Fix refcount leak to re.Pattern when using Meta(pattern=...) (#899)

Co-authored-by: Ofek Lev <ofekmeister@gmail.com>

Author: provinzkraut

src/msgspec/_core.c

@@ -1907,16 +1907,22 @@ Meta_new(PyTypeObject *type, PyObject *args, PyObject *kwargs) {
     }
 
     Meta *out = (Meta *)Meta_Type.tp_alloc(&Meta_Type, 0);
-    if (out == NULL) return NULL;
+    if (out == NULL) {
+        Py_XDECREF(regex);
+        return NULL;
+    }
 
+/* SET_FIELD handles borrowed values that need an extra INCREF.
+ * SET_FIELD_OWNED passes through references we already own. */
 #define SET_FIELD(x) do { Py_XINCREF(x); out->x = x; } while(0)
+#define SET_FIELD_OWNED(x) do { out->x = x; } while(0)
     SET_FIELD(gt);
     SET_FIELD(ge);
     SET_FIELD(lt);
     SET_FIELD(le);
     SET_FIELD(multiple_of);
     SET_FIELD(pattern);
-    SET_FIELD(regex);
+    SET_FIELD_OWNED(regex);
     SET_FIELD(min_length);
     SET_FIELD(max_length);
     SET_FIELD(tz);
@@ -1926,6 +1932,8 @@ Meta_new(PyTypeObject *type, PyObject *args, PyObject *kwargs) {
     SET_FIELD(extra_json_schema);
     SET_FIELD(extra);
 #undef SET_FIELD
+#undef SET_FIELD_OWNED
+
     return (PyObject *)out;
 }
 

tests/unit/test_struct_meta.py

@@ -1,5 +1,8 @@
 """Tests for the exposed StructMeta metaclass."""
 
+import gc
+import re
+import secrets
 from abc import ABCMeta, _abc_init, abstractmethod
 
 import pytest
@@ -624,3 +627,26 @@ def foo(self) -> int:
 
     c = Concrete(5)
     assert c.foo() == 5
+
+
+def test_struct_meta_pattern_ref_leak():
+    # ensure that we're not keeping around references to re.Pattern longer than necessary
+    # see https://github.com/jcrist/msgspec/pull/899 for details
+
+    # clear cache to get a baseline
+    re.purge()
+
+    # use a random string to create a pattern, to ensure there can never be an overlap
+    # with any cached pattern
+    pattern_string = secrets.token_hex()
+    msgspec.Meta(pattern=pattern_string)
+    # purge cache and gc again
+    re.purge()
+    gc.collect()
+    # there shouldn't be an re.Pattern with our pattern any more. if there is, it's
+    # being kept alive by some reference
+    assert not any(
+        o
+        for o in gc.get_objects()
+        if isinstance(o, re.Pattern) and o.pattern == pattern_string
+    )