1
|
/**
|
2
|
* Copyright: 2013 Regents of the University of California and the
|
3
|
* National Center for Ecological Analysis and Synthesis
|
4
|
*
|
5
|
* This program is free software; you can redistribute it and/or modify
|
6
|
* it under the terms of the GNU General Public License as published by
|
7
|
* the Free Software Foundation; either version 2 of the License, or
|
8
|
* (at your option) any later version.
|
9
|
*
|
10
|
* This program is distributed in the hope that it will be useful,
|
11
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
12
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
13
|
* GNU General Public License for more details.
|
14
|
*
|
15
|
* You should have received a copy of the GNU General Public License
|
16
|
* along with this program; if not, write to the Free Software
|
17
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA
|
18
|
*/
|
19
|
package edu.ucsb.nceas.metacat.index;
|
20
|
|
21
|
import java.io.ByteArrayInputStream;
|
22
|
import java.io.FileInputStream;
|
23
|
import java.io.FileNotFoundException;
|
24
|
import java.io.IOException;
|
25
|
import java.io.InputStream;
|
26
|
import java.util.ArrayList;
|
27
|
import java.util.Calendar;
|
28
|
import java.util.Date;
|
29
|
import java.util.HashMap;
|
30
|
import java.util.Iterator;
|
31
|
import java.util.List;
|
32
|
import java.util.Map;
|
33
|
import java.util.Set;
|
34
|
|
35
|
import javax.xml.parsers.DocumentBuilder;
|
36
|
import javax.xml.parsers.DocumentBuilderFactory;
|
37
|
import javax.xml.parsers.ParserConfigurationException;
|
38
|
import javax.xml.xpath.XPath;
|
39
|
import javax.xml.xpath.XPathExpressionException;
|
40
|
import javax.xml.xpath.XPathFactory;
|
41
|
|
42
|
import org.apache.commons.codec.EncoderException;
|
43
|
import org.apache.commons.io.output.ByteArrayOutputStream;
|
44
|
import org.apache.commons.lang.StringUtils;
|
45
|
import org.apache.commons.logging.Log;
|
46
|
import org.apache.commons.logging.LogFactory;
|
47
|
import org.apache.solr.client.solrj.SolrQuery;
|
48
|
import org.apache.solr.client.solrj.SolrServer;
|
49
|
import org.apache.solr.client.solrj.SolrServerException;
|
50
|
import org.apache.solr.client.solrj.response.QueryResponse;
|
51
|
import org.apache.solr.client.solrj.response.UpdateResponse;
|
52
|
import org.apache.solr.common.SolrDocument;
|
53
|
import org.apache.solr.common.SolrDocumentList;
|
54
|
import org.apache.solr.common.SolrInputDocument;
|
55
|
import org.apache.solr.schema.IndexSchema;
|
56
|
import org.dataone.cn.indexer.XMLNamespaceConfig;
|
57
|
import org.dataone.cn.indexer.convert.SolrDateConverter;
|
58
|
import org.dataone.cn.indexer.parser.IDocumentSubprocessor;
|
59
|
import org.dataone.cn.indexer.parser.SolrField;
|
60
|
import org.dataone.cn.indexer.resourcemap.ResourceEntry;
|
61
|
import org.dataone.cn.indexer.resourcemap.ResourceMap;
|
62
|
import org.dataone.cn.indexer.resourcemap.ResourceMapFactory;
|
63
|
import org.dataone.cn.indexer.solrhttp.SolrDoc;
|
64
|
import org.dataone.cn.indexer.solrhttp.SolrElementField;
|
65
|
import org.dataone.service.exceptions.NotFound;
|
66
|
import org.dataone.service.exceptions.NotImplemented;
|
67
|
import org.dataone.service.exceptions.ServiceFailure;
|
68
|
import org.dataone.service.exceptions.UnsupportedType;
|
69
|
import org.dataone.service.types.v1.Event;
|
70
|
import org.dataone.service.types.v1.Identifier;
|
71
|
import org.dataone.service.types.v1.SystemMetadata;
|
72
|
import org.dataone.service.util.DateTimeMarshaller;
|
73
|
import org.dataone.service.util.TypeMarshaller;
|
74
|
import org.dspace.foresite.OREParserException;
|
75
|
import org.jibx.runtime.JiBXException;
|
76
|
import org.w3c.dom.Document;
|
77
|
import org.xml.sax.SAXException;
|
78
|
|
79
|
import edu.ucsb.nceas.metacat.common.index.event.IndexEvent;
|
80
|
import edu.ucsb.nceas.metacat.common.query.SolrQueryServiceController;
|
81
|
import edu.ucsb.nceas.metacat.index.event.EventlogFactory;
|
82
|
import edu.ucsb.nceas.metacat.index.resourcemap.ResourceMapSubprocessor;
|
83
|
|
84
|
/**
|
85
|
* A class does insert, update and remove indexes to a SOLR server
|
86
|
* @author tao
|
87
|
*
|
88
|
*/
|
89
|
public class SolrIndex {
|
90
|
|
91
|
public static final String ID = "id";
|
92
|
private static final String IDQUERY = ID+":*";
|
93
|
private List<IDocumentSubprocessor> subprocessors = null;
