表情符号emojiUTF-8编码、Unicode、HTML显示

一、把含有表情的字符串编码 ---- UTF-8

- (NSString *)emojiEncode{ 
    NSString *uniStr = [NSString stringWithUTF8String:[self UTF8String]];
    NSData *uniData = [uniStr dataUsingEncoding:NSNonLossyASCIIStringEncoding];
    NSString *emojiText = [[NSString alloc] initWithData:uniData encoding:NSUTF8StringEncoding];
    return emojiText;
}

二、把请求到的含有表情字符串数据解码，解码后直接在UILabel上显示 ---- UTF-8

- (NSString *)emojiDecode{
    const char *jsonString = [self UTF8String];
    NSData *jsonData = [NSData dataWithBytes:jsonString length:strlen(jsonString)];
    NSString *emojiText = [[NSString alloc] initWithData:jsonData encoding:NSNonLossyASCIIStringEncoding];
    return emojiText;
}

注意：这样做微信端不能显示表情，也不能显示微信端上传的表情。

三、使用网页形式的表情格式(&#数字;)显示表情

//遍历含有表情的字符串，把表情替换成HTML格式的
//把UITextView或者UItextField里面text直接拿出来进行表情格式化
//把格式化之后的数据上传服务器

-(NSString *)replaceEmoji:(NSString *)text
{
    //过滤表情正则表达式
    NSRegularExpression *regex = [NSRegularExpression regularExpressionWithPattern:@"[^\\u0020-\\u007E\\u00A0-\\u00BE\\u2E80-\\uA4CF\\uF900-\\uFAFF\\uFE30-\\uFE4F\\uFF00-\\uFFEF\\u0080-\\u009F\\u2000-\\u201f\r\n]" options:NSRegularExpressionCaseInsensitive error:nil];
    // 对text字符串进行匹配
    NSArray *matches = [regex matchesInString:text
                                    options:0
                                      range:NSMakeRange(0, text.length)];
    // 遍历匹配后的每一条记录
    NSString *result = text;
    for (NSTextCheckingResult *match in matches) {
        NSRange range = [match range];
        NSString *target = [text substringWithRange:range];
        NSLog(@"%@", target);
        //把表情转换成HTML格式的
        NSString *emojiS = [target emojiUnicodeToHTML];
        //替换回原来的位置
        result = [result stringByReplacingOccurrencesOfString:target withString:emojiS];
    }
    return result;
}

//把表情转成网页能显示型的HTML格式(&#数字;)微信直接显示

#define MULITTHREEBYTEUTF16TOUNICODE(x,y) (((((x ^ 0xD800) << 2) | ((y ^ 0xDC00) >> 8)) << 8) | ((y ^ 0xDC00) & 0xFF)) + 0x10000

- (NSString *)emojiUnicodeToHTML{

    NSString *hexstr = @"";

    if ([self length] >= 2) {
    
        for (int i = 0; i < [self length] / 2 && ([self length] % 2 == 0) ; i++)
        
        {
            // three bytes
            if (([self characterAtIndex:i*2] & 0xFF00) == 0 ) {
                hexstr = [hexstr stringByAppendingFormat:@"Ox%1X 0x%1X",[self characterAtIndex:i*2],[self characterAtIndex:i*2+1]];
            }
            else
            {// four bytes
                hexstr = [hexstr stringByAppendingFormat:@"%1X ",MULITTHREEBYTEUTF16TOUNICODE([self characterAtIndex:i*2],[self characterAtIndex:i*2+1])];
            }
        
        }
        NSLog(@"(unicode) [%@]",hexstr);
    }
    else
    
    {
        NSLog(@"(unicode) U+%1X",[self characterAtIndex:0]);
    }

    NSString *html = [NSString stringWithFormat:@"%lu",strtoul([hexstr UTF8String],0,16)];

    return [NSString stringWithFormat:@"&#%@;",html];
}

//解析网页形式的表情格式在UILabel上显示
//把微信端表情转成Unicode编码（如果能支持unicode 字符串 那么直接获取U+1F604 输出即可，如果不支持，那么获取png图片显示即可，下面的convertSimpleUnicodeStr方法获取）
//把微信端的HTML形式的表情转换成iOS端显示用的Unicode格式

