万方数据知识平台 TFHpple +Xpath解析

试了一下,基本上适合所有的检索结果。

- (void)viewDidLoad
{
    [super viewDidLoad];
    // Do any additional setup after loading the view, typically from a nib.
    
    //获取数据
    NSURL *url=[NSURL URLWithString:@"http://s.g.wanfangdata.com.cn/Paper.aspx?q=hiv"];
    NSURLRequest *request=[NSURLRequest requestWithURL:url];
    NSData *data=[NSURLConnection sendSynchronousRequest:request returningResponse:nil error:nil];
    
    //解析数据
    TFHpple *hpple=[TFHpple hppleWithHTMLData:data];
    
    //XPATH字符串 越短越好,但是要能过滤出符合要求的结果
    //xpath教程:http://www.w3cschool.cn/index-14.html
    NSString *xpathStr =@"//div/ul/li [@class='title_li']   /a [not(@class)]";//解析标题 
    NSString *xpathStr2 =@"//div/ul/li [@class='zi'] ";//解析简介
    NSArray *arr=[hpple searchWithXPathQuery:xpathStr];
    NSArray *arr2=[hpple searchWithXPathQuery:xpathStr2];
    //NSLog(@"%@",arr2);
    //这个数组中就有需要的值
   
    alltilte=[[NSMutableArray alloc]init];
    alltext=[[NSMutableArray alloc]init];
    alltilte = [self SaveArry:arr];
    alltext = [self SaveAlltext:arr2];
   
    for(int i=0;i<alltilte.count;i++)
    {
        NSLog(@"标题%@",alltilte[i]);
        NSLog(@"简介%@",alltext[i]);
    }
    [self.tableView registerClass:[UITableViewCell class] forCellReuseIdentifier:@"Cell"];
    
}

-(NSMutableArray *)SaveAlltext :(NSArray *)_arry
{
    for (TFHppleElement *tmp in _arry)
    {
        NSString * string =[tmp raw];
        
        NSRange range;
        //        NSRange range2;
        range = [string rangeOfString:@"<br />"];
        string = [string substringToIndex:range.location];
        
        NSString  *temp =[string stringByReplacingOccurrencesOfString:@"<li class=\"zi\">
"
                                                           withString:@""];
        temp =[temp stringByReplacingOccurrencesOfString:@"<font color=\"red\">"
                                              withString:@""];
        temp =[temp stringByReplacingOccurrencesOfString:@"</font>"
                                              withString:@""];
        temp =[temp stringByReplacingOccurrencesOfString:@"<li class=\"zi\">"
                                              withString:@""];
        [alltext addObject:temp];
    }
    return alltext;
}



-(NSMutableArray *)SaveArry: (NSArray *)_arry
{
    
    for (TFHppleElement *tmp in _arry)
    {
        NSString * string =[tmp raw];
        NSRange range;
        NSRange range2;
        range = [string rangeOfString:@"</a>"];
        string = [string substringToIndex:range.location];
        NSString  *temp =[string stringByReplacingOccurrencesOfString:@"<font color=\"red\">" withString:@""];
        temp=[temp stringByReplacingOccurrencesOfString:@"</font>" withString:@""];
        range = [string rangeOfString:@">"];
        range2 = [string rangeOfString:@"img"];
        if(range.length!=0&&range2.length==0)
        {
            string = [temp substringFromIndex:range.location+1];
           [alltilte addObject:string];
        }
        
    }

    
    return alltilte;
}



猜你喜欢

转载自blog.csdn.net/w250130255/article/details/44342657