solr 7.0 与spring-data 3.0整合 --（4）投影及排序

往往为了利于检索，solr的schema中会有大量的字段，但是提交到客户端并不会把所有的字段信息都返回，这时候只会投影部分字段信息。

例如，完整的结构如下：

"data": [
    {
      "resId": "1713074",
      "musicId": "1462975",
      "musicName": "말해버리면 (如果说出来的话)",
      "areaName": "其他",
      "typeId": 4,
      "typeName": "MV",
      "singerName": "안녕의 온도",
      "listenNumber": 438,
      "singerTypeName": "未知",
      "publicTime": "2018-5-8",
      "resName": "말해버리면 (如果说出来的话)",
      "resNamePy": "말해버리면 (RGSCLDH)",
      "resSingerPy": "안녕의 온도(WHDWD)",
      "picture": "http://puui.qpic.cn/qqvideo_ori/0/g0026pq44za_496_280/0",
      "tags": "地区,主题,年代,2018",
      "keywords": [
        "말해버리면 (如果说出来的话)",
        "其他",
        "MV",
        "안녕의 온도",
        "未知",
        "2018-5-8",
        "地区,主题,年代,2018"
      ]
    }
  ],

但是我们只想返回给客户端：

"data": [
    {
      "musicId": "1462975",
      "musicName": "말해버리면 (如果说出来的话)",
      "typeId": 4,
      "singerName": "안녕의 온도",
      "publicTime": "2018-5-8",
      "picture": "http://puui.qpic.cn/qqvideo_ori/0/g0026pq44za_496_280/0",
    }
  ],

这时就需要用到投影(projection)了。

投影

CustomMusicRepository接口中增加：

public ScoredPage searchByNameProjection(String searchName, long pageIndex, int pageSize);

MusicRepositoryImpl实现：

@Override
public ScoredPage searchByNameProjection(String searchName, long pageIndex, int pageSize) {
    String[] words = searchName.toUpperCase().split(" ");
    String[] contidionFields = new String[]{Music.FIELD_SONG_NAME, Music.FIELD_SINGER_NAME};
    Criteria conditions = createSearchConditions(contidionFields, words);
    SimpleQuery search = new SimpleQuery(conditions);

    // 投影
    search.addProjectionOnField(Music.FIELD_SONG_NAME);
    search.setOffset(pageIndex);
    search.setRows(pageSize);
    ScoredPage page = solrTemplate.queryForPage(Music.CORE_NAME, search, Music.class);
    return page;
}

以上代码将songName加入到投影中，那么输出就只有songName才会有值，其余字段均为null。

测试代码：

@Test
public void searchByNameProjection() {
    ScoredPage musicList = musicRepository.searchByNameProjection("刘德华", 0, 10);
    for (Music music : musicList.getContent()) {
        System.out.println(music);
    }
}

结果如下：

com.ali.model.Music@64a8c844[resId=,songId=,songName=《追龙》持续超车 刘德华身兼化妆师搬运工,singerName=,keywords=]
com.ali.model.Music@3f6db3fb[resId=,songId=,songName=情深的一句,singerName=,keywords=]
com.ali.model.Music@52de51b6[resId=,songId=,songName=无心快意,singerName=,keywords=]
com.ali.model.Music@18c5069b[resId=,songId=,songName=刘德华,singerName=,keywords=]

排序

排序比较简单，将上面的方法改造一下，加入排序：

 @Override
public ScoredPage searchByNameProjection(String searchName, long pageIndex, int pageSize) {
    String[] words = searchName.toUpperCase().split(" ");
    String[] contidionFields = new String[]{Music.FIELD_SONG_NAME, Music.FIELD_SINGER_NAME};
    Criteria conditions = createSearchConditions(contidionFields, words);
    SimpleQuery search = new SimpleQuery(conditions);

    Sort sort = new Sort(Sort.Direction.DESC, Music.FIELD_SONG_ID);
    search.addSort(sort);
    
    search.addProjectionOnField(Music.FIELD_SONG_NAME);
    search.setOffset(pageIndex);
    search.setRows(pageSize);
    ScoredPage page = solrTemplate.queryForPage(Music.CORE_NAME, search, Music.class);
    return page;
}
执行上面代码，返回结果即会按照songId 倒序排列。