From 6531eb7203af79784cf2fd9d29907485fcd64daa Mon Sep 17 00:00:00 2001
From: Rohit <rohitjavvadi2@gmail.com>
Date: Mon, 15 Jun 2026 20:42:37 +0530
Subject: [PATCH] gh-151307: Bound zipfile reads for forged compressed sizes

---
 Lib/test/test_zipfile/test_core.py            | 66 +++++++++++++++++++
 Lib/zipfile/__init__.py                       | 13 +++-
 ...-06-15-20-35-00.gh-issue-151307.4Tyv93.rst |  2 +
 3 files changed, 80 insertions(+), 1 deletion(-)
 create mode 100644 Misc/NEWS.d/next/Library/2026-06-15-20-35-00.gh-issue-151307.4Tyv93.rst

diff --git a/Lib/test/test_zipfile/test_core.py b/Lib/test/test_zipfile/test_core.py
index ffed328b171fda..631515ba3d76ee 100644
--- a/Lib/test/test_zipfile/test_core.py
+++ b/Lib/test/test_zipfile/test_core.py
@@ -2632,6 +2632,72 @@ def test_full_overlap_same_name(self):
                 zipf.open(zipf.infolist()[0]).close()
             self.assertEqual(cm.filename, __file__)
 
+    def test_forged_compress_size_read_is_bounded(self):
+        # The ZIP file contains two central directory entries with the
+        # same name that point at the same local header.  The first
+        # entry claims an oversized compressed size, but the underlying
+        # read request should still stay bounded.
+        filename = b"a.txt"
+        file_data = b"x"
+        claimed_size = 0xFFFF_FFFE
+        local_header = struct.pack(
+            "<4s2B4HL2L2H",
+            b"PK\x03\x04",
+            20, 0, 0, 0, 0, 0, 0,
+            len(file_data), len(file_data),
+            len(filename), 0,
+        )
+        cd_offset = len(local_header) + len(filename) + len(file_data)
+        cd_entry_forged = struct.pack(
+            "<4s4B4HL2L5H2L",
+            b"PK\x01\x02",
+            20, 0, 20, 0,
+            0, 0, 0, 0, 0,
+            claimed_size, claimed_size,
+            len(filename),
+            0, 0, 0, 0, 0,
+            0,
+        )
+        cd_entry_normal = struct.pack(
+            "<4s4B4HL2L5H2L",
+            b"PK\x01\x02",
+            20, 0, 20, 0,
+            0, 0, 0, 0, 0,
+            len(file_data), len(file_data),
+            len(filename),
+            0, 0, 0, 0, 0,
+            0,
+        )
+        central_directory = (
+            cd_entry_forged + filename +
+            cd_entry_normal + filename
+        )
+        end_record = struct.pack(
+            zipfile.structEndArchive,
+            b"PK\x05\x06",
+            0, 0, 2, 2,
+            len(central_directory), cd_offset, 0,
+        )
+        data = local_header + filename + file_data + central_directory + end_record
+
+        class ObservedBytesIO(io.BytesIO):
+            def __init__(self, data):
+                super().__init__(data)
+                self.read_sizes = []
+
+            def read(self, n=-1):
+                self.read_sizes.append(n)
+                return super().read(n)
+
+        fileobj = ObservedBytesIO(data)
+        with zipfile.ZipFile(fileobj, "r") as zipf:
+            forged = zipf.filelist[0]
+            with self.assertWarnsRegex(UserWarning, "Overlapped entries"):
+                with self.assertRaises(EOFError):
+                    zipf.open(forged).read()
+
+        self.assertLessEqual(max(fileobj.read_sizes), len(data))
+
     @requires_zlib()
     def test_quoted_overlap(self):
         # The ZIP file contains two files. The second local header
diff --git a/Lib/zipfile/__init__.py b/Lib/zipfile/__init__.py
index 005b6f4eb84044..eec74026484abf 100644
--- a/Lib/zipfile/__init__.py
+++ b/Lib/zipfile/__init__.py
@@ -962,6 +962,10 @@ class ZipExtFile(io.BufferedIOBase):
     # Read from compressed files in 4k blocks.
     MIN_READ_SIZE = 4096
 
+    # Max chunk size to read from the underlying file object when its
+    # remaining size is unknown.
+    MAX_READ_SIZE = 1 << 24
+
     # Chunk size to read during seek
     MAX_SEEK_READ = 1 << 24
 
@@ -1000,6 +1004,11 @@ def __init__(self, fileobj, mode, zipinfo, pwd=None,
                 self._orig_file_size = zipinfo.file_size
                 self._orig_start_crc = self._running_crc
                 self._orig_crc = self._expected_crc
+                try:
+                    fileobj.seek(0, os.SEEK_END)
+                    self._compress_end = fileobj.tell()
+                finally:
+                    fileobj.seek(self._orig_compress_start)
                 self._seekable = True
         except AttributeError:
             pass
@@ -1199,7 +1208,9 @@ def _read2(self, n):
             return b''
 
         n = max(n, self.MIN_READ_SIZE)
-        n = min(n, self._compress_left)
+        n = min(n, self._compress_left, self.MAX_READ_SIZE)
+        if self._seekable:
+            n = min(n, max(0, self._compress_end - self._fileobj.tell()))
 
         data = self._fileobj.read(n)
         self._compress_left -= len(data)
diff --git a/Misc/NEWS.d/next/Library/2026-06-15-20-35-00.gh-issue-151307.4Tyv93.rst b/Misc/NEWS.d/next/Library/2026-06-15-20-35-00.gh-issue-151307.4Tyv93.rst
new file mode 100644
index 00000000000000..139ba56f9c873f
--- /dev/null
+++ b/Misc/NEWS.d/next/Library/2026-06-15-20-35-00.gh-issue-151307.4Tyv93.rst
@@ -0,0 +1,2 @@
+Bound :mod:`zipfile` member reads so a forged compressed size cannot make
+``ZipExtFile`` request multi-gigabyte reads from the underlying file object.