sin/samba-mirror
1
0
Fork 0
mirror of https://github.com/samba-team/samba.git synced 2025-01-10 01:18:15 +03:00
Code

lib: Remove an optimization in string_replace()

Browse Source 
Why? This simplifies the code.

Why do I believe we can do this? I don't think this is a very common
operation in critical code paths. Also, next_codepoint() already has
the same optimization. If this turns out to be a measurable
performance issue, we should turn next_codepoint() into a static
inline function doing the 7-bit optimized code path inlined the same
way we did it for tdb_oob(). This way all callers would benefit from
this optimization.

Signed-off-by: Volker Lendecke <vl@samba.org>
Reviewed-by: Jeremy Allison <jra@samba.org>
This commit is contained in:
Volker Lendecke 2020-09-28 12:31:13 +02:00 committed by Jeremy Allison
parent 2c04e9a6f2
commit 9d14187c95
1 changed files with 5 additions and 29 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

34
lib/util/util_str_common.c
View File

@ -72,40 +72,16 @@ _PUBLIC_ size_t ucs2_align(const void *base_ptr, const void *p, int flags)
**/
void string_replace( char *s, char oldc, char newc )
{
char *p;
/* this is quite a common operation, so we want it to be
fast. We optimise for the ascii case, knowing that all our
supported multi-byte character sets are ascii-compatible
(ie. they match for the first 128 chars) */
for (p = s; *p; p++) {
if (*p & 0x80) /* mb string - slow path. */
break;
if (*p == oldc) {
*p = newc;
}
}
if (!*p)
return;
/* Slow (mb) path. */
#ifdef BROKEN_UNICODE_COMPOSE_CHARACTERS
/* With compose characters we must restart from the beginning. JRA. */
p = s;
#endif
while (*p) {
while (*s != '\0') {
size_t c_size;
next_codepoint(p, &c_size);
next_codepoint(s, &c_size);
if (c_size == 1) {
if (*p == oldc) {
*p = newc;
if (*s == oldc) {
*s = newc;
}
}
p += c_size;
s += c_size;
}
}