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

msgspec/_core.c

@@ -1886,14 +1886,14 @@ Meta_new(PyTypeObject *type, PyObject *args, PyObject *kwargs) {
     Meta *out = (Meta *)Meta_Type.tp_alloc(&Meta_Type, 0);
     if (out == NULL) return NULL;
 
+// set fields on Meta and increase their refcount
 #define SET_FIELD(x) do { Py_XINCREF(x); 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(min_length);
     SET_FIELD(max_length);
     SET_FIELD(tz);
@@ -1903,6 +1903,11 @@ Meta_new(PyTypeObject *type, PyObject *args, PyObject *kwargs) {
     SET_FIELD(extra_json_schema);
     SET_FIELD(extra);
 #undef SET_FIELD
+
+    // set fields on Meta without increasing their refcount
+    // regex was created by a PyObject_CallOneArg call, so refcount started out as 1; no need to increase
+    out->regex = regex;
+
     return (PyObject *)out;
 }
 

tests/test_struct_meta.py

@@ -1,5 +1,9 @@
 """Tests for the exposed StructMeta metaclass."""
 
+import gc
+import re
+import secrets
+
 import pytest
 import msgspec
 from msgspec import Struct, StructMeta
@@ -362,3 +366,26 @@ class Container(metaclass=EncoderMeta):
     assert decoded.count == 1
     assert decoded.item.id == 123
     assert decoded.item.name == "test"
+
+
+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 should be no re.Pattern any more with our pattern anymore. 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
+    )