source: main/trunk/greenstone3/src/java/org/greenstone/gsdl3/service/AbstractGS2FieldSearch.java@ 32083

Last change on this file since 32083 was 32083, checked in by kjdon, 6 years ago

getFacets now includes a lang arg, so we can get teh correct displayName for the facet. when we get a facet back, look for the display name in it.
  • Property svn:keywords set to Author Date Id Revision
File size: 32.4 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;
24import java.util.List;
25import java.util.Map;
26
27import org.apache.log4j.Logger;
28import org.greenstone.gsdl3.util.FacetWrapper;
29import org.greenstone.gsdl3.util.GSXML;
30import org.greenstone.gsdl3.util.XMLConverter;
31import org.w3c.dom.Document;
32import org.w3c.dom.Element;
33import org.w3c.dom.NodeList;
34
35abstract public class AbstractGS2FieldSearch extends AbstractGS2TextSearch
36{
37 // extra services offered by mgpp collections
38 protected static final String FIELD_QUERY_SERVICE = "FieldQuery";
39 protected static final String ADVANCED_FIELD_QUERY_SERVICE = "AdvancedFieldQuery";
40 protected static final String RAW_QUERY_SERVICE = "RawQuery";
41
42 // extra parameters used
43 protected static final String LEVEL_PARAM = "level";
44 protected static final String RANK_PARAM = "sortBy";
45 protected static final String RANK_PARAM_RANK = "1";
46 protected static final String RANK_PARAM_NONE = "0";
47 protected static final String SIMPLE_FIELD_PARAM = "simpleField";
48 protected static final String ADVANCED_FIELD_PARAM = "complexField";
49
50 protected static final String RAW_PARAM = "rawquery";
51
52 // more params for field query
53 protected static final String FIELD_QUERY_PARAM = "fqv";
54 protected static final String FIELD_STEM_PARAM = "fqs";
55 protected static final String FIELD_CASE_PARAM = "fqc";
56 protected static final String FIELD_ACCENT_PARAM = "fqa";
57 protected static final String FIELD_FIELD_PARAM = "fqf";
58 protected static final String FIELD_COMBINE_PARAM = "fqk";
59 protected static final String FIELD_COMBINE_PARAM_AND = "0";
60 protected static final String FIELD_COMBINE_PARAM_OR = "1";
61 protected static final String FIELD_COMBINE_PARAM_NOT = "2";
62
63 // some stuff for config files
64 protected static final String SEARCH_TYPE_ELEM = "searchType";
65 protected static final String SEARCH_TYPE_PLAIN = "plain";
66 protected static final String SEARCH_TYPE_SIMPLE_FORM = "simpleform";
67 protected static final String SEARCH_TYPE_ADVANCED_FORM = "advancedform";
68 protected static final String SEARCH_TYPE_RAW = "raw";
69
70 protected static final String DEFAULT_LEVEL_ELEM = "defaultLevel";
71 protected static final String DEFAULT_DB_LEVEL_ELEM = "defaultDBLevel";
72 protected static final String LEVEL_ELEM = "level";
73 protected static final String FIELD_ATT = "field";
74
75 protected static final int TEXT_QUERY = 0;
76 protected static final int SIMPLE_QUERY = 1;
77 protected static final int ADVANCED_QUERY = 2;
78 protected static final int RAW_QUERY = 3;
79
80 protected String AND_OPERATOR = "&";
81 protected String OR_OPERATOR = "|";
82 protected String NOT_OPERATOR = "!";
83
84 // the default level for searching
85 protected String default_level = null;
86 // default level for collection db
87 protected String default_db_level = null;
88 // metadata for service, such as does_paging = true
89 protected Element service_metadata_list = null;
90 // which search services will we offer??