|
94
|
private SolrServer solrServer = null;
|
95
|
private XMLNamespaceConfig xmlNamespaceConfig = null;
|
96
|
private List<SolrField> sysmetaSolrFields = null;
|
97
|
|
98
|
private static DocumentBuilderFactory documentBuilderFactory = null;
|
99
|
private static DocumentBuilder builder = null;
|
100
|
|
101
|
private static XPathFactory xpathFactory = null;
|
102
|
private static XPath xpath = null;
|
103
|
Log log = LogFactory.getLog(SolrIndex.class);
|
104
|
|
105
|
static {
|
106
|
documentBuilderFactory = DocumentBuilderFactory.newInstance();
|
107
|
documentBuilderFactory.setNamespaceAware(true);
|
108
|
try {
|
109
|
builder = documentBuilderFactory.newDocumentBuilder();
|
110
|
} catch (ParserConfigurationException e) {
|
111
|
e.printStackTrace();
|
112
|
}
|
113
|
xpathFactory = XPathFactory.newInstance();
|
114
|
xpath = xpathFactory.newXPath();
|
115
|
}
|
116
|
|
117
|
/**
|
118
|
* Constructor
|
119
|
* @throws SAXException
|
120
|
* @throws IOException
|
121
|
*/
|
122
|
public SolrIndex(XMLNamespaceConfig xmlNamespaceConfig, List<SolrField> sysmetaSolrFields)
|
123
|
throws XPathExpressionException, ParserConfigurationException, IOException, SAXException {
|
124
|
this.xmlNamespaceConfig = xmlNamespaceConfig;
|
125
|
this.sysmetaSolrFields = sysmetaSolrFields;
|
126
|
init();
|
127
|
}
|
128
|
|
129
|
private void init() throws ParserConfigurationException, XPathExpressionException {
|
130
|
xpath.setNamespaceContext(xmlNamespaceConfig);
|
131
|
initExpressions();
|
132
|
}
|
133
|
|
134
|
private void initExpressions() throws XPathExpressionException {
|
135
|
for (SolrField field : sysmetaSolrFields) {
|
136
|
field.initExpression(xpath);
|
137
|
}
|
138
|
|
139
|
}
|
140
|
|
141
|
|
142
|
/**
|
143
|
* Get the list of the Subprocessors in this index.
|
144
|
* @return the list of the Subprocessors.
|
145
|
*/
|
146
|
public List<IDocumentSubprocessor> getSubprocessors() {
|
147
|
return subprocessors;
|
148
|
}
|
149
|
|
150
|
/**
|
151
|
* Set the list of Subprocessors.
|
152
|
* @param subprocessorList the list will be set.
|
153
|
*/
|
154
|
public void setSubprocessors(List<IDocumentSubprocessor> subprocessorList) {
|
155
|
for (IDocumentSubprocessor subprocessor : subprocessorList) {
|
156
|
subprocessor.initExpression(xpath);
|
157
|
}
|
158
|
this.subprocessors = subprocessorList;
|
159
|
}
|
160
|
|
161
|
/**
|
162
|
* Generate the index for the given information
|
163
|
* @param id
|
164
|
* @param systemMetadata
|
165
|
* @param dataStream
|
166
|
* @return
|
167
|
* @throws IOException
|
168
|
* @throws SAXException
|
169
|
* @throws ParserConfigurationException
|
170
|
* @throws XPathExpressionException
|
171
|
* @throws JiBXException
|
172
|
* @throws SolrServerException
|
173
|
* @throws EncoderException
|
174
|
* @throws UnsupportedType
|
175
|
* @throws NotFound
|
176
|
* @throws NotImplemented
|
177
|
*/
|
178
|
private Map<String, SolrDoc> process(String id, SystemMetadata systemMetadata, String objectPath)
|
179
|
throws IOException, SAXException, ParserConfigurationException,
|
180
|
XPathExpressionException, JiBXException, EncoderException, SolrServerException, NotImplemented, NotFound, UnsupportedType{
|
181
|
|
182
|
// Load the System Metadata document
|
183
|
ByteArrayOutputStream systemMetadataOutputStream = new ByteArrayOutputStream();
|
184
|
TypeMarshaller.marshalTypeToOutputStream(systemMetadata, systemMetadataOutputStream);
|
185
|
ByteArrayInputStream systemMetadataStream = new ByteArrayInputStream(systemMetadataOutputStream.toByteArray());
|
186
|
Document sysMetaDoc = generateXmlDocument(systemMetadataStream);
|
187
|
if (sysMetaDoc == null) {
|
188
|
log.error("Could not load System metadata for ID: " + id);
|
189
|
return null;
|
190
|
}
|
191
|
|
192
|
// Extract the field values from the System Metadata
|
193
|
List<SolrElementField> sysSolrFields = processSysmetaFields(sysMetaDoc, id);
|
194
|
SolrDoc indexDocument = new SolrDoc(sysSolrFields);
|
195
|
Map<String, SolrDoc> docs = new HashMap<String, SolrDoc>();
|
196
|
docs.put(id, indexDocument);
|
197
|
|
198
|
// Determine if subprocessors are available for this ID
|
199
|
if (subprocessors != null) {
|
200
|
// for each subprocessor loaded from the spring config
|
201
|
for (IDocumentSubprocessor subprocessor : subprocessors) {
|
202
|
// Does this subprocessor apply?
|
203
|
if (subprocessor.canProcess(sysMetaDoc)) {
|
204
|
// if so, then extract the additional information from the
|
205
|
// document.
|
206
|
try {
|
207
|
// docObject = the resource map document or science
|
208
|
// metadata document.
|
209
|
// note that resource map processing touches all objects
|
210
|
// referenced by the resource map.
|
211
|
InputStream dataStream = new FileInputStream(objectPath);
|
212
|
Document docObject = generateXmlDocument(dataStream);
|
213
|
if (docObject == null) {
|
214
|
throw new Exception("Could not load OBJECT for ID " + id );
|
215
|
} else {
|
216
|
docs = subprocessor.processDocument(id, docs, docObject);
|
217
|
}
|
218
|
} catch (Exception e) {
|
219
|
log.error(e.getMessage(), e);
|
220
|
throw new SolrServerException(e.getMessage());
|
221
|
}
|
222
|
}
|
223
|
}
|
224
|
}
|
225
|
|
226
|
// TODO: in the XPathDocumentParser class in d1_cn_index_process module,
|
227
|
// merge is only for resource map. We need more work here.
|
228
|
for (SolrDoc mergeDoc : docs.values()) {
|
229
|
if (!mergeDoc.isMerged()) {
|
230
|
mergeWithIndexedDocument(mergeDoc);
|
231
|
}
|
232
|
}
|
233
|
|
234
|
//SolrElementAdd addCommand = getAddCommand(new ArrayList<SolrDoc>(docs.values()));
|
235
|
|
236
|
return docs;
|
237
|
}
|
238
|
|
239
|
/**
|
240
|
* Merge updates with existing solr documents
|
241
|
*
|
242
|
* This method appears to re-set the data package field data into the
|
243
|
* document about to be updated in the solr index. Since packaging
|
244
|
* information is derived from the package document (resource map), this
|
245
|
* information is not present when processing a document contained in a data
|
246
|
* package. This method replaces those values from the existing solr index
|
247
|
* record for the document being processed. -- sroseboo, 1-18-12
|
248
|
*
|
249
|
* @param indexDocument
|
250
|
* @return
|
251
|
* @throws IOException
|
252
|
* @throws EncoderException
|
253
|
* @throws XPathExpressionException
|
254
|
* @throws SAXException
|
255
|
* @throws ParserConfigurationException
|
256
|
* @throws SolrServerException
|
257
|
* @throws UnsupportedType
|
258
|
* @throws NotFound
|
259
|
* @throws NotImplemented
|
260
|
*/
|
261
|
// TODO:combine merge function with resourcemap merge function
|
262
|
|
263
|
private SolrDoc mergeWithIndexedDocument(SolrDoc indexDocument) throws IOException,
|
264
|
EncoderException, XPathExpressionException, SolrServerException, ParserConfigurationException, SAXException, NotImplemented, NotFound, UnsupportedType {
|
265
|
List<String> ids = new ArrayList<String>();
|
266
|
ids.add(indexDocument.getIdentifier());
|
267
|
List<SolrDoc> indexedDocuments = ResourceMapSubprocessor.getSolrDocs(ids);
|
268
|
SolrDoc indexedDocument = indexedDocuments == null || indexedDocuments.size() <= 0 ? null
|
269
|
: indexedDocuments.get(0);
|
270
|
|
271
|
IndexSchema indexSchema = SolrQueryServiceController.getInstance().getSchema();
|
272
|
|
273
|
if (indexedDocument == null || indexedDocument.getFieldList().size() <= 0) {
|
274
|
return indexDocument;
|
275
|
} else {
|
276
|
for (SolrElementField field : indexedDocument.getFieldList()) {
|
277
|
if ((field.getName().equals(SolrElementField.FIELD_ISDOCUMENTEDBY)
|
278
|
|| field.getName().equals(SolrElementField.FIELD_DOCUMENTS) || field
|
279
|
.getName().equals(SolrElementField.FIELD_RESOURCEMAP))
|
280
|
&& !indexDocument.hasFieldWithValue(field.getName(), field.getValue())) {
|
281
|
indexDocument.addField(field);
|
282
|
} else if (!indexSchema.isCopyFieldTarget(indexSchema.getField(field.getName())) && !indexDocument.hasField(field.getName())) {
|
283
|
indexDocument.addField(field);
|
284
|
}
|
285
|
}
|
286
|
|
287
|
indexDocument.setMerged(true);
|
288
|
return indexDocument;
|
289
|
}
|
290
|
}
|
291
|
|
292
|
/*
|
293
|
* Generate a Document from the InputStream
|
294
|
*/
|
295
|
private Document generateXmlDocument(InputStream smdStream) throws SAXException {
|
296
|
Document doc = null;
|
297
|
|
298
|
try {
|
299
|
doc = builder.parse(smdStream);
|
300
|
} catch (IOException e) {
|
301
|
log.error(e.getMessage(), e);
|
302
|
}
|
303
|
|
304
|
return doc;
|
305
|
}
|
306
|
|
307
|
/*
|
308
|
* Index the fields of the system metadata
|
309
|
*/
|
310
|
private List<SolrElementField> processSysmetaFields(Document doc, String identifier) {
|
311
|
|
312
|
List<SolrElementField> fieldList = new ArrayList<SolrElementField>();
|
313
|
// solrFields is the list of fields defined in the application context
|
314
|
|
315
|
for (SolrField field : sysmetaSolrFields) {
|
316
|
try {
|
317
|
// the field.getFields method can return a single value or
|
318
|
// multiple values for multi-valued fields
|
319
|
// or can return multiple SOLR document fields.
|
320
|
fieldList.addAll(field.getFields(doc, identifier));
|
321
|
} catch (Exception e) {
|
322
|
e.printStackTrace();
|
323
|
}
|
324
|
}
|
325
|
return fieldList;
|
326
|
|
327
|
}
|
328
|
|
329
|
/**
|
330
|
* Check the parameters of the insert or update methods.
|
331
|
* @param pid
|
332
|
* @param systemMetadata
|
333
|
* @param data
|
334
|
* @throws SolrServerException
|
335
|
*/
|
336
|
private void checkParams(Identifier pid, SystemMetadata systemMetadata, String objectPath) throws SolrServerException {
|
337
|
if(pid == null || pid.getValue() == null || pid.getValue().trim().equals("")) {
|
338
|
throw new SolrServerException("The identifier of the indexed document should not be null or blank.");
|
339
|
}
|
340
|
if(systemMetadata == null) {
|
341
|
throw new SolrServerException("The system metadata of the indexed document "+pid.getValue()+ " should not be null.");
|
342
|
}
|
343
|
if(objectPath == null) {
|
344
|
throw new SolrServerException("The indexed document itself for pid "+pid.getValue()+" should not be null.");
|
345
|
}
|
346
|
}
|
347
|
|
348
|
/**
|
349
|
* Insert the indexes for a document.
|
350
|
* @param pid the id of this document
|
351
|
* @param systemMetadata the system metadata associated with the data object
|
352
|
* @param data the path to the object file itself
|
353
|
* @throws SolrServerException
|
354
|
* @throws JiBXException
|
355
|
* @throws EncoderException
|
356
|
* @throws UnsupportedType
|
357
|
* @throws NotFound
|
358
|
* @throws NotImplemented
|
359
|
*/
|
360
|
private synchronized void insert(Identifier pid, SystemMetadata systemMetadata, String objectPath)
|
361
|
throws IOException, SAXException, ParserConfigurationException,
|
362
|
XPathExpressionException, SolrServerException, JiBXException, EncoderException, NotImplemented, NotFound, UnsupportedType {
|
363
|
checkParams(pid, systemMetadata, objectPath);
|
364
|
Map<String, SolrDoc> docs = process(pid.getValue(), systemMetadata, objectPath);
|
365
|
|
366
|
//transform the Map to the SolrInputDocument which can be used by the solr server
|
367
|
if(docs != null) {
|
368
|
Set<String> ids = docs.keySet();
|
369
|
for(String id : ids) {
|
370
|
if(id != null) {
|
371
|
SolrDoc doc = docs.get(id);
|
372
|
insertToIndex(doc);
|
373
|
}
|
374
|
|
375
|
}
|
376
|
}
|
377
|
}
|
378
|
|
379
|
/**
|
380
|
* Adds the given fields to the solr index for the given pid, preserving the index values
|
381
|
* that previously existed
|
382
|
* @param pid
|
383
|
* @param fields
|
384
|
*/
|
385
|
public void insertFields(Identifier pid, Map<String, List<Object>> fields) {
|
386
|
|
387
|
try {
|
388
|
// copy the original values already indexed for this document
|
389
|
SolrQuery query = new SolrQuery("id:\"" + pid.getValue() + "\"");
|
390
|
QueryResponse res = solrServer.query(query);
|
391
|
SolrDoc doc = new SolrDoc();
|
392
|
|
393
|
// include existing values if they exist
|
394
|
IndexSchema indexSchema = SolrQueryServiceController.getInstance().getSchema();
|
395
|
|
396
|
if (res.getResults().size() > 0) {
|
397
|
SolrDocument orig = res.getResults().get(0);
|
398
|
for (String fieldName: orig.getFieldNames()) {
|
399
|
// don't transfer the copyTo fields, otherwise there are errors
|
400
|
if (indexSchema.isCopyFieldTarget(indexSchema.getField(fieldName))) {
|
401
|
continue;
|
402
|
}
|
403
|
for (Object value: orig.getFieldValues(fieldName)) {
|
404
|
String stringValue = value.toString();
|
405
|
// special handling for dates in ISO 8601
|
406
|
if (value instanceof Date) {
|
407
|
stringValue = DateTimeMarshaller.serializeDateToUTC((Date)value);
|
408
|
SolrDateConverter converter = new SolrDateConverter();
|
409
|
stringValue = converter.convert(stringValue);
|
410
|
}
|
411
|
SolrElementField field = new SolrElementField(fieldName, stringValue);
|
412
|
log.debug("Adding field: " + fieldName);
|
413
|
doc.addField(field);
|
414
