Support custom filename encoding option (#72)

extract.js

@@ -43,6 +43,8 @@ var Extract = function (opts) {
   if (!(this instanceof Extract)) return new Extract(opts)
   Writable.call(this, opts)
 
+  opts = opts || {}
+
   this._offset = 0
   this._buffer = bl()
   this._missing = 0
@@ -102,14 +104,14 @@ var Extract = function (opts) {
 
   var ongnulongpath = function () {
     var size = self._header.size
-    this._gnuLongPath = headers.decodeLongPath(b.slice(0, size))
+    this._gnuLongPath = headers.decodeLongPath(b.slice(0, size), opts.filenameEncoding)
     b.consume(size)
     onstreamend()
   }
 
   var ongnulonglinkpath = function () {
     var size = self._header.size
-    this._gnuLongLinkPath = headers.decodeLongPath(b.slice(0, size))
+    this._gnuLongLinkPath = headers.decodeLongPath(b.slice(0, size), opts.filenameEncoding)
     b.consume(size)
     onstreamend()
   }
@@ -118,7 +120,7 @@ var Extract = function (opts) {
     var offset = self._offset
     var header
     try {
-      header = self._header = headers.decode(b.slice(0, 512))
+      header = self._header = headers.decode(b.slice(0, 512), opts.filenameEncoding)
     } catch (err) {
       self.emit('error', err)
     }

headers.js

@@ -149,8 +149,8 @@ var decodeOct = function (val, offset, length) {
   }
 }
 
-var decodeStr = function (val, offset, length) {
-  return val.slice(offset, indexOf(val, 0, offset, offset + length)).toString()
+var decodeStr = function (val, offset, length, encoding) {
+  return val.slice(offset, indexOf(val, 0, offset, offset + length)).toString(encoding)
 }
 
 var addLength = function (str) {
@@ -161,8 +161,8 @@ var addLength = function (str) {
   return (len + digits) + str
 }
 
-exports.decodeLongPath = function (buf) {
-  return decodeStr(buf, 0, buf.length)
+exports.decodeLongPath = function (buf, encoding) {
+  return decodeStr(buf, 0, buf.length, encoding)
 }
 
 exports.encodePax = function (opts) { // TODO: encode more stuff in pax
@@ -240,23 +240,23 @@ exports.encode = function (opts) {
   return buf
 }
 
-exports.decode = function (buf) {
+exports.decode = function (buf, filenameEncoding) {
   var typeflag = buf[156] === 0 ? 0 : buf[156] - ZERO_OFFSET
 
-  var name = decodeStr(buf, 0, 100)
+  var name = decodeStr(buf, 0, 100, filenameEncoding)
   var mode = decodeOct(buf, 100, 8)
   var uid = decodeOct(buf, 108, 8)
   var gid = decodeOct(buf, 116, 8)
   var size = decodeOct(buf, 124, 12)
   var mtime = decodeOct(buf, 136, 12)
   var type = toType(typeflag)
-  var linkname = buf[157] === 0 ? null : decodeStr(buf, 157, 100)
+  var linkname = buf[157] === 0 ? null : decodeStr(buf, 157, 100, filenameEncoding)
   var uname = decodeStr(buf, 265, 32)
   var gname = decodeStr(buf, 297, 32)
   var devmajor = decodeOct(buf, 329, 8)
   var devminor = decodeOct(buf, 337, 8)
 
-  if (buf[345]) name = decodeStr(buf, 345, 155) + '/' + name
+  if (buf[345]) name = decodeStr(buf, 345, 155, filenameEncoding) + '/' + name
 
   // to support old tar versions that use trailing / to indicate dirs
   if (typeflag === 0 && name && name[name.length - 1] === '/') typeflag = 5

test/extract.js

@@ -589,3 +589,40 @@ test('huge', function (t) {
   var reader = fs.createReadStream(fixtures.HUGE)
   reader.pipe(gunzip).pipe(extract)
 })
+
+test('latin-1', function (t) { // can unpack filenames encoded in latin-1
+  t.plan(3)
+
+  // This is the older name for the "latin1" encoding in Node
+  var extract = tar.extract({ filenameEncoding: 'binary' })
+  var noEntries = false
+
+  extract.on('entry', function (header, stream, callback) {
+    t.deepEqual(header, {
+      name: 'En français, s\'il vous plaît?.txt',
+      mode: parseInt('644', 8),
+      uid: 0,
+      gid: 0,
+      size: 14,
+      mtime: new Date(1495941034000),
+      type: 'file',
+      linkname: null,
+      uname: 'root',
+      gname: 'root',
+      devmajor: 0,
+      devminor: 0
+    })
+
+    stream.pipe(concat(function (data) {
+      noEntries = true
+      t.same(data.toString(), 'Hello, world!\n')
+      callback()
+    }))
+  })
+
+  extract.on('finish', function () {
+    t.ok(noEntries)
+  })
+
+  extract.end(fs.readFileSync(fixtures.LATIN1_TAR))
+})

test/fixtures/index.js

@@ -15,3 +15,4 @@ exports.BASE_256_UID_GID = path.join(__dirname, 'base-256-uid-gid.tar')
 exports.LARGE_UID_GID = path.join(__dirname, 'large-uid-gid.tar')
 exports.BASE_256_SIZE = path.join(__dirname, 'base-256-size.tar')
 exports.HUGE = path.join(__dirname, 'huge.tar.gz')
+exports.LATIN1_TAR = path.join(__dirname, 'latin1.tar')