1 | package org.greenstone.gsdl3.service;
|
---|
2 |
|
---|
3 | // Greenstone classes
|
---|
4 | import java.io.BufferedReader;
|
---|
5 | import java.io.InputStream;
|
---|
6 | import java.io.InputStreamReader;
|
---|
7 | import java.io.Serializable;
|
---|
8 | import java.net.HttpURLConnection;
|
---|
9 | import java.net.URL;
|
---|
10 | import java.util.HashMap;
|
---|
11 |
|
---|
12 | import org.apache.log4j.Logger;
|
---|
13 | import org.greenstone.gsdl3.util.GSPath;
|
---|
14 | import org.greenstone.gsdl3.util.GSXML;
|
---|
15 | import org.w3c.dom.Document;
|
---|
16 | import org.w3c.dom.Element;
|
---|
17 | import org.w3c.dom.NodeList;
|
---|
18 |
|
---|
19 | /**
|
---|
20 | *
|
---|
21 | * @author Katherine Don
|
---|
22 | * @version $Revision: 25978 $
|
---|
23 | */
|
---|
24 |
|
---|
25 | public class IViaProxy extends ServiceRack
|
---|
26 | {
|
---|
27 |
|
---|
28 | static Logger logger = Logger.getLogger(org.greenstone.gsdl3.service.IViaProxy.class.getName());
|
---|
29 |
|
---|
30 | // the services on offer
|
---|
31 | // these strings must match what is found in the properties file
|
---|
32 | protected static final String TEXT_QUERY_SERVICE = "TextQuery";
|
---|
33 | protected static final String DOC_CONTENT_SERVICE = "DocumentContentRetrieve";
|
---|
34 | protected static final String DOC_META_SERVICE = "DocumentMetadataRetrieve";
|
---|
35 | protected static final String QUERY_PARAM = "query";
|
---|
36 | protected static final String FIELD_PARAM = "fields";
|
---|
37 | // have standard gs param names for hits per page, and start page
|
---|
38 | // these need to be mapped to iVia params
|
---|
39 | protected static final String GS_HITS_PARAM = "hitsPerPage";
|
---|
40 | protected static final String IM_HITS_PARAM = "no_of_records_per_page";
|
---|
41 | protected static final String GS_START_PAGE_PARAM = "startPage";
|
---|
42 | protected static final String IM_START_PAGE_PARAM = "start_page_no";
|
---|
43 |
|
---|
44 | protected String ivia_server_url = null;
|
---|
45 |
|
---|
46 | public boolean configure(Element info, Element extra_info)
|
---|
47 | {
|
---|
48 |
|
---|
49 | if (!super.configure(info, extra_info))
|
---|
50 | {
|
---|
51 | return false;
|
---|
52 | }
|
---|
53 |
|
---|
54 | Element server_elem = (Element) GSXML.getChildByTagName(info, "iViaServer");
|
---|
55 | if (server_elem == null)
|
---|
56 | {
|
---|
57 | logger.error("no iViaServer element found");
|
---|
58 | return false;
|
---|
59 | }
|
---|
60 | ivia_server_url = server_elem.getAttribute("url");
|
---|
61 | if (ivia_server_url.equals(""))
|
---|
62 | {
|
---|
63 | logger.error("no url for the iViaServer element");
|
---|
64 | return false;
|
---|
65 | }
|
---|
66 | Element tq_service = this.doc.createElement(GSXML.SERVICE_ELEM);
|
---|
67 | tq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
|
---|
68 | tq_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
|
---|
69 | this.short_service_info.appendChild(tq_service);
|
---|
70 |
|
---|
71 | Element dc_service = this.doc.createElement(GSXML.SERVICE_ELEM);
|
---|
72 | dc_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_RETRIEVE);
|
---|
73 | dc_service.setAttribute(GSXML.NAME_ATT, DOC_CONTENT_SERVICE);
|
---|
74 | this.short_service_info.appendChild(dc_service);
|
---|
75 |
|
---|
76 | Element dm_service = this.doc.createElement(GSXML.SERVICE_ELEM);
|
---|
77 | dm_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_RETRIEVE);
|
---|
78 | dm_service.setAttribute(GSXML.NAME_ATT, DOC_META_SERVICE);
|
---|
79 | this.short_service_info.appendChild(dm_service);
|
---|
80 |
|
---|
81 | //
|
---|
82 | // add some format info to service map if there is any
|
---|
83 | String path = GSPath.appendLink(GSXML.SEARCH_ELEM, GSXML.FORMAT_ELEM);
|
---|
84 | Element format = (Element) GSXML.getNodeByPath(extra_info, path);
|
---|
85 | if (format != null)
|
---|
86 | {
|
---|
87 | this.format_info_map.put(TEXT_QUERY_SERVICE, this.doc.importNode(format, true));
|
---|
88 | }
|
---|
89 |
|
---|
90 | // look for document display format
|
---|
91 | path = GSPath.appendLink(GSXML.DISPLAY_ELEM, GSXML.FORMAT_ELEM);
|
---|
92 | Element display_format = (Element) GSXML.getNodeByPath(extra_info, path);
|
---|
93 | if (display_format != null)
|
---|
94 | {
|
---|
95 | this.format_info_map.put(DOC_CONTENT_SERVICE, this.doc.importNode(display_format, true));
|
---|
96 | // shoudl we make a copy?
