1 | /*
|
---|
2 | * GS2MGSearch.java
|
---|
3 | * Copyright (C) 2002 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 | */
|
---|
19 | package org.greenstone.gsdl3.service;
|
---|
20 |
|
---|
21 | // greenstone classes
|
---|
22 | import org.greenstone.mg.*;
|
---|
23 | import org.greenstone.gdbm.*;
|
---|
24 | import org.greenstone.gsdl3.util.*;
|
---|
25 |
|
---|
26 | // xml classes
|
---|
27 | import org.w3c.dom.Document;
|
---|
28 | import org.w3c.dom.Element;
|
---|
29 | import org.w3c.dom.Node;
|
---|
30 | import org.w3c.dom.NodeList;
|
---|
31 | import org.w3c.dom.Text;
|
---|
32 |
|
---|
33 | // general java classes
|
---|
34 | import java.io.File;
|
---|
35 | import java.util.HashMap;
|
---|
36 | import java.util.Iterator;
|
---|
37 | import java.util.Map;
|
---|
38 | import java.util.Set;
|
---|
39 | import java.util.Vector;
|
---|
40 |
|
---|
41 | /**
|
---|
42 | *
|
---|
43 | * @author <a href="mailto:[email protected]">Katherine Don</a>
|
---|
44 | * @version $Revision: 3800 $
|
---|
45 | */
|
---|
46 |
|
---|
47 | public class GS2MGSearch
|
---|
48 | extends ServiceRack {
|
---|
49 |
|
---|
50 | // the services on offer
|
---|
51 | // these strings must match what is found in the properties file
|
---|
52 | private static final String TEXT_QUERY_SERVICE = "TextQuery";
|
---|
53 |
|
---|
54 | // params used
|
---|
55 | private static final String INDEX_PARAM = "index";
|
---|
56 | private static final String CASE_PARAM = "case";
|
---|
57 | private static final String STEM_PARAM = "stem";
|
---|
58 | private static final String MATCH_PARAM = "matchMode";
|
---|
59 | private static final String MATCH_PARAM_ALL = "all";
|
---|
60 | private static final String MATCH_PARAM_SOME = "some";
|
---|
61 | private static final String RANK_PARAM = "sortBy";
|
---|
62 | private static final String RANK_PARAM_RANK = "rank";
|
---|
63 | private static final String RANK_PARAM_NONE = "natural";
|
---|
64 | private static final String MAXDOCS_PARAM = "maxDocs";
|
---|
65 | private static final String BOOLEAN_PARAM_ON = "1";
|
---|
66 | private static final String BOOLEAN_PARAM_OFF = "0";
|
---|
67 | private static final String QUERY_PARAM = "query";
|
---|
68 |
|
---|
69 | // elements used in the config file that are specific to this class
|
---|
70 | private static final String DEFAULT_INDEX_ELEM = "defaultIndex";
|
---|
71 | private static final String INDEX_ELEM = "index";
|
---|
72 |
|
---|
73 | private static final String EQUIV_TERM_ELEM = "equivTerm";
|
---|
74 |
|
---|
75 | private static final String STEM_ATT = "stem";
|
---|
76 | private static final String NUM_DOCS_MATCH_ATT = "numDocsMatch";
|
---|
77 | private static final String FREQ_ATT = "freq";
|
---|
78 |
|
---|
79 | private MGWrapper mg_src_ = null;
|
---|
80 | private GDBMWrapper gdbm_src_ = null;
|
---|
81 |
|
---|
82 | private String default_index_ = null;
|
---|
83 |
|
---|
84 | private Element config_info_ = null;
|
---|
85 |
|
---|
86 |
|
---|
87 | /** constructor */
|
---|
88 | public GS2MGSearch() {
|
---|
89 | mg_src_ = new MGWrapper();
|
---|
90 | gdbm_src_ = new GDBMWrapper();
|
---|
91 | }
|
---|
92 |
|
---|
93 |
|
---|
94 | /** configure this service */
|
---|
95 | public boolean configure(Element info)
|
---|
96 | {
|
---|
97 | System.out.println("configuring GS2MGSearch");
|
---|
98 | config_info_ = info;
|
---|
99 |
|
---|
100 | // get the default index out of <defaultIndex> (buildConfig.xml)
|
---|
101 | Element def = (Element) GSXML.getChildByTagName(info, DEFAULT_INDEX_ELEM);
|
---|
102 | if (def != null) {
|
---|
103 | default_index_ = def.getAttribute(GSXML.NAME_ATT);
|
---|
104 | }
|
---|
105 | if (default_index_ == null || default_index_.equals("")) {
|
---|
106 | System.err.println("Error: default index not specified!");
|
---|
107 | return false;
|
---|
108 | }
|
---|
109 |
|
---|
110 | Element e = null;
|
---|
111 | // these entries should reflect the build config file - some services may not be available depending on how the colleciton was built.
