source: greenstone3/trunk/src/java/org/greenstone/gsdl3/service/AbstractSearch.java@ 20064

Last change on this file since 20064 was 20064, checked in by kjdon, 15 years ago

added a comment
  • Property svn:keywords set to Author Date Id Revision
File size: 16.6 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 protected String default_max_docs = "100";
97
98 protected String default_hits_per_page = "10";
99
100 public AbstractSearch()
101 {
102 }
103
104 /** sets up the short service info for TextQuery. If other services
105 * will be provided, should be added in the subclass configure
106 * also looks for search format info, and document format info
107 */
108 public boolean configure(Element info, Element extra_info)
109 {
110 if (!super.configure(info, extra_info)){
111 return false;
112 }
113
114 logger.info("Configuring AbstractSearch...");
115
116 this.config_info = info;
117
118 // 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.
119 Element tq_service = this.doc.createElement(GSXML.SERVICE_ELEM);
120 tq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
121 tq_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
122 this.short_service_info.appendChild(tq_service);
123
124 // add some format info to service map if there is any - look in extra info
125 // first look in buildConfig
126 Element format = (Element)GSXML.getChildByTagName(info, GSXML.FORMAT_ELEM);
127
128 if (format==null) {
129 String path = GSPath.appendLink(GSXML.SEARCH_ELEM, GSXML.FORMAT_ELEM);
130
131 //note by xiao: instead of retrieving the first 'format' element inside the 'search'
132 // element, we are trying to find the real format element which has at least one
133 // 'gsf:template' child element. (extra_info is collectionConfig.xml)
134 //format = (Element) GSXML.getNodeByPath(extra_info, path);
135 Element search_elem = (Element) GSXML.getChildByTagName(extra_info, GSXML.SEARCH_ELEM);
136 NodeList format_elems = null;
137 if (search_elem != null) {
138 format_elems = search_elem.getElementsByTagName(GSXML.FORMAT_ELEM);
139 }
140 for(int i=0; i<format_elems.getLength(); i++) {
141 format = (Element)format_elems.item(i);
142 if (format.getElementsByTagName("gsf:template").getLength() != 0) {
143 break;
144 }
145 }
146 }//end of if(format==null)
147 //
148 if (format != null) {
149 this.format_info_map.put(TEXT_QUERY_SERVICE, this.doc.importNode(format, true));
150 }
151
152 // look for document display format - for documentType
153 String path = GSPath.appendLink(GSXML.DISPLAY_ELEM, GSXML.FORMAT_ELEM);
154 Element display_format = (Element)GSXML.getNodeByPath(extra_info, path);
155 if (display_format != null) {
156 // check for docType option.
157 Element doc_type_opt = GSXML.getNamedElement(display_format, "gsf:option", GSXML.NAME_ATT, "documentType");
158 if (doc_type_opt != null) {
159 String value = doc_type_opt.getAttribute(GSXML.VALUE_ATT);
160 if (!value.equals("")) {
161 this.default_document_type = value;
162 }
163 }
164 }
165
166 return true;
167 }
168
169 /** returns the description of the TextQuery service. If a subclass
170 * provides other services they need to provides their own descriptions */
171 protected Element getServiceDescription(String service, String lang, String subset)
172 {
173 if (!service.equals(TEXT_QUERY_SERVICE)) {
174 return null;
175 }
176
177 Element tq_service = this.doc.createElement(GSXML.SERVICE_ELEM);
178 tq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
179 tq_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
180 if (subset==null || subset.equals(GSXML.DISPLAY_TEXT_ELEM+GSXML.LIST_MODIFIER)) {
181 tq_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_NAME, getServiceName(TEXT_QUERY_SERVICE, lang) ));
182 tq_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_SUBMIT, getServiceSubmit(TEXT_QUERY_SERVICE, lang) ));
183 tq_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_DESCRIPTION, getServiceDescription(TEXT_QUERY_SERVICE, lang)));
184 }
185 if (subset==null || subset.equals(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER)) {
186 Element param_list = this.doc.createElement(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
187 addCustomQueryParams(param_list, lang);
188 addStandardQueryParams(param_list, lang);
189 tq_service.appendChild(param_list);
190 }
191 return tq_service;
192
193 }
194
195 // perhaps these should be changed to search down the class hierarchy for
196 // values - do can just put the info in the resource bundle to use it
197 /** returns the default name for the TextQuery service */
198 protected String getServiceName(String service_id, String lang) {
199 return getTextString(service_id+".name", lang);
200 }
201
202 /** returns the default description for the TextQuery service */
203 protected String getServiceDescription(String service_id, String lang) {
204 return getTextString(service_id+".description", lang);
205 }
206
207 /** returns the default submit button text for the TextQuery service */
208 protected String getServiceSubmit(String service_id, String lang) {
209 return getTextString(service_id+".submit", lang);
210
211 }
212 /** adds the standard query params into the service description */
213 protected void addStandardQueryParams(Element param_list, String lang)
214 {
215 // this test is not so good. here we are using absence of default index
216 // to determine whether we have indexes or not. But in other places,
217 // absence of default index just means to use the first one as default.
