BSD_TAR_ZSTD
+          ? tarFile
+          : cacheFileName.replace(new RegExp(`\\${path.sep}`, 'g'), '/'),
+        '-P',
+        '-C',
+        workingDirectory.replace(new RegExp(`\\${path.sep}`, 'g'), '/'),
+        '--files-from',
+        ManifestFilename
+      )
+      break
+    case 'extract':
+      args.push(
+        '-xf',
+        BSD_TAR_ZSTD
+          ? tarFile
+          : archivePath.replace(new RegExp(`\\${path.sep}`, 'g'), '/'),
+        '-P',
+        '-C',
+        workingDirectory.replace(new RegExp(`\\${path.sep}`, 'g'), '/')
+      )
+      break
+    case 'list':
+      args.push(
+        '-tf',
+        BSD_TAR_ZSTD
+          ? tarFile
+          : archivePath.replace(new RegExp(`\\${path.sep}`, 'g'), '/'),
+        '-P'
+      )
+      break
   }
+
+  // Platform specific args
+  if (tarPath.type === ArchiveToolType.GNU) {
+    switch (process.platform) {
+      case 'win32':
+        args.push('--force-local')
+        break
+      case 'darwin':
+        args.push('--delay-directory-restore')
+        break
+    }
+  }
+
+  return args
+}
+
+// Returns commands to run tar and compression program
+async function getCommands(
+  compressionMethod: CompressionMethod,
+  type: string,
+  archivePath = ''
+): Promise<string[]> {
+  let args
+
+  const tarPath = await getTarPath()
+  const tarArgs = await getTarArgs(
+    tarPath,
+    compressionMethod,
+    type,
+    archivePath
+  )
+  const compressionArgs =
+    type !== 'create'
+      ? await getDecompressionProgram(tarPath, compressionMethod, archivePath)
+      : await getCompressionProgram(tarPath, compressionMethod)
+  const BSD_TAR_ZSTD =
+    tarPath.type === ArchiveToolType.BSD &&
+    compressionMethod !== CompressionMethod.Gzip &&
+    IS_WINDOWS
+
+  if (BSD_TAR_ZSTD && type !== 'create') {
+    args = [[...compressionArgs].join(' '), [...tarArgs].join(' ')]
+  } else {
+    args = [[...tarArgs].join(' '), [...compressionArgs].join(' ')]
+  }
+
+  if (BSD_TAR_ZSTD) {
+    return args
+  }
+
+  return [args.join(' ')]
 }
 
 function getWorkingDirectory(): string {
@@ -57,37 +166,107 @@ function getWorkingDirectory(): string {
 }
 
 // Common function for extractTar and listTar to get the compression method
-function getCompressionProgram(compressionMethod: CompressionMethod): string[] {
+async function getDecompressionProgram(
+  tarPath: ArchiveTool,
+  compressionMethod: CompressionMethod,
+  archivePath: string
+): Promise<string[]> {
   // -d: Decompress.
   // unzstd is equivalent to 'zstd -d'
   // --long=#: Enables long distance matching with # bits. Maximum is 30 (1GB) on 32-bit OS and 31 (2GB) on 64-bit.
   // Using 30 here because we also support 32-bit self-hosted runners.
+  const BSD_TAR_ZSTD =
+    tarPath.type === ArchiveToolType.BSD &&
+    compressionMethod !== CompressionMethod.Gzip &&
+    IS_WINDOWS
   switch (compressionMethod) {
     case CompressionMethod.Zstd:
-      return [
-        '--use-compress-program',
-        IS_WINDOWS ? 'zstd -d --long=30' : 'unzstd --long=30'
-      ]
+      return BSD_TAR_ZSTD
+        ? [
+            'zstd -d --long=30 --force -o',
+            TarFilename,
+            archivePath.replace(new RegExp(`\\${path.sep}`, 'g'), '/')
+          ]
+        : [
+            '--use-compress-program',
+            IS_WINDOWS ? '"zstd -d --long=30"' : 'unzstd --long=30'
+          ]
     case CompressionMethod.ZstdWithoutLong:
-      return ['--use-compress-program', IS_WINDOWS ? 'zstd -d' : 'unzstd']
+      return BSD_TAR_ZSTD
+        ? [
+            'zstd -d --force -o',
+            TarFilename,
+            archivePath.replace(new RegExp(`\\${path.sep}`, 'g'), '/')
+          ]
+        : ['--use-compress-program', IS_WINDOWS ? '"zstd -d"' : 'unzstd']
     default:
       return ['-z']
   }
 }
 
