upstream/postgresql
1
0
Fork 0
mirror of https://github.com/postgres/postgres.git synced 2026-07-07 16:50:58 -04:00
Code Issues Projects Releases Packages Wiki Activity Actions

unicode_case.c: defend against truncated UTF8.

Browse source 
Reviewed-by: Chao Li <li.evan.chao@gmail.com>
Discussion: https://postgr.es/m/c355354e6c3f4a7aafb047361b73db247260fca0.camel@j-davis.com
Backpatch-through: 17
This commit is contained in:
Jeff Davis 2026-07-07 13:35:07 -07:00
parent a53d6d8aea
commit 05da336dd7
3 changed files with 70 additions and 14 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

24
src/backend/utils/adt/pg_locale_builtin.c
View file

@ -62,21 +62,33 @@ initcap_wbnext(void *state)
while (wbstate->offset < wbstate->len)
{
char32_t u = utf8_to_unicode((const unsigned char *) wbstate->str +
int ulen = pg_utf_mblen((const unsigned char *) wbstate->str +
wbstate->offset);
bool curr_alnum = pg_u_isalnum(u, wbstate->posix);
char32_t u;
bool curr_alnum;
size_t prev_offset = wbstate->offset;
/* invalid UTF8 */
if (wbstate->offset + ulen > wbstate->len)
{
wbstate->init = true;
wbstate->offset = wbstate->len;
return prev_offset;
}
u = utf8_to_unicode((const unsigned char *) wbstate->str +
wbstate->offset);
curr_alnum = pg_u_isalnum(u, wbstate->posix);
if (!wbstate->init || curr_alnum != wbstate->prev_alnum)
{
size_t prev_offset = wbstate->offset;
wbstate->init = true;
wbstate->offset += unicode_utf8len(u);
wbstate->offset += ulen;
wbstate->prev_alnum = curr_alnum;
return prev_offset;
}
wbstate->offset += unicode_utf8len(u);
wbstate->offset += ulen;
}
return wbstate->len;

8
src/common/unicode/case_test.c
View file

@ -296,6 +296,8 @@ tfunc_fold(char *dst, size_t dstsize, const char *src,
static void
test_convert_case(void)
{
size_t needed;
/* test string with no case changes */
test_convert(tfunc_lower, "√∞", "√∞");
/* test adjust-to-cased behavior */
@ -320,6 +322,12 @@ test_convert_case(void)
/* U+FF11 FULLWIDTH ONE is alphanumeric for full case mapping */
test_convert(tfunc_title, "\uFF11a", "\uFF11a");
/* invalid UTF8: truncated multibyte sequence */
needed = unicode_strfold(NULL, 0, "abc\xCE", 4, false);
Assert(needed == 3);
/* invalid UTF8: invalid byte */
needed = unicode_strfold(NULL, 0, "abc\xF8xyz", 7, false);
Assert(needed == 3);
#ifdef USE_ICU
icu_test_full("");

52
src/common/unicode_case.c
View file

@ -189,6 +189,22 @@ unicode_strfold(char *dst, size_t dstsize, const char *src, size_t srclen,
NULL);
}
/* local version of pg_utf_mblen() to be inlinable */
static int
utf8_mblen(const unsigned char *s)
{
if ((*s & 0x80) == 0)
return 1;
else if ((*s & 0xe0) == 0xc0)
return 2;
else if ((*s & 0xf0) == 0xe0)
return 3;
else if ((*s & 0xf8) == 0xf0)
return 4;
else
return -1;
}
/*
* Implement Unicode Default Case Conversion algorithm.
*
@ -227,12 +243,18 @@ convert_case(char *dst, size_t dstsize, const char *src, size_t srclen,
while (srcoff < srclen)
{
char32_t u1 = utf8_to_unicode((const unsigned char *) src + srcoff);
int u1len = unicode_utf8len(u1);
int u1len = utf8_mblen((const unsigned char *) src + srcoff);
char32_t u1;
char32_t simple = 0;
const char32_t *special = NULL;
enum CaseMapResult casemap_result;
/* invalid UTF8 */
if (u1len < 0 || srcoff + u1len > srclen)
break;
u1 = utf8_to_unicode((const unsigned char *) src + srcoff);
if (str_casekind == CaseTitle)
{
if (srcoff == boundary)
@ -316,7 +338,14 @@ check_final_sigma(const unsigned char *str, size_t len, size_t offset)
{
if ((str[i] & 0x80) == 0 || (str[i] & 0xC0) == 0xC0)
{
char32_t curr = utf8_to_unicode(str + i);
int u1len = utf8_mblen((const unsigned char *) str + i);
char32_t curr;
/* invalid UTF8 */
if (u1len < 0 || i + u1len > len)
return false;
curr = utf8_to_unicode(str + i);
if (pg_u_prop_case_ignorable(curr))
continue;
@ -327,8 +356,8 @@ check_final_sigma(const unsigned char *str, size_t len, size_t offset)
}
else if ((str[i] & 0xC0) == 0x80)
continue;
Assert(false); /* invalid UTF-8 */
else
return false; /* invalid UTF8 */
}
/* end of string is not followed by a Cased character */
@ -340,7 +369,14 @@ check_final_sigma(const unsigned char *str, size_t len, size_t offset)
{
if ((str[i] & 0x80) == 0 || (str[i] & 0xC0) == 0xC0)
{
char32_t curr = utf8_to_unicode(str + i);
int u1len = utf8_mblen((const unsigned char *) str + i);
char32_t curr;
/* invalid UTF8 */
if (u1len < 0 || i + u1len > len)
return false;
curr = utf8_to_unicode(str + i);
if (pg_u_prop_case_ignorable(curr))
continue;
@ -351,8 +387,8 @@ check_final_sigma(const unsigned char *str, size_t len, size_t offset)
}
else if ((str[i] & 0xC0) == 0x80)
continue;
Assert(false); /* invalid UTF-8 */
else
return false; /* invalid UTF8 */
}
return true;