1 | package org.greenstone.gsdl3_extension.mat.servlet;
|
---|
2 |
|
---|
3 | import java.io.*;
|
---|
4 | import java.net.*;
|
---|
5 | import java.util.*;
|
---|
6 | import java.util.regex.Matcher;
|
---|
7 | import java.util.regex.Pattern;
|
---|
8 |
|
---|
9 | import javax.servlet.*;
|
---|
10 | import javax.servlet.http.*;
|
---|
11 | import javax.xml.parsers.*;
|
---|
12 |
|
---|
13 | import org.xml.sax.*;
|
---|
14 | import org.w3c.dom.*;
|
---|
15 |
|
---|
16 | public class MatServlet extends HttpServlet {
|
---|
17 |
|
---|
18 | private int port_number = 0;
|
---|
19 | private String oaiPrefix ="";
|
---|
20 | private String titleString = "<title>Metadata Analysis Tool - Alpha</title>";
|
---|
21 | private String h1String = "<h2>Metadata Analysis Tool - alpha 2</h2>";
|
---|
22 | private String maxRecord ="10";
|
---|
23 |
|
---|
24 | public void doGet(HttpServletRequest request,HttpServletResponse response)throws ServletException, IOException {
|
---|
25 |
|
---|
26 | PrintWriter out = response.getWriter();
|
---|
27 | String cssString = "<link rel=\"stylesheet\" href=\"http://localhost:"+port_number+"/greenstone3/mat/script/mat.css\" type=\"text/css\" >";
|
---|
28 | String headerString = "<head>" + titleString + "\n" + cssString + "</head>\n";
|
---|
29 |
|
---|
30 | port_number = request.getLocalPort();
|
---|
31 | response.setContentType("text/html");
|
---|
32 | response.setHeader("pragma", "no-cache");
|
---|
33 |
|
---|
34 | out.println("<html>");
|
---|
35 | out.println(headerString);
|
---|
36 | out.println("<body>");
|
---|
37 | out.println("<img src=http://www.cs.waikato.ac.nz/images-cs/uni.gif>");
|
---|
38 | out.println(h1String);
|
---|
39 | out.println("<p>This tool will generate statistics and visualisations of OAI repositories</p>");
|
---|
40 | out.println("<p>Enter the URL of the OAI repository to analyse, e.g.:</p>");
|
---|
41 | out.println( "<p><code>http://www.ideals.uiuc.edu/dspace-oai/request</code></p>");
|
---|
42 | out.println("<div><form method=\"post\" action=\"http://localhost:"+port_number+"/greenstone3/mat\">");
|
---|
43 | out.println("<p>OAI URL: <input type=\"text\" name=\"oaiurl\" size=\"140\"></p>");
|
---|
44 | out.println("<p><input type=\"submit\" value=\"Analyse repository\"></p>");
|
---|
45 | out.println("</form>");
|
---|
46 | out.println("<p>Or use these shortcuts:</p>");
|
---|
47 |
|
---|
48 | out.print("<table><tr>");
|
---|
49 | out.print("<div><form method=\"post\" action=\"http://localhost:"+port_number+"/greenstone3/mat\">");
|
---|
50 | out.print("<input type=\"hidden\" name=\"oaiurl\" value=\"http://www.ideals.uiuc.edu/dspace-oai/request\"> ");
|
---|
51 | out.print("<input type=\"submit\" value=\"IDEALS at U. Illinois\"></form>");
|
---|
52 |
|
---|
53 | out.print("<form method=\"post\" action=\"http://localhost:"+port_number+"/greenstone3/mat\">");
|
---|
54 | out.print("<input type=\"hidden\" name=\"oaiurl\" value=\"http://dspace.lib.cranfield.ac.uk/dspace-oai/request\"> ");
|
---|
55 | out.print( "<input type=\"submit\" value=\"QUEprints (DSpace) at Cranfield U.\"></form></div>");
|
---|
56 |
|
---|
57 | out.print("<form method=\"post\" action=\"http://localhost:"+port_number+"/greenstone3/mat\">" +
|
---|
58 | "<input type=\"hidden\" name=\"oaiurl\" value=\"http://cogprints.org/cgi/oai2\"> " +
|
---|
59 | "<input type=\"submit\" value=\"Cogprints - Cognitive Science Eprint Archive\"></form>");
|
---|
60 |
|
---|
61 | out.print("<form method=\"post\" action=\"http://localhost:"+port_number+"/greenstone3/mat\">" +
|
---|
62 | "<input type=\"hidden\" name=\"oaiurl\" value=\"http://azmemory.lib.az.us/cgi-bin/oai.exe\"> " +
|
---|
63 | "<input type=\"submit\" value=\"Arizona Memory Project\"></form>");
|
---|
64 |
|
---|
65 | out.println("</div>");
|
---|
66 | out.print("</tr></table>");
|
---|
67 |
|
---|
68 | out.println ("<table border=\"0\" align=\"center\">\r\n");
|
---|
69 | out.println ("<colgroup width=\"900\" span=\"3\">\r\n");
|
---|
70 |
|
---|
71 | out.println ("<tr>\r\n");
|
---|
72 | out.println ("<td><b>Sample Reports</b>\r\n");
|
---|
73 | out.println ("<td><b>NZ Reports</b>\r\n");
|
---|
74 | out.println ("<td><b>NZ Reports</b>\r\n");
|
---|
75 | out.println ("</tr>\r\n");
|
---|
76 |
|
---|
77 | out.println ("<tr>\r\n");
|
---|
78 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/gzzraup/Overall.html\">Cogprints,100 records</a>\r\n");
|
---|
79 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/cewmfnn/Overall.html\">The University of Auckland,1960 records</a>\r\n");
|
---|
80 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/pitnyys/Overall.html\">Christchurch Polytechnic Institute of Technology,2 records</a>\r\n");
|
---|
81 | out.println ("</tr>\r\n");
|
---|
82 |
|
---|
83 | out.println ("<tr>\r\n");
|
---|
84 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/pberglz/Overall.html\">IDEALS @ UIUC,500 records</a>\r\n");
