1 | /** |
---|
2 | * This software is copyright (c) 2012-2014 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.client; |
---|
18 | |
---|
19 | import java.util.ArrayDeque; |
---|
20 | import java.util.ArrayList; |
---|
21 | import java.util.Deque; |
---|
22 | import java.util.LinkedList; |
---|
23 | import java.util.List; |
---|
24 | import java.util.concurrent.TimeUnit; |
---|
25 | |
---|
26 | import javax.xml.XMLConstants; |
---|
27 | import javax.xml.namespace.QName; |
---|
28 | import javax.xml.parsers.DocumentBuilder; |
---|
29 | import javax.xml.parsers.DocumentBuilderFactory; |
---|
30 | import javax.xml.parsers.ParserConfigurationException; |
---|
31 | import javax.xml.stream.XMLStreamConstants; |
---|
32 | import javax.xml.stream.XMLStreamException; |
---|
33 | import javax.xml.stream.XMLStreamReader; |
---|
34 | |
---|
35 | import org.slf4j.Logger; |
---|
36 | import org.slf4j.LoggerFactory; |
---|
37 | import org.w3c.dom.Attr; |
---|
38 | import org.w3c.dom.DOMException; |
---|
39 | import org.w3c.dom.Document; |
---|
40 | import org.w3c.dom.DocumentFragment; |
---|
41 | import org.w3c.dom.Element; |
---|
42 | import org.w3c.dom.Node; |
---|
43 | |
---|
44 | |
---|
45 | /** |
---|
46 | * A client to perform SRU operations. The response of a SRU request is wrapped |
---|
47 | * in a SRU response. |
---|
48 | * <p> |
---|
49 | * This client is reusable but not thread-safe: the application may reuse a |
---|
50 | * client object, but it may not be concurrently shared between multiple |
---|
51 | * threads. |
---|
52 | * </p> |
---|
53 | */ |
---|
54 | public class SRUClient { |
---|
55 | private static final Logger logger = LoggerFactory.getLogger(SRUClient.class); |
---|
56 | private final SRUSimpleClient client; |
---|
57 | private final Handler handler; |
---|
58 | /* common */ |
---|
59 | private List<SRUDiagnostic> diagnostics; |
---|
60 | private List<SRUExtraResponseData> extraResponseData; |
---|
61 | /* scan */ |
---|
62 | private List<SRUTerm> terms; |
---|
63 | /* searchRetrieve */ |
---|
64 | private int numberOfRecords; |
---|
65 | private String resultSetId; |
---|
66 | private int resultSetIdleTime; |
---|
67 | private int nextRecordPosition; |
---|
68 | /* explain/searchRetrieve */ |
---|
69 | private List<SRURecord> records; |
---|
70 | /* statistics */ |
---|
71 | private int totalBytesTransferred; |
---|
72 | private long timeTotal; |
---|
73 | private long timeQueued; |
---|
74 | private long timeNetwork; |
---|
75 | private long timeParsing; |
---|
76 | /* other fields */ |
---|
77 | private final DocumentBuilder documentBuilder; |
---|
78 | private final Deque<Node> stack = new ArrayDeque<Node>(); |
---|
79 | |
---|
80 | |
---|
81 | /** |
---|
82 | * Constructor. |
---|
83 | * |
---|
84 | * @param config |
---|
85 | * the configuration to be used for this client. |
---|
86 | * @throws NullPointerException |
---|
87 | * if argument <code>config</code> is <node>null</code> |
---|
88 | * @throws IllegalArgumentException |
---|
89 | * if an error occurred while registering record data parsers |
---|
90 | * @see SRUClientConfig |
---|
91 | */ |
---|
92 | public SRUClient(final SRUClientConfig config) { |
---|
93 | this(config, DocumentBuilderFactory.newInstance()); |
---|
94 | } |
---|
95 | |
---|
96 | |
---|
97 | /** |
---|
98 | * Constructor. |
---|
99 | * |
---|
100 | * <p> |
---|
101 | * For internal use only. |
