IOS 用正则表达式解析HTML等文件,得到所有文本

获得网页内容

NSURL *url=[NSURL URLWithString:@"http://121.199.34.52/wordpress/?json=core.get_post_content&post_id=8764&post_type=post"];
     NSDictionary * dic=[NSJSONSerialization JSONObjectWithData:[NSData dataWithContentsOfURL:url] options:0 error:Nil];
 
  NSString *content=[dic objectForKey:@"content"];

正则表达式

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:@""];
    for (NSString *str in marr) {
           NSLog(@"呵呵-------------%@",str);
        
    }

去除字符串中所有得空格及控制字符：

str = [str stringByTrimmingCharactersInSet:[NSCharacterSet whitespaceAndNewlineCharacterSet ]];