From c284fa431aef8b20dc18acb5a390345aa3e554c9 Mon Sep 17 00:00:00 2001
From: Ayan Sinha Mahapatra <asmahapatra@aboutcode.org>
Date: Thu, 8 Jan 2026 13:08:50 +0530
Subject: [PATCH] Support setting the tarfile extract filter

This is needed for python 3.14 support.

Reference: https://github.com/aboutcode-org/commoncode/issues/88
Signed-off-by: Ayan Sinha Mahapatra <asmahapatra@aboutcode.org>
---
 CHANGELOG.rst                 |   6 ++++++
 src/commoncode/archive.py     |   8 ++++++--
 src/commoncode/testcase.py    |   8 ++++----
 tests/data/filetype/types.tar | Bin 112640 -> 109056 bytes
 tests/test_filetype.py        |   5 ++++-
 5 files changed, 20 insertions(+), 7 deletions(-)

diff --git a/CHANGELOG.rst b/CHANGELOG.rst
index 25495220..dc638663 100644
--- a/CHANGELOG.rst
+++ b/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)
 -----------------------------
 
diff --git a/src/commoncode/archive.py b/src/commoncode/archive.py
index 3c0c7583..1d3e42cc 100644
--- a/src/commoncode/archive.py
+++ b/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()
diff --git a/src/commoncode/testcase.py b/src/commoncode/testcase.py
index 525bf666..2a7b37a9 100644
--- a/src/commoncode/testcase.py
+++ b/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)
diff --git a/tests/data/filetype/types.tar b/tests/data/filetype/types.tar
index 24fc04c44da43ac26ebb53a04feca6c8d1a04342..458dc4476efa3cb3b3507135af450adc925a1bfa 100644
GIT binary patch
delta 1169
zcmZqpz}9eujnhCk$kjb0GQf4>!f0^=ATTpAVE|Hw=B8lUz|h2Wa`hb<He+)m1_gu3
zj*Kb%rNt$QMG6WGKr$^eCv~ErBs+r14>AM@K!$(-P{)czX9W1*7U_n#MubdU7>3)n
z57O)=hGyhjx9Jm;-(;p^E#f8&h6V=a<|byQCI*HK28M=4<|YgZJ|Iu9Lp?E3&=H%9
zH#431$H>IMFq!E@3T}s<l3_D7A>W<AV3|DsU6LdNIIOS+{5u(bQ(#0W7{Y@O7*C`}
z0H={|aHKDCG@F~7V@C54X%;ghbMxu57BS`{MJW;!TUc?L>biIafemMbq#<)tQ!Iwd
zu$h<`Gbosnmxj2^blp7N{2_jb!W}0^fYz8BFesRjXAMWNZg7BOP%tn*#Ne`IdbI<i
z6o-+CnISOlgGsk%qM%kiBy#|hCp@PZA+b%7*vM+kv5OlbtAVf?K`Mcn7p#E+9T<p#
zWsSkIAOffqL~M3se#y8<L}9al$Om~aF<C%FVKNgF-bBhI!)9&*4019`2cU&31Qv;G
ISY*Ni0Bb)X#{d8T

delta 846
zcmb7?Pfo%>6vn5dshGF`G?Hi+a{_sf&Yw#d&tOSobOQ!3LCT^r#5HYqZcJEs062jc
zaHB_ny(civYeYb8QfD)1lli^xd*2tmInldQI%%Z7)k^McskLMPfLh|_ZGL5kp#-f^
zASjOkqV}4MOY=CW-t_u@mviRd`t9CT`_hlA_1*7}^ttk4#v}`xvaf{Lq|>xCT)2)z
zU>pkI=Kk`r67$A%%%2Xe36Vz`S<7Gg({SWEP4tQjO<MPLMFLlctST^sQ_{A#vfAx*
z`f=5@={5#mjf`3FVUe1kRj0>)8oh8GKvB#n?QL4B5OR>Wx;S(llqQHGn_B}XdAu|n
zxMiV`am>aRjJB4~s~zunSyhTEoE?<Tk|-Y-9hK1oA^+idd`?Ro60@nb{@K5+xW8I%
ZQUVjb0)gqnu(NLSR<*M(&aamZ`xAmn^XmWr

diff --git a/tests/test_filetype.py b/tests/test_filetype.py
index c5703872..9265314c 100644
--- a/tests/test_filetype.py
+++ b/tests/test_filetype.py
@@ -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: