source: main/trunk/greenstone3/src/java/org/greenstone/gsdl3/service/AbstractGS2FieldSearch.java@ 29558

Last change on this file since 29558 was 29558, checked in by kjdon, 9 years ago

work around does_paging, does_chunking. only add in maxdocs, hitsperpage params if the service actually uses them. lucnee/solr, don't use maxdocs any more. I haven't had a chance to clean up the changes, but I need to commit, so there may be extraneous debug statements still here.
  • Property svn:keywords set to Author Date Id Revision
File size: 29.4 KB
Line 
1/*
2 * AbstractGS2FieldSearch.java
3 * Copyright (C) 2006 New Zealand Digital Library, http://www.nzdl.org
4 *
5 * This program is free software; you can redistribute it and/or modify
6 * the Free Software Foundation; either version 2 of the License, or
7 * (at your option) any later version.
8 *
9 * This program is distributed in the hope that it will be useful,
10 * but WITHOUT ANY WARRANTY; without even the implied warranty of
11 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
12 * GNU General Public License for more details.
13 *
14 * You should have received a copy of the GNU General Public License
15 * along with this program; if not, write to the Free Software
16 * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
17 */
18package org.greenstone.gsdl3.service;
19
20// Greenstone classes
21import java.io.Serializable;
22import java.util.ArrayList;
23import java.util.HashMap;
24
25import org.apache.log4j.Logger;
26import org.greenstone.gsdl3.util.FacetWrapper;
27import org.greenstone.gsdl3.util.GSXML;
28import org.greenstone.gsdl3.util.XMLConverter;
29
30import org.w3c.dom.Document;
31import org.w3c.dom.Element;
32import org.w3c.dom.NodeList;
33
34abstract public class AbstractGS2FieldSearch extends AbstractGS2TextSearch
35{
36 // extra services offered by mgpp collections
37 protected static final String FIELD_QUERY_SERVICE = "FieldQuery";
38 protected static final String ADVANCED_FIELD_QUERY_SERVICE = "AdvancedFieldQuery";
39 protected static final String RAW_QUERY_SERVICE = "RawQuery";
40
41 // extra parameters used
42 protected static final String LEVEL_PARAM = "level";
43 protected static final String RANK_PARAM = "sortBy";
44 protected static final String RANK_PARAM_RANK = "1";
45 protected static final String RANK_PARAM_NONE = "0";
46 protected static final String SIMPLE_FIELD_PARAM = "simpleField";
47 protected static final String ADVANCED_FIELD_PARAM = "complexField";
48
49 protected static final String RAW_PARAM = "rawquery";
50
51 // more params for field query
52 protected static final String FIELD_QUERY_PARAM = "fqv";
53 protected static final String FIELD_STEM_PARAM = "fqs";
54 protected static final String FIELD_CASE_PARAM = "fqc";
55 protected static final String FIELD_ACCENT_PARAM = "fqa";
56 protected static final String FIELD_FIELD_PARAM = "fqf";
57 protected static final String FIELD_COMBINE_PARAM = "fqk";
58 protected static final String FIELD_COMBINE_PARAM_AND = "0";
59 protected static final String FIELD_COMBINE_PARAM_OR = "1";
60 protected static final String FIELD_COMBINE_PARAM_NOT = "2";
61
62 // some stuff for config files
63 protected static final String SEARCH_TYPE_ELEM = "searchType";
64 protected static final String SEARCH_TYPE_PLAIN = "plain";
65 protected static final String SEARCH_TYPE_SIMPLE_FORM = "simpleform";
66 protected static final String SEARCH_TYPE_ADVANCED_FORM = "advancedform";
67 protected static final String SEARCH_TYPE_RAW = "raw";
68
69 protected static final String DEFAULT_LEVEL_ELEM = "defaultLevel";
70 protected static final String DEFAULT_DB_LEVEL_ELEM = "defaultDBLevel";
71 protected static final String LEVEL_ELEM = "level";
72 protected static final String FIELD_ATT = "field";
73
74 protected static final int TEXT_QUERY = 0;
75 protected static final int SIMPLE_QUERY = 1;
76 protected static final int ADVANCED_QUERY = 2;
77 protected static final int RAW_QUERY = 3;
78
79 protected String AND_OPERATOR = "&";
80 protected String OR_OPERATOR = "|";
81 protected String NOT_OPERATOR = "!";
82
83 // the default level for searching
84 protected String default_level = null;
85 // default level for collection db
86 protected String default_db_level = null;
87 // which search services will we offer??
