noselasd/nob
1
0
Fork 0
Code Issues Pull Requests 1 Actions Packages Projects Releases Wiki Activity

utf8 fixes

Browse Source
This commit is contained in:
noselasd
2026-08-11 19:40:36 +02:00
parent cb009866ff
commit 141220aecb
3 changed files with 29 additions and 20 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/string.c
+18 -11
View File
@@ -239,9 +239,16 @@ static const U8 utf8_lead[256] = {
*/
RuneDecode _str_utf8_decode_full(Str str, U32 offset)
{
const uint8_t b0 = str.str[offset];
const uint8_t desc = utf8_lead[b0];
const uint8_t width = UTF8_DESC_LENGTH(desc);
if (str.len == 0 || str.len <= offset) {
return (RuneDecode) {
.rune = RUNE_EOF,
.width = 0,
.valid = false,
};
}
U8 b0 = str.str[offset];
U8 desc = utf8_lead[b0];
U8 width = UTF8_DESC_LENGTH(desc);
/*
* width == 0 covers:
*
@@ -265,8 +272,8 @@ RuneDecode _str_utf8_decode_full(Str str, U32 offset)
* Its descriptor-selected range handles both the normal continuation
* range and the E0 / ED / F0 / F4 boundary cases.
*/
const Utf8Range range = utf8_ranges[UTF8_DESC_RANGE(desc)];
const uint8_t b1 = str.str[offset + 1];
Utf8Range range = utf8_ranges[UTF8_DESC_RANGE(desc)];
U8 b1 = str.str[offset + 1];
if (b1 < range.lo || b1 > range.hi)
goto invalid;
@@ -290,7 +297,7 @@ RuneDecode _str_utf8_decode_full(Str str, U32 offset)
*
* b1 has already been checked against 80..BF.
*/
const Rune rune =
Rune rune =
((Rune)(b0 & 0x1F) << 6) |
(Rune)(b1 & 0x3F);
@@ -308,12 +315,12 @@ RuneDecode _str_utf8_decode_full(Str str, U32 offset)
* b1 has already been validated using the descriptor-specific
* range. Only b2 still needs a continuation-byte check.
*/
const uint8_t b2 = str.str[offset + 2];
U8 b2 = str.str[offset + 2];
if ((b2 & 0xC0) != 0x80)
goto invalid;
const Rune rune =
Rune rune =
((Rune)(b0 & 0x0F) << 12) |
((Rune)(b1 & 0x3F) << 6) |
(Rune)(b2 & 0x3F);
@@ -332,8 +339,8 @@ RuneDecode _str_utf8_decode_full(Str str, U32 offset)
* b1 has already been validated using the descriptor-specific
* range. b2 and b3 use the ordinary continuation-byte form.
*/
const uint8_t b2 = str.str[offset + 2];
const uint8_t b3 = str.str[offset + 3];
U8 b2 = str.str[offset + 2];
U8 b3 = str.str[offset + 3];
/*
* Validate both remaining continuation bytes with one expression.
@@ -348,7 +355,7 @@ RuneDecode _str_utf8_decode_full(Str str, U32 offset)
if ((((b2 ^ 0x80) | (b3 ^ 0x80)) & 0xC0) != 0)
goto invalid;
const Rune rune =
Rune rune =
((Rune)(b0 & 0x07) << 18) |
((Rune)(b1 & 0x3F) << 12) |
((Rune)(b2 & 0x3F) << 6) |