[13242] | 1 | /*
|
---|
| 2 | * AbstractMGPPSearch.java
|
---|
| 3 | * Copyright (C) 2005 New Zealand Digital Library, http://www.nzdl.org
|
---|
| 4 | *
|
---|
| 5 | * This program is free software; you can redistribute it and/or modify
|
---|
| 6 | * it under the terms of the GNU General Public License as published by
|
---|
| 7 | * the Free Software Foundation; either version 2 of the License, or
|
---|
| 8 | * (at your option) any later version.
|
---|
| 9 | *
|
---|
| 10 | * This program is distributed in the hope that it will be useful,
|
---|
| 11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
| 12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
| 13 | * GNU General Public License for more details.
|
---|
| 14 | *
|
---|
| 15 | * You should have received a copy of the GNU General Public License
|
---|
| 16 | * along with this program; if not, write to the Free Software
|
---|
| 17 | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
| 18 | */
|
---|
| 19 | package org.greenstone.gsdl3.service;
|
---|
| 20 |
|
---|
| 21 | // greenstone classes
|
---|
| 22 | import org.greenstone.mgpp.*;
|
---|
| 23 | import org.greenstone.gsdl3.util.GSXML;
|
---|
| 24 | import org.greenstone.gsdl3.util.GSFile;
|
---|
| 25 |
|
---|
| 26 | // XML classes
|
---|
| 27 | import org.w3c.dom.Document;
|
---|
| 28 | import org.w3c.dom.Element;
|
---|
| 29 | import org.w3c.dom.NodeList;
|
---|
| 30 |
|
---|
| 31 | // java classes
|
---|
| 32 | import java.util.Iterator;
|
---|
| 33 | import java.util.Set;
|
---|
| 34 | import java.util.HashMap;
|
---|
| 35 | import java.util.Map;
|
---|
| 36 | import java.util.ArrayList;
|
---|
| 37 | import java.util.Vector;
|
---|
| 38 | import java.io.File;
|
---|
| 39 |
|
---|
| 40 | import org.apache.log4j.*;
|
---|
| 41 |
|
---|
| 42 | /** Partially implements a generic MGPP search service
|
---|
| 43 | *
|
---|
| 44 | * @author <a href="mailto:[email protected]">Katherine Don</a>
|
---|
| 45 | */
|
---|
| 46 |
|
---|
| 47 | abstract public class AbstractMGPPSearch
|
---|
| 48 | extends AbstractSearch
|
---|
| 49 | {
|
---|
| 50 |
|
---|
| 51 | static Category logger = Category.getInstance(org.greenstone.gsdl3.service.AbstractMGPPSearch.class.getName());
|
---|
| 52 |
|
---|
| 53 | // extra services offered by mgpp collections
|
---|
| 54 | private static final String FIELD_QUERY_SERVICE = "FieldQuery";
|
---|
| 55 | private static final String ADVANCED_FIELD_QUERY_SERVICE = "AdvancedFieldQuery";
|
---|
| 56 |
|
---|
| 57 | // extra parameters used
|
---|
| 58 | // private static final String INDEX_FIELD_PARAM = "index";
|
---|
| 59 | private static final String LEVEL_PARAM = "level";
|
---|
| 60 | private static final String RANK_PARAM = "sortBy";
|
---|
| 61 | private static final String RANK_PARAM_RANK = "1";
|
---|
| 62 | private static final String RANK_PARAM_NONE = "0";
|
---|
| 63 | private static final String SIMPLE_FIELD_PARAM = "simpleField";
|
---|
| 64 | private static final String ADVANCED_FIELD_PARAM = "complexField";
|
---|
| 65 |
|
---|
| 66 | // more params for field query
|
---|
| 67 | private static final String FIELD_QUERY_PARAM = "fqv";
|
---|
| 68 | private static final String FIELD_STEM_PARAM = "fqs";
|
---|
| 69 | private static final String FIELD_CASE_PARAM = "fqc";
|
---|
| 70 | private static final String FIELD_ACCENT_PARAM="fqa";
|
---|
| 71 | private static final String FIELD_FIELD_PARAM = "fqf";
|
---|
| 72 | private static final String FIELD_COMBINE_PARAM = "fqk";
|
---|
| 73 | private static final String FIELD_COMBINE_PARAM_AND = "0";
|
---|
| 74 | private static final String FIELD_COMBINE_PARAM_OR = "1";
|
---|
| 75 | private static final String FIELD_COMBINE_PARAM_NOT = "2";
|
---|
| 76 |
|
---|
| 77 | private static final String ACCENT_PARAM="accent";
|
---|
| 78 |
|
---|
| 79 | // some stuff for config files
|
---|
| 80 | private static final String SEARCH_TYPE_ELEM = "searchType";
|
---|
| 81 | private static final String SEARCH_TYPE_PLAIN = "plain";
|
---|
| 82 | private static final String SEARCH_TYPE_FORM = "form";
|
---|
| 83 | private static final String SEARCH_TYPE_FORM_SIMPLE = "simple";
|
---|
| 84 | private static final String SEARCH_TYPE_FORM_ADVANCED = "advanced";
|
---|
| 85 | protected static final String DEFAULT_INDEX_ELEM = "defaultIndex";
|
---|
| 86 | protected static final String DEFAULT_LEVEL_ELEM = "defaultLevel";
|
---|
| 87 | protected static final String LEVEL_ELEM = "level";
|
---|
| 88 |
|
---|
| 89 | protected static final String STEMINDEX_OPTION = "stemIndexes";
|
---|
| 90 | protected static final String MAXNUMERIC_OPTION = "maxnumeric";
|
---|
| 91 |
|
---|
| 92 | protected static final String EQUIV_TERM_ELEM = "equivTerm";
|
---|
| 93 |
|
---|
| 94 | protected static final String STEM_ATT = "stem";
|
---|
| 95 | protected static final String NUM_DOCS_MATCH_ATT = "numDocsMatch";
|
---|
| 96 | protected static final String FREQ_ATT = "freq";
|
---|
| 97 |
|
---|
| 98 | private static final int TEXT_QUERY = 0;
|
---|
| 99 | private static final int SIMPLE_QUERY = 1;
|
---|
| 100 | private static final int ADVANCED_QUERY = 2;
|
---|
| 101 |
|
---|
| 102 | protected static final String FIELD_ATT = "field";
|
---|
| 103 | private MGPPWrapper mgpp_src=null;
|
---|
| 104 |
|
---|
| 105 | // the default level for retrieval - and we'll use it for searching too
|
---|
| 106 | private String default_level=null;
|
---|
| 107 | // the default field for searching
|
---|
| 108 | private String default_field = null;
|
---|
| 109 | // which search services will we offer??
