support gnu long path. fixes #35

extract.js

@@ -53,6 +53,8 @@ var Extract = function(opts) {
   this._destroyed = false
   this._pax = null
   this._paxGlobal = null
+  this._gnuLongPath = null
+  this._gnuLongLinkPath = null
 
   var self = this
   var b = self._buffer
@@ -96,6 +98,20 @@ var Extract = function(opts) {
     onstreamend()
   }
 
+  var ongnulongpath = function() {
+    var size = self._header.size
+    this._gnuLongPath = headers.decodeLongPath(b.slice(0, size))
+    b.consume(size)
+    onstreamend()
+  }
+
+  var ongnulonglinkpath = function() {
+    var size = self._header.size
+    this._gnuLongLinkPath = headers.decodeLongPath(b.slice(0, size))
+    b.consume(size)
+    onstreamend()
+  }
+
   var onheader = function() {
     var offset = self._offset
     var header
@@ -111,6 +127,16 @@ var Extract = function(opts) {
       oncontinue()
       return
     }
+    if (header.type === 'gnu-long-path') {
+      self._parse(header.size, ongnulongpath)
+      oncontinue()
+      return
+    }
+    if (header.type === 'gnu-long-link-path') {
+      self._parse(header.size, ongnulonglinkpath)
+      oncontinue()
+      return
+    }
     if (header.type === 'pax-global-header') {
       self._parse(header.size, onpaxglobalheader)
       oncontinue()
@@ -122,6 +148,16 @@ var Extract = function(opts) {
       return
     }
 
+    if (self._gnuLongPath) {
+      header.name = self._gnuLongPath
+      self._gnuLongPath = null
+    }
+
+    if (self._gnuLongLinkPath) {
+      header.linkname = self._gnuLongLinkPath
+      self._gnuLongLinkPath = null
+    }
+
     if (self._pax) {
       self._header = header = mixinPax(header, self._pax)
       self._pax = null

headers.js

@@ -34,6 +34,11 @@ var toType = function(flag) {
     return 'pax-header'
     case 55:
     return 'pax-global-header'
+    case 27:
+    return 'gnu-long-link-path'
+    case 28:
+    case 30:
+    return 'gnu-long-path'
   }
 
   return null
@@ -110,6 +115,10 @@ var addLength = function(str) {
   return (len+digits)+str
 }
 
+exports.decodeLongPath = function(buf) {
+  return decodeStr(buf, 0, buf.length)
+}
+
 exports.encodePax = function(opts) { // TODO: encode more stuff in pax
   var result = ''
   if (opts.name) result += addLength(' path='+opts.name+'\n')

test/extract.js

@@ -435,3 +435,21 @@ test('space prefixed', function(t) {
 
   extract.end(fs.readFileSync(fixtures.SPACE_TAR_GZ))
 })
+
+test('gnu long path', function(t) {
+  t.plan(2)
+
+  var extract = tar.extract()
+
+  extract.on('entry', function(header, stream, callback) {
+    t.ok(header.name.length > 100)
+    callback()
+  })
+
+  extract.on('finish', function() {
+    t.ok(true)
+  })
+
+  extract.end(fs.readFileSync(fixtures.GNU_LONG_PATH))
+
+})

test/fixtures/index.js

@@ -9,3 +9,4 @@ exports.UNICODE_TAR = path.join(__dirname, 'unicode.tar')
 exports.NAME_IS_100_TAR = path.join(__dirname, 'name-is-100.tar')
 exports.INVALID_TGZ = path.join(__dirname, 'invalid.tgz')
 exports.SPACE_TAR_GZ = path.join(__dirname, 'space.tar')
+exports.GNU_LONG_PATH = path.join(__dirname, 'gnu-long-path.tar')