From 95f5606fa7f9787b618174505237c0909c4544d0 Mon Sep 17 00:00:00 2001
From: Jono Yang <jyang@nexb.com>
Date: Tue, 14 Jun 2022 16:11:24 -0700
Subject: [PATCH] Add tests for extract_tar

Signed-off-by: Jono Yang <jyang@nexb.com>
---
 src/container_inspector/utils.py |  20 ++++++---
 tests/data/tar/absolute_path.tar | Bin 0 -> 10240 bytes
 tests/data/tar/colon.tar.xz      | Bin 0 -> 188 bytes
 tests/data/tar/tar_relative.tar  | Bin 0 -> 10240 bytes
 tests/test_utils.py              |  75 +++++++++++++++++++++++++++++++
 5 files changed, 90 insertions(+), 5 deletions(-)
 create mode 100644 tests/data/tar/absolute_path.tar
 create mode 100644 tests/data/tar/colon.tar.xz
 create mode 100644 tests/data/tar/tar_relative.tar
 create mode 100644 tests/test_utils.py

diff --git a/src/container_inspector/utils.py b/src/container_inspector/utils.py
index 7c0edbe..740d0ae 100755
--- a/src/container_inspector/utils.py
+++ b/src/container_inspector/utils.py
@@ -97,36 +97,46 @@ def extract_tar(location, target_dir, skip_symlinks=True):
     with tarfile.open(location) as tarball:
         # never extract character device, block and fifo files:
         # we extract dirs, files and links only
-        to_extract = []
+        error_messages = []
         for tarinfo in tarball:
             if TRACE: logger.debug(f'_extract_tar: {tarinfo}')
 
             if tarinfo.isdev() or tarinfo.ischr() or tarinfo.isblk() or tarinfo.isfifo() or tarinfo.sparse:
+                msg = f'_extract_tar: skipping unsupported {tarinfo} file type: block, chr, dev or sparse file'
+                error_messages.append(msg)
                 if TRACE:
-                    logger.debug(f'_extract_tar: skipping unsupported {tarinfo} file type: block, chr, dev or sparse file')
+                    logger.debug(msg)
                 continue
 
             if '..' in tarinfo.name:
-                if TRACE: logger.debug(f'_extract_tar: skipping unsupported {tarinfo} with relative path')
+                msg = f'_extract_tar: skipping unsupported {tarinfo} with relative path'
+                error_messages.append(msg)
+                if TRACE:
+                    logger.debug(msg)
                 continue
 
             if tarinfo.islnk() or tarinfo.issym():
                 try:
                     target = tarball._find_link_target(tarinfo)
                     if not target:
+                        msg = f'_extract_tar: skipping link with missing target: {tarinfo}'
+                        error_messages.append(msg)
                         if TRACE:
-                            logger.debug(f'_extract_tar: skipping link with missing target: {tarinfo}')
+                            logger.debug(msg)
                         continue
 
                 except Exception:
                     import traceback
+                    msg = f'_extract_tar: skipping link with missing target: {tarinfo}: {traceback.format_exc()}'
+                    error_messages.append(msg)
                     if TRACE:
-                        logger.debug(f'_extract_tar: skipping link with missing target: {tarinfo}: {traceback.format_exc()}')
+                        logger.debug(msg)
                     continue
 
             tarinfo.mode = 0o755
             tarinfo.name = tarinfo.name.lstrip('/')
             tarball.extract(member=tarinfo, path=target_dir, set_attrs=False,)
+        return error_messages
 
 
 def extract_tar_with_symlinks(location, target_dir):
diff --git a/tests/data/tar/absolute_path.tar b/tests/data/tar/absolute_path.tar
new file mode 100644
index 0000000000000000000000000000000000000000..0b168da538ab9a466b06f31a9d2ca85fbb218f83
GIT binary patch
literal 10240
zcmeIyOAdlC5P;#XJw;C-v`{!tQHUh?2-N8DEx53R#9fJze>UwDQpz_Bi96qt;a(lv
zJ`s^3^-&rlrm=XBem>>nnh&9r)|oPCo@tRPUHT<yBAsWp?R{{zmr^vny&SIXFuNEe
zk5gTDLH;MxB&fgLyQhorUSolaJpWSv@>759?^9V}N+Ht-XIX5y@%mfYv|YVhu&}TV
zfw2FpU+Dko|4;v~_4QTs|60^M2LS{SKmY**5I_I{1Q0*~0R#|0009ILKmY**5I_I{
L1Q0*~foK9R;hsut

