1 |
|
---|
2 | /*
|
---|
3 | * GS2MGPPSearch.java
|
---|
4 | * Copyright (C) 2002 New Zealand Digital Library, http://www.nzdl.org
|
---|
5 | *
|
---|
6 | * This program is free software; you can redistribute it and/or modify
|
---|
7 | * it under the terms of the GNU General Public License as published by
|
---|
8 | * the Free Software Foundation; either version 2 of the License, or
|
---|
9 | * (at your option) any later version.
|
---|
10 | *
|
---|
11 | * This program is distributed in the hope that it will be useful,
|
---|
12 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
13 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
14 | * GNU General Public License for more details.
|
---|
15 | *
|
---|
16 | * You should have received a copy of the GNU General Public License
|
---|
17 | * along with this program; if not, write to the Free Software
|
---|
18 | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
19 | */
|
---|
20 | package org.greenstone.gsdl3.service;
|
---|
21 |
|
---|
22 |
|
---|
23 | // Greenstone classes
|
---|
24 | import org.greenstone.mgpp.*;
|
---|
25 | import org.greenstone.gsdl3.util.*;
|
---|
26 |
|
---|
27 | // XML classes
|
---|
28 | import org.w3c.dom.Element;
|
---|
29 | import org.w3c.dom.Node;
|
---|
30 | import org.w3c.dom.NodeList;
|
---|
31 |
|
---|
32 | // General Java classes
|
---|
33 | import java.io.File;
|
---|
34 | import java.util.HashMap;
|
---|
35 | import java.util.Vector;
|
---|
36 | import java.util.Set;
|
---|
37 | import java.util.Map;
|
---|
38 | import java.util.Iterator;
|
---|
39 |
|
---|
40 |
|
---|
41 | /**
|
---|
42 | * A ServiceRack class for searching in greenstone 2 MGPP collections
|
---|
43 | *
|
---|
44 | * @author <a href="mailto:[email protected]">Katherine Don</a>
|
---|
45 | * @version $Revision: 5644 $
|
---|
46 | * @see ServiceRack
|
---|
47 | */
|
---|
48 | public class GS2MGPPSearch
|
---|
49 | extends GS2Search {
|
---|
50 |
|
---|
51 | // extra services offered by mgpp collections
|
---|
52 | private static final String FIELD_QUERY_SERVICE = "FieldQuery";
|
---|
53 | private static final String ADVANCED_FIELD_QUERY_SERVICE = "AdvancedFieldQuery";
|
---|
54 |
|
---|
55 | // extra parameters used
|
---|
56 | private static final String INDEX_FIELD_PARAM = "index";
|
---|
57 | private static final String LEVEL_PARAM = "level";
|
---|
58 | private static final String RANK_PARAM = "sortBy";
|
---|
59 | private static final String RANK_PARAM_RANK = "1";
|
---|
60 | private static final String RANK_PARAM_NONE = "0";
|
---|
61 | private static final String SIMPLE_FIELD_PARAM = "simpleField";
|
---|
62 | private static final String ADVANCED_FIELD_PARAM = "complexField";
|
---|
63 | // more params for field query
|
---|
64 | private static final String FIELD_QUERY_PARAM = "fqv";
|
---|
65 | private static final String FIELD_STEM_PARAM = "fqs";
|
---|
66 | private static final String FIELD_CASE_PARAM = "fqc";
|
---|
67 | private static final String FIELD_FIELD_PARAM = "fqf";
|
---|
68 | private static final String FIELD_COMBINE_PARAM = "fqk";
|
---|
69 | private static final String FIELD_COMBINE_PARAM_AND = "0";
|
---|
70 | private static final String FIELD_COMBINE_PARAM_OR = "1";
|
---|
71 | private static final String FIELD_COMBINE_PARAM_NOT = "2";
|
---|
72 |
|
---|
73 | // some stuff for config files
|
---|
74 | private static final String SEARCH_TYPE_ELEM = "searchType";
|
---|
75 | private static final String SEARCH_TYPE_PLAIN = "plain";
|
---|
76 | private static final String SEARCH_TYPE_FORM = "form";
|
---|
77 | private static final String SEARCH_TYPE_FORM_SIMPLE = "simple";
|
---|
78 | private static final String SEARCH_TYPE_FORM_ADVANCED = "advanced";
|
---|
79 |
|
---|
80 |
|
---|
81 | private static final int TEXT_QUERY = 0;
|
---|
82 | private static final int SIMPLE_QUERY = 1;
|
---|
83 | private static final int ADVANCED_QUERY = 2;
|
---|
84 |
|
---|
85 | protected static final String FIELD_ATT = "field";
|
---|
86 | private MGPPWrapper mgpp_src=null;
|
---|
87 |
|
---|
88 | // the default level for retrieval - and we'll use it for searching too
|
---|
89 | private String default_level=null;
|
---|
90 | // which search services will we offer??
