Strip common path prefixes from linkname as well as name when extract…

…ing a tarfile (#688)

Stripping common path prefixes from the member name breaks hardlink name resolution unless you perform the same operation on the member linknames

LP: #1606700

snapcraft/internal/sources.py

@@ -296,17 +296,25 @@ def filter_members(tar):
                 for m in members:
                     if m.name == common:
                         continue
-                    if m.name.startswith(common + '/'):
-                        m.name = m.name[len(common + '/'):]
-                    # strip leading '/', './' or '../' as many times as needed
-                    m.name = re.sub(r'^(\.{0,2}/)*', r'', m.name)
+                    self._strip_prefix(common, m)
                     # We mask all files to be writable to be able to easily
                     # extract on top.
                     m.mode = m.mode | 0o200
                     yield m
 
             tar.extractall(members=filter_members(tar), path=dst)
 
+    def _strip_prefix(self, common, member):
+        if member.name.startswith(common + '/'):
+            member.name = member.name[len(common + '/'):]
+        # strip leading '/', './' or '../' as many times as needed
+        member.name = re.sub(r'^(\.{0,2}/)*', r'', member.name)
+        # do the same for linkname if this is a hardlink
+        if member.islnk() and not member.issym():
+            if member.linkname.startswith(common + '/'):
+                member.linkname = member.linkname[len(common + '/'):]
+            member.linkname = re.sub(r'^(\.{0,2}/)*', r'', member.linkname)
+
 
 class Zip(FileBase):
 

snapcraft/tests/test_plugin_tar_content.py

@@ -82,3 +82,114 @@ class Options:
 
         self.assertTrue(
             os.path.exists(os.path.join(self.build_prefix)))
+
+    def test_common_prefix_stripping(self):
+        class Options:
+            source = os.path.join('src', 'test.tar')
+            destination = 'destdir1'
+
+        # create tar file for testing
+        os.makedirs('src/test_prefix')
+        file_to_tar = os.path.join('src', 'test_prefix', 'test.txt')
+        open(file_to_tar, 'w').close()
+        tar = tarfile.open(os.path.join('src', 'test.tar'), 'w')
+        tar.add(file_to_tar)
+        tar.close()
+
+        t = TarContentPlugin('tar_content', Options(),
+                             self.project_options)
+        os.mkdir(t.sourcedir)
+        t.pull()
+        t.build()
+
+        # the 'test_prefix' part of the path should have been removed
+        self.assertTrue(
+            os.path.exists(os.path.join(self.install_prefix, 'destdir1')))
+        self.assertTrue(
+            os.path.exists(
+                os.path.join(self.install_prefix, 'destdir1/test.txt')))
+
+    def test_common_prefix_stripping_symlink(self):
+        class Options:
+            source = os.path.join('src', 'test.tar')
+            destination = 'destdir1'
+
+        # create tar file for testing
+        os.makedirs('src/test_prefix')
+        file_to_tar = os.path.join('src', 'test_prefix', 'test.txt')
+        open(file_to_tar, 'w').close()
+
+        file_to_link = os.path.join('src', 'test_prefix', 'link.txt')
+        os.symlink("./test.txt", file_to_link)
+        self.assertTrue(os.path.islink(file_to_link))
+
+        def check_for_symlink(tarinfo):
+            self.assertTrue(tarinfo.issym())
+            self.assertEqual(file_to_link, tarinfo.name)
+            self.assertEqual(file_to_tar, os.path.normpath(
+                os.path.join(
+                    os.path.dirname(file_to_tar), tarinfo.linkname)))
+            return tarinfo
+
+        tar = tarfile.open(os.path.join('src', 'test.tar'), 'w')
+        tar.add(file_to_tar)
+        tar.add(file_to_link, filter=check_for_symlink)
+        tar.close()
+
+        t = TarContentPlugin('tar_content', Options(),
+                             self.project_options)
+        os.mkdir(t.sourcedir)
+        t.pull()
+        t.build()
+
+        # the 'test_prefix' part of the path should have been removed
+        self.assertTrue(
+            os.path.exists(os.path.join(self.install_prefix, 'destdir1')))
+        self.assertTrue(
+            os.path.exists(
+                os.path.join(self.install_prefix, 'destdir1/test.txt')))
+        self.assertTrue(
+            os.path.exists(
+                os.path.join(self.install_prefix, 'destdir1/link.txt')))
+
+    def test_common_prefix_stripping_hardlink(self):
+        class Options:
+            source = os.path.join('src', 'test.tar')
+            destination = 'destdir1'
+
+        # create tar file for testing
+        os.makedirs('src/test_prefix')
+        file_to_tar = os.path.join('src', 'test_prefix', 'test.txt')
+        open(file_to_tar, 'w').close()
+
+        file_to_link = os.path.join('src', 'test_prefix', 'link.txt')
+        os.link(file_to_tar, file_to_link)
+        self.assertTrue(os.path.exists(file_to_link))
+
+        def check_for_hardlink(tarinfo):
+            self.assertTrue(tarinfo.islnk())
+            self.assertFalse(tarinfo.issym())
+            self.assertEqual(file_to_link, tarinfo.name)
+            self.assertEqual(file_to_tar, tarinfo.linkname)
+            return tarinfo
+
+        tar = tarfile.open(os.path.join('src', 'test.tar'), 'w')
+        tar.add(file_to_tar)
+        tar.add(file_to_link, filter=check_for_hardlink)
+        tar.close()
+
+        t = TarContentPlugin('tar_content', Options(),
+                             self.project_options)
+        os.mkdir(t.sourcedir)
+        t.pull()
+        t.build()
+
+        # the 'test_prefix' part of the path should have been removed
+        self.assertTrue(
+            os.path.exists(os.path.join(self.install_prefix, 'destdir1')))
+        self.assertTrue(
+            os.path.exists(
+                os.path.join(self.install_prefix, 'destdir1/test.txt')))
+        self.assertTrue(
+            os.path.exists(
+                os.path.join(self.install_prefix, 'destdir1/link.txt')))