IOS使用正则表达式去掉html中的标签元素,获得纯文本

 

content是根据网址获得的网页源码字符串

NSRegularExpression *regularExpretion=[ NSRegularExpression regularExpressionWithPattern : @"<[^>]*>|\n"

                                                                                    options : 0

                                                                                      error :nil ];

     

     content=[regularExpretion stringByReplacingMatchesInString :content options : NSMatchingReportProgress range :NSMakeRange( 0 , content .length ) withTemplate : @"-" ]; //替换所有html和换行匹配元素为"-"

     

     regularExpretion=[ NSRegularExpression regularExpressionWithPattern : @"-{1,}" options : 0 error :nil ] ;

     content=[regularExpretion stringByReplacingMatchesInString :content options : NSMatchingReportProgress range :NSMakeRange( 0 , content .length ) withTemplate : @"-" ]; //把多个"-"匹配为一个"-"

     

     //根据"-"分割到数组

     NSArray *arr=[ NSArray array ];

     content=[ NSString stringWithString :content];

     arr =  [content componentsSeparatedByString : @"-" ];

     NSMutableArray *marr=[ NSMutableArray arrayWithArray :arr];

     [marr removeObject : @"" ];

     return  marr;

 

 

//正则去除网络标签
-(NSString *)getZZwithString:(NSString *)string{
    NSRegularExpression *regularExpretion=[NSRegularExpression regularExpressionWithPattern:@"<[^>]*>|\n"
                                                                                    options:0
                                                                                      error:nil];
    string=[regularExpretion stringByReplacingMatchesInString:string options:NSMatchingReportProgress range:NSMakeRange(0, string.length) withTemplate:@""];
    return string;
}