|
1
23 package org.ofbiz.product.product;
24
25 import java.util.HashMap  ;
26 import java.util.HashSet ;
27 import java.util.Iterator ;
28 import java.util.List ;
29 import java.util.Map ;
30 import java.util.Set ;
31 import java.util.StringTokenizer ;
32 import java.util.TreeSet ;
33
34 import org.ofbiz.base.util.Debug;
35 import org.ofbiz.base.util.UtilMisc;
36 import org.ofbiz.base.util.UtilProperties;
37 import org.ofbiz.base.util.UtilValidate;
38 import org.ofbiz.entity.GenericDelegator;
39 import org.ofbiz.entity.GenericEntityException;
40 import org.ofbiz.entity.GenericValue;
41
42
50 public class KeywordSearch {
51
52 public static final String module = KeywordSearch.class.getName();
53
54 public static Set thesaurusRelsToInclude = new HashSet ();
55 public static Set thesaurusRelsForReplace = new HashSet ();
56
57 static {
58 thesaurusRelsToInclude.add("KWTR_UF");
59 thesaurusRelsToInclude.add("KWTR_USE");
60 thesaurusRelsToInclude.add("KWTR_CS");
61 thesaurusRelsToInclude.add("KWTR_NT");
62 thesaurusRelsToInclude.add("KWTR_BT");
63 thesaurusRelsToInclude.add("KWTR_RT");
64
65 thesaurusRelsForReplace.add("KWTR_USE");
66 thesaurusRelsForReplace.add("KWTR_CS");
67 }
68
69 public static String getSeparators() {
70 String seps = UtilProperties.getPropertyValue("prodsearch", "index.keyword.separators", ";: ,.!?\t\"\'\r\n\\/()[]{}*%<>-+_");
72 return seps;
73 }
74
75 public static String getStopWordBagOr() {
76 return UtilProperties.getPropertyValue("prodsearch", "stop.word.bag.or");
77 }
78 public static String getStopWordBagAnd() {
79 return UtilProperties.getPropertyValue("prodsearch", "stop.word.bag.and");
80 }
81
82 public static boolean getRemoveStems() {
83 String removeStemsStr = UtilProperties.getPropertyValue("prodsearch", "remove.stems");
84 return "true".equals(removeStemsStr);
85 }
86 public static Set getStemSet() {
87 String stemBag = UtilProperties.getPropertyValue("prodsearch", "stem.bag");
88 Set stemSet = new TreeSet ();
89 if (UtilValidate.isNotEmpty(stemBag)) {
90 String curToken;
91 StringTokenizer tokenizer = new StringTokenizer (stemBag, ": ");
92 while (tokenizer.hasMoreTokens()) {
93 curToken = tokenizer.nextToken();
94 stemSet.add(curToken);
95 }
96 }
97 return stemSet;
98 }
99
100 public static void processForKeywords(String str, Map keywords, boolean forSearch, boolean anyPrefix, boolean anySuffix, boolean isAnd) {
101 String separators = getSeparators();
102 String stopWordBagOr = getStopWordBagOr();
103 String stopWordBagAnd = getStopWordBagAnd();
104
105 boolean removeStems = getRemoveStems();
106 Set stemSet = getStemSet();
107
108 processForKeywords(str, keywords, separators, stopWordBagAnd, stopWordBagOr, removeStems, stemSet, forSearch, anyPrefix, anySuffix, isAnd);
109 }
110
111 public static void processKeywordsForIndex(String str, Map keywords, String separators, String stopWordBagAnd, String stopWordBagOr, boolean removeStems, Set stemSet) {
112 processForKeywords(str, keywords, separators, stopWordBagAnd, stopWordBagOr, removeStems, stemSet, false, false, false, false);
113 }
114
115 public static void processForKeywords(String str, Map keywords, String separators, String stopWordBagAnd, String stopWordBagOr, boolean removeStems, Set stemSet, boolean forSearch, boolean anyPrefix, boolean anySuffix, boolean isAnd) {
116 Set keywordSet = makeKeywordSet(str, separators, forSearch);
117 fixupKeywordSet(keywordSet, keywords, stopWordBagAnd, stopWordBagOr, removeStems, stemSet, forSearch, anyPrefix, anySuffix, isAnd);
118 }
119
120 public static void fixupKeywordSet(Set keywordSet, Map keywords, String stopWordBagAnd, String stopWordBagOr, boolean removeStems, Set stemSet, boolean forSearch, boolean anyPrefix, boolean anySuffix, boolean isAnd) {
121 if (keywordSet == null) {
122 return;
123 }
124
125 Iterator keywordIter = keywordSet.iterator();
126 while (keywordIter.hasNext()) {
127 String token = (String ) keywordIter.next();
128
129
131 String colonToken = ":" + token + ":";
133 if (forSearch) {