|
---|
85 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/pljncki/Overall.html\">Auckland University of Technology,320 records</a>\r\n");
|
---|
86 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/detbosl/Overall.html\">Manukau Institute of Technology,13 records</a>\r\n");
|
---|
87 | out.println ("</tr>\r\n");
|
---|
88 |
|
---|
89 | out.println ("<tr>\r\n");
|
---|
90 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/upqzqen/Overall.html\">NZ research,4600 records</a>\r\n");
|
---|
91 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/ovdikqt/Overall.html\">University of Canterbury,640 records</a>\r\n");
|
---|
92 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/cderfvb/Overall.html\">NorthTec,19 records</a>\r\n");
|
---|
93 | out.println ("</tr>\r\n");
|
---|
94 |
|
---|
95 | out.println ("<tr>\r\n");
|
---|
96 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/nwjrwew/Overall.html\">ResearchBank,6000 records</a>\r\n");
|
---|
97 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/ixrwnqs/Overall.html\">Lincoln University,430 records</a>\r\n");
|
---|
98 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/polkiuj/Overall.html\">Open Polytechnic of New Zealand,14 records</a>\r\n");
|
---|
99 | out.println ("</tr>\r\n");
|
---|
100 |
|
---|
101 | out.println ("<tr>\r\n");
|
---|
102 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/cfxnxcd/Overall.html\">MINDS @ UW,6000 records</a>\r\n");
|
---|
103 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/icsqvwj/Overall.html\">Massey University,290 records</a>\r\n");
|
---|
104 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/qwertyu/Overall.html\">Unitec New Zealand,55 records</a>\r\n");
|
---|
105 | out.println ("</tr>\r\n");
|
---|
106 |
|
---|
107 | out.println ("<tr>\r\n");
|
---|
108 | out.println ("<td> \r\n");
|
---|
109 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/kengnzg/Overall.html\">University of Otago,670 records</a>\r\n");
|
---|
110 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/gbnhyuj/Overall.html\">Universal College of Learning,12 records</a>\r\n");
|
---|
111 | out.println ("</tr>\r\n");
|
---|
112 |
|
---|
113 | out.println ("<tr>\r\n");
|
---|
114 | out.println ("<td> \r\n");
|
---|
115 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/ylxiywr/Overall.html\">Victoria University of Wellington,220 records</a>\r\n");
|
---|
116 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/poiuytr/Overall.html\">Whitireia Community Polytechnic,59 records</a>\r\n");
|
---|
117 | out.println ("</tr>\r\n");
|
---|
118 |
|
---|
119 | out.println ("<tr>\r\n");
|
---|
120 | out.println ("<td> \r\n");
|
---|
121 |
|
---|
122 | out.println ("<td><a href=\"http://www.nzdl.org/greenstone3/mat/nwgfril/Overall.html\">University of Waikato,270 records</a>\r\n");
|
---|
123 | out.println ("</tr>\r\n");
|
---|
124 | out.println ("</tr>\r\n</table>");
|
---|
125 |
|
---|
126 | out.println("<p style=\"font-size:small;\" align=\"right\">powered by <a href=\"http://www.greenstone.org/greenstone3-home\">Greenstone3</a></p>");
|
---|
127 | out.println("<p align=\"left\">Comments to <a href=\"http://www.cs.waikato.ac.nz/~daven/\">Dave Nichols</a></p>");
|
---|
128 | out.println ("</body>");
|
---|
129 | out.println ("</html>");
|
---|
130 | out.close();
|
---|
131 | }
|
---|
132 |
|
---|
133 | protected void doPost(HttpServletRequest req, HttpServletResponse res)throws ServletException, IOException {
|
---|
134 |
|
---|
135 | String cssString = "<link rel=\"stylesheet\" href=\"http://localhost:"+port_number+"/greenstone3/mat/script/mat.css\" type=\"text/css\" >";
|
---|
136 | String headerString = "<head>" + titleString + "\n" + cssString + "</head>\n";
|
---|
137 | String javaScript = "<script type=\"text/javascript\" src=\"http://localhost:"+port_number+"/greenstone3/mat/script/status3.js\"></script>";
|
---|
138 | String headerString2 = "<head>" + titleString + "\n" + javaScript + cssString+"</head>\n";
|
---|
139 |
|
---|
140 | port_number = req.getLocalPort();
|
---|
141 | res.setContentType("text/html");
|
---|
142 | res.setHeader("pragma", "no-cache");
|
---|
143 | PrintWriter out = res.getWriter();
|
---|
144 |
|
---|
145 | out.println("<html>");
|
---|
146 |
|
---|
147 | if ( req.getParameter("metadataPrefix") != null) {
|
---|
148 |
|
---|
149 | oaiPrefix = req.getParameter("metadataPrefix");
|
---|
150 | String maxRecords = req.getParameter("maxrecords");
|
---|
151 |
|
---|
152 | Pattern pa = Pattern.compile("[0-9]{1,5}");
|
---|
153 | Matcher ma = pa.matcher(maxRecords);
|
---|
154 |
|
---|
155 | out.println(headerString2);
|
---|
156 | out.println("<body onLoad=\"autoSubmit()\">");
|
---|
157 | out.println(h1String);
|
---|
158 | out.println("<p>Please wait .... <br> It's downloading OAI records </p>");
|
---|
