source: trunk/gsdl3/src/java/org/greenstone/gsdl3/service/AbstractSearch.java@ 14002

Last change on this file since 14002 was 14002, checked in by qq6, 17 years ago

add if-statements whencreating INDEX_LANGUAGE_PARAM, INDEX_PARAM ,INDEX_SUBCOLLECTION_PARAM
  • Property svn:keywords set to Author Date Id Revision
File size: 15.5 KB
Line 
1/*
2 * AbstractSearch.java
3 * Copyright (C) 2005 New Zealand Digital Library, http://www.nzdl.org
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * it under the terms of the GNU General Public License as published by
7 * the Free Software Foundation; either version 2 of the License, or
8 * (at your option) any later version.
9 *
10 * This program is distributed in the hope that it will be useful,
11 * but WITHOUT ANY WARRANTY; without even the implied warranty of
12 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
13 * GNU General Public License for more details.
14 *
15 * You should have received a copy of the GNU General Public License
16 * along with this program; if not, write to the Free Software
17 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
18 */
19package org.greenstone.gsdl3.service;
20
21// Greenstone classes
22import org.greenstone.gsdl3.util.GSXML;
23import org.greenstone.gsdl3.util.GSPath;
24
25// XML classes
26import org.w3c.dom.Document;
27import org.w3c.dom.Element;
28import org.w3c.dom.NodeList;
29
30// java classes
31import java.util.ArrayList;
32import java.util.HashMap;
33
34import org.apache.log4j.*;
35
36/** Partially implements a generic search service
37 *
38 * @author <a href="mailto:[email protected]">Katherine Don</a>
39 */
40
41public abstract class AbstractSearch
42 extends ServiceRack
43{
44
45 static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.AbstractSearch.class.getName());
46
47
48 // the search service
49 protected static final String TEXT_QUERY_SERVICE = "TextQuery";
50
51 // compulsory params
52 protected static final String INDEX_PARAM = "index";
53 protected static final String QUERY_PARAM = "query";
54
55 // optional standard params - some of these have to be implemented
56 protected static final String INDEX_SUBCOLLECTION_PARAM = "indexSubcollection";
57 protected static final String INDEX_LANGUAGE_PARAM = "indexLanguage";
58 protected static final String MAXDOCS_PARAM = "maxDocs";
59 protected static final String HITS_PER_PAGE_PARAM = "hitsPerPage";
60 protected static final String START_PAGE_PARAM = "startPage";
61
62 protected static final String INDEX_SUBCOLLECTION_ELEM = "indexSubcollection";
63 protected static final String INDEX_LANGUAGE_ELEM = "indexLanguage";
64
65
66 // some other common params that may be used
67 protected static final String CASE_PARAM = "case";
68 protected static final String STEM_PARAM = "stem";
69 protected static final String ACCENT_PARAM="accent";
70
71 protected static final String BOOLEAN_PARAM_ON = "1";
72 protected static final String BOOLEAN_PARAM_OFF = "0";
73 protected static final String MATCH_PARAM = "matchMode";
74 protected static final String MATCH_PARAM_ALL = "all";
75 protected static final String MATCH_PARAM_SOME = "some";
76
77 /** can more than one index be searched at the smae time? */
78 protected boolean does_multi_index_search = false;
79 /** does this service support paging of results? */
80 protected boolean does_paging = false;
81 /** does this service support asking for a subset of results? */
82 protected boolean does_chunking = false;
83 /** the default document type - use if all documents are the same type
84 */
85 protected String default_document_type = null;
86 /** the default index, or comma separated list if more than one is
87 * the default (with start and end commas, eg ,TI,SU,).
88 * Should be set by configure()
89 */
90 protected String default_index = "";
91
92 protected String default_index_subcollection = "";
93
94 protected String default_index_language = "";
95
96 public AbstractSearch()
97 {
98 }
99
100 /** sets up the short service info for TextQuery. If other services
101 * will be provided, should be added in the subclass configure
102 * also looks for search format info, and document format info
103 */
104 public boolean configure(Element info, Element extra_info)
105 {
106 if (!super.configure(info, extra_info)){
107 return false;
108 }
109
110 logger.info("Configuring AbstractSearch...");
111
112 this.config_info = info;
113
114 // set up short_service_info_ - for now just has id and type. the name (lang dependent) will be added in if the list is requested.