91 protected boolean plain_search = false;
92 protected boolean simple_form_search = false;
93 protected boolean advanced_form_search = false;
94 protected boolean raw_search = false;
95 //Parameters to get full field with highlighted terms
96 protected String hldocOID = null;
97 //index Field for highlighting
98 protected String indexField = null;
99
100 static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.AbstractGS2FieldSearch.class.getName());
101
102 /** constructor */
103 public AbstractGS2FieldSearch()
104 {
105 super();
106 paramDefaults.put(RANK_PARAM, RANK_PARAM_RANK);
107 paramDefaults.put(FIELD_COMBINE_PARAM, FIELD_COMBINE_PARAM_AND);
108 // paramDefaults.put(FIELD_CASE_PARAM, BOOLEAN_PARAM_ON);
109 // paramDefaults.put(FIELD_STEM_PARAM, BOOLEAN_PARAM_OFF);
110 // paramDefaults.put(FIELD_ACCENT_PARAM, BOOLEAN_PARAM_ON);
111 }
112
113 public void cleanUp()
114 {
115 super.cleanUp();
116 }
117
118 /** configure this service */
119 public boolean configure(Element info, Element extra_info)
120 {
121 if (!super.configure(info, extra_info))
122 {
123 return false;
124 }
125
126 // Get the default level out of <defaultLevel> (buildConfig.xml)
127 Element def = (Element) GSXML.getChildByTagName(info, DEFAULT_LEVEL_ELEM);
128 if (def != null)
129 {
130 this.default_level = def.getAttribute(GSXML.SHORTNAME_ATT);
131 }
132 if (this.default_level == null || this.default_level.equals(""))
133 {
134 logger.error("default level not specified!, assuming Doc");
135 this.default_level = "Doc";
136 }
137
138 // Get the default DB level
139 def = (Element) GSXML.getChildByTagName(info, DEFAULT_DB_LEVEL_ELEM);
140 if (def != null)
141 {
142 this.default_db_level = def.getAttribute(GSXML.SHORTNAME_ATT);
143 }
144 if (this.default_db_level == null || this.default_db_level.equals(""))
145 {
146 logger.error("default database level (defaultDBLevel) not specified!, assuming Sec");
147 this.default_db_level = "Sec";
148 }
149
150 // get stuff from from extra info (which is collectionConfig.xml)
151 if (extra_info != null)
152 {
153
154 // the search element
155 Element config_search = (Element) GSXML.getChildByTagName(extra_info, GSXML.SEARCH_ELEM);
156
157 NodeList search_types = config_search.getElementsByTagName(SEARCH_TYPE_ELEM);
158 if (search_types == null)
159 {
160 // none specified, assume plain only
161 this.plain_search = true;
162 }
163 else
164 {
165 for (int i = 0; i < search_types.getLength(); i++)
166 {
167 Element t = (Element) search_types.item(i);
168
169 String type_name = t.getAttribute(GSXML.NAME_ATT);
170 if (type_name.equals(SEARCH_TYPE_PLAIN))
171 {
172 this.plain_search = true;
173 }
174 else if (type_name.equals(SEARCH_TYPE_SIMPLE_FORM))
175 {
176 this.simple_form_search = true;
177 }
178 else if (type_name.equals(SEARCH_TYPE_ADVANCED_FORM))
179 {
180 this.advanced_form_search = true;
181 }
182 else if (type_name.equals(SEARCH_TYPE_RAW))
183 {
184 this.raw_search = true;
185 }
186 }
187 }
188
189 // AbstractGS2TextSearch has set up the TextQuery service, but we may not want it
190 if (!this.plain_search)
191 {
192 // need to remove the TextQuery service
193 Element tq_service = GSXML.getNamedElement(short_service_info, GSXML.SERVICE_ELEM, GSXML.NAME_ATT, QUERY_SERVICE);
194 short_service_info.removeChild(tq_service);
195 }
196
197 Document owner = info.getOwnerDocument();
198
199 NodeList levels = info.getElementsByTagName(GSXML.LEVEL_ELEM);
200
201 for (int i = 0; i < levels.getLength(); i++)
202 {
203 Element lev = (Element) levels.item(i);
204 String name = lev.getAttribute(GSXML.NAME_ATT);
205 Element node_extra = GSXML.getNamedElement(config_search, GSXML.LEVEL_ELEM, GSXML.NAME_ATT, name);
206 if (node_extra == null)
207 {
208 logger.error("haven't found extra info for level named " + name);
209 continue;
210 }
211
212 // get the display elements if any - displayName
213 NodeList display_names = node_extra.getElementsByTagName(GSXML.DISPLAY_TEXT_ELEM);
214 if (display_names != null)
215 {
216 for (int j = 0; j < display_names.getLength(); j++)
217 {
218 Element e = (Element) display_names.item(j);
219 lev.appendChild(owner.importNode(e, true));
220 }
221 }
222 } // for each level
223 }
224 else
225 {
226 // for soem reason we don't have the collectionConfig file. assume plain only
