函數memstr

大家用了memchr、strchr、strstr之後，有沒有想要一個叫memstr的函數？從內存裡面找特定的字符串？

glibc裡面沒有提供memstr，我這裡提供一個吧。驗證過的可靠版本：

[cpp]
//find 'substr' from a fixed-length buffer   
//('full_data' will be treated as binary data buffer)  
//return NULL if not found  
char* memstr(char* full_data, int full_data_len, char* substr) 
{ 
    if (full_data == NULL || full_data_len <= 0 || substr == NULL) { 
        return NULL; 
    } 
 
    if (*substr == '\0') { 
        return NULL; 
    } 
 
    int sublen = strlen(substr); 
 
    int i; 
    char* cur = full_data; 
    int last_possible = full_data_len - sublen + 1; 
    for (i = 0; i < last_possible; i++) { 
        if (*cur == *substr) { 
            //assert(full_data_len - i >= sublen);  
            if (memcmp(cur, substr, sublen) == 0) { 
                //found  
                return cur; 
            } 
        } 
        cur++; 
    } 
 
    return NULL; 
} 

//find 'substr' from a fixed-length buffer
//('full_data' will be treated as binary data buffer)
//return NULL if not found
char* memstr(char* full_data, int full_data_len, char* substr)
{
    if (full_data == NULL || full_data_len <= 0 || substr == NULL) {
        return NULL;
    }

    if (*substr == '\0') {
        return NULL;
    }

    int sublen = strlen(substr);

    int i;
    char* cur = full_data;
    int last_possible = full_data_len - sublen + 1;
    for (i = 0; i < last_possible; i++) {
        if (*cur == *substr) {
            //assert(full_data_len - i >= sublen);
            if (memcmp(cur, substr, sublen) == 0) {
                //found
                return cur;
            }
        }
        cur++;
    }

    return NULL;
}

這個函數調用了memcmp，應該可以利用memcmp的加速能力。

ps：研究過memcmp的同學應該都知道，memcmp會針對硬件做優化，比如一次比較多個字節什麼的。