|
---|
| 110 | private boolean plain_search = false;
|
---|
| 111 | private boolean simple_form_search = false;
|
---|
| 112 | private boolean advanced_form_search = false;
|
---|
| 113 |
|
---|
| 114 | // stem indexes available
|
---|
| 115 | private boolean does_case=true;
|
---|
| 116 | private boolean does_stem=true;
|
---|
| 117 | private boolean does_accent=false;
|
---|
| 118 |
|
---|
| 119 | // maxnumeric - not used yet. needs to be passed to MGPPWrapper.
|
---|
| 120 | private int maxnumeric = 4;
|
---|
| 121 |
|
---|
| 122 | /** the stem used for the index files */
|
---|
| 123 | protected String index_stem = null;
|
---|
| 124 |
|
---|
| 125 | public AbstractMGPPSearch()
|
---|
| 126 | {
|
---|
| 127 | this.mgpp_src = new MGPPWrapper();
|
---|
| 128 | }
|
---|
| 129 |
|
---|
| 130 | public void cleanUp() {
|
---|
| 131 | super.cleanUp();
|
---|
| 132 | this.mgpp_src.unloadIndexData();
|
---|
| 133 | }
|
---|
| 134 |
|
---|
| 135 | public boolean configure(Element info, Element extra_info)
|
---|
| 136 | {
|
---|
| 137 | // Do generic configuration
|
---|
| 138 | if (super.configure(info, extra_info) == false)
|
---|
| 139 | return false;
|
---|
| 140 |
|
---|
| 141 | // Do specific configuration
|
---|
| 142 | logger.info("Configuring AbstractMGPPSearch...");
|
---|
| 143 |
|
---|
| 144 | // do we support any of the extended features?
|
---|
| 145 | does_chunking = true;
|
---|
| 146 | // Get the default index out of <defaultIndex> (buildConfig.xml)
|
---|
| 147 | Element def = (Element) GSXML.getChildByTagName(info, DEFAULT_INDEX_ELEM);
|
---|
| 148 | if (def != null) {
|
---|
| 149 | this.default_index = def.getAttribute(GSXML.NAME_ATT);
|
---|
| 150 | } // otherwise its "" and will be the first one in the list
|
---|
| 151 |
|
---|
| 152 | // the index stem is either specified in the config file or is "index"
|
---|
| 153 | Element index_stem_elem = (Element) GSXML.getChildByTagName(info, GSXML.INDEX_STEM_ELEM);
|
---|
| 154 | if (index_stem_elem != null) {
|
---|
| 155 | this.index_stem = index_stem_elem.getAttribute(GSXML.NAME_ATT);
|
---|
| 156 | }
|
---|
| 157 | if (this.index_stem == null || this.index_stem.equals("")) {
|
---|
| 158 | logger.warn("AbstractMGPPSearch.configure(): indexStem element not found, stem will default to collection name");
|
---|
| 159 | this.index_stem = this.cluster_name;
|
---|
| 160 | }
|
---|
| 161 |
|
---|
| 162 | // get index options
|
---|
| 163 | Element index_option_list = (Element) GSXML.getChildByTagName(info, GSXML.INDEX_OPTION_ELEM + GSXML.LIST_MODIFIER);
|
---|
| 164 | if (index_option_list != null) {
|
---|
| 165 | NodeList options = index_option_list.getElementsByTagName(GSXML.INDEX_OPTION_ELEM);
|
---|
| 166 | for (int i=0; i<options.getLength(); i++) {
|
---|
| 167 | Element opt = (Element)options.item(i);
|
---|
| 168 | String name = opt.getAttribute(GSXML.NAME_ATT);
|
---|
| 169 | String value = opt.getAttribute(GSXML.VALUE_ATT);
|
---|
| 170 | if (name.equals(MAXNUMERIC_OPTION)) {
|
---|
| 171 | int maxnum = Integer.parseInt(value);
|
---|
| 172 | if (4 <= maxnum && maxnum < 512) {
|
---|
| 173 | maxnumeric = maxnum;
|
---|
| 174 | }
|
---|
| 175 | }
|
---|
| 176 | else if (name.equals(STEMINDEX_OPTION)) {
|
---|
| 177 | int stemindex = Integer.parseInt(value);
|
---|
| 178 | // stem and case are true by default, accent folding false by defualt
|
---|
| 179 | if ((stemindex & 1) == 0) {
|
---|
| 180 | does_case = false;
|
---|
| 181 | }
|
---|
| 182 | if ((stemindex & 2) == 0) {
|
---|
| 183 | does_stem = false;
|
---|
| 184 | }
|
---|
| 185 | if ((stemindex & 4) != 0) {
|
---|
| 186 | does_accent = true;
|
---|
| 187 | }
|
---|
| 188 | }
|
---|
| 189 | }
|
---|
| 190 | }
|
---|
| 191 |
|
---|
| 192 | // the generic config has set up the text query service, but we may not want it
|
---|
| 193 | Element search_type_list = (Element) GSXML.getChildByTagName(info, SEARCH_TYPE_ELEM + GSXML.LIST_MODIFIER);
|
---|
| 194 | if (search_type_list == null) {
|
---|
| 195 | // assume form and plain
|
---|
| 196 | this.plain_search = true;
|
---|
| 197 | this.simple_form_search = true;
|
---|
| 198 | this.advanced_form_search = true;
|
---|
| 199 | } else {
|
---|
| 200 | NodeList types = search_type_list.getElementsByTagName(SEARCH_TYPE_ELEM);
|
---|
| 201 | for (int i=0; i<types.getLength(); i++) {
|
---|
| 202 | Element t = (Element)types.item(i);
|
---|
| 203 | String type_name = t.getAttribute(GSXML.NAME_ATT);