---|
102 | * </p> |
---|
103 | * |
---|
104 | * @param defaultVersion |
---|
105 | * the default version to use for SRU requests; may be overridden |
---|
106 | * by individual requests |
---|
107 | * @param parsers |
---|
108 | * a <code>Map</code> to store record schema to record data |
---|
109 | * parser mappings |
---|
110 | * @param documentBuilderFactory |
---|
111 | * the Document Builder factory to be used to create Document |
---|
112 | * Builders |
---|
113 | */ |
---|
114 | SRUClient(final SRUClientConfig config, |
---|
115 | final DocumentBuilderFactory documentBuilderFactory) { |
---|
116 | if (config == null) { |
---|
117 | throw new NullPointerException("config == null"); |
---|
118 | } |
---|
119 | if (documentBuilderFactory == null) { |
---|
120 | throw new NullPointerException("documentBuilderFactory == null"); |
---|
121 | } |
---|
122 | this.client = new SRUSimpleClient(config); |
---|
123 | this.handler = new Handler(config); |
---|
124 | try { |
---|
125 | synchronized (documentBuilderFactory) { |
---|
126 | documentBuilderFactory.setNamespaceAware(true); |
---|
127 | documentBuilderFactory.setCoalescing(true); |
---|
128 | this.documentBuilder = |
---|
129 | documentBuilderFactory.newDocumentBuilder(); |
---|
130 | } // documentBuilderFactory (documentBuilderFactory) |
---|
131 | } catch (ParserConfigurationException e) { |
---|
132 | throw new Error("error initialzing document builder factory", e); |
---|
133 | } |
---|
134 | reset(); |
---|
135 | } |
---|
136 | |
---|
137 | |
---|
138 | /** |
---|
139 | * Perform a <em>explain</em> operation. |
---|
140 | * |
---|
141 | * @param request |
---|
142 | * an instance of a {@link SRUExplainRequest} object |
---|
143 | * @return a {@link SRUExplainResponse} object |
---|
144 | * @throws SRUClientException |
---|
145 | * if an unrecoverable error occurred |
---|
146 | * @throws NullPointerException |
---|
147 | * if any required argument is <code>null</code> |
---|
148 | */ |
---|
149 | public SRUExplainResponse explain(SRUExplainRequest request) |
---|
150 | throws SRUClientException { |
---|
151 | if (request == null) { |
---|
152 | throw new NullPointerException("request == null"); |
---|
153 | } |
---|
154 | try { |
---|
155 | client.explain(request, handler); |
---|
156 | SRURecord record = null; |
---|
157 | if ((records != null) && !records.isEmpty()) { |
---|
158 | record = records.get(0); |
---|
159 | } |
---|
160 | return new SRUExplainResponse(request, |
---|
161 | diagnostics, |
---|
162 | extraResponseData, |
---|
163 | totalBytesTransferred, |
---|
164 | timeTotal, |
---|
165 | timeQueued, |
---|
166 | timeNetwork, |
---|
167 | timeParsing, |
---|
168 | record); |
---|
169 | } finally { |
---|
170 | reset(); |
---|
171 | } |
---|
172 | } |
---|
173 | |
---|
174 | |
---|
175 | /** |
---|
176 | * Perform a <em>scan</em> operation. |
---|
177 | * |
---|
178 | * @param request |
---|
179 | * an instance of a {@link SRUScanRequest} object |
---|
180 | * @return a {@link SRUScanResponse} object |
---|
181 | * @throws SRUClientException |
---|
182 | * if an unrecoverable error occurred |
---|
183 | * @throws NullPointerException |
---|
184 | * if any required argument is <code>null</code> |
---|
185 | */ |
---|
186 | public SRUScanResponse scan(SRUScanRequest request) |
---|
187 | throws SRUClientException { |
---|
188 | if (request == null) { |