88 protected boolean plain_search = false;
89 protected boolean simple_form_search = false;
90 protected boolean advanced_form_search = false;
91 protected boolean raw_search = false;
92
93 static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.AbstractGS2FieldSearch.class.getName());
94
95 /** constructor */
96 public AbstractGS2FieldSearch()
97 {
98 super();
99 paramDefaults.put(RANK_PARAM, RANK_PARAM_RANK);
100 paramDefaults.put(FIELD_COMBINE_PARAM, FIELD_COMBINE_PARAM_AND);
101 // paramDefaults.put(FIELD_CASE_PARAM, BOOLEAN_PARAM_ON);
102 // paramDefaults.put(FIELD_STEM_PARAM, BOOLEAN_PARAM_OFF);
103 // paramDefaults.put(FIELD_ACCENT_PARAM, BOOLEAN_PARAM_ON);
104 }
105
106 public void cleanUp()
107 {
108 super.cleanUp();
109 }
110
111 /** configure this service */
112 public boolean configure(Element info, Element extra_info)
113 {
114 if (!super.configure(info, extra_info))
115 {
116 return false;
117 }
118
119 // Get the default level out of <defaultLevel> (buildConfig.xml)
120 Element def = (Element) GSXML.getChildByTagName(info, DEFAULT_LEVEL_ELEM);
121 if (def != null)
122 {
123 this.default_level = def.getAttribute(GSXML.SHORTNAME_ATT);
124 }
125 if (this.default_level == null || this.default_level.equals(""))
126 {
127 logger.error("default level not specified!, assuming Doc");
128 this.default_level = "Doc";
129 }
130
131 // Get the default DB level
132 def = (Element) GSXML.getChildByTagName(info, DEFAULT_DB_LEVEL_ELEM);
133 if (def != null)
134 {
135 this.default_db_level = def.getAttribute(GSXML.SHORTNAME_ATT);
136 }
137 if (this.default_db_level == null || this.default_db_level.equals(""))
138 {
139 logger.error("default database level (defaultDBLevel) not specified!, assuming Sec");
140 this.default_db_level = "Sec";
141 }
142
143 // get stuff from from extra info (which is collectionConfig.xml)
144 if (extra_info != null)
145 {
146
147 // the search element
148 Element config_search = (Element) GSXML.getChildByTagName(extra_info, GSXML.SEARCH_ELEM);
149
150 NodeList search_types = config_search.getElementsByTagName(SEARCH_TYPE_ELEM);
151 if (search_types == null)
152 {
153 // none specified, assume plain only
154 this.plain_search = true;
155 }
156 else
157 {
158 for (int i = 0; i < search_types.getLength(); i++)
159 {
160 Element t = (Element) search_types.item(i);
161
162 String type_name = t.getAttribute(GSXML.NAME_ATT);
163 if (type_name.equals(SEARCH_TYPE_PLAIN))
164 {
165 this.plain_search = true;
166 }
167 else if (type_name.equals(SEARCH_TYPE_SIMPLE_FORM))
168 {
169 this.simple_form_search = true;
170 }
171 else if (type_name.equals(SEARCH_TYPE_ADVANCED_FORM))
172 {
173 this.advanced_form_search = true;
174 }
175 else if (type_name.equals(SEARCH_TYPE_RAW))
176 {
177 this.raw_search = true;
178 }
179 }
180 }
181
182 // AbstractGS2TextSearch has set up the TextQuery service, but we may not want it
183 if (!this.plain_search)
184 {
185 // need to remove the TextQuery service
186 Element tq_service = GSXML.getNamedElement(short_service_info, GSXML.SERVICE_ELEM, GSXML.NAME_ATT, QUERY_SERVICE);
187 short_service_info.removeChild(tq_service);
188 }
189
190 Document owner = info.getOwnerDocument();
191
192 NodeList levels = info.getElementsByTagName(GSXML.LEVEL_ELEM);
193
194 for (int i = 0; i < levels.getLength(); i++)
195 {
196 Element lev = (Element) levels.item(i);
197 String name = lev.getAttribute(GSXML.NAME_ATT);
198 Element node_extra = GSXML.getNamedElement(config_search, GSXML.LEVEL_ELEM, GSXML.NAME_ATT, name);
199 if (node_extra == null)
200 {
201 logger.error("haven't found extra info for level named " + name);
202 continue;
203 }
204
205 // get the display elements if any - displayName
206 NodeList display_names = node_extra.getElementsByTagName(GSXML.DISPLAY_TEXT_ELEM);
207 if (display_names != null)
208 {
209 for (int j = 0; j < display_names.getLength(); j++)
210 {
211 Element e = (Element) display_names.item(j);
212 lev.appendChild(owner.importNode(e, true));
213 }
214 }
215 } // for each level
216 }
217 else
218 {
219 // for soem reason we don't have the collectionConfig file. assume plain only
220 this.plain_search = true;
221 }
222
223 // the format info is the same for all services
224 Element format_info = (Element) format_info_map.get(QUERY_SERVICE);
225
226 // set up the extra services which are available for this collection
227 if (this.simple_form_search)
228 {
229 // set up short_service_info_ - for now just has id and type - name will be added in on the fly
230 Element fq_service = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
231 fq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
232 fq_service.setAttribute(GSXML.NAME_ATT, FIELD_QUERY_SERVICE);
233 this.short_service_info.appendChild(fq_service);
234
235 if (format_info != null)
236 {
237 this.format_info_map.put(FIELD_QUERY_SERVICE, format_info);
238 }
239 }
240
241 if (this.advanced_form_search)
242 {
243 Element afq_service = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
244 afq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
245 afq_service.setAttribute(GSXML.NAME_ATT, ADVANCED_FIELD_QUERY_SERVICE);
246 this.short_service_info.appendChild(afq_service);
247
248 if (format_info != null)
249 {
250 this.format_info_map.put(ADVANCED_FIELD_QUERY_SERVICE, format_info);
251 }
252 }
253
254 if (this.raw_search)
255 {
256 Element rq_service = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
257 rq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
258 rq_service.setAttribute(GSXML.NAME_ATT, RAW_QUERY_SERVICE);
259 this.short_service_info.appendChild(rq_service);
260
261 if (format_info != null)
262 {
263 this.format_info_map.put(RAW_QUERY_SERVICE, format_info);
264 }
265 }
266
267 return true;
268 }
269
270 // if field case/stem/accent defaults are not specified explicitly, we should use case/stem/accent defaults
271 protected void getSearchParamDefaults(Element search_elem) {
272
273 boolean found_f_stem = false;
274 boolean found_f_case = false;
275 boolean found_f_accent = false;
276 NodeList param_defaults_list = GSXML.getChildrenByTagName(search_elem, GSXML.PARAM_DEFAULT_ELEM);
277 for (int i=0; i<param_defaults_list.getLength(); i++) {
278 Element paramdef = (Element)param_defaults_list.item(i);
279 String name = paramdef.getAttribute(GSXML.NAME_ATT);
280 String val = paramdef.getAttribute(GSXML.VALUE_ATT);
281 if (!name.equals("") && !val.equals("")) {
282 paramDefaults.put(name, val);
283 if (name.equals(FIELD_STEM_PARAM)) {
284 found_f_stem = true;
285 } else if (name.equals(FIELD_CASE_PARAM)) {
286 found_f_case = true;
287 } else if (name.equals(FIELD_ACCENT_PARAM)) {
288 found_f_accent = true;
289 }
290 if (!found_f_stem) {
291 paramDefaults.put (FIELD_STEM_PARAM, paramDefaults.get(STEM_PARAM));
292 }
293 if (!found_f_case) {
294 paramDefaults.put (FIELD_CASE_PARAM, paramDefaults.get(CASE_PARAM));
295 }
296 if (!found_f_accent) {
297 paramDefaults.put (FIELD_ACCENT_PARAM, paramDefaults.get(ACCENT_PARAM));
298 }
299
300 }
301 }
302 }
303 protected Element getServiceDescription(Document doc, String service_id, String lang, String subset)
304 {
305 // should we check that the service is actually on offer? presumably we wont get asked for services that we haven't advertised previously.