227 this.plain_search = true;
228 }
229
230 // the format info is the same for all services
231 Element format_info = (Element) format_info_map.get(QUERY_SERVICE);
232
233 // set up the extra services which are available for this collection
234 if (this.simple_form_search)
235 {
236 // set up short_service_info_ - for now just has id and type - name will be added in on the fly
237 Element fq_service = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
238 fq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
239 fq_service.setAttribute(GSXML.NAME_ATT, FIELD_QUERY_SERVICE);
240 this.short_service_info.appendChild(fq_service);
241
242 if (format_info != null)
243 {
244 this.format_info_map.put(FIELD_QUERY_SERVICE, format_info);
245 }
246 }
247
248 if (this.advanced_form_search)
249 {
250 Element afq_service = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
251 afq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
252 afq_service.setAttribute(GSXML.NAME_ATT, ADVANCED_FIELD_QUERY_SERVICE);
253 this.short_service_info.appendChild(afq_service);
254
255 if (format_info != null)
256 {
257 this.format_info_map.put(ADVANCED_FIELD_QUERY_SERVICE, format_info);
258 }
259 }
260
261 if (this.raw_search)
262 {
263 Element rq_service = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
264 rq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
265 rq_service.setAttribute(GSXML.NAME_ATT, RAW_QUERY_SERVICE);
266 this.short_service_info.appendChild(rq_service);
267
268 if (format_info != null)
269 {
270 this.format_info_map.put(RAW_QUERY_SERVICE, format_info);
271 }
272 }
273
274 return true;
275 }
276
277 // if field case/stem/accent defaults are not specified explicitly, we should use case/stem/accent defaults
278 protected void getSearchParamDefaults(Element search_elem) {
279
280 boolean found_f_stem = false;
281 boolean found_f_case = false;
282 boolean found_f_accent = false;
283 NodeList param_defaults_list = GSXML.getChildrenByTagName(search_elem, GSXML.PARAM_DEFAULT_ELEM);
284 for (int i=0; i<param_defaults_list.getLength(); i++) {
285 Element paramdef = (Element)param_defaults_list.item(i);
286 String name = paramdef.getAttribute(GSXML.NAME_ATT);
287 String val = paramdef.getAttribute(GSXML.VALUE_ATT);
288 if (!name.equals("") && !val.equals("")) {
289 paramDefaults.put(name, val);
290 if (name.equals(FIELD_STEM_PARAM)) {
291 found_f_stem = true;
292 } else if (name.equals(FIELD_CASE_PARAM)) {
293 found_f_case = true;
294 } else if (name.equals(FIELD_ACCENT_PARAM)) {
295 found_f_accent = true;
296 }
297 if (!found_f_stem) {
298 paramDefaults.put (FIELD_STEM_PARAM, paramDefaults.get(STEM_PARAM));
299 }
300 if (!found_f_case) {
301 paramDefaults.put (FIELD_CASE_PARAM, paramDefaults.get(CASE_PARAM));
302 }
303 if (!found_f_accent) {
304 paramDefaults.put (FIELD_ACCENT_PARAM, paramDefaults.get(ACCENT_PARAM));
305 }
306
307 }
308 }
309 }
310 protected Element getServiceDescription(Document doc, String service_id, String lang, String subset)
311 {
312 // should we check that the service is actually on offer? presumably we wont get asked for services that we haven't advertised previously.
313
314 if (!service_id.equals(FIELD_QUERY_SERVICE) && !service_id.equals(ADVANCED_FIELD_QUERY_SERVICE) && !service_id.equals(RAW_QUERY_SERVICE))
315 {
316 return super.getServiceDescription(doc, service_id, lang, subset);
317 }
318
319 Element service = doc.createElement(GSXML.SERVICE_ELEM);
320 service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
321 service.setAttribute(GSXML.NAME_ATT, service_id);
322 if (subset == null || subset.equals(GSXML.DISPLAY_TEXT_ELEM + GSXML.LIST_MODIFIER))
323 {
324 service.appendChild(GSXML.createDisplayTextElement(doc, GSXML.DISPLAY_TEXT_NAME, getTextString(service_id + ".name", lang)));
325 service.appendChild(GSXML.createDisplayTextElement(doc, GSXML.DISPLAY_TEXT_SUBMIT, getTextString(service_id + ".submit", lang)));
326 service.appendChild(GSXML.createDisplayTextElement(doc, GSXML.DISPLAY_TEXT_DESCRIPTION, getTextString(service_id + ".description", lang)));
327
328 }
329 if (subset == null || subset.equals(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER))
330 {
331 Element param_list = doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
