[Python-checkins] bpo-23689: re module, fix memory leak when a match is terminated by a signal or memory allocation failure (GH-32283)
serhiy-storchaka
webhook-mailer at python.org
Sun Apr 3 12:16:41 EDT 2022
https://github.com/python/cpython/commit/6e3eee5c11b539e9aab39cff783acf57838c355a
commit: 6e3eee5c11b539e9aab39cff783acf57838c355a
branch: main
author: Ma Lin <animalize at users.noreply.github.com>
committer: serhiy-storchaka <storchaka at gmail.com>
date: 2022-04-03T19:16:20+03:00
summary:
bpo-23689: re module, fix memory leak when a match is terminated by a signal or memory allocation failure (GH-32283)
files:
A Misc/NEWS.d/next/Library/2022-04-03-13-19-08.bpo-23689.TFSc3E.rst
M Lib/re/_compiler.py
M Lib/re/_constants.py
M Lib/test/test_re.py
M Modules/_sre.c
M Modules/clinic/_sre.c.h
M Modules/sre.h
M Modules/sre_constants.h
M Modules/sre_lib.h
diff --git a/Lib/re/_compiler.py b/Lib/re/_compiler.py
index 62da8e55d72ab..bedd4b8f40016 100644
--- a/Lib/re/_compiler.py
+++ b/Lib/re/_compiler.py
@@ -67,14 +67,21 @@
_ignorecase_fixes = {i: tuple(j for j in t if i != j)
for t in _equivalences for i in t}
+class _CompileData:
+ __slots__ = ('code', 'repeat_count')
+ def __init__(self):
+ self.code = []
+ self.repeat_count = 0
+
def _combine_flags(flags, add_flags, del_flags,
TYPE_FLAGS=_parser.TYPE_FLAGS):
if add_flags & TYPE_FLAGS:
flags &= ~TYPE_FLAGS
return (flags | add_flags) & ~del_flags
-def _compile(code, pattern, flags):
+def _compile(data, pattern, flags):
# internal: compile a (sub)pattern
+ code = data.code
emit = code.append
_len = len
LITERAL_CODES = _LITERAL_CODES
@@ -147,7 +154,7 @@ def _compile(code, pattern, flags):
skip = _len(code); emit(0)
emit(av[0])
emit(av[1])
- _compile(code, av[2], flags)
+ _compile(data, av[2], flags)
emit(SUCCESS)
code[skip] = _len(code) - skip
else:
@@ -155,7 +162,11 @@ def _compile(code, pattern, flags):
skip = _len(code); emit(0)
emit(av[0])
emit(av[1])
- _compile(code, av[2], flags)
+ # now op is in (MIN_REPEAT, MAX_REPEAT, POSSESSIVE_REPEAT)
+ if op != POSSESSIVE_REPEAT:
+ emit(data.repeat_count)
+ data.repeat_count += 1
+ _compile(data, av[2], flags)
code[skip] = _len(code) - skip
emit(REPEATING_CODES[op][1])
elif op is SUBPATTERN:
@@ -164,7 +175,7 @@ def _compile(code, pattern, flags):
emit(MARK)
emit((group-1)*2)
# _compile_info(code, p, _combine_flags(flags, add_flags, del_flags))
- _compile(code, p, _combine_flags(flags, add_flags, del_flags))
+ _compile(data, p, _combine_flags(flags, add_flags, del_flags))
if group:
emit(MARK)
emit((group-1)*2+1)
@@ -176,7 +187,7 @@ def _compile(code, pattern, flags):
# pop their stack if they reach it
emit(ATOMIC_GROUP)
skip = _len(code); emit(0)
- _compile(code, av, flags)
+ _compile(data, av, flags)
emit(SUCCESS)
code[skip] = _len(code) - skip
elif op in SUCCESS_CODES:
@@ -191,13 +202,13 @@ def _compile(code, pattern, flags):
if lo != hi:
raise error("look-behind requires fixed-width pattern")