115 Element tq_service = this.doc.createElement(GSXML.SERVICE_ELEM);
116 tq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
117 tq_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
118 this.short_service_info.appendChild(tq_service);
119
120 // add some format info to service map if there is any - look in extra info
121 // first look in buildConfig
122 Element format = (Element)GSXML.getChildByTagName(info, GSXML.FORMAT_ELEM);
123
124 if (format==null) {
125 String path = GSPath.appendLink(GSXML.SEARCH_ELEM, GSXML.FORMAT_ELEM);
126 format = (Element) GSXML.getNodeByPath(extra_info, path);
127 }
128 //Element format = (Element) GSXML.getChildByTagName(info, GSXML.FORMAT_ELEM);
129 if (format != null) {
130 this.format_info_map.put(TEXT_QUERY_SERVICE, this.doc.importNode(format, true));
131 }
132
133 // look for document display format - for documentType
134 String path = GSPath.appendLink(GSXML.DISPLAY_ELEM, GSXML.FORMAT_ELEM);
135 Element display_format = (Element)GSXML.getNodeByPath(extra_info, path);
136 if (display_format != null) {
137 // check for docType option.
138 Element doc_type_opt = GSXML.getNamedElement(display_format, "gsf:option", GSXML.NAME_ATT, "documentType");
139 if (doc_type_opt != null) {
140 String value = doc_type_opt.getAttribute(GSXML.VALUE_ATT);
141 if (!value.equals("")) {
142 this.default_document_type = value;
143 }
144 }
145 }
146
147 return true;
148 }
149
150 /** returns the description of the TextQuery service. If a subclass
151 * provides other services they need to provides their own descriptions */
152 protected Element getServiceDescription(String service, String lang, String subset)
153 {
154 if (!service.equals(TEXT_QUERY_SERVICE)) {
155 return null;
156 }
157
158 Element tq_service = this.doc.createElement(GSXML.SERVICE_ELEM);
159 tq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
160 tq_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
161 if (subset==null || subset.equals(GSXML.DISPLAY_TEXT_ELEM+GSXML.LIST_MODIFIER)) {
162 tq_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_NAME, getServiceName(TEXT_QUERY_SERVICE, lang) ));
163 tq_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_SUBMIT, getServiceSubmit(TEXT_QUERY_SERVICE, lang) ));
164 tq_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_DESCRIPTION, getServiceDescription(TEXT_QUERY_SERVICE, lang)));
165 }
166 if (subset==null || subset.equals(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER)) {
167 Element param_list = this.doc.createElement(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
168 addCustomQueryParams(param_list, lang);
169 addStandardQueryParams(param_list, lang);
170 tq_service.appendChild(param_list);
171 }
172 return tq_service;
173
174 }
175
176 // perhaps these should be changed to search down the class hierarchy for
177 // values - do can just put the info in the resource bundle to use it
178 /** returns the default name for the TextQuery service */
179 protected String getServiceName(String service_id, String lang) {
180 return getTextString(service_id+".name", lang);
181 }
182
183 /** returns the default description for the TextQuery service */
184 protected String getServiceDescription(String service_id, String lang) {
185 return getTextString(service_id+".description", lang);
186 }
187
188 /** returns the default submit button text for the TextQuery service */
189 protected String getServiceSubmit(String service_id, String lang) {
190 return getTextString(service_id+".submit", lang);
191
192 }
193 /** adds the standard query params into the service description */
194 protected void addStandardQueryParams(Element param_list, String lang)
195 {
196 if (!default_index.equals("")){
197 createParameter(INDEX_PARAM, param_list, lang);
198 }
199 if (!default_index_subcollection.equals("")){
200 createParameter(INDEX_SUBCOLLECTION_PARAM,param_list, lang);
201 }
202 if (!default_index_language.equals("")){
203 createParameter(INDEX_LANGUAGE_PARAM,param_list, lang);
204 }
205 if (does_chunking) {
206 createParameter(MAXDOCS_PARAM, param_list, lang);
207 }
208 if (does_paging) {
209 createParameter(HITS_PER_PAGE_PARAM, param_list, lang);
210 createParameter(START_PAGE_PARAM, param_list, lang);
211 }
212 createParameter(QUERY_PARAM, param_list, lang);
213 }
214
215 /** adds any service specific query params into the service
216 * default implementation: add nothing. subclasses may need to
217 * override this to add in their specific parameters
218 */
219 protected void addCustomQueryParams(Element param_list, String lang)
220 {
221 // default behaviour, do nothing
222 }
223
224 /** default implementations for the standard parameters plus some
225 * other common ones
226 * index, maxDocs, hitsPerPage, startPage, query, case, stem,
227 */
228 protected void createParameter(String name, Element param_list, String lang) {
229 Element param = null;
230 if (name.equals(QUERY_PARAM)) {
231 param = GSXML.createParameterDescription(this.doc, QUERY_PARAM, getTextString("param."+QUERY_PARAM, lang), GSXML.PARAM_TYPE_STRING, null, null, null);
232 param_list.appendChild(param);
233 } else if (name.equals(INDEX_PARAM)) {
234
235 // should we make these class fields?
