BRVAH分组功能原理分析

最近在diycode社区遇到一位同学提问，所以特写此文章来分析BRVAH分组功能的实现。如果还什么疑问都可以在这里进行提问 因为开源项目和技术分享收到 Google 的面试邀请，大家有什么想要讨论的么？

问题分析的步骤：

1. 如何使用
2. 原理分析

Adapter:

public class SectionAdapter extends BaseSectionQuickAdapter<MySection> { 

public SectionAdapter(int layoutResId, int sectionHeadResId, List data) { 
    super(layoutResId, sectionHeadResId, data); 
} 
@Override 
protected void convert(BaseViewHolder helper, MySection item) { 
    helper.setImageUrl(R.id.iv, (String) item.t); 
} 
@Override 
protected void convertHead(BaseViewHolder   helper,final MySection item) {
     helper.setText(R.id.header, item.header); 
 }

adapter的构造需要传入三个参数，分别是内容的布局和头部的布局和数据源，数据源需要继承SectionEntity如下：

Entity:

public class MySection extends SectionEntity<Video> {    
public MySection(boolean isHeader, String header, boolean isMroe) {        
    super(isHeader, header);        
}    
public MySection(Video t) {        
    super(t);    
}    
}

填充数据

public static List<MySection> getSampleData() {
    List<MySection> list = new ArrayList<>();
    list.add(new MySection(true, "Section 1"));
    list.add(new MySection(new Video(HTTPS_AVATARS1_GITHUBUSERCONTENT_COM_LINK, CYM_CHAD)));    
    list.add(new MySection(new Video(HTTPS_AVATARS1_GITHUBUSERCONTENT_COM_LINK, CYM_CHAD)));    
    list.add(new MySection(new Video(HTTPS_AVATARS1_GITHUBUSERCONTENT_COM_LINK, CYM_CHAD)));    
    list.add(new MySection(new Video(HTTPS_AVATARS1_GITHUBUSERCONTENT_COM_LINK, CYM_CHAD)));    
    list.add(new MySection(true, "Section 2"));    
    list.add(new MySection(new Video(HTTPS_AVATARS1_GITHUBUSERCONTENT_COM_LINK, CYM_CHAD)));    
    list.add(new MySection(new Video(HTTPS_AVATARS1_GITHUBUSERCONTENT_COM_LINK, CYM_CHAD)));    
    list.add(new MySection(new Video(HTTPS_AVATARS1_GITHUBUSERCONTENT_COM_LINK, CYM_CHAD)));    
    list.add(new MySection(true, "Section 3"));    
    list.add(new MySection(new Video(HTTPS_AVATARS1_GITHUBUSERCONTENT_COM_LINK, CYM_CHAD)));    
    list.add(new MySection(new Video(HTTPS_AVATARS1_GITHUBUSERCONTENT_COM_LINK, CYM_CHAD)));    
    list.add(new MySection(true, "Section 4"));    
    list.add(new MySection(new Video(HTTPS_AVATARS1_GITHUBUSERCONTENT_COM_LINK, CYM_CHAD)));    
    list.add(new MySection(new Video(HTTPS_AVATARS1_GITHUBUSERCONTENT_COM_LINK, CYM_CHAD)));    
    list.add(new MySection(true, "Section 5"));    
    list.add(new MySection(new Video(HTTPS_AVATARS1_GITHUBUSERCONTENT_COM_LINK, CYM_CHAD)));    
    list.add(new MySection(new Video(HTTPS_AVATARS1_GITHUBUSERCONTENT_COM_LINK, CYM_CHAD)));    
    return list;
}

其实头部和内容部分就是通过不同的type来实现的，我们可以查看BaseSectionQuickAdapter源码

    @Override
    protected int getDefItemViewType(int position) {
        return ((SectionEntity) mData.get(position)).isHeader ? SECTION_HEADER_VIEW : 0;
    }

它是通过SectionEntity的isHeader属性来区别是否是头部的

public abstract class SectionEntity<T> {    
    public boolean isHeader;    
    public T t;    
    public String header;    
    public SectionEntity(boolean isHeader, String header) {        
        this.isHeader = isHeader;        
        this.header = header;        
        this.t = null;    
    }    
    public SectionEntity(T t) {        
        this.isHeader = false;        
        this.header = null;        
        this.t = t;    
    }
}

这就是为什么要求开发者的实体类必须继承SectionEntity的原因了，因为需要通过它的isHeader这个属性来改变type，onCreateViewHolder通过不同的type来加载不同的布局。

@Override
protected BaseViewHolder onCreateDefViewHolder(ViewGroup parent, int viewType) {    
    if (viewType == SECTION_HEADER_VIEW)        
        return new BaseViewHolder(getItemView(mSectionHeadResId, parent));    

    return super.onCreateDefViewHolder(parent, viewType);
}

然后在onBindViewHolder里面通过type来区分头部和内容部分调用不同的方法

protected void convert(BaseViewHolder holder, Object item) {    
    switch (holder.getItemViewType()) {        
        case SECTION_HEADER_VIEW:            
        setFullSpan(holder);            
        convertHead(holder, (T) item);            
        break;        
        default:            
        convert(holder, (T) item);            
        break;    
    }
}

protected abstract void convertHead(BaseViewHolder helper, T item);

protected abstract void convert(BaseViewHolder helper, T item);

在此有同学肯定会想这个setFullSpan方法是干嘛的呀？

因为要考虑到RecyclerView的setLayoutManager的三种不同类型的LayoutManager

• LinearLayoutManager
• GridLayoutManager
• StaggeredGridLayoutManager

LinearLayoutManager：本身就可以占据一行，所以不需要做特殊的处理
GridLayoutManager：需要特殊处理,才能实现头部占据一行的效果，代码如下：

  @Override
  public void onAttachedToRecyclerView(RecyclerView recyclerView) {    
    super.onAttachedToRecyclerView(recyclerView);    
    RecyclerView.LayoutManager manager = recyclerView.getLayoutManager();    
    if (manager instanceof GridLayoutManager) {        
      final GridLayoutManager gridManager = ((GridLayoutManager) manager);        
      gridManager.setSpanSizeLookup(new GridLayoutManager.SpanSizeLookup() {            
        @Override            
        public int getSpanSize(int position) {                
          int type = getItemViewType(position);                
          return (type == SECTION_HEADER_VIEW) ? gridManager.getSpanCount() : 1;            
        });    
      }
  }

StaggeredGridLayoutManager：需要特殊处理,才能实现头部占据一行的效果，代码如下：

    protected void setFullSpan(RecyclerView.ViewHolder holder) {    
        if (holder.itemView.getLayoutParams() instanceof  StaggeredGridLayoutManager.LayoutParams) {        
            StaggeredGridLayoutManager.LayoutParams params = (StaggeredGridLayoutManager.LayoutParams) 
            holder.itemView.getLayoutParams();        
            params.setFullSpan(true);    
      }
    } 

这就是为什么要在conver方法里判断是head类型的代码块中要调用setFullSpan方法的原因了。

如果还什么疑问都可以在这里进行提问 因为开源项目和技术分享收到 Google 的面试邀请，大家有什么想要讨论的么？