|
---|
91 | private boolean plain_search = false;
|
---|
92 | private boolean simple_form_search = false;
|
---|
93 | private boolean advanced_form_search = false;
|
---|
94 |
|
---|
95 | /** constructor */
|
---|
96 | public GS2MGPPSearch()
|
---|
97 | {
|
---|
98 | this.mgpp_src = new MGPPWrapper();
|
---|
99 | }
|
---|
100 |
|
---|
101 |
|
---|
102 | /** configure this service */
|
---|
103 | public boolean configure(Element info, Element extra_info)
|
---|
104 | {
|
---|
105 | // Do generic configuration
|
---|
106 | if (super.configure(info, extra_info) == false)
|
---|
107 | return false;
|
---|
108 |
|
---|
109 | // the format info is the same for all services
|
---|
110 | Element format_info = (Element)format_info_map.get(TEXT_QUERY_SERVICE);
|
---|
111 |
|
---|
112 | // the generic config has set up the text query service, but we may not want it
|
---|
113 | Element search_type_list = (Element) GSXML.getChildByTagName(info, SEARCH_TYPE_ELEM + GSXML.LIST_MODIFIER);
|
---|
114 | if (search_type_list == null) {
|
---|
115 | // assume form and plain
|
---|
116 | this.plain_search = true;
|
---|
117 | this.simple_form_search = true;
|
---|
118 | this.advanced_form_search = true;
|
---|
119 | } else {
|
---|
120 | NodeList types = search_type_list.getElementsByTagName(SEARCH_TYPE_ELEM);
|
---|
121 | for (int i=0; i<types.getLength(); i++) {
|
---|
122 | Element t = (Element)types.item(i);
|
---|
123 | String type_name = t.getAttribute(GSXML.NAME_ATT);
|
---|
124 | if (type_name.equals(SEARCH_TYPE_PLAIN)) {
|
---|
125 | this.plain_search = true;
|
---|
126 | } else if (type_name.equals(SEARCH_TYPE_FORM)) {
|
---|
127 | String type_type = t.getAttribute(GSXML.TYPE_ATT);
|
---|
128 | if (type_type.equals("")) {
|
---|
129 | this.simple_form_search = true;
|
---|
130 | this.advanced_form_search = true;
|
---|
131 | } else if (type_type.equals(SEARCH_TYPE_FORM_SIMPLE)) {
|
---|
132 | this.simple_form_search = true;
|
---|
133 | } else if (type_type.equals(SEARCH_TYPE_FORM_ADVANCED)) {
|
---|
134 | this.advanced_form_search = true;
|
---|
135 |
|
---|
136 | }
|
---|
137 | }
|
---|
138 | }
|
---|
139 | }
|
---|
140 |
|
---|
141 | if (!this.plain_search) {
|
---|
142 | // need to remove the TextQuery service
|
---|
143 | Element tq_service = GSXML.getNamedElement(short_service_info, GSXML.SERVICE_ELEM, GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
|
---|
144 | short_service_info.removeChild(tq_service);
|
---|
145 |
|
---|
146 | }
|
---|
147 |
|
---|
148 |
|
---|
149 | // Do specific configuration
|
---|
150 | System.out.println("Configuring GS2MGPPSearch...");
|
---|
151 |
|
---|
152 | // Get the default level out of <defaultLevel> (buildConfig.xml)
|
---|
153 | Element def = (Element) GSXML.getChildByTagName(info, DEFAULT_LEVEL_ELEM);
|
---|
154 | if (def != null) {
|
---|
155 | this.default_level = def.getAttribute(GSXML.NAME_ATT);
|
---|
156 | }
|
---|
157 | if (this.default_level == null || this.default_level.equals("")) {
|
---|
158 | System.err.println("Error: default level not specified!");
|
---|
159 | return false;
|
---|
160 | }
|
---|
161 |
|
---|
162 | // the default level is also the level which gdbm is expecting
|
---|
163 | // this must not be overwritten
|
---|
164 | this.mgpp_src.setReturnLevel(this.default_level);
|
---|
165 | // return term info
|
---|
166 | this.mgpp_src.setReturnTerms(true);
|
---|
167 | // set the default - this may be overwritten by query params
|
---|
168 | this.mgpp_src.setQueryLevel(this.default_level);
|
---|
169 |
|
---|
170 | // set up the extra services which are available for this collection
|
---|
171 | // check the config info - if there is no field list, then there is no fielded searching
|
---|
172 |
|
---|
173 | Element field_list = (Element) GSXML.getChildByTagName(info, GSXML.FIELD_ELEM+GSXML.LIST_MODIFIER);
|
---|
174 | if (field_list==null) {
|
---|
175 | // nothing more to do
|
---|
176 | return true;
|
---|
177 | }
|
---|
178 | // else set up the fielded query services
|
---|
179 |
|
---|
180 | if (this.simple_form_search) {
|
---|
181 | // set up short_service_info_ - for now just has id and type - name will be added in on teh fly
|
---|
182 | Element fq_service = this.doc.createElement(GSXML.SERVICE_ELEM);
|
---|
183 | fq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
|
---|
184 | fq_service.setAttribute(GSXML.NAME_ATT, FIELD_QUERY_SERVICE);
|
---|
185 | this.short_service_info.appendChild(fq_service);
|
---|
186 |
|
---|
187 | if (format_info != null) {
|
---|
188 | this.format_info_map.put(FIELD_QUERY_SERVICE, format_info);
|
---|
189 | }
|
---|
190 | }
|
---|
191 |
|
---|
192 | if (this.advanced_form_search) {
|
---|
193 | Element afq_service = this.doc.createElement(GSXML.SERVICE_ELEM);
|
---|
194 | afq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
|
---|
195 | afq_service.setAttribute(GSXML.NAME_ATT, ADVANCED_FIELD_QUERY_SERVICE);
|
---|
196 | this.short_service_info.appendChild(afq_service);
|
---|
197 |
|
---|
198 | if (format_info != null) {
|
---|
199 | this.format_info_map.put(ADVANCED_FIELD_QUERY_SERVICE, format_info);
|
---|
200 | }
|
---|
201 | }
|
---|
202 | return true;
|
---|
203 | }
|
---|
204 |
|
---|
205 | protected Element getServiceDescription(String service_id, String lang, String subset) {
|
---|
206 | // should we check that the service is actually on offer? presumably we wont get asked for services that we haven't advertised previously.
