Improve core logic

.github/workflows/workflow.yml

@@ -7,7 +7,7 @@ jobs:
     strategy:
       matrix:
         os: [macOS-latest]
-        node: [8.5.0, 13]
+        node: [8.3.0, 13]
       fail-fast: false
     runs-on: ${{ matrix.os }}
     steps:

package.json

@@ -24,15 +24,19 @@
   },
   "dependencies": {
     "archiver": "^3.0.0",
-    "debug": "^4.1.1",
+    "common-path-prefix": "^2.0.0",
+    "cp-file": "^7.0.0",
     "elf-tools": "^1.1.1",
+    "end-of-stream": "^1.4.1",
     "glob": "^7.1.3",
-    "npm-packlist": "^1.1.12",
-    "p-all": "^2.0.0",
+    "make-dir": "^3.0.0",
+    "p-map": "^3.0.0",
+    "path-exists": "^4.0.0",
+    "pkg-dir": "^4.2.0",
     "precinct": "^6.1.1",
-    "read-pkg-up": "^7.0.0",
     "require-package-name": "^2.0.1",
     "resolve": "^1.10.0",
+    "util.promisify": "^1.0.0",
     "yargs": "^14.2.0"
   },
   "devDependencies": {
@@ -62,11 +66,10 @@
     "pkg": "^4.3.7",
     "prettier": "^1.16.4",
     "rimraf": "^3.0.0",
-    "tmp-promise": "^2.0.2",
-    "util.promisify": "^1.0.0"
+    "tmp-promise": "^2.0.2"
   },
   "engines": {
-    "node": ">=8.5.0"
+    "node": ">=8.3.0"
   },
   "homepage": "https://github.com/netlify/zip-it-and-ship-it#README",
   "keywords": [
@@ -79,7 +82,7 @@
     "static"
   ],
   "license": "MIT",
-  "main": "index.js",
+  "main": "src/main.js",
   "pkg": {
     "targets": [
       "node10"
@@ -93,7 +96,7 @@
     "prepublishOnly": "run-s pkg && git push && git push --tags && gh-release -a build/zip-it-and-ship-it_$(git describe --abbrev=0 --tags)_Linux-64bit.tar.gz,build/zip-it-and-ship-it_$(git describe --abbrev=0 --tags)_macOS-64bit.tar.gz",
     "test": "run-s test:* test:dev:*",
     "test-ci": "run-s test:* test:ci:*",
-    "test:lint": "eslint --ignore-path .gitignore --fix --cache --format=codeframe --max-warnings=0 \"src/**/*.js\" \"*.js\"",
+    "test:lint": "eslint --ignore-path .gitignore --fix --cache --format=codeframe --max-warnings=0 \"src/**/*.js\"",
     "test:prettier": "prettier --ignore-path .gitignore --write --loglevel warn \"src/**/*.js\" \"*.{js,md,yml,json}\"",
     "test:dev:ava": "ava",
     "test:ci:ava": "nyc -r lcovonly -r text -r json ava",

src/archive.js

@@ -0,0 +1,34 @@
+const { createWriteStream } = require('fs')
+
+const archiver = require('archiver')
+const endOfStream = require('end-of-stream')
+const promisify = require('util.promisify')
+
+const pEndOfStream = promisify(endOfStream)
+
+// Start zipping files
+const startZip = function(destPath) {
+  const output = createWriteStream(destPath)
+  const archive = archiver('zip', { level: 9 })
+  archive.pipe(output)
+  return { archive, output }
+}
+
+// Add new file to zip
+const addZipFile = function(archive, file, name, stat) {
+  // Ensure sha256 stability regardless of mtime
+  archive.file(file, { name, mode: stat.mode, date: new Date(0), stats: stat })
+}
+
+// Add new file content to zip
+const addZipContent = function(archive, content, name) {
+  archive.append(content, { name, date: new Date(0) })
+}
+
+// End zipping files
+const endZip = async function(archive, output) {
+  archive.finalize()
+  await pEndOfStream(output)
+}
+
+module.exports = { startZip, addZipFile, addZipContent, endZip }

src/bin.test.js

@@ -22,7 +22,7 @@ test('CLI | --help', async t => {
   t.true(stdout.includes('Options:'))
 })
 