|
---|
| 204 | if (type_name.equals(SEARCH_TYPE_PLAIN)) {
|
---|
| 205 | this.plain_search = true;
|
---|
| 206 | } else if (type_name.equals(SEARCH_TYPE_FORM)) {
|
---|
| 207 | String type_type = t.getAttribute(GSXML.TYPE_ATT);
|
---|
| 208 | if (type_type.equals("")) {
|
---|
| 209 | this.simple_form_search = true;
|
---|
| 210 | this.advanced_form_search = true;
|
---|
| 211 | } else if (type_type.equals(SEARCH_TYPE_FORM_SIMPLE)) {
|
---|
| 212 | this.simple_form_search = true;
|
---|
| 213 | } else if (type_type.equals(SEARCH_TYPE_FORM_ADVANCED)) {
|
---|
| 214 | this.advanced_form_search = true;
|
---|
| 215 |
|
---|
| 216 | }
|
---|
| 217 | }
|
---|
| 218 | }
|
---|
| 219 | }
|
---|
| 220 |
|
---|
| 221 | if (!this.plain_search) {
|
---|
| 222 | // need to remove the TextQuery service
|
---|
| 223 | Element tq_service = GSXML.getNamedElement(short_service_info, GSXML.SERVICE_ELEM, GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
|
---|
| 224 | short_service_info.removeChild(tq_service);
|
---|
| 225 |
|
---|
| 226 | }
|
---|
| 227 |
|
---|
| 228 |
|
---|
| 229 | // Get the default level out of <defaultLevel> (buildConfig.xml)
|
---|
| 230 | def = (Element) GSXML.getChildByTagName(info, DEFAULT_LEVEL_ELEM);
|
---|
| 231 | if (def != null) {
|
---|
| 232 | this.default_level = def.getAttribute(GSXML.NAME_ATT);
|
---|
| 233 | }
|
---|
| 234 | if (this.default_level == null || this.default_level.equals("")) {
|
---|
| 235 | logger.error("default level not specified!");
|
---|
| 236 | return false;
|
---|
| 237 | }
|
---|
| 238 |
|
---|
| 239 | // the default level is also the level which gdbm is expecting
|
---|
| 240 | // this must not be overwritten
|
---|
| 241 | this.mgpp_src.setReturnLevel(this.default_level);
|
---|
| 242 | // return term info
|
---|
| 243 | this.mgpp_src.setReturnTerms(true);
|
---|
| 244 | // set the default - this may be overwritten by query params
|
---|
| 245 | this.mgpp_src.setQueryLevel(this.default_level);
|
---|
| 246 |
|
---|
| 247 | // set up the extra services which are available for this collection
|
---|
| 248 | // check the config info - if there is no field list, then there is no fielded searching
|
---|
| 249 |
|
---|
| 250 | Element field_list = (Element) GSXML.getChildByTagName(info, GSXML.FIELD_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 251 | if (field_list==null) {
|
---|
| 252 | // nothing more to do
|
---|
| 253 | return true;
|
---|
| 254 | }
|
---|
| 255 |
|
---|
| 256 | // the format info is the same for all services
|
---|
| 257 | Element format_info = (Element)format_info_map.get(TEXT_QUERY_SERVICE);
|
---|
| 258 |
|
---|
| 259 | // find the default field - use the first one
|
---|
| 260 | Element first_field = (Element)GSXML.getChildByTagName(field_list, GSXML.FIELD_ELEM);
|
---|
| 261 | default_field = first_field.getAttribute(GSXML.SHORTNAME_ATT);
|
---|
| 262 | // else set up the fielded query services
|
---|
| 263 |
|
---|
| 264 | if (this.simple_form_search) {
|
---|
| 265 | // set up short_service_info_ - for now just has id and type - name will be added in on teh fly
|
---|
| 266 | Element fq_service = this.doc.createElement(GSXML.SERVICE_ELEM);
|
---|
| 267 | fq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
|
---|
| 268 | fq_service.setAttribute(GSXML.NAME_ATT, FIELD_QUERY_SERVICE);
|
---|
| 269 | this.short_service_info.appendChild(fq_service);
|
---|
| 270 |
|
---|
| 271 | if (format_info != null) {
|
---|
| 272 | this.format_info_map.put(FIELD_QUERY_SERVICE, format_info);
|
---|
| 273 | }
|
---|
| 274 | }
|
---|
| 275 |
|
---|
| 276 | if (this.advanced_form_search) {
|
---|
| 277 | Element afq_service = this.doc.createElement(GSXML.SERVICE_ELEM);
|
---|
| 278 | afq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
|
---|
| 279 | afq_service.setAttribute(GSXML.NAME_ATT, ADVANCED_FIELD_QUERY_SERVICE);
|
---|
| 280 | this.short_service_info.appendChild(afq_service);
|
---|
| 281 |
|
---|
| 282 | if (format_info != null) {
|
---|
| 283 | this.format_info_map.put(ADVANCED_FIELD_QUERY_SERVICE, format_info);
|
---|
| 284 | }
|
---|
| 285 | }
|
---|
| 286 | return true;
|
---|
| 287 |
|
---|
| 288 | }
|
---|
| 289 |
|
---|
| 290 | protected Element getServiceDescription(String service_id, String lang, String subset) {
|
---|
| 291 | // should we check that the service is actually on offer? presumably we wont get asked for services that we haven't advertised previously.