|
---|
207 |
|
---|
208 | if (!service_id.equals(FIELD_QUERY_SERVICE) && !service_id.equals(ADVANCED_FIELD_QUERY_SERVICE)) {
|
---|
209 | return super.getServiceDescription(service_id, lang, subset);
|
---|
210 | }
|
---|
211 |
|
---|
212 |
|
---|
213 | Element service = this.doc.createElement(GSXML.SERVICE_ELEM);
|
---|
214 | service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
|
---|
215 | service.setAttribute(GSXML.NAME_ATT, service_id);
|
---|
216 | if (subset == null || subset.equals(GSXML.DISPLAY_TEXT_ELEM)) {
|
---|
217 | service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_NAME, getTextString(service_id+".name", lang)));
|
---|
218 | service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_SUBMIT, getTextString(service_id+".submit", lang)));
|
---|
219 | service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_DESCRIPTION, getTextString(service_id+".description", lang)));
|
---|
220 |
|
---|
221 | }
|
---|
222 | if (subset == null || subset.equals(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER)) {
|
---|
223 | Element param_list = this.doc.createElement(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
|
---|
224 | service.appendChild(param_list);
|
---|
225 | if (service_id.equals(FIELD_QUERY_SERVICE)) {
|
---|
226 |
|
---|
227 | createFieldQueryParamList(param_list, lang);
|
---|
228 | } else {
|
---|
229 | createAdvancedFieldQueryParamList(param_list, lang);
|
---|
230 | }
|
---|
231 | }
|
---|
232 | return service;
|
---|
233 |
|
---|
234 | }
|
---|
235 |
|
---|
236 | /** creates a new param element and adds it to the param list */
|
---|
237 | protected void createParameter(String id, Element param_list, String lang) {
|
---|
238 | Element param=null;
|
---|
239 |
|
---|
240 | if (id.equals(LEVEL_PARAM)) {
|
---|
241 | // the level info - read from config file
|
---|
242 | Element level_list = (Element)GSXML.getChildByTagName(this.config_info, LEVEL_ELEM+GSXML.LIST_MODIFIER);
|
---|
243 | NodeList levels = level_list.getElementsByTagName(LEVEL_ELEM);
|
---|
244 | int len = levels.getLength();
|
---|
245 | if (len > 1) { // add level param to list only if more than one index specified
|
---|
246 | String [] levs = new String[len];
|
---|
247 | String [] lev_names = new String[len];
|
---|
248 | for (int i=0; i<len; i++) {
|
---|
249 | levs[i] = ((Element)levels.item(i)).getAttribute(GSXML.NAME_ATT);
|
---|
250 | lev_names[i] = getTextString("level."+levs[i], lang);
|
---|
251 |
|
---|
252 | }
|
---|
253 | // the first one is the default
|
---|
254 | param = GSXML.createParameterDescription(this.doc, LEVEL_PARAM, getTextString("param."+LEVEL_PARAM, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, levs[0], levs, lev_names);
|
---|
255 |
|
---|
256 | }
|
---|
257 | } else if (id.equals(RANK_PARAM)) {
|
---|
258 | String [] vals1 = {RANK_PARAM_RANK, RANK_PARAM_NONE };
|
---|
259 | String [] vals1_texts = { getTextString("param."+RANK_PARAM+"."+RANK_PARAM_RANK, lang), getTextString("param."+RANK_PARAM+"."+RANK_PARAM_NONE, lang)};
|
---|
260 |
|
---|
261 | param = GSXML.createParameterDescription(this.doc, RANK_PARAM, getTextString("param."+RANK_PARAM, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, RANK_PARAM_RANK, vals1, vals1_texts );
|
---|
262 |
|
---|
263 | } else if (id.equals(FIELD_QUERY_PARAM)) {
|
---|
264 | param = GSXML.createParameterDescription(this.doc, FIELD_QUERY_PARAM, getTextString("param."+FIELD_QUERY_PARAM, lang), GSXML.PARAM_TYPE_STRING, null, null, null);
|
---|
265 |
|
---|
266 |
|
---|
267 | } else if (id.equals(FIELD_CASE_PARAM) || id.equals(FIELD_STEM_PARAM)) {
|
---|
268 | String[] bool_ops = {"0", "1"};
|
---|