306
307 if (!service_id.equals(FIELD_QUERY_SERVICE) && !service_id.equals(ADVANCED_FIELD_QUERY_SERVICE) && !service_id.equals(RAW_QUERY_SERVICE))
308 {
309 return super.getServiceDescription(doc, service_id, lang, subset);
310 }
311
312 Element service = doc.createElement(GSXML.SERVICE_ELEM);
313 service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
314 service.setAttribute(GSXML.NAME_ATT, service_id);
315 if (subset == null || subset.equals(GSXML.DISPLAY_TEXT_ELEM + GSXML.LIST_MODIFIER))
316 {
317 service.appendChild(GSXML.createDisplayTextElement(doc, GSXML.DISPLAY_TEXT_NAME, getTextString(service_id + ".name", lang)));
318 service.appendChild(GSXML.createDisplayTextElement(doc, GSXML.DISPLAY_TEXT_SUBMIT, getTextString(service_id + ".submit", lang)));
319 service.appendChild(GSXML.createDisplayTextElement(doc, GSXML.DISPLAY_TEXT_DESCRIPTION, getTextString(service_id + ".description", lang)));
320
321 }
322 if (subset == null || subset.equals(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER))
323 {
324 Element param_list = doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
325 service.appendChild(param_list);
326 if (service_id.equals(FIELD_QUERY_SERVICE))
327 {
328
329 addCustomQueryParams(param_list, lang);
330 if (!default_index_subcollection.equals(""))
331 {
332 createParameter(INDEX_SUBCOLLECTION_PARAM, param_list, lang);
333 }
334 if (!default_index_language.equals(""))
335 {
336 createParameter(INDEX_LANGUAGE_PARAM, param_list, lang);
337 }
338
339 if (does_chunking) {
340 createParameter(MAXDOCS_PARAM, param_list, lang);
341 }
342 if (does_paging)
343 {
344 createParameter(HITS_PER_PAGE_PARAM, param_list, lang);
345 createParameter(START_PAGE_PARAM, param_list, lang);
346 }
347
348 // create a multi param for the fields etc
349 // text box, field
350 Element multiparam = null;
351 Element param = null;
352 multiparam = GSXML.createParameterDescription(doc, SIMPLE_FIELD_PARAM, "", GSXML.PARAM_TYPE_MULTI, null, null, null);
353 multiparam.setAttribute("occurs", "4");
354 param_list.appendChild(multiparam);
355
356 // the components
357
358 createParameter(FIELD_QUERY_PARAM, multiparam, lang);
359 createParameter(FIELD_FIELD_PARAM, multiparam, lang);
360
361 }
362 else if (service_id.equals(ADVANCED_FIELD_QUERY_SERVICE))
363 {
364 addCustomQueryParamsAdvField(param_list, lang);
365 if (!default_index_subcollection.equals(""))
366 {
367 createParameter(INDEX_SUBCOLLECTION_PARAM, param_list, lang);
368 }
369 if (!default_index_language.equals(""))
370 {
371 createParameter(INDEX_LANGUAGE_PARAM, param_list, lang);
372 }
373
374 // create a multi param for the fields etc
375 // text box, stem, case, field
376
377 Element multiparam = null;
378 Element param = null;
379
380 multiparam = GSXML.createParameterDescription(doc, ADVANCED_FIELD_PARAM, "", GSXML.PARAM_TYPE_MULTI, null, null, null);
381 multiparam.setAttribute("occurs", "4");
382 param_list.appendChild(multiparam);
383
384 createParameter(FIELD_COMBINE_PARAM, multiparam, lang);
385 createParameter(FIELD_QUERY_PARAM, multiparam, lang);
386 if (this.does_case)
387 {
388 createParameter(FIELD_CASE_PARAM, multiparam, lang);
389 }
390 if (this.does_stem)
391 {
392 createParameter(FIELD_STEM_PARAM, multiparam, lang);
393 }
394 if (this.does_accent)
395 {
396 createParameter(FIELD_ACCENT_PARAM, multiparam, lang);
397 }
398 if (does_chunking) {
399 createParameter(MAXDOCS_PARAM, param_list, lang);
400 }
401 if (does_paging)
402 {
403 createParameter(HITS_PER_PAGE_PARAM, param_list, lang);
404 createParameter(START_PAGE_PARAM, param_list, lang);
405 }
406 createParameter(FIELD_FIELD_PARAM, multiparam, lang);
407
408 }
409 else if (service_id.equals(RAW_QUERY_SERVICE))
410 {
411 if (does_chunking)
412 {
413 createParameter(MAXDOCS_PARAM, param_list, lang);
414 }
415 if (does_paging)
416 {
417 createParameter(HITS_PER_PAGE_PARAM, param_list, lang);
418 createParameter(START_PAGE_PARAM, param_list, lang);
419 }
420 createParameter(RAW_PARAM, param_list, lang);
421 service.appendChild(param_list);
422 }
423 }
424 return service;
425
426 }
427
428 /** add in the level params to TextQuery */
429 protected void addCustomQueryParams(Element param_list, String lang)
430 {
431 createParameter(LEVEL_PARAM, param_list, lang);
432 super.addCustomQueryParams(param_list, lang);
433 }
434 /** add in the level params to TextQuery */
435 protected void addCustomQueryParamsAdvField(Element param_list, String lang)
436 {
437 createParameter(LEVEL_PARAM, param_list, lang);
438 }
439
440 /** create a param and add to the list */
441 protected void createParameter(String name, Element param_list, String lang)
442 {
443 Document doc = param_list.getOwnerDocument();
444 Element param = null;
445 String param_default = paramDefaults.get(name);
446 if (name.equals(LEVEL_PARAM))
447 {
448 ArrayList<String> level_ids = new ArrayList<String>();
449 ArrayList<String> level_names = new ArrayList<String>();
450 getLevelData(level_ids, level_names, lang);
451 if (level_ids.size() > 1)
452 {
453 // the first one is the default
454 //param = GSXML.createParameterDescription2(doc, LEVEL_PARAM, getTextString("param."+LEVEL_PARAM, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, (String)level_ids.get(0), level_ids, level_names);
455 param = GSXML.createParameterDescription2(doc, LEVEL_PARAM, getTextString("param." + LEVEL_PARAM, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, this.default_level, level_ids, level_names);
456 }
457 else
458 {
459 // we need to set the level, but hidden, in case there is an invalid level saved
460 //param = GSXML.createParameterDescription(doc, LEVEL_PARAM, "", GSXML.PARAM_TYPE_INVISIBLE, (String)level_ids.get(0), null, null);
461 param = GSXML.createParameterDescription(doc, LEVEL_PARAM, "", GSXML.PARAM_TYPE_INVISIBLE, this.default_level, null, null);
462 }
463 }
464 else if (name.equals(RANK_PARAM))
465 {
466 String[] vals1 = { RANK_PARAM_RANK, RANK_PARAM_NONE };
467 String[] vals1_texts = { getTextString("param." + RANK_PARAM + "." + RANK_PARAM_RANK, lang), getTextString("param." + RANK_PARAM + "." + RANK_PARAM_NONE, lang) };
468
469 param = GSXML.createParameterDescription(doc, RANK_PARAM, getTextString("param." + RANK_PARAM, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, param_default, vals1, vals1_texts);
470
471 }
472 else if (name.equals(FIELD_QUERY_PARAM))
473 {
474 param = GSXML.createParameterDescription(doc, FIELD_QUERY_PARAM, getTextString("param." + FIELD_QUERY_PARAM, lang), GSXML.PARAM_TYPE_STRING, null, null, null);
475
476 }
477 else if (name.equals(FIELD_CASE_PARAM) || name.equals(FIELD_STEM_PARAM) || name.equals(FIELD_ACCENT_PARAM))
478 {
479 String[] bool_ops = { "0", "1" };
480 String[] bool_texts = { getTextString("param.boolean.off", lang, "AbstractTextSearch"), getTextString("param.boolean.on", lang, "AbstractTextSearch") };
481 param = GSXML.createParameterDescription(doc, name, getTextString("param." + name, lang), GSXML.PARAM_TYPE_BOOLEAN, param_default, bool_ops, bool_texts);
482
483 }
484 else if (name.equals(FIELD_FIELD_PARAM))
485 {
486 ArrayList<String> fields = new ArrayList<String>();
487 ArrayList<String> field_names = new ArrayList<String>();
488 getIndexData(fields, field_names, lang);
489 // the field list - read from config file
490
491 // Fix for http://trac.greenstone.org/ticket/245 "java crash, index out of bounds"
492 // org.greenstone.gsdl3.service.AbstractGS2FieldSearch.createParameter(AbstractGS2FieldSearch.java:362)
493 // Changed from:
494 // param = GSXML.createParameterDescription2(doc, name, getTextString("param."+name, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, (String)fields.get(0), fields, field_names );
495 String default_value = (fields.size() > 0) ? fields.get(0) : null;
496 // don't want to access element 0 if fields.size()==0, and
497 // GSXML.createParameterDescription2 checks for default_value==null condition
498 param = GSXML.createParameterDescription2(doc, name, getTextString("param." + name, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, default_value, fields, field_names);
499
500 }
501 else if (name.equals(FIELD_COMBINE_PARAM))
502 {
503
504 String[] vals = { FIELD_COMBINE_PARAM_AND, FIELD_COMBINE_PARAM_OR, FIELD_COMBINE_PARAM_NOT };
505 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) };
506
507 param = GSXML.createParameterDescription(doc, FIELD_COMBINE_PARAM, "", GSXML.PARAM_TYPE_ENUM_SINGLE, param_default, vals, val_texts);
508 param.setAttribute(GSXML.PARAM_IGNORE_POS_ATT, "0");
509 }
510
511 if (param != null)
512 {
513 param_list.appendChild(param);
514 }
515 else
516 {
517 super.createParameter(name, param_list, lang);