332 service.appendChild(param_list);
333 if (service_id.equals(FIELD_QUERY_SERVICE))
334 {
335
336 addCustomQueryParams(param_list, lang);
337 if (!default_index_subcollection.equals(""))
338 {
339 createParameter(INDEX_SUBCOLLECTION_PARAM, param_list, lang);
340 }
341 if (!default_index_language.equals(""))
342 {
343 createParameter(INDEX_LANGUAGE_PARAM, param_list, lang);
344 }
345
346 if (does_chunking) {
347 createParameter(MAXDOCS_PARAM, param_list, lang);
348 }
349 if (does_paging)
350 {
351 createParameter(HITS_PER_PAGE_PARAM, param_list, lang);
352 createParameter(START_PAGE_PARAM, param_list, lang);
353 }
354
355 // create a multi param for the fields etc
356 // text box, field
357 Element multiparam = null;
358 Element param = null;
359 multiparam = GSXML.createParameterDescription(doc, SIMPLE_FIELD_PARAM, "", GSXML.PARAM_TYPE_MULTI, null, null, null);
360 multiparam.setAttribute("occurs", "4");
361 param_list.appendChild(multiparam);
362
363 // the components
364
365 createParameter(FIELD_QUERY_PARAM, multiparam, lang);
366 createParameter(FIELD_FIELD_PARAM, multiparam, lang);
367
368 }
369 else if (service_id.equals(ADVANCED_FIELD_QUERY_SERVICE))
370 {
371 addCustomQueryParamsAdvField(param_list, lang);
372 if (!default_index_subcollection.equals(""))
373 {
374 createParameter(INDEX_SUBCOLLECTION_PARAM, param_list, lang);
375 }
376 if (!default_index_language.equals(""))
377 {
378 createParameter(INDEX_LANGUAGE_PARAM, param_list, lang);
379 }
380
381 // create a multi param for the fields etc
382 // text box, stem, case, field
383
384 Element multiparam = null;
385 Element param = null;
386
387 multiparam = GSXML.createParameterDescription(doc, ADVANCED_FIELD_PARAM, "", GSXML.PARAM_TYPE_MULTI, null, null, null);
388 multiparam.setAttribute("occurs", "4");
389 param_list.appendChild(multiparam);
390
391 createParameter(FIELD_COMBINE_PARAM, multiparam, lang);
392 createParameter(FIELD_QUERY_PARAM, multiparam, lang);
393 if (this.does_case)
394 {
395 createParameter(FIELD_CASE_PARAM, multiparam, lang);
396 }
397 if (this.does_stem)
398 {
399 createParameter(FIELD_STEM_PARAM, multiparam, lang);
400 }
401 if (this.does_accent)
402 {
403 createParameter(FIELD_ACCENT_PARAM, multiparam, lang);
404 }
405 if (does_chunking) {
406 createParameter(MAXDOCS_PARAM, param_list, lang);
407 }
408 if (does_paging)
409 {
410 createParameter(HITS_PER_PAGE_PARAM, param_list, lang);
411 createParameter(START_PAGE_PARAM, param_list, lang);
412 }
413 createParameter(FIELD_FIELD_PARAM, multiparam, lang);
414
415 }
416 else if (service_id.equals(RAW_QUERY_SERVICE))
417 {
418 if (does_chunking)
419 {
420 createParameter(MAXDOCS_PARAM, param_list, lang);
421 }
422 if (does_paging)
423 {
424 createParameter(HITS_PER_PAGE_PARAM, param_list, lang);
425 createParameter(START_PAGE_PARAM, param_list, lang);
426 }
427 createParameter(RAW_PARAM, param_list, lang);
428 service.appendChild(param_list);
429 }
430 }
431 if (subset == null || subset.equals(GSXML.METADATA_ELEM + GSXML.LIST_MODIFIER)) {
432 if (service_metadata_list == null) {
433 Document ml_doc = XMLConverter.newDOM();
434 service_metadata_list = ml_doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
435 if (does_paging) {
436 service_metadata_list.appendChild(GSXML.createMetadataElement(ml_doc, "does_paging", "true"));
437 }
438 }
439 service.appendChild(doc.importNode(service_metadata_list, true));
440 }
441 return service;
442
443 }
444
445 /** add in the level params to TextQuery */
446 protected void addCustomQueryParams(Element param_list, String lang)
447 {
448 createParameter(LEVEL_PARAM, param_list, lang);
449 super.addCustomQueryParams(param_list, lang);
450 }
451 /** add in the level params to TextQuery */
452 protected void addCustomQueryParamsAdvField(Element param_list, String lang)
453 {
454 createParameter(LEVEL_PARAM, param_list, lang);
455 }
456
457 /** create a param and add to the list */
458 protected void createParameter(String name, Element param_list, String lang)
459 {
460 Document doc = param_list.getOwnerDocument();
461 Element param = null;
462 String param_default = paramDefaults.get(name);
463 if (name.equals(LEVEL_PARAM))
464 {
465 ArrayList<String> level_ids = new ArrayList<String>();