269 | String[] bool_texts = {getTextString("param.boolean.off", lang),getTextString("param.boolean.on", lang)};
|
---|
270 | param = GSXML.createParameterDescription(this.doc, id, getTextString("param."+id, lang), GSXML.PARAM_TYPE_BOOLEAN, BOOLEAN_PARAM_ON, bool_ops, bool_texts);
|
---|
271 |
|
---|
272 | } else if (id.equals(FIELD_FIELD_PARAM) || id.equals(INDEX_FIELD_PARAM)) {
|
---|
273 |
|
---|
274 | // the field list - read from config file
|
---|
275 | Element field_list = (Element)GSXML.getChildByTagName(this.config_info, GSXML.FIELD_ELEM+GSXML.LIST_MODIFIER);
|
---|
276 | NodeList fields = field_list.getElementsByTagName(GSXML.FIELD_ELEM);
|
---|
277 | int len = fields.getLength();
|
---|
278 | String [] f_names = new String [len];
|
---|
279 | String [] f_texts = new String [len];
|
---|
280 | for (int i=0; i< len;i++) {
|
---|
281 | f_names[i] = ((Element)fields.item(i)).getAttribute(GSXML.SHORTNAME_ATT);
|
---|
282 | // should these be changed to a text element based on lang?
|
---|
283 | // or is the name of a metadata element eg dc:Title its
|
---|
284 | // name in all langs
|
---|
285 | f_texts[i] = ((Element)fields.item(i)).getAttribute(GSXML.NAME_ATT);
|
---|
286 |
|
---|
287 | }
|
---|
288 | param = GSXML.createParameterDescription(this.doc, id, getTextString("param."+id, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, f_names[0], f_names, f_texts );
|
---|
289 |
|
---|
290 | } else if (id.equals(FIELD_COMBINE_PARAM)) {
|
---|
291 |
|
---|
292 | String []vals = {FIELD_COMBINE_PARAM_AND, FIELD_COMBINE_PARAM_OR, FIELD_COMBINE_PARAM_NOT};
|
---|
293 | String []val_texts = {getTextString("param."+FIELD_COMBINE_PARAM+"."+FIELD_COMBINE_PARAM_AND, lang), getTextString("param."+FIELD_COMBINE_PARAM+"."+FIELD_COMBINE_PARAM_OR, lang), getTextString("param."+FIELD_COMBINE_PARAM+"."+FIELD_COMBINE_PARAM_NOT, lang)};
|
---|
294 |
|
---|
295 |
|
---|
296 | param = GSXML.createParameterDescription(this.doc, FIELD_COMBINE_PARAM, "", GSXML.PARAM_TYPE_ENUM_SINGLE, FIELD_COMBINE_PARAM_AND, vals, val_texts);
|
---|
297 | param.setAttribute(GSXML.PARAM_IGNORE_POS_ATT, "0");
|
---|
298 |
|
---|
299 |
|
---|
300 | }
|
---|
301 |
|
---|
302 | // add the param to the list
|
---|
303 | if (param !=null) {
|
---|
304 | param_list.appendChild(param);
|
---|
305 | }
|
---|
306 | else {
|
---|
307 | super.createParameter(id, param_list, lang);
|
---|
308 | }
|
---|
309 | }
|
---|
310 |
|
---|
311 |
|
---|
312 | /** this creates all teh params and appends them to param_list.
|
---|
313 | * if display=true it creates the text strings version
|
---|
314 | * otherwise it creates the description version
|
---|
315 | */
|
---|
316 | protected boolean createTextQueryParamList(Element param_list,
|
---|
317 |
|
---|
318 | String lang)
|
---|
319 | {
|
---|
320 | // the order they are specified here is the order they appear on
|
---|
321 | // the query form
|
---|
322 | //createParameter(INDEX_PARAM, param_list, display, lang); - there is only ever one index now
|
---|
323 | createParameter(INDEX_FIELD_PARAM, param_list, lang);
|
---|
324 | createParameter(LEVEL_PARAM, param_list, lang);
|
---|
325 | createParameter(CASE_PARAM, param_list, lang);
|
---|
326 | createParameter(STEM_PARAM, param_list, lang);
|
---|
327 | createParameter(MATCH_PARAM, param_list, lang);
|
---|
328 | createParameter(RANK_PARAM, param_list, lang);
|
---|
329 | createParameter(MAXDOCS_PARAM, param_list, lang);
|
---|
330 | createParameter(QUERY_PARAM, param_list, lang);
|
---|
331 |
|
---|
332 | return true;
|
---|
333 | }
|
---|
334 |
|
---|
335 |
|
---|
336 |
|
---|
337 | /** this creates all teh params and appends them to param_list.