218 if (!default_index.equals("")){
219 createParameter(INDEX_PARAM, param_list, lang);
220 }
221 if (!default_index_subcollection.equals("")){
222 createParameter(INDEX_SUBCOLLECTION_PARAM,param_list, lang);
223 }
224 if (!default_index_language.equals("")){
225 createParameter(INDEX_LANGUAGE_PARAM,param_list, lang);
226 }
227 if (does_chunking) {
228 createParameter(MAXDOCS_PARAM, param_list, lang);
229 }
230 if (does_paging) {
231 createParameter(HITS_PER_PAGE_PARAM, param_list, lang);
232 createParameter(START_PAGE_PARAM, param_list, lang);
233 }
234 createParameter(QUERY_PARAM, param_list, lang);
235 }
236
237 /** adds any service specific query params into the service
238 * default implementation: add nothing. subclasses may need to
239 * override this to add in their specific parameters
240 */
241 protected void addCustomQueryParams(Element param_list, String lang)
242 {
243 // default behaviour, do nothing
244 }
245
246 /** default implementations for the standard parameters plus some
247 * other common ones
248 * index, maxDocs, hitsPerPage, startPage, query, case, stem,
249 */
250 protected void createParameter(String name, Element param_list, String lang) {
251 Element param = null;
252 if (name.equals(QUERY_PARAM)) {
253 param = GSXML.createParameterDescription(this.doc, QUERY_PARAM, getTextString("param."+QUERY_PARAM, lang), GSXML.PARAM_TYPE_STRING, null, null, null);
254 param_list.appendChild(param);
255 } else if (name.equals(INDEX_PARAM)) {
256
257 // should we make these class fields?
258 ArrayList index_ids = new ArrayList();
259 ArrayList index_names = new ArrayList();
260 getIndexData(index_ids, index_names, lang);
261 String param_type = GSXML.PARAM_TYPE_ENUM_SINGLE;
262 if (does_multi_index_search) {
263 param_type = GSXML.PARAM_TYPE_ENUM_MULTI;
264 }
265 param = GSXML.createParameterDescription2(this.doc, INDEX_PARAM, getTextString("param."+INDEX_PARAM, lang), param_type, this.default_index, index_ids, index_names);
266 param_list.appendChild(param);
267 }
268 else if (name.equals(INDEX_SUBCOLLECTION_PARAM)){
269 Element index_sub_list = (Element)GSXML.getChildByTagName(this.config_info, INDEX_SUBCOLLECTION_ELEM+GSXML.LIST_MODIFIER);
270 if (index_sub_list == null) return;
271 ArrayList index_sub_ids = new ArrayList();
272 ArrayList index_sub_names = new ArrayList();
273 getIndexSubcollectionData(index_sub_ids, index_sub_names, lang);
274 String param_type = GSXML.PARAM_TYPE_ENUM_SINGLE;
275 if (does_multi_index_search) {
276 param_type = GSXML.PARAM_TYPE_ENUM_MULTI;
277 }
278 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);
279 param_list.appendChild(param);
280 }
281 else if(name.equals(INDEX_LANGUAGE_PARAM)){
282 Element index_lang_list = (Element)GSXML.getChildByTagName(this.config_info, INDEX_LANGUAGE_ELEM+GSXML.LIST_MODIFIER);
283 if (index_lang_list == null) return;
284 ArrayList index_lang_ids = new ArrayList();
285 ArrayList index_lang_names = new ArrayList();
286 getIndexLanguageData(index_lang_ids, index_lang_names, lang);
287 String param_type = GSXML.PARAM_TYPE_ENUM_SINGLE;
288 if (does_multi_index_search) {
289 param_type = GSXML.PARAM_TYPE_ENUM_MULTI;
290 }
291 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);
292 param_list.appendChild(param);
293 }
294 else if (name.equals(MAXDOCS_PARAM)) {
295 param = GSXML.createParameterDescription(this.doc, name, getTextString("param."+name, lang), GSXML.PARAM_TYPE_INTEGER, this.default_max_docs, null, null);
296 param_list.appendChild(param);
297 }
298 else if(name.equals(HITS_PER_PAGE_PARAM)){
299 param = GSXML.createParameterDescription(this.doc, name, getTextString("param."+name, lang), GSXML.PARAM_TYPE_INTEGER, this.default_hits_per_page, null, null);
300 param_list.appendChild(param);
301 }
302 else if (name.equals(CASE_PARAM) || name.equals(STEM_PARAM) || name.equals(ACCENT_PARAM)) {
303 String[] bool_ops = {"0", "1"};
304 String[] bool_texts = {getTextString("param.boolean.off", lang),getTextString("param.boolean.on", lang)};
