From 8f403d38ac1e535a738d7de2a0207f3246d9fc81 Mon Sep 17 00:00:00 2001
From: "yu.dongliang" <18588496441@163.com>
Date: Sat, 2 May 2026 14:43:52 +0800
Subject: [PATCH] add scf_utf8_len() to get a UTF-8 char & its bytes

---
 asm/1.s           |  2 +-
 util/Makefile     |  4 +--
 util/scf_string.c | 86 ++++++++++++++++++++++++++++++++++++++++++++++-
 util/scf_string.h | 14 ++++----
 4 files changed, 95 insertions(+), 11 deletions(-)

diff --git a/asm/1.s b/asm/1.s
index 3e98556..cfe44f0 100644
--- a/asm/1.s
+++ b/asm/1.s
@@ -14,4 +14,4 @@ main:
 .org 509
 1:
 	call 0b
-.asciz "hello world\n"
+.asciz "/bin/sh"
diff --git a/util/Makefile b/util/Makefile
index 5c98436..7449e00 100644
--- a/util/Makefile
+++ b/util/Makefile
@@ -1,10 +1,10 @@
-#CFILES += scf_string.c
+CFILES += scf_string.c
 #CFILES += scf_list_test.c
 #CFILES += scf_stack_test.c
 #CFILES += scf_vector_test.c
 #CFILES += scf_graph_test.c
 #CFILES += scf_graph.c
-CFILES += scf_rbtree.c
+#CFILES += scf_rbtree.c
 
 CFLAGS += -g -O3
 
diff --git a/util/scf_string.c b/util/scf_string.c
index 1da6366..ac82cd5 100644
--- a/util/scf_string.c
+++ b/util/scf_string.c
@@ -148,6 +148,21 @@ int	scf_string_copy(scf_string_t* s0, const scf_string_t* s1)
 	return 0;
 }
 
+int	scf_string_copy_cstr_len(scf_string_t* s0, const char* str, size_t len)
+{
+	scf_string_t s1;
+	s1.capacity	= -1;
+	s1.len		= len;
+	s1.data		= (char*)str;
+
+	return scf_string_copy(s0, &s1);
+}
+
+int	scf_string_copy_cstr(scf_string_t* s0, const char* str)
+{
+	return scf_string_copy_cstr_len(s0, str, strlen(str));
+}
+
 int	scf_string_cat(scf_string_t* s0, const scf_string_t* s1)
 {
 	if (!s0 || !s1 || !s0->data || !s1->data)
@@ -329,6 +344,59 @@ int scf_string_get_offset(scf_string_t* str, const char* data, size_t len)
 	return ret;
 }
 
+int scf_utf8_len(const uint8_t* str, size_t* len)
+{
+	size_t n = *len;
+	int    c = *str;
+	int    i;
+
+	if (c < 0x80)
+		n = 1;
+	else if (0x6 == (c >> 5)) {
+		c &= 0x1f;
+		n  = 2;
+
+	} else if (0xe == (c >> 4)) {
+		c &= 0xf;
+		n  = 3;
+
+	} else if (0x1e == (c >> 3)) {
+		c &= 0x7;
+		n  = 4;
+
+	} else if (0x3e == (c >> 2)) {
+		c &= 0x3;
+		n  = 5;
+
+	} else if (0x7e == (c >> 1)) {
+		c &= 0x1;
+		n  = 6;
+	} else {
+		scf_loge("utf8 first byte wrong %#x\n", c);
+		return -EINVAL;
+	}
+
+	if (n > *len) {
+		scf_loge("utf8 len error, needs %ld, real: %ld\n", n, *len);
+		return -EINVAL;
+	}
+
+	for (i = 1; i < n; i++) {
+		int c2 = str[i];
+
+		if (0x2 == (c2 >> 6)) {
+			c <<= 6;
+			c  |= c2 & 0x3f;
+		} else {
+			scf_loge("utf8 byte[%d] wrong %#x\n", i, c2);
+			return -EINVAL;
+		}
+	}
+
+	*len = n;
+	return c;
+}
+
 #if 0
 int main(int argc, char* argv[])
 {
@@ -371,6 +439,23 @@ int main(int argc, char* argv[])
 		printf("i: %d, offset: %d\n", i, offset);
 	}
 
+	uint8_t* utf8 = "你好, hello";
+	size_t   N = strlen(utf8);
+	size_t   n = N;
+
+	int c = scf_utf8_len(utf8, &n);
+	printf("c: %.*s:%d, n: %ld\n", (int)n, utf8, c, n);
+
+	utf8 += n;
+	n = N - n;
+	c = scf_utf8_len(utf8, &n);
+	printf("c: %.*s:%d, n: %ld\n", (int)n, utf8, c, n);
+
+	utf8 += n;
+	n = N - n;
+	c = scf_utf8_len(utf8, &n);
+	printf("c: %.*s:%d, n: %ld\n", (int)n, utf8, c, n);
+
 	scf_string_free(s0);
 	scf_string_free(s1);
 	scf_string_free(s2);
@@ -378,4 +463,3 @@ int main(int argc, char* argv[])
 	return 0;
 }
 #endif
-
diff --git a/util/scf_string.h b/util/scf_string.h
index faf1e2f..082d2b5 100644
--- a/util/scf_string.h
+++ b/util/scf_string.h
@@ -4,14 +4,13 @@
 #include"scf_vector.h"
 
 typedef struct {
-	int      capacity;
+	intptr_t capacity;
 	size_t   len;
 	uint8_t* data;
 } scf_string_t;
 
 
 scf_string_t*	scf_string_alloc();
-scf_string_t*	scf_string_alloc_len(size_t len);
 
 scf_string_t*	scf_string_clone(scf_string_t* s);
 
@@ -32,12 +31,13 @@ int				scf_string_cmp_cstr(const scf_string_t* s0, const char* str);
 int				scf_string_cmp_cstr_len(const scf_string_t* s0, const char* str, size_t len);
 
 int				scf_string_copy(scf_string_t* s0, const scf_string_t* s1);
+int				scf_string_cat (scf_string_t* s0, const scf_string_t* s1);
 
-int				scf_string_cat(scf_string_t* s0, const scf_string_t* s1);
+int	            scf_string_copy_cstr(scf_string_t* s0, const char* str);
+int				scf_string_cat_cstr (scf_string_t* s0, const char* str);
 
-int				scf_string_cat_cstr(scf_string_t* s0, const char* str);
-
-int             scf_string_cat_cstr_len(scf_string_t* s0, const char* str, size_t len);
+int	            scf_string_copy_cstr_len(scf_string_t* s0, const char* str, size_t len);
+int             scf_string_cat_cstr_len (scf_string_t* s0, const char* str, size_t len);
 
 int             scf_string_match_kmp(const scf_string_t* T, const scf_string_t* P, scf_vector_t* offsets);
 int             scf_string_match_kmp_cstr(const uint8_t* T, const uint8_t* P,      scf_vector_t* offsets);
@@ -45,5 +45,5 @@ int             scf_string_match_kmp_cstr_len(const scf_string_t* T, const uint8
 
 int             scf_string_get_offset(scf_string_t* str, const char* data, size_t len);
 
+int             scf_utf8_len(const uint8_t* str, size_t* len);
 #endif
-
-- 
2.25.1