|
---|
112 | // set up short_service_info_ - for now just has name and type
|
---|
113 | e = doc_.createElement(GSXML.SERVICE_ELEM);
|
---|
114 | e.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
|
---|
115 | e.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
|
---|
116 | short_service_info_.appendChild(e);
|
---|
117 |
|
---|
118 | // set up service_info_map_ - for now, just has the same elements as above
|
---|
119 | // should have full details about each service incl params lists etc.
|
---|
120 | e = doc_.createElement(GSXML.SERVICE_ELEM);
|
---|
121 | e.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
|
---|
122 | e.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
|
---|
123 | Element param_list = doc_.createElement(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
|
---|
124 | createTextQueryParamList(param_list, false, null);
|
---|
125 | e.appendChild(param_list);
|
---|
126 | service_info_map_.put(TEXT_QUERY_SERVICE, e);
|
---|
127 |
|
---|
128 | // Open GDBM database for querying
|
---|
129 | String gdbm_db_file = GSFile.GDBMDatabaseFile(site_home_, cluster_name_);
|
---|
130 | if (gdbm_src_.openDatabase(gdbm_db_file, GDBMWrapper.READER)) {
|
---|
131 | return true;
|
---|
132 | }
|
---|
133 | else {
|
---|
134 | System.err.println("Error: Could not open gdbm database!");
|
---|
135 | return false;
|
---|
136 | }
|
---|
137 | }
|
---|
138 |
|
---|
139 |
|
---|
140 | /** creates a new param element and adds it to the param list */
|
---|
141 | protected void createParameter(String name, Element param_list, boolean display, String lang)
|
---|
142 | {
|
---|
143 | Element param=null;
|
---|
144 |
|
---|
145 | if (name.equals(INDEX_PARAM)) {
|
---|
146 | // the index info - read from config file
|
---|
147 | Element index_list = (Element)GSXML.getChildByTagName(config_info_, INDEX_ELEM+GSXML.LIST_MODIFIER);
|
---|
148 | NodeList indexes = index_list.getElementsByTagName(INDEX_ELEM);
|
---|
149 | int len = indexes.getLength();
|
---|
150 | if (len > 1) { // add index param to list only if more than one index specified
|
---|
151 | String [] inds = new String[len];
|
---|
152 | for (int i=0; i<len; i++) {
|
---|
153 | inds[i] = ((Element)indexes.item(i)).getAttribute(GSXML.NAME_ATT);
|
---|
154 | }
|
---|
155 | if (display) {
|
---|
156 | // use the same index names for now - should get these out of the config info
|
---|
157 | param = GSXML.createParameterDisplay(doc_, INDEX_PARAM, getTextString("param."+INDEX_PARAM, lang), inds, inds);
|
---|
158 | } else {
|
---|
159 | param = GSXML.createParameter(doc_, INDEX_PARAM, GSXML.PARAM_TYPE_ENUM_SINGLE, default_index_, inds);
|
---|
160 | }
|
---|
161 | }
|
---|
162 | }
|
---|
163 | else if (name.equals(CASE_PARAM)) {
|
---|
164 | if (display) {
|
---|
165 | String[] bool_ops = {"0", "1"};
|
---|
166 | String[] bool_texts = {getTextString("param.boolean.off", lang),getTextString("param.boolean.on", lang)};
|
---|
167 | param = GSXML.createParameterDisplay(doc_, CASE_PARAM, getTextString("param."+CASE_PARAM, lang), bool_ops, bool_texts);
|
---|
168 | } else {
|
---|
169 | param = GSXML.createParameter(doc_, CASE_PARAM, GSXML.PARAM_TYPE_BOOLEAN, BOOLEAN_PARAM_ON, null);
|
---|
170 | }
|
---|
171 | }
|
---|
172 | else if (name.equals(STEM_PARAM)) {
|
---|
173 | if (display) {
|
---|
174 | String[] bool_ops = {"0", "1"};
|
---|
175 | String[] bool_texts = {getTextString("param.boolean.off", lang),getTextString("param.boolean.on", lang)};