236 ArrayList index_ids = new ArrayList();
237 ArrayList index_names = new ArrayList();
238 getIndexData(index_ids, index_names, lang);
239 String param_type = GSXML.PARAM_TYPE_ENUM_SINGLE;
240 if (does_multi_index_search) {
241 param_type = GSXML.PARAM_TYPE_ENUM_MULTI;
242 }
243 param = GSXML.createParameterDescription2(this.doc, INDEX_PARAM, getTextString("param."+INDEX_PARAM, lang), param_type, this.default_index, index_ids, index_names);
244 param_list.appendChild(param);
245 }
246 else if (name.equals(INDEX_SUBCOLLECTION_PARAM)){
247 Element index_sub_list = (Element)GSXML.getChildByTagName(this.config_info, INDEX_SUBCOLLECTION_ELEM+GSXML.LIST_MODIFIER);
248 if (index_sub_list == null) return;
249 ArrayList index_sub_ids = new ArrayList();
250 ArrayList index_sub_names = new ArrayList();
251 getIndexSubcollectionData(index_sub_ids, index_sub_names, lang);
252 String param_type = GSXML.PARAM_TYPE_ENUM_SINGLE;
253 if (does_multi_index_search) {
254 param_type = GSXML.PARAM_TYPE_ENUM_MULTI;
255 }
256 param = GSXML.createParameterDescription2(this.doc, INDEX_SUBCOLLECTION_PARAM, getTextString("param."+INDEX_SUBCOLLECTION_PARAM, lang), param_type, this.default_index_subcollection, index_sub_ids, index_sub_names);
257 param_list.appendChild(param);
258 }
259 else if(name.equals(INDEX_LANGUAGE_PARAM)){
260 Element index_lang_list = (Element)GSXML.getChildByTagName(this.config_info, INDEX_LANGUAGE_ELEM+GSXML.LIST_MODIFIER);
261 if (index_lang_list == null) return;
262 ArrayList index_lang_ids = new ArrayList();
263 ArrayList index_lang_names = new ArrayList();
264 getIndexLanguageData(index_lang_ids, index_lang_names, lang);
265 String param_type = GSXML.PARAM_TYPE_ENUM_SINGLE;
266 if (does_multi_index_search) {
267 param_type = GSXML.PARAM_TYPE_ENUM_MULTI;
268 }
269 param = GSXML.createParameterDescription2(this.doc, INDEX_LANGUAGE_PARAM, getTextString("param."+INDEX_LANGUAGE_PARAM, lang), param_type, this.default_index_language, index_lang_ids, index_lang_names);
270 param_list.appendChild(param);
271 }
272 else if (name.equals(MAXDOCS_PARAM) || name.equals(HITS_PER_PAGE_PARAM)) {
273 String default_val = "100";
274 if (name.equals(HITS_PER_PAGE_PARAM)) {
275 default_val = "10";
276 }
277 param = GSXML.createParameterDescription(this.doc, name, getTextString("param."+name, lang), GSXML.PARAM_TYPE_INTEGER, default_val, null, null);
278 param_list.appendChild(param);
279 } else if (name.equals(CASE_PARAM) || name.equals(STEM_PARAM) || name.equals(ACCENT_PARAM)) {
280 String[] bool_ops = {"0", "1"};
281 String[] bool_texts = {getTextString("param.boolean.off", lang),getTextString("param.boolean.on", lang)};
282 param = GSXML.createParameterDescription(this.doc, name, getTextString("param."+name, lang), GSXML.PARAM_TYPE_BOOLEAN, BOOLEAN_PARAM_ON, bool_ops, bool_texts);
283 param_list.appendChild(param);
284 } else if (name.equals(MATCH_PARAM)) {
285 String[] vals = {MATCH_PARAM_SOME, MATCH_PARAM_ALL };
286 String[] val_texts = {getTextString("param."+MATCH_PARAM+"."+MATCH_PARAM_SOME, lang), getTextString("param."+MATCH_PARAM+"."+MATCH_PARAM_ALL, lang)};
287 param = GSXML.createParameterDescription(this.doc, MATCH_PARAM, getTextString("param."+MATCH_PARAM, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, MATCH_PARAM_SOME, vals, val_texts);
288 param_list.appendChild(param);
