From b1531cb6957d29ec4a80b7ef7a9b9b04824e045b Mon Sep 17 00:00:00 2001
From: Kevin Lindsay <hello@kevinlindsay.io>
Date: Tue, 29 Dec 2020 05:22:18 -0500
Subject: [PATCH] feat(headers.js): allow unknown formats (#131)

Co-authored-by: Kevin Lindsay <klindsay@surge.solutions>
---
 extract.js                       |   2 +-
 headers.js                       |   6 ++-
 test/extract.js                  |  74 +++++++++++++++++++++++++++++++
 test/fixtures/index.js           |   2 +
 test/fixtures/unknown-format.tar | Bin 0 -> 3072 bytes
 5 files changed, 81 insertions(+), 3 deletions(-)
 create mode 100644 test/fixtures/unknown-format.tar

diff --git a/extract.js b/extract.js
index 6278c38..11b13b7 100644
--- a/extract.js
+++ b/extract.js
@@ -120,7 +120,7 @@ var Extract = function (opts) {
     var offset = self._offset
     var header
     try {
-      header = self._header = headers.decode(b.slice(0, 512), opts.filenameEncoding)
+      header = self._header = headers.decode(b.slice(0, 512), opts.filenameEncoding, opts.allowUnknownFormat)
     } catch (err) {
       self.emit('error', err)
     }
diff --git a/headers.js b/headers.js
index 2787cfd..aba4ca4 100644
--- a/headers.js
+++ b/headers.js
@@ -237,7 +237,7 @@ exports.encode = function (opts) {
   return buf
 }
 
-exports.decode = function (buf, filenameEncoding) {
+exports.decode = function (buf, filenameEncoding, allowUnknownFormat) {
   var typeflag = buf[156] === 0 ? 0 : buf[156] - ZERO_OFFSET
 
   var name = decodeStr(buf, 0, 100, filenameEncoding)
@@ -270,7 +270,9 @@ exports.decode = function (buf, filenameEncoding) {
     // 'gnu'/'oldgnu' format. Similar to ustar, but has support for incremental and
     // multi-volume tarballs.
   } else {
-    throw new Error('Invalid tar header: unknown format.')
+    if (!allowUnknownFormat) {
+      throw new Error('Invalid tar header: unknown format.')
+    }
   }
 
   // to support old tar versions that use trailing / to indicate dirs
diff --git a/test/extract.js b/test/extract.js
index e2f92bd..30ad2e1 100644
--- a/test/extract.js
+++ b/test/extract.js
@@ -679,3 +679,77 @@ test('v7 unsupported', function (t) { // correctly fails to parse v7 tarballs
 
   extract.end(fs.readFileSync(fixtures.V7_TAR))
 })
+
+test('unknown format doesn\'t extract by default', function (t) {
+  t.plan(1)
+
+  var extract = tar.extract()
+
+  extract.on('error', function (err) {
+    t.ok(!!err)
+    extract.destroy()
+  })
+
+  extract.end(fs.readFileSync(fixtures.UNKNOWN_FORMAT))
+})
+
+test('unknown format attempts to extract if allowed', function (t) {
+  t.plan(5)
+
+  var extract = tar.extract({ allowUnknownFormat: true })
+  var noEntries = false
+
+  var onfile1 = function (header, stream, callback) {
+    t.deepEqual(header, {
+      name: 'file-1.txt',
+      mode: parseInt('644', 8),
+      uid: 501,
+      gid: 20,
+      size: 12,
+      mtime: new Date(1387580181000),
+      type: 'file',
+      linkname: null,
+      uname: 'maf',
+      gname: 'staff',
+      devmajor: 0,
+      devminor: 0
+    })
+
+    extract.on('entry', onfile2)
+    stream.pipe(concat(function (data) {
+      t.same(data.toString(), 'i am file-1\n')
+      callback()
+    }))
+  }
+
+  var onfile2 = function (header, stream, callback) {
+    t.deepEqual(header, {
+      name: 'file-2.txt',
+      mode: parseInt('644', 8),
+      uid: 501,
+      gid: 20,
+      size: 12,
+      mtime: new Date(1387580181000),
+      type: 'file',
+      linkname: null,
+      uname: 'maf',
+      gname: 'staff',
+      devmajor: 0,
+      devminor: 0
+    })
+
+    stream.pipe(concat(function (data) {
+      noEntries = true
+      t.same(data.toString(), 'i am file-2\n')
+      callback()
+    }))
+  }
+
+  extract.once('entry', onfile1)
+
+  extract.on('finish', function () {
+    t.ok(noEntries)
+  })
+
+  extract.end(fs.readFileSync(fixtures.UNKNOWN_FORMAT))
+})
diff --git a/test/fixtures/index.js b/test/fixtures/index.js
index 879de1f..569d9ab 100644
--- a/test/fixtures/index.js
+++ b/test/fixtures/index.js
@@ -21,5 +21,7 @@ exports.INCOMPLETE_TAR = path.join(__dirname, 'incomplete.tar')
 exports.GNU_TAR = path.join(__dirname, 'gnu.tar')
 // Created using gnu tar: tar cf gnu-incremental.tar -G --format gnu --owner=myuser:12345 --group=mygroup:67890 test.txt
 exports.GNU_INCREMENTAL_TAR = path.join(__dirname, 'gnu-incremental.tar')
+// Created from multi-file.tar, removing the magic and recomputing the checksum
+exports.UNKNOWN_FORMAT = path.join(__dirname, 'unknown-format.tar')
 // Created using gnu tar: tar cf v7.tar --format v7 test.txt
 exports.V7_TAR = path.join(__dirname, 'v7.tar')
diff --git a/test/fixtures/unknown-format.tar b/test/fixtures/unknown-format.tar
new file mode 100644
index 0000000000000000000000000000000000000000..397db9b273617844f5ce9b7098b2939f05294179
GIT binary patch
literal 3072
zcmeHHF$%*l49uKY=mQBSRZ8CDp@u*+)kB|O$6f*n-3+Eki>yNyr?cd^y`B@#>63<L
z24GR8A^UZ);vqH8P&QU#5i?d(F#@y6Bj+_^bSVt%Azsik@J=EA%BY>x<^Vch?fJY>
t*eRAF7<=aF+}bEmo?bfK|KopX)AC(a-~UN`|L-QlF;~D9a0UKUfiJY{E<gYP

literal 0
HcmV?d00001