From 998bccac244d12c98a443af56332d263a7ba90b0 Mon Sep 17 00:00:00 2001
From: lipengyu <lipengyu@kylinos.cn>
Date: Sat, 29 Aug 2026 09:33:41 +0800
Subject: [PATCH] gh-156539: Validate duplicate ZIP members individually in
 testzip

Pass each ZipInfo directly to open() so testzip() does not resolve duplicate filenames to the last matching member.

Co-authored-by: lipengyu <lipengyu@kylinos.cn>
---
 Lib/test/test_zipfile/test_core.py            | 24 +++++++++++++++++++
 Lib/zipfile/__init__.py                       |  5 ++--
 ...-08-29-09-28-53.gh-issue-156539.mH4AGb.rst |  3 +++
 3 files changed, 29 insertions(+), 3 deletions(-)
 create mode 100644 Misc/NEWS.d/next/Library/2026-08-29-09-28-53.gh-issue-156539.mH4AGb.rst

diff --git a/Lib/test/test_zipfile/test_core.py b/Lib/test/test_zipfile/test_core.py
index fdf2cd26f8c7c6..58cf972089f263 100644
--- a/Lib/test/test_zipfile/test_core.py
+++ b/Lib/test/test_zipfile/test_core.py
@@ -4335,6 +4335,30 @@ def test_closed_zip_raises_ValueError(self):
             f.write('zipfile test data')
         self.assertRaises(ValueError, zipf.write, TESTFN)
 
+    def test_testzip_with_duplicate_names(self):
+        data = io.BytesIO()
+        with zipfile.ZipFile(data, mode="w") as zipf:
+            zipf.writestr("duplicate", b"corrupt")
+            with warnings.catch_warnings():
+                warnings.simplefilter("ignore")
+                zipf.writestr("duplicate", b"valid")
+
+        zipdata = bytearray(data.getvalue())
+        with zipfile.ZipFile(io.BytesIO(zipdata)) as zipf:
+            zinfo = zipf.infolist()[0]
+            file_header = struct.unpack_from(
+                zipfile.structFileHeader, zipdata, zinfo.header_offset
+            )
+            name_length, extra_length = file_header[-2:]
+        data_offset = (zinfo.header_offset + zipfile.sizeFileHeader
+                       + name_length + extra_length)
+        zipdata[data_offset] ^= 1
+
+        with zipfile.ZipFile(io.BytesIO(zipdata)) as zipf:
+            self.assertRaises(zipfile.BadZipFile, zipf.read,
+                              zipf.infolist()[0])
+            self.assertEqual("duplicate", zipf.testzip())
+
     def test_bad_constructor_mode(self):
         """Check that bad modes passed to ZipFile constructor are caught."""
         self.assertRaises(ValueError, zipfile.ZipFile, TESTFN, "q")
diff --git a/Lib/zipfile/__init__.py b/Lib/zipfile/__init__.py
index 0accf324c90e3f..d0c4a92c59a0a5 100644
--- a/Lib/zipfile/__init__.py
+++ b/Lib/zipfile/__init__.py
@@ -2145,7 +2145,7 @@ def testzip(self):
             try:
                 # Read by chunks, to avoid an OverflowError or a
                 # MemoryError with very large embedded files.
-                with self.open(zinfo.filename, "r") as f:
+                with self.open(zinfo, "r") as f:
                     while f.read(chunk_size):     # Check CRC-32
                         pass
             except BadZipFile:
@@ -2401,8 +2401,7 @@ def remove(self, zinfo_or_arcname):
             except KeyError:
                 pass
 
-            # Avoid missing entry if there is another entry having the same name,
-            # to prevent an error on `testzip()`.
+            # Keep the last remaining entry with this name in NameToInfo.
             # Reverse the order as NameToInfo normally stores the last added one.
             for zi in reversed(self.filelist):
                 if zi.filename == zinfo.filename:
diff --git a/Misc/NEWS.d/next/Library/2026-08-29-09-28-53.gh-issue-156539.mH4AGb.rst b/Misc/NEWS.d/next/Library/2026-08-29-09-28-53.gh-issue-156539.mH4AGb.rst
new file mode 100644
index 00000000000000..08963764677bf7
--- /dev/null
+++ b/Misc/NEWS.d/next/Library/2026-08-29-09-28-53.gh-issue-156539.mH4AGb.rst
@@ -0,0 +1,3 @@
+Fix :meth:`zipfile.ZipFile.testzip` to validate every member in a ZIP
+archive with duplicate names. It could previously miss corruption in an
+earlier member.