518 }
519 }
520
521 // should cache some of this
522 protected void getLevelData(ArrayList<String> level_ids, ArrayList<String> level_names, String lang)
523 {
524 Element level_list = (Element) GSXML.getChildByTagName(this.config_info, LEVEL_ELEM + GSXML.LIST_MODIFIER);
525 NodeList levels = level_list.getElementsByTagName(LEVEL_ELEM);
526 for (int i = 0; i < levels.getLength(); i++)
527 {
528 Element level = (Element) levels.item(i);
529 String shortname = level.getAttribute(GSXML.SHORTNAME_ATT);
530 if (shortname.equals(""))
531 {
532 continue;
533 }
534 level_ids.add(shortname);
535 String display_name = GSXML.getDisplayText(level, GSXML.DISPLAY_TEXT_NAME, lang, "en");
536 if (display_name.equals(""))
537 {
538 // we'll use the name, and the dictionary
539 display_name = level.getAttribute(GSXML.NAME_ATT);
540 if (display_name.equals(""))
541 {
542 display_name = shortname;
543 }
544 else
545 {
546 display_name = getTextString("level." + display_name, lang);
547 }
548 }
549 level_names.add(display_name);
550 }
551 }
552
553 // the following three functions are needed so the base class can
554 // call the process+SERVICE_NAME methods
555 /** process a text query */
556 protected Element processTextQuery(Element request)
557 {
558 return processAnyQuery(request, TEXT_QUERY);
559 }
560
561 /** process a field query */
562 protected Element processFieldQuery(Element request)
563 {
564 return processAnyQuery(request, SIMPLE_QUERY);
565 }
566
567 /** process an advanced field query */
568 protected Element processAdvancedFieldQuery(Element request)
569 {
570 return processAnyQuery(request, ADVANCED_QUERY);
571 }
572
573 protected Element processRawQuery(Element request)
574 {
575 return processAnyQuery(request, RAW_QUERY);
576 }
577
578 /** process a query */
579 protected Element processAnyQuery(Element request, int query_type)
580 {
581 String service_name = null;
582 String empty_query_test_param = null;
583 // set up the type specific bits
584 switch (query_type)
585 {
586 case TEXT_QUERY:
587 service_name = QUERY_SERVICE;
588 empty_query_test_param = QUERY_PARAM;
589 break;
590 case SIMPLE_QUERY:
591 service_name = FIELD_QUERY_SERVICE;
592 empty_query_test_param = FIELD_QUERY_PARAM;
593 break;
594 case ADVANCED_QUERY:
595 service_name = ADVANCED_FIELD_QUERY_SERVICE;
596 empty_query_test_param = FIELD_QUERY_PARAM;
597 break;
598 case RAW_QUERY:
599 service_name = RAW_QUERY_SERVICE;
600 empty_query_test_param = RAW_PARAM;
601 break;
602 default:
603 // should never get here
604 logger.error("wrong query type!!");
605 return null;
606 }
607
608 // Create a new (empty) result message
609 Document result_doc = XMLConverter.newDOM();
610 Element result = result_doc.createElement(GSXML.RESPONSE_ELEM);
611 result.setAttribute(GSXML.FROM_ATT, service_name);
612 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
613
614 // Get the parameters of the request
615 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
616 if (param_list == null)
617 {
618 logger.error("TextQuery request had no paramList.");
619 return result; // Return the empty result
620 }
621
622 // Process the request parameters
623 HashMap<String, Serializable> params = GSXML.extractParams(param_list, false);
624
625 // Make sure a query has been specified
626 String query = (String) params.get(empty_query_test_param);
627 if (query == null || query.equals(""))
628 {
629 return result; // Return the empty result
630 }
631
632 // If a field hasn't been specified, use the default - for textQuery
633 String field = (String) params.get(INDEX_PARAM);
634 if (field == null)
635 {
636 field = default_index;
637 }
638
639 // set up the appropriate query system
640 if (!setUpQueryer(params))
641 {
642 return result;
643 }
644
645 // if field search, create the query string
646 switch (query_type)
647 {
648 case TEXT_QUERY:
649 query = addFieldInfo(query, field);
650 break;
651 case SIMPLE_QUERY:
652 query = parseFieldQueryParams(params);
653 break;
654 case ADVANCED_QUERY:
655 query = parseAdvancedFieldQueryParams(params);
656 break;
657 }
658
659 // run the query
660 Object query_result = runQuery(query);
661
662 // build up the response
663
664 // Create a metadata list to store information about the query results
665 // should we be using metadataList? or something else?
