source: trunk/gsdl3/src/java/org/greenstone/gsdl3/service/AbstractSearch.java@ 13270

Last change on this file since 13270 was 13270, checked in by shaoqun, 17 years ago

replace Category class which is deprecated with Logger class
  • Property svn:keywords set to Author Date Id Revision
File size: 13.1 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 MAXDOCS_PARAM = "maxDocs";
57 protected static final String HITS_PER_PAGE_PARAM = "hitsPerPage";
58 protected static final String START_PAGE_PARAM = "startPage";
59
60 // some other common params that may be used
61 protected static final String CASE_PARAM = "case";
62 protected static final String STEM_PARAM = "stem";
63 protected static final String ACCENT_PARAM="accent";
64
65 protected static final String BOOLEAN_PARAM_ON = "1";
66 protected static final String BOOLEAN_PARAM_OFF = "0";
67 protected static final String MATCH_PARAM = "matchMode";
68 protected static final String MATCH_PARAM_ALL = "all";
69 protected static final String MATCH_PARAM_SOME = "some";
70
71 /** can more than one index be searched at the smae time? */
72 protected boolean does_multi_index_search = false;
73 /** does this service support paging of results? */
74 protected boolean does_paging = false;
75 /** does this service support asking for a subset of results? */
76 protected boolean does_chunking = false;
77 /** the default document type - use if all documents are the same type
78 */
79 protected String default_document_type = null;
80 /** the default index, or comma separated list if more than one is
81 * the default (with start and end commas, eg ,TI,SU,).
82 * Should be set by configure()
83 */
84 protected String default_index = "";
85
86 public AbstractSearch()
87 {
88 }
89
90 /** sets up the short service info for TextQuery. If other services
91 * will be provided, should be added in the subclass configure
92 * also looks for search format info, and document format info
93 */
94 public boolean configure(Element info, Element extra_info)
95 {
96 if (!super.configure(info, extra_info)){
97 return false;
98 }
99
100 logger.info("Configuring AbstractSearch...");
101
102 this.config_info = info;
103
104 // 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.
105 Element tq_service = this.doc.createElement(GSXML.SERVICE_ELEM);
106 tq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
107 tq_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
108 this.short_service_info.appendChild(tq_service);
109
110 // add some format info to service map if there is any - look in extra info
111 // first look in buildConfig
112 Element format = (Element)GSXML.getChildByTagName(info, GSXML.FORMAT_ELEM);
113
114 if (format==null) {
115 String path = GSPath.appendLink(GSXML.SEARCH_ELEM, GSXML.FORMAT_ELEM);
116 format = (Element) GSXML.getNodeByPath(extra_info, path);
117 }
118 //Element format = (Element) GSXML.getChildByTagName(info, GSXML.FORMAT_ELEM);
119 if (format != null) {
120 this.format_info_map.put(TEXT_QUERY_SERVICE, this.doc.importNode(format, true));
121 }
122
123 // look for document display format - for documentType
124 String path = GSPath.appendLink(GSXML.DISPLAY_ELEM, GSXML.FORMAT_ELEM);
125 Element display_format = (Element)GSXML.getNodeByPath(extra_info, path);
126 if (display_format != null) {
127 // check for docType option.
128 Element doc_type_opt = GSXML.getNamedElement(display_format, "gsf:option", GSXML.NAME_ATT, "documentType");
129 if (doc_type_opt != null) {
130 String value = doc_type_opt.getAttribute(GSXML.VALUE_ATT);
131 if (!value.equals("")) {
132 this.default_document_type = value;
133 }
134 }
135 }
136
137 return true;
138 }
139
140 /** returns the description of the TextQuery service. If a subclass
141 * provides other services they need to provides their own descriptions */
142 protected Element getServiceDescription(String service, String lang, String subset)
143 {
144 if (!service.equals(TEXT_QUERY_SERVICE)) {
145 return null;
146 }
147
148 Element tq_service = this.doc.createElement(GSXML.SERVICE_ELEM);
149 tq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
150 tq_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
151 if (subset==null || subset.equals(GSXML.DISPLAY_TEXT_ELEM+GSXML.LIST_MODIFIER)) {
152 tq_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_NAME, getServiceName(TEXT_QUERY_SERVICE, lang) ));
153 tq_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_SUBMIT, getServiceSubmit(TEXT_QUERY_SERVICE, lang) ));
154 tq_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_DESCRIPTION, getServiceDescription(TEXT_QUERY_SERVICE, lang)));
155 }
156 if (subset==null || subset.equals(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER)) {
157 Element param_list = this.doc.createElement(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
158 addCustomQueryParams(param_list, lang);
159 addStandardQueryParams(param_list, lang);
160 tq_service.appendChild(param_list);
161 }
162 return tq_service;
163
164 }
165
166 // perhaps these should be changed to search down the class hierarchy for
167 // values - do can just put the info in the resource bundle to use it
168 /** returns the default name for the TextQuery service */
169 protected String getServiceName(String service_id, String lang) {
170 return getTextString(service_id+".name", lang);
171 }
172
173 /** returns the default description for the TextQuery service */
174 protected String getServiceDescription(String service_id, String lang) {
175 return getTextString(service_id+".description", lang);
176 }
177
178 /** returns the default submit button text for the TextQuery service */
179 protected String getServiceSubmit(String service_id, String lang) {
180 return getTextString(service_id+".submit", lang);
181
182 }
183 /** adds the standard query params into the service description */
184 protected void addStandardQueryParams(Element param_list, String lang)
185 {
186 createParameter(INDEX_PARAM, param_list, lang);
187 if (does_chunking) {
188 createParameter(MAXDOCS_PARAM, param_list, lang);
189 }
190 if (does_paging) {
191 createParameter(HITS_PER_PAGE_PARAM, param_list, lang);
192 createParameter(START_PAGE_PARAM, param_list, lang);
193 }
194 createParameter(QUERY_PARAM, param_list, lang);
195 }
196
197 /** adds any service specific query params into the service
198 * default implementation: add nothing. subclasses may need to
199 * override this to add in their specific parameters
200 */
201 protected void addCustomQueryParams(Element param_list, String lang)
202 {
203 // default behaviour, do nothing
204 }
205
206 /** default implementations for the standard parameters plus some
207 * other common ones
208 * index, maxDocs, hitsPerPage, startPage, query, case, stem,
209 */
210 protected void createParameter(String name, Element param_list, String lang) {
211 Element param = null;
212 if (name.equals(QUERY_PARAM)) {
213 param = GSXML.createParameterDescription(this.doc, QUERY_PARAM, getTextString("param."+QUERY_PARAM, lang), GSXML.PARAM_TYPE_STRING, null, null, null);
214 param_list.appendChild(param);
215 } else if (name.equals(INDEX_PARAM)) {
216
217 // should we make these class fields?
218 ArrayList index_ids = new ArrayList();
219 ArrayList index_names = new ArrayList();
220 getIndexData(index_ids, index_names, lang);
221 String param_type = GSXML.PARAM_TYPE_ENUM_SINGLE;
222 if (does_multi_index_search) {
223 param_type = GSXML.PARAM_TYPE_ENUM_MULTI;
224 }
225 param = GSXML.createParameterDescription2(this.doc, INDEX_PARAM, getTextString("param."+INDEX_PARAM, lang), param_type, this.default_index, index_ids, index_names);
226 param_list.appendChild(param);
227 } else if (name.equals(MAXDOCS_PARAM)) {