159 | out.println("<input type=\"button\" onClick=\"showdiv()\" value=\"show debug infomation\">");
|
---|
160 |
|
---|
161 | if(ma.matches()){
|
---|
162 | int num = Integer.parseInt(maxRecords);
|
---|
163 | if(num>0){
|
---|
164 | downloadCollection(out, req, res, oaiPrefix, maxRecords);
|
---|
165 | }
|
---|
166 | else{
|
---|
167 | downloadCollection(out, req, res, oaiPrefix, maxRecord);
|
---|
168 | }
|
---|
169 | }
|
---|
170 | else{
|
---|
171 | downloadCollection(out, req, res, oaiPrefix, maxRecord);
|
---|
172 | }
|
---|
173 |
|
---|
174 | }
|
---|
175 | else if(req.getParameter("matShell") != null){
|
---|
176 | out.println(headerString2);
|
---|
177 | out.println("<body onLoad=\"autoSubmit2()\">");
|
---|
178 | out.println(h1String);
|
---|
179 | out.println("<p>Please wait ...</p><p>It's building collection now.</p>");
|
---|
180 | out.println("<input type=\"button\" onClick=\"showdiv()\" value=\"show debug infomation\">");
|
---|
181 | buildCollection(out,req.getParameter("matShell"),req.getParameter("collectionName"),req.getParameter("collectionURL"),req.getParameter("oaiPrefix"));
|
---|
182 | }
|
---|
183 |
|
---|
184 | else if (req.getParameter("collName") != null){
|
---|
185 | out.println(headerString);
|
---|
186 | out.write("<script type=\"text/javascript\" src=\"http://localhost:"+port_number+"/greenstone3/mat/script/getInfomation.js\"></script>\r\n");
|
---|
187 | out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/yahoo/yahoo-min.js\"></script>\r\n");
|
---|
188 | out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/event/event-min.js\"></script>\r\n");
|
---|
189 | out.write("<script type=\"text/javascript\" src=\"http://yui.yahooapis.com/2.4.1/build/connection/connection-min.js\"></script>\r\n");
|
---|
190 | out.println("<body>");
|
---|
191 | out.println(h1String);
|
---|
192 | analyzeCollection(out, req.getParameter("collName"),req.getParameter("collURL"),req.getParameter("collHost"),req.getParameter("oaiPrefix"));
|
---|
193 | }
|
---|
194 |
|
---|
195 | else {
|
---|
196 | out.println(headerString);
|
---|
197 | out.println("<body>");
|
---|
198 | out.println(h1String);
|
---|
199 |
|
---|
200 | java.net.URL oaiURL;
|
---|
201 | String oaiURLString = req.getParameter("oaiurl");
|
---|
202 | //System.out.println("OAIURL:"+oaiURLString);
|
---|
203 |
|
---|
204 | if (! (oaiURLString.startsWith("http://"))) {
|
---|
205 | oaiURLString = "http://" + oaiURLString; // add on protocol if missing
|
---|
206 | }
|
---|
207 |
|
---|
208 | // URL checks
|
---|
209 | // check if Java can make a URL from the string
|
---|
210 |
|
---|
211 | try {
|
---|
212 | oaiURL = new URL(oaiURLString);
|
---|
213 | }
|
---|
214 | catch (MalformedURLException e) {
|
---|
215 | out.println("<p>Malformed URL Exception caught: " + e.getMessage() + "</p>");
|
---|
216 | out.println("<p>The system cannot recognise the URL you have entered.</p>");
|
---|
217 | return; // go no further
|
---|
218 | }
|
---|
219 |
|
---|
220 | //out.println("<p>host: " + oaiURL.getHost() + "</p>" );
|
---|
221 | //out.println("<p>protocol: " + oaiURL.getProtocol() + "</p>" );
|
---|
222 |
|
---|
223 | // disallow anything with waikato as part of the host
|
---|
224 | // note: this blocks things like waikato.uiuc.edu -
|
---|
225 | // but guess this is highly unlikely to occur
|
---|
226 | // (also turns off the researchcommons)
|
---|
227 | // could use either of these approaches, waikato or waikato.ac.nz
|
---|
228 |
|
---|
229 | if ( oaiURL.getHost().indexOf("researchcommons.waikato.ac.nz") == -1 ) { // not the RC
|
---|
230 | /*
|
---|
231 | if((oaiURL.getHost().indexOf("waikato") != -1 ) ||
|
---|
232 | (oaiURL.getHost().indexOf("waikato.ac.nz") != -1 ) ) {
|
---|
233 | out.println("<p>This service cannot be used to access Waikato URLs</p>");
|
---|
234 | return;
|
---|
235 | }*/
|
---|
236 | }
|
---|
237 |
|
---|
238 | // need to prevent machine names on their own, e.g. smith
|
---|
239 | // being allowed through
|
---|
240 |
|
---|
241 | // approach 1: explicit blacklisting
|
---|
242 |
|
---|
243 | if (oaiURL.getHost().equals("smith") ||
|
---|
244 | oaiURL.getHost().equals("wesson") ) {
|
---|
245 | out.println("<p>This service cannot be used to access these URLs</p>");
|
---|
246 | return;
|
---|
247 | }
|
---|
248 |
|
---|
249 | // approach 2: require at least one . in the host URL
|
---|
250 | if (oaiURL.getHost().indexOf(".") == -1 ) { // i.e. no . in URL host
|
---|
251 | out.println("<p>This service cannot be used to access URLs of this form.</p>");
|
---|
252 | return;
|
---|
253 | }
|
---|
254 |
|
---|
255 | // jones.cs would get through to this point, might this mean something
|
---|
256 | // from the perspective of the host machine?? Does it matter?
|
---|
257 | // will automatic domain completion be applied? by java? by 'the network'?
|
---|
258 | // require 2 dots in the host? does that help at all ?