666 Element metadata_list = result_doc.createElement(GSXML.METADATA_ELEM + GSXML.LIST_MODIFIER);
667 result.appendChild(metadata_list);
668
669 // Add a metadata element specifying the number of matching documents
670 long totalDocs = numDocsMatched(query_result);
671
672 GSXML.addMetadata(metadata_list, "numDocsMatched", "" + totalDocs);
673
674 // Create a document list to store the matching documents, and add them
675 String[] docs = getDocIDs(query_result);
676 String[] doc_ranks = getDocRanks(query_result);
677
678 // add a metadata item to specify docs returned
679 if (does_chunking) // this means we have a max docs param, and might ask for only a subset of results
680 {
681 logger.error("does_chunking = true");
682 int docs_returned = docs.length;
683 String maxdocs_str = (String) params.get(MAXDOCS_PARAM);
684 if (maxdocs_str != null)
685 {
686 int maxdocs = Integer.parseInt(maxdocs_str);
687 if (maxdocs > 0) { // maxdocs==-1 means return all
688 docs_returned = (maxdocs < (int) totalDocs ? maxdocs : (int) totalDocs);
689 }
690 }
691 GSXML.addMetadata(metadata_list, "numDocsReturned", "" + docs_returned);
692 }
693
694
695 // add a metadata item to specify what actual query was done - eg if stuff was stripped out etc. and then we can use the query later, cos we don't know which parameter was the query
696 GSXML.addMetadata(metadata_list, "query", query);
697 if (docs.length > 0)
698 {
699 Element document_list = result_doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
700 result.appendChild(document_list);
701 for (int d = 0; d < docs.length; d++)
702 {
703 String doc_id = internalNum2OID(docs[d]);
704 Element doc_node = createDocNode(result_doc, doc_id, doc_ranks[d]);
705 document_list.appendChild(doc_node);
706 }
707 }
708
709 // Create a term list to store the term information, and add it
710 Element term_list = result_doc.createElement(GSXML.TERM_ELEM + GSXML.LIST_MODIFIER);
711 result.appendChild(term_list);
712 addTermInfo(term_list, params, query_result);
713
714 if(does_faceting)
715 {
716 ArrayList<FacetWrapper> facets = getFacets(query_result);
717 if(facets != null)
718 {
719 Element facet_list = result_doc.createElement(GSXML.FACET_ELEM + GSXML.LIST_MODIFIER);
720 result.appendChild(facet_list);
721
722 for(FacetWrapper currentFacet : facets)
723 {
724 Element facet_elem = result_doc.createElement(GSXML.FACET_ELEM);
725 facet_elem.setAttribute(GSXML.NAME_ATT, currentFacet.getName());
726 facet_list.appendChild(facet_elem);
727
728 HashMap<String, Long> countMap = currentFacet.getCounts();
729
730 for(String countName : countMap.keySet())
731 {
732 long countValue = countMap.get(countName);
733 if(countValue > 0)
734 {
735 Element count_elem = result_doc.createElement(GSXML.COUNT_ELEM);
736 count_elem.setAttribute(GSXML.NAME_ATT, countName);
737 count_elem.setTextContent("" + countValue);
738
739 facet_elem.appendChild(count_elem);
740 }
741 }
742 }
743 }
744 }
745
746 return result;
747
748 }
749
750 /** methods to handle actually doing the query */
751 /** do any initialisation of the query object */
752 abstract protected boolean setUpQueryer(HashMap<String, Serializable> params);
753
754 /** do the query */
755 abstract protected Object runQuery(String query);
756
757 /** get the total number of docs that match */
758 abstract protected long numDocsMatched(Object query_result);
759
760 /** get the list of doc ids */
761 abstract protected String[] getDocIDs(Object query_result);
762
763 /** get the list of doc ranks */
764 abstract protected String[] getDocRanks(Object query_result);
765