305 param = GSXML.createParameterDescription(this.doc, name, getTextString("param."+name, lang), GSXML.PARAM_TYPE_BOOLEAN, BOOLEAN_PARAM_ON, bool_ops, bool_texts);
306 param_list.appendChild(param);
307 } else if (name.equals(MATCH_PARAM)) {
308 String[] vals = {MATCH_PARAM_SOME, MATCH_PARAM_ALL };
309 String[] val_texts = {getTextString("param."+MATCH_PARAM+"."+MATCH_PARAM_SOME, lang), getTextString("param."+MATCH_PARAM+"."+MATCH_PARAM_ALL, lang)};
310 param = GSXML.createParameterDescription(this.doc, MATCH_PARAM, getTextString("param."+MATCH_PARAM, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, MATCH_PARAM_SOME, vals, val_texts);
311 param_list.appendChild(param);
312 } else if (name.equals(START_PAGE_PARAM)) {
313 // start page - set to 1 for the search page
314 param = GSXML.createParameterDescription(this.doc, START_PAGE_PARAM, "", GSXML.PARAM_TYPE_INVISIBLE, "1", null, null);
315 param_list.appendChild(param);
316 }
317
318
319 }
320 /** create an element to go into the search results list. A node element
321 * has the form
322 * <docNode nodeId='xxx' nodeType='leaf' docType='hierarchy' rank='0.23'/>
323 */
324 protected Element createDocNode(String node_id, String rank) {
325 Element node = this.doc.createElement(GSXML.DOC_NODE_ELEM);
326 node.setAttribute(GSXML.NODE_ID_ATT, node_id);
327 node.setAttribute(GSXML.NODE_RANK_ATT, rank);
328 String doc_type = null;
329 if (default_document_type != null) {
330 doc_type = default_document_type;
331 } else {
332 doc_type = getDocType(node_id);
333 }
334 node.setAttribute(GSXML.DOC_TYPE_ATT, doc_type);
335 String node_type = getNodeType(node_id, doc_type);
336 node.setAttribute(GSXML.NODE_TYPE_ATT, node_type);
337 return node;
338 }
339
340 /** returns the node type of the specified node.
341 should be one of
342 GSXML.NODE_TYPE_LEAF,
343 GSXML.NODE_TYPE_INTERNAL,
344 GSXML.NODE_TYPE_ROOT
345 */
346 protected String getNodeType(String node_id, String doc_type) {
347 if (doc_type.equals(GSXML.DOC_TYPE_SIMPLE)) {
348 return GSXML.NODE_TYPE_LEAF;
349 }
350
351 if (!hasParent(node_id)) {
352 return GSXML.NODE_TYPE_ROOT;
353 }
354 if (doc_type.equals(GSXML.DOC_TYPE_PAGED)) {
355 return GSXML.NODE_TYPE_LEAF;
356 }
357 if (!hasChildren(node_id)) {
358 return GSXML.NODE_TYPE_LEAF;
359 }
360 return GSXML.NODE_TYPE_INTERNAL;
361
362 }
363
364
365 /** returns the document type of the doc that the specified node
366 belongs to. should be one of
367 GSXML.DOC_TYPE_SIMPLE,
368 GSXML.DOC_TYPE_PAGED,
369 GSXML.DOC_TYPE_HIERARCHY
370 default implementation returns GSXML.DOC_TYPE_SIMPLE, over ride
371 if documents can be hierarchical
372 */
373 protected String getDocType(String node_id) {
374 return GSXML.DOC_TYPE_SIMPLE;
375 }
376
377 /** returns true if the node has child nodes
378 * default implementation returns false, over ride if documents can be
379 * hierarchical
380 */
381 protected boolean hasChildren(String node_id) {
382 return false;
383 }
384 /** returns true if the node has a parent
385 * default implementation returns false, over ride if documents can be
386 * hierarchical*/
387 protected boolean hasParent(String node_id) {
388 return false;
389 }
390
391 /** do the actual query
392 * must be implemented by subclass */
393 abstract protected Element processTextQuery(Element request);
394
395 /** get the details about the indexes available
396 * must be implemented by subclass
397 * there must be at least one index */
398 abstract protected void getIndexData(ArrayList index_ids, ArrayList index_names, String lang);
399
400 /** get the details about the indexexSubcollections available
401 * might be implemented by subclass
402 */
403 protected void getIndexSubcollectionData(ArrayList index_ids, ArrayList index_names, String lang){}
404
405 /** get the details about the indexes available
406 * might be implemented by subclass
407 */
408 protected void getIndexLanguageData(ArrayList index_ids, ArrayList index_names, String lang){}
409
410
411}
412
Note: See TracBrowser for help on using the repository browser.