|
---|
259 | // domain suffix whitelisting is impractical here, as we'd have to
|
---|
260 | // list all countries in the world
|
---|
261 |
|
---|
262 | // do an OAI verb=identify check to make sure it is an OAI server
|
---|
263 | // trim the URL back and then add on verb=Identify
|
---|
264 |
|
---|
265 |
|
---|
266 | //System.setProperty("http.proxyHost", "wwwcache.cs.waikato.ac.nz");
|
---|
267 | //System.setProperty("http.proxyPort","80");
|
---|
268 | //System.setProperty("http.nonProxyHosts", "localhost|*.waikato.ac.nz");
|
---|
269 |
|
---|
270 | Properties systemSettings = System.getProperties();
|
---|
271 | systemSettings.put("http.proxyHost", "wwwcache.cs.waikato.ac.nz");
|
---|
272 | systemSettings.put("http.proxyPort", "80");
|
---|
273 |
|
---|
274 | URL url = oaiURL;
|
---|
275 | String identifyVerb = "";
|
---|
276 |
|
---|
277 | if (! url.getPath().endsWith("?")) {
|
---|
278 | identifyVerb += "?";
|
---|
279 | }
|
---|
280 | identifyVerb += "verb=Identify";
|
---|
281 |
|
---|
282 | URL identifyURL = new URL (url.toString() + identifyVerb);
|
---|
283 | HttpURLConnection connection = (HttpURLConnection)identifyURL.openConnection();
|
---|
284 | //URLConnection connection = identifyURL.openConnection();
|
---|
285 | connection.connect();
|
---|
286 |
|
---|
287 | //BufferedReader inIdentify = new BufferedReader(new InputStreamReader(
|
---|
288 | //connection.getInputStream()));
|
---|
289 |
|
---|
290 | Document identifyDocument;
|
---|
291 |
|
---|
292 | try {
|
---|
293 | DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
|
---|
294 | DocumentBuilder builder = factory.newDocumentBuilder();
|
---|
295 | identifyDocument = builder.parse( connection.getInputStream() );
|
---|
296 |
|
---|
297 | Element oaiElement = identifyDocument.getDocumentElement();
|
---|
298 |
|
---|
299 | if (oaiElement.getTagName() == "OAI-PMH" ) {
|
---|
300 | //out.println("<p>OAI-PMH element found...</p>");
|
---|
301 | }
|
---|
302 | else {
|
---|
303 | out.println("<p>Error: OAI-PMH element not found..exiting</p>");
|
---|
304 | return;
|
---|
305 | }
|
---|
306 |
|
---|
307 | NodeList identifyNodeList = identifyDocument.getElementsByTagName("Identify");
|
---|
308 | Node identifyNode = null;
|
---|
309 |
|
---|
310 | if (identifyNodeList.getLength() == 1 ) {
|
---|
311 | identifyNode = identifyNodeList.item(0);
|
---|
312 | System.out.println("<Identify>\n");
|
---|
313 | }
|
---|
314 | else {
|
---|
315 | out.println("<p>Error: Identify node not found... exiting</p>");
|
---|
316 | return;
|
---|
317 | }
|
---|
318 |
|
---|
319 | NodeList identifyChildList = identifyNode.getChildNodes();
|
---|
320 | out.println("<table rules =\"none\">");
|
---|
321 |
|
---|
322 | for (int i=0; i < identifyChildList.getLength(); i++) {
|
---|
323 | if (identifyChildList.item(i).getNodeName() == "repositoryName" ) {
|
---|
324 | out.println("<tr><td><strong>Repository Name:</strong></td><td> " + identifyChildList.item(i).getTextContent() + "</td></tr>");
|
---|
325 | }
|
---|
326 | if (identifyChildList.item(i).getNodeName() == "baseURL" ) {
|
---|
327 | String baseURL = identifyChildList.item(i).getTextContent();
|
---|
328 |
|
---|
329 | if(baseURL.startsWith("http://")){
|
---|
330 | out.println("<tr><td><strong>Base URL:</strong></td><td><code><a href=\"" + baseURL + "\">" + baseURL + "</a></code></td></tr>");
|
---|
331 | }
|
---|
332 | else{
|
---|
333 | out.println("<tr><td><strong>Base URL:</strong></td><td><code><a href=\"" + oaiURLString + "\">" + oaiURLString + "</a></code></td></tr>");
|
---|
334 | }
|
---|
335 | }
|
---|
336 | }
|
---|
337 | out.println("</table>");
|
---|
338 | }
|
---|
339 | catch (SAXParseException spe) {
|
---|
340 | out.println(spe.getMessage());
|
---|
341 | }
|
---|
342 | catch (SAXException sxe) {
|
---|
343 | out.println(sxe.toString());
|
---|
344 | }
|
---|
345 | catch (ParserConfigurationException pce) {
|
---|
346 | out.println(pce.toString());
|
---|
347 | }
|
---|
348 | catch (IOException ioe) {
|
---|
349 | out.println(ioe.toString());
|
---|
350 | }
|
---|
351 |
|
---|
352 | // now get the metadata prefixes
|
---|
353 |
|
---|
354 | String metadataFormats = "";
|
---|
355 |
|
---|
356 | if (! url.getPath().endsWith("?")) {
|
---|
357 | metadataFormats += "?";