|
---|
97 | }
|
---|
98 |
|
---|
99 | return true;
|
---|
100 |
|
---|
101 | }
|
---|
102 |
|
---|
103 | protected Element getServiceDescription(String service, String lang, String subset)
|
---|
104 | {
|
---|
105 |
|
---|
106 | if (service.equals(TEXT_QUERY_SERVICE))
|
---|
107 | {
|
---|
108 | Element tq_service = this.doc.createElement(GSXML.SERVICE_ELEM);
|
---|
109 | tq_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_QUERY);
|
---|
110 | tq_service.setAttribute(GSXML.NAME_ATT, TEXT_QUERY_SERVICE);
|
---|
111 | if (subset == null || subset.equals(GSXML.DISPLAY_TEXT_ELEM + GSXML.LIST_MODIFIER))
|
---|
112 | {
|
---|
113 | tq_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_NAME, getTextString(TEXT_QUERY_SERVICE + ".name", lang)));
|
---|
114 | tq_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_SUBMIT, getTextString(TEXT_QUERY_SERVICE + ".submit", lang)));
|
---|
115 | tq_service.appendChild(GSXML.createDisplayTextElement(this.doc, GSXML.DISPLAY_TEXT_DESCRIPTION, getTextString(TEXT_QUERY_SERVICE + ".description", lang)));
|
---|
116 | }
|
---|
117 | if (subset == null || subset.equals(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER))
|
---|
118 | {
|
---|
119 | Element param_list = this.doc.createElement(GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
|
---|
120 | tq_service.appendChild(param_list);
|
---|
121 | Element param = GSXML.createParameterDescription(this.doc, QUERY_PARAM, getTextString("param." + QUERY_PARAM, lang), GSXML.PARAM_TYPE_STRING, null, null, null);
|
---|
122 | param_list.appendChild(param);
|
---|
123 | String[] field_ids = { "kw", "au", "su", "ti", "de", "fu" };
|
---|
124 | String[] field_names = { getTextString("param." + FIELD_PARAM + ".kw", lang), getTextString("param." + FIELD_PARAM + ".au", lang), getTextString("param." + FIELD_PARAM + ".su", lang), getTextString("param." + FIELD_PARAM + ".ti", lang), getTextString("param." + FIELD_PARAM + ".de", lang), getTextString("param." + FIELD_PARAM + ".fu", lang) };
|
---|
125 |
|
---|
126 | param = GSXML.createParameterDescription(this.doc, FIELD_PARAM, getTextString("param." + FIELD_PARAM, lang), GSXML.PARAM_TYPE_ENUM_MULTI, "kw,au,su,ti,de,fu", field_ids, field_names);
|
---|
127 | param_list.appendChild(param);
|
---|
128 |
|
---|
129 | String[] hits_options = { "10", "30", "50" };
|
---|
130 | param = GSXML.createParameterDescription(this.doc, GS_HITS_PARAM, getTextString("param." + GS_HITS_PARAM, lang), GSXML.PARAM_TYPE_ENUM_SINGLE, "10", hits_options, hits_options);
|
---|
131 | param_list.appendChild(param);
|
---|
132 |
|
---|
133 | param = GSXML.createParameterDescription(this.doc, GS_START_PAGE_PARAM, "", GSXML.PARAM_TYPE_INVISIBLE, "1", null, null);
|
---|
134 | param_list.appendChild(param);
|
---|
135 | }
|
---|
136 | return tq_service;
|
---|
137 | }
|
---|
138 | if (service.equals(DOC_META_SERVICE))
|
---|
139 | {
|
---|
140 | Element dm_service = this.doc.createElement(GSXML.SERVICE_ELEM);
|
---|
141 | dm_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_RETRIEVE);
|
---|
142 | dm_service.setAttribute(GSXML.NAME_ATT, DOC_META_SERVICE);
|
---|
143 | return dm_service;
|
---|
144 |
|
---|
145 | }
|
---|
146 | if (service.equals(DOC_CONTENT_SERVICE))
|
---|
147 | {
|
---|
148 | Element dc_service = this.doc.createElement(GSXML.SERVICE_ELEM);
|
---|
149 | dc_service.setAttribute(GSXML.TYPE_ATT, GSXML.SERVICE_TYPE_RETRIEVE);
|
---|
150 | dc_service.setAttribute(GSXML.NAME_ATT, DOC_CONTENT_SERVICE);
|
---|
151 | return dc_service;
|
---|
152 |
|
---|
153 | }
|
---|
154 | return null;
|
---|
155 | }
|
---|
156 |
|
---|
157 | /** Process a text query - implemented by concrete subclasses */
|
---|
158 | protected Element processTextQuery(Element request)
|
---|
159 | {
|
---|
160 |
|
---|
161 | // Create a new (empty) result message
|
---|
162 | Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
|
---|
163 | result.setAttribute(GSXML.FROM_ATT, TEXT_QUERY_SERVICE);
|
---|
164 | result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
|
---|
165 | Element doc_node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
|
---|
166 | result.appendChild(doc_node_list);
|
---|
167 |
|
---|
168 | Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
|
---|
169 | if (param_list == null)
|
---|
170 | {
|
---|
171 | logger.error("TextQuery request had no paramList.");
|
---|
172 | return result; // Return the empty result
|
---|
173 | }
|
---|
174 |
|
---|
175 | // Process the request parameters
|
---|
176 | HashMap<String, Serializable> params = GSXML.extractParams(param_list, false);
|
---|
177 |
|
---|
178 | // Make sure a query has been specified
|
---|
179 | String query = (String) params.get(QUERY_PARAM);
|
---|
180 | if (query == null || query.equals(""))
|
---|
181 | {
|
---|
182 | return result; // Return the empty result
|
---|
183 | }
|
---|
184 | // tidy whitespace
|
---|
185 | query = query.replaceAll("\\s+", "+");
|
---|
186 | String url_string = ivia_server_url + "/cgi-bin/canned_search?theme=gsdl3&query=" + query;
|
---|
187 |
|
---|
188 | // check for fields
|
---|
189 | String fields = (String) params.get(FIELD_PARAM);
|
---|
190 | if (fields != null && !fields.equals(""))
|
---|
191 | {
|
---|
192 | url_string += "&fields=" + fields;
|
---|
193 | }
|
---|
194 |
|
---|
195 | //check for hits per page
|
---|
196 | String hits_per_page = (String) params.get(GS_HITS_PARAM);
|
---|
197 | if (hits_per_page != null && !hits_per_page.equals(""))
|
---|
198 | {
|
---|
199 | url_string += "&" + IM_HITS_PARAM + "=" + hits_per_page;
|
---|
200 | }
|
---|
201 |
|
---|
202 | // check for start page
|
---|
203 | String start_page = (String) params.get(GS_START_PAGE_PARAM);
|
---|
204 | if (start_page != null && !start_page.equals(""))
|
---|
205 | {
|
---|
206 | url_string += "&" + IM_START_PAGE_PARAM + "=" + start_page;
|
---|
207 | }
|
---|
208 | String results_num = null;
|
---|
209 | String doc_ids = null;
|
---|
210 | try
|
---|
211 | {
|
---|
212 | logger.debug("IViaProxy, sending " + url_string);
|
---|
213 | BufferedReader reader = makeConnection(url_string);
|
---|
214 | results_num = reader.readLine();
|
---|
215 | doc_ids = reader.readLine();
|
---|
216 |
|
---|
217 | }
|
---|
218 | catch (Exception e)
|
---|
219 | {
|
---|
220 | logger.error("exception happened during query");
|
---|
221 | e.printStackTrace();
|
---|
222 | return result;
|
---|
223 | }
|
---|
224 |
|
---|
225 | if (results_num.startsWith("Resources: "))
|
---|
226 | {
|
---|
227 | results_num = results_num.substring(11);
|
---|
228 | }
|
---|
229 | else
|
---|
230 | {
|
---|
231 | logger.error("badly formatted results line: " + results_num);
|
---|
232 | return result;
|
---|
233 | }
|
---|
234 | if (doc_ids.startsWith("Ids: "))
|
---|
235 | {
|
---|
236 | doc_ids = doc_ids.substring(5).trim();
|
---|
237 | }
|
---|
238 | else
|
---|
239 | {
|
---|
240 | logger.error("badly formatted docs line: " + doc_ids);
|
---|
241 | return result;
|
---|
242 | }
|
---|
243 |
|
---|
244 | // get the num docs and add to a metadata list
|
---|
245 | Element metadata_list = this.doc.createElement(GSXML.METADATA_ELEM + GSXML.LIST_MODIFIER);
|
---|
246 | result.appendChild(metadata_list);
|
---|
247 |
|
---|
248 | // Add a metadata element specifying the number of matching documents
|
---|
249 | long numdocs = Long.parseLong(results_num);
|
---|
250 | GSXML.addMetadata(this.doc, metadata_list, "numDocsMatched", "" + numdocs);
|
---|
251 | String[] ids = doc_ids.split(" ");
|
---|
252 |
|
---|
253 | for (int d = 0; d < ids.length; d++)
|
---|
254 | {
|
---|
255 | Element doc_node = this.doc.createElement(GSXML.DOC_NODE_ELEM);
|
---|
256 | doc_node.setAttribute(GSXML.NODE_ID_ATT, ids[d]);
|
---|
257 | doc_node_list.appendChild(doc_node);
|
---|
258 | }
|
---|
259 | logger.debug("IViaProxy result:");
|
---|
260 | logger.debug(this.converter.getString(result));
|
---|
261 | return result;
|
---|
262 |
|
---|
263 | }
|
---|
264 |
|
---|
265 | protected Element processDocumentMetadataRetrieve(Element request)
|
---|
266 | {
|
---|
267 | Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
|
---|
268 | result.setAttribute(GSXML.FROM_ATT, DOC_META_SERVICE);
|
---|
269 | result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
|
---|
270 |
|
---|
271 | // Get the parameters of the request
|
---|
272 | Element param_list = (Element) GSXML.getChildByTagName(request, GSXML.PARAM_ELEM + GSXML.LIST_MODIFIER);
|
---|
273 | if (param_list == null)
|
---|
274 | {
|
---|
275 | logger.error("missing paramList.\n");
|
---|
276 | return result; // Return the empty result
|
---|
277 | }
|
---|
278 |
|
---|
279 | // The metadata information required
|
---|
280 | StringBuffer field_list = new StringBuffer();
|
---|
281 | Element param = GSXML.getFirstElementChild(param_list);//(Element) param_list.getFirstChild();
|
---|
282 | while (param != null)
|
---|
283 | {
|
---|
284 | // Identify the metadata information desired
|
---|
285 | if (param.getAttribute(GSXML.NAME_ATT).equals("metadata"))
|
---|
286 | {
|
---|
287 | String metadata = GSXML.getValue(param);
|
---|
288 | if (isAcceptableMetadata(metadata))
|
---|
289 | {
|
---|
290 | field_list.append(metadata);
|
---|
291 | field_list.append(",");
|
---|
292 | }
|
---|
293 | }
|
---|
294 | param = (Element) param.getNextSibling();
|
---|
295 | }
|
---|
296 |
|
---|
297 | if (field_list.length() == 0)
|
---|
298 | {
|
---|
299 | logger.error("no metadata specified.\n");
|
---|
300 | return result;
|
---|
301 | }
|
---|
302 |
|
---|
303 | // Get the documents
|
---|
304 | Element request_node_list = (Element) GSXML.getChildByTagName(request, GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
|
---|
305 | if (request_node_list == null)
|
---|
306 | {
|
---|
307 | logger.error("DocumentMetadataRetrieve request had no " + GSXML.DOC_NODE_ELEM + "List.\n");
|
---|
308 | return result;
|
---|
309 | }
|
---|
310 |
|
---|
311 | StringBuffer record_id_list = new StringBuffer();
|
---|
312 |
|
---|
313 | NodeList request_nodes = request_node_list.getChildNodes();
|
---|
314 | for (int i = 0; i < request_nodes.getLength(); i++)
|
---|
315 | {
|
---|
316 | Element request_node = (Element) request_nodes.item(i);
|
---|
317 | String node_id = request_node.getAttribute(GSXML.NODE_ID_ATT);
|
---|
318 | record_id_list.append(node_id);
|
---|
319 | record_id_list.append(",");
|
---|
320 | }
|
---|
321 |
|
---|
322 | // do the query to the iVia server
|
---|
323 | String url_string = ivia_server_url + "/cgi-bin/view_record_set?theme=gsdl3&record_id_list=" + record_id_list.toString() + "&field_list=" + field_list.toString();
|
---|
324 |
|
---|
325 | Element node_list = this.doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
|
---|
326 | result.appendChild(node_list);
|
---|
327 | try
|
---|
328 | {
|
---|
329 | BufferedReader reader = makeConnection(url_string);
|
---|
330 | String line;
|
---|
331 | while ((line = reader.readLine()) != null)
|
---|
332 | {
|
---|
333 | if (!line.startsWith("Record:"))
|
---|
334 | {
|
---|
335 | continue;
|
---|
336 | }
|
---|
337 | // the first line is the record
|
---|
338 | line = line.substring(8);
|
---|
339 | Element doc_node = this.doc.createElement(GSXML.DOC_NODE_ELEM);
|
---|
340 | doc_node.setAttribute(GSXML.NODE_ID_ATT, line);
|
---|
341 | Element meta_list = this.doc.createElement(GSXML.METADATA_ELEM + GSXML.LIST_MODIFIER);
|
---|
342 | doc_node.appendChild(meta_list);
|
---|
343 | while ((line = reader.readLine()) != null)
|
---|
344 | {
|
---|
345 | //metadata entry
|
---|
346 | int col_pos = line.indexOf(':');
|
---|
347 | if (col_pos == -1)
|
---|
348 | {
|
---|
349 | // end of the metadata for this doc
|
---|
350 | break;
|
---|
351 | }
|
---|
352 | String name = line.substring(0, col_pos);
|
---|
353 | String value = line.substring(col_pos + 2); // includes a space
|
---|
354 | GSXML.addMetadata(this.doc, meta_list, name, value);
|
---|
355 | }
|
---|
356 | node_list.appendChild(doc_node);
|
---|
357 |
|
---|
358 | }
|
---|
359 | }
|
---|
360 | catch (Exception e)
|
---|
361 | {
|
---|
362 | logger.error("exception happened");
|
---|
363 | e.printStackTrace();
|
---|
364 | }
|
---|
365 | logger.debug("IViaProxy: returning result: ");
|
---|
366 | logger.debug(this.converter.getPrettyString(result));
|
---|
367 | return result;
|
---|
368 |
|
---|
369 | }
|
---|
370 |
|
---|
371 | protected Element processDocumentContentRetrieve(Element request)
|
---|
372 | {
|
---|
373 | Element result = this.doc.createElement(GSXML.RESPONSE_ELEM);
|
---|
374 | result.setAttribute(GSXML.FROM_ATT, DOC_CONTENT_SERVICE);
|
---|
375 | result.setAttribute(GSXML.TYPE_ATT, GSXML.REQUEST_TYPE_PROCESS);
|
---|
376 |
|
---|
377 | // Get the request doc_list
|
---|
378 | Element query_doc_list = (Element) GSXML.getChildByTagName(request, GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
|
---|
379 | if (query_doc_list == null)
|
---|
380 | {
|
---|
381 | logger.error("DocumentContentRetrieve request specified no doc nodes.\n");
|
---|
382 | return result;
|
---|
383 | }
|
---|
384 |
|
---|
385 | Element doc_list = this.doc.createElement(GSXML.DOC_NODE_ELEM + GSXML.LIST_MODIFIER);
|
---|
386 | result.appendChild(doc_list);
|
---|
387 |
|
---|
388 | // Get the documents
|
---|
389 | String[] doc_ids = GSXML.getAttributeValuesFromList(query_doc_list, GSXML.NODE_ID_ATT);
|
---|
390 | for (int i = 0; i < doc_ids.length; i++)
|
---|
391 | {
|
---|
392 | String doc_id = doc_ids[i];
|
---|
393 | Element doc_node = getDocument(doc_id);
|
---|
394 | doc_list.appendChild(doc_node);
|
---|
395 | }
|
---|
396 | return result;
|
---|
397 |
|
---|
398 | }
|
---|
399 |
|
---|
400 | /**
|
---|
401 | * gets a document by sending a request to iVia, then processes it and
|
---|
402 | * creates a documentNode around the text
|
---|
403 | */
|
---|
404 | protected Element getDocument(String doc_id)
|
---|
405 | {
|
---|
406 |
|
---|
407 | String url_string = ivia_server_url + "/cgi-bin/view_record?theme=gsdl3&record_id=" + doc_id;
|
---|
408 | StringBuffer buffer = new StringBuffer();
|
---|
409 | try
|
---|
410 | {
|
---|
411 | BufferedReader reader = makeConnection(url_string);
|
---|
412 |
|
---|
413 | String line;
|
---|
414 | while ((line = reader.readLine()) != null)
|
---|
415 | {
|
---|
416 | buffer.append(line);
|
---|
417 | }
|
---|
418 |
|
---|
419 | }
|
---|
420 | catch (Exception e)
|
---|
421 | {
|
---|
422 | logger.error("exception happened");
|
---|
423 | e.printStackTrace();
|
---|
424 | }
|
---|
425 |
|
---|
426 | String node_content = buffer.toString();
|
---|
427 | String escaped_content = GSXML.xmlSafe(node_content);
|
---|
428 |
|
---|
429 | StringBuffer processed_content = new StringBuffer(escaped_content.length());
|
---|
430 | processed_content.append("<nodeContent>");
|
---|
431 | int pos = 0;
|
---|
432 | int lastpos = 0;
|
---|
433 | while ((pos = escaped_content.indexOf("<a ", lastpos)) != -1)
|
---|
434 | {
|
---|
435 | processed_content.append(escaped_content.substring(lastpos, pos));
|
---|
436 | int endpos = escaped_content.indexOf("</a>", pos);
|
---|
437 | if (endpos == -1)
|
---|
438 | {
|
---|
439 | break;
|
---|
440 | }
|
---|
441 | String link = escaped_content.substring(pos, endpos + 10);
|
---|
442 | link = convertLink(link);
|
---|
443 | processed_content.append(link);
|
---|
444 | lastpos = endpos + 10;
|
---|
445 | }
|
---|
446 | processed_content.append(escaped_content.substring(lastpos)); // get the last bit
|
---|
447 | processed_content.append("</nodeContent>");
|
---|
448 |
|
---|
449 | Element doc_node = this.doc.createElement(GSXML.DOC_NODE_ELEM);
|
---|
450 | doc_node.setAttribute(GSXML.NODE_ID_ATT, doc_id);
|
---|
451 |
|
---|
452 | Document content_doc = this.converter.getDOM(processed_content.toString());
|
---|
453 | if (content_doc != null)
|
---|
454 | {
|
---|
455 | Element content_element = content_doc.getDocumentElement();
|
---|
456 | doc_node.appendChild(this.doc.importNode(content_element, true));
|
---|
457 | }
|
---|
458 | else
|
---|
459 | {
|
---|
460 | logger.error("Couldn't parse the node content");
|
---|
461 | }
|
---|
462 | return doc_node;
|
---|
463 |
|
---|
464 | }
|
---|
465 |
|
---|
466 | /** converts a url from an <a> element into a greenstone suitable one */
|
---|
467 | protected String convertLink(String aref)
|
---|
468 | {
|
---|
469 |
|
---|
470 | if (aref.indexOf("href="http") != -1)
|
---|
471 | {
|
---|
472 | return aref; // an external link
|
---|
473 | }
|
---|
474 | String type = "other";
|
---|
475 | if (aref.indexOf("/cgi-bin/canned_search") != -1)
|
---|
476 | {
|
---|
477 | type = "query";
|
---|
478 | }
|
---|
479 | else if (aref.indexOf("/cgi-bin/click_through") != -1)
|
---|
480 | {
|
---|
481 | type = "external";
|
---|
482 | }
|
---|
483 | else if (aref.indexOf("/cgi-bin/view_record") != -1)
|
---|
484 | {
|
---|
485 | type = "document";
|
---|
486 | }
|
---|
487 |
|
---|
488 | int href_start = aref.indexOf("href="") + 11;
|
---|
489 | int href_end = aref.indexOf(">", href_start);
|
---|
490 | String href = aref.substring(href_start, href_end);
|
---|
491 | String link_content = aref.substring(href_end + 4, aref.length() - 10);
|
---|
492 |
|
---|
493 | if (type.equals("external"))
|
---|
494 | {
|
---|
495 | // the external link is everything after the http at the end.
|
---|
496 | String address = href.substring(href.lastIndexOf("http"));
|
---|
497 | address = address.replaceAll("%3[aA]", ":");
|
---|
498 | address = address.replaceAll("%2[fF]", "/");
|
---|
499 |
|
---|
500 | return "<a href=\"" + address + "\">" + link_content + "</a>";
|
---|
501 | }
|
---|
502 | if (type.equals("other"))
|
---|
503 | {
|
---|
504 | return "other type of link (" + link_content + ")";
|
---|
505 | }
|
---|
506 | StringBuffer result = new StringBuffer();
|
---|
507 | result.append("<link type='");
|
---|
508 | result.append(type);
|
---|
509 | result.append("'");
|
---|
510 | if (type.equals("query"))
|
---|
511 | {
|
---|
512 | result.append(" service='TextQuery'");
|
---|
513 | }
|
---|
514 | result.append(">");
|
---|
515 | // add in the parameters
|
---|
516 | href = href.substring(href.indexOf("?") + 1);
|
---|
517 | String[] params = href.split("&");
|
---|
518 | for (int i = 0; i < params.length; i++)
|
---|
519 | {
|
---|
520 | String param = params[i];
|
---|
521 | int eq_pos = param.indexOf("=");
|
---|
522 | if (eq_pos != -1)
|
---|
523 | {
|
---|
524 |
|
---|
525 | result.append("<param name='" + param.substring(0, eq_pos) + "' value='" + param.substring(eq_pos + 1) + "'/>");
|
---|
526 | }
|
---|
527 | }
|
---|
528 | result.append(link_content);
|
---|
529 | result.append("</link>");
|
---|
530 |
|
---|
531 | return result.toString();
|
---|
532 |
|
---|
533 | }
|
---|
534 |
|
---|
535 | // iVia craps out if we ask for a metadata which is not valid. So need
|
---|
536 | // to make sure we only ask for acceptable fields.
|
---|
537 | protected boolean isAcceptableMetadata(String meta)
|
---|
538 | {
|
---|
539 | String valid_metadata = ",title,url,ivia_description,keywords,subjects,";
|
---|
540 | if (valid_metadata.indexOf("," + meta + ",") != -1)
|
---|
541 | {
|
---|
542 | return true;
|
---|
543 | }
|
---|
544 | return false;
|
---|
545 | }
|
---|
546 |
|
---|
547 | protected BufferedReader makeConnection(String url_string)
|
---|
548 | {
|
---|
549 | BufferedReader reader = null;
|
---|
550 | try
|
---|
551 | {
|
---|
552 | URL url = new URL(url_string);
|
---|
553 | HttpURLConnection connection = (HttpURLConnection) url.openConnection();
|
---|
554 | InputStream input = connection.getInputStream();
|
---|
555 | reader = new BufferedReader(new InputStreamReader(input));
|
---|
556 | }
|
---|
557 | catch (java.net.MalformedURLException e)
|
---|
558 | {
|
---|
559 |
|
---|
560 | logger.error("Malformed URL: " + url_string);
|
---|
561 | }
|
---|
562 | catch (java.io.IOException e)
|
---|
563 | {
|
---|
564 | logger.error("An error occurred during IO to url " + url_string);
|
---|
565 | }
|
---|
566 | return reader;
|
---|
567 | }
|
---|
568 |
|
---|
569 | }
|
---|