我正在研究字符串的“胖指针”概念。基本上我有标题结构保持容量和长度信息。我用预设的字符长度分配它,然后将指针返回到第一个字符。当我想要标题信息时,我减去“sizeof”标题。
除调整大小功能外,所有功能都按照我期望的方式正常工作:
typedef uint8_t* utf8;
/*
* Resize string
*/
bool string_resize( utf8 *str, size_t room ) {
utf8* p = str;
struct string_header *hdr = (string_header_t *) (*p - sizeof(string_header_t));
size_t cap = hdr->capacity;
size_t len = hdr->length;
/* Backup the current capacity if the process fails */
size_t bck = cap;
if ( len + room <= cap ) {
//printf("::hit\n");
return true;
}
cap = len + room;
if ( cap < MAX_PREALLOC ) {
cap *= 2;
} else {
cap += MAX_PREALLOC;
}
hdr->capacity = cap;
void * new = realloc( hdr, sizeof(string_header_t) + cap + 1 );
if ( new == NULL ) {
hdr->capacity = bck;
return false;
}
*str = (utf8) new + sizeof(string_header_t);
/* Remove garbage if there is any after the string content */
memset( *str+len, 0, cap-len + 1 );
return true;
}
Valgrind 返回我在未由 malloc 分配的内存中读取的错误(尝试访问字符串的新部分时总是发生)。
如您所见,我使用(没有 typedef)uint8_t** 所以我应该将正确的指针传递给指向函数的指针,然后更改它。
非常感谢任何帮助。
[更新 1] 字符串操作上下文的附加函数:
typedef struct string_header {
size_t capacity;
size_t length;
} string_header_t;
/*
* Allocate the string with the prefered length.
*/
utf8 string_alloc( size_t len ) {
struct string_header *hdr = calloc(1, sizeof(string_header_t) + sizeof(uint8_t) * len);
assert( hdr );
hdr->capacity = len;
hdr->length = 0;
return ((utf8) hdr) + sizeof(string_header_t);
}
/*
* Allocate the new string with the initial default capacity.
*/
utf8 string_new() {
return string_alloc( INITIAL_CAPACITY );
}
/*
* Delete the string.
*/
void string_dealloc( utf8 self ) {
if ( self == NULL )
return;
string_header_t *hdr = (string_header_t *) (self - sizeof(string_header_t));
free(hdr);
}
static inline void string_push( utf8 s, char c ) {
string_header_t* hdr = (string_header_t *) (s - sizeof(string_header_t));
//*(s + hdr->length++) = (uint8_t) c;
size_t len = hdr->length++;
s[len] = c;
}
bool string_append_char( utf8 str, char c ) {
if ( string_resize(&str, 1) != ARDP_SUCCESS )
return ARDP_FAILURE;
string_push( str, c );
return ARDP_SUCCESS;
}
bool string_append_utf8( utf8 s, int cp ) {
if ( cp < 0 or cp > 0x10ffff ) {
return false;
}
else if ( cp < 0x80 ) {
return string_append_char(s, cp & 0x7F);
}
else if ( cp < 0x800 ) {
if ( string_resize( &s, 2 ) isnt ARDP_SUCCESS )
return false;
string_push( s, 0xC0 | ((cp >> 6) & 0x1F) );
string_push( s, 0x80 | (cp & 0x3F) );
}
else if ( cp < 0x10000 ) {
if ( string_resize( &s, 3 ) isnt ARDP_SUCCESS )
return false;
string_push( s, 0xE0 | ((cp >> 12) & 0xF) );
string_push( s, 0x80 | ((cp >> 6) & 0x3F) );
string_push( s, 0x80 | (cp & 0x3F) );
}
else {
if ( string_resize( &s, 4 ) isnt ARDP_SUCCESS )
return false;
string_push( s, 0xF0 | ((cp >> 18) & 0x7) );
string_push( s, 0x80 | ((cp >> 12) & 0x3F) );
string_push( s, 0x80 | ((cp >> 6) & 0x3F) );
string_push( s, 0x80 | (cp & 0x3F) );
}
return true;
}
bool string_finish( utf8 str ) {
if ( string_resize(&str, 1) )
return false;
string_header_t *hdr = (string_header_t *) (str - sizeof(string_header_t));
*(str + hdr->length) = '\0';
return true;
}
[update 2] Valgrind 日志(几乎都和这个一样):
==96370== Invalid read of size 8
==96370== at 0x100011201: string_append_char (string.c:68)
==96370== by 0x100000AE7: test_string (example.c:84)
==96370== by 0x100000BEA: main (example.c:106)
==96370== Address 0x100aac6d0 is 0 bytes inside a block of size 24 free'd
==96370== at 0x1000098B8: realloc (in /usr/local/Cellar/valgrind/HEAD/lib/valgrind/vgpreload_memcheck-amd64-darwin.so)
==96370== by 0x100011243: string_append_char (string.c:92)
==96370== by 0x100000ADA: test_string (example.c:83)
==96370== by 0x100000BEA: main (example.c:106)
==96370== Block was alloc'd at
==96370== at 0x100009551: calloc (in /usr/local/Cellar/valgrind/HEAD/lib/valgrind/vgpreload_memcheck-amd64-darwin.so)
==96370== by 0x1000110F2: string_new (string.c:38)
==96370== by 0x100000A5A: test_string (example.c:72)
==96370== by 0x100000BEA: main (example.c:106)
==96370== Invalid write of size 8
==96370== at 0x100011274: string_append_char (string.h:44)
==96370== by 0x100000AE7: test_string (example.c:84)
==96370== by 0x100000BEA: main (example.c:106)
==96370== Address 0x100aac6d8 is 8 bytes inside a block of size 24 free'd
==96370== at 0x1000098B8: realloc (in /usr/local/Cellar/valgrind/HEAD/lib/valgrind/vgpreload_memcheck-amd64-darwin.so)
==96370== by 0x100011243: string_append_char (string.c:92)
==96370== by 0x100000ADA: test_string (example.c:83)
==96370== by 0x100000BEA: main (example.c:106)
==96370== Block was alloc'd at
==96370== at 0x100009551: calloc (in /usr/local/Cellar/valgrind/HEAD/lib/valgrind/vgpreload_memcheck-amd64-darwin.so)
==96370== by 0x1000110F2: string_new (string.c:38)
==96370== by 0x100000A5A: test_string (example.c:72)
==96370== by 0x100000BEA: main (example.c:106)
[更新 3] 一些示例代码:
void test_string(void) {
utf8 str = string_new();
string_debug( str );
string_append_char( str, 'h');
string_append_char( str, 't');
string_append_char( str, 't');
string_append_char( str, 'p');
string_append_char( str, ':');
string_append_char( str, '/');
string_append_char( str, '/');
string_append_char( str, 'g');
string_append_char( str, 'o');
string_append_char( str, 'o');
string_append_char( str, 'g');
string_append_char( str, 'l');
string_append_char( str, 'e');
string_append_char( str, '.');
string_append_char( str, 'c');
string_append_char( str, 'o');
string_append_char( str, 'm');
string_append_char( str, '/');
string_append_char( str, '?');
string_append_char( str, 's');
string_append_char( str, '=');
string_append_char( str, 'f');
string_append_char( str, 'i');
string_append_char( str, 's');
string_append_char( str, 'h');
//string_finish(str);
printf("String %s", str);
string_dealloc(str);
}
最佳答案
您正在尝试使用字符串指针作为字符串结构的代理。但是字符串结构可能会被重新分配,因此字符串的地址可能会改变。为了让(例如)string_append_char
的调用者知道更改,他们必须有某种机制来接收字符串指针的新值。但他们没有;他们传入一个 uint8_t*
并返回一个 bool
。如果附加导致重新分配,一旦 string_append_char
返回,新地址将丢失。
您可以通过传递句柄(即 uint8_t**
)而不是简单的 uint8_t*
来做到这一点。但在很多方面,这都违背了界面的意义。至少,您最终会使用 &str
和其他 str
进行一些调用,这将使您的代码脆弱且难以阅读。
实际上,您还不如直接使用string
结构,并包含一个内联函数来提取C 字符串指针,类似于C++
接口(interface)。额外的间接级别可能看起来有点低效,但事实证明使用它编程要容易得多。
关于c - Realloc 在字符串调整大小函数中失败,我们在Stack Overflow上找到一个类似的问题: https://stackoverflow.com/questions/34467050/