feat: adding support for filename encodings

Author: martinheidegger

.github/workflows/ci.yml

@@ -38,4 +38,4 @@ jobs:
       - name: Lint
         run: yarn lint
       - name: Test
-        run: yarn coverage
+        run: yarn ci:coverage

cli.js

@@ -7,12 +7,18 @@ var extract = require('./')
 var args = process.argv.slice(2)
 var source = args[0]
 var dest = args[1] || process.cwd()
+var encodingReg = /^--encoding=*/
+var encodingIndex = args.findIndex(entry => encodingReg.test(entry))
+var encoding
+if (encodingIndex !== -1) {
+  encoding = args.splice(encodingIndex, 1)[0].replace(encodingReg, '')
+}
 if (!source) {
-  console.error('Usage: extract-zip foo.zip <targetDirectory>')
+  console.error('Usage: extract-zip foo.zip <targetDirectory> [--encoding=<encoding>]')
   process.exit(1)
 }
 
-extract(source, { dir: dest })
+extract(source, { dir: dest, encoding })
   .catch(function (err) {
     console.error('error!', err)
     process.exit(1)

index.d.ts

@@ -6,6 +6,7 @@ import { Entry, ZipFile } from 'yauzl';
 
 declare namespace extract {
     interface Options {
+        encoding?: string;
         dir: string;
         defaultDirMode?: number;
         defaultFileMode?: number;

index.js

@@ -3,7 +3,7 @@ const debug = require('debug')('extract-zip')
 const { createWriteStream, promises: fs } = require('fs')
 const getStream = require('get-stream')
 const path = require('path')
-const { promisify } = require('util')
+const { promisify, TextDecoder } = require('util')
 const stream = require('stream')
 const yauzl = require('yauzl')
 
@@ -14,12 +14,13 @@ class Extractor {
   constructor (zipPath, opts) {
     this.zipPath = zipPath
     this.opts = opts
+    this.decoder = new TextDecoder(opts.encoding)
   }
 
   async extract () {
     debug('opening', this.zipPath, 'with opts', this.opts)
 
-    this.zipfile = await openZip(this.zipPath, { lazyEntries: true })
+    this.zipfile = await openZip(this.zipPath, { lazyEntries: true, decodeStrings: false })
     this.canceled = false
 
     return new Promise((resolve, reject) => {
@@ -37,6 +38,9 @@ class Extractor {
       })
 
       this.zipfile.on('entry', async entry => {
+        entry.fileName = this.decoder.decode(entry.fileName)
+        entry.fileNameLength = entry.fileName.length
+
         /* istanbul ignore if */
         if (this.canceled) {
           debug('skipping entry', entry.fileName, { cancelled: this.canceled })

index.test-d.ts

@@ -10,6 +10,7 @@ let options: extract.Options = {
 };
 options = {
   dir,
+  encoding: 'shift-jis',
   defaultDirMode: 0o700,
   defaultFileMode,
   onEntry: (entry: Entry, zipfile: ZipFile): void => {

package.json

@@ -14,6 +14,7 @@
     "lint:js": "eslint .",
     "lint:ts": "eslint --config .eslintrc.typescript.js --ext .ts .",
     "test": "yarn lint && ava",
+    "ci:coverage": "full-icu node_modules/nyc/bin/nyc.js ava",
     "tsd": "tsd"
   },
   "files": [
@@ -29,7 +30,7 @@
     "extract"
   ],
   "engines": {
-    "node": ">= 10.17.0"
+    "node": ">= 10.18.0"
   },
   "dependencies": {
     "debug": "^4.1.1",
@@ -40,6 +41,7 @@
     "@types/yauzl": "^2.9.1"
   },
   "devDependencies": {
+    "@leichtgewicht/full-icu": "^1.3.3",
     "@typescript-eslint/eslint-plugin": "^3.2.0",
     "@typescript-eslint/parser": "^3.2.0",
     "ava": "^3.5.1",

readme.md

@@ -42,16 +42,19 @@ async function main () {
 ### Options
 
 - `dir` (required) - the path to the directory where the extracted files are written
+- `encoding` - string - [encoding][] to be used for file names, defaults to `utf-8`
 - `defaultDirMode` - integer - Directory Mode (permissions), defaults to `0o755`
 - `defaultFileMode` - integer - File Mode (permissions), defaults to `0o644`
 - `onEntry` - function - if present, will be called with `(entry, zipfile)`, entry is every entry from the zip file forwarded from the `entry` event from yauzl. `zipfile` is the `yauzl` instance
 
 Default modes are only used if no permissions are set in the zip file.
 
+[encoding]: https://developer.mozilla.org/en-US/docs/Web/API/TextDecoder/encoding
+
 ## CLI Usage
 
 ```
-extract-zip foo.zip <targetDirectory>
+extract-zip foo.zip <targetDirectory> [--encoding=<encoding>]
 ```
 
 If not specified, `targetDirectory` will default to `process.cwd()`.

test/index.js

@@ -11,6 +11,7 @@ const subdirZip = path.join(__dirname, 'file-in-subdir-without-subdir-entry.zip'
 const symlinkDestZip = path.join(__dirname, 'symlink-dest.zip')
 const symlinkZip = path.join(__dirname, 'symlink.zip')
 const brokenZip = path.join(__dirname, 'broken.zip')
+const mojibakeZip = path.join(__dirname, 'mojibake.zip')
 
 const relativeTarget = './cats'
 
@@ -161,3 +162,11 @@ test('extract broken zip', async t => {
     message: 'invalid central directory file header signature: 0x2014b00'
   })
 })
+
+test('extract mojibake', async t => {
+  const dirPath = await mkdtemp(t, 'mojibake-zip')
+  await extract(mojibakeZip, { dir: dirPath, encoding: 'windows-949' })
+  await pathExists(t, path.join(dirPath, '새 텍스트 문서.txt'), 'file created')
+  await pathExists(t, path.join(dirPath, '새 폴더'), 'folder created')
+  await pathExists(t, path.join(dirPath, '새 폴더', '한글문서.txt'), 'subfile created')
+})