HTML解析类TFHpple框架

TFhpple框架github

• 结构如下

          <div class="album_box clearfix">
            <div class="a_left">
              <a href="/album/550fcb41dfa688445500002c">
                ![](https://img.haomeiwen.com/i1933166/da7fb99c21b410d3.jpg?imageMogr2/auto-orient/strip%7CimageView2/2/w/1240)
                <p>连载中</p>
              </a>
            </div>
             
            <div class="a_right">
             <h1><a href="/album/550fcb41dfa688445500002c">青春的疲倦</a></h1>
             <p>作者：<a href="/profile/1179643" class="blue">骆瑞生Rhettson</a>
                    
             </p>
             <p>分集：6<span class="genre">类型：故事</span></p>
             <div class="a_article">
              <p><a href="/album/550fcb41dfa688445500002c">骆瑞生的小说、文章与诗词。</a></p>
                          </div>
              <div class="a_idea">
               <a>阅读：<span class="num">7739</span></a>
               <a>推荐：<span class="num">28</span></a>
              </div>
             </div>
             
          </div>
         <!--album_box end-->
                   <!--album_box start-->
          <div class="album_box clearfix">
             
            <div class="a_left">
              <a href="/album/551bb249dfa688b519000002">
                ![](https://img.haomeiwen.com/i1933166/80773ffc1dae17a1.jpeg?imageMogr2/auto-orient/strip%7CimageView2/2/w/1240)
                <p>连载中</p>
              </a>
            </div>
             
            <div class="a_right">
             <h1><a href="/album/551bb249dfa688b519000002">实时评论——书籍电影</a></h1>
             <p>作者：<a href="/profile/567917" class="blue">YishuiH-AN易水涵</a>
                    
             </p>
             <p>分集：38<span class="genre">类型：随笔</span></p>
             <div class="a_article">
              <p><a href="/album/551bb249dfa688b519000002">这里是一些电影与书籍的整合推荐及书影评论 当然也会择选一些以往好的文艺产物分享 如果你需要文艺的情怀来冲淡当下的物欲生活 不妨进来看看这些作家、歌手、导演、专辑、电影、小说、散文……</a></p>
                          </div>
              <div class="a_idea">
               <a>阅读：<span class="num">185033</span></a>
               <a>推荐：<span class="num">5231</span></a>
              </div>
             </div>
             
          </div>
         <!--album_box end-->
                   <!--album_box start-->
          <div class="album_box clearfix">
             
            <div class="a_left">
              <a href="/album/5515ebeddfa688bb24000041">
                ![](https://img.haomeiwen.com/i1933166/a1b5845ff3fb1f54.jpeg?imageMogr2/auto-orient/strip%7CimageView2/2/w/1240)
                <p>连载中</p>
              </a>
            </div>
             
            <div class="a_right">
             <h1><a href="/album/5515ebeddfa688bb24000041">梦想开始的地方</a></h1>
             <p>作者：<a href="/profile/1167833" class="blue">郭道甲</a>
                    
             </p>
             <p>分集：52<span class="genre">类型：故事</span></p>
             <div class="a_article">
              <p><a href="/album/5515ebeddfa688bb24000041">每一篇故事，都有你曾经的影子。</a></p>
                          </div>
              <div class="a_idea">
               <a>阅读：<span class="num">166656</span></a>
               <a>推荐：<span class="num">6158</span></a>
              </div>
             </div>
             
          </div>
         

• 以获取文章列表标题 简介 内容为例
  自定义model如下

@property (nonatomic, copy)NSString *title;
@property (nonatomic, copy)NSString *desc;
@property (nonatomic, copy)NSString *content;

• 开始解析

- (NSMutableArray *)praseWithTFHpple:(TFHpple *)hppleObject {
    NSMutableArray<ArticleListModel *> *tempArray = [@[] mutableCopy];
    NSArray *divArr = [hppleObject searchWithXPathQuery:@"//div"];
    for (TFHppleElement *element in divArr) {
        if ([[element objectForKey:@"class"] isEqualToString:@"moving_article clearfix"]) {
            ArticleListModel *articleModel = [[ArticleListModel alloc] init];
            //获取title
            TFHppleElement *h2Element = [[element searchWithXPathQuery:@"//h2"] firstObject];
            articleModel.title = [h2Element content];   
            //获取desc
            NSArray *pArray = [element searchWithXPathQuery:@"//p"];
            for (TFHppleElement *pElement in pArray) {
                if ([[pElement objectForKey:@"class"] isEqualToString:@"clearfix"]) {
                    articleModel.desc = [[pElement content] stringByReplacingOccurrencesOfString:@"\n" withString:@""];
                }
            } 
            //获取content
            NSArray *_divArray = [element searchWithXPathQuery:@"//div"];
            for (TFHppleElement *_divElement in _divArray) {
                if ([[_divElement objectForKey:@"class"] isEqualToString:@"long"]) {
                    articleModel.content = [_divElement content];
                }
            }  
           NSLog(@"%@",[articleModel descroption]);
        }
    }
    return tempArray;
}