|
---|
358 | }
|
---|
359 | metadataFormats += "verb=ListMetadataFormats";
|
---|
360 |
|
---|
361 | URL metadataFormatsURL = new URL (url.toString() + metadataFormats);
|
---|
362 | URLConnection connection2 = metadataFormatsURL.openConnection();
|
---|
363 | connection.connect();
|
---|
364 | Document document2;
|
---|
365 |
|
---|
366 | try {
|
---|
367 | DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
|
---|
368 | DocumentBuilder builder = factory.newDocumentBuilder();
|
---|
369 | document2 = builder.parse( connection2.getInputStream() );
|
---|
370 |
|
---|
371 | NodeList prefixList = document2.getElementsByTagName("metadataPrefix");
|
---|
372 |
|
---|
373 | // create UI based on response
|
---|
374 |
|
---|
375 | if ( prefixList.getLength() > 0 ) {
|
---|
376 |
|
---|
377 | out.println("<p>Choose one metadata prefix to use:</p>");
|
---|
378 | out.println("<form method=\"post\" action=\"http://localhost:"+port_number+"/greenstone3/mat\">");
|
---|
379 | out.println("<table rules =\"none\">");
|
---|
380 |
|
---|
381 | for (int i=0; i < prefixList.getLength(); i++) {
|
---|
382 | out.println("<tr><td>");
|
---|
383 | String id = "radioID" + i;
|
---|
384 | String prefix = prefixList.item(i).getTextContent();
|
---|
385 |
|
---|
386 | if (prefix.equals("oai_dc")) {
|
---|
387 | out.println("<label for=\"" + id + "\">" + prefix + " (Dublin Core)</label>");
|
---|
388 | //System.out.println("<MetadataElement>Dublin Core<\\MetadataElement>\n");
|
---|
389 | }
|
---|
390 | else {
|
---|
391 | out.println("<label for=\"" + id + "\">" + prefix + "</label>");
|
---|
392 | //System.out.println("<MetadataElement>"+prefix+"<\\MetadataElement>\n");
|
---|
393 | }
|
---|
394 |
|
---|
395 | out.print("</td><td><input type=\"radio\" name=\"metadataPrefix\" value=\"" + prefix + "\" ");
|
---|
396 |
|
---|
397 | if (prefix.equals("oai_dc")) {
|
---|
398 | out.println("checked=\"checked\" ");
|
---|
399 | }
|
---|
400 | out.println(" id=\"" + id + "\"><br>");
|
---|
401 | out.println("</td></tr>");
|
---|
402 | }
|
---|
403 |
|
---|
404 | //System.out.println("</Identify>");
|
---|
405 | out.println("</table>");
|
---|
406 |
|
---|
407 | // need to pass oaiurl through as well (again)
|
---|
408 |
|
---|
409 | out.println("<input type=\"hidden\" name=\"oaiURL\" value=\"" + url.toString() + "\"><br>");
|
---|
410 | out.println("Max records: <input type=\"text\" name=\"maxrecords\" value=\"500\"><br>");
|
---|
411 |
|
---|
412 | // submit button
|
---|
413 |
|
---|
414 | out.println("<p><input type=\"submit\" value=\"Continue\"></p>");
|
---|
415 | out.println("</form>");
|
---|
416 |
|
---|
417 | out.println("<p> Warning: Generating the statistics and visualization will take some time: </p> ");
|
---|
418 | out.println("<table border = \"1\">");
|
---|
419 | out.println("<tr align=\"right\"><td> No.of Records <td> Estimated Time");
|
---|
420 | out.println("<tr align=\"right\"><td>100<td> 5 minutes");
|
---|
421 | out.println("<tr align=\"right\"><td>500 <td> 10 minutes");
|
---|
422 | out.println("<tr align=\"right\"><td>1000 <td> 18 minutes");
|
---|
423 | out.println("<tr align=\"right\"><td>2000<td> 30 minutes");
|
---|
424 | out.println("</table>");
|
---|
425 |
|
---|
426 | out.println("<p> This tool is designed to work with Dublin Core metadata: note that the mapping of qualified Dublin Core to simple Dublin Core (as in <code>oai_dc</code>) may affect the results.");
|
---|
427 | }
|
---|
428 | else {
|
---|
429 | out.println("<p>Error: no metadata prefixes found... exiting</p>");
|
---|
430 | return;
|
---|
431 | }
|
---|
432 | }
|
---|
433 | catch (SAXParseException spe) {
|
---|
434 | out.println(spe.getMessage());
|
---|
435 | }
|
---|
436 | catch (SAXException sxe) {
|
---|
437 | out.println(sxe.toString());
|
---|
438 | }
|
---|
439 | catch (ParserConfigurationException pce) {
|
---|
440 | out.println(pce.toString());
|
---|
441 | }
|
---|
442 | catch (IOException ioe) {
|
---|
443 | out.println(ioe.toString());
|
---|
444 | }
|
---|
445 |
|
---|
446 | // check we get a valid XML document back
|
---|
447 | // check OAI-PMH element
|
---|
448 | // check we have a <repositoryName> and <baseURL> elements
|
---|
449 | // maybe check the baseURL against the query we issued?