|
---|
338 | * if display=true it creates the text strings version
|
---|
339 | * otherwise it creates the description version
|
---|
340 | */
|
---|
341 | protected boolean createFieldQueryParamList(Element param_list,
|
---|
342 | String lang)
|
---|
343 | {
|
---|
344 | // add in the index param
|
---|
345 | //createParameter(INDEX_PARAM, param_list, display, lang);
|
---|
346 | createParameter(LEVEL_PARAM, param_list, lang);
|
---|
347 | createParameter(MATCH_PARAM, param_list, lang);
|
---|
348 | createParameter(CASE_PARAM, param_list, lang);
|
---|
349 | createParameter(STEM_PARAM, param_list, lang);
|
---|
350 | createParameter(RANK_PARAM, param_list, lang);
|
---|
351 | createParameter(MAXDOCS_PARAM, param_list, lang);
|
---|
352 |
|
---|
353 | // create a multi param for the fields etc
|
---|
354 | // text box, field
|
---|
355 | Element multiparam = null;
|
---|
356 | Element param=null;
|
---|
357 | multiparam = GSXML.createParameterDescription(this.doc, SIMPLE_FIELD_PARAM, "", GSXML.PARAM_TYPE_MULTI, null, null, null);
|
---|
358 | multiparam.setAttribute("occurs", "4");
|
---|
359 | param_list.appendChild(multiparam);
|
---|
360 |
|
---|
361 | // the components
|
---|
362 | createParameter(FIELD_QUERY_PARAM, multiparam, lang);
|
---|
363 | createParameter(FIELD_FIELD_PARAM, multiparam, lang);
|
---|
364 |
|
---|
365 | return true;
|
---|
366 | }
|
---|
367 |
|
---|
368 | protected boolean createAdvancedFieldQueryParamList(Element param_list,
|
---|
369 | String lang) {
|
---|
370 |
|
---|
371 | // first do index and level params
|
---|
372 | //createParameter(INDEX_PARAM, param_list, display, lang);
|
---|
373 | createParameter(LEVEL_PARAM, param_list, lang);
|
---|
374 | createParameter(RANK_PARAM, param_list, lang);
|
---|
375 | createParameter(MAXDOCS_PARAM, param_list, lang);
|
---|
376 |
|
---|
377 |
|
---|
378 | // create a multi param for the fields etc
|
---|
379 | // text box, stem, case, field
|
---|
380 |
|
---|
381 | Element multiparam = null;
|
---|
382 | Element param=null;
|
---|
383 |
|
---|
384 | multiparam = GSXML.createParameterDescription(this.doc, ADVANCED_FIELD_PARAM, "", GSXML.PARAM_TYPE_MULTI, null, null, null);
|
---|
385 | multiparam.setAttribute("occurs", "4");
|
---|
386 | param_list.appendChild(multiparam);
|
---|
387 |
|
---|
388 | createParameter(FIELD_COMBINE_PARAM, multiparam, lang);
|
---|
389 | createParameter(FIELD_QUERY_PARAM, multiparam, lang);
|
---|
390 | createParameter(FIELD_CASE_PARAM, multiparam, lang);
|
---|
391 | createParameter(FIELD_STEM_PARAM, multiparam, lang);
|
---|
392 | createParameter(FIELD_FIELD_PARAM, multiparam, lang);
|
---|
393 |
|
---|
394 |
|
---|
395 | return true;
|
---|
396 | }
|
---|
397 |
|
---|
398 | // the following three functions are needed so the base class can
|
---|
399 | // call the process+SERVICE_NAME methods
|
---|
400 | /** process a text query */
|
---|
401 | protected Element processTextQuery(Element request) {
|
---|
402 | return processAnyQuery(request, TEXT_QUERY);
|
---|
403 | }
|
---|
404 |
|
---|
405 | /** process a field query */
|
---|
406 | protected Element processFieldQuery(Element request) {
|
---|
407 | return processAnyQuery(request, SIMPLE_QUERY);
|
---|
408 | }
|
---|
409 |
|
---|
410 | /** process an advanced field query */
|
---|
411 | protected Element processAdvancedFieldQuery(Element request) {
|
---|
412 | return processAnyQuery(request, ADVANCED_QUERY);
|
---|
413 | }
|
---|
414 |
|
---|
415 | /** process a query */
|
---|
416 | protected Element processAnyQuery(Element request, int query_type)
|
---|
417 | {
|
---|
418 |
|
---|
419 | String service_name=null;
|
---|
420 | String empty_query_test_param=null;
|
---|
421 | // set up the type specific bits
|
---|
422 | switch (query_type) {
|
---|
423 | case TEXT_QUERY:
|
---|
424 | service_name = TEXT_QUERY_SERVICE;
|
---|
425 | empty_query_test_param = QUERY_PARAM;
|
---|
426 | break;