emit(lo) # look behind
- _compile(code, av[1], flags)
+ _compile(data, av[1], flags)
emit(SUCCESS)
code[skip] = _len(code) - skip
elif op is CALL:
emit(op)
skip = _len(code); emit(0)
- _compile(code, av, flags)
+ _compile(data, av, flags)
emit(SUCCESS)
code[skip] = _len(code) - skip
elif op is AT:
@@ -216,7 +227,7 @@ def _compile(code, pattern, flags):
for av in av[1]:
skip = _len(code); emit(0)
# _compile_info(code, av, flags)
- _compile(code, av, flags)
+ _compile(data, av, flags)
emit(JUMP)
tailappend(_len(code)); emit(0)
code[skip] = _len(code) - skip
@@ -244,12 +255,12 @@ def _compile(code, pattern, flags):
emit(op)
emit(av[0]-1)
skipyes = _len(code); emit(0)
- _compile(code, av[1], flags)
+ _compile(data, av[1], flags)
if av[2]:
emit(JUMP)
skipno = _len(code); emit(0)
code[skipyes] = _len(code) - skipyes + 1
- _compile(code, av[2], flags)
+ _compile(data, av[2], flags)
code[skipno] = _len(code) - skipno
else:
code[skipyes] = _len(code) - skipyes + 1
@@ -608,17 +619,17 @@ def isstring(obj):
def _code(p, flags):
flags = p.state.flags | flags
- code = []
+ data = _CompileData()
# compile info block
- _compile_info(code, p, flags)
+ _compile_info(data.code, p, flags)
# compile the pattern
- _compile(code, p.data, flags)
+ _compile(data, p.data, flags)
- code.append(SUCCESS)
+ data.code.append(SUCCESS)
- return code
+ return data
def _hex_code(code):
return '[%s]' % ', '.join('%#0*x' % (_sre.CODESIZE*2+2, x) for x in code)
@@ -719,7 +730,7 @@ def print_2(*args):
else:
print_(FAILURE)
i += 1
- elif op in (REPEAT, REPEAT_ONE, MIN_REPEAT_ONE,
+ elif op in (REPEAT_ONE, MIN_REPEAT_ONE,
POSSESSIVE_REPEAT, POSSESSIVE_REPEAT_ONE):
skip, min, max = code[i: i+3]
if max == MAXREPEAT:
@@ -727,6 +738,13 @@ def print_2(*args):
print_(op, skip, min, max, to=i+skip)
dis_(i+3, i+skip)
i += skip
+ elif op is REPEAT:
+ skip, min, max, repeat_index = code[i: i+4]
+ if max == MAXREPEAT:
+ max = 'MAXREPEAT'
+ print_(op, skip, min, max, repeat_index, to=i+skip)
+ dis_(i+4, i+skip)
+ i += skip
elif op is GROUPREF_EXISTS:
arg, skip = code[i: i+2]
print_(op, arg, skip, to=i+skip)
@@ -781,11 +799,11 @@ def compile(p, flags=0):
else:
pattern = None
- code = _code(p, flags)
+ data = _code(p, flags)
if flags & SRE_FLAG_DEBUG:
print()
- dis(code)
+ dis(data.code)
# map in either direction
groupindex = p.state.groupdict
@@ -794,7 +812,6 @@ def compile(p, flags=0):
indexgroup[i] = k
return _sre.compile(
- pattern, flags | p.state.flags, code,
- p.state.groups-1,
- groupindex, tuple(indexgroup)
- )
+ pattern, flags | p.state.flags, data.code,
+ p.state.groups-1, groupindex, tuple(indexgroup),
+ data.repeat_count)
diff --git a/Lib/re/_constants.py b/Lib/re/_constants.py
index c735edfea1f13..5317fd53e9c5a 100644
--- a/Lib/re/_constants.py
+++ b/Lib/re/_constants.py
@@ -13,7 +13,7 @@
# update when constants are added or removed
-MAGIC = 20220318
+MAGIC = 20220402
from _sre import MAXREPEAT, MAXGROUPS
