IOS 去掉代码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:@"<[^>]*>| "  options:0  error:nil];

 

//替换所有html和换行匹配元素为"-"     

   content=[regularExpretion stringByReplacingMatchesInString:content options:NSMatchingReportProgress range:NSMakeRange(0, content.length) withTemplate:@"-"];

     

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

 

//把多个"-"匹配为一个"-"

   content=[regularExpretion stringByReplacingMatchesInString:content options:NSMatchingReportProgress range:NSMakeRange(0, content.length) withTemplate:@"-"];

     

    //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;