source: main/trunk/greenstone3/src/java/org/greenstone/gsdl3/service/IViaSearch.java@ 22085

Last change on this file since 22085 was 22085, checked in by sjm84, 14 years ago

Created a util package from classes that could be useful outside of their original packages

  • Property svn:keywords set to Author Date Id Revision
File size: 6.6 KB
Line 
1package org.greenstone.gsdl3.service;
2
3// Greenstone classes
4import org.greenstone.util.Misc;
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
12//Java classes
13import java.util.ArrayList;
14import java.util.HashMap;
15import java.io.File;
16import java.io.BufferedReader;
17import java.net.Authenticator;
18
19import org.apache.log4j.*;
20
21/**
22 *
23 * @author <a href="mailto:[email protected]">Katherine Don</a>
24 * @author <a href="mailto:[email protected]">Chi-Yu Huang</a>
25 */
26
27public class IViaSearch
28 extends AbstractSearch {
29
30
31 static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.IViaSearch.class.getName());
32
33 // have standard gs param names for hits per page, and start page
34 // these need to be mapped to iVia params
35 protected static final String IM_HITS_PARAM = "no_of_records_per_page";
36 protected static final String IM_START_PAGE_PARAM = "start_page_no";
37
38 protected String ivia_server_url = null;
39 protected ArrayList index_ids = null;
40 public IViaSearch()
41 {
42 }
43
44 //Configure IViaSearch Service
45 public boolean configure(Element info, Element extra_info)
46 {
47 if (!super.configure(info, extra_info)){
48 return false;
49 }
50
51 Element server_elem = (Element)GSXML.getChildByTagName(info, "iViaServer");
52 if (server_elem == null) {
53 logger.error("no iViaServer element found");
54 return false;
55 }
56 ivia_server_url = server_elem.getAttribute("url");
57 if (ivia_server_url.equals("")) {
58 logger.error("no url for the iViaServer element");
59 return false;
60 }
61 does_paging = true;
62 does_multi_index_search = true;
63 this.default_index = ",kw,au,su,ti,de,fu,"; // all of them
64 index_ids = new ArrayList();
65 index_ids.add("kw");
66 index_ids.add("au");
67 index_ids.add("su");
68 index_ids.add("ti");
69 index_ids.add("de");
70 index_ids.add("fu");
71
72 return true;
73 }
74
75 /** Process a text query - implemented by concrete subclasses */
76 protected Element processTextQuery(Element request) {
77 // Create a new (empty) result message
78 Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
79 result.setAttribute(GSXML.FROM_ATT, TEXT_QUERY_SERVICE);
80 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
81 Element doc_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
82 result.appendChild(doc_node_list);
83 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
84 if (param_list == null) {
85 logger.error("TextQuery request had no paramList.");
86 return result; // Return the empty result
87 }
88
89 // Process the request parameters
90 HashMap params = GSXML.extractParams(param_list, false);
91
92 // Make sure a query has been specified
93 String query = (String) params.get(QUERY_PARAM);
94 if (query == null || query.equals("")) {
95 return result; // Return the empty result
96 }
97
98 // tidy whitespace
99 query = query.replaceAll("\\s+", "+");
100 String url_string = ivia_server_url+"/cgi-bin/canned_search?theme=gsdl3&query="+query;
101
102 // check for fields
103 String fields = (String) params.get(INDEX_PARAM);
104 fields = checkFieldParam(fields); // removes invalid fields
105 if (!fields.equals("")) {
106 url_string += "&fields="+fields;
107 }
108 //check for hits per page
109 String hits_per_page = (String) params.get(HITS_PER_PAGE_PARAM);
110 if (hits_per_page != null && !hits_per_page.equals("")) {
111 url_string += "&"+IM_HITS_PARAM+"="+hits_per_page;
112 }
113
114 // check for start page
115 String start_page = (String) params.get(START_PAGE_PARAM);
116 if (start_page != null && !start_page.equals("")) {
117 url_string += "&"+IM_START_PAGE_PARAM+"="+start_page;
118 }
119 String results_num = null;
120 String doc_ids = null;
121 BufferedReader reader = null;
122 try {
123 logger.debug("sending "+url_string);
124 reader = Misc.makeHttpConnection(url_string);
125 results_num = reader.readLine();
126 doc_ids = reader.readLine();
127 } catch (java.net.MalformedURLException e) {
128 GSXML.addError(this.doc, result, "Malformed URL: "+url_string);
129 return result;
130 } catch (java.io.IOException e) {
131 GSXML.addError(this.doc, result, "IOException during connection to "+url_string+": "+e.toString());
132 return result;
133 }
134
135 if (results_num.startsWith("Resources: ") && doc_ids.startsWith("Ids: ")) {
136 results_num = results_num.substring(11);
137 doc_ids = doc_ids.substring(5).trim();
138
139 } else {
140 logger.error("badly formatted results:");
141 StringBuffer result_string = new StringBuffer();
142 result_string.append("Error: badly formatted result from IVia server:\n ");
143 result_string.append(results_num);
144 result_string.append(doc_ids);
145 String line;
146 try {
147 while((line = reader.readLine()) != null) {
148 result_string.append(line);
149 }
150 } catch (Exception e) {
151 result_string.append("Exception: "+e);
152 }
153 GSXML.addError(this.doc, result, result_string.toString());
154
155 return result;
156 }
157
158 // get the num docs and add to a metadata list
159 Element metadata_list = this.doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
160 result.appendChild(metadata_list);
161
162 // Add a metadata element specifying the number of matching documents
163 long numdocs = Long.parseLong(results_num);
164 GSXML.addMetadata(this.doc, metadata_list, "numDocsMatched", ""+numdocs);
165 String [] ids = doc_ids.split(" ");
166
167 for (int d=0; d<ids.length; d++) {
168 Element doc_node = this.doc.createElement(GSXML.DOC_NODE_ELEM);
169 doc_node.setAttribute(GSXML.NODE_ID_ATT, ids[d]);
170 doc_node_list.appendChild(doc_node);
171 }
172 return result;
173 }
174
175 protected String checkFieldParam(String fields) {
176
177 if (fields == null) {
178 // return the default
179 return "";
180 }
181 StringBuffer new_fields = new StringBuffer();
182 String [] ids = fields.split(",");
183 for (int i=0; i<ids.length; i++) {
184 if(index_ids.contains(ids[i])) {
185 new_fields.append(ids[i]);
186 new_fields.append(",");
187 }
188 }
189 if (new_fields.length() == 0) {
190 return "";
191 }
192 return new_fields.toString();
193 }
194 /**
195 An IVia server has a fixed list of fields to search (I think) so we can hard code them here rather than reading them in from a config file
196 */
197 protected void getIndexData(ArrayList index_ids, ArrayList index_names,String lang){
198 index_ids.addAll(this.index_ids);
199 index_names.add(getTextString("param."+INDEX_PARAM+".kw", lang));
200 index_names.add(getTextString("param."+INDEX_PARAM+".au", lang));
201 index_names.add(getTextString("param."+INDEX_PARAM+".su", lang));
202 index_names.add(getTextString("param."+INDEX_PARAM+".ti", lang));
203 index_names.add(getTextString("param."+INDEX_PARAM+".de", lang));
204 index_names.add(getTextString("param."+INDEX_PARAM+".fu", lang));
205 }
206
207}
Note: See TracBrowser for help on using the repository browser.