[11014] | 1 | /*
|
---|
| 2 | * CrossCollectionSearch.java
|
---|
| 3 | * Copyright (C) 2002 New Zealand Digital Library, http://www.nzdl.org
|
---|
| 4 | *
|
---|
| 5 | * This program is free software; you can redistribute it and/or modify
|
---|
| 6 | * it under the terms of the GNU General Public License as published by
|
---|
| 7 | * the Free Software Foundation; either version 2 of the License, or
|
---|
| 8 | * (at your option) any later version.
|
---|
| 9 | *
|
---|
| 10 | * This program is distributed in the hope that it will be useful,
|
---|
| 11 | * but WITHOUT ANY WARRANTY; without even the implied warranty of
|
---|
| 12 | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
---|
| 13 | * GNU General Public License for more details.
|
---|
| 14 | *
|
---|
| 15 | * You should have received a copy of the GNU General Public License
|
---|
| 16 | * along with this program; if not, write to the Free Software
|
---|
| 17 | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
|
---|
| 18 | */
|
---|
| 19 | package org.greenstone.gsdl3.service;
|
---|
| 20 |
|
---|
[26577] | 21 | import java.util.ArrayList;
|
---|
| 22 | import java.util.HashMap;
|
---|
| 23 | import java.util.Iterator;
|
---|
| 24 | import java.util.Map;
|
---|
| 25 | import java.util.Set;
|
---|
| 26 |
|
---|
| 27 | import org.apache.log4j.Logger;
|
---|
| 28 | import org.greenstone.gsdl3.util.GSPath;
|
---|
[11014] | 29 | import org.greenstone.gsdl3.util.GSXML;
|
---|
[24993] | 30 | import org.greenstone.gsdl3.util.UserContext;
|
---|
[28966] | 31 | import org.greenstone.gsdl3.util.XMLConverter;
|
---|
| 32 | import org.w3c.dom.Document;
|
---|
[11014] | 33 | import org.w3c.dom.Element;
|
---|
[16688] | 34 | import org.w3c.dom.Node;
|
---|
[11014] | 35 | import org.w3c.dom.NodeList;
|
---|
| 36 |
|
---|
| 37 | /**
|
---|
[30958] | 38 | * This ServiceRack gets specified in siteConfig.xml. So it is loaded by the MessaegRouter, and two services get activated: TextQuery, DocumentMetadataRetrieve.
|
---|
| 39 | These are located at MR level, not inside a collection. QueryAction will send messages to "TextQuery", rather than eg "mgppdemo/TextQuery".
|
---|
| 40 | These two services will requery the MR for search results/document metadata based on collections or documents listed.
|
---|
[11014] | 41 | */
|
---|
| 42 |
|
---|
[24979] | 43 | public class CrossCollectionSearch extends ServiceRack
|
---|
| 44 | {
|
---|
[11014] | 45 |
|
---|
[24979] | 46 | static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.CrossCollectionSearch.class.getName());
|
---|
| 47 | protected static final String QUERY_PARAM = "query";
|
---|
| 48 | protected static final String COLLECTION_PARAM = "collection";
|
---|
[31181] | 49 | protected static final String GROUP_PARAM = "group";
|
---|
[30150] | 50 | protected static final String MAXDOCS_PARAM = "maxDocs"; // matches standard maxDocs, but in this case, means max docs per collection
|
---|
| 51 | protected static final String HITS_PER_PAGE_PARAM = "hitsPerPage";
|
---|
| 52 | protected static final String MAXDOCS_DEFAULT = "20";
|
---|
[24979] | 53 | // the services on offer - these proxy the actual collection ones
|
---|
| 54 | protected static final String TEXT_QUERY_SERVICE = "TextQuery";
|
---|
| 55 | protected static final String DOCUMENT_METADATA_RETRIEVE_SERVICE = "DocumentMetadataRetrieve";
|
---|
[11014] | 56 |
|
---|
[24979] | 57 | protected String[] coll_ids_list = null;
|
---|
| 58 | protected String[] coll_ids_list_no_all = null;
|
---|
[26577] | 59 | // maps lang to coll names list
|
---|
| 60 | protected HashMap<String, String[]> coll_names_map = null;
|
---|
[11014] | 61 |
|
---|
[26577] | 62 | //protected String[] coll_names_list = null;
|
---|
| 63 |
|
---|
[24979] | 64 | /** constructor */
|
---|
| 65 | public CrossCollectionSearch()
|
---|
| 66 | {
|
---|
| 67 | }
|
---|
[11014] | 68 |
|
---|
[24979] | 69 | public boolean configure(Element info, Element extra_info)
|
---|
| 70 | {
|
---|
| 71 | // any parameters? colls to include??