466 ArrayList<String> level_names = new ArrayList<String>();
467 getLevelData(level_ids, level_names, lang);
468 if (level_ids.size() > 1)
469 {
470 // the first one is the default
471 //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);
472 param = GSXML.createParameterDescription2(doc, LEVEL_PARAM, getTextString("param." + LEVEL_PARAM, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, this.default_level, level_ids, level_names);
473 }
474 else
475 {
476 // we need to set the level, but hidden, in case there is an invalid level saved
477 //param = GSXML.createParameterDescription(doc, LEVEL_PARAM, "", GSXML.PARAM_TYPE_INVISIBLE, (String)level_ids.get(0), null, null);
478 param = GSXML.createParameterDescription(doc, LEVEL_PARAM, "", GSXML.PARAM_TYPE_INVISIBLE, this.default_level, null, null);
479 }
480 }
481 else if (name.equals(RANK_PARAM))
482 {
483 String[] vals1 = { RANK_PARAM_RANK, RANK_PARAM_NONE };
484 String[] vals1_texts = { getTextString("param." + RANK_PARAM + "." + RANK_PARAM_RANK, lang), getTextString("param." + RANK_PARAM + "." + RANK_PARAM_NONE, lang) };
485
486 param = GSXML.createParameterDescription(doc, RANK_PARAM, getTextString("param." + RANK_PARAM, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, param_default, vals1, vals1_texts);
487
488 }
489 else if (name.equals(FIELD_QUERY_PARAM))
490 {
491 param = GSXML.createParameterDescription(doc, FIELD_QUERY_PARAM, getTextString("param." + FIELD_QUERY_PARAM, lang), GSXML.PARAM_TYPE_STRING, null, null, null);
492
493 }
494 else if (name.equals(FIELD_CASE_PARAM) || name.equals(FIELD_STEM_PARAM) || name.equals(FIELD_ACCENT_PARAM))
495 {
496 String[] bool_ops = { "0", "1" };
497 String[] bool_texts = { getTextString("param.boolean.off", lang), getTextString("param.boolean.on", lang) };
498 param = GSXML.createParameterDescription(doc, name, getTextString("param." + name, lang), GSXML.PARAM_TYPE_BOOLEAN, param_default, bool_ops, bool_texts);
499
500 }
501 else if (name.equals(FIELD_FIELD_PARAM))
502 {
503 ArrayList<String> fields = new ArrayList<String>();
504 ArrayList<String> field_names = new ArrayList<String>();
505 getIndexData(fields, field_names, lang);
506 // the field list - read from config file
507
508 // Fix for http://trac.greenstone.org/ticket/245 "java crash, index out of bounds"
509 // org.greenstone.gsdl3.service.AbstractGS2FieldSearch.createParameter(AbstractGS2FieldSearch.java:362)
510 // Changed from:
511 // param = GSXML.createParameterDescription2(doc, name, getTextString("param."+name, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, (String)fields.get(0), fields, field_names );
512 String default_value = (fields.size() > 0) ? fields.get(0) : null;
513 // don't want to access element 0 if fields.size()==0, and
514 // GSXML.createParameterDescription2 checks for default_value==null condition
515 param = GSXML.createParameterDescription2(doc, name, getTextString("param." + name, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, default_value, fields, field_names);
516
517 }
518 else if (name.equals(FIELD_COMBINE_PARAM))
519 {
520
521 String[] vals = { FIELD_COMBINE_PARAM_AND, FIELD_COMBINE_PARAM_OR, FIELD_COMBINE_PARAM_NOT };
522 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) };
523
524 param = GSXML.createParameterDescription(doc, FIELD_COMBINE_PARAM, "", GSXML.PARAM_TYPE_ENUM_SINGLE, param_default, vals, val_texts);
525 param.setAttribute(GSXML.PARAM_IGNORE_POS_ATT, "0");
526 }
527
528 if (param != null)
529 {
530 param_list.appendChild(param);
531 }
532 else
533 {
534 super.createParameter(name, param_list, lang);
535 }
536 }
537
538 // should cache some of this
539 protected void getLevelData(ArrayList<String> level_ids, ArrayList<String> level_names, String lang)
540 {
541 Element level_list = (Element) GSXML.getChildByTagName(this.config_info, LEVEL_ELEM + GSXML.LIST_MODIFIER);
542 NodeList levels = level_list.getElementsByTagName(LEVEL_ELEM);
543 for (int i = 0; i < levels.getLength(); i++)
544 {
545 Element level = (Element) levels.item(i);
546 String shortname = level.getAttribute(GSXML.SHORTNAME_ATT);
547 if (shortname.equals(""))
548 {
549 continue;
550 }
551 level_ids.add(shortname);
552 String display_name = getDisplayText(level, GSXML.DISPLAY_TEXT_NAME, lang, "en");