---|
189 | throw new NullPointerException("request == null"); |
---|
190 | } |
---|
191 | try { |
---|
192 | client.scan(request, handler); |
---|
193 | return new SRUScanResponse(request, |
---|
194 | diagnostics, |
---|
195 | extraResponseData, |
---|
196 | totalBytesTransferred, |
---|
197 | timeTotal, |
---|
198 | timeQueued, |
---|
199 | timeNetwork, |
---|
200 | timeParsing, |
---|
201 | terms); |
---|
202 | } finally { |
---|
203 | reset(); |
---|
204 | } |
---|
205 | |
---|
206 | } |
---|
207 | |
---|
208 | |
---|
209 | /** |
---|
210 | * Perform a <em>searchRetrieve</em> operation. |
---|
211 | * |
---|
212 | * @param request |
---|
213 | * an instance of a {@link SRUSearchRetrieveRequest} object |
---|
214 | * @return a {@link SRUSearchRetrieveRequest} object |
---|
215 | * @throws SRUClientException |
---|
216 | * if an unrecoverable error occurred |
---|
217 | * @throws NullPointerException |
---|
218 | * if any required argument is <code>null</code> |
---|
219 | */ |
---|
220 | public SRUSearchRetrieveResponse searchRetrieve( |
---|
221 | SRUSearchRetrieveRequest request) throws SRUClientException { |
---|
222 | if (request == null) { |
---|
223 | throw new NullPointerException("request == null"); |
---|
224 | } |
---|
225 | try { |
---|
226 | client.searchRetrieve(request, handler); |
---|
227 | return new SRUSearchRetrieveResponse(request, |
---|
228 | diagnostics, |
---|
229 | extraResponseData, |
---|
230 | totalBytesTransferred, |
---|
231 | timeTotal, |
---|
232 | timeQueued, |
---|
233 | timeNetwork, |
---|
234 | timeParsing, |
---|
235 | numberOfRecords, |
---|
236 | resultSetId, |
---|
237 | resultSetIdleTime, |
---|
238 | records, |
---|
239 | nextRecordPosition); |
---|
240 | } finally { |
---|
241 | reset(); |
---|
242 | } |
---|
243 | } |
---|
244 | |
---|
245 | |
---|
246 | void setTimeQueued(long timeQueued) { |
---|
247 | this.timeQueued = TimeUnit.NANOSECONDS.toMillis(timeQueued); |
---|
248 | } |
---|
249 | |
---|
250 | |
---|
251 | private void addTerm(SRUTerm term) { |
---|
252 | if (terms == null) { |
---|
253 | terms = new LinkedList<SRUTerm>(); |
---|
254 | } |
---|
255 | terms.add(term); |
---|
256 | } |
---|
257 | |
---|
258 | |
---|
259 | private void addRecord(SRURecord record) { |
---|
260 | if (records == null) { |
---|
261 | records = new LinkedList<SRURecord>(); |
---|
262 | } |
---|
263 | records.add(record); |
---|
264 | } |
---|
265 | |
---|
266 | |
---|
267 | private void reset() { |
---|
268 | /* common */ |
---|
269 | diagnostics = null; |
---|
270 | extraResponseData = null; |
---|
271 | /* scan */ |
---|
272 | terms = null; |
---|
273 | /* searchRetrieve */ |
---|
274 | numberOfRecords = -1; |
---|
275 | resultSetId = null; |
---|
276 | resultSetIdleTime = -1; |
---|
277 | nextRecordPosition = -1; |
---|
278 | /* explain/searchRetrieve */ |
---|
279 | records = null; |
---|
280 | /* statistics */ |
---|
281 | totalBytesTransferred = -1; |
---|
282 | timeQueued = -1; |
---|
283 | timeTotal = -1; |
---|
284 | timeNetwork = -1; |
---|
285 | timeParsing = -1; |
---|
286 | } |
---|
287 | |
---|
288 | |
---|
289 | |
---|
290 | private class Handler extends SRUDefaultHandlerAdapter { |
---|
291 | private final List<SRUExtraResponseDataParser> parsers; |
---|
292 | |
---|
293 | |
---|
294 | private Handler(SRUClientConfig config) { |
---|
295 | this.parsers = config.getExtraResponseDataParsers(); |