134 if ((isAnd && stopWordBagAnd.indexOf(colonToken) >= 0) || (!isAnd && stopWordBagOr.indexOf(colonToken) >= 0)) {
135 continue;
136 }
137 } else {
138 if (stopWordBagOr.indexOf(colonToken) >= 0 && stopWordBagAnd.indexOf(colonToken) >= 0) {
139 continue;
140 }
141 }
142
143 if (removeStems) {
145 Iterator stemIter = stemSet.iterator();
146 while (stemIter.hasNext()) {
147 String stem = (String ) stemIter.next();
148 if (token.endsWith(stem)) {
149 token = token.substring(0, token.length() - stem.length());
150 }
151 }
152 }
153
154 if (token.length() == 0) {
156 continue;
157 }
158
159 if (token.length() == 1 && Character.isLetter(token.charAt(0))) {
161 continue;
162 }
163
164 if (forSearch) {
165 StringBuffer strSb = new StringBuffer ();
166 if (anyPrefix) strSb.append('%');
167 strSb.append(token);
168 if (anySuffix) strSb.append('%');
169 int dblPercIdx = -1;
171 while ((dblPercIdx = strSb.indexOf("%%")) >= 0) {
172 strSb.replace(dblPercIdx, dblPercIdx+2, "%");
174 }
176 token = strSb.toString();
177 }
178
179 Long curWeight = (Long ) keywords.get(token);
181 if (curWeight == null) {
182 keywords.put(token, new Long (1));
183 } else {
184 keywords.put(token, new Long (curWeight.longValue() + 1));
185 }
186 }
187 }
188
189 public static Set makeKeywordSet(String str, String separators, boolean forSearch) {
190 if (separators == null) separators = getSeparators();
191
192 Set keywords = new TreeSet ();
193 if (str.length() > 0) {
194 if (forSearch) {
195 StringBuffer sb = new StringBuffer (separators);
197 if (sb.indexOf("%") >= 0) sb.deleteCharAt(sb.indexOf("%"));
198 if (sb.indexOf("_") >= 0) sb.deleteCharAt(sb.indexOf("_"));
199 if (sb.indexOf("*") >= 0) sb.deleteCharAt(sb.indexOf("*"));
200 if (sb.indexOf("?") >= 0) sb.deleteCharAt(sb.indexOf("?"));
201 separators = sb.toString();
202 }
203
204 StringTokenizer tokener = new StringTokenizer (str, separators, false);
205 while (tokener.hasMoreTokens()) {
206 String token = tokener.nextToken().toLowerCase();
208
209 if (forSearch) {
210 token = token.replace('*', '%');
212 token = token.replace('?', '_');
213 }
214
215 keywords.add(token);
216 }
217 }
218 return keywords;
219 }
220
221 public static Set fixKeywordsForSearch(Set keywordSet, boolean anyPrefix, boolean anySuffix, boolean removeStems, boolean isAnd) {
222 Map keywords = new HashMap ();
223 fixupKeywordSet(keywordSet, keywords, getStopWordBagAnd(), getStopWordBagOr(), removeStems, getStemSet(), true, anyPrefix, anySuffix, isAnd);
224 return keywords.keySet();
225 }
226
227 public static boolean expandKeywordForSearch(String enteredKeyword, Set addToSet, GenericDelegator delegator) {
228 boolean replaceEnteredKeyword = false;
229
230 try {
231 List thesaurusList = delegator.findByAndCache("KeywordThesaurus", UtilMisc.toMap("enteredKeyword", enteredKeyword));
232 Iterator thesaurusIter = thesaurusList.iterator();
233 while (thesaurusIter.hasNext()) {
234 GenericValue keywordThesaurus = (GenericValue) thesaurusIter.next();
235 String relationshipEnumId = (String ) keywordThesaurus.get("relationshipEnumId");
236 if (thesaurusRelsToInclude.contains(relationshipEnumId)) {
237 addToSet.addAll(makeKeywordSet(keywordThesaurus.getString("alternateKeyword"), null, true));
238 if (thesaurusRelsForReplace.contains(relationshipEnumId)) {
239 replaceEnteredKeyword = true;
240 }
241 }
242 }
243 } catch (GenericEntityException e) {
244 Debug.logError(e, "Error expanding entered keyword", module);
245 }
246
247 Debug.logInfo("Expanded keyword [" + enteredKeyword + "], got set: " + addToSet, module);
248 return replaceEnteredKeyword;
249 }
250
251 public static void induceKeywords(GenericValue product) throws GenericEntityException {
252 if (product == null) return;
253 KeywordIndex.indexKeywords(product, false);
254 }
255
256 public static void induceKeywords(GenericValue product, boolean doAll) throws GenericEntityException {
257 if (product == null) return;
258 KeywordIndex.indexKeywords(product, doAll);
259 }
260 }
261 |
Popular Tags |
Java API By Example, From Geeks To Geeks.