traverseda
/
cpython
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

bpo-32677: Optimize str.isascii() (GH-5356)

This commit is contained in:
INADA Naoki 2018-01-28 09:59:12 +09:00 committed by GitHub
parent ea8fc52e75
commit bea57060c8
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 43 additions and 4 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
Lib/test/string_tests.py
View File

@ -916,6 +916,13 @@ class BaseTest:
self.checkequal(True, '\x00\x7f', 'isascii') self.checkequal(True, '\x00\x7f', 'isascii')
self.checkequal(False, '\x80', 'isascii') self.checkequal(False, '\x80', 'isascii')
self.checkequal(False, '\xe9', 'isascii') self.checkequal(False, '\xe9', 'isascii')
# bytes.isascii() and bytearray.isascii() has optimization which
# check 4 or 8 bytes at once. So check some alignments.
for p in range(8):
self.checkequal(True, ' '*p + '\x7f', 'isascii')
self.checkequal(False, ' '*p + '\x80', 'isascii')
self.checkequal(True, ' '*p + '\x7f' + ' '*8, 'isascii')
self.checkequal(False, ' '*p + '\x80' + ' '*8, 'isascii')
def test_isdigit(self): def test_isdigit(self):
self.checkequal(False, '', 'isdigit') self.checkequal(False, '', 'isdigit')

40
Objects/bytes_methods.c
View File

@ -98,19 +98,51 @@ PyDoc_STRVAR_shared(_Py_isascii__doc__,
Return True if B is empty or all characters in B are ASCII,\n\ Return True if B is empty or all characters in B are ASCII,\n\
False otherwise."); False otherwise.");
// Optimization is copied from ascii_decode in unicodeobject.c
/* Mask to quickly check whether a C 'long' contains a
non-ASCII, UTF8-encoded char. */
#if (SIZEOF_LONG == 8)
# define ASCII_CHAR_MASK 0x8080808080808080UL
#elif (SIZEOF_LONG == 4)
# define ASCII_CHAR_MASK 0x80808080UL
#else
# error C 'long' size should be either 4 or 8!
#endif
PyObject* PyObject*
_Py_bytes_isascii(const char *cptr, Py_ssize_t len) _Py_bytes_isascii(const char *cptr, Py_ssize_t len)
{ {
const unsigned char *p = (unsigned char *) cptr; const char *p = cptr;
const unsigned char *e = p + len; const char *end = p + len;
for (; p < e; p++) { const char *aligned_end = (const char *) _Py_ALIGN_DOWN(end, SIZEOF_LONG);
if (*p >= 128) {
while (p < end) {
/* Fast path, see in STRINGLIB(utf8_decode) in stringlib/codecs.h
for an explanation. */
if (_Py_IS_ALIGNED(p, SIZEOF_LONG)) {
/* Help allocation */
const char *_p = p;
while (_p < aligned_end) {
unsigned long value = *(unsigned long *) _p;
if (value & ASCII_CHAR_MASK) {
Py_RETURN_FALSE;
}
_p += SIZEOF_LONG;
}
p = _p;
if (_p == end)
break;
}
if ((unsigned char)*p & 0x80) {
Py_RETURN_FALSE; Py_RETURN_FALSE;
} }
p++;
} }
Py_RETURN_TRUE; Py_RETURN_TRUE;
} }
#undef ASCII_CHAR_MASK
PyDoc_STRVAR_shared(_Py_isdigit__doc__, PyDoc_STRVAR_shared(_Py_isdigit__doc__,
"B.isdigit() -> bool\n\ "B.isdigit() -> bool\n\