|
---|
| 72 | logger.info("Configuring CrossCollectionSearch...");
|
---|
| 73 | // query service
|
---|
[28966] | 74 | Element ccs_service = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
|
---|
[24979] | 75 | ccs_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
|
---|
| 76 | ccs_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
|
---|
| 77 | this.short_service_info.appendChild(ccs_service);
|
---|
[11014] | 78 |
|
---|
[24979] | 79 | // metadata service
|
---|
[28966] | 80 | Element dmr_service = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
|
---|
[24979] | 81 | dmr_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_RETRIEVE);
|
---|
| 82 | dmr_service.setAttribute(GSXML.NAME_ATT, DOCUMENT_METADATA_RETRIEVE_SERVICE);
|
---|
| 83 | this.short_service_info.appendChild(dmr_service);
|
---|
| 84 |
|
---|
| 85 | // get any format info
|
---|
| 86 | Element format_info = (Element) GSXML.getChildByTagName(info, GSXML.FORMAT_ELEM);
|
---|
| 87 | if (format_info != null)
|
---|
| 88 | {
|
---|
[28966] | 89 | this.format_info_map.put(TEXT_QUERY_SERVICE, this.desc_doc.importNode(format_info, true));
|
---|
[24979] | 90 | }
|
---|
| 91 | else
|
---|
| 92 | {
|
---|
| 93 | // add in a default format statement
|
---|
[30588] | 94 | //"xmlns:gsf='" + GSXML.GSF_NAMESPACE + "' xmlns:xsl='" + GSXML.XSL_NAMESPACE + "
|
---|
| 95 | String format_string = "<format "+GSXML.STD_NAMESPACES_ATTS + "><gsf:template match='documentNode'><td><a><xsl:attribute name='href'>?a=d&c=<xsl:value-of select='@collection'/>&d=<xsl:value-of select='@nodeID'/><xsl:if test=\"@nodeType='leaf'\">&sib=1</xsl:if>&dt=<xsl:value-of select='@docType'/>&p.a=q&p.s=" + TEXT_QUERY_SERVICE + "&p.c=";
|
---|
[24979] | 96 | if (this.cluster_name != null)
|
---|
| 97 | {
|
---|
| 98 | format_string += this.cluster_name;
|
---|
| 99 | }
|
---|
| 100 | format_string += "</xsl:attribute><gsf:icon/></a></td><td><gsf:metadata name='Title'/> (<xsl:value-of select='@collection'/>) </td></gsf:template></format>";
|
---|
[28966] | 101 | this.format_info_map.put(TEXT_QUERY_SERVICE, this.desc_doc.importNode(this.converter.getDOM(format_string).getDocumentElement(), true));
|
---|
[24979] | 102 | }
|
---|
| 103 | return true;
|
---|
[11014] | 104 | }
|
---|
| 105 |
|
---|
[28966] | 106 | protected Element getServiceDescription(Document doc, String service, String lang, String subset)
|
---|
[24979] | 107 | {
|
---|
| 108 | if (service.equals(TEXT_QUERY_SERVICE))
|
---|
| 109 | {
|
---|
[11014] | 110 |
|
---|
[28966] | 111 | Element ccs_service = doc.createElement(GSXML.SERVICE_ELEM);
|
---|
[24979] | 112 | ccs_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
|
---|
| 113 | ccs_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
|
---|
| 114 |
|
---|
| 115 | // display info
|
---|
| 116 | if (subset == null || subset.equals(GSXML.DISPLAY_TEXT_ELEM + GSXML.LIST_MODIFIER))
|
---|
| 117 | {
|
---|
[28966] | 118 | ccs_service.appendChild(GSXML.createDisplayTextElement(doc, GSXML.DISPLAY_TEXT_NAME, getTextString(TEXT_QUERY_SERVICE + ".name", lang)));
|
---|
| 119 | ccs_service.appendChild(GSXML.createDisplayTextElement(doc, GSXML.DISPLAY_TEXT_SUBMIT, getTextString(TEXT_QUERY_SERVICE + ".submit", lang)));
|
---|
| 120 | ccs_service.appendChild(GSXML.createDisplayTextElement(doc, GSXML.DISPLAY_TEXT_DESCRIPTION, getTextString(TEXT_QUERY_SERVICE + ".description", lang)));
|
---|
[24979] | 121 | }
|
---|
| 122 | // param info
|
---|
| 123 | if (subset == null || subset.equals(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER))
|
---|
| 124 | {
|
---|
[28966] | 125 | Element param_list = doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
|
---|
[24979] | 126 | // collection list
|
---|
| 127 | if (coll_ids_list == null)
|
---|
| 128 | {
|
---|
[26406] | 129 | initCollectionList(lang);
|
---|
[24979] | 130 | }
|
---|
[26577] | 131 | if (!coll_names_map.containsKey(lang))
|
---|
| 132 | {
|
---|
| 133 | addCollectionNames(lang);
|
---|
[26406] | 134 | }
|
---|
[28966] | 135 | Element param = GSXML.createParameterDescription(doc, COLLECTION_PARAM, getTextString("param." + COLLECTION_PARAM, lang), GSXML.PARAM_TYPE_ENUM_MULTI, "all", coll_ids_list, coll_names_map.get(lang));
