diff options
Diffstat (limited to 'Modules/_sre.c')
| -rw-r--r-- | Modules/_sre.c | 129 |
1 files changed, 100 insertions, 29 deletions
diff --git a/Modules/_sre.c b/Modules/_sre.c index 4b376ec078..d09249672f 100644 --- a/Modules/_sre.c +++ b/Modules/_sre.c @@ -35,7 +35,7 @@ * other compatibility work. */ -static char copyright[] = +static const char copyright[] = " SRE 2.2.2 Copyright (c) 1997-2002 by Secret Labs AB "; #define PY_SSIZE_T_CLEAN @@ -62,9 +62,6 @@ static char copyright[] = /* -------------------------------------------------------------------- */ /* optional features */ -/* enables fast searching */ -#define USE_FAST_SEARCH - /* enables copy/deepcopy handling (work in progress) */ #undef USE_BUILTIN_COPY @@ -717,7 +714,7 @@ _sre_SRE_Pattern_search_impl(PatternObject *self, PyObject *string, } static PyObject* -call(char* module, char* function, PyObject* args) +call(const char* module, const char* function, PyObject* args) { PyObject* name; PyObject* mod; @@ -1059,7 +1056,6 @@ pattern_subx(PatternObject* self, PyObject* ptemplate, PyObject* string, PyObject* joiner; PyObject* item; PyObject* filter; - PyObject* args; PyObject* match; void* ptr; Py_ssize_t status; @@ -1161,13 +1157,7 @@ pattern_subx(PatternObject* self, PyObject* ptemplate, PyObject* string, match = pattern_new_match(self, &state, 1); if (!match) goto error; - args = PyTuple_Pack(1, match); - if (!args) { - Py_DECREF(match); - goto error; - } - item = PyObject_CallObject(filter, args); - Py_DECREF(args); + item = _PyObject_CallArg1(filter, match); Py_DECREF(match); if (!item) goto error; @@ -1516,14 +1506,12 @@ _sre_compile_impl(PyObject *module, PyObject *pattern, int flags, self->groups = groups; - Py_XINCREF(groupindex); + Py_INCREF(groupindex); self->groupindex = groupindex; - Py_XINCREF(indexgroup); + Py_INCREF(indexgroup); self->indexgroup = indexgroup; - self->weakreflist = NULL; - if (!_validate(self)) { Py_DECREF(self); return NULL; @@ -1592,7 +1580,7 @@ _sre_compile_impl(PyObject *module, PyObject *pattern, int flags, skip = *code; \ VTRACE(("%lu (skip to %p)\n", \ (unsigned long)skip, code+skip)); \ - if (skip-adj > (Py_uintptr_t)(end - code)) \ + if (skip-adj > (uintptr_t)(end - code)) \ FAIL; \ code++; \ } while (0) @@ -1626,7 +1614,7 @@ _validate_charset(SRE_CODE *code, SRE_CODE *end) case SRE_OP_CHARSET: offset = 256/SRE_CODE_BITS; /* 256-bit bitmap */ - if (offset > (Py_uintptr_t)(end - code)) + if (offset > (uintptr_t)(end - code)) FAIL; code += offset; break; @@ -1634,7 +1622,7 @@ _validate_charset(SRE_CODE *code, SRE_CODE *end) case SRE_OP_BIGCHARSET: GET_ARG; /* Number of blocks */ offset = 256/sizeof(SRE_CODE); /* 256-byte table */ - if (offset > (Py_uintptr_t)(end - code)) + if (offset > (uintptr_t)(end - code)) FAIL; /* Make sure that each byte points to a valid block */ for (i = 0; i < 256; i++) { @@ -1643,7 +1631,7 @@ _validate_charset(SRE_CODE *code, SRE_CODE *end) } code += offset; offset = arg * (256/SRE_CODE_BITS); /* 256-bit bitmap times arg */ - if (offset > (Py_uintptr_t)(end - code)) + if (offset > (uintptr_t)(end - code)) FAIL; code += offset; break; @@ -1794,11 +1782,11 @@ _validate_inner(SRE_CODE *code, SRE_CODE *end, Py_ssize_t groups) GET_ARG; prefix_len = arg; GET_ARG; /* Here comes the prefix string */ - if (prefix_len > (Py_uintptr_t)(newcode - code)) + if (prefix_len > (uintptr_t)(newcode - code)) FAIL; code += prefix_len; /* And here comes the overlap table */ - if (prefix_len > (Py_uintptr_t)(newcode - code)) + if (prefix_len > (uintptr_t)(newcode - code)) FAIL; /* Each overlap value should be < prefix_len */ for (i = 0; i < prefix_len; i++) { @@ -1927,7 +1915,7 @@ _validate_inner(SRE_CODE *code, SRE_CODE *end, Py_ssize_t groups) to allow