|
---|
| 292 |
|
---|
| 293 | if (!service_id.equals(FIELD_QUERY_SERVICE) && !service_id.equals(ADVANCED_FIELD_QUERY_SERVICE)) {
|
---|
| 294 | return super.getServiceDescription(service_id, lang, subset);
|
---|
| 295 | }
|
---|
| 296 |
|
---|
| 297 |
|
---|
| 298 | Element service = this.doc.createElement(GSXML.SERVICE_ELEM);
|
---|
| 299 | service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
|
---|
| 300 | service.setAttribute(GSXML.NAME_ATT, service_id);
|
---|
| 301 | if (subset == null || subset.equals(GSXML.DISPLAY_TEXT_ELEM+GSXML.LIST_MODIFIER)) {
|
---|
| 302 | service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_NAME, getTextString(service_id+".name", lang)));
|
---|
| 303 | service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_SUBMIT, getTextString(service_id+".submit", lang)));
|
---|
| 304 | service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_DESCRIPTION, getTextString(service_id+".description", lang)));
|
---|
| 305 |
|
---|
| 306 | }
|
---|
| 307 | if (subset == null || subset.equals(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER)) {
|
---|
| 308 | Element param_list = this.doc.createElement(GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 309 | service.appendChild(param_list);
|
---|
| 310 | if (service_id.equals(FIELD_QUERY_SERVICE)) {
|
---|
| 311 |
|
---|
| 312 | addCustomQueryParams(param_list, lang);
|
---|
| 313 | createParameter(MAXDOCS_PARAM, param_list, lang);
|
---|
| 314 | // create a multi param for the fields etc
|
---|
| 315 | // text box, field
|
---|
| 316 | Element multiparam = null;
|
---|
| 317 | Element param=null;
|
---|
| 318 | multiparam = GSXML.createParameterDescription(this.doc, SIMPLE_FIELD_PARAM, "", GSXML.PARAM_TYPE_MULTI, null, null, null);
|
---|
| 319 | multiparam.setAttribute("occurs", "4");
|
---|
| 320 | param_list.appendChild(multiparam);
|
---|
| 321 |
|
---|
| 322 | // the components
|
---|
| 323 | createParameter(FIELD_QUERY_PARAM, multiparam, lang);
|
---|
| 324 | createParameter(FIELD_FIELD_PARAM, multiparam, lang);
|
---|
| 325 |
|
---|
| 326 | } else {
|
---|
| 327 | createParameter(LEVEL_PARAM, param_list, lang);
|
---|
| 328 | createParameter(RANK_PARAM, param_list, lang);
|
---|
| 329 | createParameter(MAXDOCS_PARAM, param_list, lang);
|
---|
| 330 |
|
---|
| 331 |
|
---|
| 332 | // create a multi param for the fields etc
|
---|
| 333 | // text box, stem, case, field
|
---|
| 334 |
|
---|
| 335 | Element multiparam = null;
|
---|
| 336 | Element param=null;
|
---|
| 337 |
|
---|
| 338 | multiparam = GSXML.createParameterDescription(this.doc, ADVANCED_FIELD_PARAM, "", GSXML.PARAM_TYPE_MULTI, null, null, null);
|
---|
| 339 | multiparam.setAttribute("occurs", "4");
|
---|
| 340 | param_list.appendChild(multiparam);
|
---|
| 341 |
|
---|
| 342 | createParameter(FIELD_COMBINE_PARAM, multiparam, lang);
|
---|
| 343 | createParameter(FIELD_QUERY_PARAM, multiparam, lang);
|
---|
| 344 | if (this.does_case) {
|
---|
| 345 | createParameter(FIELD_CASE_PARAM, multiparam, lang);
|
---|
| 346 | }
|
---|
| 347 | if (this.does_stem) {
|
---|
| 348 | createParameter(FIELD_STEM_PARAM, multiparam, lang);
|
---|
| 349 | }
|
---|
| 350 | if (this.does_accent) {
|
---|
| 351 | createParameter(FIELD_ACCENT_PARAM, multiparam, lang);
|
---|
| 352 | }
|
---|
| 353 | createParameter(FIELD_FIELD_PARAM, multiparam, lang);
|
---|
| 354 |
|
---|
| 355 | }
|
---|
| 356 | }
|
---|
| 357 | return service;
|
---|
| 358 |
|
---|
| 359 | }
|
---|
| 360 |
|
---|
| 361 | /** add in the mgpp specific params to TextQuery */
|
---|
| 362 | protected void addCustomQueryParams(Element param_list, String lang)
|
---|
| 363 | {
|
---|
| 364 | createParameter(LEVEL_PARAM, param_list, lang);
|
---|
| 365 | if (this.does_case){
|
---|
| 366 | createParameter(CASE_PARAM, param_list, lang);
|
---|
| 367 | }
|
---|
| 368 | if (this.does_stem){
|
---|
| 369 | createParameter(STEM_PARAM, param_list, lang);
|
---|
| 370 | }
|
---|
| 371 | if (this.does_accent){
|
---|
| 372 | createParameter(ACCENT_PARAM, param_list, lang);
|
---|
| 373 | }
|
---|
| 374 | createParameter(MATCH_PARAM, param_list, lang);
|
---|
| 375 | createParameter(RANK_PARAM, param_list, lang);
|
---|
| 376 | }
|
---|
| 377 |
|
---|
| 378 | /** create a param and add to the list */
|
---|
| 379 | protected void createParameter(String name, Element param_list, String lang)
|
---|
| 380 | {
|
---|
| 381 | Element param = null;
|
---|
| 382 | if (name.equals(LEVEL_PARAM)) {
|
---|
| 383 | ArrayList level_ids = new ArrayList();
|
---|
| 384 | ArrayList level_names = new ArrayList();
|
---|
| 385 | getLevelData(level_ids, level_names, lang);
|
---|
| 386 | if (level_ids.size()>1) {