228 param = GSXML.createParameterDescription(this.doc, MAXDOCS_PARAM, getTextString("param."+MAXDOCS_PARAM, lang), GSXML.PARAM_TYPE_INTEGER, "10", null, null);
229 param_list.appendChild(param);
230 } else if (name.equals(CASE_PARAM) || name.equals(STEM_PARAM) || name.equals(ACCENT_PARAM)) {
231 String[] bool_ops = {"0", "1"};
232 String[] bool_texts = {getTextString("param.boolean.off", lang),getTextString("param.boolean.on", lang)};
233 param = GSXML.createParameterDescription(this.doc, name, getTextString("param."+name, lang), GSXML.PARAM_TYPE_BOOLEAN, BOOLEAN_PARAM_ON, bool_ops, bool_texts);
234 param_list.appendChild(param);
235 } else if (name.equals(MATCH_PARAM)) {
236 String[] vals = {MATCH_PARAM_SOME, MATCH_PARAM_ALL };
237 String[] val_texts = {getTextString("param."+MATCH_PARAM+"."+MATCH_PARAM_SOME, lang), getTextString("param."+MATCH_PARAM+"."+MATCH_PARAM_ALL, lang)};
238 param = GSXML.createParameterDescription(this.doc, MATCH_PARAM, getTextString("param."+MATCH_PARAM, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, MATCH_PARAM_SOME, vals, val_texts);
239 param_list.appendChild(param);
240 } else if (name.equals(HITS_PER_PAGE_PARAM)) {
241 // hits per page
242 String [] hits_options = {"10", "30", "50"};
243 param = GSXML.createParameterDescription(this.doc, HITS_PER_PAGE_PARAM, getTextString("param."+HITS_PER_PAGE_PARAM, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, "10", hits_options, hits_options);
244 param_list.appendChild(param);
245 } else if (name.equals(START_PAGE_PARAM)) {
246 // start page - set to 1 for the search page
247 param = GSXML.createParameterDescription(this.doc, START_PAGE_PARAM, "", GSXML.PARAM_TYPE_INVISIBLE, "1", null, null);
248 param_list.appendChild(param);
249 }
250
251
252 }
253 /** create an element to go into the search results list. A node element
254 * has the form
255 * <docNode nodeId='xxx' nodeType='leaf' docType='hierarchy' rank='0.23'/>
256 */
257 protected Element createDocNode(String node_id, String rank) {
258 Element node = this.doc.createElement(GSXML.DOC_NODE_ELEM);
259 node.setAttribute(GSXML.NODE_ID_ATT, node_id);
260 node.setAttribute(GSXML.NODE_RANK_ATT, rank);
261 String doc_type = null;
262 if (default_document_type != null) {
263 doc_type = default_document_type;
264 } else {
265 doc_type = getDocType(node_id);
266 }
267 node.setAttribute(GSXML.DOC_TYPE_ATT, doc_type);
268 String node_type = getNodeType(node_id, doc_type);
269 node.setAttribute(GSXML.NODE_TYPE_ATT, node_type);
270 return node;
271 }
272
273 /** returns the node type of the specified node.
274 should be one of
275 GSXML.NODE_TYPE_LEAF,
276 GSXML.NODE_TYPE_INTERNAL,
277 GSXML.NODE_TYPE_ROOT
278 */
279 protected String getNodeType(String node_id, String doc_type) {
280 if (doc_type.equals(GSXML.DOC_TYPE_SIMPLE)) {
281 return GSXML.NODE_TYPE_LEAF;
282 }
283
284 if (!hasParent(node_id)) {
285 return GSXML.NODE_TYPE_ROOT;
286 }
287 if (doc_type.equals(GSXML.DOC_TYPE_PAGED)) {
288 return GSXML.NODE_TYPE_LEAF;
289 }
290 if (!hasChildren(node_id)) {
291 return GSXML.NODE_TYPE_LEAF;
292 }
293 return GSXML.NODE_TYPE_INTERNAL;
294
295 }
296
297
298 /** returns the document type of the doc that the specified node
299 belongs to. should be one of
300 GSXML.DOC_TYPE_SIMPLE,
301 GSXML.DOC_TYPE_PAGED,
302 GSXML.DOC_TYPE_HIERARCHY
303 default implementation returns GSXML.DOC_TYPE_SIMPLE, over ride
304 if documents can be hierarchical
305 */
306 protected String getDocType(String node_id) {
307 return GSXML.DOC_TYPE_SIMPLE;
308 }
309
310 /** returns true if the node has child nodes
311 * default implementation returns false, over ride if documents can be
312 * hierarchical
313 */
314 protected boolean hasChildren(String node_id) {
315 return false;
316 }
317 /** returns true if the node has a parent
318 * default implementation returns false, over ride if documents can be
319 * hierarchical*/
320 protected boolean hasParent(String node_id) {
321 return false;
322 }
323
324 /** do the actual query
325 * must be implemented by subclass */
326 abstract protected Element processTextQuery(Element request);
327
328 /** get the details about the indexes available
329 * must be implemented by subclass
330 * there must be at least one index */
331 abstract protected void getIndexData(ArrayList index_ids, ArrayList index_names, String lang);
332
333}
334
Note: See TracBrowser for help on using the repository browser.