|
---|
176 | param = GSXML.createParameterDisplay(doc_, STEM_PARAM, getTextString("param."+STEM_PARAM, lang), bool_ops, bool_texts);
|
---|
177 | } else {
|
---|
178 | param = GSXML.createParameter(doc_, STEM_PARAM, GSXML.PARAM_TYPE_BOOLEAN, BOOLEAN_PARAM_ON, null);
|
---|
179 | }
|
---|
180 | }
|
---|
181 | else if (name.equals(MATCH_PARAM)) {
|
---|
182 | String[] vals = {MATCH_PARAM_ALL, MATCH_PARAM_SOME};
|
---|
183 | if (display) {
|
---|
184 | String[] val_texts = {getTextString("param."+MATCH_PARAM+"."+MATCH_PARAM_ALL, lang),getTextString("param."+MATCH_PARAM+"."+MATCH_PARAM_SOME, lang)};
|
---|
185 |
|
---|
186 | param = GSXML.createParameterDisplay(doc_, MATCH_PARAM, getTextString("param."+MATCH_PARAM, lang), vals, val_texts);
|
---|
187 | } else {
|
---|
188 | param = GSXML.createParameter(doc_, MATCH_PARAM, GSXML.PARAM_TYPE_ENUM_SINGLE, MATCH_PARAM_ALL, vals);
|
---|
189 | }
|
---|
190 | }
|
---|
191 | else if (name.equals(MAXDOCS_PARAM)) {
|
---|
192 | if (display) {
|
---|
193 | param = GSXML.createParameterDisplay(doc_, MAXDOCS_PARAM, getTextString("param."+MAXDOCS_PARAM, lang), null, null);
|
---|
194 | } else {
|
---|
195 | param = GSXML.createParameter(doc_, MAXDOCS_PARAM, GSXML.PARAM_TYPE_INTEGER, "10", null);
|
---|
196 | }
|
---|
197 | }
|
---|
198 | else if (name.equals(QUERY_PARAM)) {
|
---|
199 | if (display) {
|
---|
200 | param = GSXML.createParameterDisplay(doc_, QUERY_PARAM, getTextString("param."+QUERY_PARAM, lang), null, null);
|
---|
201 | } else {
|
---|
202 | param = GSXML.createParameter(doc_, QUERY_PARAM, GSXML.PARAM_TYPE_STRING, null, null);
|
---|
203 | }
|
---|
204 | }
|
---|
205 |
|
---|
206 | // add the param to the list
|
---|
207 | if (param != null) {
|
---|
208 | param_list.appendChild(param);
|
---|
209 | }
|
---|
210 | }
|
---|
211 |
|
---|
212 |
|
---|
213 | /** this creates all teh params and appends them to param_list.
|
---|
214 | * if display=true it creates the text strings version
|
---|
215 | * otherwise it creates the description version
|
---|
216 | */
|
---|
217 | protected boolean createTextQueryParamList(Element param_list, boolean display, String lang)
|
---|
218 | {
|
---|
219 | // the order they are specified here is the order they appear on
|
---|
220 | // the query form
|
---|
221 | createParameter(INDEX_PARAM, param_list, display, lang);
|
---|
222 | createParameter(CASE_PARAM, param_list, display, lang);
|
---|
223 | createParameter(STEM_PARAM, param_list, display, lang);
|
---|
224 | createParameter(MATCH_PARAM, param_list, display, lang);
|
---|
225 | createParameter(MAXDOCS_PARAM, param_list, display, lang);
|
---|
226 | createParameter(QUERY_PARAM, param_list, display, lang);
|
---|
227 | return true;
|
---|
228 | }
|
---|
229 |
|
---|
230 |
|
---|
231 | /** creates a display element containing all the text strings needed to display the service page, in the language specified */
|
---|
232 | protected Element createServiceDisplay(String service, String lang) {
|
---|
233 | Element display = doc_.createElement(GSXML.DISPLAY_ELEM);
|
---|
234 | display.appendChild(GSXML.createTextElement(doc_, GSXML.DISPLAY_NAME_ELEM,
|
---|
235 | getTextString(service+".name", lang)));
|
---|
236 | display.appendChild(GSXML.createTextElement(doc_, GSXML.DISPLAY_SUBMIT_ELEM,
|
---|
237 | getTextString(service+".submit", lang)));
|
---|
238 |
|
---|
239 | // now need to add in the params
|
---|
240 | if (service.equals(TEXT_QUERY_SERVICE)) {