|
---|
| 387 | // the first one is the default
|
---|
| 388 | param = GSXML.createParameterDescription2(this.doc, LEVEL_PARAM, getTextString("param."+LEVEL_PARAM, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, (String)level_ids.get(0), level_ids, level_names);
|
---|
| 389 | }
|
---|
| 390 |
|
---|
| 391 | } else if (name.equals(RANK_PARAM)) {
|
---|
| 392 | String [] vals1 = {RANK_PARAM_RANK, RANK_PARAM_NONE };
|
---|
| 393 | String [] vals1_texts = { getTextString("param."+RANK_PARAM+"."+RANK_PARAM_RANK, lang), getTextString("param."+RANK_PARAM+"."+RANK_PARAM_NONE, lang)};
|
---|
| 394 |
|
---|
| 395 | param = GSXML.createParameterDescription(this.doc, RANK_PARAM, getTextString("param."+RANK_PARAM, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, RANK_PARAM_RANK, vals1, vals1_texts );
|
---|
| 396 |
|
---|
| 397 | } else if (name.equals(FIELD_QUERY_PARAM)) {
|
---|
| 398 | param = GSXML.createParameterDescription(this.doc, FIELD_QUERY_PARAM, getTextString("param."+FIELD_QUERY_PARAM, lang), GSXML.PARAM_TYPE_STRING, null, null, null);
|
---|
| 399 |
|
---|
| 400 |
|
---|
| 401 | } else if (name.equals(FIELD_CASE_PARAM) || name.equals(FIELD_STEM_PARAM) || name.equals(FIELD_ACCENT_PARAM)) {
|
---|
| 402 | String[] bool_ops = {"0", "1"};
|
---|
| 403 | String[] bool_texts = {getTextString("param.boolean.off", lang, "AbstractSearch"),getTextString("param.boolean.on", lang, "AbstractSearch")};
|
---|
| 404 | param = GSXML.createParameterDescription(this.doc, name, getTextString("param."+name, lang), GSXML.PARAM_TYPE_BOOLEAN, BOOLEAN_PARAM_ON, bool_ops, bool_texts);
|
---|
| 405 |
|
---|
| 406 | } else if (name.equals(FIELD_FIELD_PARAM)) {
|
---|
| 407 | ArrayList fields = new ArrayList();
|
---|
| 408 | ArrayList field_names = new ArrayList();
|
---|
| 409 | getIndexData(fields, field_names, lang);
|
---|
| 410 | // the field list - read from config file
|
---|
| 411 | param = GSXML.createParameterDescription2(this.doc, name, getTextString("param."+name, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, (String)fields.get(0), fields, field_names );
|
---|
| 412 |
|
---|
| 413 | } else if (name.equals(FIELD_COMBINE_PARAM)) {
|
---|
| 414 |
|
---|
| 415 | String []vals = {FIELD_COMBINE_PARAM_AND, FIELD_COMBINE_PARAM_OR, FIELD_COMBINE_PARAM_NOT};
|
---|
| 416 | 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)};
|
---|
| 417 |
|
---|
| 418 |
|
---|
| 419 | param = GSXML.createParameterDescription(this.doc, FIELD_COMBINE_PARAM, "", GSXML.PARAM_TYPE_ENUM_SINGLE, FIELD_COMBINE_PARAM_AND, vals, val_texts);
|
---|
| 420 | param.setAttribute(GSXML.PARAM_IGNORE_POS_ATT, "0");
|
---|
| 421 | }
|
---|
| 422 |
|
---|
| 423 | if (param != null) {
|
---|
| 424 | param_list.appendChild(param);
|
---|
| 425 | } else {
|
---|
| 426 | super.createParameter(name, param_list, lang);
|
---|
| 427 | }
|
---|
| 428 | }
|
---|
| 429 |
|
---|
| 430 | // should cache some of this
|
---|
| 431 | protected void getLevelData(ArrayList level_ids, ArrayList level_names, String lang)
|
---|
| 432 | {
|
---|
| 433 | Element level_list = (Element)GSXML.getChildByTagName(this.config_info, LEVEL_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 434 | NodeList levels = level_list.getElementsByTagName(LEVEL_ELEM);
|
---|
| 435 | for (int i=0; i<levels.getLength(); i++) {
|
---|
| 436 | level_ids.add(((Element)levels.item(i)).getAttribute(GSXML.NAME_ATT));
|
---|
| 437 | level_names.add(getTextString("level."+level_ids.get(i), lang));
|
---|
| 438 | }
|
---|
| 439 |
|
---|
| 440 | }
|
---|
| 441 | protected void getIndexData(ArrayList index_ids, ArrayList index_names, String lang){
|
---|
| 442 | // we currently only have one index, and use the field data as the index
|
---|
| 443 |
|
---|
| 444 | // the field list - read from config file
|
---|
| 445 | Element field_list = (Element)GSXML.getChildByTagName(this.config_info, GSXML.FIELD_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 446 | NodeList fields = field_list.getElementsByTagName(GSXML.FIELD_ELEM);
|
---|
| 447 | for (int i=0; i< fields.getLength();i++) {
|
---|
| 448 | String shortname = ((Element)fields.item(i)).getAttribute(GSXML.SHORTNAME_ATT);
|
---|
| 449 | String name = ((Element)fields.item(i)).getAttribute(GSXML.NAME_ATT);
|
---|
| 450 | if (name.equals("")) {
|
---|
| 451 | // no name, ignore
|
---|
| 452 | continue;
|
---|
| 453 | }
|
---|
| 454 | // TODO change field so that name is the id, and full metadata name is somthing else
|
---|
| 455 | if (shortname.equals("")) {
|
---|
| 456 | shortname = name;
|
---|
| 457 | }
|
---|
| 458 | index_ids.add(shortname);
|
---|
| 459 | // should these be changed to a text element based on lang?
