75aa201631
Resolves: RHEL-13159,RHEL-20322,RHEL-27512,RHEL-30372,RHEL-31070,RHEL-31219,RHEL-33890,RHEL-35703,RHEL-38864,RHEL-40878,RHEL-6589
55 lines
1.9 KiB
Diff
55 lines
1.9 KiB
Diff
From d26e1ba4539a3a33224ca1019b9e6cf5590744f5 Mon Sep 17 00:00:00 2001
|
|
From: =?UTF-8?q?Zbigniew=20J=C4=99drzejewski-Szmek?= <zbyszek@in.waw.pl>
|
|
Date: Fri, 23 Jun 2023 11:10:42 -0600
|
|
Subject: [PATCH] basic/utf8: make utf8_encoded_to_unichar() return length of
|
|
the codepoint
|
|
|
|
(cherry picked from commit 9579e9a5308573c3c9c82f1978456cc71f68760c)
|
|
|
|
Related: RHEL-31219
|
|
---
|
|
src/basic/utf8.c | 10 ++++------
|
|
1 file changed, 4 insertions(+), 6 deletions(-)
|
|
|
|
diff --git a/src/basic/utf8.c b/src/basic/utf8.c
|
|
index 2532fcf81a..9d9e76904e 100644
|
|
--- a/src/basic/utf8.c
|
|
+++ b/src/basic/utf8.c
|
|
@@ -90,7 +90,7 @@ int utf8_encoded_to_unichar(const char *str, char32_t *ret_unichar) {
|
|
switch (len) {
|
|
case 1:
|
|
*ret_unichar = (char32_t)str[0];
|
|
- return 0;
|
|
+ return 1;
|
|
case 2:
|
|
unichar = str[0] & 0x1f;
|
|
break;
|
|
@@ -119,15 +119,14 @@ int utf8_encoded_to_unichar(const char *str, char32_t *ret_unichar) {
|
|
}
|
|
|
|
*ret_unichar = unichar;
|
|
-
|
|
- return 0;
|
|
+ return len;
|
|
}
|
|
|
|
bool utf8_is_printable_newline(const char* str, size_t length, bool allow_newline) {
|
|
assert(str);
|
|
|
|
for (const char *p = str; length > 0;) {
|
|
- int encoded_len, r;
|
|
+ int encoded_len;
|
|
char32_t val;
|
|
|
|
encoded_len = utf8_encoded_valid_unichar(p, length);
|
|
@@ -135,8 +134,7 @@ bool utf8_is_printable_newline(const char* str, size_t length, bool allow_newlin
|
|
return false;
|
|
assert(encoded_len > 0 && (size_t) encoded_len <= length);
|
|
|
|
- r = utf8_encoded_to_unichar(p, &val);
|
|
- if (r < 0 ||
|
|
+ if (utf8_encoded_to_unichar(p, &val) < 0 ||
|
|
unichar_is_control(val) ||
|
|
(!allow_newline && val == '\n'))
|
|
return false;
|