匹配HTML格式表情正则.png

- (NSString *)digitalEmojiToUnicode:(NSString *)string
{
    //把微信端上传的表情处理成iOS字符串
    NSString *text = [string htmlToString];
    //匹配HTML格式表情正则
    NSString *prefix = @"[^&#]*\\;";
    NSRegularExpression *regular = [NSRegularExpression regularExpressionWithPattern:prefix
                                                                         options:NSRegularExpressionCaseInsensitive
                                                                           error:nil];
    // 对text字符串进行匹配
    NSArray *matches = [regular matchesInString:text
                                    options:0
                                      range:NSMakeRange(0, text.length)];
    // 遍历匹配后的每一条记录
    NSString *result =  text；
    for (NSTextCheckingResult *match in matches) {
        NSRange range = [match range];
        NSString *target = [text substringWithRange:range];
        NSLog(@"%@", target);
        //把HTML格式的表情转换为Unicode格式的
        NSString *emojiS = [target emojiHTMLToUnicode];
        //iOS 端直接支持unicode字符 (有一些高级表情会失败，返回nil)
        NSString *convertUnicode = [emojiS convertSimpleUnicodeStr];
        if (convertUnicode.length ==0) {
            convertUnicode = @"❓";（高级表情显示不出来）
        }
        //把表情替换回原来的位置，然后就能直接用UILabel显示表情了
        result = [result stringByReplacingOccurrencesOfString:[@"&#" stringByAppendingString:target] withString:convertUnicode];
    
    }
    return result;
}

注意：上面代码里面的正则表达式不对，符号被改了，看图片。

- (NSString *)emojiHTMLToUnicode{

    NSString *result = [self stringByReplacingOccurrencesOfString:@"&#" withString:@""];
    result = [result stringByReplacingOccurrencesOfString:@";" withString:@""];
    NSString *hexString = [NSString stringWithFormat:@"U+%@",[[NSString alloc] initWithFormat:@"%1X",[result intValue]]];
    return hexString;
}

//这段函数作用是将 U+1F591 转换为 \U0001F591
//如果为 U+2500 那么就是补全为 \U00002500
//返回的sendStr就是可以直接在UILabel上显示的表情

- (NSString *)convertSimpleUnicodeStr{
    NSString *strUrl = [self stringByReplacingOccurrencesOfString:@"U+" withString:@""];
    unsigned long  unicodeIntValue= strtoul([strUrl UTF8String],0,16);
    //   UTF32Char inputChar = unicodeIntValue ;// 变成utf32
    unsigned long inputChar = unicodeIntValue ;// 变成utf32
    //    inputChar = NSSwapHostIntToLittle(inputChar); // 转换成Little 如果需要
inputChar = NSSwapHostLongToLittle(inputChar); // 转换成Little 如果需要
NSString *sendStr = [[NSString alloc] initWithBytes:&inputChar length:4 encoding:NSUTF32LittleEndianStringEncoding];
NSLog(@"%@",sendStr);
return sendStr;

}

//网页内容转String

- (NSString *)htmlToString{

    NSData *data = [self dataUsingEncoding:NSUTF8StringEncoding];
    if (data.length > 0) {
        return  [[NSAttributedString alloc] initWithData:data options:@{NSDocumentTypeDocumentAttribute:NSHTMLTextDocumentType,NSCharacterEncodingDocumentAttribute:[NSNumber numberWithInteger:NSUTF8StringEncoding]} documentAttributes:nil error:nil].string;
    }
    return @"";
}

注意：这个方法是把网页格式的数据变成字符串

慢慢来，一步一个巴掌印。。。。。