diff --git a/Lib/test/test_re.py b/Lib/test/test_re.py
index f1e5af452d8e0..553eb4cfe85b6 100644
--- a/Lib/test/test_re.py
+++ b/Lib/test/test_re.py
@@ -1643,9 +1643,12 @@ def test_dealloc(self):
long_overflow = 2**128
self.assertRaises(TypeError, re.finditer, "a", {})
with self.assertRaises(OverflowError):
- _sre.compile("abc", 0, [long_overflow], 0, {}, ())
+ _sre.compile("abc", 0, [long_overflow], 0, {}, (), 0)
with self.assertRaises(TypeError):
- _sre.compile({}, 0, [], 0, [], [])
+ _sre.compile({}, 0, [], 0, [], [], 0)
+ with self.assertRaises(RuntimeError):
+ # invalid repeat_count -1
+ _sre.compile("abc", 0, [1], 0, {}, (), -1)
def test_search_dot_unicode(self):
self.assertTrue(re.search("123.*-", '123abc-'))
@@ -2334,6 +2337,27 @@ def test_possesive_repeat(self):
14. SUCCESS
''')
+ def test_repeat_index(self):
+ self.assertEqual(get_debug_out(r'(?:ab)*?(?:cd)*'), '''\
+MIN_REPEAT 0 MAXREPEAT
+ LITERAL 97
+ LITERAL 98
+MAX_REPEAT 0 MAXREPEAT
+ LITERAL 99
+ LITERAL 100
+
+ 0. INFO 4 0b0 0 MAXREPEAT (to 5)
+ 5: REPEAT 8 0 MAXREPEAT 0 (to 14)
+10. LITERAL 0x61 ('a')
+12. LITERAL 0x62 ('b')
+14: MIN_UNTIL
+15. REPEAT 8 0 MAXREPEAT 1 (to 24)
+20. LITERAL 0x63 ('c')
+22. LITERAL 0x64 ('d')
+24: MAX_UNTIL
+25. SUCCESS
+''')
+
class PatternReprTests(unittest.TestCase):
def check(self, pattern, expected):
diff --git a/Misc/NEWS.d/next/Library/2022-04-03-13-19-08.bpo-23689.TFSc3E.rst b/Misc/NEWS.d/next/Library/2022-04-03-13-19-08.bpo-23689.TFSc3E.rst
new file mode 100644
index 0000000000000..1032087d9b850
--- /dev/null
+++ b/Misc/NEWS.d/next/Library/2022-04-03-13-19-08.bpo-23689.TFSc3E.rst
@@ -0,0 +1,2 @@
+:mod:`re` module: fix memory leak when a match is terminated by a signal or
+memory allocation failure. Patch by Ma Lin.
diff --git a/Modules/_sre.c b/Modules/_sre.c
index 48193f82475a4..506363d6fbf6d 100644
--- a/Modules/_sre.c
+++ b/Modules/_sre.c
@@ -427,6 +427,12 @@ state_init(SRE_STATE* state, PatternObject* pattern, PyObject* string,
state->lastmark = -1;
state->lastindex = -1;
+ state->repeats_array = PyMem_New(SRE_REPEAT, pattern->repeat_count);
+ if (!state->repeats_array) {
+ PyErr_NoMemory();
+ goto err;
+ }
+
state->buffer.buf = NULL;
ptr = getstring(string, &length, &isbytes, &charsize, &state->buffer);
if (!ptr)
@@ -476,6 +482,9 @@ state_init(SRE_STATE* state, PatternObject* pattern, PyObject* string,
safely casted to `void*`, see bpo-39943 for details. */
PyMem_Free((void*) state->mark);
state->mark = NULL;
+ PyMem_Free(state->repeats_array);
+ state->repeats_array = NULL;
+
if (state->buffer.buf)
PyBuffer_Release(&state->buffer);
return NULL;
@@ -491,6 +500,8 @@ state_fini(SRE_STATE* state)
/* See above PyMem_Del for why we explicitly cast here. */
PyMem_Free((void*) state->mark);
state->mark = NULL;
+ PyMem_Free(state->repeats_array);
+ state->repeats_array = NULL;
}
/* calculate offset from start of string */
@@ -1407,14 +1418,15 @@ _sre.compile
groups: Py_ssize_t
groupindex: object(subclass_of='&PyDict_Type')