|
---|
241 | createTextQueryParamList(display, true, lang);
|
---|
242 | }
|
---|
243 |
|
---|
244 | return display;
|
---|
245 | }
|
---|
246 |
|
---|
247 |
|
---|
248 | /** process a text query */
|
---|
249 | protected Element processTextQuery(Element request)
|
---|
250 | {
|
---|
251 | Element result = doc_.createElement(GSXML.RESPONSE_ELEM);
|
---|
252 | String from = GSPath.appendLink(cluster_name_, TEXT_QUERY_SERVICE);
|
---|
253 | result.setAttribute(GSXML.FROM_ATT, from);
|
---|
254 | result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_QUERY);
|
---|
255 |
|
---|
256 | // get param list
|
---|
257 | Element param_list = (Element)GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
|
---|
258 | if (param_list==null) {
|
---|
259 | System.err.println("GS2MGSearch, TextQuery Error: no param list in request!");
|
---|
260 | return result; // empty result
|
---|
261 | }
|
---|
262 | HashMap params = GSXML.extractParams(param_list);
|
---|
263 | String query = (String)params.get(QUERY_PARAM);
|
---|
264 | if (query == null) {
|
---|
265 | // no query, no result
|
---|
266 | return result;
|
---|
267 | }
|
---|
268 |
|
---|
269 | String index = (String) params.get(INDEX_PARAM);
|
---|
270 | if (index == null) { // if it is not present, use the default index
|
---|
271 | index = default_index_;
|
---|
272 | }
|
---|
273 |
|
---|
274 | // now set up the mg stuff
|
---|
275 | String basedir = GSFile.collectionBaseDir(site_home_, cluster_name_) +
|
---|
276 | File.separatorChar; // Needed for MG
|
---|
277 | String textdir = GSFile.collectionTextPath(cluster_name_);
|
---|
278 | String indexpath = GSFile.collectionIndexPath(cluster_name_, index);
|
---|
279 |
|
---|
280 | // set the mg query parameters to the values the user has specified
|
---|
281 | setStandardQueryParams(params);
|
---|
282 | mg_src_.setIndex(indexpath);
|
---|
283 |
|
---|
284 | System.out.println("GS2MGSearch, query string: " + query);
|
---|
285 | mg_src_.runQuery(basedir + File.separatorChar, textdir, query);
|
---|
286 | MGQueryResult mqr = mg_src_.getQueryResult();
|
---|
287 | long totalDocs = mqr.getTotalDocs();
|
---|
288 | // System.out.println("Matching documents: " + totalDocs);
|
---|
289 |
|
---|
290 | // get the docnums out, and convert to HASH ids
|
---|
291 | Vector docs = mqr.getDocs();
|
---|
292 | if (docs.size() == 0) {
|
---|
293 | // no docs found
|
---|
294 | System.out.println("No results found...\n");
|
---|
295 | }
|
---|
296 |
|
---|
297 | // Create a metadata list to store information about the query results
|
---|
298 | Element metadata_list = GSXML.addMetaList(doc_, result);
|
---|
299 |
|
---|
300 | // Add a metadata element specifying the number of matching documents
|
---|
301 | Element num_matches_elem = doc_.createElement(GSXML.METADATA_ELEM);
|
---|
302 | num_matches_elem.setAttribute(GSXML.NAME_ATT, "numDocsMatched");
|
---|
303 | num_matches_elem.setAttribute(GSXML.VALUE_ATT, "" + totalDocs);
|
---|
304 | metadata_list.appendChild(num_matches_elem);
|
---|
305 |
|
---|
306 | // Response content: documents and terms
|
---|
307 | Element c = doc_.createElement(GSXML.CONTENT_ELEM);
|
---|
308 | result.appendChild(c);
|
---|
309 |
|
---|
310 | Element document_list = doc_.createElement(GSXML.DOCUMENT_ELEM+GSXML.LIST_MODIFIER);
|
---|
311 | c.appendChild(document_list);
|
---|
312 | // add each document
|
---|
313 | for (int d = 0; d < docs.size(); d++) {
|
---|
