ios 计算utf8 的字符个数.

支持所有语言的utf8编码的字符串.

汇编: 时间效率

//
//  LenUTF8.s
//  TestASM
//
//  Created by doorxp on 7/5/13.
//  Copyright (c) 2013 doorxp. All rights reserved.
//
//extern int lenutf(const char *src, bool noAsciiAs2);
//@src 要计算长度的字符串utf8编码
//
//@noAsciiAs2  true 非ascii做为两个字节对待,false 一个字符算一个
//
    .globl _lenutf
    .align 2
_lenutf:
START:
    EOR R5, R5              //清零计数器
    TEQ R0, #0              //
    BEQ END                 //NULL退出
LOOP:
    LDRB R3,[R0],#1
    TEQ R3, #0
    BEQ END

    AND R4, R3, #0xE0
    EORS R4, R4, #0xC0
    ADDEQ R5,R5, #1
    ADDEQ R0, R0,#0x1
    BEQ ADDAGIN

    AND R4, R3, #0xF0
    EORS R4, R4, #0xE0
    ADDEQ R5,R5, #1
    ADDEQ R0,R0, #2
    BEQ ADDAGIN
    
    AND R4, R3, #0xF8
    EORS R4, R4, #0xF0
    ADDEQ R5,R5, #1
    ADDEQ R0,R0, #0x3
    BEQ ADDAGIN
    
    AND R4, R3, #0xFC
    EORS R4, R4, #0xF8
    ADDEQ R5,R5, #1
    ADDEQ R0,R0,#4
    BEQ ADDAGIN
    
    AND R4, R3, #0xFE
    EORS R4, R4, #0xFC
    ADDEQ R5,R5, #1
    ADDEQ R0,R0, #5
    BEQ ADDAGIN
    
    ADD R5, R5, #1
    B LOOP
ADDAGIN:
    TEQ R1, #0
    ADDNE R5, R5, #1
    B LOOP
END:
    MOV R0, R5
    MOV PC,LR

c++:

size_t lengthOf(const string& src,  bool noAsciiHas2)
{
    register size_t siz = 0;
    register unsigned char item;
    register int i = 0;
    for (; i<src.size(); i++)
    {
        item = src[i];
        if (!((item & 0XE0) ^ 0XC0))
        {
            siz ++;
            if (noAsciiHas2)
            {
                siz ++ ;
            }
            i++;
        }
        else if (!((item & 0XF0) ^ 0XE0))
		{
            siz ++;
            if (noAsciiHas2)
            {
                siz ++ ;
            }
            i+=2;
		}
		else if (!((item & 0XF8) ^ 0XF0))
		{
            siz ++;
            if (noAsciiHas2)
            {
                siz ++ ;
            }
            i+=3;
		}
		else if (!((item & 0XFC) ^ 0XF8))
		{
            siz ++;
            if (noAsciiHas2)
            {
                siz ++ ;
            }
            i+=4;
		}
		else if (!((item & 0XFE) ^ 0XFC))
		{
            siz ++;
            if (noAsciiHas2)
            {
                siz ++ ;
            }
            i+=5;
		}
		else
		{
            siz ++;
		}
    }
    return siz;
}

代码简单版:

int calcCharCount(const char * pszText)
{
    int n = 0;
    char ch = 0;
    while ((ch = *pszText))
    {
        CC_BREAK_IF(! ch);

        if (0x80 != (0xC0 & ch))
        {
            ++n;
        }
        ++pszText;
    }
    return n;
}