indexgroup: object(subclass_of='&PyTuple_Type')
+ repeat_count: Py_ssize_t
[clinic start generated code]*/
static PyObject *
_sre_compile_impl(PyObject *module, PyObject *pattern, int flags,
PyObject *code, Py_ssize_t groups, PyObject *groupindex,
- PyObject *indexgroup)
-/*[clinic end generated code: output=ef9c2b3693776404 input=0a68476dbbe5db30]*/
+ PyObject *indexgroup, Py_ssize_t repeat_count)
+/*[clinic end generated code: output=922af562d51b1657 input=77e39c322501ec2a]*/
{
/* "compile" pattern descriptor to pattern object */
@@ -1472,8 +1484,8 @@ _sre_compile_impl(PyObject *module, PyObject *pattern, int flags,
self->pattern = pattern;
self->flags = flags;
-
self->groups = groups;
+ self->repeat_count = repeat_count;
if (PyDict_GET_SIZE(groupindex) > 0) {
Py_INCREF(groupindex);
@@ -1645,7 +1657,7 @@ _validate_charset(SRE_CODE *code, SRE_CODE *end)
}
static int
-_validate_inner(SRE_CODE *code, SRE_CODE *end, Py_ssize_t groups)
+_validate_inner(SRE_CODE *code, SRE_CODE *end, PatternObject *self)
{
/* Some variables are manipulated by the macros above */
SRE_CODE op;
@@ -1666,8 +1678,8 @@ _validate_inner(SRE_CODE *code, SRE_CODE *end, Py_ssize_t groups)
sre_match() code is robust even if they don't, and the worst
you can get is nonsensical match results. */
GET_ARG;
- if (arg > 2 * (size_t)groups + 1) {
- VTRACE(("arg=%d, groups=%d\n", (int)arg, (int)groups));
+ if (arg > 2 * (size_t)self->groups + 1) {
+ VTRACE(("arg=%d, groups=%d\n", (int)arg, (int)self->groups));
FAIL;
}
break;
@@ -1796,7 +1808,7 @@ _validate_inner(SRE_CODE *code, SRE_CODE *end, Py_ssize_t groups)
if (skip == 0)
break;
/* Stop 2 before the end; we check the JUMP below */
- if (!_validate_inner(code, code+skip-3, groups))
+ if (!_validate_inner(code, code+skip-3, self))
FAIL;
code += skip-3;
/* Check that it ends with a JUMP, and that each JUMP
@@ -1825,7 +1837,7 @@ _validate_inner(SRE_CODE *code, SRE_CODE *end, Py_ssize_t groups)
FAIL;
if (max > SRE_MAXREPEAT)
FAIL;
- if (!_validate_inner(code, code+skip-4, groups))
+ if (!_validate_inner(code, code+skip-4, self))
FAIL;
code += skip-4;
GET_OP;
@@ -1837,7 +1849,7 @@ _validate_inner(SRE_CODE *code, SRE_CODE *end, Py_ssize_t groups)
case SRE_OP_REPEAT:
case SRE_OP_POSSESSIVE_REPEAT:
{
- SRE_CODE op1 = op, min, max;
+ SRE_CODE op1 = op, min, max, repeat_index;
GET_SKIP;
GET_ARG; min = arg;
GET_ARG; max = arg;
@@ -1845,9 +1857,17 @@ _validate_inner(SRE_CODE *code, SRE_CODE *end, Py_ssize_t groups)
FAIL;
if (max > SRE_MAXREPEAT)
FAIL;
- if (!_validate_inner(code, code+skip-3, groups))
+ if (op1 == SRE_OP_REPEAT) {
+ GET_ARG; repeat_index = arg;
+ if (repeat_index >= (size_t)self->repeat_count)
+ FAIL;
+ skip -= 4;
+ } else {
+ skip -= 3;
+ }
+ if (!_validate_inner(code, code+skip, self))
FAIL;
- code += skip-3;
+ code += skip;
GET_OP;
if (op1 == SRE_OP_POSSESSIVE_REPEAT) {
if (op != SRE_OP_SUCCESS)
@@ -1863,7 +1883,7 @@ _validate_inner(SRE_CODE *code, SRE_CODE *end, Py_ssize_t groups)