|
}
|
415
|
}
|
416
|
}
|
417
|
|
418
|
// add the additional fields we are trying to include in the index
|
419
|
for (String fieldName: fields.keySet()) {
|
420
|
List<Object> values = fields.get(fieldName);
|
421
|
for (Object value: values) {
|
422
|
if (!doc.hasFieldWithValue(fieldName, value.toString())) {
|
423
|
if (indexSchema.getField(fieldName).multiValued()) {
|
424
|
doc.addField(new SolrElementField(fieldName, value.toString()));
|
425
|
} else {
|
426
|
doc.updateOrAddField(fieldName, value.toString());
|
427
|
}
|
428
|
}
|
429
|
}
|
430
|
}
|
431
|
|
432
|
// make sure there is an id in the solrdoc so it is added to the index
|
433
|
if (!doc.hasField(ID)) {
|
434
|
doc.updateOrAddField(ID, pid.getValue());
|
435
|
}
|
436
|
|
437
|
// insert the whole thing
|
438
|
insertToIndex(doc);
|
439
|
} catch (Exception e) {
|
440
|
String error = "SolrIndex.insetFields - could not update the solr index: " + e.getMessage();
|
441
|
writeEventLog(null, pid, error);
|
442
|
log.error(error, e);
|
443
|
}
|
444
|
|
445
|
}
|
446
|
|
447
|
/*
|
448
|
* Insert a SolrDoc to the solr server.
|
449
|
*/
|
450
|
private synchronized void insertToIndex(SolrDoc doc) throws SolrServerException, IOException {
|
451
|
if(doc != null ) {
|
452
|
SolrInputDocument solrDoc = new SolrInputDocument();
|
453
|
List<SolrElementField> list = doc.getFieldList();
|
454
|
if(list != null) {
|
455
|
//solrDoc.addField(METACATPIDFIELD, pid);
|
456
|
Iterator<SolrElementField> iterator = list.iterator();
|
457
|
while (iterator.hasNext()) {
|
458
|
SolrElementField field = iterator.next();
|
459
|
if(field != null) {
|
460
|
String value = field.getValue();
|
461
|
String name = field.getName();
|
462
|
//System.out.println("add name/value pair - "+name+"/"+value);
|
463
|
solrDoc.addField(name, value);
|
464
|
}
|
465
|
}
|
466
|
}
|
467
|
if(!solrDoc.isEmpty()) {
|
468
|
/*IndexEvent event = new IndexEvent();
|
469
|
event.setDate(Calendar.getInstance().getTime());
|
470
|
Identifier pid = new Identifier();
|
471
|
pid.setValue(doc.getIdentifier());
|
472
|
event.setIdentifier(pid);*/
|
473
|
try {
|
474
|
UpdateResponse response = solrServer.add(solrDoc);
|
475
|
solrServer.commit();
|
476
|
/*event.setType(IndexEvent.SUCCESSINSERT);
|
477
|
event.setDescription("Successfully insert the solr index for the id "+pid.getValue());
|
478
|
try {
|
479
|
EventlogFactory.createIndexEventLog().write(event);
|
480
|
} catch (Exception e) {
|
481
|
log.error("SolrIndex.insertToIndex - IndexEventLog can't log the index inserting event :"+e.getMessage());
|
482
|
}*/
|
483
|
} catch (SolrServerException e) {
|
484
|
/*event.setAction(Event.CREATE);
|
485
|
event.setDescription("Failed to insert the solr index for the id "+pid.getValue()+" since "+e.getMessage());
|
486
|
try {
|
487
|
EventlogFactory.createIndexEventLog().write(event);
|
488
|
} catch (Exception ee) {
|
489
|
log.error("SolrIndex.insertToIndex - IndexEventLog can't log the index inserting event :"+ee.getMessage());
|
490
|
}*/
|
491
|
throw e;
|
492
|
} catch (IOException e) {
|
493
|
/*event.setAction(Event.CREATE);
|
494
|
event.setDescription("Failed to insert the solr index for the id "+pid.getValue()+" since "+e.getMessage());
|
495
|
try {
|
496
|
EventlogFactory.createIndexEventLog().write(event);
|
497
|
} catch (Exception ee) {
|
498
|
log.error("SolrIndex.insertToIndex - IndexEventLog can't log the index inserting event :"+ee.getMessage());
|
499
|
}*/
|
500
|
throw e;
|
501
|
|
502
|
}
|
503
|
//System.out.println("=================the response is:\n"+response.toString());
|
504
|
}
|
505
|
}
|
506
|
}
|
507
|
|
508
|
/**
|
509
|
* Update the solr index. This method handles the three scenarios:
|
510
|
* 1. Remove an existing doc - if the the system metadata shows the value of the archive is true,
|
511
|
* remove the index for the previous version(s) and generate new index for the doc.
