Merge pull request #2512 from kamil-certat/http_collector_dirs

FIX: Support for extracting data from archives with dirs

Author: sebix

CHANGELOG.md

@@ -13,6 +13,7 @@
 
 ### Core
 - `intelmq.lib.utils.drop_privileges`: When IntelMQ is called as `root` and dropping the privileges to user `intelmq`, also set the non-primary groups associated with the `intelmq` user. Makes the behaviour of running intelmqctl as `root` closer to the behaviour of `sudo -u intelmq ...` (PR#2507 by Mikk Margus Möll).
+- `intelmq.lib.utils.unzip`: Ignore directories themselves when extracting data to prevent the extraction of empty data for a directory entries (PR#2512 by Kamil Mankowski).
 
 ### Development
 

intelmq/lib/utils.py

@@ -538,7 +538,7 @@ def extract_tar(file):
     def extract(filename):
         return tar.extractfile(filename).read()
 
-    return tuple(file.name for file in tar.getmembers()), tar, extract
+    return tuple(file.name for file in tar.getmembers() if file.isfile()), tar, extract
 
 
 def extract_gzip(file):
@@ -547,7 +547,7 @@ def extract_gzip(file):
 
 def extract_zip(file):
     zfp = zipfile.ZipFile(io.BytesIO(file), "r")
-    return zfp.namelist(), zfp, zfp.read
+    return [member.filename for member in zfp.infolist() if not member.is_dir()], zfp, zfp.read
 
 
 def unzip(file: bytes, extract_files: Union[bool, list], logger=None,

intelmq/tests/assets/subdir.tar.gz

@@ -0,0 +1,3 @@
+SPDX-FileCopyrightText: 2024 CERT.at GmbH
+
+SPDX-License-Identifier: AGPL-3.0-or-later

intelmq/tests/assets/subdir.tar.gz.license

@@ -0,0 +1,3 @@
+SPDX-FileCopyrightText: 2024 CERT.at GmbH
+
+SPDX-License-Identifier: AGPL-3.0-or-later

intelmq/tests/assets/subdir.zip

@@ -143,6 +143,25 @@ def test_zip(self, mocker):
         self.assertMessageEqual(0, output0)
         self.assertMessageEqual(1, output1)
 
+    def test_zip_subdirs(self, mocker):
+        """
+        Test unzipping when the zip has subdirectories
+        """
+        prepare_mocker(mocker)
+        self.run_bot(parameters={'http_url': 'http://localhost/subdir.zip',
+                                 'name': 'Example feed',
+                                 },
+                     iterations=1)
+
+        output0 = OUTPUT[0].copy()
+        output0['feed.url'] = 'http://localhost/subdir.zip'
+        output0['extra.file_name'] = 'subdir/bar'
+        output1 = OUTPUT[1].copy()
+        output1['feed.url'] = 'http://localhost/subdir.zip'
+        output1['extra.file_name'] = 'subdir/foo'
+        self.assertMessageEqual(0, output0)
+        self.assertMessageEqual(1, output1)
+
     @test.skip_exotic()
     def test_pgp(self, mocker):
         """

intelmq/tests/assets/subdir.zip.license

@@ -260,6 +260,14 @@ def test_unzip_tar_gz_return_names(self):
         self.assertEqual(tuple(result), (('bar', b'bar text\n'),
                                          ('foo', b'foo text\n')))
 
+    def test_unzip_tar_gz_with_subdir(self):
+        """ Test the unzip function with a tar gz file containing a subdirectory and return_names. Test that the directories themselves are ignored. """
+        filename = os.path.join(os.path.dirname(__file__), '../assets/subdir.tar.gz')
+        with open(filename, 'rb') as fh:
+            result = utils.unzip(fh.read(), extract_files=True, return_names=True)
+        self.assertEqual(tuple(result), (('subdir/foo', b'foo text\n'),
+                                         ('subdir/bar', b'bar text\n')))
+
     def test_unzip_gz(self):
         """ Test the unzip function with a gz file. """
         filename = os.path.join(os.path.dirname(__file__), '../assets/foobar.gz')
@@ -289,6 +297,14 @@ def test_unzip_zip_return_names(self):
         self.assertEqual(tuple(result), (('bar', b'bar text\n'),
                                          ('foo', b'foo text\n')))
 
+    def test_unzip_zip_with_subdir(self):
+        """ Test the unzip function with a zip containing a subdirectory and returning names. Test that the directories themselves are ignored."""
+        filename = os.path.join(os.path.dirname(__file__), '../assets/subdir.zip')
+        with open(filename, 'rb') as fh:
+            result = utils.unzip(fh.read(), extract_files=True, return_names=True)
+        self.assertEqual(tuple(result), (('subdir/bar', b'bar text\n'),
+                                         ('subdir/foo', b'foo text\n')))
+
     def test_file_name_from_response(self):
         """ test file_name_from_response """
         response = requests.Response()