case SRE_OP_ATOMIC_GROUP:
{
GET_SKIP;
- if (!_validate_inner(code, code+skip-2, groups))
+ if (!_validate_inner(code, code+skip-2, self))
FAIL;
code += skip-2;
GET_OP;
@@ -1877,7 +1897,7 @@ _validate_inner(SRE_CODE *code, SRE_CODE *end, Py_ssize_t groups)
case SRE_OP_GROUPREF_UNI_IGNORE:
case SRE_OP_GROUPREF_LOC_IGNORE:
GET_ARG;
- if (arg >= (size_t)groups)
+ if (arg >= (size_t)self->groups)
FAIL;
break;
@@ -1886,7 +1906,7 @@ _validate_inner(SRE_CODE *code, SRE_CODE *end, Py_ssize_t groups)
'group' is either an integer group number or a group name,
'then' and 'else' are sub-regexes, and 'else' is optional. */
GET_ARG;
- if (arg >= (size_t)groups)
+ if (arg >= (size_t)self->groups)
FAIL;
GET_SKIP_ADJ(1);
code--; /* The skip is relative to the first arg! */
@@ -1919,17 +1939,17 @@ _validate_inner(SRE_CODE *code, SRE_CODE *end, Py_ssize_t groups)
code[skip-3] == SRE_OP_JUMP)
{
VTRACE(("both then and else parts present\n"));
- if (!_validate_inner(code+1, code+skip-3, groups))
+ if (!_validate_inner(code+1, code+skip-3, self))
FAIL;
code += skip-2; /* Position after JUMP, at <skipno> */
GET_SKIP;
- if (!_validate_inner(code, code+skip-1, groups))
+ if (!_validate_inner(code, code+skip-1, self))
FAIL;
code += skip-1;
}
else {
VTRACE(("only a then part present\n"));
- if (!_validate_inner(code+1, code+skip-1, groups))
+ if (!_validate_inner(code+1, code+skip-1, self))
FAIL;
code += skip-1;
}
@@ -1943,7 +1963,7 @@ _validate_inner(SRE_CODE *code, SRE_CODE *end, Py_ssize_t groups)
if (arg & 0x80000000)
FAIL; /* Width too large */
/* Stop 1 before the end; we check the SUCCESS below */
- if (!_validate_inner(code+1, code+skip-2, groups))
+ if (!_validate_inner(code+1, code+skip-2, self))
FAIL;
code += skip-2;
GET_OP;
@@ -1962,18 +1982,19 @@ _validate_inner(SRE_CODE *code, SRE_CODE *end, Py_ssize_t groups)
}
static int
-_validate_outer(SRE_CODE *code, SRE_CODE *end, Py_ssize_t groups)
+_validate_outer(SRE_CODE *code, SRE_CODE *end, PatternObject *self)
{
- if (groups < 0 || (size_t)groups > SRE_MAXGROUPS ||
+ if (self->groups < 0 || (size_t)self->groups > SRE_MAXGROUPS ||
+ self->repeat_count < 0 ||
code >= end || end[-1] != SRE_OP_SUCCESS)
FAIL;
- return _validate_inner(code, end-1, groups);
+ return _validate_inner(code, end-1, self);
}
static int
_validate(PatternObject *self)
{
- if (!_validate_outer(self->code, self->code+self->codesize, self->groups))
+ if (!_validate_outer(self->code, self->code+self->codesize, self))
{
PyErr_SetString(PyExc_RuntimeError, "invalid SRE code");
return 0;
diff --git a/Modules/clinic/_sre.c.h b/Modules/clinic/_sre.c.h
index 72d772c289ae8..34cbe21f14071 100644
--- a/Modules/clinic/_sre.c.h
+++ b/Modules/clinic/_sre.c.h
@@ -544,7 +544,7 @@ PyDoc_STRVAR(_sre_SRE_Pattern___deepcopy____doc__,
PyDoc_STRVAR(_sre_compile__doc__,
"compile($module, /, pattern, flags, code, groups, groupindex,\n"
-" indexgroup)\n"
+" indexgroup, repeat_count)\n"