553 if (display_name.equals(""))
554 {
555 // we'll use the name, and look it up in the dictionary
556 display_name = level.getAttribute(GSXML.NAME_ATT);
557 if (display_name.equals(""))
558 {
559 display_name = shortname;
560 }
561 else
562 {
563 display_name = getTextString("level." + display_name, lang);
564 }
565 }
566 level_names.add(display_name);
567 }
568 }
569
570 // the following three functions are needed so the base class can
571 // call the process+SERVICE_NAME methods
572 /** process a text query */
573 protected Element processTextQuery(Element request)
574 {
575 return processAnyQuery(request, TEXT_QUERY);
576 }
577
578 /** process a field query */
579 protected Element processFieldQuery(Element request)
580 {
581 return processAnyQuery(request, SIMPLE_QUERY);
582 }
583
584 /** process an advanced field query */
585 protected Element processAdvancedFieldQuery(Element request)
586 {
587 return processAnyQuery(request, ADVANCED_QUERY);
588 }
589
590 protected Element processRawQuery(Element request)
591 {
592 return processAnyQuery(request, RAW_QUERY);
593 }
594
595 /** process a query */
596 protected Element processAnyQuery(Element request, int query_type)
597 {
598
599 String service_name = null;
600 String empty_query_test_param = null;
601 // set up the type specific bits
602 switch (query_type)
603 {
604 case TEXT_QUERY:
605 service_name = QUERY_SERVICE;
606 empty_query_test_param = QUERY_PARAM;
607 break;
608 case SIMPLE_QUERY:
609 service_name = FIELD_QUERY_SERVICE;
610 empty_query_test_param = FIELD_QUERY_PARAM;
611 break;
612 case ADVANCED_QUERY:
613 service_name = ADVANCED_FIELD_QUERY_SERVICE;
614 empty_query_test_param = FIELD_QUERY_PARAM;
615 break;
616 case RAW_QUERY:
617 service_name = RAW_QUERY_SERVICE;
618 empty_query_test_param = RAW_PARAM;
619 break;
620 default:
621 // should never get here
622 logger.error("wrong query type!!");
623 return null;
624 }
625
626 // Create a new (empty) result message
627 Document result_doc = XMLConverter.newDOM();
628 Element result = result_doc.createElement(GSXML.RESPONSE_ELEM);
629 result.setAttribute(GSXML.FROM_ATT, service_name);
630 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
631
632 // Get the parameters of the request
633 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
634 if (param_list == null)
635 {
636 logger.error("TextQuery request had no paramList.");
637 return result; // Return the empty result
638 }
639
640 // Process the request parameters
641
642 HashMap<String, Serializable> params = GSXML.extractParams(param_list, false);
643 //Save as variable to make it accessable from GS2SolrSearch
644 hldocOID = (String) params.get("hldocOID");
645 //System.out.println("@@@@@Extracted hldocOID is " + hldocOID);
646 // Make sure a query has been specified
647 String query = (String) params.get(empty_query_test_param);
648 if (query == null || query.equals(""))
649 {
650 return result; // Return the empty result
651 }
652
653 // If a field hasn't been specified, use the default - for textQuery
654 String field = (String) params.get(INDEX_PARAM);
655 //Save as variable to make it accessable from GS2SolrSearch
656
657 if (field == null)
658 {
659 field = default_index;
660 }
661 indexField = field;
662 // set up the appropriate query system
663 if (!setUpQueryer(params))
664 {
665 return result;
666 }
667
668 // if field search, create the query string
669 switch (query_type)
670 {
671 case TEXT_QUERY:
672 query = addFieldInfo(query, field);
673 break;
674 case SIMPLE_QUERY:
675 query = parseFieldQueryParams(params);
676 break;
677 case ADVANCED_QUERY:
678 query = parseAdvancedFieldQueryParams(params);
679 break;
680 }
681
682 // run the query
683 Object query_result = runQuery(query);
684
685 // We want highlighted text to be returned right now!
686 if (hldocOID != null && does_full_field_highlighting)
687 {
688 Element nodeContentElement = result_doc.createElement(GSXML.NODE_CONTENT_ELEM);
689 nodeContentElement.setTextContent((String) query_result);
690 result.appendChild(nodeContentElement);
691 return result;
692 }
693
694
695 // build up the response
696
697 // Create a metadata list to store information about the query results
698 // should we be using metadataList? or something else?