314 | long docnum = ((MGDocInfo) docs.elementAt(d)).num_;
|
---|
315 | String id = gdbm_src_.docnum2Oid(docnum);
|
---|
316 | System.out.println("Docnum: " + docnum + " ID: " + id);
|
---|
317 | Node no = GSXML.createDocumentElement(doc_, id);
|
---|
318 | document_list.appendChild(no);
|
---|
319 | }
|
---|
320 |
|
---|
321 | Element term_list = doc_.createElement(GSXML.TERM_ELEM+GSXML.LIST_MODIFIER);
|
---|
322 | c.appendChild(term_list);
|
---|
323 | // get the terms out, and add to query result metadata
|
---|
324 | Vector terms = mqr.getTerms();
|
---|
325 | for (int t = 0; t < terms.size(); t++) {
|
---|
326 | MGTermInfo term_info = (MGTermInfo) terms.get(t);
|
---|
327 |
|
---|
328 | String term = term_info.term_;
|
---|
329 | int stem_method = term_info.stem_method_;
|
---|
330 | Vector equiv_terms = term_info.equiv_terms_;
|
---|
331 |
|
---|
332 | Element term_elem = doc_.createElement(GSXML.TERM_ELEM);
|
---|
333 | term_elem.setAttribute(GSXML.NAME_ATT, term);
|
---|
334 | term_elem.setAttribute(STEM_ATT, "" + stem_method);
|
---|
335 |
|
---|
336 | Element equiv_term_list = doc_.createElement(EQUIV_TERM_ELEM+GSXML.LIST_MODIFIER);
|
---|
337 | term_elem.appendChild(equiv_term_list);
|
---|
338 |
|
---|
339 | long total_term_freq = 0;
|
---|
340 | for (int et = 0; et < equiv_terms.size(); et++) {
|
---|
341 | MGEquivTermInfo equiv_term_info = (MGEquivTermInfo) equiv_terms.get(et);
|
---|
342 |
|
---|
343 | Element equiv_term_elem = doc_.createElement(GSXML.TERM_ELEM);
|
---|
344 | equiv_term_elem.setAttribute(GSXML.NAME_ATT, equiv_term_info.term_);
|
---|
345 | equiv_term_elem.setAttribute(NUM_DOCS_MATCH_ATT, "" + equiv_term_info.match_docs_);
|
---|
346 | equiv_term_elem.setAttribute(FREQ_ATT, "" + equiv_term_info.term_freq_);
|
---|
347 | equiv_term_list.appendChild(equiv_term_elem);
|
---|
348 |
|
---|
349 | total_term_freq += equiv_term_info.term_freq_;
|
---|
350 | }
|
---|
351 |
|
---|
352 | term_elem.setAttribute(FREQ_ATT, "" + total_term_freq);
|
---|
353 | term_list.appendChild(term_elem);
|
---|
354 | }
|
---|
355 |
|
---|
356 | return result;
|
---|
357 | }
|
---|
358 |
|
---|
359 |
|
---|
360 | // should probably use a list rather than map
|
---|
361 | protected boolean setStandardQueryParams(HashMap params)
|
---|
362 | {
|
---|
363 | // set the default ones
|
---|
364 | mg_src_.setReturnTerms(true);
|
---|
365 | Set entries = params.entrySet();
|
---|
366 | Iterator i = entries.iterator();
|
---|
367 | while (i.hasNext()) {
|
---|
368 | Map.Entry m = (Map.Entry)i.next();
|
---|
369 | String name = (String)m.getKey();
|
---|
370 | String value = (String)m.getValue();
|
---|
371 |
|
---|
372 | if (name.equals(CASE_PARAM)) {
|
---|
373 | boolean val = (value.equals(BOOLEAN_PARAM_ON) ? true : false);
|
---|
374 | mg_src_.setCase(val);
|
---|
375 | }
|
---|
376 | else if (name.equals(STEM_PARAM)) {
|
---|
377 | boolean val = (value.equals(BOOLEAN_PARAM_ON) ? true : false);
|
---|
378 | mg_src_.setStem(val);
|
---|
379 | }
|
---|
380 | else if (name.equals(MATCH_PARAM)) {
|
---|
381 | int mode;
|
---|
382 | if (value.equals(MATCH_PARAM_ALL)) mode = 1;
|
---|
383 | else mode = 0;
|
---|
384 | mg_src_.setMatchMode(mode);
|
---|
385 | }
|
---|
386 | else if (name.equals(MAXDOCS_PARAM)) {
|
---|
387 | int docs = Integer.parseInt(value);
|
---|
388 | mg_src_.setMaxDocs(docs);
|
---|
389 | } // ignore any others
|
---|
390 | }
|
---|
391 | return true;
|
---|
392 | }
|
---|
393 | }
|
---|