+// Used for creating the archive
+// -T#: Compress using # working thread. If # is 0, attempt to detect and use the number of physical CPU cores.
+// zstdmt is equivalent to 'zstd -T0'
+// --long=#: Enables long distance matching with # bits. Maximum is 30 (1GB) on 32-bit OS and 31 (2GB) on 64-bit.
+// Using 30 here because we also support 32-bit self-hosted runners.
+// Long range mode is added to zstd in v1.3.2 release, so we will not use --long in older version of zstd.
+async function getCompressionProgram(
+  tarPath: ArchiveTool,
+  compressionMethod: CompressionMethod
+): Promise<string[]> {
+  const cacheFileName = utils.getCacheFileName(compressionMethod)
+  const BSD_TAR_ZSTD =
+    tarPath.type === ArchiveToolType.BSD &&
+    compressionMethod !== CompressionMethod.Gzip &&
+    IS_WINDOWS
+  switch (compressionMethod) {
+    case CompressionMethod.Zstd:
+      return BSD_TAR_ZSTD
+        ? [
+            'zstd -T0 --long=30 --force -o',
+            cacheFileName.replace(new RegExp(`\\${path.sep}`, 'g'), '/'),
+            TarFilename
+          ]
+        : [
+            '--use-compress-program',
+            IS_WINDOWS ? '"zstd -T0 --long=30"' : 'zstdmt --long=30'
+          ]
+    case CompressionMethod.ZstdWithoutLong:
+      return BSD_TAR_ZSTD
+        ? [
+            'zstd -T0 --force -o',
+            cacheFileName.replace(new RegExp(`\\${path.sep}`, 'g'), '/'),
+            TarFilename
+          ]
+        : ['--use-compress-program', IS_WINDOWS ? '"zstd -T0"' : 'zstdmt']
+    default:
+      return ['-z']
+  }
+}
+
+// Executes all commands as separate processes
+async function execCommands(commands: string[], cwd?: string): Promise<void> {
+  for (const command of commands) {
+    try {
+      await exec(command, undefined, {cwd})
+    } catch (error) {
+      throw new Error(
+        `${command.split(' ')[0]} failed with error: ${error?.message}`
+      )
+    }
+  }
+}
+
+// List the contents of a tar
 export async function listTar(
   archivePath: string,
   compressionMethod: CompressionMethod
 ): Promise<void> {
-  const args = [
-    ...getCompressionProgram(compressionMethod),
-    '-tf',
-    archivePath.replace(new RegExp(`\\${path.sep}`, 'g'), '/'),
-    '-P'
-  ]
-  await execTar(args, compressionMethod)
+  const commands = await getCommands(compressionMethod, 'list', archivePath)
+  await execCommands(commands)
 }
 
+// Extract a tar
 export async function extractTar(
   archivePath: string,
   compressionMethod: CompressionMethod
@@ -95,61 +274,21 @@ export async function extractTar(
   // Create directory to extract tar into
   const workingDirectory = getWorkingDirectory()
   await io.mkdirP(workingDirectory)
-  const args = [
-    ...getCompressionProgram(compressionMethod),
-    '-xf',
-    archivePath.replace(new RegExp(`\\${path.sep}`, 'g'), '/'),
-    '-P',
-    '-C',
-    workingDirectory.replace(new RegExp(`\\${path.sep}`, 'g'), '/')
-  ]
-  await execTar(args, compressionMethod)
+  const commands = await getCommands(compressionMethod, 'extract', archivePath)
+  await execCommands(commands)
 }
 
+// Create a tar
 export async function createTar(
   archiveFolder: string,
   sourceDirectories: string[],
   compressionMethod: CompressionMethod
 ): Promise<void> {
   // Write source directories to manifest.txt to avoid command length limits
-  const manifestFilename = 'manifest.txt'
-  const cacheFileName = utils.getCacheFileName(compressionMethod)
   writeFileSync(
-    path.join(archiveFolder, manifestFilename),
+    path.join(archiveFolder, ManifestFilename),
     sourceDirectories.join('\n')
   )
-  const workingDirectory = getWorkingDirectory()
-
-  // -T#: Compress using # working thread. If # is 0, attempt to detect and use the number of physical CPU cores.
-  // zstdmt is equivalent to 'zstd -T0'
-  // --long=#: Enables long distance matching with # bits. Maximum is 30 (1GB) on 32-bit OS and 31 (2GB) on 64-bit.
-  // Using 30 here because we also support 32-bit self-hosted runners.
-  // Long range mode is added to zstd in v1.3.2 release, so we will not use --long in older version of zstd.
-  function getCompressionProgram(): string[] {
-    switch (compressionMethod) {
-      case CompressionMethod.Zstd:
-        return [
-          '--use-compress-program',
-          IS_WINDOWS ? 'zstd -T0 --long=30' : 'zstdmt --long=30'
-        ]
-      case CompressionMethod.ZstdWithoutLong:
-        return ['--use-compress-program', IS_WINDOWS ? 'zstd -T0' : 'zstdmt']
-      default:
-        return ['-z']
-    }
-  }
-  const args = [
-    '--posix',
-    ...getCompressionProgram(),
-    '-cf',
-    cacheFileName.replace(new RegExp(`\\${path.sep}`, 'g'), '/'),
-    '--exclude',
-    cacheFileName.replace(new RegExp(`\\${path.sep}`, 'g'), '/'),
-    '-P',
-    '-C',
-    workingDirectory.replace(new RegExp(`\\${path.sep}`, 'g'), '/'),
-    '--files-from',
-    manifestFilename
-  ]
-  await execTar(args, compressionMethod, archiveFolder)
+  const commands = await getCommands(compressionMethod, 'create')
+  await execCommands(commands, archiveFolder)
 }