source: main/trunk/greenstone3/src/java/org/greenstone/gsdl3/service/IViaSearch.java@ 26198

Last change on this file since 26198 was 25727, checked in by kjdon, 12 years ago

getting rid of my email address

  • 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.io.Serializable;
18import java.net.Authenticator;
19
20import org.apache.log4j.*;
21
22/**
23 *
24 * @author Katherine Don
25 * @author <a href="mailto:[email protected]">Chi-Yu Huang</a>
26 */
27
28public class IViaSearch
29 extends AbstractTextSearch {
30
31
32 static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.IViaSearch.class.getName());
33
34 // have standard gs param names for hits per page, and start page
35 // these need to be mapped to iVia params
36 protected static final String IM_HITS_PARAM = "no_of_records_per_page";
37 protected static final String IM_START_PAGE_PARAM = "start_page_no";
38
39 protected String ivia_server_url = null;
40 protected ArrayList<String> index_ids = null;
41 public IViaSearch()
42 {
43 }
44
45 //Configure IViaSearch Service
46 public boolean configure(Element info, Element extra_info)
47 {
48 if (!super.configure(info, extra_info)){
49 return false;
50 }
51
52 Element server_elem = (Element)GSXML.getChildByTagName(info, "iViaServer");
53 if (server_elem == null) {
54 logger.error("no iViaServer element found");
55 return false;
56 }
57 ivia_server_url = server_elem.getAttribute("url");
58 if (ivia_server_url.equals("")) {
59 logger.error("no url for the iViaServer element");
60 return false;
61 }
62 does_paging = true;
63 does_multi_index_search = true;
64 this.default_index = ",kw,au,su,ti,de,fu,"; // all of them
65 index_ids = new ArrayList<String>();
66 index_ids.add("kw");
67 index_ids.add("au");
68 index_ids.add("su");
69 index_ids.add("ti");
70 index_ids.add("de");
71 index_ids.add("fu");
72
73 return true;
74 }
75
76 /** Process a text query - implemented by concrete subclasses */
77 protected Element processTextQuery(Element request) {
78 // Create a new (empty) result message
79 Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
80 result.setAttribute(GSXML.FROM_ATT, QUERY_SERVICE);
81 result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
82 Element doc_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM+GSXML.LIST_MODIFIER);
83 result.appendChild(doc_node_list);
84 Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM+GSXML.LIST_MODIFIER);
85 if (param_list == null) {
86 logger.error("TextQuery request had no paramList.");
87 return result; // Return the empty result
88 }
89
90 // Process the request parameters
91 HashMap<String, Serializable> params = GSXML.extractParams(param_list, false);
92
93 // Make sure a query has been specified
94 String query = (String) params.get(QUERY_PARAM);
95 if (query == null || query.equals("")) {
96 return result; // Return the empty result
97 }
98
99 // tidy whitespace
100 query = query.replaceAll("\\s+", "+");
101 String url_string = ivia_server_url+"/cgi-bin/canned_search?theme=gsdl3&query="+query;
102
103 // check for fields
104 String fields = (String) params.get(INDEX_PARAM);
105 fields = checkFieldParam(fields); // removes invalid fields
106 if (!fields.equals("")) {
107 url_string += "&fields="+fields;
108 }
109 //check for hits per page
110 String hits_per_page = (String) params.get(HITS_PER_PAGE_PARAM);
111 if (hits_per_page != null && !hits_per_page.equals("")) {
112 url_string += "&"+IM_HITS_PARAM+"="+hits_per_page;
113 }
114
115 // check for start page
116 String start_page = (String) params.get(START_PAGE_PARAM);
117 if (start_page != null && !start_page.equals("")) {
118 url_string += "&"+IM_START_PAGE_PARAM+"="+start_page;
119 }
120 String results_num = null;
121 String doc_ids = null;
122 BufferedReader reader = null;
123 try {
124 logger.debug("sending "+url_string);
125 reader = Misc.makeHttpConnection(url_string);
126 results_num = reader.readLine();
127 doc_ids = reader.readLine();
128 } catch (java.net.MalformedURLException e) {
129 GSXML.addError(this.doc, result, "Malformed URL: "+url_string);
130 return result;
131 } catch (java.io.IOException e) {
132 GSXML.addError(this.doc, result, "IOException during connection to "+url_string+": "+e.toString());
133 return result;
134 }
135
136 if (results_num.startsWith("Resources: ") && doc_ids.startsWith("Ids: ")) {
137 results_num = results_num.substring(11);
138 doc_ids = doc_ids.substring(5).trim();
139
140 } else {
141 logger.error("badly formatted results:");
142 StringBuffer result_string = new StringBuffer();
143 result_string.append("Error: badly formatted result from IVia server:\n ");
144 result_string.append(results_num);
145 result_string.append(doc_ids);
146 String line;
147 try {
148 while((line = reader.readLine()) != null) {
149 result_string.append(line);
150 }
151 } catch (Exception e) {
152 result_string.append("Exception: "+e);
153 }
154 GSXML.addError(this.doc, result, result_string.toString());
155
156 return result;
157 }
158
159 // get the num docs and add to a metadata list
160 Element metadata_list = this.doc.createElement(GSXML.METADATA_ELEM+GSXML.LIST_MODIFIER);
161 result.appendChild(metadata_list);
162
163 // Add a metadata element specifying the number of matching documents
164 long numdocs = Long.parseLong(results_num);
165 GSXML.addMetadata(this.doc, metadata_list, "numDocsMatched", ""+numdocs);
166 String [] ids = doc_ids.split(" ");
167
168 for (int d=0; d<ids.length; d++) {
169 Element doc_node = this.doc.createElement(GSXML.DOC_NODE_ELEM);
170 doc_node.setAttribute(GSXML.NODE_ID_ATT, ids[d]);
171 doc_node_list.appendChild(doc_node);
172 }
173 return result;
174 }
175
176 protected String checkFieldParam(String fields) {
177
178 if (fields == null) {
179 // return the default
180 return "";
181 }
182 StringBuffer new_fields = new StringBuffer();
183 String [] ids = fields.split(",");
184 for (int i=0; i<ids.length; i++) {
185 if(index_ids.contains(ids[i])) {
186 new_fields.append(ids[i]);
187 new_fields.append(",");
188 }
189 }
190 if (new_fields.length() == 0) {
191 return "";
192 }
193 return new_fields.toString();
194 }
195 /**
196 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
197 */
198 protected void getIndexData(ArrayList<String> index_ids, ArrayList<String> index_names,String lang){
199 index_ids.addAll(this.index_ids);
200 index_names.add(getTextString("param."+INDEX_PARAM+".kw", lang));
201 index_names.add(getTextString("param."+INDEX_PARAM+".au", lang));
202 index_names.add(getTextString("param."+INDEX_PARAM+".su", lang));
203 index_names.add(getTextString("param."+INDEX_PARAM+".ti", lang));
204 index_names.add(getTextString("param."+INDEX_PARAM+".de", lang));
205 index_names.add(getTextString("param."+INDEX_PARAM+".fu", lang));
206 }
207
208}
Note: See TracBrowser for help on using the repository browser.