---|
296 | } |
---|
297 | |
---|
298 | |
---|
299 | @Override |
---|
300 | public void onDiagnostics(List<SRUDiagnostic> diagnostics) |
---|
301 | throws SRUClientException { |
---|
302 | SRUClient.this.diagnostics = diagnostics; |
---|
303 | } |
---|
304 | |
---|
305 | |
---|
306 | @Override |
---|
307 | public void onExtraResponseData(XMLStreamReader reader) |
---|
308 | throws XMLStreamException, SRUClientException { |
---|
309 | if (reader.getEventType() == XMLStreamConstants.START_DOCUMENT) { |
---|
310 | reader.next(); |
---|
311 | } |
---|
312 | |
---|
313 | while (reader.hasNext()) { |
---|
314 | SRUExtraResponseData data = null; |
---|
315 | |
---|
316 | XmlStreamReaderUtils.consumeWhitespace(reader); |
---|
317 | switch (reader.getEventType()) { |
---|
318 | case XMLStreamConstants.START_ELEMENT: |
---|
319 | final QName root = reader.getName(); |
---|
320 | logger.debug("@start: {}", root); |
---|
321 | |
---|
322 | /* |
---|
323 | * The content model of "extraResponseData" is a sequence |
---|
324 | * of elements. Parse each child element into a separate |
---|
325 | * entity, i.e. if a parse is available for handling the |
---|
326 | * element use it, otherwise parse into a document fragment. |
---|
327 | */ |
---|
328 | if ((parsers != null) && !parsers.isEmpty()) { |
---|
329 | for (SRUExtraResponseDataParser parser: parsers) { |
---|
330 | if (parser.supports(root)) { |
---|
331 | logger.debug("parsing extra response data with parser '{}'", |
---|
332 | parser.getClass().getName()); |
---|
333 | data = parser.parse(reader); |
---|
334 | break; |
---|
335 | } |
---|
336 | } |
---|
337 | } |
---|
338 | if (data == null) { |
---|
339 | logger.debug("parsing of extra response data (generic)"); |
---|
340 | data = parseExtraResponseAsDocumentFragment( |
---|
341 | root, documentBuilder, stack, reader); |
---|
342 | } |
---|
343 | break; |
---|
344 | case XMLStreamConstants.END_DOCUMENT: |
---|
345 | break; |
---|
346 | default: |
---|
347 | throw new SRUClientException("expected a start element at " + |
---|
348 | "this location (event code = " + |
---|
349 | reader.getEventType() + ")"); |
---|
350 | } |
---|
351 | |
---|
352 | if (data != null) { |
---|
353 | if (SRUClient.this.extraResponseData == null) { |
---|
354 | SRUClient.this.extraResponseData = |
---|
355 | new ArrayList<SRUExtraResponseData>(); |
---|
356 | } |
---|
357 | SRUClient.this.extraResponseData.add(data); |
---|
358 | } |
---|
359 | } // while |
---|
360 | |
---|
361 | } |
---|
362 | |
---|
363 | |
---|
364 | @Override |
---|
365 | public void onTerm(String value, int numberOfRecords, |
---|
366 | String displayTerm, SRUWhereInList whereInList) |
---|
367 | throws SRUClientException { |
---|
368 | SRUClient.this.addTerm(new SRUTerm(value, numberOfRecords, |
---|
369 | displayTerm, whereInList)); |
---|
370 | } |
---|
371 | |
---|
372 | |
---|
373 | @Override |
---|
374 | public void onExtraTermData(String value, XMLStreamReader reader) |
---|
375 | throws XMLStreamException, SRUClientException { |
---|
376 | final List<SRUTerm> terms = SRUClient.this.terms; |
---|
377 | if ((terms != null) && !terms.isEmpty()) { |
---|
378 | SRUTerm term = terms.get(terms.size() - 1); |
---|
379 | term.setExtraTermData(copyStaxToDocumentFragment( |
---|
380 | documentBuilder, stack, reader)); |
---|
381 | } else { |
---|
382 | /* |
---|
383 | * should never happen ... |
