Skip to content

gh-153161: Cache the match object in the compiled re pattern #153162

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
serhiy-storchaka wants to merge 1 commit into
base: main
Choose a base branch
from
+45 −7
Open

gh-153161: Cache the match object in the compiled re pattern #153162

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
3 changes: 3 additions & 0 deletions Misc/NEWS.d/next/Library/2026-07-06-12-00-00.gh-issue-153161.MtchC5.rst
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,3 @@
Speed up successful matching in :mod:`re` by 15-18 ns: a dead match object
is kept in the compiled pattern and reused for the next match instead of
allocating a new one.
47 changes: 40 additions & 7 deletions Modules/_sre/sre.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -44,6 +44,7 @@ static const char copyright[] =
#include "pycore_long.h" // _PyLong_GetZero()
#include "pycore_list.h" // _PyList_AppendTakeRef()
#include "pycore_moduleobject.h" // _PyModule_GetState()
#include "pycore_object.h" // _Py_NewReference()
#include "pycore_tuple.h" // _PyTuple_FromPairSteal
#include "pycore_unicodeobject.h" // _PyUnicode_Copy
#include "pycore_unicodectype.h" // _PyUnicode_IsXidStart()
Expand Down Expand Up @@ -875,6 +876,12 @@ static int
pattern_clear(PyObject *op)
{
PatternObject *self = _PatternObject_CAST(op);
PyObject *cached = _Py_atomic_exchange_ptr(&self->cached_match, NULL);
if (cached != NULL) {
PyTypeObject *tp = Py_TYPE(cached);
PyObject_GC_Del(cached);
Py_DECREF(tp);
}
Py_CLEAR(self->groupindex);
Py_CLEAR(self->indexgroup);
Py_CLEAR(self->pattern);
Expand Down Expand Up @@ -1788,6 +1795,7 @@ _sre_compile_impl(PyObject *module, PyObject *pattern, int flags,
if (!self)
return NULL;
self->weakreflist = NULL;
self->cached_match = NULL;
self->pattern = NULL;
self->groupindex = NULL;
self->indexgroup = NULL;
Expand Down Expand Up @@ -2505,7 +2513,24 @@ match_dealloc(PyObject *self)
{
PyTypeObject *tp = Py_TYPE(self);
PyObject_GC_UnTrack(self);
(void)match_clear(self);
MatchObject *match = _MatchObject_CAST(self);
PatternObject *pattern = match->pattern;
Py_CLEAR(match->string);
Py_CLEAR(match->regs);
if (pattern != NULL) {
match->pattern = NULL;
/* Cache the dead object (it keeps only its type reference) in the
pattern for the next match. The pattern reference is given up
either way; if it was the last one, pattern_clear() frees the
object just cached. */
PyObject *expected = NULL;
if (_Py_atomic_compare_exchange_ptr(&pattern->cached_match,
&expected, self)) {
Py_DECREF(pattern);
return;
}
Py_DECREF(pattern);
}
tp->tp_free(self);
Py_DECREF(tp);
}
Expand Down Expand Up @@ -2950,12 +2975,20 @@ pattern_new_match(_sremodulestate* module_state,
if (status > 0) {

/* create match object (with room for extra group marks) */
/* coverity[ampersand_in_size] */
match = PyObject_GC_NewVar(MatchObject,
module_state->Match_Type,
2*(pattern->groups+1));
if (!match)
return NULL;
match = (MatchObject *)_Py_atomic_exchange_ptr(&pattern->cached_match,
NULL);
if (match != NULL) {
assert(Py_SIZE(match) == 2*(pattern->groups+1));
_Py_NewReference((PyObject *)match);
}
else {
/* coverity[ampersand_in_size] */
match = PyObject_GC_NewVar(MatchObject,
module_state->Match_Type,
2*(pattern->groups+1));
if (!match)
return NULL;
}

Py_INCREF(pattern);
match->pattern = pattern;
Expand Down
2 changes: 2 additions & 0 deletions Modules/_sre/sre.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,6 +33,8 @@ typedef struct {
PyObject* pattern; /* pattern source (or None) */
int flags; /* flags used when compiling pattern source */
PyObject *weakreflist; /* List of weak references */
PyObject *cached_match; /* dead match object kept for reuse; owns only
a reference to its type */
int isbytes; /* pattern type (1 - bytes, 0 - string, -1 - None) */
#ifdef Py_DEBUG
/* for simulation of user interruption */
Expand Down
Loading