Java tutorial
/* * Copyright (c) 2013-2016 Cinchapi Inc. * * Licensed 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 com.cinchapi.concourse.server.storage.db; import static com.cinchapi.concourse.server.GlobalState.STOPWORDS; import java.util.Collection; import java.util.Collections; import java.util.Map; import java.util.Map.Entry; import java.util.Set; import javax.annotation.Nullable; import javax.annotation.concurrent.ThreadSafe; import com.cinchapi.concourse.annotate.DoNotInvoke; import com.cinchapi.concourse.annotate.PackagePrivate; import com.cinchapi.concourse.server.model.Position; import com.cinchapi.concourse.server.model.PrimaryKey; import com.cinchapi.concourse.server.model.Text; import com.cinchapi.concourse.util.TStrings; import com.google.common.collect.HashMultimap; import com.google.common.collect.Maps; import com.google.common.collect.Multimap; import com.google.common.collect.Sets; import com.google.common.collect.TreeMultimap; /** * A collection of n-gram indexes that enable fulltext infix searching. For * every word in a {@link Value}, each substring index is mapped to a * {@link Position}. The entire SearchIndex contains a collection of these * mappings. * * @author Jeff Nelson */ @PackagePrivate @ThreadSafe final class SearchRecord extends Record<Text, Text, Position> { /** * DO NOT INVOKE. Use {@link Record#createSearchRecord(Text)} or * {@link Record#createSearchRecordPartial(Text, Text)} instead. * * @param locator * @param key */ @PackagePrivate @DoNotInvoke SearchRecord(Text locator, @Nullable Text key) { super(locator, key); } /** * Return the Set of primary keys for records that match {@code query}. * * @param query * @return the Set of PrimaryKeys */ public Set<PrimaryKey> search(Text query) { read.lock(); try { Multimap<PrimaryKey, Integer> reference = HashMultimap.create(); String[] toks = query.toString().toLowerCase() .split(TStrings.REGEX_GROUP_OF_ONE_OR_MORE_WHITESPACE_CHARS); boolean initial = true; int offset = 0; for (String tok : toks) { Multimap<PrimaryKey, Integer> temp = HashMultimap.create(); if (STOPWORDS.contains(tok)) { // When skipping a stop word, we must record an offset to // correctly determine if the next term match is in the // correct relative position to the previous term match ++offset; continue; } Set<Position> positions = get(Text.wrap(tok)); for (Position position : positions) { PrimaryKey key = position.getPrimaryKey(); int pos = position.getIndex(); if (initial) { temp.put(key, pos); } else { for (int current : reference.get(key)) { if (pos == current + 1 + offset) { temp.put(key, pos); } } } } initial = false; reference = temp; offset = 0; } // Result Scoring: Scoring is simply the number of times the query // appears in a document [e.g. the number of Positions mapped from // key: #reference.get(key).size()]. The total number of positions // in #reference is equal to the total number of times a document // appears in the corpus [e.g. reference.asMap().values().size()]. Multimap<Integer, PrimaryKey> sorted = TreeMultimap.create(Collections.<Integer>reverseOrder(), PrimaryKey.Sorter.INSTANCE); for (Entry<PrimaryKey, Collection<Integer>> entry : reference.asMap().entrySet()) { sorted.put(entry.getValue().size(), entry.getKey()); } return Sets.newLinkedHashSet(sorted.values()); } finally { read.unlock(); } } @Override protected Map<Text, Set<Position>> mapType() { return Maps.newHashMap(); } }