---|
384 | */ |
---|
385 | throw new SRUClientException( |
---|
386 | "internal error; 'terms' is null or empty"); |
---|
387 | } |
---|
388 | } |
---|
389 | |
---|
390 | |
---|
391 | @Override |
---|
392 | public void onStartRecords(int numberOfRecords, String resultSetId, |
---|
393 | int resultSetIdleTime) throws SRUClientException { |
---|
394 | SRUClient.this.resultSetId = resultSetId; |
---|
395 | SRUClient.this.resultSetIdleTime = resultSetIdleTime; |
---|
396 | } |
---|
397 | |
---|
398 | |
---|
399 | @Override |
---|
400 | public void onFinishRecords(int nextRecordPosition) |
---|
401 | throws SRUClientException { |
---|
402 | SRUClient.this.nextRecordPosition = nextRecordPosition; |
---|
403 | } |
---|
404 | |
---|
405 | |
---|
406 | @Override |
---|
407 | public void onRecord(String identifier, int position, |
---|
408 | SRURecordData data) throws SRUClientException { |
---|
409 | SRUClient.this.addRecord( |
---|
410 | new SRURecord(data, identifier, position)); |
---|
411 | } |
---|
412 | |
---|
413 | |
---|
414 | @Override |
---|
415 | public void onSurrogateRecord(String identifier, int position, |
---|
416 | SRUDiagnostic data) throws SRUClientException { |
---|
417 | SRUClient.this.addRecord(new SRURecord( |
---|
418 | new SRUSurrogateRecordData(data), identifier, position)); |
---|
419 | } |
---|
420 | |
---|
421 | |
---|
422 | @Override |
---|
423 | public void onExtraRecordData(String identifier, int position, |
---|
424 | XMLStreamReader reader) throws XMLStreamException, |
---|
425 | SRUClientException { |
---|
426 | final List<SRURecord> records = SRUClient.this.records; |
---|
427 | if ((records != null) && !records.isEmpty()) { |
---|
428 | final SRURecord record = records.get(records.size() - 1); |
---|
429 | record.setExtraRecordData(copyStaxToDocumentFragment( |
---|
430 | documentBuilder, stack, reader)); |
---|
431 | } else { |
---|
432 | /* |
---|
433 | * should never happen ... |
---|
434 | */ |
---|
435 | throw new SRUClientException( |
---|
436 | "internal error; 'records' is null or empty"); |
---|
437 | } |
---|
438 | } |
---|
439 | |
---|
440 | |
---|
441 | @Override |
---|
442 | public void onRequestStatistics(int totalBytesTransferred, |
---|
443 | long millisTotal, long millisNetwork, long millisProcessing) { |
---|
444 | SRUClient.this.totalBytesTransferred = totalBytesTransferred; |
---|
445 | if (SRUClient.this.timeQueued > 0) { |
---|
446 | SRUClient.this.timeTotal = timeQueued + millisTotal; |
---|
447 | } else { |
---|
448 | SRUClient.this.timeTotal = millisTotal; |
---|
449 | } |
---|
450 | SRUClient.this.timeNetwork = millisNetwork; |
---|
451 | SRUClient.this.timeParsing = millisProcessing; |
---|
452 | } |
---|
453 | } // inner class Handler |
---|
454 | |
---|
455 | |
---|
456 | private static SRUExtraResponseData parseExtraResponseAsDocumentFragment( |
---|
457 | QName name, DocumentBuilder builder, Deque<Node> stack, XMLStreamReader reader) |
---|
458 | throws XMLStreamException { |
---|
459 | return new SRUGenericExtraResponseData(name, |
---|
460 | copyStaxToDocumentFragment(builder, stack, reader)); |
---|
461 | } |
---|
462 | |
---|
463 | |
---|
464 | private static DocumentFragment copyStaxToDocumentFragment( |
---|
465 | DocumentBuilder builder, Deque<Node> stack, XMLStreamReader reader) |
---|
466 | throws XMLStreamException { |
---|
467 | try { |
---|