699 Element metadata_list = result_doc.createElement(GSXML.METADATA_ELEM + GSXML.LIST_MODIFIER);
700 result.appendChild(metadata_list);
701
702 // Add a metadata element specifying the number of matching documents
703 long totalDocs = numDocsMatched(query_result);
704
705 GSXML.addMetadata(metadata_list, "numDocsMatched", "" + totalDocs);
706
707 // Create a document list to store the matching documents, and add them
708 String[] docs = getDocIDs(query_result);
709 String[] doc_ranks = getDocRanks(query_result);
710
711 // add a metadata item to specify docs returned
712 if (does_chunking) // this means we have a max docs param, and might ask for only a subset of results
713 {
714 logger.error("does_chunking = true");
715 int docs_returned = docs.length;
716 String maxdocs_str = (String) params.get(MAXDOCS_PARAM);
717 if (maxdocs_str != null)
718 {
719 int maxdocs = Integer.parseInt(maxdocs_str);
720 if (maxdocs > 0) { // maxdocs==-1 means return all
721 docs_returned = (maxdocs < (int) totalDocs ? maxdocs : (int) totalDocs);
722 }
723 }
724 GSXML.addMetadata(metadata_list, "numDocsReturned", "" + docs_returned);
725 }
726
727 Map<String, Map<String, List<String>>> hlResults = null;
728 if (does_highlight_snippets)
729 {
730 hlResults = getHighlightSnippets(query_result);
731 }
732
733 // 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
734 GSXML.addMetadata(metadata_list, "query", query);
735 if (docs.length > 0)
736 {
737 Element document_list = result_doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
738 result.appendChild(document_list);
739 Element snippet_list = result_doc.createElement(GSXML.HL_SNIPPET_ELEM + GSXML.LIST_MODIFIER);
740 result.appendChild(snippet_list);
741 for (int d = 0; d < docs.length; d++)
742 {
743 String doc_id = internalNum2OID(docs[d]);
744 Element doc_node = createDocNode(result_doc, doc_id, doc_ranks[d]);
745 if (hlResults != null && hlResults.get(docs[d]) != null && hlResults.get(docs[d]).get(indexField) != null) {
746 for (String snippet : hlResults.get(docs[d]).get(indexField)){
747 //remove html tags
748 snippet = snippet.replaceAll("\\<.*?>", "");
749 //remove truncated tags
750 snippet = snippet.replaceAll(".*>", "");
751 snippet = snippet.replaceAll("\\<.*", "");
752 //remove unwanted symbols at start of line
753 snippet = snippet.replaceAll("^[ .,»):;-–]+", "");
754 //highlighting tags transformation
755 snippet = snippet.replaceAll("&lt;", "<");
756 snippet = snippet.replaceAll("&gt;", ">");
757 Element snippet_node = result_doc.createElement(GSXML.HL_SNIPPET_ELEM);
758 snippet_node.setAttribute(GSXML.NODE_ID_ATT, doc_id);
759 snippet_node.setTextContent(snippet);
760 snippet_list.appendChild(snippet_node);
761 }
762 }
763 document_list.appendChild(doc_node);
764 }
765 }
766
767 // Create a term list to store the term information, and add it
768 Element term_list = result_doc.createElement(GSXML.TERM_ELEM + GSXML.LIST_MODIFIER);
769 result.appendChild(term_list);
770 addTermInfo(term_list, params, query_result);
771
772 if(does_faceting)
773 {
774 String lang = request.getAttribute(GSXML.LANG_ATT);
775 ArrayList<FacetWrapper> facets = getFacets(query_result, lang);
776 if(facets != null)
777 {
778 Element facet_list = result_doc.createElement(GSXML.FACET_ELEM + GSXML.LIST_MODIFIER);
779 result.appendChild(facet_list);
780
781 for(FacetWrapper currentFacet : facets)
782 {
783 Element facet_elem = result_doc.createElement(GSXML.FACET_ELEM);
784 facet_elem.setAttribute(GSXML.NAME_ATT, currentFacet.getName());
785 String display_name = currentFacet.getDisplayName();
786 if (display_name != null && !display_name.equals("")) {
787 facet_elem.appendChild(GSXML.createDisplayTextElement(result_doc, GSXML.DISPLAY_TEXT_NAME, display_name));
788 }
789 facet_list.appendChild(facet_elem);
790
791 HashMap<String, Long> countMap = currentFacet.getCounts();
792
793 for(String countName : countMap.keySet())
794 {
795 long countValue = countMap.get(countName);
796 if(countValue > 0)
797 {
798 Element count_elem = result_doc.createElement(GSXML.COUNT_ELEM);
799 count_elem.setAttribute(GSXML.NAME_ATT, countName);
800 count_elem.setTextContent("" + countValue);
801
802 facet_elem.appendChild(count_elem);
803 }
804 }
805 }
806 }
807 }
808
809
810 return result;
811
812 }
813
814 /** methods to handle actually doing the query */