|
---|
| 460 | // or is the name of a metadata element eg dc:Title its
|
---|
| 461 | // name in all langs
|
---|
| 462 | index_names.add(name);
|
---|
| 463 | }
|
---|
| 464 | }
|
---|
| 465 |
|
---|
| 466 | // the following three functions are needed so the base class can
|
---|
| 467 | // call the process+SERVICE_NAME methods
|
---|
| 468 | /** process a text query */
|
---|
| 469 | protected Element processTextQuery(Element request) {
|
---|
| 470 | return processAnyQuery(request, TEXT_QUERY);
|
---|
| 471 | }
|
---|
| 472 |
|
---|
| 473 | /** process a field query */
|
---|
| 474 | protected Element processFieldQuery(Element request) {
|
---|
| 475 | return processAnyQuery(request, SIMPLE_QUERY);
|
---|
| 476 | }
|
---|
| 477 |
|
---|
| 478 | /** process an advanced field query */
|
---|
| 479 | protected Element processAdvancedFieldQuery(Element request) {
|
---|
| 480 | return processAnyQuery(request, ADVANCED_QUERY);
|
---|
| 481 | }
|
---|
| 482 |
|
---|
| 483 | /** process a query */
|
---|
| 484 | protected Element processAnyQuery(Element request, int query_type)
|
---|
| 485 | {
|
---|
| 486 |
|
---|
| 487 | String service_name=null;
|
---|
| 488 | String empty_query_test_param=null;
|
---|
| 489 | // set up the type specific bits
|
---|
| 490 | switch (query_type) {
|
---|
| 491 | case TEXT_QUERY:
|
---|
| 492 | service_name = TEXT_QUERY_SERVICE;
|
---|
| 493 | empty_query_test_param = QUERY_PARAM;
|
---|
| 494 | break;
|
---|
| 495 | case SIMPLE_QUERY:
|
---|
| 496 | service_name = FIELD_QUERY_SERVICE;
|
---|
| 497 | empty_query_test_param = FIELD_QUERY_PARAM;
|
---|
| 498 | break;
|
---|
| 499 | case ADVANCED_QUERY:
|
---|
| 500 | service_name = ADVANCED_FIELD_QUERY_SERVICE;
|
---|
| 501 | empty_query_test_param = FIELD_QUERY_PARAM;
|
---|
| 502 | break;
|
---|
| 503 | default:
|
---|
| 504 | // should never get here
|
---|
| 505 | logger.error("wrong query type!!");
|
---|
| 506 | return null;
|
---|
| 507 | }
|
---|
| 508 |
|
---|
| 509 | // Create a new (empty) result message
|
---|
| 510 | Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
|
---|
| 511 | result.setAttribute(GSXML.FROM_ATT, service_name);
|
---|
| 512 | result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
|
---|
| 513 |
|
---|
| 514 | // Get the parameters of the request
|
---|
| 515 | Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 516 | if (param_list == null) {
|
---|
| 517 | logger.error("TextQuery request had no paramList.");
|
---|
| 518 | return result; // Return the empty result
|
---|
| 519 | }
|
---|
| 520 |
|
---|
| 521 | // Process the request parameters
|
---|
| 522 | HashMap params = GSXML.extractParams(param_list, false);
|
---|
| 523 |
|
---|
| 524 | // Make sure a query has been specified
|
---|
| 525 | String query = (String) params.get(empty_query_test_param);
|
---|
| 526 | if (query == null || query.equals("")) {
|
---|
| 527 | return result; // Return the empty result
|
---|
| 528 | }
|
---|
| 529 |
|
---|
| 530 | // If a field hasn't been specified, use the default - for textQuery
|
---|
| 531 | String field = (String) params.get(INDEX_PARAM);
|
---|
| 532 | if (field == null) {
|
---|
| 533 | field = default_field;
|
---|
| 534 | }
|
---|
| 535 |
|
---|
| 536 | // set up mgpp_src
|
---|
| 537 | String indexdir = GSFile.collectionBaseDir(this.site_home, this.cluster_name) +
|
---|
| 538 | File.separatorChar + GSFile.collectionIndexPath(this.index_stem, this.default_index);
|
---|
| 539 | this.mgpp_src.loadIndexData(indexdir);
|
---|
| 540 | setStandardQueryParams(params);
|
---|
| 541 |
|
---|
| 542 | // if field search, create the query string
|
---|
| 543 | switch (query_type) {
|
---|
| 544 | case TEXT_QUERY:
|
---|
| 545 | query = addFieldInfo(query, field);
|
---|
| 546 | break;
|
---|
| 547 | case SIMPLE_QUERY:
|
---|
| 548 | query = parseFieldQueryParams(params);
|
---|
| 549 | break;
|
---|
| 550 | case ADVANCED_QUERY:
|
---|
| 551 | query = parseAdvancedFieldQueryParams(params);
|
---|
| 552 | break;
|
---|
| 553 | }
|
---|
| 554 |
|
---|
| 555 | // run the query
|
---|
| 556 | this.mgpp_src.runQuery(query);
|
---|
| 557 | MGPPQueryResult mqr= this.mgpp_src.getQueryResult();
|
---|
| 558 |
|
---|
| 559 | // build up the response
|
---|
| 560 |
|
---|
| 561 | // Create a metadata list to store information about the query results
|
---|
| 562 | // should we be using metadataList? or something else?