|
---|
450 |
|
---|
451 |
|
---|
452 | // if we get here we have a valid non-waikato non-local OAI server
|
---|
453 | // /* process stuff here */
|
---|
454 | // test with University home page - as it doesn't require
|
---|
455 | // dealing with the proxy server
|
---|
456 | // URL url = new URL("http://waikato.ac.nz");
|
---|
457 | // URL url = new URL(oaiURL);
|
---|
458 | // need to go through the proxy here
|
---|
459 | // http://dn.codegear.com/article/29783
|
---|
460 | // http://java.ittoolbox.com/groups/technical-functional/java-l/response-to-proxy-authentication-exception-71438
|
---|
461 | // http://www.jguru.com/faq/view.jsp?EID=13186
|
---|
462 | // http://www.developer.com/java/other/article.php/1551421
|
---|
463 | // System.setProperty("http.proxyHost","http://proxy.scms.waikato.ac.nz");
|
---|
464 | // System.setProperty("http.proxyPort","80");
|
---|
465 | // System.setProperty("http.nonProxyHosts", "localhost|*.waikato.ac.nz");
|
---|
466 | // build a GS 3 collection from the OAI URL
|
---|
467 | // http://www.javaworld.com/javaworld/jw-12-2000/jw-1229-traps.html
|
---|
468 | // http://codon.kribb.re.kr/wiki/display/Java/Using+Runtime.exec+to+invoke+child+process
|
---|
469 |
|
---|
470 | out.print("</body></html>");
|
---|
471 | out.close();
|
---|
472 | } // endif
|
---|
473 | } // end doPost
|
---|
474 |
|
---|
475 | protected void buildCollection( PrintWriter out, String matShell2, String collName, String oaiURLString, String oaiPrefix) throws ServletException, IOException{
|
---|
476 |
|
---|
477 | out.println("<br><br><div id=\"hideshow\" style=\"visibility:hidden\">");
|
---|
478 | out.println("<input type=\"button\" onClick=\"hidediv()\" value=\"hide debug infomation\"/>");
|
---|
479 | out.println("<p>Building collection...</p>");
|
---|
480 |
|
---|
481 | String host = "http://localhost:"+port_number+"/greenstone3/Collection_Analysis/";
|
---|
482 |
|
---|
483 | out.println("<form method=\"post\" name=\"aForm\">");
|
---|
484 | out.println("<input type=\"hidden\" name=\"collName\" value=\"" + collName + "\"><br>");
|
---|
485 | out.println("<input type=\"hidden\" name=\"collURL\" value=\"" + oaiURLString + "\"><br>");
|
---|
486 | out.println("<input type=\"hidden\" name=\"collHost\" value=\"" + host + "\"><br>");
|
---|
487 | out.println("<input type=\"hidden\" name=\"oaiPrefix\" value=\"" + oaiPrefix + "\"><br>");
|
---|
488 | out.println("</form>");
|
---|
489 | out.println("<p><pre>" + matShell2 + "</pre></p>");
|
---|
490 | out.flush();
|
---|
491 |
|
---|
492 | Process p2 = processShell(matShell2, out);
|
---|
493 |
|
---|
494 | if (p2.exitValue() == 0){
|
---|
495 | out.println("<p>Collection built.</p>");
|
---|
496 | }
|
---|
497 | else{
|
---|
498 | out.println("<p><b>Collection not built.</b></p>");
|
---|
499 | }
|
---|
500 |
|
---|
501 | out.print("</div></body></html>");
|
---|
502 | p2 = null;
|
---|
503 | p2.destroy();
|
---|
504 | out.close();
|
---|
505 | }
|
---|
506 |
|
---|
507 | private void downloadCollection( PrintWriter out, HttpServletRequest req, HttpServletResponse res, String oaiPrefix, String Records) throws ServletException, IOException
|
---|
508 | {
|
---|
509 | String oaiURLString = req.getParameter("oaiURL");
|
---|
510 | String metadataprefix = req.getParameter("metadataPrefix");
|
---|
511 | String collName = generateCollName();
|
---|
512 | String host = "http://localhost:"+port_number+"/";
|
---|
513 |
|
---|
514 | out.println("<br><br><div id=\"hideshow\" style=\"visibility:hidden\">");
|
---|
515 | out.println("<input type=\"button\" onClick=\"hidediv()\" value=\"hide debug infomation\"/>");
|
---|
516 | out.println("<p>Downloading OAI documents...</p>");
|
---|
517 | out.flush();
|
---|
518 | // move to correct directory
|
---|
519 |
|
---|
520 | String maxRecords = Records;
|
---|
521 | String cacheDir = "/tmp/oai-mat-5/"+ collName;
|
---|
522 |
|
---|
523 | // downloading
|
---|
524 | String gs3Root = "/research/cc108/greenstone3Project";
|
---|
525 | String collectDir = gs3Root + "/web/sites/localsite/collect";
|
---|
526 | String logFile = "/tmp/mat3.txt";
|
---|
527 |
|
---|
528 | String matShell = "source /research/cc108/greenstone3Project/gs2build/bin/script/mat-colbuild-download.bash "
|
---|
529 | + collName
|
---|
530 | + " "
|
---|
531 | + oaiURLString
|
---|
532 | + " "
|
---|
533 | + cacheDir
|
---|
534 | + " "
|
---|
535 | + maxRecords
|
---|
536 | + " "
|
---|
537 | + metadataprefix
|
---|
538 | + " "
|
---|
539 | + gs3Root
|
---|
540 | + " "
|
---|
541 | + collectDir
|
---|
542 | + " "
|
---|
543 | + logFile;
|
---|
544 |
|
---|
545 | out.println("<p><pre>" + matShell + "</pre></p>");
|
---|
546 | out.flush();
|
---|
547 | Process p = processShell(matShell, out);
|
---|
548 |
|
---|
549 | if (p.exitValue() == 0){
|
---|
550 | out.println("<p>Collection downloaded.</p>");
|
---|
551 | }
|
---|
552 | else {
|
---|
553 | out.println("<p><b>Collection not downloaded properly.</b></p>");
|
---|
554 | }
|
---|
555 | out.flush();
|
---|
556 | //p = null;
|
---|
557 | p.destroy();
|
---|
558 | // return the web page
|
---|
559 | //out.println(req.getRequestURL());
|
---|
560 | // building
|
---|
561 |
|
---|
562 | String matShell2 = "source /research/cc108/greenstone3Project/gs2build/bin/script/mat-colbuild.bash "
|
---|
563 | + collName
|
---|
564 | + " "
|
---|
565 | + oaiURLString
|
---|
566 | + " "
|
---|
567 | + cacheDir
|
---|
568 | + " "
|
---|
569 | + maxRecords
|
---|
570 | + " "
|
---|
571 | + metadataprefix
|
---|
572 | + " "
|
---|
573 | + gs3Root
|
---|
574 | + " "
|
---|
575 | + collectDir
|
---|
576 | + " "
|
---|
577 | + logFile;
|
---|
578 |
|
---|
579 | out.println("<form method=\"post\" name=\"collectionForm\">");
|
---|
580 | out.println("<input type=\"hidden\" name=\"matShell\" value=\"" + matShell2 + "\">");
|
---|
581 | out.println("<input type=\"hidden\" name=\"collectionName\" value=\"" + collName + "\">");
|
---|
582 | out.println("<input type=\"hidden\" name=\"collectionURL\" value=\"" + oaiURLString + "\">");
|
---|
583 | out.println("<input type=\"hidden\" name=\"oaiPrefix\" value=\"" + oaiPrefix + "\">");
|
---|
584 | out.println("<input type=\"hidden\" name=\"host\" value=\"" + host + "\">");
|
---|
585 | //out.println("<input type=\"submit\" value=\"start\"><br>");
|
---|
586 | out.println("</form>");
|
---|
587 | out.println("</div>");
|
---|
588 | out.print("</body></html>");
|
---|
589 | out.flush();
|
---|
590 | out.close();
|
---|
591 |
|
---|
592 | // return page
|
---|
593 | //javascript to submit button in 5 secs
|
---|
594 | //form
|
---|
595 | // hidden inputs
|
---|
596 | // collname
|
---|
597 | // host
|
---|
598 | /////////////////////////////////////////////////////////////////////////////////////////////
|
---|
599 | ////////////////////////////////////////////////////////////////////////////////////////////
|
---|
600 |
|
---|
601 | // specify plugins as parameter to mkcol.pl
|
---|
602 | //String makeColl = "perl -S mkcol.pl -creator [email protected] " + collName;
|
---|
603 | //out.println("<p><pre>" + makeColl + "</pre></p>");
|
---|
604 | // specify the OAI server in a config file???
|
---|
605 | // downloadfrom.pl -download_mode OAI -cache_dir -gli -url oaiURL.toString() -get_doc no
|
---|
606 | // what about that 500 doc limit in OAIDownload.pm
|
---|
607 | /*
|
---|
608 | String downloadFrom = "downloadfrom.pl -download_mode OAI -cache_dir "
|
---|
609 | + cacheDir
|
---|
610 | + " -url "
|
---|
611 | + oaiURL.toString()
|
---|
612 | + " -max_records 10"
|
---|
613 | + " -proxy_on"
|
---|
614 | + " -proxy_host "
|
---|
615 | + proxyHost
|
---|
616 | + " -proxy_port "
|
---|
617 | + proxyPort;
|
---|
618 | */
|
---|
619 |
|
---|
620 | // -proxy_on -proxy_host wwwcache.cs.waikato.ac.nz -proxy_port 80
|
---|
621 | // out.println("<p><pre>" + downloadFrom + "</pre></p>");
|
---|
622 | // /home/daven/research/greenstone3/gs2build/bin/script/build [options] collection-name
|
---|
623 |
|
---|
624 | /*
|
---|
625 | String build = "build -indextype lucene -download file://"
|
---|
626 | + cacheDir
|
---|
627 | + " -log_events " //event log goes to greenstone3/gs2build/etc/events.txt
|
---|
628 | + collName;
|
---|
629 | */
|
---|
630 |
|
---|
631 | // out.println("<p><pre>" + build + "</pre></p>");
|
---|
632 | // perl -S importfrom.pl collname
|
---|
633 | // String importFrom = " perl -S importfrom.pl " + collName;
|
---|
634 | // perl -S buildcol.pl collname
|
---|
635 | // String buildColl = "perl -S buildcol.pl " + collName;
|
---|
636 | // need to be in correct directory
|
---|
637 | // String changeDir = "cd $GSDLHOME/collect/" + collName;
|
---|
638 | // build.pl ??
|
---|
639 | // rebuild ?
|
---|
640 | // rm -r index/*
|
---|
641 | // String removeOld = "rm -r index/*";
|
---|
642 | // mv building/* index/
|
---|
643 | // String moveToIndex = "mv building/* index/";
|
---|
644 |
|
---|
645 | //////////////////////////////////////
|
---|
646 | /*
|
---|
647 | URLConnection connection = url.openConnection();
|
---|
648 | connection.connect();
|
---|
649 | Map headerMap = connection.getHeaderFields(); // gets the HTTP headers
|
---|
650 |
|
---|
651 | out.print("<html><head><title>Analysing...</title></head><body>");
|
---|
652 | out.println ("<img src=http://www.cs.waikato.ac.nz/images-cs/uni.gif>");
|
---|
653 | out.println ("<h1>Metadata Analysis Tool - Alpha</h1>");
|
---|
654 | out.println("<p>Analysing the OAI URL: <code>" + url.toString() + "</code></p>");
|
---|
655 | out.println("<code><pre>");
|
---|
656 |
|
---|
657 | Iterator keyValuePairs = headerMap.entrySet().iterator();
|
---|
658 | out.println("size = " + headerMap.size());
|
---|
659 | for (int i = 0; i < headerMap.size(); i++) {
|
---|
660 | out.println("i = " + i);
|
---|
661 | Map.Entry entry = (Map.Entry) keyValuePairs.next();
|
---|
662 | out.println(entry.getKey());
|
---|
663 | out.println(entry.getValue());
|
---|
664 | out.println();
|
---|
665 | }
|
---|
666 |
|
---|
667 | out.println(headerMap.toString());
|
---|
668 | out.println("</pre></code>");
|
---|
669 | */
|
---|
670 |
|
---|
671 | /*
|
---|
672 | // get the source HTML and insert it into the page - messy
|
---|
673 | out.println("<p>HTML source:</p>");
|
---|
674 | out.println("<pre><code>");
|
---|
675 | BufferedReader in = new BufferedReader(new InputStreamReader(
|
---|
676 | connection.getInputStream()));
|
---|
677 | String inputLine;
|
---|
678 | while ((inputLine = in.readLine()) != null)
|
---|
679 | out.println(inputLine);
|
---|
680 | in.close();
|
---|
681 | out.println("</pre></code>");
|
---|
682 | */
|
---|
683 |
|
---|
684 | //out.print("</body></html>");
|
---|
685 |
|
---|
686 | }
|
---|
687 |
|
---|
688 | /* produce a random 7 letter collection name */
|
---|
689 |
|
---|
690 | private String generateCollName () {
|
---|
691 | Random random = new Random();
|
---|
692 | StringBuffer message = new StringBuffer();
|
---|
693 | int offset = 97; // = "a"
|
---|
694 | message.append( (char) ( random.nextInt( 26 ) + offset ) );
|
---|
695 | message.append( (char) ( random.nextInt( 26 ) + offset ) );
|
---|
696 | message.append( (char) ( random.nextInt( 26 ) + offset ) );
|
---|
697 | message.append( (char) ( random.nextInt( 26 ) + offset ) );
|
---|
698 | message.append( (char) ( random.nextInt( 26 ) + offset ) );
|
---|
699 | message.append( (char) ( random.nextInt( 26 ) + offset ) );
|
---|
700 | message.append( (char) ( random.nextInt( 26 ) + offset ) );
|
---|
701 | return message.toString();
|
---|
702 | }
|
---|
703 |
|
---|
704 | private Process processShell( String command, PrintWriter out) {
|
---|
705 |
|
---|
706 | String s= "";
|
---|
707 | try {
|
---|
708 |
|
---|
709 | String[] args = new String[]{"sh", "-c", command};
|
---|
710 | Process p = Runtime.getRuntime().exec(args);
|
---|
711 | BufferedReader stdInput = new BufferedReader(new InputStreamReader(p.getInputStream()));
|
---|
712 | BufferedReader stdError = new BufferedReader(new InputStreamReader(p.getErrorStream()));
|
---|
713 | StringBuffer stdInputBuffer = new StringBuffer();
|
---|
714 |
|
---|
715 | while ((s = stdInput.readLine()) != null) {
|
---|
716 | stdInputBuffer.append(s+"\n");
|
---|
717 | if(s.contains("-->")){
|
---|
718 | s = s.replace("-->", "");
|
---|
719 | out.println("<!-- " + s + " -->");
|
---|
720 | }
|
---|
721 | else{
|
---|
722 | out.println("<!-- " + s + " -->");
|
---|
723 | }
|
---|
724 | out.flush();
|
---|
725 | }
|
---|
726 |
|
---|
727 | StringBuffer stdErrorBuffer = new StringBuffer();
|
---|
728 |
|
---|
729 | while ((s = stdError.readLine()) != null) {
|
---|
730 | stdErrorBuffer.append(s+"\n");
|
---|
731 |
|
---|
732 | if(s.contains("-->")){
|
---|
733 | s = s.replace("-->", "");
|
---|
734 | out.println("<!-- " + s + " -->");
|
---|
735 | }
|
---|
736 | else{
|
---|
737 | out.println("<!-- " + s + " -->");
|
---|
738 | }
|
---|
739 | out.flush();
|
---|
740 | }
|
---|
741 |
|
---|
742 | out.println("<p>Here is the standard output:</p>\n");
|
---|
743 | out.println("<p><pre>" + stdInputBuffer + "</pre></p>");
|
---|
744 | out.println("<p>Here is the standard error (if any):</p>\n");
|
---|
745 | out.println("<p><pre>" + stdErrorBuffer + "</pre></p>");
|
---|
746 | out.flush();
|
---|
747 | //if (p.exitValue() != 0)
|
---|
748 | if (false){
|
---|
749 | out.println("<p>An error occurred while building the collection.</p>");
|
---|
750 | out.println("<p>Here is the standard output:</p>\n");
|
---|
751 | out.println("<p><pre>" + stdInputBuffer + "</pre></p>");
|
---|
752 | out.println("<p>Here is the standard error (if any):</p>\n");
|
---|
753 | out.println("<p><pre>" + stdErrorBuffer + "</pre></p>");
|
---|
754 | }
|
---|
755 |
|
---|
756 | InputStream is = p.getInputStream();
|
---|
757 | is.close();
|
---|
758 |
|
---|
759 | OutputStream os = p.getOutputStream();
|
---|
760 | os.close();
|
---|
761 |
|
---|
762 | InputStream es = p.getErrorStream();
|
---|
763 | es.close();
|
---|
764 |
|
---|
765 | stdInput.close();
|
---|
766 | stdError.close();
|
---|
767 |
|
---|
768 | return p;
|
---|
769 | }
|
---|
770 | catch (IOException e) {
|
---|
771 | out.println("exception happened - here's what I know: ");
|
---|
772 | out.println(e.toString());
|
---|
773 | out.flush();
|
---|
774 | }
|
---|
775 | return null;
|
---|
776 | }
|
---|
777 |
|
---|
778 | private void analyzeCollection(PrintWriter out, String collectionName,String collectionURL,String collectionHost, String Prefix){
|
---|
779 |
|
---|
780 | String collName = collectionName;
|
---|
781 | String oaiURLString = collectionURL;
|
---|
782 | String host = collectionHost;
|
---|
783 |
|
---|
784 | try{
|
---|
785 | DescribeMessager dm = new DescribeMessager(collName,oaiURLString);
|
---|
786 | out.println("<p>Generating statistics and visualisations...</p>");
|
---|
787 | out.flush();
|
---|
788 | out.println("<p>please wait.</p>");
|
---|
789 | out.flush();
|
---|
790 | boolean status = dm.describeMatadata(out,collName,oaiURLString,Prefix);
|
---|
791 |
|
---|
792 | if(status){
|
---|
793 | out.println("<script type=\"text/javascript\" language=\"JavaScript\">");
|
---|
794 | out.println("reconfig();");
|
---|
795 | out.println("</script>");
|
---|
796 | out.println("<a href=\""+host+collName+"/Overall.html\">View the report</a>");
|
---|
797 | }
|
---|
798 | out.println("</body></html>");
|
---|
799 | }catch(Exception e){e.printStackTrace(out);
|
---|
800 | out.println("<p><pre>"+e.toString()+"</pre></p>");
|
---|
801 | }
|
---|
802 | out.close();
|
---|
803 | }
|
---|
804 | }
|
---|