arbitrary jumps anywhere in the code; so we just look for a JUMP opcode preceding our skip target. */ - if (skip >= 3 && skip-3 < (Py_uintptr_t)(end - code) && + if (skip >= 3 && skip-3 < (uintptr_t)(end - code) && code[skip-3] == SRE_OP_JUMP) { VTRACE(("both then and else parts present\n")); @@ -2057,8 +2045,9 @@ match_getindex(MatchObject* self, PyObject* index) /* Default value */ return 0; - if (PyLong_Check(index)) - return PyLong_AsSsize_t(index); + if (PyIndex_Check(index)) { + return PyNumber_AsSsize_t(index, NULL); + } i = -1; @@ -2135,6 +2124,12 @@ match_group(MatchObject* self, PyObject* args) return result; } +static PyObject* +match_getitem(MatchObject* self, PyObject* name) +{ + return match_getslice(self, name, Py_None); +} + /*[clinic input] _sre.SRE_Match.groups @@ -2657,6 +2652,75 @@ pattern_scanner(PatternObject *self, PyObject *string, Py_ssize_t pos, Py_ssize_ return (PyObject*) scanner; } +static Py_hash_t +pattern_hash(PatternObject *self) +{ + Py_hash_t hash, hash2; + + hash = PyObject_Hash(self->pattern); + if (hash == -1) { + return -1; + } + + hash2 = _Py_HashBytes(self->code, sizeof(self->code[0]) * self->codesize); + hash ^= hash2; + + hash ^= self->flags; + hash ^= self->isbytes; + hash ^= self->codesize; + + if (hash == -1) { + hash = -2; + } + return hash; +} + +static PyObject* +pattern_richcompare(PyObject *lefto, PyObject *righto, int op) +{ + PatternObject *left, *right; + int cmp; + + if (op != Py_EQ && op != Py_NE) { + Py_RETURN_NOTIMPLEMENTED; + } + + if (Py_TYPE(lefto) != &Pattern_Type || Py_TYPE(righto) != &Pattern_Type) { + Py_RETURN_NOTIMPLEMENTED; + } + + if (lefto == righto) { + /* a pattern is equal to itself */ + return PyBool_FromLong(op == Py_EQ); + } + + left = (PatternObject *)lefto; + right = (PatternObject *)righto; + + cmp = (left->flags == right->flags + && left->isbytes == right->isbytes + && left->codesize == right->codesize); + if (cmp) { + /* Compare the code and the pattern because the same pattern can + produce different codes depending on the locale used to compile the + pattern when the re.LOCALE flag is used. Don't compare groups, + indexgroup nor groupindex: they are derivated from the pattern. */ + cmp = (memcmp(left->code, right->code, + sizeof(left->code[0]) * left->codesize) == 0); + } + if (cmp) { + cmp = PyObject_RichCompareBool(left->pattern, right->pattern, + Py_EQ); + if (cmp < 0) { + return NULL; + } + } + if (op == Py_NE) { + cmp = !cmp; + } + return PyBool_FromLong(cmp); +} + #include "clinic/_sre.c.h" static PyMethodDef pattern_methods[] = { @@ -2701,7 +2765,7 @@ static PyTypeObject Pattern_Type = { 0, /* tp_as_number */ 0, /* tp_as_sequence */ 0, /* tp_as_mapping */ - 0, /* tp_hash */ + (hashfunc)pattern_hash, /* tp_hash */ 0, /* tp_call */ 0, /* tp_str */ 0, /* tp_getattro */ @@ -2711,7 +2775,7 @@ static PyTypeObject Pattern_Type = { pattern_doc, /* tp_doc */ 0, /* tp_traverse */ 0, /* tp_clear */ - 0, /* tp_richcompare */ + pattern_richcompare, /* tp_richcompare */ offsetof(PatternObject, weakreflist), /* tp_weaklistoffset */ 0, /* tp_iter */ 0, /* tp_iternext */ @@ -2720,6 +2784,13 @@ static PyTypeObject Pattern_Type = { pattern_getset, /* tp_getset */ }; +/* Match objects do not support length or assignment, but do support + __getitem__. */ +static PyMappingMethods match_as_mapping = { + NULL, + (binaryfunc)match_getitem, + NULL +}; static PyMethodDef match_methods[] = { {"group", (PyCFunction) match_group, METH_VARARGS, match_group_doc}, @@ -2765,7 +2836,7 @@ static PyTypeObject Match_Type = { (reprfunc)match_repr, /* tp_repr */ 0, /* tp_as_number */ 0, /* tp_as_sequence */ - 0, /* tp_as_mapping */ + &match_as_mapping, /* tp_as_mapping */ 0, /* tp_hash */ 0, /* tp_call */ 0, /* tp_str */ |