summaryrefslogtreecommitdiffstats
path: root/Objects
diff options
context:
space:
mode:
authorAntoine Pitrou <solipsis@pitrou.net>2011-10-11 18:29:21 (GMT)
committerAntoine Pitrou <solipsis@pitrou.net>2011-10-11 18:29:21 (GMT)
commit2c3b2302adb1bb00b6050afc30eacbc023379b93 (patch)
treec41e4702224848fd137a0e6043949d1fa88b3b77 /Objects
parent60dd7dc23c37bbb1c6ecc83a866f1a379afce443 (diff)
downloadcpython-2c3b2302adb1bb00b6050afc30eacbc023379b93.zip
cpython-2c3b2302adb1bb00b6050afc30eacbc023379b93.tar.gz
cpython-2c3b2302adb1bb00b6050afc30eacbc023379b93.tar.bz2
Issue #13134: optimize finding single-character strings using memchr
Diffstat (limited to 'Objects')
-rw-r--r--Objects/stringlib/fastsearch.h73
1 files changed, 73 insertions, 0 deletions
diff --git a/Objects/stringlib/fastsearch.h b/Objects/stringlib/fastsearch.h
index d35cba3..33ab6ff 100644
--- a/Objects/stringlib/fastsearch.h
+++ b/Objects/stringlib/fastsearch.h
@@ -32,6 +32,60 @@
#define STRINGLIB_BLOOM(mask, ch) \
((mask & (1UL << ((ch) & (STRINGLIB_BLOOM_WIDTH -1)))))
+
+Py_LOCAL_INLINE(Py_ssize_t)
+STRINGLIB(fastsearch_memchr_1char)(const STRINGLIB_CHAR* s, Py_ssize_t n,
+ STRINGLIB_CHAR ch, unsigned char needle,
+ Py_ssize_t maxcount, int mode)
+{
+ void *candidate;
+ const STRINGLIB_CHAR *found;
+
+#define DO_MEMCHR(memchr, s, needle, nchars) do { \
+ candidate = memchr((const void *) (s), (needle), (nchars) * sizeof(STRINGLIB_CHAR)); \
+ found = (const STRINGLIB_CHAR *) \
+ ((Py_ssize_t) candidate & (~ ((Py_ssize_t) sizeof(STRINGLIB_CHAR) - 1))); \
+ } while (0)
+
+ if (mode == FAST_SEARCH) {
+ const STRINGLIB_CHAR *_s = s;
+ const STRINGLIB_CHAR *e = s + n;
+ while (_s < e) {
+ DO_MEMCHR(memchr, _s, needle, e - _s);
+ if (found == NULL)
+ return -1;
+ if (sizeof(STRINGLIB_CHAR) == 1 || *found == ch)
+ return (found - _s);
+ /* False positive */
+ _s = found + 1;
+ }
+ return -1;
+ }
+#ifdef HAVE_MEMRCHR
+ /* memrchr() is a GNU extension, available since glibc 2.1.91.
+ it doesn't seem as optimized as memchr(), but is still quite
+ faster than our hand-written loop in FASTSEARCH below */
+ else if (mode == FAST_RSEARCH) {
+ while (n > 0) {
+ DO_MEMCHR(memrchr, s, needle, n);
+ if (found == NULL)
+ return -1;
+ n = found - s;
+ if (sizeof(STRINGLIB_CHAR) == 1 || *found == ch)
+ return n;
+ /* False positive */
+ }
+ return -1;
+ }
+#endif
+ else {
+ assert(0); /* Should never get here */
+ return 0;
+ }
+
+#undef DO_MEMCHR
+}
+
Py_LOCAL_INLINE(Py_ssize_t)
FASTSEARCH(const STRINGLIB_CHAR* s, Py_ssize_t n,
const STRINGLIB_CHAR* p, Py_ssize_t m,
@@ -51,6 +105,25 @@ FASTSEARCH(const STRINGLIB_CHAR* s, Py_ssize_t n,
if (m <= 0)
return -1;
/* use special case for 1-character strings */
+ if (n > 10 && (mode == FAST_SEARCH
+#ifdef HAVE_MEMRCHR
+ || mode == FAST_RSEARCH
+#endif
+ )) {
+ /* use memchr if we can choose a needle without two many likely
+ false positives */
+ unsigned char needle;
+ int use_needle = 1;
+ needle = p[0] & 0xff;
+ if (needle == 0 && sizeof(STRINGLIB_CHAR) > 1) {
+ needle = (p[0] >> 8) & 0xff;
+ if (needle >= 32)
+ use_needle = 0;
+ }
+ if (use_needle)
+ return STRINGLIB(fastsearch_memchr_1char)
+ (s, n, p[0], needle, maxcount, mode);
+ }
if (mode == FAST_COUNT) {
for (i = 0; i < n; i++)
if (s[i] == p[0]) {