From db90058840223f15d7460086fd4b81a7aa3c886b Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?=D0=A1=D1=82=D1=80=D0=B0=D1=85=D0=B8=D1=9A=D0=B0=20=D0=A0?=
 =?UTF-8?q?=D0=B0=D0=B4=D0=B8=D1=9B?= <contact@strahinja.org>
Date: Sat, 6 Jul 2024 12:23:21 +0000
Subject: [PATCH] utf8.[ch]: *_char_* -> *_rune_*; (u32_to_u8,u32_strlen): Add
 max

---
 utf8.c | 20 ++++++++++----------
 utf8.h | 13 ++++++++-----
 2 files changed, 18 insertions(+), 15 deletions(-)

diff --git a/utf8.c b/utf8.c
index af1215f..aff8559 100644
--- a/utf8.c
+++ b/utf8.c
@@ -42,7 +42,7 @@ const int utf_length_table[256] = {
 const int extract_masks[6] = {XMASK1, XMASK2, XMASK3, XMASK4, XMASK5, XMASK6};
 
 int
-u8_char_to_u32(u32* to, const u8* from, size_t* from_delta)
+u8_rune_to_u32(u32* to, const u8* from, size_t* from_delta)
 {
 	const u8* pfrom = from;
 	if (!from)
@@ -64,7 +64,7 @@ u8_char_to_u32(u32* to, const u8* from, size_t* from_delta)
 }
 
 int
-u32_char_to_u8(u8* to, const u32 from)
+u32_rune_to_u8(u8* to, const u32 from)
 {
 	u32 cfrom = from;
 	u8 start  = 0;
@@ -109,7 +109,7 @@ u32_char_to_u8(u8* to, const u32 from)
 }
 
 int
-u8_to_u32(u32* to, const u8* from, size_t* from_delta)
+u8_to_u32(u32* to, const u8* from, const ssize_t max, size_t* from_delta)
 {
 	const u8* pfrom = from;
 	u32* pto	= to;
@@ -117,9 +117,9 @@ u8_to_u32(u32* to, const u8* from, size_t* from_delta)
 		return 1;
 	size_t delta = 0;
 	*from_delta  = 0;
-	while (*pfrom)
+	while (pfrom - from < max && *pfrom)
 	{
-		int result = u8_char_to_u32(pto, pfrom, &delta);
+		int result = u8_rune_to_u32(pto, pfrom, &delta);
 		if (result)
 			return result;
 		pto++;
@@ -131,16 +131,16 @@ u8_to_u32(u32* to, const u8* from, size_t* from_delta)
 }
 
 int
-u32_to_u8(u8* to, const u32* from)
+u32_to_u8(u8* to, const u32* from, const ssize_t max)
 {
 	const u32* pfrom = from;
 	u8* pto		 = to;
 	int len		 = 0;
 	if (!from)
 		return 0;
-	while (*pfrom)
+	while (pfrom - from < max && *pfrom)
 	{
-		size_t delta = u32_char_to_u8(pto, *pfrom);
+		size_t delta = u32_rune_to_u8(pto, *pfrom);
 		pto += delta;
 		len += delta;
 		pfrom++;
@@ -149,10 +149,10 @@ u32_to_u8(u8* to, const u32* from)
 }
 
 size_t
-u32_strlen(const u32* s)
+u32_strlen(const u32* s, const ssize_t max)
 {
 	const u32* ps = s;
-	while (ps && *ps)
+	while (ps && ps - s < max && *ps)
 		ps++;
 	return ps - s;
 }
diff --git a/utf8.h b/utf8.h
index a3a08ce..8691a66 100644
--- a/utf8.h
+++ b/utf8.h
@@ -5,6 +5,9 @@
 #include <stdint.h>
 #include <sys/types.h>
 
+/* maximum chars for UTF-8 representations of Unicode chars, per Unicode char */
+#define UTF8REPMAX 6
+
 #define XMASK1 0x7F /* b01111111 */
 #define XMASK2 0x1F /* b00011111 */
 #define XMASK3 0x0F /* b00001111 */
@@ -34,8 +37,8 @@
 typedef uint8_t u8;
 typedef uint32_t u32;
 
-int u8_char_to_u32(u32* to, const u8* from, size_t* from_delta);
-int u32_char_to_u8(u8* to, const u32 from);
-int u8_to_u32(u32* to, const u8* from, size_t* from_delta);
-int u32_to_u8(u8* to, const u32* from);
-size_t u32_strlen(const u32* s);
+int u8_rune_to_u32(u32* to, const u8* from, size_t* from_delta);
+int u32_rune_to_u8(u8* to, const u32 from); /* max == 1 */
+int u8_to_u32(u32* to, const u8* from, const ssize_t max, size_t* from_delta);
+int u32_to_u8(u8* to, const u32* from, const ssize_t max);
+size_t u32_strlen(const u32* s, const ssize_t max);
-- 
2.45.2