815 /** do any initialisation of the query object */
816 abstract protected boolean setUpQueryer(HashMap<String, Serializable> params);
817
818 /** do the query */
819 abstract protected Object runQuery(String query);
820
821 /** get the total number of docs that match */
822 abstract protected long numDocsMatched(Object query_result);
823
824 /** get the list of doc ids */
825 abstract protected String[] getDocIDs(Object query_result);
826
827 /** get the list of doc ranks */
828 abstract protected String[] getDocRanks(Object query_result);
829
830 /** get the list of facets */
831 abstract protected ArrayList<FacetWrapper> getFacets(Object query_result, String lang);
832
833 /** get the map of highlighting snippets */
834 abstract protected Map<String, Map<String, List<String>>> getHighlightSnippets(Object query_result);
835
836 /** add in term info if available */
837 abstract protected boolean addTermInfo(Element term_list, HashMap<String, Serializable> params, Object query_result);
838
839 /**
840 * combines all the field params into a single query - for simple field
841 * query
842 */
843 /** We assume the combination (AND/OR) is done by the match param */
844 protected String parseFieldQueryParams(HashMap<String, Serializable> params)
845 {
846
847 StringBuffer final_query = new StringBuffer(256);
848 String text_line = (String) params.get(FIELD_QUERY_PARAM);
849 String[] texts = text_line.split(",", -1);
850 String field_line = (String) params.get(FIELD_FIELD_PARAM);
851 String[] fields = field_line.split(",", -1);
852
853 for (int i = 0; i < texts.length; i++)
854 {
855 String q = texts[i].trim();
856 if (!q.equals(""))
857 {
858 final_query.append(" " + addFieldInfo(q, fields[i]));
859 }
860 }
861
862 return final_query.toString();
863 }
864
865 abstract protected String addFieldInfo(String query, String field);
866
867 /**
868 * combines all the field params into a single query - for advanced field
869 * query
870 */
871 protected String parseAdvancedFieldQueryParams(HashMap<String, Serializable> params)
872 {
873
874 StringBuffer final_query = new StringBuffer(256);
875 String text_line = (String) params.get(FIELD_QUERY_PARAM);
876 String[] texts = text_line.split(",", -1);
877 String field_line = (String) params.get(FIELD_FIELD_PARAM);
878 String[] fields = field_line.split(",", -1);
879 String[] cases = null;
880 String[] stems = null;
881 String[] accents = null;
882 if (does_case)
883 {
884 String case_line = (String) params.get(FIELD_CASE_PARAM);
885 if (case_line != null)
886 cases = case_line.split(",", -1);
887 }
888 if (does_stem)
889 {
890 String stem_line = (String) params.get(FIELD_STEM_PARAM);
891 if (stem_line != null)
892 stems = stem_line.split(",", -1);
893 }
894 if (does_accent)
895 {
896 String accent_line = (String) params.get(FIELD_ACCENT_PARAM);
897 if (accent_line != null)
898 accents = accent_line.split(",", -1);
899 }
900 String combine_line = (String) params.get(FIELD_COMBINE_PARAM);
901 String[] combines = combine_line.split(",", -1);
902 String combine = "";
903 for (int i = 0; i < texts.length; i++)
904 {
905 if (i == 0)
906 {// assume first one is blank
907 combine = "";
908 }
909 else
910 {
911 String x = combines[i];
912 if (x.equals(FIELD_COMBINE_PARAM_AND))
913 {
914 combine = AND_OPERATOR;
915 }
916 else if (x.equals(FIELD_COMBINE_PARAM_OR))
917 {
918 combine = OR_OPERATOR;
919 }
920 else if (x.equals(FIELD_COMBINE_PARAM_NOT))
921 {
922 combine = NOT_OPERATOR;
923 }
924
925 }
926
927 String q = texts[i].trim();
928 boolean modified = false;
929 if (!q.equals(""))
930 {
931 String c = null;
932 String s = null;
933 String a = null;
934 if (does_case)
935 {
936 modified = true;
937 c = cases[i];
938 }
939 if (does_stem)
940 {
941 modified = true;
942 s = stems[i];
943 }
944 if (does_accent)
945 {
946 modified = true;
947 a = accents[i];
948 }
949 if (modified)
950 {
951 q = addStemOptions(q, s, c, a);
952 }
953 addQueryElem(final_query, q, fields[i], combine);
954 }
955 }
956 return final_query.toString();
957 }
958
959 abstract protected void addQueryElem(StringBuffer final_query, String query, String field, String combine);
960
961 abstract protected String addStemOptions(String query, String stem, String casef, String accent);
962
963}
Note: See TracBrowser for help on using the repository browser.