-test.skip('CLI | Normal execution', async t => {
+test('CLI | Normal execution', async t => {
   const tmpDir = await tmpName({ prefix: 'zip-it-test' })
   const { stdout } = await execa.command(`${BINARY_PATH} ${join(FIXTURES_DIR, 'simple')} ${tmpDir}`)
   const zipped = JSON.parse(stdout)

src/dependencies.js

@@ -0,0 +1,213 @@
+const { dirname } = require('path')
+
+const precinct = require('precinct')
+const resolve = require('resolve')
+const requirePackageName = require('require-package-name')
+const promisify = require('util.promisify')
+const glob = require('glob')
+
+const pResolve = promisify(resolve)
+const pGlob = promisify(glob)
+
+// Retrieve all the files recursively required by a Node.js file
+const getDependencies = async function(handler, packageRoot) {
+  const packageJson = getPackageJson(packageRoot)
+
+  const state = { localFiles: [], modulePaths: [] }
+
+  try {
+    return await getFileDependencies(handler, packageJson, state)
+  } catch (error) {
+    error.message = `In file "${handler}": ${error.message}`
+    throw error
+  }
+}
+
+const getPackageJson = function(packageRoot) {
+  if (packageRoot === undefined) {
+    return {}
+  }
+
+  return require(`${packageRoot}/package.json`)
+}
+
+const getFileDependencies = async function(path, packageJson, state) {
+  if (state.localFiles.includes(path)) {
+    return []
+  }
+
+  state.localFiles.push(path)
+
+  const basedir = dirname(path)
+  // This parses JavaScript in `path` to retrieve all the `require()` statements
+  // TODO: `precinct.paperwork()` uses `fs.readFileSync()` under the hood,
+  // but should use `fs.readFile()` instead
+  const dependencies = precinct.paperwork(path, { includeCore: false })
+
+  const depsPaths = await Promise.all(
+    dependencies.map(dependency => getImportDependencies(dependency, basedir, packageJson, state))
+  )
+  return [].concat(...depsPaths)
+}
+
+// `require()` statements can be either `require('moduleName')` or
+// `require(path)`
+const getImportDependencies = function(dependency, basedir, packageJson, state) {
+  if (LOCAL_IMPORT_REGEXP.test(dependency)) {
+    return getLocalImportDependencies(dependency, basedir, packageJson, state)
+  }
+
+  return getModuleDependencies(dependency, basedir, state, packageJson)
+}
+
+const LOCAL_IMPORT_REGEXP = /^(\.|\/)/
+
+// When a file requires another one, we apply the top-level logic recursively
+const getLocalImportDependencies = async function(dependency, basedir, packageJson, state) {
+  const dependencyPath = await pResolve(dependency, { basedir })
+  const depsPath = await getFileDependencies(dependencyPath, packageJson, state)
+  return [dependencyPath, ...depsPath]
+}
+
+// When a file requires a module, we find its path inside `node_modules` and
+// use all its published files. We also recurse on the module's dependencies.
+const getModuleDependencies = async function(dependency, basedir, state, packageJson) {
+  const moduleName = requirePackageName(dependency.replace(BACKSLASH_REGEXP, '/'))
+
+  try {
+    return await getModuleNameDependencies(moduleName, basedir, state)
+  } catch (error) {
+    return handleModuleNotFound({ error, moduleName, packageJson })
+  }
+}
+
+const BACKSLASH_REGEXP = /\\/g
+
+const getModuleNameDependencies = async function(moduleName, basedir, state) {
+  if (EXCLUDED_MODULES.includes(moduleName)) {
+    return []
+  }
+
+  // Find the Node.js module directory path
+  const packagePath = await pResolve(`${moduleName}/package.json`, { basedir })
+  const modulePath = dirname(packagePath)
+
+  if (state.modulePaths.includes(modulePath)) {
+    return []
+  }
+
+  state.modulePaths.push(modulePath)
+
+  const pkg = require(packagePath)
+
+  const [publishedFiles, depsPaths] = await Promise.all([
+    getPublishedFiles(modulePath),
+    getNestedModules(modulePath, state, pkg)