|
512
|
* 2. Add a new doc - if the system metadata shows the value of the archive is false, generate the
|
513
|
* index for the doc.
|
514
|
*/
|
515
|
public void update(Identifier pid, SystemMetadata systemMetadata) {
|
516
|
String objectPath = null;
|
517
|
try {
|
518
|
objectPath = DistributedMapsFactory.getObjectPathMap().get(pid);
|
519
|
update(pid, systemMetadata, objectPath);
|
520
|
EventlogFactory.createIndexEventLog().remove(pid);
|
521
|
} catch (Exception e) {
|
522
|
String error = "SolrIndex.update - could not update the solr index since " + e.getMessage();
|
523
|
writeEventLog(systemMetadata, pid, error);
|
524
|
log.error(error, e);
|
525
|
}
|
526
|
}
|
527
|
|
528
|
|
529
|
/**
|
530
|
* Update the solr index. This method handles the three scenarios:
|
531
|
* 1. Remove an existing doc - if the the system metadata shows the value of the archive is true,
|
532
|
* remove the index for the previous version(s) and generate new index for the doc.
|
533
|
* 2. Add a new doc - if the system metadata shows the value of the archive is false, generate the
|
534
|
* index for the doc.
|
535
|
* @param pid
|
536
|
* @param systemMetadata
|
537
|
* @param data
|
538
|
* @throws SolrServerException
|
539
|
* @throws ServiceFailure
|
540
|
* @throws XPathExpressionException
|
541
|
* @throws NotImplemented
|
542
|
* @throws NotFound
|
543
|
* @throws UnsupportedType
|
544
|
* @throws IOException
|
545
|
* @throws SAXException
|
546
|
* @throws ParserConfigurationException
|
547
|
* @throws OREParserException
|
548
|
* @throws JiBXException
|
549
|
* @throws EncoderException
|
550
|
*/
|
551
|
void update(Identifier pid, SystemMetadata systemMetadata, String objectPath) throws SolrServerException,
|
552
|
ServiceFailure, XPathExpressionException, NotImplemented, NotFound, UnsupportedType,
|
553
|
IOException, SAXException, ParserConfigurationException, OREParserException, JiBXException, EncoderException {
|
554
|
checkParams(pid, systemMetadata, objectPath);
|
555
|
boolean isArchive = systemMetadata.getArchived() != null && systemMetadata.getArchived();
|
556
|
if(isArchive ) {
|
557
|
//delete the index for the archived objects
|
558
|
remove(pid.getValue(), systemMetadata);
|
559
|
log.info("SolrIndex.update============================= archive the idex for the identifier "+pid);
|
560
|
} else {
|
561
|
//generate index for either add or update.
|
562
|
insert(pid, systemMetadata, objectPath);
|
563
|
log.info("SolrIndex.update============================= insert index for the identifier "+pid);
|
564
|
}
|
565
|
}
|
566
|
|
567
|
|
568
|
|
569
|
/*
|
570
|
* Is the pid a resource map
|
571
|
*/
|
572
|
private boolean isDataPackage(String pid, SystemMetadata sysmeta) throws FileNotFoundException, ServiceFailure {
|
573
|
boolean isDataPackage = false;
|
574
|
//SystemMetadata sysmeta = DistributedMapsFactory.getSystemMetadata(pid);
|
575
|
if(sysmeta != null) {
|
576
|
isDataPackage = IndexGeneratorTimerTask.isResourceMap(sysmeta.getFormatId());
|
577
|
}
|
578
|
return isDataPackage;
|
579
|
}
|
580
|
|
581
|
private boolean isPartOfDataPackage(String pid) throws XPathExpressionException, NotImplemented, NotFound, UnsupportedType, SolrServerException, IOException, ParserConfigurationException, SAXException {
|
582
|
SolrDoc dataPackageIndexDoc = ResourceMapSubprocessor.getSolrDoc(pid);
|
583
|
if (dataPackageIndexDoc != null) {
|
584
|
String resourceMapId = dataPackageIndexDoc
|
585
|
.getFirstFieldValue(SolrElementField.FIELD_RESOURCEMAP);
|
586
|
return StringUtils.isNotEmpty(resourceMapId);
|
587
|
} else {
|
588
|
return false;
|
589
|
}
|
590
|
}
|
591
|
/**
|
592
|
* Remove the indexed associated with specified pid.
