1 | package org.greenstone.gsdl3.service;
|
---|
2 |
|
---|
3 | // Greenstone classes
|
---|
4 | import java.io.File;
|
---|
5 | import java.io.Serializable;
|
---|
6 | import java.util.ArrayList;
|
---|
7 | import java.util.HashMap;
|
---|
8 |
|
---|
9 | import org.apache.log4j.Logger;
|
---|
10 | import org.apache.lucene.index.IndexReader;
|
---|
11 | import org.apache.lucene.index.Term;
|
---|
12 | import org.apache.lucene.search.IndexSearcher;
|
---|
13 | import org.apache.lucene.search.Query;
|
---|
14 | import org.apache.lucene.search.TermQuery;
|
---|
15 | import org.apache.lucene.search.TopDocs;
|
---|
16 | import org.apache.lucene.store.Directory;
|
---|
17 | import org.apache.lucene.store.FSDirectory;
|
---|
18 | import org.greenstone.gsdl3.util.GSFile;
|
---|
19 | import org.greenstone.gsdl3.util.GSXML;
|
---|
20 | import org.w3c.dom.Element;
|
---|
21 | import org.w3c.dom.NodeList;
|
---|
22 |
|
---|
23 | public class LuceneSearch extends AbstractTextSearch
|
---|
24 | {
|
---|
25 | static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.LuceneSearch.class.getName());
|
---|
26 |
|
---|
27 | protected static final String INDEX_ELEM = "index";
|
---|
28 |
|
---|
29 | protected ArrayList<String> index_ids;
|
---|
30 |
|
---|
31 | public LuceneSearch()
|
---|
32 | {
|
---|
33 | index_ids = new ArrayList<String>();
|
---|
34 | }
|
---|
35 |
|
---|
36 | public boolean configure(Element info, Element extra_info)
|
---|
37 | {
|
---|
38 | if (!super.configure(info, extra_info))
|
---|
39 | {
|
---|
40 | return false;
|
---|
41 | }
|
---|
42 |
|
---|
43 | default_index = "idx";
|
---|
44 |
|
---|
45 | // cache index info read from config file
|
---|
46 | Element index_list = (Element) GSXML.getChildByTagName(this.config_info, INDEX_ELEM + GSXML.LIST_MODIFIER);
|
---|
47 | if (index_list != null)
|
---|
48 | {
|
---|
49 | NodeList indexes = index_list.getElementsByTagName(INDEX_ELEM);
|
---|
50 | int len = indexes.getLength();
|
---|
51 | // now add even if there is only one
|
---|
52 | for (int i = 0; i < len; i++)
|
---|
53 | {
|
---|
54 | Element index = (Element) indexes.item(i);
|
---|
55 | index_ids.add(index.getAttribute(GSXML.NAME_ATT));
|
---|
56 | }
|
---|
57 | }
|
---|
58 | else
|
---|
59 | {
|
---|
60 | // there is only one index, so we assume the default
|
---|
61 | index_ids.add(this.default_index);
|
---|
62 | }
|
---|
63 |
|
---|
64 | return true;
|
---|
65 | }
|
---|
66 |
|
---|
67 | protected void getIndexData(ArrayList<String> index_ids, ArrayList<String> index_names, String lang)
|
---|
68 | {
|
---|
69 | // copying exercise for index_ids,
|
---|
70 | for (int i = 0; i < this.index_ids.size(); i++)
|
---|
71 | {
|
---|
72 | index_ids.add(this.index_ids.get(i));
|
---|
73 | }
|
---|
74 |
|
---|
75 | // But need to work out display name from scratch as this uses
|
---|
76 | // the 'lang' parameter
|
---|
77 |
|
---|
78 | Element index_list = (Element) GSXML.getChildByTagName(this.config_info, INDEX_ELEM + GSXML.LIST_MODIFIER);
|
---|
79 | if (index_list != null)
|
---|
80 | {
|
---|
81 | NodeList indexes = index_list.getElementsByTagName(INDEX_ELEM);
|
---|
82 | int len = indexes.getLength();
|
---|
83 | // now add even if there is only one
|
---|
84 | for (int i = 0; i < len; i++)
|
---|
85 | {
|
---|
86 | Element index = (Element) indexes.item(i);
|
---|
87 | index_names.add(GSXML.getDisplayText(index, GSXML.DISPLAY_TEXT_NAME, lang, "en"));
|
---|
88 |
|
---|
89 | }
|
---|
90 | }
|
---|
91 | else
|
---|
92 | {
|
---|
93 | // there is only one index, so we assume the default
|
---|
94 | index_names.add("default index");
|
---|
95 | }
|
---|
96 | }
|
---|
97 |
|
---|
98 | protected void initResultElement(Element result, Element doc_node_list, Element metadata_list)
|
---|
99 | {
|
---|
100 | // Create a new (empty) result message
|
---|
101 | result.setAttribute(GSXML.FROM_ATT, QUERY_SERVICE);
|
---|
102 | result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
|
---|
103 | result.appendChild(doc_node_list);
|
---|