|
---|
| 563 | Element metadata_list = this.doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 564 | result.appendChild(metadata_list);
|
---|
| 565 |
|
---|
| 566 | // Add a metadata element specifying the number of matching documents
|
---|
| 567 | long totalDocs = mqr.getTotalDocs();
|
---|
| 568 | GSXML.addMetadata(this.doc, metadata_list, "numDocsMatched", ""+totalDocs);
|
---|
| 569 |
|
---|
| 570 | // Create a document list to store the matching documents, and add them
|
---|
| 571 | Vector docs = mqr.getDocs();
|
---|
| 572 |
|
---|
| 573 | // add a metadata item to specify docs returned
|
---|
| 574 | GSXML.addMetadata(this.doc, metadata_list, "numDocsReturned", ""+docs.size());
|
---|
| 575 | // 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
|
---|
| 576 | GSXML.addMetadata(this.doc, metadata_list, "query", query);
|
---|
| 577 | if (docs.size()>0) {
|
---|
| 578 | Element document_list = this.doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 579 | result.appendChild(document_list);
|
---|
| 580 | for (int d = 0; d < docs.size(); d++) {
|
---|
| 581 | long docnum = ((MGPPDocInfo) docs.elementAt(d)).num_;
|
---|
| 582 | float rank = ((MGPPDocInfo) docs.elementAt(d)).rank_;
|
---|
| 583 | String doc_id = MGPPNum2OID(docnum);
|
---|
| 584 | Element doc_node = createDocNode(doc_id, Float.toString(rank));
|
---|
| 585 | //doc_node.setAttribute("rank", Float.toString(rank));
|
---|
| 586 | document_list.appendChild(doc_node);
|
---|
| 587 | }
|
---|
| 588 | }
|
---|
| 589 | // Create a term list to store the term information, and add it
|
---|
| 590 | String query_level = (String)params.get(LEVEL_PARAM); // the current query level
|
---|
| 591 | Element term_list = this.doc.createElement(GSXML.TERM_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 592 | result.appendChild(term_list);
|
---|
| 593 | Vector terms = mqr.getTerms();
|
---|
| 594 | for (int t = 0; t < terms.size(); t++) {
|
---|
| 595 | MGPPTermInfo term_info = (MGPPTermInfo) terms.get(t);
|
---|
| 596 |
|
---|
| 597 | Element term_elem = this.doc.createElement(GSXML.TERM_ELEM);
|
---|
| 598 | term_elem.setAttribute(GSXML.NAME_ATT, term_info.term_);
|
---|
| 599 | term_elem.setAttribute(STEM_ATT, "" + term_info.stem_method_);
|
---|
| 600 | term_elem.setAttribute(FREQ_ATT, "" + term_info.term_freq_);
|
---|
| 601 | term_elem.setAttribute(NUM_DOCS_MATCH_ATT, "" + term_info.match_docs_);
|
---|
| 602 | field = term_info.tag_;
|
---|
| 603 | if (field.equals(query_level)) {
|
---|
| 604 | // ignore
|
---|
| 605 | field = "";
|
---|
| 606 | }
|
---|
| 607 | term_elem.setAttribute(FIELD_ATT, field);
|
---|
| 608 |
|
---|
| 609 | Vector equiv_terms = term_info.equiv_terms_;
|
---|
| 610 | Element equiv_term_list = this.doc.createElement(EQUIV_TERM_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 611 | term_elem.appendChild(equiv_term_list);
|
---|
| 612 |
|
---|
| 613 | for (int et = 0; et < equiv_terms.size(); et++) {
|
---|
| 614 | String equiv_term = (String) equiv_terms.get(et);
|
---|
| 615 |
|
---|
| 616 | Element equiv_term_elem = this.doc.createElement(GSXML.TERM_ELEM);
|
---|
| 617 | equiv_term_elem.setAttribute(GSXML.NAME_ATT, equiv_term);
|
---|
| 618 | equiv_term_elem.setAttribute(NUM_DOCS_MATCH_ATT, "");
|
---|
| 619 | equiv_term_elem.setAttribute(FREQ_ATT, "");
|
---|
| 620 | equiv_term_list.appendChild(equiv_term_elem);
|
---|
| 621 | }
|
---|
| 622 |
|
---|
| 623 | term_list.appendChild(term_elem);
|
---|
| 624 | }
|
---|
| 625 |
|
---|
| 626 | return result;
|
---|
| 627 | }
|
---|
| 628 |
|
---|
| 629 | // should probably use a list rather than map
|
---|
| 630 | protected boolean setStandardQueryParams(HashMap params) {
|
---|
| 631 |
|
---|
| 632 | Set entries = params.entrySet();
|
---|
| 633 | Iterator i = entries.iterator();
|
---|
| 634 | while (i.hasNext()) {
|
---|
| 635 | Map.Entry m = (Map.Entry)i.next();
|
---|
| 636 | String name = (String)m.getKey();
|
---|
| 637 | String value = (String)m.getValue();
|
---|
| 638 |
|
---|
| 639 | if (name.equals(CASE_PARAM)) {
|
---|
| 640 | boolean val = (value.equals(BOOLEAN_PARAM_ON)?true:false);
|
---|
| 641 | this.mgpp_src.setCase(val);
|
---|
| 642 | } else if (name.equals(STEM_PARAM)) {
|
---|
| 643 | boolean val = (value.equals(BOOLEAN_PARAM_ON)?true:false);
|
---|
| 644 | this.mgpp_src.setStem(val);
|
---|
| 645 | } else if (name.equals(MAXDOCS_PARAM)&& !value.equals("")) {
|
---|
| 646 | int docs = Integer.parseInt(value);
|
---|
| 647 | this.mgpp_src.setMaxDocs(docs);
|
---|
| 648 | } else if (name.equals(LEVEL_PARAM)) {
|
---|
| 649 | this.mgpp_src.setQueryLevel(value);
|
---|
| 650 | } else if (name.equals(MATCH_PARAM)) {
|
---|
| 651 | int mode;
|
---|
| 652 | if (value.equals(MATCH_PARAM_ALL)) mode=1;
|
---|
| 653 | else mode=0;
|
---|
| 654 | this.mgpp_src.setMatchMode(mode);
|
---|
| 655 | } else if (name.equals(RANK_PARAM)) {
|
---|
