企業搜索引擎開發之鏈接器connector（三十）

時間 2019-11-05

原文原文鏈接

鏈接器裏面採用的什麼樣的數據結構，咱們先從Document迭代器開始入手，具體的Document迭代器類都實現了DocumentList接口，該接口定義了兩個方法html

public interface DocumentList {

  public Document nextDocument() throws RepositoryException;

  public String checkpoint() throws RepositoryException;
}

前者用於獲取Document對象，後者獲取斷點狀態json

上文中分析的DiffingConnectorDocumentList類即實現了DocumentList接口，從List<CheckpointAndChange> guaranteedChanges集合的迭代器中迭代獲取CheckpointAndChange對象而後包裝爲Document類型對象數據結構

Document也是一接口類型ide

public interface Document {

  public Property findProperty(String name) throws RepositoryException;

  public Set<String> getPropertyNames() throws RepositoryException;
}

從Document接口定義的方法能夠看出，Document接口相似於Map容器結構，若是進一步考察String類型的key對應的value類型Property，能夠發現Document接口很相似於HashMap結構ui

public interface Property {

  public Value nextValue() throws RepositoryException;
}

下面繼續考察Document接口的具體實現類，以JsonDocument類說明： this

/**
 *省略了其餘部分紅員屬性及方法
 * A simple {@link Document} implementation created from a {@link JSONObject}.
 */
public class JsonDocument implements Document {

   private final Map<String, List<Value>> properties;

/**
   * Constructor used by {@link DBHandle} when deserializing a
   * {@code DocumentHandle} from the recovery file.
   */
  public JsonDocument(JSONObject jsonObject) {
    this(buildJsonProperties(jsonObject), jsonObject);
  }

  /**
   * Constructor used by the {@link DocumentBuilder} for creating a
   * {@link JsonDocument} object used by {@link RepositoryHandler}
   * for building a collection over JsonDocument.
   */
  public JsonDocument(Map<String, List<Value>> properties,
                      JSONObject jsonObject) {
    this.properties = properties;
    this.jsonObject = jsonObject;
    objectId = getSingleValueString(SpiConstants.PROPNAME_DOCID);
    if (Strings.isNullOrEmpty(objectId)) {
      throw new IllegalArgumentException(
          "Unable to parse for docID from the properties:" + properties);
    }
  }

@Override
  public Set<String> getPropertyNames() {
    return properties.keySet();
  }

  @Override
  public Property findProperty(String name) throws RepositoryException {
    List<Value> property = properties.get(name);
    if (name.equals(SpiConstants.PROPNAME_CONTENT) && filterMimeType()) {
        property = null;
    }
    return (property == null) ? null : new SimpleProperty(property);
  }

}

JsonDocument類還有什麼好說的呢，內部實際是對Map<String, List<Value>> properties的封裝搜索引擎

屬性類型SimpleProperty實現了Property接口spa

/**
 * Simple implementation of the {@link Property} interface.
 * Implementors may use this directly or for reference.
 *
 * @since 1.0
 */
public class SimpleProperty implements Property {

  final Iterator<Value> iterator;

  /**
   * Constructs a property with a single value.
   *
   * @param value the property's {@link Value}
   * @since 2.4
   */
  public SimpleProperty(Value value) {
    this(Collections.singletonList(value));
  }

  /**
   * Constructs a property with multiple values.
   *
   * @param values a {@code List} of the property's {@link Value Values}
   */
  public SimpleProperty(List<Value> values) {
    this.iterator = values.iterator();
  }

  @Override
  public Value nextValue() {
    return (iterator.hasNext()) ? iterator.next() : null;
  }
}