468 | final Document doc = builder.newDocument(); |
---|
469 | stack.push(doc.createDocumentFragment()); |
---|
470 | |
---|
471 | boolean stop = false; |
---|
472 | while (!stop && reader.hasNext()) { |
---|
473 | final Node parent = stack.peek(); |
---|
474 | switch (reader.getEventType()) { |
---|
475 | case XMLStreamConstants.START_ELEMENT: |
---|
476 | stack.push(createElementNode(parent, doc, reader)); |
---|
477 | break; |
---|
478 | case XMLStreamConstants.END_ELEMENT: |
---|
479 | stack.pop(); |
---|
480 | if (stack.size() == 1) { |
---|
481 | stop = true; |
---|
482 | } |
---|
483 | break; |
---|
484 | case XMLStreamConstants.CHARACTERS: |
---|
485 | parent.appendChild(doc.createTextNode(reader.getText())); |
---|
486 | break; |
---|
487 | case XMLStreamConstants.COMMENT: |
---|
488 | parent.appendChild(doc.createComment(reader.getText())); |
---|
489 | break; |
---|
490 | case XMLStreamConstants.CDATA: |
---|
491 | parent.appendChild(doc.createCDATASection(reader.getText())); |
---|
492 | break; |
---|
493 | default: |
---|
494 | break; |
---|
495 | } |
---|
496 | reader.next(); |
---|
497 | } // while |
---|
498 | if (stack.size() != 1) { |
---|
499 | throw new XMLStreamException( |
---|
500 | "internal error; stack should hold only one element"); |
---|
501 | } |
---|
502 | return (DocumentFragment) stack.pop(); |
---|
503 | } catch (DOMException e) { |
---|
504 | throw new XMLStreamException( |
---|
505 | "error creating document fragment", e); |
---|
506 | } |
---|
507 | } |
---|
508 | |
---|
509 | |
---|
510 | private static Element createElementNode(Node parent, Document doc, |
---|
511 | XMLStreamReader reader) throws XMLStreamException, DOMException { |
---|
512 | Element element = doc.createElementNS(reader.getNamespaceURI(), |
---|
513 | reader.getLocalName()); |
---|
514 | |
---|
515 | if ((reader.getPrefix() != null) && !reader.getPrefix().isEmpty()) { |
---|
516 | element.setPrefix(reader.getPrefix()); |
---|
517 | } |
---|
518 | |
---|
519 | parent.appendChild(element); |
---|
520 | |
---|
521 | // add namespace declarations |
---|
522 | for (int i = 0; i < reader.getNamespaceCount(); i++) { |
---|
523 | final String uri = reader.getNamespaceURI(i); |
---|
524 | final String prefix = reader.getNamespacePrefix(i); |
---|
525 | |
---|
526 | if ((prefix != null) && !prefix.isEmpty()) { |
---|
527 | element.setAttributeNS(XMLConstants.XMLNS_ATTRIBUTE_NS_URI, |
---|
528 | XMLConstants.XMLNS_ATTRIBUTE + ":" + prefix, |
---|
529 | uri); |
---|
530 | } else { |
---|
531 | if (uri != null) { |
---|
532 | element.setAttributeNS(XMLConstants.XMLNS_ATTRIBUTE_NS_URI, |
---|
533 | XMLConstants.XMLNS_ATTRIBUTE, |
---|
534 | uri); |
---|
535 | } |
---|
536 | } |
---|
537 | } |
---|
538 | |
---|
539 | // add other attributes |
---|
540 | for (int i = 0; i < reader.getAttributeCount(); i++) { |
---|
541 | String name = reader.getAttributeLocalName(i); |
---|
542 | String prefix = reader.getAttributePrefix(i); |
---|
543 | if (prefix != null && prefix.length() > 0) { |
---|
544 | name = prefix + ":" + name; |
---|
545 | } |
---|
546 | |
---|
547 | Attr attr = doc.createAttributeNS( |
---|
548 | reader.getAttributeNamespace(i), name); |
---|
549 | attr.setValue(reader.getAttributeValue(i)); |
---|
550 | element.setAttributeNode(attr); |
---|
551 | } |
---|
552 | return element; |
---|
553 | } |
---|
554 | |
---|
555 | } // class SRUClient |
---|