+  ])
+  return [...publishedFiles, ...depsPaths]
+}
+
+const EXCLUDED_MODULES = ['aws-sdk']
+
+// We use all the files published by the Node.js except some that are not needed
+const getPublishedFiles = async function(modulePath) {
+  const ignore = getIgnoredFiles(modulePath)
+  const publishedFiles = await pGlob(`${modulePath}/**`, {
+    ignore,
+    nodir: true,
+    absolute: true,
+    dot: true
+  })
+  return publishedFiles
+}
+
+const getIgnoredFiles = function(modulePath) {
+  return IGNORED_FILES.map(ignoreFile => `${modulePath}/${ignoreFile}`)
+}
+
+// To make the zip archive smaller, we remove those.
+const IGNORED_FILES = [
+  'node_modules/**',
+  '.npmignore',
+  'package-lock.json',
+  'yarn.lock',
+  '*.log',
+  '*.lock',
+  '*~',
+  '*.map',
+  '*.ts',
+  '*.patch'
+]
+
+// Apply the Node.js module logic recursively on its own dependencies, using
+// the `package.json` `dependencies`, `peerDependencies` and
+// `optionalDependencies` keys
+const getNestedModules = async function(modulePath, state, pkg) {
+  const dependencies = getNestedDependencies(pkg)
+
+  const depsPaths = await Promise.all(
+    dependencies.map(dependency => getModuleDependencies(dependency, modulePath, state, pkg))
+  )
+  return [].concat(...depsPaths)
+}
+
+const getNestedDependencies = function({ dependencies = {}, peerDependencies = {}, optionalDependencies = {} }) {
+  return [
+    ...Object.keys(dependencies),
+    ...Object.keys(peerDependencies).filter(shouldIncludePeerDependency),
+    ...Object.keys(optionalDependencies)
+  ]
+}
+
+// Workaround for https://github.com/netlify/zip-it-and-ship-it/issues/73
+// TODO: remove this after adding proper modules exclusion as outlined in
+// https://github.com/netlify/zip-it-and-ship-it/issues/68
+const shouldIncludePeerDependency = function(name) {
+  return !EXCLUDED_PEER_DEPENDENCIES.includes(name)
+}
+
+const EXCLUDED_PEER_DEPENDENCIES = ['prisma2']
+
+// Modules can be required conditionally (inside an `if` or `try`/`catch` block).
+// When a `require()` statement is found but the module is not found, it is
+// possible that that block either always evaluates to:
+//  - `false`: in which case, we should not bundle the dependency
+//  - `true`: in which case, we should report the dependency as missing
+// Those conditional modules might be:
+//  - present in the `package.json` `dependencies`
+//  - present in the `package.json` `optionalDependencies`
+//  - present in the `package.json` `peerDependencies`
+//  - not present in the `package.json`, if the module author wants its users
+//    to explicitly install it as an optional dependency.
+// The current implementation:
+//  - when parsing `require()` statements inside function files, always consider
+//    conditional modules to be included, i.e. report them if not found.
+//    This is because our current parsing logic does not know whether a
+//    `require()` is conditional or not.
+//  - when parsing module dependencies, ignore `require()` statements if not
+//    present in the `package.json` `*dependencies`. I.e. user must manually
+//    install them if the module is used.
+// `optionalDependencies`:
+//  - are not reported when missing
+//  - are included in module dependencies
+const handleModuleNotFound = function({ error, moduleName, packageJson }) {
+  if (error.code === 'MODULE_NOT_FOUND' && isOptionalModule(moduleName, packageJson)) {
+    return []
+  }
+
+  throw error
+}
+
+const isOptionalModule = function(
+  moduleName,
+  { optionalDependencies = {}, peerDependenciesMeta = {}, peerDependencies = {} }
+) {
+  return (
+    optionalDependencies[moduleName] !== undefined ||
+    (peerDependenciesMeta[moduleName] &&
+      peerDependenciesMeta[moduleName].optional &&
+      peerDependencies[moduleName] !== undefined)
+  )
+}
+
+module.exports = { getDependencies }