766 /** get the list of facets */
767 abstract protected ArrayList<FacetWrapper> getFacets(Object query_result);
768
769 /** add in term info if available */
770 abstract protected boolean addTermInfo(Element term_list, HashMap<String, Serializable> params, Object query_result);
771
772 /**
773 * combines all the field params into a single query - for simple field
774 * query
775 */
776 /** We assume the combination (AND/OR) is done by the match param */
777 protected String parseFieldQueryParams(HashMap<String, Serializable> params)
778 {
779
780 StringBuffer final_query = new StringBuffer(256);
781 String text_line = (String) params.get(FIELD_QUERY_PARAM);
782 String[] texts = text_line.split(",", -1);
783 String field_line = (String) params.get(FIELD_FIELD_PARAM);
784 String[] fields = field_line.split(",", -1);
785
786 for (int i = 0; i < texts.length; i++)
787 {
788 String q = texts[i].trim();
789 if (!q.equals(""))
790 {
791 final_query.append(" " + addFieldInfo(q, fields[i]));
792 }
793 }
794
795 return final_query.toString();
796 }
797
798 abstract protected String addFieldInfo(String query, String field);
799
800 /**
801 * combines all the field params into a single query - for advanced field
802 * query
803 */
804 protected String parseAdvancedFieldQueryParams(HashMap<String, Serializable> params)
805 {
806
807 StringBuffer final_query = new StringBuffer(256);
808 String text_line = (String) params.get(FIELD_QUERY_PARAM);
809 String[] texts = text_line.split(",", -1);
810 String field_line = (String) params.get(FIELD_FIELD_PARAM);
811 String[] fields = field_line.split(",", -1);
812 String[] cases = null;
813 String[] stems = null;
814 String[] accents = null;
815 if (does_case)
816 {
817 String case_line = (String) params.get(FIELD_CASE_PARAM);
818 if (case_line != null)
819 cases = case_line.split(",", -1);
820 }
821 if (does_stem)
822 {
823 String stem_line = (String) params.get(FIELD_STEM_PARAM);
824 if (stem_line != null)
825 stems = stem_line.split(",", -1);
826 }
827 if (does_accent)
828 {
829 String accent_line = (String) params.get(FIELD_ACCENT_PARAM);
830 if (accent_line != null)
831 accents = accent_line.split(",", -1);
832 }
833 String combine_line = (String) params.get(FIELD_COMBINE_PARAM);
834 String[] combines = combine_line.split(",", -1);
835 String combine = "";
836 for (int i = 0; i < texts.length; i++)
837 {
838 if (i == 0)
839 {// assume first one is blank
840 combine = "";
841 }
842 else
843 {
844 String x = combines[i];
845 if (x.equals(FIELD_COMBINE_PARAM_AND))
846 {
847 combine = AND_OPERATOR;
848 }
849 else if (x.equals(FIELD_COMBINE_PARAM_OR))
850 {
851 combine = OR_OPERATOR;
852 }
853 else if (x.equals(FIELD_COMBINE_PARAM_NOT))
854 {
855 combine = NOT_OPERATOR;
856 }
857
858 }
859
860 String q = texts[i].trim();
861 boolean modified = false;
862 if (!q.equals(""))
863 {
864 String c = null;
865 String s = null;
866 String a = null;
867 if (does_case)
868 {
869 modified = true;
870 c = cases[i];
871 }
872 if (does_stem)
873 {
874 modified = true;
875 s = stems[i];
876 }
877 if (does_accent)
878 {
879 modified = true;
880 a = accents[i];
881 }
882 if (modified)
883 {
884 q = addStemOptions(q, s, c, a);
885 }
886 addQueryElem(final_query, q, fields[i], combine);
887 }
888 }
889 return final_query.toString();
890 }
891
892 abstract protected void addQueryElem(StringBuffer final_query, String query, String field, String combine);
893
894 abstract protected String addStemOptions(String query, String stem, String casef, String accent);
895
896}
Note: See TracBrowser for help on using the repository browser.