|
---|
427 | case SIMPLE_QUERY:
|
---|
428 | service_name = FIELD_QUERY_SERVICE;
|
---|
429 | empty_query_test_param = FIELD_QUERY_PARAM;
|
---|
430 | break;
|
---|
431 | case ADVANCED_QUERY:
|
---|
432 | service_name = ADVANCED_FIELD_QUERY_SERVICE;
|
---|
433 | empty_query_test_param = FIELD_QUERY_PARAM;
|
---|
434 | break;
|
---|
435 | default:
|
---|
436 | // should never get here
|
---|
437 | System.err.println("GS2MGPPSearch: wrong query type!!");
|
---|
438 | return null;
|
---|
439 | }
|
---|
440 |
|
---|
441 | // Create a new (empty) result message
|
---|
442 | Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
|
---|
443 | result.setAttribute(GSXML.FROM_ATT, service_name);
|
---|
444 | result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
|
---|
445 |
|
---|
446 | // Get the parameters of the request
|
---|
447 | Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
|
---|
448 | if (param_list == null) {
|
---|
449 | System.err.println("Error: TextQuery request had no paramList.");
|
---|
450 | return result; // Return the empty result
|
---|
451 | }
|
---|
452 |
|
---|
453 | // Process the request parameters
|
---|
454 | HashMap params = GSXML.extractParams(param_list, false);
|
---|
455 |
|
---|
456 | // Make sure a query has been specified
|
---|
457 | String query = (String) params.get(empty_query_test_param);
|
---|
458 | if (query == null || query.equals("")) {
|
---|
459 | return result; // Return the empty result
|
---|
460 | }
|
---|
461 |
|
---|
462 | // If an field hasn't been specified, use all fields
|
---|
463 | String field = (String) params.get(INDEX_FIELD_PARAM);
|
---|
464 | if (field == null) {
|
---|
465 | field = "ZZ";
|
---|
466 | }
|
---|
467 |
|
---|
468 | // set up mgpp_src
|
---|
469 | String indexdir = GSFile.collectionBaseDir(this.site_home, this.cluster_name) +
|
---|
470 | File.separatorChar + GSFile.collectionIndexPath(this.cluster_name, this.default_index);
|
---|
471 | this.mgpp_src.loadIndexData(indexdir);
|
---|
472 | setStandardQueryParams(params);
|
---|
473 |
|
---|
474 | // if field search, create the query string
|
---|
475 | switch (query_type) {
|
---|
476 | case TEXT_QUERY:
|
---|
477 | query = addFieldInfo(query, field);
|
---|
478 | break;
|
---|
479 | case SIMPLE_QUERY:
|
---|
480 | query = parseFieldQueryParams(params);
|
---|
481 | break;
|
---|
482 | case ADVANCED_QUERY:
|
---|
483 | query = parseAdvancedFieldQueryParams(params);
|
---|
484 | break;
|
---|
485 | }
|
---|
486 | // run the query
|
---|
487 | this.mgpp_src.runQuery(query);
|
---|
488 | MGPPQueryResult mqr= this.mgpp_src.getQueryResult();
|
---|
489 |
|
---|
490 | // build up the response
|
---|
491 |
|
---|
492 | // Create a metadata list to store information about the query results
|
---|
493 | // should we be using metadataList? or something else?
|
---|
494 | Element metadata_list = this.doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
|
---|
495 | result.appendChild(metadata_list);
|
---|
496 |
|
---|
497 | // Add a metadata element specifying the number of matching documents
|
---|
498 | long totalDocs = mqr.getTotalDocs();
|
---|
499 | GSXML.addMetadata(this.doc, metadata_list, "numDocsMatched", ""+totalDocs);
|
---|
500 |
|
---|
501 | // Create a document list to store the matching documents, and add them
|
---|
502 | Vector docs = mqr.getDocs();
|
---|
503 |
|
---|
504 | // add a metadata item to specify docs returned
|
---|
505 | GSXML.addMetadata(this.doc, metadata_list, "numDocsReturned", ""+docs.size());
|
---|
506 | Element document_list = this.doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
|
---|
507 | result.appendChild(document_list);
|
---|
508 | for (int d = 0; d < docs.size(); d++) {
|
---|
509 | long docnum = ((MGPPDocInfo) docs.elementAt(d)).num_;
|
---|
510 | String doc_id = this.gdbm_src.docnum2Oid(docnum);
|
---|
511 | Element doc_node = createDocumentNodeElement(doc_id);
|
---|