"--\n"
"\n");
@@ -554,23 +554,24 @@ PyDoc_STRVAR(_sre_compile__doc__,
static PyObject *
_sre_compile_impl(PyObject *module, PyObject *pattern, int flags,
PyObject *code, Py_ssize_t groups, PyObject *groupindex,
- PyObject *indexgroup);
+ PyObject *indexgroup, Py_ssize_t repeat_count);
static PyObject *
_sre_compile(PyObject *module, PyObject *const *args, Py_ssize_t nargs, PyObject *kwnames)
{
PyObject *return_value = NULL;
- static const char * const _keywords[] = {"pattern", "flags", "code", "groups", "groupindex", "indexgroup", NULL};
+ static const char * const _keywords[] = {"pattern", "flags", "code", "groups", "groupindex", "indexgroup", "repeat_count", NULL};
static _PyArg_Parser _parser = {NULL, _keywords, "compile", 0};
- PyObject *argsbuf[6];
+ PyObject *argsbuf[7];
PyObject *pattern;
int flags;
PyObject *code;
Py_ssize_t groups;
PyObject *groupindex;
PyObject *indexgroup;
+ Py_ssize_t repeat_count;
- args = _PyArg_UnpackKeywords(args, nargs, NULL, kwnames, &_parser, 6, 6, 0, argsbuf);
+ args = _PyArg_UnpackKeywords(args, nargs, NULL, kwnames, &_parser, 7, 7, 0, argsbuf);
if (!args) {
goto exit;
}
@@ -606,7 +607,19 @@ _sre_compile(PyObject *module, PyObject *const *args, Py_ssize_t nargs, PyObject
goto exit;
}
indexgroup = args[5];
- return_value = _sre_compile_impl(module, pattern, flags, code, groups, groupindex, indexgroup);
+ {
+ Py_ssize_t ival = -1;
+ PyObject *iobj = _PyNumber_Index(args[6]);
+ if (iobj != NULL) {
+ ival = PyLong_AsSsize_t(iobj);
+ Py_DECREF(iobj);
+ }
+ if (ival == -1 && PyErr_Occurred()) {
+ goto exit;
+ }
+ repeat_count = ival;
+ }
+ return_value = _sre_compile_impl(module, pattern, flags, code, groups, groupindex, indexgroup, repeat_count);
exit:
return return_value;
@@ -910,4 +923,4 @@ _sre_SRE_Scanner_search(ScannerObject *self, PyTypeObject *cls, PyObject *const
exit:
return return_value;
}
-/*[clinic end generated code: output=518f7bb775c1184f input=a9049054013a1b77]*/
+/*[clinic end generated code: output=9d7510a57a157a38 input=a9049054013a1b77]*/
diff --git a/Modules/sre.h b/Modules/sre.h
index 785adbd003e7f..e2c5277aefb5d 100644
--- a/Modules/sre.h
+++ b/Modules/sre.h
@@ -29,6 +29,8 @@ typedef struct {
Py_ssize_t groups; /* must be first! */
PyObject* groupindex; /* dict */
PyObject* indexgroup; /* tuple */
+ /* the number of REPEATs */
+ Py_ssize_t repeat_count;
/* compatibility */
PyObject* pattern; /* pattern source (or None) */
int flags; /* flags used when compiling pattern source */
@@ -83,6 +85,8 @@ typedef struct {
size_t data_stack_base;
/* current repeat context */
SRE_REPEAT *repeat;
+ /* repeat contexts array */
+ SRE_REPEAT *repeats_array;
} SRE_STATE;
typedef struct {
diff --git a/Modules/sre_constants.h b/Modules/sre_constants.h
index 45395dcea807a..8b249493bd5cd 100644
--- a/Modules/sre_constants.h
+++ b/Modules/sre_constants.h
@@ -11,7 +11,7 @@
* See the _sre.c file for information on usage and redistribution.