104 | result.appendChild(metadata_list);
|
---|
105 | }
|
---|
106 |
|
---|
107 | protected boolean hasParamList(Element request, Element metadata_list)
|
---|
108 | {
|
---|
109 | // Get the parameters of the request
|
---|
110 | Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
|
---|
111 | if (param_list == null)
|
---|
112 | {
|
---|
113 | logger.error("TextQuery request had no paramList.");
|
---|
114 | GSXML.addMetadata(this.doc, metadata_list, "numDocsMatched", "0");
|
---|
115 | return false; // signal that an empty result should be return
|
---|
116 | }
|
---|
117 |
|
---|
118 | return true;
|
---|
119 | }
|
---|
120 |
|
---|
121 | protected boolean hasQueryString(Element param_list, Element metadata_list)
|
---|
122 | {
|
---|
123 | // Process the request parameters to make sure a query has been specified
|
---|
124 | HashMap<String, Serializable> params = GSXML.extractParams(param_list, false);
|
---|
125 | String query_string = (String) params.get(QUERY_PARAM);
|
---|
126 |
|
---|
127 | if (query_string == null || query_string.equals(""))
|
---|
128 | {
|
---|
129 | logger.error("TextQuery request had no query string.");
|
---|
130 | GSXML.addMetadata(this.doc, metadata_list, "numDocsMatched", "0");
|
---|
131 | return false; // signal that an empty result should be return
|
---|
132 | }
|
---|
133 |
|
---|
134 | return true;
|
---|
135 | }
|
---|
136 |
|
---|
137 | /** Process a text query - implemented by concrete subclasses */
|
---|
138 | protected Element processTextQuery(Element request)
|
---|
139 | {
|
---|
140 | Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
|
---|
141 | Element doc_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
|
---|
142 | Element metadata_list = this.doc.createElement(GSXML.METADATA_ELEM + GSXML.LIST_MODIFIER);
|
---|
143 | initResultElement(result, doc_node_list, metadata_list);
|
---|
144 |
|
---|
145 | if (!hasParamList(request, metadata_list))
|
---|
146 | {
|
---|
147 | return result;
|
---|
148 | }
|
---|
149 |
|
---|
150 | Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
|
---|
151 | if (!hasQueryString(param_list, metadata_list))
|
---|
152 | {
|
---|
153 | return result;
|
---|
154 | }
|
---|
155 |
|
---|
156 | HashMap<String, Serializable> params = GSXML.extractParams(param_list, false);
|
---|
157 | String query_string = (String) params.get(QUERY_PARAM);
|
---|
158 |
|
---|
159 | // Get the index
|
---|
160 | String index = (String) params.get(INDEX_PARAM);
|
---|
161 | if (index == null || index.equals(""))
|
---|
162 | {
|
---|
163 | index = this.default_index; // assume the default
|
---|
164 | }
|
---|
165 |
|
---|
166 | try
|
---|
167 | {
|
---|
168 | String index_dir = GSFile.collectionIndexDir(this.site_home, this.cluster_name);
|
---|
169 | index_dir += File.separator + index;
|
---|
170 | Directory index_dir_dir = FSDirectory.open(new File(index_dir));
|
---|
171 | IndexSearcher searcher = new IndexSearcher(index_dir_dir);
|
---|
172 |
|
---|
173 | Term term = new Term("content", query_string);
|
---|
174 |
|
---|
175 | Query query = new TermQuery(term);
|
---|
176 |
|
---|
177 | TopDocs hits = searcher.search(query, Integer.MAX_VALUE);
|
---|
178 |
|
---|
179 | GSXML.addMetadata(this.doc, metadata_list, "numDocsMatched", "" + hits.scoreDocs.length);
|
---|
180 |
|
---|
181 | IndexReader reader = searcher.getIndexReader();
|
---|
182 |
|
---|
183 | for (int i = 0; i < hits.scoreDocs.length; i++)
|
---|
184 | {
|
---|
185 | int lucene_doc_num = hits.scoreDocs[i].doc;
|
---|
186 | org.apache.lucene.document.Document luc_doc = reader.document(lucene_doc_num);
|
---|
187 | String node_id = luc_doc.get("nodeID");
|
---|
188 | Element node = this.doc.createElement(GSXML.DOC_NODE_ELEM);
|
---|
189 | node.setAttribute(GSXML.NODE_ID_ATT, node_id);
|
---|
190 | doc_node_list.appendChild(node);
|
---|
191 | }
|
---|
192 | }
|
---|
193 | catch (Exception e)
|
---|
194 | {
|
---|
195 | e.printStackTrace();
|
---|
196 | }
|
---|
197 |
|
---|
198 | return result;
|
---|
199 | }
|
---|
200 | }
|
---|