[9886] | 1 | package org.greenstone.gsdl3.service;
|
---|
| 2 |
|
---|
| 3 | // Greenstone classes
|
---|
| 4 | import org.greenstone.gsdl3.util.*;
|
---|
| 5 |
|
---|
| 6 | //Fedora Web Services classes
|
---|
| 7 | import fedora.client.utility.AutoFinder;
|
---|
| 8 | import fedora.client.APIAStubFactory;
|
---|
| 9 | import fedora.client.Downloader;
|
---|
| 10 | import fedora.server.access.FedoraAPIA;
|
---|
| 11 | import fedora.server.types.gen.FieldSearchQuery;
|
---|
| 12 | import fedora.server.types.gen.FieldSearchResult;
|
---|
| 13 | import fedora.server.types.gen.ListSession;
|
---|
| 14 | import fedora.server.types.gen.ObjectFields;
|
---|
| 15 | import fedora.server.types.gen.Condition;
|
---|
| 16 | import fedora.server.types.gen.ComparisonOperator;
|
---|
| 17 |
|
---|
| 18 | import javax.xml.transform.stream.StreamResult;
|
---|
| 19 | import javax.xml.transform.stream.StreamSource;
|
---|
| 20 | import javax.xml.transform.Templates;
|
---|
| 21 | import javax.xml.transform.Transformer;
|
---|
| 22 | import javax.xml.transform.TransformerFactory;
|
---|
| 23 |
|
---|
| 24 | // XML classes
|
---|
| 25 | import org.w3c.dom.Element;
|
---|
| 26 | import org.w3c.dom.Document;
|
---|
| 27 | import org.w3c.dom.NodeList;
|
---|
| 28 |
|
---|
| 29 | //Java classes
|
---|
| 30 | import java.util.ArrayList;
|
---|
| 31 | import java.util.HashMap;
|
---|
| 32 | import java.util.List;
|
---|
| 33 | import java.util.Properties;
|
---|
| 34 | import java.io.File;
|
---|
| 35 | import java.io.InputStream;
|
---|
| 36 | import java.io.BufferedReader;
|
---|
| 37 | import java.io.InputStreamReader;
|
---|
| 38 | import java.io.IOException;
|
---|
| 39 | import java.net.HttpURLConnection;
|
---|
| 40 | import java.net.URLConnection;
|
---|
| 41 | import java.net.URL;
|
---|
| 42 | import java.net.Authenticator;
|
---|
| 43 | import java.net.MalformedURLException;
|
---|
| 44 | import java.lang.Object;
|
---|
| 45 |
|
---|
[20294] | 46 | import org.apache.log4j.*;
|
---|
[9886] | 47 |
|
---|
| 48 | /**
|
---|
| 49 | *
|
---|
| 50 | *
|
---|
| 51 | */
|
---|
| 52 |
|
---|
| 53 | public class FedoraSearch
|
---|
| 54 | extends AbstractSearch {
|
---|
| 55 |
|
---|
[20294] | 56 | static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.FedoraSearch.class.getName());
|
---|
[9886] | 57 | // the services on offer
|
---|
| 58 | // these strings must match what is found in the properties file
|
---|
| 59 | protected static final String FIELD_PARAM = "fields";
|
---|
| 60 |
|
---|
| 61 | //Parameters connect to Proxy Server
|
---|
| 62 | private String proxyUser = null;
|
---|
| 63 | private String proxyPassword = null;
|
---|
| 64 |
|
---|
| 65 | private int maxResult, matchNum;
|
---|
| 66 |
|
---|
| 67 | protected String fedora_server_url = null;
|
---|
| 68 |
|
---|
| 69 | public FedoraSearch()
|
---|
| 70 | {
|
---|
[28969] | 71 | QUERY_SERVICE = "TextQuery";
|
---|
[9886] | 72 | }
|
---|
| 73 |
|
---|
| 74 | //Configure FedoraSearch Service
|
---|
| 75 | public boolean configure(Element info, Element extra_info)
|
---|
| 76 | {
|
---|
| 77 | Element server_elem = (Element)GSXML.getChildByTagName(info, "fedoraServer");
|
---|
| 78 | if (server_elem == null) {
|
---|
[20294] | 79 | logger.error("no fedoraServer element found");
|
---|
[9886] | 80 | return false;
|
---|
| 81 | }
|
---|
| 82 | fedora_server_url = server_elem.getAttribute("url");
|
---|
[20294] | 83 | logger.info("Fedora Server:" + fedora_server_url);
|
---|
[9886] | 84 |
|
---|
| 85 | if (fedora_server_url.equals("")) {
|
---|
[20294] | 86 | logger.error("no url for the fedoraServer element");
|
---|
[9886] | 87 | return false;
|
---|
| 88 | }
|
---|
[28969] | 89 | Element tq_service = this.desc_doc.createElement(GSXML.SERVICE_ELEM);
|
---|
[9886] | 90 | tq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
|
---|
[28969] | 91 | tq_service.setAttribute(GSXML.NAME_ATT, QUERY_SERVICE);
|
---|
[9886] | 92 | this.short_service_info.appendChild(tq_service);
|
---|
| 93 |
|
---|
| 94 | // add some format info to service map if there is any
|
---|
| 95 | String path = GSPath.appendLink(GSXML.SEARCH_ELEM, GSXML.FORMAT_ELEM);
|
---|
| 96 | Element format = (Element) GSXML.getNodeByPath(extra_info, path);
|
---|
| 97 | if (format != null) {
|
---|
[28969] | 98 | this.format_info_map.put(QUERY_SERVICE, this.desc_doc.importNode(format, true));
|
---|
[9886] | 99 | }
|
---|
| 100 | return true;
|
---|
| 101 | }
|
---|
| 102 |
|
---|
| 103 | /** Process a text query - implemented by concrete subclasses */
|
---|
| 104 | protected Element processTextQuery(Element request) {
|
---|
| 105 | // Create a new (empty) result message
|
---|
[28969] | 106 | Document result_doc = XMLConverter.newDOM();
|
---|
| 107 | Element result = result_doc.createElement(GSXML.RESPONSE_ELEM);
|
---|
| 108 | result.setAttribute(GSXML.FROM_ATT, QUERY_SERVICE);
|
---|
[9886] | 109 | result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
|
---|
| 110 |
|
---|
[28969] | 111 | Element doc_node_list = result_doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
|
---|
[9886] | 112 | result.appendChild(doc_node_list);
|
---|
| 113 |
|
---|
| 114 | Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
|
---|
| 115 | if (param_list == null) {
|
---|
[20294] | 116 | logger.error("TextQuery request had no paramList.");
|
---|
[9886] | 117 | return result; // Return the empty result
|
---|
| 118 | }
|
---|
| 119 |
|
---|
| 120 | // Process the request parameters
|
---|
| 121 | HashMap params = GSXML.extractParams(param_list, false);
|
---|
| 122 |
|
---|
| 123 | // Make sure a query has been specified
|
---|
| 124 | String query = (String) params.get(QUERY_PARAM);
|
---|
| 125 | if (query == null || query.equals("")) {
|
---|
| 126 | return result; // Return the empty result
|
---|
| 127 | }
|
---|
| 128 | //Check Max number of Documents hit
|
---|
| 129 | String maxDocs = (String) params.get(MAXDOCS_PARAM);
|
---|
| 130 | String hits_per_page = (String) params.get(HITS_PER_PAGE_PARAM);
|
---|
| 131 |
|
---|
| 132 | //Check the start_page number
|
---|
| 133 | String start_page = (String) params.get(START_PAGE_PARAM);
|
---|
| 134 |
|
---|
| 135 | // tidy whitespace
|
---|
| 136 | //query = query.replaceAll("\\s+", "+");
|
---|
| 137 | matchNum = 0;
|
---|
| 138 | maxResult = 20;
|
---|
| 139 |
|
---|
| 140 | try {
|
---|
| 141 | //Search Fedora Server
|
---|
| 142 | String host = "toetoe";
|
---|
| 143 | int port = 8200;
|
---|
| 144 | String user = "fedoraAdmin";
|
---|
| 145 | String pass = "fedoraAdmin";
|
---|
| 146 |
|
---|
| 147 | AutoFinder finder=new AutoFinder(host, port, user, pass);
|
---|
| 148 |
|
---|
| 149 | String[] fieldsArray = new String[] {"pid", "label", "fType", "bDef",
|
---|
| 150 | "bMech", "cModel", "state", "ownerId", "cDate", "mDate",
|
---|
| 151 | "dcmDate", "title", "creator", "subject", "description",
|
---|
| 152 | "publisher", "contributor", "date", "type", "format",
|
---|
| 153 | "identifier", "source", "language", "relation", "coverage",
|
---|
| 154 | "rights"};
|
---|
| 155 |
|
---|
| 156 | String match_fields = join(fieldsArray, "|");
|
---|
| 157 |
|
---|
| 158 | FieldSearchQuery query_fedora=fedoraSearch(query, match_fields);
|
---|
| 159 |
|
---|
| 160 | FieldSearchResult result_fedora = finder.findObjects(fieldsArray, maxResult, query_fedora);
|
---|
| 161 |
|
---|
| 162 | while (result_fedora != null) {
|
---|
[20294] | 163 | logger.info("Search Result:" + result_fedora.getResultList().length);
|
---|
[9886] | 164 | for (int i=0; i<result_fedora.getResultList().length; i++) {
|
---|
| 165 | ObjectFields o = result_fedora.getResultList()[i];
|
---|
| 166 | matchNum++;
|
---|
| 167 | String pid = o.getPid();
|
---|
| 168 | String fedora_url = fedora_server_url +"/fedora/get/" + pid;
|
---|
| 169 | String label = o.getLabel();
|
---|
| 170 | String[] creator = o.getCreator();
|
---|
| 171 | String[] subject = o.getSubject();
|
---|
| 172 | String[] publisher = o.getPublisher();
|
---|
| 173 | String[] contributor = o.getContributor();
|
---|
| 174 | String[] title = o.getTitle();
|
---|
| 175 | String[] description = o.getDescription();
|
---|
| 176 | String[] format = o.getFormat();
|
---|
| 177 | String[] identifier = o.getIdentifier();
|
---|
| 178 | String[] source = o.getSource();
|
---|
| 179 | String[] language = o.getLanguage();
|
---|
| 180 |
|
---|
[28969] | 181 | Element doc_node = result_doc.createElement(GSXML.DOC_NODE_ELEM);
|
---|
[9886] | 182 | doc_node_list.appendChild(doc_node);
|
---|
| 183 |
|
---|
[28969] | 184 | Element metadata_list = result_doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
|
---|
[9886] | 185 | doc_node.appendChild(metadata_list);
|
---|
| 186 |
|
---|
| 187 | if (fedora_url != null) {
|
---|
[28966] | 188 | GSXML.addMetadata( metadata_list, "URL", fedora_url);
|
---|
[9886] | 189 | }
|
---|
| 190 | if (o.getDescription() != null) {
|
---|
[28966] | 191 | GSXML.addMetadata(metadata_list, "Description", description[0]);
|
---|
[9886] | 192 | }
|
---|
| 193 | if (title[0] != null) {
|
---|
[28966] | 194 | GSXML.addMetadata(metadata_list, "Title", title[0]);
|
---|
[9886] | 195 | }
|
---|
| 196 | }
|
---|
| 197 | ListSession sess=result_fedora.getListSession();
|
---|
| 198 | if (sess!=null) {
|
---|
| 199 | result_fedora = finder.resumeFindObjects(sess.getToken());
|
---|
| 200 | } else {
|
---|
| 201 | result_fedora = null;
|
---|
| 202 | }
|
---|
| 203 | }
|
---|
| 204 | } catch (Exception e) {
|
---|
[20294] | 205 | logger.error("ERROR: " + e.getClass().getName()
|
---|
[9886] | 206 | + ((e.getMessage()==null) ? "" : ": " + e.getMessage()));
|
---|
| 207 | }
|
---|
| 208 |
|
---|
[28969] | 209 | Element metadata_list = result_doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
|
---|
[9886] | 210 | result.appendChild(metadata_list);
|
---|
| 211 |
|
---|
| 212 | // Add a metadata element specifying the number of matching documents
|
---|
[20294] | 213 | //logger.info("FedoraSearch Number of Documents:" + matchNum);
|
---|
[28966] | 214 | GSXML.addMetadata(metadata_list, "numDocsMatched", ""+matchNum);
|
---|
[9886] | 215 |
|
---|
| 216 | return result;
|
---|
| 217 | }
|
---|
| 218 |
|
---|
| 219 | protected FieldSearchQuery fedoraSearch(String query_terms, String matchFields) {
|
---|
| 220 |
|
---|
| 221 | FieldSearchQuery fedora_query=new FieldSearchQuery();
|
---|
| 222 | try {
|
---|
| 223 | // check if query contains any fields name (pid, label, fType....)
|
---|
| 224 | if (query_terms.matches("^("+ matchFields+").+")){
|
---|
[20294] | 225 | logger.info("Implementing Specific Query....");
|
---|
[9886] | 226 |
|
---|
| 227 | List conditions=fedora.server.search.Condition.getConditions(query_terms);
|
---|
| 228 |
|
---|
| 229 | Condition[] cond=new Condition[conditions.size()];
|
---|
| 230 | for (int i=0; i<conditions.size(); i++) {
|
---|
| 231 | fedora.server.search.Condition c = (fedora.server.search.Condition) conditions.get(i);
|
---|
| 232 | try {
|
---|
| 233 | String operator = transOperator(c.getOperator().getAbbreviation());
|
---|
| 234 | cond[i] = new Condition();
|
---|
| 235 | cond[i].setProperty(c.getProperty());
|
---|
| 236 | cond[i].setOperator(ComparisonOperator.fromValue(operator));
|
---|
| 237 | cond[i].setValue(c.getValue());
|
---|
| 238 | } catch (Exception e){
|
---|
| 239 | e.printStackTrace();
|
---|
| 240 | }
|
---|
| 241 | }
|
---|
| 242 | fedora_query.setConditions(cond);
|
---|
| 243 | } else {
|
---|
[20294] | 244 | logger.info("Implementing Simple Query...");
|
---|
[9886] | 245 | fedora_query.setTerms(query_terms);
|
---|
| 246 | }
|
---|
| 247 | } catch (Exception e){
|
---|
| 248 | e.printStackTrace();
|
---|
| 249 | }
|
---|
| 250 | return fedora_query;
|
---|
| 251 | }
|
---|
| 252 |
|
---|
| 253 | protected void getIndexData(ArrayList index_ids, ArrayList index_names,String lang){
|
---|
| 254 | index_ids.add("kw");
|
---|
| 255 | index_ids.add("au");
|
---|
| 256 | index_ids.add("su");
|
---|
| 257 | index_ids.add("ti");
|
---|
| 258 | index_ids.add("de");
|
---|
| 259 | index_ids.add("fu");
|
---|
| 260 | index_names.add(getTextString("param."+FIELD_PARAM+".kw", lang));
|
---|
| 261 | index_names.add(getTextString("param."+FIELD_PARAM+".au", lang));
|
---|
| 262 | index_names.add(getTextString("param."+FIELD_PARAM+".su", lang));
|
---|
| 263 | index_names.add(getTextString("param."+FIELD_PARAM+".ti", lang));
|
---|
| 264 | index_names.add(getTextString("param."+FIELD_PARAM+".de", lang));
|
---|
| 265 | index_names.add(getTextString("param."+FIELD_PARAM+".fu", lang));
|
---|
| 266 | }
|
---|
| 267 |
|
---|
| 268 | protected String getDocType (String node_id){
|
---|
| 269 | return node_id;
|
---|
| 270 | }
|
---|
| 271 | protected boolean hasChildren(String node_id){
|
---|
| 272 | return false;
|
---|
| 273 | }
|
---|
| 274 |
|
---|
| 275 | protected boolean hasParent(String node_id){
|
---|
| 276 | return false;
|
---|
| 277 | }
|
---|
| 278 | private String transOperator(String operator){
|
---|
| 279 | String oper_char ="";
|
---|
| 280 | if (operator == "~") {
|
---|
| 281 | oper_char = "has";
|
---|
| 282 | } else if (operator == "=") {
|
---|
| 283 | oper_char = "eq";
|
---|
| 284 | } else if (operator == ">"){
|
---|
| 285 | oper_char = "gt";
|
---|
| 286 | } else if (operator == ">="){
|
---|
| 287 | oper_char = "ge";
|
---|
| 288 | } else if (operator == "<"){
|
---|
| 289 | oper_char = "lt";
|
---|
| 290 | } else if (operator == "<="){
|
---|
| 291 | oper_char = "le";
|
---|
| 292 | }
|
---|
| 293 | return oper_char;
|
---|
| 294 | }
|
---|
| 295 | protected String join( String[] array, String delim ) {
|
---|
| 296 | String j = "";
|
---|
| 297 | for ( int i=0; i<array.length; i++ ) {
|
---|
| 298 | if (i!=0) j += delim;
|
---|
| 299 | j += array[i];
|
---|
| 300 | }
|
---|
| 301 | return j;
|
---|
| 302 | }
|
---|
| 303 | }
|
---|