| 656 | if (value.equals(RANK_PARAM_RANK)) {
|
---|
| 657 | this.mgpp_src.setSortByRank(true);
|
---|
| 658 | } else if (value.equals(RANK_PARAM_NONE)) {
|
---|
| 659 | this.mgpp_src.setSortByRank(false);
|
---|
| 660 | }
|
---|
| 661 | } // ignore any others
|
---|
| 662 | }
|
---|
| 663 | return true;
|
---|
| 664 | }
|
---|
| 665 |
|
---|
| 666 | protected String addFieldInfo(String query, String field) {
|
---|
| 667 | if (field.equals("") || field.equals("ZZ")) {
|
---|
| 668 | return query;
|
---|
| 669 | }
|
---|
| 670 | return "["+query+"]:"+field;
|
---|
| 671 | }
|
---|
| 672 | /** combines all the field params into a single query
|
---|
| 673 | * - for simple field query */
|
---|
| 674 | protected String parseFieldQueryParams(HashMap params) {
|
---|
| 675 |
|
---|
| 676 | StringBuffer final_query = new StringBuffer(256);
|
---|
| 677 | String text_line = (String)params.get(FIELD_QUERY_PARAM);
|
---|
| 678 | String[] texts = text_line.split(",", -1);
|
---|
| 679 | String field_line = (String)params.get(FIELD_FIELD_PARAM);
|
---|
| 680 | String[] fields = field_line.split(",", -1);
|
---|
| 681 | String combine="&";
|
---|
| 682 | String match = (String)params.get(MATCH_PARAM);
|
---|
| 683 | if (match.equals(MATCH_PARAM_SOME)) {
|
---|
| 684 | combine = "|";
|
---|
| 685 | }
|
---|
| 686 |
|
---|
| 687 | for (int i=0; i<texts.length; i++) {
|
---|
| 688 |
|
---|
| 689 | String q = texts[i].trim();
|
---|
| 690 | if (!q.equals("")) {
|
---|
| 691 | addQueryElem(final_query, q, fields[i], combine);
|
---|
| 692 | }
|
---|
| 693 | }
|
---|
| 694 |
|
---|
| 695 | return final_query.toString();
|
---|
| 696 | }
|
---|
| 697 |
|
---|
| 698 | /** combines all the field params into a single query
|
---|
| 699 | * - for advanced field query */
|
---|
| 700 | protected String parseAdvancedFieldQueryParams(HashMap params) {
|
---|
| 701 |
|
---|
| 702 | StringBuffer final_query = new StringBuffer(256);
|
---|
| 703 | String text_line = (String)params.get(FIELD_QUERY_PARAM);
|
---|
| 704 | String[] texts = text_line.split(",", -1);
|
---|
| 705 | String field_line = (String)params.get(FIELD_FIELD_PARAM);
|
---|
| 706 | String[] fields = field_line.split(",", -1);
|
---|
| 707 | String case_line = (String)params.get(FIELD_CASE_PARAM);
|
---|
| 708 | String[] cases = case_line.split(",", -1);
|
---|
| 709 | String stem_line = (String)params.get(FIELD_STEM_PARAM);
|
---|
| 710 | String[] stems = stem_line.split(",", -1);
|
---|
| 711 | String combine_line = (String)params.get(FIELD_COMBINE_PARAM);
|
---|
| 712 | String [] combines = combine_line.split(",", -1);
|
---|
| 713 | String combine = "&";
|
---|
| 714 | for (int i=0; i<texts.length; i++) {
|
---|
| 715 | if (i==0) {// assume first one is blank
|
---|
| 716 | combine = "";
|
---|
| 717 | } else {
|
---|
| 718 | String x = combines[i];
|
---|
| 719 | if (x.equals(FIELD_COMBINE_PARAM_AND)) {
|
---|
| 720 | combine = "&";
|
---|
| 721 | } else if (x.equals(FIELD_COMBINE_PARAM_OR)) {
|
---|
| 722 | combine = "|";
|
---|
| 723 | } else if (x.equals(FIELD_COMBINE_PARAM_NOT)) {
|
---|
| 724 | combine = "!";
|
---|
| 725 | }
|
---|
| 726 |
|
---|
| 727 | }
|
---|
| 728 |
|
---|
| 729 | String q = texts[i].trim();
|
---|
| 730 | if (!q.equals("")) {
|
---|
| 731 | q = addStemAndCase(q, stems[i], cases[i]);
|
---|
| 732 | addQueryElem(final_query, q, fields[i], combine);
|
---|
| 733 | }
|
---|
| 734 | }
|
---|
| 735 |
|
---|
| 736 | return final_query.toString();
|
---|
| 737 | }
|
---|
| 738 |
|
---|
| 739 | protected void addQueryElem(StringBuffer s, String q, String f, String c) {
|
---|
| 740 |
|
---|
| 741 | String combine="";
|
---|
| 742 | if (s.length()>0) {
|
---|
| 743 | combine = " "+c+" ";
|
---|
| 744 | }
|
---|
| 745 | if (f.equals("")||f.equals("ZZ")) {
|
---|
| 746 | s.append(combine+q);
|
---|
| 747 | } else {
|
---|
| 748 | s.append(combine+"["+q+"]:"+f);
|
---|
| 749 | }
|
---|
| 750 | }
|
---|
| 751 |
|
---|
| 752 | protected String addStemAndCase(String q, String s, String c) {
|
---|
| 753 | String mods = "#";
|
---|
| 754 | if (c.equals("1")) {
|
---|
| 755 | mods += "i";
|
---|
| 756 | } else {
|
---|
| 757 | mods += "c";
|
---|
| 758 | }
|
---|
| 759 | if (s.equals("1")) {
|
---|
| 760 | mods += "s";
|
---|
| 761 | } else {
|
---|
| 762 | mods+= "u";
|
---|
| 763 | }
|
---|
| 764 | StringBuffer temp = new StringBuffer();
|
---|
| 765 | String [] terms = q.split(" ");
|
---|
| 766 | for (int i=0; i<terms.length; i++) {
|
---|
| 767 | String t = terms[i].trim();
|
---|
| 768 |
|
---|
| 769 | if (!t.equals("") && !t.equals("TX")) {
|
---|
| 770 | temp.append(" "+t+mods);
|
---|
| 771 | }
|
---|
| 772 | }
|
---|
| 773 | return temp.toString();
|
---|
| 774 | }
|
---|
| 775 |
|
---|
| 776 | /** convert MGPP internal id to Greenstone oid */
|
---|
| 777 | abstract protected String MGPPNum2OID(long docnum);
|
---|
| 778 |
|
---|
| 779 | }
|
---|