|
593
|
* @param pid the pid which the indexes are associated with
|
594
|
* @throws IOException
|
595
|
* @throws SolrServerException
|
596
|
* @throws ParserConfigurationException
|
597
|
* @throws SAXException
|
598
|
* @throws UnsupportedType
|
599
|
* @throws NotFound
|
600
|
* @throws NotImplemented
|
601
|
* @throws XPathExpressionException
|
602
|
* @throws ServiceFailure
|
603
|
* @throws OREParserException
|
604
|
*/
|
605
|
private void remove(String pid, SystemMetadata sysmeta) throws IOException, SolrServerException, ServiceFailure, XPathExpressionException, NotImplemented, NotFound, UnsupportedType, SAXException, ParserConfigurationException, OREParserException {
|
606
|
if (isDataPackage(pid, sysmeta)) {
|
607
|
removeDataPackage(pid);
|
608
|
} else if (isPartOfDataPackage(pid)) {
|
609
|
removeFromDataPackage(pid);
|
610
|
} else {
|
611
|
removeFromIndex(pid);
|
612
|
}
|
613
|
}
|
614
|
|
615
|
/*
|
616
|
* Remove a resource map pid
|
617
|
*/
|
618
|
private void removeDataPackage(String pid) throws ServiceFailure, SAXException, XPathExpressionException, NotImplemented, NotFound, UnsupportedType, SolrServerException, IOException, ParserConfigurationException, OREParserException {
|
619
|
Document resourceMapDoc = generateXmlDocument(DistributedMapsFactory.getDataObject(pid));
|
620
|
//ResourceMap resourceMap = new ResourceMap(resourceMapDoc);
|
621
|
ResourceMap resourceMap = ResourceMapFactory.buildResourceMap(resourceMapDoc);
|
622
|
List<String> documentIds = resourceMap.getAllDocumentIDs();
|
623
|
List<SolrDoc> indexDocuments =ResourceMapSubprocessor.getSolrDocs(documentIds);
|
624
|
removeFromIndex(pid);
|
625
|
//List<SolrDoc> docsToUpdate = new ArrayList<SolrDoc>();
|
626
|
// for each document in data package:
|
627
|
for (SolrDoc indexDoc : indexDocuments) {
|
628
|
|
629
|
if (indexDoc.getIdentifier().equals(pid)) {
|
630
|
continue; // skipping the resource map, no need update
|
631
|
// it.
|
632
|
// will
|
633
|
// be removed.
|
634
|
}
|
635
|
|
636
|
// Remove resourceMap reference
|
637
|
indexDoc.removeFieldsWithValue(SolrElementField.FIELD_RESOURCEMAP,
|
638
|
resourceMap.getIdentifier());
|
639
|
|
640
|
// // Remove documents/documentedby values for this resource
|
641
|
// map
|
642
|
for (ResourceEntry entry : resourceMap.getMappedReferences()) {
|
643
|
if (indexDoc.getIdentifier().equals(entry.getIdentifier())) {
|
644
|
for (String documentedBy : entry.getDocumentedBy()) {
|
645
|
// Using removeOneFieldWithValue in-case same
|
646
|
// documents
|
647
|
// are in more than one data package. just
|
648
|
// remove
|
649
|
// one
|
650
|
// instance of data package info.
|
651
|
indexDoc.removeOneFieldWithValue(SolrElementField.FIELD_ISDOCUMENTEDBY,
|
652
|
documentedBy);
|
653
|
}
|
654
|
for (String documents : entry.getDocuments()) {
|
655
|
indexDoc.removeOneFieldWithValue(SolrElementField.FIELD_DOCUMENTS,
|
656
|
documents);
|
657
|
}
|
658
|
break;
|
659
|
}
|
660
|
}
|
661
|
removeFromIndex(indexDoc.getIdentifier());
|
662
|
insertToIndex(indexDoc);
|
663
|
//docsToUpdate.add(indexDoc);
|
664
|
}
|
665
|
//SolrElementAdd addCommand = new SolrElementAdd(docsToUpdate);
|
666
|
//httpService.sendUpdate(solrIndexUri, addCommand);
|
667
|
}
|
668
|
|
669
|
/*
|
670
|
* Remove a pid which is part of resource map.
|
671
|
*/
|
672
|
private void removeFromDataPackage(String pid) throws XPathExpressionException, NotImplemented, NotFound, UnsupportedType, SolrServerException, IOException, ParserConfigurationException, SAXException {
|
673
|
SolrDoc indexedDoc = ResourceMapSubprocessor.getSolrDoc(pid);
|
674
|
removeFromIndex(pid);
|
675
|
List<SolrDoc> docsToUpdate = new ArrayList<SolrDoc>();
|
676
|
|
677
|
List<String> documents = indexedDoc.getAllFieldValues(SolrElementField.FIELD_DOCUMENTS);
|
678
|
for (String documentsValue : documents) {
|
679
|
SolrDoc solrDoc = ResourceMapSubprocessor.getSolrDoc(documentsValue);
|
680
|
solrDoc.removeFieldsWithValue(SolrElementField.FIELD_ISDOCUMENTEDBY, pid);
|
681
|
removeFromIndex(documentsValue);
|
682
|
insertToIndex(solrDoc);
|
683
|
}
|
684
|
|
685
|
List<String> documentedBy = indexedDoc
|
686
|
.getAllFieldValues(SolrElementField.FIELD_ISDOCUMENTEDBY);
|
687
|
for (String documentedByValue : documentedBy) {