512 | document_list.appendChild(doc_node);
|
---|
513 | }
|
---|
514 |
|
---|
515 | // Create a term list to store the term information, and add it
|
---|
516 | String query_level = (String)params.get(LEVEL_PARAM); // the current query level
|
---|
517 | Element term_list = this.doc.createElement(GSXML.TERM_ELEM+GSXML.LIST_MODIFIER);
|
---|
518 | result.appendChild(term_list);
|
---|
519 | Vector terms = mqr.getTerms();
|
---|
520 | for (int t = 0; t < terms.size(); t++) {
|
---|
521 | MGPPTermInfo term_info = (MGPPTermInfo) terms.get(t);
|
---|
522 |
|
---|
523 | Element term_elem = this.doc.createElement(GSXML.TERM_ELEM);
|
---|
524 | term_elem.setAttribute(GSXML.NAME_ATT, term_info.term_);
|
---|
525 | term_elem.setAttribute(STEM_ATT, "" + term_info.stem_method_);
|
---|
526 | term_elem.setAttribute(FREQ_ATT, "" + term_info.term_freq_);
|
---|
527 | term_elem.setAttribute(NUM_DOCS_MATCH_ATT, "" + term_info.match_docs_);
|
---|
528 | field = term_info.tag_;
|
---|
529 | if (field.equals(query_level)) {
|
---|
530 | // ignore
|
---|
531 | field = "";
|
---|
532 | }
|
---|
533 | term_elem.setAttribute(FIELD_ATT, field);
|
---|
534 |
|
---|
535 | Vector equiv_terms = term_info.equiv_terms_;
|
---|
536 | Element equiv_term_list = this.doc.createElement(EQUIV_TERM_ELEM+GSXML.LIST_MODIFIER);
|
---|
537 | term_elem.appendChild(equiv_term_list);
|
---|
538 |
|
---|
539 | for (int et = 0; et < equiv_terms.size(); et++) {
|
---|
540 | String equiv_term = (String) equiv_terms.get(et);
|
---|
541 |
|
---|
542 | Element equiv_term_elem = this.doc.createElement(GSXML.TERM_ELEM);
|
---|
543 | equiv_term_elem.setAttribute(GSXML.NAME_ATT, equiv_term);
|
---|
544 | equiv_term_elem.setAttribute(NUM_DOCS_MATCH_ATT, "");
|
---|
545 | equiv_term_elem.setAttribute(FREQ_ATT, "");
|
---|
546 | equiv_term_list.appendChild(equiv_term_elem);
|
---|
547 | }
|
---|
548 |
|
---|
549 | term_list.appendChild(term_elem);
|
---|
550 | }
|
---|
551 |
|
---|
552 | return result;
|
---|
553 | }
|
---|
554 |
|
---|
555 | // should probably use a list rather than map
|
---|
556 | protected boolean setStandardQueryParams(HashMap params) {
|
---|
557 |
|
---|
558 | Set entries = params.entrySet();
|
---|
559 | Iterator i = entries.iterator();
|
---|
560 | while (i.hasNext()) {
|
---|
561 | Map.Entry m = (Map.Entry)i.next();
|
---|
562 | String name = (String)m.getKey();
|
---|
563 | String value = (String)m.getValue();
|
---|
564 |
|
---|
565 | if (name.equals(CASE_PARAM)) {
|
---|
566 | boolean val = (value.equals(BOOLEAN_PARAM_ON)?true:false);
|
---|
567 | this.mgpp_src.setCase(val);
|
---|
568 | } else if (name.equals(STEM_PARAM)) {
|
---|
569 | boolean val = (value.equals(BOOLEAN_PARAM_ON)?true:false);
|
---|
570 | this.mgpp_src.setStem(val);
|
---|
571 | } else if (name.equals(MAXDOCS_PARAM)&& !value.equals("")) {
|
---|
572 | int docs = Integer.parseInt(value);
|
---|
573 | this.mgpp_src.setMaxDocs(docs);
|
---|
574 | } else if (name.equals(LEVEL_PARAM)) {
|
---|
575 | this.mgpp_src.setQueryLevel(value);
|
---|
576 | } else if (name.equals(MATCH_PARAM)) {
|
---|
577 | int mode;
|
---|
578 | if (value.equals(MATCH_PARAM_ALL)) mode=1;
|
---|
579 | else mode=0;
|
---|
580 | this.mgpp_src.setMatchMode(mode);
|
---|
581 | } else if (name.equals(RANK_PARAM)) {
|
---|
582 | if (value.equals(RANK_PARAM_RANK)) {
|
---|
583 | this.mgpp_src.setSortByRank(true);
|
---|
584 | } else if (value.equals(RANK_PARAM_NONE)) {
|
---|
585 | this.mgpp_src.setSortByRank(false);
|
---|
586 | }
|
---|
587 | } // ignore any others
|
---|
588 | }
|
---|
589 | return true;
|
---|
590 | }
|
---|
591 |
|
---|
592 | protected String addFieldInfo(String query, String field) {
|
---|
593 | if (field.equals("") || field.equals("ZZ")) {
|
---|
594 | return query;
|
---|
595 | }
|
---|
596 | return "["+query+"]:"+field;
|
---|
597 | }
|
---|
598 | /** combines all the field params into a single query
|
---|
