From a6c942cbbad909fe81aa76afeb58d5c2f077d107 Mon Sep 17 00:00:00 2001
From: Stan Ulbrych <stan@ulbrych.org>
Date: Tue, 17 Mar 2026 11:50:30 +0000
Subject: [PATCH 1/4] Limit the growth of `encodings.search_function` cache

---
 Lib/encodings/__init__.py                             |  5 +++--
 Lib/test/test_codecs.py                               | 11 +++++++++++
 .../2026-03-17-11-46-20.gh-issue-146054.udYcqn.rst    |  1 +
 3 files changed, 15 insertions(+), 2 deletions(-)
 create mode 100644 Misc/NEWS.d/next/Library/2026-03-17-11-46-20.gh-issue-146054.udYcqn.rst

diff --git a/Lib/encodings/__init__.py b/Lib/encodings/__init__.py
index e205ec326376d8..40f0737f7d4ce0 100644
--- a/Lib/encodings/__init__.py
+++ b/Lib/encodings/__init__.py
@@ -34,6 +34,7 @@
 from . import aliases
 
 _cache = {}
+_MAXCACHE = 500
 _unknown = '--unknown--'
 _import_tail = ['*']
 _aliases = aliases.aliases
@@ -110,8 +111,8 @@ def search_function(encoding):
         mod = None
 
     if mod is None:
-        # Cache misses
-        _cache[encoding] = None
+        if len(_cache) < _MAXCACHE:
+            _cache[encoding] = None
         return None
 
     # Now ask the module for the registry entry
diff --git a/Lib/test/test_codecs.py b/Lib/test/test_codecs.py
index c31faec9ee5214..79c8a7ef886482 100644
--- a/Lib/test/test_codecs.py
+++ b/Lib/test/test_codecs.py
@@ -3908,5 +3908,16 @@ def test_encodings_normalize_encoding(self):
             self.assertEqual(normalize('utf\xE9\u20AC\U0010ffff-8'), 'utf_8')
 
 
+class CodecCacheTest(unittest.TestCase):
+    def test_cache_bounded(self):
+        for i in range(encodings._MAXCACHE + 1000):
+            try:
+                b'x'.decode(f'nonexist_{i}')
+            except LookupError:
+                pass
+
+        self.assertLessEqual(len(encodings._cache), encodings._MAXCACHE)
+
+
 if __name__ == "__main__":
     unittest.main()
diff --git a/Misc/NEWS.d/next/Library/2026-03-17-11-46-20.gh-issue-146054.udYcqn.rst b/Misc/NEWS.d/next/Library/2026-03-17-11-46-20.gh-issue-146054.udYcqn.rst
new file mode 100644
index 00000000000000..c549b197a10b5c
--- /dev/null
+++ b/Misc/NEWS.d/next/Library/2026-03-17-11-46-20.gh-issue-146054.udYcqn.rst
@@ -0,0 +1 @@
+Limit the size of :func:`encodings.search_function` cache.

From ed48ab8ca680045868fe7619ca9b56792e5691e8 Mon Sep 17 00:00:00 2001
From: Stan Ulbrych <stan@ulbrych.org>
Date: Tue, 17 Mar 2026 11:52:35 +0000
Subject: [PATCH 2/4] Add OSS Fuzz issue

---
 .../next/Library/2026-03-17-11-46-20.gh-issue-146054.udYcqn.rst  | 1 +
 1 file changed, 1 insertion(+)

diff --git a/Misc/NEWS.d/next/Library/2026-03-17-11-46-20.gh-issue-146054.udYcqn.rst b/Misc/NEWS.d/next/Library/2026-03-17-11-46-20.gh-issue-146054.udYcqn.rst
index c549b197a10b5c..8692c7f171d0fb 100644
--- a/Misc/NEWS.d/next/Library/2026-03-17-11-46-20.gh-issue-146054.udYcqn.rst
+++ b/Misc/NEWS.d/next/Library/2026-03-17-11-46-20.gh-issue-146054.udYcqn.rst
@@ -1 +1,2 @@
 Limit the size of :func:`encodings.search_function` cache.
+Found by OSS Fuzz in :oss-fuzz:`493449985`.

From d57d3979f1bb82ce820689fade010247957ddcf3 Mon Sep 17 00:00:00 2001
From: Stan Ulbrych <stan@ulbrych.org>
Date: Tue, 17 Mar 2026 12:21:45 +0000
Subject: [PATCH 3/4] LRU cache

---
 Lib/encodings/__init__.py | 7 +++++--
 1 file changed, 5 insertions(+), 2 deletions(-)

diff --git a/Lib/encodings/__init__.py b/Lib/encodings/__init__.py
index 40f0737f7d4ce0..fbf5a3a7f2ce18 100644
--- a/Lib/encodings/__init__.py
+++ b/Lib/encodings/__init__.py
@@ -111,8 +111,9 @@ def search_function(encoding):
         mod = None
 
     if mod is None:
-        if len(_cache) < _MAXCACHE:
-            _cache[encoding] = None
+        if len(_cache) >= _MAXCACHE:
+            _cache.clear()
+        _cache[encoding] = None
         return None
 
     # Now ask the module for the registry entry
@@ -133,6 +134,8 @@ def search_function(encoding):
         entry = codecs.CodecInfo(*entry)
 
     # Cache the codec registry entry
+    if len(_cache) >= _MAXCACHE:
+        _cache.clear()
     _cache[encoding] = entry
 
     # Register its aliases (without overwriting previously registered

From ef2851732705d9ace2b80b9d6aa2d32d174b97a0 Mon Sep 17 00:00:00 2001
From: Stan Ulbrych <stan@ulbrych.org>
Date: Tue, 17 Mar 2026 12:23:54 +0000
Subject: [PATCH 4/4] Accidentally removed comment.

---
 Lib/encodings/__init__.py | 1 +
 1 file changed, 1 insertion(+)

diff --git a/Lib/encodings/__init__.py b/Lib/encodings/__init__.py
index fbf5a3a7f2ce18..169c48324f227b 100644
--- a/Lib/encodings/__init__.py
+++ b/Lib/encodings/__init__.py
@@ -111,6 +111,7 @@ def search_function(encoding):
         mod = None
 
     if mod is None:
+        # Cache misses
         if len(_cache) >= _MAXCACHE:
             _cache.clear()
         _cache[encoding] = None