1 /* 2 Egothor Software License version 1.00 3 Copyright (C) 1997-2004 Leo Galambos. 4 Copyright (C) 2002-2004 "Egothor developers" 5 on behalf of the Egothor Project. 6 All rights reserved. 7 8 This software is copyrighted by the "Egothor developers". If this 9 license applies to a single file or document, the "Egothor developers" 10 are the people or entities mentioned as copyright holders in that file 11 or document. If this license applies to the Egothor project as a 12 whole, the copyright holders are the people or entities mentioned in 13 the file CREDITS. This file can be found in the same location as this 14 license in the distribution. 15 16 Redistribution and use in source and binary forms, with or without 17 modification, are permitted provided that the following conditions are 18 met: 19 1. Redistributions of source code must retain the above copyright 20 notice, the list of contributors, this list of conditions, and the 21 following disclaimer. 22 2. Redistributions in binary form must reproduce the above copyright 23 notice, the list of contributors, this list of conditions, and the 24 disclaimer that follows these conditions in the documentation 25 and/or other materials provided with the distribution. 26 3. The name "Egothor" must not be used to endorse or promote products 27 derived from this software without prior written permission. For 28 written permission, please contact Leo.G@seznam.cz 29 4. Products derived from this software may not be called "Egothor", 30 nor may "Egothor" appear in their name, without prior written 31 permission from Leo.G@seznam.cz. 32 33 In addition, we request that you include in the end-user documentation 34 provided with the redistribution and/or in the software itself an 35 acknowledgement equivalent to the following: 36 "This product includes software developed by the Egothor Project. 37 http://egothor.sf.net/" 38 39 THIS SOFTWARE IS PROVIDED ``AS IS'' AND ANY EXPRESSED OR IMPLIED 40 WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF 41 MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. 42 IN NO EVENT SHALL THE EGOTHOR PROJECT OR ITS CONTRIBUTORS BE LIABLE 43 FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR 44 CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF 45 SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR 46 BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, 47 WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE 48 OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN 49 IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE. 50 51 This software consists of voluntary contributions made by many 52 individuals on behalf of the Egothor Project and was originally 53 created by Leo Galambos (Leo.G@seznam.cz). 54 */ 55 package org.egothor.stemmer; 56 57 import java.util.ArrayList; 58 import java.util.Arrays; 59 import java.util.Iterator; 60 import java.util.List; 61 62 /** 63 * The Lift class is a data structure that is a variation of a Patricia trie. 64 * 65 * <p>Lift's <i>raison d'etre</i> is to implement reduction of the trie via the Lift-Up method., 66 * which makes the data structure less liable to overstemming. 67 */ 68 public class Lift extends Reduce { 69 boolean changeSkip; 70 71 /** 72 * Constructor for the Lift object. 73 * 74 * @param changeSkip when set to <code>true</code>, comparison of two Cells takes a skip command 75 * into account 76 */ Lift(boolean changeSkip)77 public Lift(boolean changeSkip) { 78 this.changeSkip = changeSkip; 79 } 80 81 /** 82 * Optimize (eliminate rows with no content) the given Trie and return the reduced Trie. 83 * 84 * @param orig the Trie to optimized 85 * @return the reduced Trie 86 */ 87 @Override optimize(Trie orig)88 public Trie optimize(Trie orig) { 89 List<CharSequence> cmds = orig.cmds; 90 List<Row> rows = new ArrayList<>(); 91 List<Row> orows = orig.rows; 92 int[] remap = new int[orows.size()]; 93 94 for (int j = orows.size() - 1; j >= 0; j--) { 95 liftUp(orows.get(j), orows); 96 } 97 98 Arrays.fill(remap, -1); 99 rows = removeGaps(orig.root, orows, new ArrayList<Row>(), remap); 100 101 return new Trie(orig.forward, remap[orig.root], cmds, rows); 102 } 103 104 /** 105 * Reduce the trie using Lift-Up reduction. 106 * 107 * <p>The Lift-Up reduction propagates all leaf-values (patch commands), where possible, to higher 108 * levels which are closer to the root of the trie. 109 * 110 * @param in the Row to consider when optimizing 111 * @param nodes contains the patch commands 112 */ liftUp(Row in, List<Row> nodes)113 public void liftUp(Row in, List<Row> nodes) { 114 Iterator<Cell> i = in.cells.values().iterator(); 115 for (; i.hasNext(); ) { 116 Cell c = i.next(); 117 if (c.ref >= 0) { 118 Row to = nodes.get(c.ref); 119 int sum = to.uniformCmd(changeSkip); 120 if (sum >= 0) { 121 if (sum == c.cmd) { 122 if (changeSkip) { 123 if (c.skip != to.uniformSkip + 1) { 124 continue; 125 } 126 c.skip = to.uniformSkip + 1; 127 } else { 128 c.skip = 0; 129 } 130 c.cnt += to.uniformCnt; 131 c.ref = -1; 132 } else if (c.cmd < 0) { 133 c.cnt = to.uniformCnt; 134 c.cmd = sum; 135 c.ref = -1; 136 if (changeSkip) { 137 c.skip = to.uniformSkip + 1; 138 } else { 139 c.skip = 0; 140 } 141 } 142 } 143 } 144 } 145 } 146 } 147