lucene/facet/src/java/org/apache/lucene/facet/taxonomy/TaxonomyWriter.java

/*
 * Licensed to the Apache Software Foundation (ASF) under one or more
 * contributor license agreements.  See the NOTICE file distributed with
 * this work for additional information regarding copyright ownership.
 * The ASF licenses this file to You under the Apache License, Version 2.0
 * (the "License"); you may not use this file except in compliance with
 * the License.  You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */
package org.apache.lucene.facet.taxonomy;

import java.io.Closeable;
import java.io.IOException;
import java.util.Map;
import org.apache.lucene.index.IndexWriter;
import org.apache.lucene.index.TwoPhaseCommit;

/**
 * TaxonomyWriter is the interface which the faceted-search library uses to dynamically build the
 * taxonomy at indexing time.
 *
 * <p>Notes about concurrent access to the taxonomy:
 *
 * <p>An implementation must allow multiple readers and a single writer to be active concurrently.
 * Readers follow so-called "point in time" semantics, i.e., a reader object will only see taxonomy
 * entries which were available at the time it was created. What the writer writes is only available
 * to (new) readers after the writer's commit() is called.
 *
 * <p>Faceted search keeps two indices - namely Lucene's main index, and this taxonomy index. When
 * one or more readers are active concurrently with the writer, care must be taken to avoid an
 * inconsistency between the state of these two indices: When writing to the indices, the taxonomy
 * must always be committed to disk *before* the main index, because the main index refers to
 * categories listed in the taxonomy. Such control can best be achieved by turning off the main
 * index's "autocommit" feature, and explicitly calling commit() for both indices (first for the
 * taxonomy, then for the main index). In old versions of Lucene (2.2 or earlier), when autocommit
 * could not be turned off, a more complicated solution needs to be used. E.g., use some sort of
 * (possibly inter-process) locking to ensure that a reader is being opened only right after both
 * indices have been flushed (and before anything else is written to them).
 *
 * @lucene.experimental
 */
public interface TaxonomyWriter extends Closeable, TwoPhaseCommit {

  /**
   * addCategory() adds a category with a given path name to the taxonomy, and returns its ordinal.
   * If the category was already present in the taxonomy, its existing ordinal is returned.
   *
   * <p>Before adding a category, addCategory() makes sure that all its ancestor categories exist in
   * the taxonomy as well. As result, the ordinal of a category is guaranteed to be smaller then the
   * ordinal of any of its descendants.
   */
  public int addCategory(FacetLabel categoryPath) throws IOException;

  /**
   * getParent() returns the ordinal of the parent category of the category with the given ordinal.
   *
   * <p>When a category is specified as a path name, finding the path of its parent is as trivial as
   * dropping the last component of the path. getParent() is functionally equivalent to calling
   * getPath() on the given ordinal, dropping the last component of the path, and then calling
   * getOrdinal() to get an ordinal back.
   *
   * <p>If the given ordinal is the ROOT_ORDINAL, an INVALID_ORDINAL is returned. If the given
   * ordinal is a top-level category, the ROOT_ORDINAL is returned. If an invalid ordinal is given
   * (negative or beyond the last available ordinal), an IndexOutOfBoundsException is thrown.
   * However, it is expected that getParent will only be called for ordinals which are already known
   * to be in the taxonomy. TODO (Facet): instead of a getParent(ordinal) method, consider having a
   *
   * <p>getCategory(categorypath, prefixlen) which is similar to addCategory except it doesn't add
   * new categories; This method can be used to get the ordinals of all prefixes of the given
   * category, and it can use exactly the same code and cache used by addCategory() so it means less
   * code.
   */
  public int getParent(int ordinal) throws IOException;

  /**
   * getSize() returns the number of categories in the taxonomy.
   *
   * <p>Because categories are numbered consecutively starting with 0, it means the taxonomy
   * contains ordinals 0 through getSize()-1.
   *
   * <p>Note that the number returned by getSize() is often slightly higher than the number of
   * categories inserted into the taxonomy; This is because when a category is added to the
   * taxonomy, its ancestors are also added automatically (including the root, which always get
   * ordinal 0).
   */
  public int getSize();

  /** Sets the commit user data iterable. See {@link IndexWriter#setLiveCommitData}. */
  public void setLiveCommitData(Iterable<Map.Entry<String, String>> commitUserData);

  /** Returns the commit user data iterable that was set on {@link #setLiveCommitData(Iterable)}. */
  public Iterable<Map.Entry<String, String>> getLiveCommitData();
}