literal 0
HcmV?d00001

diff --git a/tests/data/tar/colon.tar.xz b/tests/data/tar/colon.tar.xz
new file mode 100644
index 0000000000000000000000000000000000000000..e8481c9d0a801606e767c8d1b948866542b6b5cb
GIT binary patch
literal 188
zcmV;t07L)%H+ooF000E$*0e?f03iVu0001VFXf})2mb(hT>vqg$<C7A*Et1+*B>N&
z*e=+F66_D+%jGGkz9*+9E>NZG3X>-d?0b#LMD=mVu?LRMsHOL*=z%ICUB<6S>XQRI
z9LBF!xU*##W2M(YqbgeOu{$jZ{KXfAT;%+fM{(U5m?`$kNV)-8pqy?xo4pACNZJK<
qHPZk90001^xl4oFsT14)0hIxO5C8zyZ{b3*#Ao{g000001X)@t=2GAQ

literal 0
HcmV?d00001

diff --git a/tests/data/tar/tar_relative.tar b/tests/data/tar/tar_relative.tar
new file mode 100644
index 0000000000000000000000000000000000000000..0caff864506dd5c31a8b108fe68f77c2e28faf32
GIT binary patch
literal 10240
zcmeIz@2bKe9LDioN>|{f?a$?iBTwf=WVqj`%g?QdLr|O-%l6_c1I5YI4?a$*(%F-X
z{wVe*9Ckh`xk#~SsZwd%wwUTvp6k@r7FwIybXwQjG)^~Lql-<o=qg@(Y#hYJw4`zH
z@x_q2dmqDidNW|&+Mmt68M>xd;s@EdHLZxX<P=b<OpiMT`StOB1C+g$bFeW4S>!U9
zqVezi*Cyq^Ynq1q=Qsy*{`(v=F~=GOQuuGb{c;%ld#ksP_P6rinTq`9I4|Tshca^@
ofB*srAb<b@2q1s}0tg_000IagfB*srAb<b@2q1s}0-qPS0x}t4$N&HU

literal 0
HcmV?d00001

diff --git a/tests/test_utils.py b/tests/test_utils.py
new file mode 100644
index 0000000..7446744
--- /dev/null
+++ b/tests/test_utils.py
@@ -0,0 +1,75 @@
+#
+# Copyright (c) nexB Inc. and others. All rights reserved.
+# SPDX-License-Identifier: Apache-2.0
+# See http://www.apache.org/licenses/LICENSE-2.0 for the license text.
+# See https://github.com/nexB/container-inspector for support or download.
+# See https://aboutcode.org for more information about nexB OSS projects.
+#
+
+import os
+
+from commoncode import fileutils
+from commoncode import testcase
+
+from container_inspector import utils
+
+
+def check_files(target_dir, expected):
+    """
+    Walk test_dir.
+    Check that all dirs are readable.
+    Check that all files are:
+     * non-special,
+     * readable,
+     * have a posix path that ends with one of the expected tuple paths.
+    """
+    result = []
+
+    test_dir_path = fileutils.as_posixpath(target_dir)
+    for top, _, files in os.walk(target_dir):
+        for f in files:
+            location = os.path.join(top, f)
+            path = fileutils.as_posixpath(location)
+            path = path.replace(test_dir_path, '').strip('/')
+            result.append(path)
+
+    expected_content = sorted(expected)
+    result = sorted(result)
+
+    assert result == expected_content
+
+
+class TestUtils(testcase.FileBasedTesting):
+    test_data_dir = os.path.join(os.path.dirname(__file__), 'data')
+
+    def test_extract_tree_with_colon_in_filenames(self):
+        expected = (
+            'colon/libc6:amd64.list',
+        )
+        test_dir = self.get_test_loc('tar/colon.tar.xz')
+        temp_dir = self.get_temp_dir()
+        errors = utils.extract_tar(location=test_dir, target_dir=temp_dir)
+        check_files(temp_dir, expected)
+        assert not errors
+
+    def test_extract_tar_relative(self):
+        expected = ()
+        test_dir = self.get_test_loc('tar/tar_relative.tar')
+        temp_dir = self.get_temp_dir()
+        errors = utils.extract_tar(location=test_dir, target_dir=temp_dir)
+        check_files(temp_dir, expected)
+        assert errors
+        for error in errors:
+            assert 'skipping unsupported' in error
+            assert 'with relative path' in error
+
+    def test_extract_tar_absolute(self):
+        expected = (
+            'tmp/subdir/a.txt',
+            'tmp/subdir/b.txt',
+        )
+        test_dir = self.get_test_loc('tar/absolute_path.tar')
+        temp_dir = self.get_temp_dir()
+        errors = utils.extract_tar(location=test_dir, target_dir=temp_dir)
+        check_files(temp_dir, expected)
+        assert not errors