599 | * - for simple field query */
|
---|
600 | protected String parseFieldQueryParams(HashMap params) {
|
---|
601 |
|
---|
602 | StringBuffer final_query = new StringBuffer(256);
|
---|
603 | String text_line = (String)params.get(FIELD_QUERY_PARAM);
|
---|
604 | String[] texts = text_line.split(",", -1);
|
---|
605 | String field_line = (String)params.get(FIELD_FIELD_PARAM);
|
---|
606 | String[] fields = field_line.split(",", -1);
|
---|
607 | String combine="&";
|
---|
608 | String match = (String)params.get(MATCH_PARAM);
|
---|
609 | if (match.equals(MATCH_PARAM_SOME)) {
|
---|
610 | combine = "|";
|
---|
611 | }
|
---|
612 |
|
---|
613 | for (int i=0; i<texts.length; i++) {
|
---|
614 |
|
---|
615 | String q = texts[i].trim();
|
---|
616 | if (!q.equals("")) {
|
---|
617 | addQueryElem(final_query, q, fields[i], combine);
|
---|
618 | }
|
---|
619 | }
|
---|
620 |
|
---|
621 | return final_query.toString();
|
---|
622 | }
|
---|
623 |
|
---|
624 | /** combines all the field params into a single query
|
---|
625 | * - for advanced field query */
|
---|
626 | protected String parseAdvancedFieldQueryParams(HashMap params) {
|
---|
627 |
|
---|
628 | StringBuffer final_query = new StringBuffer(256);
|
---|
629 | String text_line = (String)params.get(FIELD_QUERY_PARAM);
|
---|
630 | String[] texts = text_line.split(",", -1);
|
---|
631 | String field_line = (String)params.get(FIELD_FIELD_PARAM);
|
---|
632 | String[] fields = field_line.split(",", -1);
|
---|
633 | String case_line = (String)params.get(FIELD_CASE_PARAM);
|
---|
634 | String[] cases = case_line.split(",", -1);
|
---|
635 | String stem_line = (String)params.get(FIELD_STEM_PARAM);
|
---|
636 | String[] stems = stem_line.split(",", -1);
|
---|
637 | String combine_line = (String)params.get(FIELD_COMBINE_PARAM);
|
---|
638 | String [] combines = combine_line.split(",", -1);
|
---|
639 | String combine = "&";
|
---|
640 | for (int i=0; i<texts.length; i++) {
|
---|
641 | if (i==0) {// assume first one is blank
|
---|
642 | combine = "";
|
---|
643 | } else {
|
---|
644 | String x = combines[i];
|
---|
645 | if (x.equals(FIELD_COMBINE_PARAM_AND)) {
|
---|
646 | combine = "&";
|
---|
647 | } else if (x.equals(FIELD_COMBINE_PARAM_OR)) {
|
---|
648 | combine = "|";
|
---|
649 | } else if (x.equals(FIELD_COMBINE_PARAM_NOT)) {
|
---|
650 | combine = "!";
|
---|
651 | }
|
---|
652 |
|
---|
653 | }
|
---|
654 |
|
---|
655 | String q = texts[i].trim();
|
---|
656 | if (!q.equals("")) {
|
---|
657 | q = addStemAndCase(q, stems[i], cases[i]);
|
---|
658 | addQueryElem(final_query, q, fields[i], combine);
|
---|
659 | }
|
---|
660 | }
|
---|
661 |
|
---|
662 | return final_query.toString();
|
---|
663 | }
|
---|
664 |
|
---|
665 | protected void addQueryElem(StringBuffer s, String q, String f, String c) {
|
---|
666 |
|
---|
667 | String combine="";
|
---|
668 | if (s.length()>0) {
|
---|
669 | combine = " "+c+" ";
|
---|
670 | }
|
---|
671 | if (f.equals("")||f.equals("ZZ")) {
|
---|
672 | s.append(combine+q);
|
---|
673 | } else {
|
---|
674 | s.append(combine+"["+q+"]:"+f);
|
---|
675 | }
|
---|
676 | }
|
---|
677 |
|
---|
678 | protected String addStemAndCase(String q, String s, String c) {
|
---|
679 | String mods = "#";
|
---|
680 | if (c.equals("1")) {
|
---|
681 | mods += "i";
|
---|
682 | } else {
|
---|
683 | mods += "c";
|
---|
684 | }
|
---|
685 | if (s.equals("1")) {
|
---|
686 | mods += "s";
|
---|
687 | } else {
|
---|
688 | mods+= "u";
|
---|
689 | }
|
---|
690 | StringBuffer temp = new StringBuffer();
|
---|
691 | String [] terms = q.split(" ");
|
---|
692 | for (int i=0; i<terms.length; i++) {
|
---|
693 | String t = terms[i].trim();
|
---|
694 |
|
---|
695 | if (!t.equals("") && !t.equals("TX")) {
|
---|
696 | temp.append(" "+t+mods);
|
---|
697 | }
|
---|
698 | }
|
---|
699 | return temp.toString();
|
---|
700 | }
|
---|
701 |
|
---|
702 |
|
---|
703 | }
|
---|
704 |
|
---|
705 |
|
---|