|
---|
[24979] | 136 | param_list.appendChild(param);
|
---|
[30150] | 137 | // max docs param
|
---|
| 138 | param = GSXML.createParameterDescription(doc, MAXDOCS_PARAM, getTextString("param." + MAXDOCS_PARAM, lang), GSXML.PARAM_TYPE_INTEGER, MAXDOCS_DEFAULT, null, null);
|
---|
| 139 | param_list.appendChild(param);
|
---|
[24979] | 140 | // query param
|
---|
[28966] | 141 | param = GSXML.createParameterDescription(doc, QUERY_PARAM, getTextString("param." + QUERY_PARAM, lang), GSXML.PARAM_TYPE_STRING, null, null, null);
|
---|
[24979] | 142 | param_list.appendChild(param);
|
---|
| 143 | ccs_service.appendChild(param_list);
|
---|
| 144 | }
|
---|
| 145 |
|
---|
| 146 | logger.debug("service description=" + this.converter.getPrettyString(ccs_service));
|
---|
| 147 | return ccs_service;
|
---|
[11014] | 148 | }
|
---|
[24979] | 149 | // these ones are probably never called, but put them here just in case
|
---|
[28966] | 150 | Element service_elem = doc.createElement(GSXML.SERVICE_ELEM);
|
---|
[24979] | 151 | service_elem.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_RETRIEVE);
|
---|
| 152 | service_elem.setAttribute(GSXML.NAME_ATT, service);
|
---|
| 153 | return service_elem;
|
---|
| 154 |
|
---|
[11014] | 155 | }
|
---|
| 156 |
|
---|
[24979] | 157 | protected Element processTextQuery(Element request)
|
---|
| 158 | {
|
---|
| 159 | // Create a new (empty) result message
|
---|
[28966] | 160 | Document result_doc = XMLConverter.newDOM();
|
---|
| 161 | Element result = result_doc.createElement(GSXML.RESPONSE_ELEM);
|
---|
[24979] | 162 | result.setAttribute(GSXML.FROM_ATT, TEXT_QUERY_SERVICE);
|
---|
| 163 | result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
|
---|
[11014] | 164 |
|
---|
[24993] | 165 | UserContext userContext = new UserContext(request);
|
---|
[26577] | 166 |
|
---|
[24979] | 167 | // Get the parameters of the request
|
---|
| 168 | Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
|
---|
| 169 | if (param_list == null)
|
---|
| 170 | {
|
---|
| 171 | logger.error("TextQuery request had no paramList.");
|
---|
| 172 | return result; // Return the empty result
|
---|
| 173 | }
|
---|
| 174 | // get the collection list
|
---|
| 175 | String[] colls_list = coll_ids_list_no_all;
|
---|
| 176 | Element coll_param = GSXML.getNamedElement(param_list, GSXML.PARAM_ELEM, GSXML.NAME_ATT, COLLECTION_PARAM);
|
---|
| 177 | if (coll_param != null)
|
---|
| 178 | {
|
---|
| 179 | String coll_list = GSXML.getValue(coll_param);
|
---|
| 180 | if (!coll_list.equals("all") && !coll_list.equals(""))
|
---|
| 181 | {
|
---|
| 182 | colls_list = coll_list.split(",");
|
---|
| 183 | }
|
---|
| 184 | }
|
---|
[31181] | 185 |
|
---|
| 186 | colls_list = mergeGroups(userContext, param_list, colls_list);
|
---|
[30150] | 187 |
|
---|
| 188 | String maxdocs = MAXDOCS_DEFAULT;
|
---|
| 189 | Element maxdocs_param = GSXML.getNamedElement(param_list, GSXML.PARAM_ELEM, GSXML.NAME_ATT, MAXDOCS_PARAM);
|
---|
| 190 | if (maxdocs_param != null) {
|
---|
| 191 | maxdocs = GSXML.getValue(maxdocs_param);
|
---|
| 192 | }
|
---|
| 193 |
|
---|
[28966] | 194 | Document msg_doc = XMLConverter.newDOM();
|
---|
| 195 | Element query_message = msg_doc.createElement(GSXML.MESSAGE_ELEM);
|
---|
[24979] | 196 | // we are sending the same request to each collection - build up the to
|
---|
| 197 | // attribute for the request
|
---|
| 198 | StringBuffer to_att = new StringBuffer();
|
---|
| 199 | for (int i = 0; i < colls_list.length; i++)
|
---|
| 200 | {
|
---|
| 201 | if (i > 0)
|
---|
| 202 | {
|
---|
| 203 | to_att.append(",");
|
---|
| 204 | }
|
---|
| 205 | to_att.append(GSPath.appendLink(colls_list[i], "TextQuery"));
|
---|
[11014] | 206 |
|
---|
[24979] | 207 | }
|
---|
| 208 | // send the query to all colls
|
---|
[28966] | 209 | Element query_request = GSXML.createBasicRequest(msg_doc, GSXML.REQUEST_TYPE_PROCESS, to_att.toString(), userContext);
|
---|
[24979] | 210 | query_message.appendChild(query_request);
|
---|
| 211 | // should we add params individually?
|
---|
[28966] | 212 | Element new_param_list = msg_doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
|
---|
[24979] | 213 | query_request.appendChild(new_param_list);
|
---|
[28966] | 214 | new_param_list.appendChild(msg_doc.importNode(GSXML.getNamedElement(param_list, GSXML.PARAM_ELEM, GSXML.NAME_ATT, QUERY_PARAM), true));
|
---|
[30150] | 215 |
|
---|
| 216 | // for cross coll search, we only want maxdocs from each collection
|
---|
| 217 | // some colls use maxdocs, some use hits per page so lets send both
|
---|
| 218 | new_param_list.appendChild(GSXML.createParameter(msg_doc, MAXDOCS_PARAM, maxdocs));
|
---|
| 219 | new_param_list.appendChild(GSXML.createParameter(msg_doc, HITS_PER_PAGE_PARAM, maxdocs));
|
---|
[24979] | 220 | Element query_result = (Element) this.router.process(query_message);
|
---|
[30150] | 221 | // create the doc list for the response
|
---|
[28966] | 222 | Element doc_node_list = result_doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
|
---|
[24979] | 223 | result.appendChild(doc_node_list);
|
---|
[31262] | 224 | Element result_snippet_list = result_doc.createElement(GSXML.HL_SNIPPET_ELEM + GSXML.LIST_MODIFIER);
|
---|
| 225 | result.appendChild(result_snippet_list);
|
---|
| 226 | NodeList hl_snippet_list = query_result.getElementsByTagName(GSXML.HL_SNIPPET_ELEM);
|
---|
| 227 | if (hl_snippet_list != null){
|
---|
| 228 | for (int hls = 0; hls < hl_snippet_list.getLength(); hls++){
|
---|
| 229 | result_snippet_list.appendChild(result_doc.importNode(hl_snippet_list.item(hls), true));
|
---|
| 230 | }
|
---|
| 231 | }
|
---|
[11014] | 232 |
|
---|
[24979] | 233 | NodeList responses = query_result.getElementsByTagName(GSXML.RESPONSE_ELEM);
|
---|
[30150] | 234 | int num_docs = 0;
|
---|
[24979] | 235 | for (int k = 0; k < responses.getLength(); k++)
|
---|
| 236 | {
|
---|
| 237 | String coll_name = GSPath.removeLastLink(((Element) responses.item(k)).getAttribute(GSXML.FROM_ATT));
|
---|
| 238 | NodeList nodes = ((Element) responses.item(k)).getElementsByTagName(GSXML.DOC_NODE_ELEM);
|
---|
| 239 | if (nodes == null || nodes.getLength() == 0)
|
---|
| 240 | continue;
|
---|
[30150] | 241 | num_docs += nodes.getLength();
|
---|
[24979] | 242 | Element last_node = null;
|
---|
| 243 | Element this_node = null;
|
---|
| 244 | for (int n = 0; n < nodes.getLength(); n++)
|
---|
| 245 | {
|
---|
| 246 | this_node = (Element) nodes.item(n);
|
---|
| 247 | this_node.setAttribute("collection", coll_name);
|
---|
[31262] | 248 |
|
---|
[24979] | 249 | if (k == 0)
|
---|
| 250 | {
|
---|
[13978] | 251 |
|
---|
[28966] | 252 | doc_node_list.appendChild(result_doc.importNode(this_node, true));
|
---|
[24979] | 253 | }
|
---|
| 254 | else
|
---|
| 255 | {
|
---|
| 256 | if (last_node == null)
|
---|
| 257 | {
|
---|
| 258 | last_node = (Element) GSXML.getChildByTagName(doc_node_list, GSXML.DOC_NODE_ELEM);
|
---|
| 259 | }
|
---|
| 260 | last_node = GSXML.insertIntoOrderedList(doc_node_list, GSXML.DOC_NODE_ELEM, last_node, this_node, "rank", true);
|
---|
| 261 | }
|
---|
[13978] | 262 |
|
---|
[24979] | 263 | }
|
---|
| 264 | }
|
---|
[30150] | 265 | // just send back num docs returned. Too hard to work out number of matches etc as each index type
|
---|
| 266 | // does it differently
|
---|
[30161] | 267 | Element metadata_list = result_doc.createElement(GSXML.METADATA_ELEM + GSXML.LIST_MODIFIER);
|
---|
| 268 | result.appendChild(metadata_list);
|
---|
[30150] | 269 | GSXML.addMetadata(metadata_list, "numDocsReturned", "" + num_docs);
|
---|
[24979] | 270 | return result;
|
---|
[11014] | 271 | }
|
---|
[13978] | 272 |
|
---|
[24979] | 273 | // protected Element processAdvTextQuery(Element request)
|
---|
| 274 | // {
|
---|
[13978] | 275 |
|
---|
[24979] | 276 | // }
|
---|
[31181] | 277 |
|
---|
| 278 |
|
---|
| 279 |
|
---|
[26406] | 280 | protected boolean initCollectionList(String lang)
|
---|
[24979] | 281 | {
|
---|
[24993] | 282 | UserContext userContext = new UserContext();
|
---|
[26406] | 283 | userContext.setLanguage(lang);
|
---|
[24993] | 284 | userContext.setUserID("");
|
---|
| 285 |
|
---|
[24979] | 286 | // first, get the message router info
|
---|
[28966] | 287 | Document msg_doc = XMLConverter.newDOM();
|
---|
| 288 | Element coll_list_message = msg_doc.createElement(GSXML.MESSAGE_ELEM);
|
---|
| 289 | Element coll_list_request = GSXML.createBasicRequest(msg_doc, GSXML.REQUEST_TYPE_DESCRIBE, "", userContext); // uid
|
---|
[24979] | 290 | coll_list_message.appendChild(coll_list_request);
|
---|
| 291 | logger.debug("coll list request = " + this.converter.getPrettyString(coll_list_request));
|
---|
| 292 | Element coll_list_response = (Element) this.router.process(coll_list_message);
|
---|
| 293 | if (coll_list_response == null)
|
---|
| 294 | {
|
---|
| 295 | logger.error("couldn't query the message router!");
|
---|
| 296 | return false;
|
---|
| 297 | }
|
---|
| 298 | logger.debug("coll list response = " + this.converter.getPrettyString(coll_list_response));
|
---|
| 299 | // second, get some info from each collection. we want the coll name
|
---|
| 300 | // and whether its got a text query service
|
---|
[13978] | 301 |
|
---|
[24979] | 302 | NodeList colls = coll_list_response.getElementsByTagName(GSXML.COLLECTION_ELEM);
|
---|
[26406] | 303 | // we can send the same request to multiple collections at once by using a comma separated list
|
---|
[28966] | 304 | Element metadata_message = msg_doc.createElement(GSXML.MESSAGE_ELEM);
|
---|
[26406] | 305 | StringBuffer colls_sb = new StringBuffer();
|
---|
[24979] | 306 | for (int i = 0; i < colls.getLength(); i++)
|
---|
| 307 | {
|
---|
| 308 | Element c = (Element) colls.item(i);
|
---|
| 309 | String name = c.getAttribute(GSXML.NAME_ATT);
|
---|
[26577] | 310 | if (i != 0)
|
---|
| 311 | {
|
---|
| 312 | colls_sb.append(",");
|
---|
[26406] | 313 | }
|
---|
| 314 | colls_sb.append(name);
|
---|
| 315 | //Element metadata_request = GSXML.createBasicRequest(this.doc, GSXML.REQUEST_TYPE_DESCRIBE, name, userContext);
|
---|
| 316 | //metadata_message.appendChild(metadata_request);
|
---|
[24979] | 317 | }
|
---|
[26578] | 318 |
|
---|
[28966] | 319 | Element metadata_request = GSXML.createBasicRequest(msg_doc, GSXML.REQUEST_TYPE_DESCRIBE, colls_sb.toString(), userContext);
|
---|
[26406] | 320 | metadata_message.appendChild(metadata_request);
|
---|
[24979] | 321 | logger.debug("metadata request = " + this.converter.getPrettyString(metadata_message));
|
---|
| 322 | Element metadata_response = (Element) this.router.process(metadata_message);
|
---|
| 323 | logger.debug("metadata response = " + this.converter.getPrettyString(metadata_response));
|
---|
| 324 | NodeList coll_responses = metadata_response.getElementsByTagName(GSXML.RESPONSE_ELEM);
|
---|
[25635] | 325 | ArrayList<String> valid_colls = new ArrayList<String>();
|
---|
| 326 | ArrayList<String> valid_coll_names = new ArrayList<String>();
|
---|
[24979] | 327 | for (int i = 0; i < coll_responses.getLength(); i++)
|
---|
| 328 | {
|
---|
| 329 | Element response = (Element) coll_responses.item(i);
|
---|
| 330 | Element coll = (Element) GSXML.getChildByTagName(response, GSXML.COLLECTION_ELEM);
|
---|
| 331 | Element service_list = (Element) GSXML.getChildByTagName(coll, GSXML.SERVICE_ELEM + GSXML.LIST_MODIFIER);
|
---|
| 332 | if (service_list == null)
|
---|
| 333 | continue;
|
---|
| 334 | Element query_service = GSXML.getNamedElement(service_list, GSXML.SERVICE_ELEM, GSXML.NAME_ATT, TEXT_QUERY_SERVICE); // should be AbstractTextSearch.TEXT_QUERY_SERVICE
|
---|
| 335 | if (query_service == null)
|
---|
| 336 | continue;
|
---|
| 337 | // use the name of the response in case we are talking to a remote collection, not the name of the collection.
|
---|
| 338 | String coll_id = response.getAttribute(GSXML.FROM_ATT);
|
---|
[30630] | 339 | String coll_name = getDisplayText(coll, GSXML.DISPLAY_TEXT_NAME, lang, "en");
|
---|
[24979] | 340 | valid_colls.add(coll_id);
|
---|
| 341 | valid_coll_names.add(coll_name);
|
---|
| 342 | }
|
---|
| 343 |
|
---|
[26578] | 344 | this.coll_names_map = new HashMap<String, String[]>();
|
---|
| 345 |
|
---|
[24979] | 346 | // ids no all has the list without 'all' option.
|
---|
| 347 | this.coll_ids_list_no_all = new String[1];
|
---|
[25635] | 348 | this.coll_ids_list_no_all = valid_colls.toArray(coll_ids_list_no_all);
|
---|
[24979] | 349 |
|
---|
| 350 | valid_colls.add(0, "all");
|
---|
[26406] | 351 | valid_coll_names.add(0, getTextString("param." + COLLECTION_PARAM + ".all", lang));
|
---|
| 352 |
|
---|
[24979] | 353 | this.coll_ids_list = new String[1];
|
---|
[25635] | 354 | this.coll_ids_list = valid_colls.toArray(coll_ids_list);
|
---|
[26406] | 355 |
|
---|
| 356 | String[] coll_names_list = new String[1];
|
---|
| 357 | coll_names_list = valid_coll_names.toArray(coll_names_list);
|
---|
| 358 | this.coll_names_map.put(lang, coll_names_list);
|
---|
[24979] | 359 | return true;
|
---|
[11014] | 360 | }
|
---|
| 361 |
|
---|
[26577] | 362 | protected void addCollectionNames(String lang)
|
---|
| 363 | {
|
---|
[26406] | 364 |
|
---|
[26577] | 365 | UserContext userContext = new UserContext();
|
---|
| 366 | userContext.setLanguage(lang);
|
---|
| 367 | userContext.setUserID("");
|
---|
[26406] | 368 |
|
---|
[26577] | 369 | ArrayList<String> coll_names = new ArrayList<String>();
|
---|
| 370 | coll_names.add(getTextString("param." + COLLECTION_PARAM + ".all", lang));
|
---|
[26406] | 371 |
|
---|
[26577] | 372 | // need to request MR for collection descriptions
|
---|
[28966] | 373 | Document msg_doc = XMLConverter.newDOM();
|
---|
| 374 | Element metadata_message = msg_doc.createElement(GSXML.MESSAGE_ELEM);
|
---|
[26406] | 375 |
|
---|
[26577] | 376 | // get a comma separated list of coll ids to send to MR
|
---|
| 377 | // the first item is the place holder for 'all'
|
---|
| 378 | StringBuffer colls_sb = new StringBuffer();
|
---|
| 379 | for (int i = 1; i < coll_ids_list.length; i++)
|
---|
| 380 | {
|
---|
| 381 | if (i != 1)
|
---|
| 382 | {
|
---|
| 383 | colls_sb.append(",");
|
---|
| 384 | }
|
---|
| 385 | colls_sb.append(coll_ids_list[i]);
|
---|
| 386 | }
|
---|
[28966] | 387 | Element metadata_request = GSXML.createBasicRequest(msg_doc, GSXML.REQUEST_TYPE_DESCRIBE, colls_sb.toString(), userContext);
|
---|
[26577] | 388 | // param_list to request just displayTextList
|
---|
[28966] | 389 | Element param_list = msg_doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
|
---|
| 390 | Element param = GSXML.createParameter(msg_doc, GSXML.SUBSET_PARAM, GSXML.DISPLAY_TEXT_ELEM + GSXML.LIST_MODIFIER);
|
---|
[26577] | 391 | param_list.appendChild(param);
|
---|
| 392 | metadata_request.appendChild(param_list);
|
---|
| 393 | metadata_message.appendChild(metadata_request);
|
---|
| 394 | logger.debug("coll names metadata request = " + this.converter.getPrettyString(metadata_message));
|
---|
| 395 | Element metadata_response = (Element) this.router.process(metadata_message);
|
---|
| 396 | logger.debug("coll names metadata response = " + this.converter.getPrettyString(metadata_response));
|
---|
| 397 | NodeList coll_responses = metadata_response.getElementsByTagName(GSXML.RESPONSE_ELEM);
|
---|
| 398 | for (int i = 0; i < coll_responses.getLength(); i++)
|
---|
| 399 | {
|
---|
| 400 | Element response = (Element) coll_responses.item(i);
|
---|
| 401 | Element coll = (Element) GSXML.getChildByTagName(response, GSXML.COLLECTION_ELEM);
|
---|
[30630] | 402 | String coll_name = getDisplayText(coll, GSXML.DISPLAY_TEXT_NAME, lang, "en");
|
---|
[26577] | 403 | coll_names.add(coll_name);
|
---|
| 404 | }
|
---|
[26406] | 405 |
|
---|
[26577] | 406 | String[] coll_names_list = new String[1];
|
---|
| 407 | coll_names_list = coll_names.toArray(coll_names_list);
|
---|
| 408 | this.coll_names_map.put(lang, coll_names_list);
|
---|
[26406] | 409 |
|
---|
[26577] | 410 | }
|
---|
[26406] | 411 |
|
---|
[24979] | 412 | protected Element processDocumentMetadataRetrieve(Element request)
|
---|
| 413 | {
|
---|
| 414 | // Create a new (empty) result message
|
---|
[28966] | 415 | Document result_doc = XMLConverter.newDOM();
|
---|
| 416 | Element result = result_doc.createElement(GSXML.RESPONSE_ELEM);
|
---|
[24979] | 417 | result.setAttribute(GSXML.FROM_ATT, DOCUMENT_METADATA_RETRIEVE_SERVICE);
|
---|
| 418 | result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
|
---|
| 419 |
|
---|
[24993] | 420 | UserContext userContext = new UserContext(request);
|
---|
[24979] | 421 | // Get the parameters of the request
|
---|
| 422 | Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
|
---|
| 423 | if (param_list == null)
|
---|
| 424 | {
|
---|
| 425 | logger.error("DocumentMetadataRetrieve request had no paramList.");
|
---|
| 426 | return result; // Return the empty result
|
---|
| 427 | }
|
---|
| 428 |
|
---|
| 429 | NodeList query_doc_list = request.getElementsByTagName(GSXML.DOC_NODE_ELEM);
|
---|
| 430 | if (query_doc_list.getLength() == 0)
|
---|
| 431 | {
|
---|
| 432 | logger.error("DocumentMetadataRetrieve request had no documentNodes.");
|
---|
| 433 | return result; // Return the empty result
|
---|
| 434 | }
|
---|
| 435 |
|
---|
| 436 | // the resulting doc node list
|
---|
[28966] | 437 | Element result_node_list = result_doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
|
---|
[24979] | 438 | result.appendChild(result_node_list);
|
---|
| 439 |
|
---|
| 440 |
|
---|
| 441 | // organise the nodes into collection lists
|
---|
[25635] | 442 | HashMap<String, Node> coll_map = new HashMap<String, Node>();
|
---|
[24979] | 443 |
|
---|
| 444 | for (int i = 0; i < query_doc_list.getLength(); i++)
|
---|
| 445 | {
|
---|
| 446 | Element doc_node = (Element) query_doc_list.item(i);
|
---|
| 447 | String coll_name = doc_node.getAttribute("collection");
|
---|
| 448 | Element coll_items = (Element) coll_map.get(coll_name);
|
---|
| 449 | if (coll_items == null)
|
---|
| 450 | {
|
---|
[28966] | 451 | coll_items = result_doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
|
---|
[24979] | 452 | coll_map.put(coll_name, coll_items);
|
---|
| 453 | }
|
---|
[28966] | 454 | coll_items.appendChild(result_doc.importNode(doc_node, true));
|
---|
[24979] | 455 | }
|
---|
| 456 |
|
---|
| 457 | // create teh individual requests
|
---|
[28966] | 458 | Document msg_doc = XMLConverter.newDOM();
|
---|
| 459 | Element meta_request_message = msg_doc.createElement(GSXML.MESSAGE_ELEM);
|
---|
| 460 |
|
---|
[24979] | 461 | Set mapping_set = coll_map.entrySet();
|
---|
| 462 | Iterator iter = mapping_set.iterator();
|
---|
| 463 |
|
---|
| 464 | while (iter.hasNext())
|
---|
| 465 | {
|
---|
| 466 | Map.Entry e = (Map.Entry) iter.next();
|
---|
| 467 | String cname = (String) e.getKey();
|
---|
| 468 | Element doc_nodes = (Element) e.getValue();
|
---|
[28966] | 469 | Element meta_request = GSXML.createBasicRequest(msg_doc, GSXML.REQUEST_TYPE_PROCESS, GSPath.appendLink(cname, DOCUMENT_METADATA_RETRIEVE_SERVICE), userContext);
|
---|
| 470 | meta_request.appendChild(msg_doc.importNode(doc_nodes, true));
|
---|
[30958] | 471 | meta_request.appendChild(msg_doc.importNode(param_list, true));
|
---|
[24979] | 472 | meta_request_message.appendChild(meta_request);
|
---|
| 473 |
|
---|
| 474 | }
|
---|
| 475 |
|
---|
| 476 | Node meta_result_node = this.router.process(meta_request_message);
|
---|
[28966] | 477 | Element meta_result = GSXML.nodeToElement(meta_result_node);
|
---|
[24979] | 478 |
|
---|
| 479 | // now need to put the doc nodes back in the right order
|
---|
| 480 | // go through the original list again. keep an element pointer to
|
---|
| 481 | // the next element in each collections list
|
---|
| 482 | NodeList meta_responses = meta_result.getElementsByTagName(GSXML.RESPONSE_ELEM);
|
---|
| 483 | for (int i = 0; i < meta_responses.getLength(); i++)
|
---|
| 484 | {
|
---|
| 485 | String collname = GSPath.removeLastLink(((Element) meta_responses.item(i)).getAttribute(GSXML.FROM_ATT));
|
---|
| 486 | Element first_elem = (Element) GSXML.getNodeByPath(meta_responses.item(i), "documentNodeList/documentNode");
|
---|
| 487 | coll_map.put(collname, first_elem);
|
---|
| 488 | }
|
---|
| 489 |
|
---|
| 490 | for (int i = 0; i < query_doc_list.getLength(); i++)
|
---|
| 491 | {
|
---|
| 492 | Element doc_node = (Element) query_doc_list.item(i);
|
---|
[28966] | 493 | Element new_node = (Element) result_doc.importNode(doc_node, false);
|
---|
[24979] | 494 | result_node_list.appendChild(new_node);
|
---|
| 495 | String coll_name = doc_node.getAttribute("collection");
|
---|
| 496 |
|
---|
| 497 | Element meta_elem = (Element) coll_map.get(coll_name);
|
---|
| 498 | GSXML.mergeMetadataLists(new_node, meta_elem);
|
---|
| 499 | coll_map.put(coll_name, meta_elem.getNextSibling());
|
---|
| 500 | }
|
---|
| 501 | return result;
|
---|
[11014] | 502 | }
|
---|
[31181] | 503 |
|
---|
| 504 | private String[] mergeGroups(UserContext userContext, Element paramList, String[] collArray){
|
---|
| 505 | Document doc = XMLConverter.newDOM();
|
---|
[31344] | 506 | boolean allSelected = false;
|
---|
[31343] | 507 | Element groupParamList = extractGroupParams(paramList, collArray, doc);
|
---|
[31181] | 508 | Element collParam = GSXML.getNamedElement(paramList, GSXML.PARAM_ELEM, GSXML.NAME_ATT, COLLECTION_PARAM);
|
---|
[31344] | 509 | if (collParam != null) {
|
---|
| 510 | String collValue = GSXML.getValue(collParam);
|
---|
| 511 | if (collValue != null) {
|
---|
| 512 | allSelected = collValue.matches(".*\\ball\\b.*");
|
---|
| 513 | }
|
---|
| 514 | }
|
---|
[31343] | 515 | //Group param not empty and coll param null or not 'all'
|
---|
| 516 | if ( allSelected || !groupParamList.hasChildNodes())
|
---|
[31181] | 517 | {
|
---|
| 518 | return collArray;
|
---|
| 519 | }
|
---|
[31343] | 520 | //GroupInfo service to get uniq collections
|
---|
| 521 | String uniqCollServiceName = "UniqueCollections";
|
---|
| 522 | Element infoResponse = getMRInfo(userContext);
|
---|
| 523 | Element serviceList = (Element) GSXML.getChildByTagName(infoResponse, GSXML.SERVICE_ELEM + GSXML.LIST_MODIFIER);
|
---|
| 524 | if (serviceList == null) {
|
---|
| 525 | logger.error("Service list is null!");
|
---|
| 526 | return collArray;
|
---|
| 527 | }
|
---|
| 528 | Element groupInfoService = GSXML.getNamedElement(serviceList, GSXML.SERVICE_ELEM, GSXML.NAME_ATT, uniqCollServiceName);
|
---|
| 529 | if (groupInfoService == null){
|
---|
| 530 | logger.error("UniqueCollections service unavailable; Check your groupConfig.xml");
|
---|
| 531 | return collArray;
|
---|
| 532 | }
|
---|
| 533 | Element groupQueryMessage = doc.createElement(GSXML.MESSAGE_ELEM);
|
---|
| 534 | Element groupQueryRequest = GSXML.createBasicRequest(doc, GSXML.REQUEST_TYPE_PROCESS, uniqCollServiceName, userContext);
|
---|
| 535 | groupQueryMessage.appendChild(groupQueryRequest);
|
---|
| 536 | groupQueryRequest.appendChild(groupParamList);
|
---|
| 537 | if (collParam != null){
|
---|
| 538 | groupParamList.appendChild(doc.importNode(GSXML.getNamedElement(paramList, GSXML.PARAM_ELEM, GSXML.NAME_ATT, GSXML.COLLECTION_ELEM), true));
|
---|
| 539 | }
|
---|
| 540 | Element groupQueryResult = (Element) this.router.process(groupQueryMessage);
|
---|
| 541 | Element groupQueryResponse = (Element) GSXML.getChildByTagName(groupQueryResult, GSXML.RESPONSE_ELEM);
|
---|
| 542 | Element collList = (Element) GSXML.getChildByTagName(groupQueryResponse, GSXML.COLLECTION_ELEM + GSXML.LIST_MODIFIER);
|
---|
| 543 | NodeList collections = GSXML.getChildrenByTagName(collList, GSXML.COLLECTION_ELEM);
|
---|
| 544 | collArray = new String[collections.getLength()];
|
---|
| 545 | for (int i = 0; i < collections.getLength(); i++){
|
---|
| 546 | String collName = ((Element) collections.item(i)).getAttribute(GSXML.NAME_ATT);
|
---|
| 547 | collArray[i] = collName;
|
---|
| 548 | }
|
---|
[31181] | 549 | return collArray;
|
---|
| 550 |
|
---|
| 551 | }
|
---|
[31343] | 552 |
|
---|
| 553 | private Element extractGroupParams(Element paramList, String[] collArray, Document doc) {
|
---|
| 554 | Element groupParamList = doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
|
---|
| 555 | Element groupParam = null;
|
---|
| 556 | Element groupNode = GSXML.getNamedElement(paramList, GSXML.PARAM_ELEM, GSXML.NAME_ATT, GROUP_PARAM);
|
---|
| 557 | if (groupNode != null){
|
---|
| 558 | groupParam = (Element) doc.importNode(groupNode, true);
|
---|
| 559 | groupParamList.appendChild(groupParam);
|
---|
| 560 | } else {
|
---|
| 561 | groupParam = doc.createElement(GSXML.PARAM_ELEM);
|
---|
| 562 | groupParam.setAttribute(GSXML.NAME_ATT, GSXML.GROUP_ELEM);
|
---|
| 563 | groupParamList.appendChild(groupParam);
|
---|
| 564 | }
|
---|
| 565 | String prefix = GSXML.GROUP_ELEM + "/";
|
---|
| 566 | for (int i = 0; i < collArray.length; i++) {
|
---|
| 567 | String collectionParam = collArray[i];
|
---|
| 568 | if (collectionParam.startsWith(prefix)){
|
---|
| 569 | String value = groupParam.getAttribute(GSXML.VALUE_ATT);
|
---|
| 570 | value += "," + collectionParam.substring(prefix.length() - 1 );
|
---|
| 571 | groupParam.setAttribute(GSXML.VALUE_ATT, value);
|
---|
| 572 | }
|
---|
| 573 | }
|
---|
| 574 | return groupParamList;
|
---|
| 575 | }
|
---|
| 576 |
|
---|
[31181] | 577 | private Element getMRInfo(UserContext userContext){
|
---|
| 578 | Document doc = XMLConverter.newDOM();
|
---|
| 579 | Element infoMessage = doc.createElement(GSXML.MESSAGE_ELEM);
|
---|
| 580 | Element infoRequest = GSXML.createBasicRequest(doc, GSXML.REQUEST_TYPE_DESCRIBE, "", userContext);
|
---|
| 581 | Element paramList = doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
|
---|
| 582 | infoRequest.appendChild(paramList);
|
---|
| 583 | GSXML.addParameterToList(paramList, GSXML.SUBSET_PARAM, GSXML.SERVICE_ELEM + GSXML.LIST_MODIFIER);
|
---|
| 584 | infoMessage.appendChild(infoRequest);
|
---|
| 585 | Element responseMessage = (Element) this.router.process(infoMessage);
|
---|
| 586 | if (responseMessage == null)
|
---|
| 587 | {
|
---|
| 588 | logger.error("couldn't query the message router!");
|
---|
| 589 | return null;
|
---|
| 590 | }
|
---|
| 591 | Element infoResponse = (Element) GSXML.getChildByTagName(responseMessage, GSXML.RESPONSE_ELEM);
|
---|
| 592 | if (infoResponse == null)
|
---|
| 593 | {
|
---|
| 594 | logger.error("response from message router is null!");
|
---|
| 595 | return null;
|
---|
| 596 | }
|
---|
| 597 |
|
---|
| 598 | return infoResponse;
|
---|
| 599 |
|
---|
| 600 | }
|
---|
| 601 |
|
---|
[11014] | 602 | }
|
---|