289 } else if (name.equals(START_PAGE_PARAM)) {
290 // start page - set to 1 for the search page
291 param = GSXML.createParameterDescription(this.doc, START_PAGE_PARAM, "", GSXML.PARAM_TYPE_INVISIBLE, "1", null, null);
292 param_list.appendChild(param);
293 }
294
295
296 }
297 /** create an element to go into the search results list. A node element
298 * has the form
299 * <docNode nodeId='xxx' nodeType='leaf' docType='hierarchy' rank='0.23'/>
300 */
301 protected Element createDocNode(String node_id, String rank) {
302 Element node = this.doc.createElement(GSXML.DOC_NODE_ELEM);
303 node.setAttribute(GSXML.NODE_ID_ATT, node_id);
304 node.setAttribute(GSXML.NODE_RANK_ATT, rank);
305 String doc_type = null;
306 if (default_document_type != null) {
307 doc_type = default_document_type;
308 } else {
309 doc_type = getDocType(node_id);
310 }
311 node.setAttribute(GSXML.DOC_TYPE_ATT, doc_type);
312 String node_type = getNodeType(node_id, doc_type);
313 node.setAttribute(GSXML.NODE_TYPE_ATT, node_type);
314 return node;
315 }
316
317 /** returns the node type of the specified node.
318 should be one of
319 GSXML.NODE_TYPE_LEAF,
320 GSXML.NODE_TYPE_INTERNAL,
321 GSXML.NODE_TYPE_ROOT
322 */
323 protected String getNodeType(String node_id, String doc_type) {
324 if (doc_type.equals(GSXML.DOC_TYPE_SIMPLE)) {
325 return GSXML.NODE_TYPE_LEAF;
326 }
327
328 if (!hasParent(node_id)) {
329 return GSXML.NODE_TYPE_ROOT;
330 }
331 if (doc_type.equals(GSXML.DOC_TYPE_PAGED)) {
332 return GSXML.NODE_TYPE_LEAF;
333 }
334 if (!hasChildren(node_id)) {
335 return GSXML.NODE_TYPE_LEAF;
336 }
337 return GSXML.NODE_TYPE_INTERNAL;
338
339 }
340
341
342 /** returns the document type of the doc that the specified node
343 belongs to. should be one of
344 GSXML.DOC_TYPE_SIMPLE,
345 GSXML.DOC_TYPE_PAGED,
346 GSXML.DOC_TYPE_HIERARCHY
347 default implementation returns GSXML.DOC_TYPE_SIMPLE, over ride
348 if documents can be hierarchical
349 */
350 protected String getDocType(String node_id) {
351 return GSXML.DOC_TYPE_SIMPLE;
352 }
353
354 /** returns true if the node has child nodes
355 * default implementation returns false, over ride if documents can be
356 * hierarchical
357 */
358 protected boolean hasChildren(String node_id) {
359 return false;
360 }
361 /** returns true if the node has a parent
362 * default implementation returns false, over ride if documents can be
363 * hierarchical*/
364 protected boolean hasParent(String node_id) {
365 return false;
366 }
367
368 /** do the actual query
369 * must be implemented by subclass */
370 abstract protected Element processTextQuery(Element request);
371
372 /** get the details about the indexes available
373 * must be implemented by subclass
374 * there must be at least one index */
375 abstract protected void getIndexData(ArrayList index_ids, ArrayList index_names, String lang);
376
377 /** get the details about the indexexSubcollections available
378 * might be implemented by subclass
379 */
380 protected void getIndexSubcollectionData(ArrayList index_ids, ArrayList index_names, String lang){}
381
382 /** get the details about the indexes available
383 * might be implemented by subclass
384 */
385 protected void getIndexLanguageData(ArrayList index_ids, ArrayList index_names, String lang){}
386
387
388}
389
Note: See TracBrowser for help on using the repository browser.