1 | /*
|
---|
2 | * AbstractMGSearch.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.mg.*;
|
---|
23 | import org.greenstone.gsdl3.util.GSXML;
|
---|
24 | import org.greenstone.gsdl3.util.GSFile;
|
---|
25 | // XML classes
|
---|
26 | import org.w3c.dom.Document;
|
---|
27 | import org.w3c.dom.Element;
|
---|
28 | import org.w3c.dom.NodeList;
|
---|
29 |
|
---|
30 | // java
|
---|
31 | import java.util.Vector;
|
---|
32 | import java.util.ArrayList;
|
---|
33 | import java.util.HashMap;
|
---|
34 | import java.util.Map;
|
---|
35 | import java.util.Set;
|
---|
36 | import java.util.Iterator;
|
---|
37 | import java.io.File;
|
---|
38 |
|
---|
39 | /** Partially implements a generic MG search service
|
---|
40 | *
|
---|
41 | * @author <a href="mailto:[email protected]">Katherine Don</a>
|
---|
42 | */
|
---|
43 |
|
---|
44 | abstract public class AbstractMGSearch
|
---|
45 | extends AbstractSearch
|
---|
46 | {
|
---|
47 | protected static final String CASE_PARAM = "case";
|
---|
48 | protected static final String STEM_PARAM = "stem";
|
---|
49 | protected static final String MATCH_PARAM = "matchMode";
|
---|
50 | protected static final String MATCH_PARAM_ALL = "all";
|
---|
51 | protected static final String MATCH_PARAM_SOME = "some";
|
---|
52 |
|
---|
53 | protected static final String BOOLEAN_PARAM_ON = "1";
|
---|
54 | protected static final String BOOLEAN_PARAM_OFF = "0";
|
---|
55 |
|
---|
56 | protected static final String EQUIV_TERM_ELEM = "equivTerm";
|
---|
57 |
|
---|
58 | protected static final String STEM_ATT = "stem";
|
---|
59 | protected static final String NUM_DOCS_MATCH_ATT = "numDocsMatch";
|
---|
60 | protected static final String FREQ_ATT = "freq";
|
---|
61 |
|
---|
62 | // Elements used in the config file that are specific to this class
|
---|
63 | protected static final String DEFAULT_INDEX_ELEM = "defaultIndex";
|
---|
64 | protected static final String INDEX_STEM_ELEM = "indexStem";
|
---|
65 | protected static final String INDEX_ELEM = "index";
|
---|
66 |
|
---|
67 | /** the default index */
|
---|
68 | protected String default_index = null;
|
---|
69 | /** the stem used for the index files */
|
---|
70 | protected String index_stem = null;
|
---|
71 | protected MGWrapper mg_src = null;
|
---|
72 |
|
---|
73 |
|
---|
74 | public AbstractMGSearch()
|
---|
75 | {
|
---|
76 | this.mg_src = new MGWrapper();
|
---|
77 | this.dictionary_name = "MGSearch";
|
---|
78 | }
|
---|
79 |
|
---|
80 | public boolean configure(Element info, Element extra_info)
|
---|
81 | {
|
---|
82 | // Get the default index out of <defaultIndex> (buildConfig.xml)
|
---|
83 | Element def = (Element) GSXML.getChildByTagName(info, DEFAULT_INDEX_ELEM);
|
---|
84 | if (def != null) {
|
---|
85 | this.default_index = def.getAttribute(GSXML.NAME_ATT);
|
---|
86 | }
|
---|
87 | if (this.default_index == null || this.default_index.equals("")) {
|
---|
88 | System.err.println("Error: default index not specified!");
|
---|
89 | return false;
|
---|
90 | }
|
---|
91 |
|
---|
92 | // the index stem is either the collection name or is specified in the config file
|
---|
93 | Element index_stem_elem = (Element) GSXML.getChildByTagName(info, INDEX_STEM_ELEM);
|
---|
94 | if (index_stem_elem != null) {
|
---|
95 | this.index_stem = index_stem_elem.getAttribute(GSXML.NAME_ATT);
|
---|
96 | }
|
---|
97 | if (this.index_stem == null || this.index_stem.equals("")) {
|
---|
98 | System.err.println("AbstractMGSearch.configure(): indexStem element not found, stem will default to collection name");
|
---|
99 | this.index_stem = this.cluster_name;
|
---|
100 | }
|
---|
101 |
|
---|
102 | // get display info from extra info
|
---|
103 | if (extra_info !=null) {
|
---|
104 | Document owner = info.getOwnerDocument();
|
---|
105 | // so far we have index specific display elements, and global format elements
|
---|
106 | NodeList indexes = info.getElementsByTagName(GSXML.INDEX_ELEM);
|
---|
107 | Element config_search = (Element)GSXML.getChildByTagName(extra_info, GSXML.SEARCH_ELEM);
|
---|
108 |
|
---|
109 | for (int i=0; i<indexes.getLength();i++) {
|
---|
110 | Element ind = (Element)indexes.item(i);
|
---|
111 | String name = ind.getAttribute(GSXML.NAME_ATT);
|
---|
112 | Element node_extra = GSXML.getNamedElement(config_search,
|
---|
113 | GSXML.INDEX_ELEM,
|
---|
114 | GSXML.NAME_ATT,
|
---|
115 | name);
|
---|
116 | if (node_extra == null) {
|
---|
117 | System.err.println("GS2Search: haven't found extra info for index named "+name);
|
---|
118 | continue;
|
---|
119 | }
|
---|
120 |
|
---|
121 | // get the display elements if any - displayName
|
---|
122 | NodeList display_names = node_extra.getElementsByTagName(GSXML.DISPLAY_TEXT_ELEM);
|
---|
123 | if (display_names !=null) {
|
---|
124 | for (int j=0; j<display_names.getLength(); j++) {
|
---|
125 | Element e = (Element)display_names.item(j);
|
---|
126 | ind.appendChild(owner.importNode(e, true));
|
---|
127 | }
|
---|
128 | }
|
---|
129 | } // for each index
|
---|
130 | }
|
---|
131 | return super.configure(info, extra_info);
|
---|
132 | }
|
---|
133 |
|
---|
134 | protected void addCustomQueryParams(Element param_list, String lang)
|
---|
135 | {
|
---|
136 | createParameter(CASE_PARAM, param_list, lang);
|
---|
137 | createParameter(STEM_PARAM, param_list, lang);
|
---|
138 | createParameter(MATCH_PARAM, param_list, lang);
|
---|
139 | }
|
---|
140 |
|
---|
141 | protected void getIndexData(ArrayList index_ids, ArrayList index_names, String lang)
|
---|
142 | {
|
---|
143 | // the index info - read from config file - cache it??
|
---|
144 | Element index_list = (Element)GSXML.getChildByTagName(this.config_info, INDEX_ELEM+GSXML.LIST_MODIFIER);
|
---|
145 | NodeList indexes = index_list.getElementsByTagName(INDEX_ELEM);
|
---|
146 | int len = indexes.getLength();
|
---|
147 | // now add even if there is only one
|
---|
148 | for (int i=0; i<len; i++) {
|
---|
149 | Element index = (Element)indexes.item(i);
|
---|
150 | index_ids.add(index.getAttribute(GSXML.NAME_ATT));
|
---|
151 | index_names.add(GSXML.getDisplayText(index, GSXML.DISPLAY_TEXT_NAME, lang, "en"));
|
---|
152 |
|
---|
153 | }
|
---|
154 |
|
---|
155 | }
|
---|
156 |
|
---|
157 | /** do the actual query */
|
---|
158 | protected Element processTextQuery(Element request)
|
---|
159 | {
|
---|
160 |
|
---|
161 | // Create a new (empty) result message
|
---|
162 | Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
|
---|
163 | result.setAttribute(GSXML.FROM_ATT, TEXT_QUERY_SERVICE);
|
---|
164 | result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
|
---|
165 |
|
---|
166 | // Get the parameters of the request
|
---|
167 | Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
|
---|
168 | if (param_list == null) {
|
---|
169 | System.err.println("Error: TextQuery request had no paramList.");
|
---|
170 | return result; // Return the empty result
|
---|
171 | }
|
---|
172 |
|
---|
173 | // Process the request parameters
|
---|
174 | HashMap params = GSXML.extractParams(param_list, false);
|
---|
175 |
|
---|
176 | // Make sure a query has been specified
|
---|
177 | String query = (String) params.get(QUERY_PARAM);
|
---|
178 | if (query == null || query.equals("")) {
|
---|
179 | return result; // Return the empty result
|
---|
180 | }
|
---|
181 |
|
---|
182 | // If an index hasn't been specified, use the default
|
---|
183 | String index = (String) params.get(INDEX_PARAM);
|
---|
184 | if (index == null) {
|
---|
185 | index = this.default_index;
|
---|
186 | }
|
---|
187 |
|
---|
188 | // The location of the MG index and text files
|
---|
189 | String basedir = GSFile.collectionBaseDir(this.site_home, this.cluster_name) + File.separatorChar; // Needed for MG
|
---|
190 | String textdir = GSFile.collectionTextPath(this.index_stem);
|
---|
191 | String indexpath = GSFile.collectionIndexPath(this.index_stem, index);
|
---|
192 | this.mg_src.setIndex(indexpath);
|
---|
193 |
|
---|
194 | // set the mg query parameters to the values the user has specified
|
---|
195 | setStandardQueryParams(params);
|
---|
196 | this.mg_src.runQuery(basedir, textdir, query);
|
---|
197 | MGQueryResult mqr = this.mg_src.getQueryResult();
|
---|
198 | long totalDocs = mqr.getTotalDocs();
|
---|
199 |
|
---|
200 | // Get the docnums out, and convert to HASH ids
|
---|
201 | Vector docs = mqr.getDocs();
|
---|
202 | if (docs.size() == 0) {
|
---|
203 | System.err.println("GS2MGSearch: Warning: No results found...\n");
|
---|
204 | }
|
---|
205 |
|
---|
206 | // Create a metadata list to store information about the query results
|
---|
207 | Element metadata_list = this.doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
|
---|
208 | result.appendChild(metadata_list);
|
---|
209 |
|
---|
210 | // Add a metadata element specifying the number of matching documents
|
---|
211 | // because teh total number is just the number returned, use numDocsReturned, not numDocsMatched
|
---|
212 | GSXML.addMetadata(this.doc, metadata_list, "numDocsReturned", ""+totalDocs);
|
---|
213 | // 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
|
---|
214 | GSXML.addMetadata(this.doc, metadata_list, "query", query);
|
---|
215 |
|
---|
216 | // Create a document list to store the matching documents, and add them
|
---|
217 | Element document_list = this.doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
|
---|
218 | result.appendChild(document_list);
|
---|
219 | for (int d = 0; d < docs.size(); d++) {
|
---|
220 | long docnum = ((MGDocInfo) docs.elementAt(d)).num_;
|
---|
221 | float rank = ((MGDocInfo) docs.elementAt(d)).rank_;
|
---|
222 | String doc_id = MGNum2OID(docnum);
|
---|
223 | Element doc_node = createDocNode(doc_id, Float.toString(rank));
|
---|
224 | document_list.appendChild(doc_node);
|
---|
225 | }
|
---|
226 |
|
---|
227 | // Create a term list to store the term information, and add it
|
---|
228 | Element term_list = this.doc.createElement(GSXML.TERM_ELEM+GSXML.LIST_MODIFIER);
|
---|
229 | result.appendChild(term_list);
|
---|
230 | Vector terms = mqr.getTerms();
|
---|
231 | for (int t = 0; t < terms.size(); t++) {
|
---|
232 | MGTermInfo term_info = (MGTermInfo) terms.get(t);
|
---|
233 |
|
---|
234 | String term = term_info.term_;
|
---|
235 | int stem_method = term_info.stem_method_;
|
---|
236 | Vector equiv_terms = term_info.equiv_terms_;
|
---|
237 |
|
---|
238 | Element term_elem = this.doc.createElement(GSXML.TERM_ELEM);
|
---|
239 | term_elem.setAttribute(GSXML.NAME_ATT, term);
|
---|
240 | term_elem.setAttribute(STEM_ATT, "" + stem_method);
|
---|
241 |
|
---|
242 | Element equiv_term_list = this.doc.createElement(EQUIV_TERM_ELEM+GSXML.LIST_MODIFIER);
|
---|
243 | term_elem.appendChild(equiv_term_list);
|
---|
244 |
|
---|
245 | long total_term_freq = 0;
|
---|
246 | for (int et = 0; et < equiv_terms.size(); et++) {
|
---|
247 | MGEquivTermInfo equiv_term_info = (MGEquivTermInfo) equiv_terms.get(et);
|
---|
248 |
|
---|
249 | Element equiv_term_elem = this.doc.createElement(GSXML.TERM_ELEM);
|
---|
250 | equiv_term_elem.setAttribute(GSXML.NAME_ATT, equiv_term_info.term_);
|
---|
251 | equiv_term_elem.setAttribute(NUM_DOCS_MATCH_ATT, "" + equiv_term_info.match_docs_);
|
---|
252 | equiv_term_elem.setAttribute(FREQ_ATT, "" + equiv_term_info.term_freq_);
|
---|
253 | equiv_term_list.appendChild(equiv_term_elem);
|
---|
254 |
|
---|
255 | total_term_freq += equiv_term_info.term_freq_;
|
---|
256 | }
|
---|
257 |
|
---|
258 | term_elem.setAttribute(FREQ_ATT, "" + total_term_freq);
|
---|
259 | term_list.appendChild(term_elem);
|
---|
260 | }
|
---|
261 | return result;
|
---|
262 | }
|
---|
263 |
|
---|
264 | // should probably use a list rather than map
|
---|
265 | protected boolean setStandardQueryParams(HashMap params)
|
---|
266 | {
|
---|
267 | // set the default ones
|
---|
268 | this.mg_src.setReturnTerms(true);
|
---|
269 | this.mg_src.setCase(true); // turn casefolding on by default
|
---|
270 | Set entries = params.entrySet();
|
---|
271 | Iterator i = entries.iterator();
|
---|
272 | while (i.hasNext()) {
|
---|
273 | Map.Entry m = (Map.Entry)i.next();
|
---|
274 | String name = (String)m.getKey();
|
---|
275 | String value = (String)m.getValue();
|
---|
276 |
|
---|
277 | if (name.equals(CASE_PARAM)) {
|
---|
278 | boolean val = (value.equals(BOOLEAN_PARAM_ON) ? true : false);
|
---|
279 | this.mg_src.setCase(val);
|
---|
280 | }
|
---|
281 | else if (name.equals(STEM_PARAM)) {
|
---|
282 | boolean val = (value.equals(BOOLEAN_PARAM_ON) ? true : false);
|
---|
283 | this.mg_src.setStem(val);
|
---|
284 | }
|
---|
285 | else if (name.equals(MATCH_PARAM)) {
|
---|
286 | int mode = (value.equals(MATCH_PARAM_ALL) ? 1 : 0);
|
---|
287 | this.mg_src.setMatchMode(mode);
|
---|
288 | }
|
---|
289 | else if (name.equals(MAXDOCS_PARAM)) {
|
---|
290 | int docs = Integer.parseInt(value);
|
---|
291 | this.mg_src.setMaxDocs(docs);
|
---|
292 | } // ignore any others
|
---|
293 | }
|
---|
294 | return true;
|
---|
295 | }
|
---|
296 |
|
---|
297 | /** convert MG internal id to Greenstone oid */
|
---|
298 | abstract protected String MGNum2OID(long docnum);
|
---|
299 | }
|
---|