es 插件

类 若 实现NativeScriptFactory接口。A factory to create instances of either {@link ExecutableScript} or {@link SearchScript}

只是一个工厂类，仍需要 创建 上面二者之一。实际中 需 创建 类 继承 SearchScript接口的实现类AbstractSearchScript 的 子类@ AbstractLongSearchScript @AbstractDoubleSearchScript。

我们使用 它是因为public List<NativeScriptFactory> getNativeScripts() 需要返回的是工厂。

NativeScriptFactory

/**
* A factory to create instances of either {@link ExecutableScript} or {@link SearchScript}. Note,
* if this factory creates {@link SearchScript}, it must extend {@link AbstractSearchScript}.
*
* @see AbstractExecutableScript
* @see AbstractSearchScript
* @see AbstractLongSearchScript
* @see AbstractDoubleSearchScript
*/
public interface NativeScriptFactory {

ExecutableScript （一般不用，忽略）

* An executable script, can't be used concurrently.

SearchScript  接口

AbstractSearchScript （核心类，提供了绝大部分功能的实现）

/**
* A base class for any script type that is used during the search process (custom score, aggs, and so on).
* <p>
* If the script returns a specific numeric type, consider overriding the type specific base classes
* such as {@link AbstractDoubleSearchScript} and {@link AbstractLongSearchScript}
* for better performance.
* <p>
* The use is required to implement the {@link #run()} method.
*/
public abstract class AbstractSearchScript extends AbstractExecutableScript implements LeafSearchScript {

它 的核心是 属性：

private LeafSearchLookup lookup;
private Scorer scorer;

所有方法的实现同和这两个属性有关。

setLookup（）实现lookup的初始化

通过SearchLookup调用lookup.getLeafSearchLookup(context)实现

searchLookUp则通过DefaultSearchContext.lookup()实现初始化

　　lookup():　　getQueryShardContext().lookup();

DefaultSearchContext则通过createContext实现初始化

也就是通过QueryShardContext.lookup() 实现。

QueryShardContext ： lookup = new SearchLookup(getMapperService(), indexFieldDataService, types);

其主要属性 及初始化：

public class SearchLookup {

final DocLookup docMap;

final SourceLookup sourceLookup;

final FieldsLookup fieldsLookup;

public SearchLookup(MapperService mapperService, IndexFieldDataService fieldDataService, @Nullable String[] types) {
docMap = new DocLookup(mapperService, fieldDataService, types);
sourceLookup = new SourceLookup();
fieldsLookup = new FieldsLookup(mapperService, types);
}

public class FieldsLookup {

private final MapperService mapperService;
@Nullable
private final String[] types;

FieldsLookup(MapperService mapperService, @Nullable String[] types) {
this.mapperService = mapperService;
this.types = types;
}

然后追踪 传参的来源：

queryShardContext.setTypes(ShardSearchRequest.types());

LocalTransport.sendRequest()

　　targetTransport.receiveMessage(version, data, action, requestId, this);

　　　　processReceivedMessage(data, action, sourceTransport, version, requestId);

　　　　　　StreamInput stream = StreamInput.wrap(data);

　　　　　　handleRequest(stream, requestId, data.length, sourceTransport, version);

　　　　　　　　request.readFrom(stream);

　　　　　　　　　　TaskId.readFromStream(in);

　　　　　　　　　　　　ShardSearchTransportRequest.readFrom()

　　　　　　　　　　　　　　shardSearchLocalRequest.innerReadFrom(in);

　　　　　　　　　　　　　　　　　　types = in.readStringArray();

总结： 数据有了，直接用

plsSearchScript 继承自AbstractSearchScript 类。

覆写了run（），run方法会执行plsExScript接口的run().

我们只需要提供一个实现plsExScirpt接口的类