1 | /** |
---|
2 | * This software is copyright (c) 2013-2016 by |
---|
3 | * - Institut fuer Deutsche Sprache (http://www.ids-mannheim.de) |
---|
4 | * This is free software. You can redistribute it |
---|
5 | * and/or modify it under the terms described in |
---|
6 | * the GNU General Public License v3 of which you |
---|
7 | * should have received a copy. Otherwise you can download |
---|
8 | * it from |
---|
9 | * |
---|
10 | * http://www.gnu.org/licenses/gpl-3.0.txt |
---|
11 | * |
---|
12 | * @copyright Institut fuer Deutsche Sprache (http://www.ids-mannheim.de) |
---|
13 | * |
---|
14 | * @license http://www.gnu.org/licenses/gpl-3.0.txt |
---|
15 | * GNU General Public License v3 |
---|
16 | */ |
---|
17 | package eu.clarin.sru.server.fcs; |
---|
18 | |
---|
19 | import java.net.URI; |
---|
20 | import java.util.List; |
---|
21 | import java.util.Map; |
---|
22 | |
---|
23 | import javax.servlet.ServletContext; |
---|
24 | import javax.xml.XMLConstants; |
---|
25 | import javax.xml.stream.XMLStreamException; |
---|
26 | import javax.xml.stream.XMLStreamWriter; |
---|
27 | |
---|
28 | import org.slf4j.Logger; |
---|
29 | import org.slf4j.LoggerFactory; |
---|
30 | import org.z3950.zing.cql.CQLNode; |
---|
31 | import org.z3950.zing.cql.CQLTermNode; |
---|
32 | import eu.clarin.sru.server.SRUConfigException; |
---|
33 | import eu.clarin.sru.server.SRUConstants; |
---|
34 | import eu.clarin.sru.server.SRUDiagnosticList; |
---|
35 | import eu.clarin.sru.server.SRUException; |
---|
36 | import eu.clarin.sru.server.SRUExplainResult; |
---|
37 | import eu.clarin.sru.server.SRUQueryParserRegistry; |
---|
38 | import eu.clarin.sru.server.SRURequest; |
---|
39 | import eu.clarin.sru.server.SRUScanResultSet; |
---|
40 | import eu.clarin.sru.server.SRUSearchEngine; |
---|
41 | import eu.clarin.sru.server.SRUServer; |
---|
42 | import eu.clarin.sru.server.SRUServerConfig; |
---|
43 | import eu.clarin.sru.server.utils.SRUSearchEngineBase; |
---|
44 | |
---|
45 | |
---|
46 | /** |
---|
47 | * A base class for implementing a simple search engine to be used as a |
---|
48 | * CLARIN-FCS endpoint. |
---|
49 | * |
---|
50 | */ |
---|
51 | public abstract class SimpleEndpointSearchEngineBase extends |
---|
52 | SRUSearchEngineBase { |
---|
53 | private static final String X_FCS_ENDPOINT_DESCRIPTION = |
---|
54 | "x-fcs-endpoint-description"; |
---|
55 | private static final String ED_NS = |
---|
56 | "http://clarin.eu/fcs/endpoint-description"; |
---|
57 | private static final String ED_PREFIX = "ed"; |
---|
58 | private static final int ED_VERSION = 2; |
---|
59 | private static final Logger logger = |
---|
60 | LoggerFactory.getLogger(SimpleEndpointSearchEngineBase.class); |
---|
61 | protected EndpointDescription endpointDescription; |
---|
62 | |
---|
63 | |
---|
64 | /** |
---|
65 | * This method should not be overridden. Perform your custom initialization |
---|
66 | * in the {@link #doInit(ServletContext, SRUServerConfig, Map)} method |
---|
67 | * Instead. |
---|
68 | * |
---|
69 | * @see #doInit(ServletContext, SRUServerConfig, Map) |
---|
70 | */ |
---|
71 | @Override |
---|
72 | public final void init(ServletContext context, |
---|
73 | SRUServerConfig config, |
---|
74 | SRUQueryParserRegistry.Builder parserReqistryBuilder, |
---|
75 | Map<String, String> params) throws SRUConfigException { |
---|
76 | logger.debug("initializing"); |
---|
77 | super.init(context, config, parserReqistryBuilder, params); |
---|
78 | |
---|
79 | parserReqistryBuilder.register(new FCSQueryParser()); |
---|
80 | |
---|
81 | logger.debug("initializing search engine implementation"); |
---|
82 | doInit(context, config, parserReqistryBuilder, params); |
---|
83 | |
---|
84 | logger.debug("initizalizing endpoint description"); |
---|
85 | this.endpointDescription = |
---|
86 | createEndpointDescription(context, config, params); |
---|
87 | if (this.endpointDescription == null) { |
---|
88 | logger.error("SimpleEndpointSearchEngineBase implementation " + |
---|
89 | "error: createEndpointDescription() returned null"); |
---|
90 | throw new SRUConfigException("createEndpointDescription() " + |
---|
91 | "returned no valid implementation of an EndpointDescription"); |
---|
92 | } |
---|
93 | } |
---|
94 | |
---|
95 | |
---|
96 | /** |
---|
97 | * This method should not be overridden. Perform you custom cleanup in the |
---|
98 | * {@link #doDestroy()} method. |
---|
99 | * |
---|
100 | * @see #doDestroy() |
---|
101 | */ |
---|
102 | @Override |
---|
103 | public final void destroy() { |
---|
104 | logger.debug("performing cleanup of endpoint description"); |
---|
105 | endpointDescription.destroy(); |
---|
106 | logger.debug("performing cleanup of search engine"); |
---|
107 | doDestroy(); |
---|
108 | super.destroy(); |
---|
109 | } |
---|
110 | |
---|
111 | |
---|
112 | @Override |
---|
113 | public final SRUExplainResult explain(SRUServerConfig config, |
---|
114 | SRURequest request, SRUDiagnosticList diagnostics) |
---|
115 | throws SRUException { |
---|
116 | |
---|
117 | final boolean provideEndpointDescription = |
---|
118 | parseBoolean(request.getExtraRequestData( |
---|
119 | X_FCS_ENDPOINT_DESCRIPTION)); |
---|
120 | |
---|
121 | if (provideEndpointDescription) { |
---|
122 | return new SRUExplainResult(diagnostics) { |
---|
123 | @Override |
---|
124 | public boolean hasExtraResponseData() { |
---|
125 | return provideEndpointDescription; |
---|
126 | } |
---|
127 | |
---|
128 | |
---|
129 | @Override |
---|
130 | public void writeExtraResponseData(XMLStreamWriter writer) |
---|
131 | throws XMLStreamException { |
---|
132 | writeEndpointDescription(writer); |
---|
133 | } |
---|
134 | }; |
---|
135 | } else { |
---|
136 | return null; |
---|
137 | } |
---|
138 | } |
---|
139 | |
---|
140 | |
---|
141 | /** |
---|
142 | * Handle a <em>scan</em> operation. This implementation provides support to |
---|
143 | * CLARIN FCS resource enumeration. If you want to provide custom scan |
---|
144 | * behavior for a different index, override the |
---|
145 | * {@link #doScan(SRUServerConfig, SRURequest, SRUDiagnosticList)} method. |
---|
146 | * |
---|
147 | * @see #doScan(SRUServerConfig, SRURequest, SRUDiagnosticList) |
---|
148 | */ |
---|
149 | @Override |
---|
150 | public final SRUScanResultSet scan(SRUServerConfig config, |
---|
151 | SRURequest request, SRUDiagnosticList diagnostics) |
---|
152 | throws SRUException { |
---|
153 | return doScan(config, request, diagnostics); |
---|
154 | } |
---|
155 | |
---|
156 | |
---|
157 | protected abstract EndpointDescription createEndpointDescription( |
---|
158 | ServletContext context, SRUServerConfig config, |
---|
159 | Map<String, String> params) throws SRUConfigException; |
---|
160 | |
---|
161 | |
---|
162 | /** |
---|
163 | * Initialize the search engine. This initialization should be tailed |
---|
164 | * towards your environment and needs. |
---|
165 | * |
---|
166 | * @param context |
---|
167 | * the {@link ServletContext} for the Servlet |
---|
168 | * @param config |
---|
169 | * the {@link SRUServerConfig} object for this search engine |
---|
170 | * @param parsersRegistryBuilder |
---|
171 | * the {@link SRUQueryParserRegistry.Builder} object to be used |
---|
172 | * for this search engine. Use to register additional query |
---|
173 | * parsers with the {@link SRUServer}. |
---|
174 | * @param params |
---|
175 | * additional parameters gathered from the Servlet configuration |
---|
176 | * and Servlet context. |
---|
177 | * @throws SRUConfigException |
---|
178 | * if an error occurred |
---|
179 | */ |
---|
180 | protected abstract void doInit(ServletContext context, |
---|
181 | SRUServerConfig config, |
---|
182 | SRUQueryParserRegistry.Builder queryParsersBuilder, |
---|
183 | Map<String, String> params) throws SRUConfigException; |
---|
184 | |
---|
185 | |
---|
186 | /** |
---|
187 | * Destroy the search engine. Override this method for any cleanup the |
---|
188 | * search engine needs to perform upon termination. |
---|
189 | */ |
---|
190 | protected void doDestroy() { |
---|
191 | } |
---|
192 | |
---|
193 | |
---|
194 | /** |
---|
195 | * Handle a <em>scan</em> operation. The default implementation is a no-op. |
---|
196 | * Override this method, if you want to provide a custom behavior. |
---|
197 | * |
---|
198 | * @param config |
---|
199 | * the <code>SRUEndpointConfig</code> object that contains the |
---|
200 | * endpoint configuration |
---|
201 | * @param request |
---|
202 | * the <code>SRURequest</code> object that contains the request |
---|
203 | * made to the endpoint |
---|
204 | * @param diagnostics |
---|
205 | * the <code>SRUDiagnosticList</code> object for storing |
---|
206 | * non-fatal diagnostics |
---|
207 | * @return a <code>SRUScanResultSet</code> object or <code>null</code> if |
---|
208 | * this operation is not supported by this search engine |
---|
209 | * @throws SRUException |
---|
210 | * if an fatal error occurred |
---|
211 | * @see SRUSearchEngine#scan(SRUServerConfig, SRURequest, SRUDiagnosticList) |
---|
212 | * @deprecated override |
---|
213 | * {@link #scan(SRUServerConfig, SRURequest, SRUDiagnosticList)} |
---|
214 | */ |
---|
215 | @Deprecated |
---|
216 | protected SRUScanResultSet doScan(SRUServerConfig config, |
---|
217 | SRURequest request, SRUDiagnosticList diagnostics) |
---|
218 | throws SRUException { |
---|
219 | final CQLNode scanClause = request.getScanClause(); |
---|
220 | if (scanClause instanceof CQLTermNode) { |
---|
221 | final CQLTermNode root = (CQLTermNode) scanClause; |
---|
222 | final String index = root.getIndex(); |
---|
223 | throw new SRUException(SRUConstants.SRU_UNSUPPORTED_INDEX, index, |
---|
224 | "scan operation on index '" + index + "' is not supported"); |
---|
225 | } else { |
---|
226 | throw new SRUException(SRUConstants.SRU_QUERY_FEATURE_UNSUPPORTED, |
---|
227 | "Scan clause too complex."); |
---|
228 | } |
---|
229 | } |
---|
230 | |
---|
231 | |
---|
232 | /** |
---|
233 | * Convince method for parsing a string to boolean. Values <code>1</code>, |
---|
234 | * <code>true</code>, <code>yes</code> yield a <em>true</em> boolean value |
---|
235 | * as a result, all others (including <code>null</code>) a <em>false</em> |
---|
236 | * boolean value. |
---|
237 | * |
---|
238 | * @param value |
---|
239 | * the string to parse |
---|
240 | * @return <code>true</code> if the supplied string was considered something |
---|
241 | * representing a <em>true</em> boolean value, <code>false</code> |
---|
242 | * otherwise |
---|
243 | */ |
---|
244 | protected static boolean parseBoolean(String value) { |
---|
245 | if (value != null) { |
---|
246 | return value.equals("1") || Boolean.parseBoolean(value); |
---|
247 | } |
---|
248 | return false; |
---|
249 | } |
---|
250 | |
---|
251 | |
---|
252 | private void writeEndpointDescription(XMLStreamWriter writer) |
---|
253 | throws XMLStreamException { |
---|
254 | writer.setPrefix(ED_PREFIX, ED_NS); |
---|
255 | writer.writeStartElement(ED_NS, "EndpointDescription"); |
---|
256 | writer.writeNamespace(ED_PREFIX, ED_NS); |
---|
257 | writer.writeAttribute("version", Integer.toString(ED_VERSION)); |
---|
258 | |
---|
259 | // Capabilities |
---|
260 | writer.writeStartElement(ED_NS, "Capabilities"); |
---|
261 | for (URI capability : endpointDescription.getCapabilities()) { |
---|
262 | writer.writeStartElement(ED_NS, "Capability"); |
---|
263 | writer.writeCharacters(capability.toString()); |
---|
264 | writer.writeEndElement(); // "Capability" element |
---|
265 | } |
---|
266 | writer.writeEndElement(); // "Capabilities" element |
---|
267 | |
---|
268 | // SupportedDataViews |
---|
269 | writer.writeStartElement(ED_NS, "SupportedDataViews"); |
---|
270 | for (DataView dataView : endpointDescription.getSupportedDataViews()) { |
---|
271 | writer.writeStartElement(ED_NS, "SupportedDataView"); |
---|
272 | writer.writeAttribute("id", dataView.getIdentifier()); |
---|
273 | String s; |
---|
274 | switch (dataView.getDeliveryPolicy()) { |
---|
275 | case SEND_BY_DEFAULT: |
---|
276 | s = "send-by-default"; |
---|
277 | break; |
---|
278 | case NEED_TO_REQUEST: |
---|
279 | s = "need-to-request"; |
---|
280 | break; |
---|
281 | default: |
---|
282 | throw new XMLStreamException( |
---|
283 | "invalid value for payload delivery policy: " + |
---|
284 | dataView.getDeliveryPolicy()); |
---|
285 | } // switch |
---|
286 | writer.writeAttribute("delivery-policy", s); |
---|
287 | writer.writeCharacters(dataView.getMimeType()); |
---|
288 | writer.writeEndElement(); // "SupportedDataView" element |
---|
289 | } |
---|
290 | writer.writeEndElement(); // "SupportedDataViews" element |
---|
291 | |
---|
292 | // SupportedLayers |
---|
293 | final List<Layer> layers = endpointDescription.getSupportedLayers(); |
---|
294 | if (layers != null) { |
---|
295 | writer.writeStartElement(ED_NS, "SupportedLayers"); |
---|
296 | for (Layer layer : layers) { |
---|
297 | writer.writeStartElement(ED_NS, "SupportedLayer"); |
---|
298 | writer.writeAttribute("id", layer.getId()); |
---|
299 | writer.writeAttribute("result-id", |
---|
300 | layer.getResultId().toString()); |
---|
301 | if (layer.getContentEncoding() == Layer.ContentEncoding.EMPTY) { |
---|
302 | writer.writeAttribute("type", "empty"); |
---|
303 | } |
---|
304 | if (layer.getQualifier() != null) { |
---|
305 | writer.writeAttribute("qualifier", layer.getQualifier()); |
---|
306 | } |
---|
307 | if (layer.getAltValueInfo() != null) { |
---|
308 | writer.writeAttribute("alt-value-info", |
---|
309 | layer.getAltValueInfo()); |
---|
310 | if (layer.getAltValueInfoURI() != null) { |
---|
311 | writer.writeAttribute("alt-value-info-uri", |
---|
312 | layer.getAltValueInfoURI().toString()); |
---|
313 | } |
---|
314 | } |
---|
315 | writer.writeCharacters(layer.getType()); |
---|
316 | writer.writeEndElement(); // "SupportedLayer" element |
---|
317 | } |
---|
318 | writer.writeEndElement(); // "SupportedLayers" element |
---|
319 | |
---|
320 | } |
---|
321 | |
---|
322 | // Resources |
---|
323 | try { |
---|
324 | List<ResourceInfo> resources = |
---|
325 | endpointDescription.getResourceList( |
---|
326 | EndpointDescription.PID_ROOT); |
---|
327 | writeResourceInfos(writer, resources); |
---|
328 | } catch (SRUException e) { |
---|
329 | throw new XMLStreamException("error retriving top-level resources", |
---|
330 | e); |
---|
331 | } |
---|
332 | writer.writeEndElement(); // "EndpointDescription" element |
---|
333 | } |
---|
334 | |
---|
335 | |
---|
336 | private void writeResourceInfos(XMLStreamWriter writer, |
---|
337 | List<ResourceInfo> resources) throws XMLStreamException { |
---|
338 | if (resources == null) { |
---|
339 | throw new NullPointerException("resources == null"); |
---|
340 | } |
---|
341 | if (!resources.isEmpty()) { |
---|
342 | writer.writeStartElement(ED_NS, "Resources"); |
---|
343 | |
---|
344 | for (ResourceInfo resource : resources) { |
---|
345 | writer.writeStartElement(ED_NS, "Resource"); |
---|
346 | writer.writeAttribute("pid", resource.getPid()); |
---|
347 | |
---|
348 | // title |
---|
349 | final Map<String, String> title = resource.getTitle(); |
---|
350 | for (Map.Entry<String, String> i : title.entrySet()) { |
---|
351 | writer.setPrefix(XMLConstants.XML_NS_PREFIX, |
---|
352 | XMLConstants.XML_NS_URI); |
---|
353 | writer.writeStartElement(ED_NS, "Title"); |
---|
354 | writer.writeAttribute(XMLConstants.XML_NS_URI, "lang", i.getKey()); |
---|
355 | writer.writeCharacters(i.getValue()); |
---|
356 | writer.writeEndElement(); // "title" element |
---|
357 | } |
---|
358 | |
---|
359 | // description |
---|
360 | final Map<String, String> description = resource.getDescription(); |
---|
361 | if (description != null) { |
---|
362 | for (Map.Entry<String, String> i : description.entrySet()) { |
---|
363 | writer.writeStartElement(ED_NS, "Description"); |
---|
364 | writer.writeAttribute(XMLConstants.XML_NS_URI, "lang", |
---|
365 | i.getKey()); |
---|
366 | writer.writeCharacters(i.getValue()); |
---|
367 | writer.writeEndElement(); // "Description" element |
---|
368 | } |
---|
369 | } |
---|
370 | |
---|
371 | // landing page |
---|
372 | final String landingPageURI = resource.getLandingPageURI(); |
---|
373 | if (landingPageURI != null) { |
---|
374 | writer.writeStartElement(ED_NS, "LandingPageURI"); |
---|
375 | writer.writeCharacters(landingPageURI); |
---|
376 | writer.writeEndElement(); // "LandingPageURI" element |
---|
377 | } |
---|
378 | |
---|
379 | // languages |
---|
380 | final List<String> languages = resource.getLanguages(); |
---|
381 | writer.writeStartElement(ED_NS, "Languages"); |
---|
382 | for (String i : languages) { |
---|
383 | writer.writeStartElement(ED_NS, "Language"); |
---|
384 | writer.writeCharacters(i); |
---|
385 | writer.writeEndElement(); // "Language" element |
---|
386 | |
---|
387 | } |
---|
388 | writer.writeEndElement(); // "Languages" element |
---|
389 | |
---|
390 | // available data views |
---|
391 | StringBuilder sb = new StringBuilder(); |
---|
392 | for (DataView dataview : resource.getAvailableDataViews()) { |
---|
393 | if (sb.length() > 0) { |
---|
394 | sb.append(" "); |
---|
395 | } |
---|
396 | sb.append(dataview.getIdentifier()); |
---|
397 | } |
---|
398 | writer.writeEmptyElement(ED_NS, "AvailableDataViews"); |
---|
399 | writer.writeAttribute("ref", sb.toString()); |
---|
400 | |
---|
401 | final List<Layer> layers = resource.getAvailableLayers(); |
---|
402 | if (layers != null) { |
---|
403 | sb = new StringBuilder(); |
---|
404 | for (Layer layer : resource.getAvailableLayers()) { |
---|
405 | if (sb.length() > 0) { |
---|
406 | sb.append(" "); |
---|
407 | } |
---|
408 | sb.append(layer.getId()); |
---|
409 | } |
---|
410 | writer.writeEmptyElement(ED_NS, "AvailableLayers"); |
---|
411 | writer.writeAttribute("ref", sb.toString()); |
---|
412 | } |
---|
413 | |
---|
414 | // child resources |
---|
415 | List<ResourceInfo> subs = resource.getSubResources(); |
---|
416 | if ((subs != null) && !subs.isEmpty()) { |
---|
417 | writeResourceInfos(writer, subs); |
---|
418 | } |
---|
419 | |
---|
420 | writer.writeEndElement(); // "Resource" element |
---|
421 | } |
---|
422 | writer.writeEndElement(); // "Resources" element |
---|
423 | } |
---|
424 | } |
---|
425 | |
---|
426 | } // class SimpleEndpointSearchEngineBase |
---|