Merge pull request #95 from aboutcode-org/update-tar-extract

Support setting the tarfile extract filter

Author: AyanSinhaMahapatra

CHANGELOG.rst

@@ -1,6 +1,12 @@
 Release notes
 =============
 
+Version 32.4.2 - (2025-01-08)
+-----------------------------
+
+- Support setting the tar archive filter in python3.14
+  https://github.com/aboutcode-org/commoncode/issues/88
+
 Version 32.4.1 - (2025-01-07)
 -----------------------------
 

src/commoncode/archive.py

@@ -14,6 +14,7 @@
 from os import path
 
 from commoncode.system import on_windows
+from commoncode.system import py314
 
 """
 Mimimal tar and zip file handling, primarily for testing.
@@ -39,7 +40,7 @@ def _extract_tar_raw(test_path, target_dir, to_bytes, *args, **kwargs):
 extract_tar_uni = partial(_extract_tar_raw, to_bytes=False)
 
 
-def extract_tar(location, target_dir, verbatim=False, *args, **kwargs):
+def extract_tar(location, target_dir, verbatim=False, filter=None, *args, **kwargs):
     """
     Extract a tar archive at location in the target_dir directory.
     If `verbatim` is True preserve the permissions.
@@ -58,7 +59,10 @@ def extract_tar(location, target_dir, verbatim=False, *args, **kwargs):
                     if not verbatim:
                         tarinfo.mode = 0o755
                     to_extract.append(tarinfo)
-            tar.extractall(target_dir, members=to_extract)
+            if py314 and filter:
+                tar.extractall(target_dir, members=to_extract, filter=filter)
+            else:
+                tar.extractall(target_dir, members=to_extract)
         finally:
             if tar:
                 tar.close()

src/commoncode/testcase.py

@@ -194,7 +194,7 @@ def remove_vcs(self, test_dir):
             for tf in tilde_files:
                 os.remove(tf)
 
-    def __extract(self, test_path, extract_func=None, verbatim=False):
+    def __extract(self, test_path, extract_func=None, verbatim=False, filter=None):
         """
         Given an archive file identified by test_path relative
         to a test files directory, return a new temp directory where the
@@ -206,7 +206,7 @@ def __extract(self, test_path, extract_func=None, verbatim=False):
         target_path = path.basename(test_path)
         target_dir = self.get_temp_dir(target_path)
         original_archive = self.get_test_loc(test_path)
-        extract_func(original_archive, target_dir, verbatim=verbatim)
+        extract_func(original_archive, target_dir, verbatim=verbatim, filter=filter)
         return target_dir
 
     def extract_test_zip(self, test_path, *args, **kwargs):
@@ -215,8 +215,8 @@ def extract_test_zip(self, test_path, *args, **kwargs):
     def extract_test_zip_raw(self, test_path, *args, **kwargs):
         return self.__extract(test_path, extract_zip_raw)
 
-    def extract_test_tar(self, test_path, verbatim=False):
-        return self.__extract(test_path, extract_tar, verbatim)
+    def extract_test_tar(self, test_path, verbatim=False, filter=None):
+        return self.__extract(test_path, extract_tar, verbatim, filter)
 
     def extract_test_tar_raw(self, test_path, *args, **kwargs):
         return self.__extract(test_path, extract_tar_raw)

tests/data/filetype/types.tar

@@ -34,7 +34,9 @@ def test_get_size_on_directory(self):
         assert filetype.get_size(test_dir) == 12400
 
     def test_get_type(self):
-        test_dir = self.extract_test_tar("filetype/types.tar", verbatim=True)
+        test_dir = self.extract_test_tar(
+            "filetype/types.tar", verbatim=True, filter="fully_trusted"
+        )
         results = []
         for root, dirs, files in os.walk(test_dir):
             for d in dirs:
@@ -61,6 +63,7 @@ def test_get_type(self):
         if on_posix:
             expected += [
                 ("2-SYMTYPE", "l"),
+                ("6-FIFOTYPE", "s"),
             ]
 
         try: