source: trunk/gsdl3/src/java/org/greenstone/gsdl3/service/LuceneSearch.java@ 9272

Last change on this file since 9272 was 9272, checked in by kjdon, 19 years ago

made this inherit from AbstractSearch

  • Property svn:keywords set to Author Date Id Revision
File size: 4.1 KB
Line 
1package org.greenstone.gsdl3.service;
2
3// Greenstone classes
4import org.greenstone.gdbm.*;
5import org.greenstone.gsdl3.util.*;
6
7// XML classes
8import org.w3c.dom.Element;
9import org.w3c.dom.Document;
10import org.w3c.dom.NodeList;
11
12import java.util.HashMap;
13import java.util.ArrayList;
14
15import org.apache.lucene.analysis.Analyzer;
16import org.apache.lucene.analysis.standard.StandardAnalyzer;
17import org.apache.lucene.document.*; //Document;
18import org.apache.lucene.search.Searcher;
19import org.apache.lucene.search.IndexSearcher;
20import org.apache.lucene.search.Query;
21import org.apache.lucene.search.Hits;
22import org.apache.lucene.queryParser.QueryParser;
23import org.apache.lucene.search.TermQuery;
24import org.apache.lucene.index.Term;
25
26import java.io.File;
27/**
28 *
29 * @author <a href="mailto:[email protected]">Katherine Don</a>
30 * @version $Revision: 9272 $
31 */
32
33public class LuceneSearch
34 extends AbstractSearch {
35
36 protected static final String INDEX_ELEM = "index";
37
38 protected void getIndexData(ArrayList index_ids, ArrayList index_names, String lang)
39 {
40 // the index info - read from config file - cache it??
41 Element index_list = (Element)GSXML.getChildByTagName(this.config_info, INDEX_ELEM+GSXML.LIST_MODIFIER);
42 if (index_list != null) {
43 NodeList indexes = index_list.getElementsByTagName(INDEX_ELEM);
44 int len = indexes.getLength();
45 // now add even if there is only one
46 for (int i=0; i<len; i++) {
47 Element index = (Element)indexes.item(i);
48 index_ids.add(index.getAttribute(GSXML.NAME_ATT));
49 index_names.add(GSXML.getDisplayText(index, GSXML.DISPLAY_TEXT_NAME, lang, "en"));
50
51 }
52 } else {
53 // there is only one index, so we assume the default
54 index_ids.add("idx");
55 index_names.add("Default index");
56 }
57
58 }
59
60 /** Process a text query - implemented by concrete subclasses */
61 protected Element processTextQuery(Element request) {
62
63 // Create a new (empty) result message
64 Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
65 result.setAttribute(GSXML.FROM_ATT, TEXT_QUERY_SERVICE);
66 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
67 Element doc_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
68 result.appendChild(doc_node_list);
69 Element metadata_list = this.doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
70 result.appendChild(metadata_list);
71 // Get the parameters of the request
72 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
73 if (param_list == null) {
74 System.err.println("LuceneSearch Error: TextQuery request had no paramList.");
75 GSXML.addMetadata(this.doc, metadata_list, "numDocsMatched", "0");
76 return result; // Return the empty result
77 }
78
79 // Process the request parameters
80 HashMap params = GSXML.extractParams(param_list, false);
81
82 // Make sure a query has been specified
83 String query_string = (String) params.get(QUERY_PARAM);
84 if (query_string == null || query_string.equals("")) {
85 System.err.println("LuceneSearch Error: TextQuery request had no query string.");
86 GSXML.addMetadata(this.doc, metadata_list, "numDocsMatched", "0");
87 return result; // Return the empty result
88 }
89
90 // Get the index
91 String index = (String) params.get(INDEX_PARAM);
92 try {
93 String index_dir = GSFile.collectionIndexDir(this.site_home, this.cluster_name);
94 index_dir += File.separator+index;
95 Searcher searcher = new IndexSearcher(index_dir);
96 Analyzer analyzer = new StandardAnalyzer();
97
98 Term term = new Term("content", query_string);
99
100 Query query = new TermQuery(term);
101
102 Hits hits = searcher.search(query);
103 GSXML.addMetadata(this.doc, metadata_list, "numDocsMatched", ""+hits.length());
104
105 for (int i=0; i<hits.length(); i++) {
106 org.apache.lucene.document.Document luc_doc = hits.doc(i);
107 String node_id = luc_doc.get("nodeID");
108 Element node = this.doc.createElement(GSXML.DOC_NODE_ELEM);
109 node.setAttribute(GSXML.NODE_ID_ATT, node_id);
110 doc_node_list.appendChild(node);
111 }
112 } catch (Exception e) {
113 e.printStackTrace();
114 }
115
116 return result;
117
118 }
119
120
121}
Note: See TracBrowser for help on using the repository browser.