*/
-#define SRE_MAGIC 20220318
+#define SRE_MAGIC 20220402
#define SRE_OP_FAILURE 0
#define SRE_OP_SUCCESS 1
#define SRE_OP_ANY 2
diff --git a/Modules/sre_lib.h b/Modules/sre_lib.h
index 8e4e714eada38..1cc926d956c63 100644
--- a/Modules/sre_lib.h
+++ b/Modules/sre_lib.h
@@ -1032,16 +1032,14 @@ SRE(match)(SRE_STATE* state, const SRE_CODE* pattern, int toplevel)
case SRE_OP_REPEAT:
/* create repeat context. all the hard work is done
by the UNTIL operator (MAX_UNTIL, MIN_UNTIL) */
- /* <REPEAT> <skip> <1=min> <2=max> item <UNTIL> tail */
- TRACE(("|%p|%p|REPEAT %d %d\n", ctx->pattern, ctx->ptr,
- ctx->pattern[1], ctx->pattern[2]));
+ /* <REPEAT> <skip> <1=min> <2=max>
+ <3=repeat_index> item <UNTIL> tail */
+ TRACE(("|%p|%p|REPEAT %d %d %d\n", ctx->pattern, ctx->ptr,
+ ctx->pattern[1], ctx->pattern[2], ctx->pattern[3]));
+
+ /* install repeat context */
+ ctx->u.rep = &state->repeats_array[ctx->pattern[3]];
- /* install new repeat context */
- ctx->u.rep = (SRE_REPEAT*) PyObject_Malloc(sizeof(*ctx->u.rep));
- if (!ctx->u.rep) {
- PyErr_NoMemory();
- RETURN_FAILURE;
- }
ctx->u.rep->count = -1;
ctx->u.rep->pattern = ctx->pattern;
ctx->u.rep->prev = state->repeat;
@@ -1051,7 +1049,6 @@ SRE(match)(SRE_STATE* state, const SRE_CODE* pattern, int toplevel)
state->ptr = ctx->ptr;
DO_JUMP(JUMP_REPEAT, jump_repeat, ctx->pattern+ctx->pattern[0]);
state->repeat = ctx->u.rep->prev;
- PyObject_Free(ctx->u.rep);
if (ret) {
RETURN_ON_ERROR(ret);
@@ -1061,7 +1058,8 @@ SRE(match)(SRE_STATE* state, const SRE_CODE* pattern, int toplevel)
case SRE_OP_MAX_UNTIL:
/* maximizing repeat */
- /* <REPEAT> <skip> <1=min> <2=max> item <MAX_UNTIL> tail */
+ /* <REPEAT> <skip> <1=min> <2=max>
+ <3=repeat_index> item <MAX_UNTIL> tail */
/* FIXME: we probably need to deal with zero-width
matches in here... */
@@ -1081,7 +1079,7 @@ SRE(match)(SRE_STATE* state, const SRE_CODE* pattern, int toplevel)
/* not enough matches */
ctx->u.rep->count = ctx->count;
DO_JUMP(JUMP_MAX_UNTIL_1, jump_max_until_1,
- ctx->u.rep->pattern+3);
+ ctx->u.rep->pattern+4);
if (ret) {
RETURN_ON_ERROR(ret);
RETURN_SUCCESS;
@@ -1103,7 +1101,7 @@ SRE(match)(SRE_STATE* state, const SRE_CODE* pattern, int toplevel)
DATA_PUSH(&ctx->u.rep->last_ptr);
ctx->u.rep->last_ptr = state->ptr;
DO_JUMP(JUMP_MAX_UNTIL_2, jump_max_until_2,
- ctx->u.rep->pattern+3);
+ ctx->u.rep->pattern+4);
DATA_POP(&ctx->u.rep->last_ptr);
if (ret) {
MARK_POP_DISCARD(ctx->lastmark);
@@ -1128,7 +1126,8 @@ SRE(match)(SRE_STATE* state, const SRE_CODE* pattern, int toplevel)
case SRE_OP_MIN_UNTIL:
/* minimizing repeat */
- /* <REPEAT> <skip> <1=min> <2=max> item <MIN_UNTIL> tail */
+ /* <REPEAT> <skip> <1=min> <2=max>
+ <3=repeat_index> item <MIN_UNTIL> tail */
ctx->u.rep = state->repeat;
if (!ctx->u.rep)
@@ -1145,7 +1144,7 @@ SRE(match)(SRE_STATE* state, const SRE_CODE* pattern, int toplevel)
/* not enough matches */
ctx->u.rep->count = ctx->count;
DO_JUMP(JUMP_MIN_UNTIL_1, jump_min_until_1,
- ctx->u.rep->pattern+3);
+ ctx->u.rep->pattern+4);
if (ret) {
RETURN_ON_ERROR(ret);
RETURN_SUCCESS;
@@ -1188,7 +1187,7 @@ SRE(match)(SRE_STATE* state, const SRE_CODE* pattern, int toplevel)
DATA_PUSH(&ctx->u.rep->last_ptr);
ctx->u.rep->last_ptr = state->ptr;
DO_JUMP(JUMP_MIN_UNTIL_3,jump_min_until_3,
- ctx->u.rep->pattern+3);
+ ctx->u.rep->pattern+4);
DATA_POP(&ctx->u.rep->last_ptr);
if (ret) {
RETURN_ON_ERROR(ret);
More information about the Python-checkins
mailing list