|
688
|
SolrDoc solrDoc = ResourceMapSubprocessor.getSolrDoc(documentedByValue);
|
689
|
solrDoc.removeFieldsWithValue(SolrElementField.FIELD_DOCUMENTS, pid);
|
690
|
//docsToUpdate.add(solrDoc);
|
691
|
removeFromIndex(documentedByValue);
|
692
|
insertToIndex(solrDoc);
|
693
|
}
|
694
|
|
695
|
//SolrElementAdd addCommand = new SolrElementAdd(docsToUpdate);
|
696
|
//httpService.sendUpdate(solrIndexUri, addCommand);
|
697
|
}
|
698
|
|
699
|
/*
|
700
|
* Remove a pid from the solr index
|
701
|
*/
|
702
|
private synchronized void removeFromIndex(String pid) throws SolrServerException, IOException {
|
703
|
if(pid != null && !pid.trim().equals("")) {
|
704
|
/*IndexEvent event = new IndexEvent();
|
705
|
event.setDate(Calendar.getInstance().getTime());
|
706
|
Identifier identifier = new Identifier();
|
707
|
identifier.setValue(pid);
|
708
|
event.setIdentifier(identifier);*/
|
709
|
try {
|
710
|
solrServer.deleteById(pid);
|
711
|
solrServer.commit();
|
712
|
/*event.setType(IndexEvent.SUCCESSDELETE);
|
713
|
event.setDescription("Successfully remove the solr index for the id "+identifier.getValue());
|
714
|
try {
|
715
|
EventlogFactory.createIndexEventLog().write(event);
|
716
|
} catch (Exception e) {
|
717
|
log.error("SolrIndex.removeFromIndex - IndexEventLog can't log the index deleting event :"+e.getMessage());
|
718
|
}*/
|
719
|
} catch (SolrServerException e) {
|
720
|
/*event.setAction(Event.DELETE);
|
721
|
event.setDescription("Failurely remove the solr index for the id "+identifier.getValue()+" since "+e.getMessage());
|
722
|
try {
|
723
|
EventlogFactory.createIndexEventLog().write(event);
|
724
|
} catch (Exception ee) {
|
725
|
log.error("SolrIndex.removeFromIndex - IndexEventLog can't log the index deleting event :"+ee.getMessage());
|
726
|
}*/
|
727
|
throw e;
|
728
|
|
729
|
} catch (IOException e) {
|
730
|
/*event.setAction(Event.DELETE);
|
731
|
event.setDescription("Failurely remove the solr index for the id "+identifier.getValue()+" since "+e.getMessage());
|
732
|
try {
|
733
|
EventlogFactory.createIndexEventLog().write(event);
|
734
|
} catch (Exception ee) {
|
735
|
log.error("SolrIndex.removeFromIndex - IndexEventLog can't log the index deleting event :"+ee.getMessage());
|
736
|
}*/
|
737
|
throw e;
|
738
|
}
|
739
|
|
740
|
}
|
741
|
}
|
742
|
|
743
|
/**
|
744
|
* Get the solrServer
|
745
|
* @return
|
746
|
*/
|
747
|
public SolrServer getSolrServer() {
|
748
|
return solrServer;
|
749
|
}
|
750
|
|
751
|
/**
|
752
|
* Set the solrServer.
|
753
|
* @param solrServer
|
754
|
*/
|
755
|
public void setSolrServer(SolrServer solrServer) {
|
756
|
this.solrServer = solrServer;
|
757
|
}
|
758
|
|
759
|
/**
|
760
|
* Get all indexed ids in the solr server.
|
761
|
* @return an empty list if there is no index.
|
762
|
* @throws SolrServerException
|
763
|
*/
|
764
|
public List<String> getSolrIds() throws SolrServerException {
|
765
|
List<String> list = new ArrayList<String>();
|
766
|
SolrQuery query = new SolrQuery(IDQUERY);
|
767
|
query.setRows(Integer.MAX_VALUE);
|
768
|
query.setFields(ID);
|
769
|
QueryResponse response = solrServer.query(query);
|
770
|
SolrDocumentList docs = response.getResults();
|
771
|
if(docs != null) {
|
772
|
for(SolrDocument doc :docs) {
|
773
|
String identifier = (String)doc.getFieldValue(ID);
|
774
|
//System.out.println("======================== "+identifier);
|
775
|
list.add(identifier);
|
776
|
}
|
777
|
}
|
778
|
return list;
|
779
|
}
|
780
|
|
781
|
private void writeEventLog(SystemMetadata systemMetadata, Identifier pid, String error) {
|
782
|
IndexEvent event = new IndexEvent();
|
783
|
event.setIdentifier(pid);
|
784
|
event.setDate(Calendar.getInstance().getTime());
|
785
|
String action = null;
|
786
|
if (systemMetadata == null ) {
|
787
|
action = Event.CREATE.xmlValue();
|
788
|
event.setAction(Event.CREATE);
|
789
|
}
|
790
|
else if(systemMetadata.getArchived()) {
|
791
|
action = Event.DELETE.xmlValue();
|
792
|
event.setAction(Event.DELETE);
|
793
|
} else {
|
794
|
action = Event.CREATE.xmlValue();
|
795
|
event.setAction(Event.CREATE);
|
796
|
}
|
797
|
event.setDescription("Failed to "+action+"the solr index for the id "+pid.getValue()+" since "+error);
|
798
|
try {
|
799
|
EventlogFactory.createIndexEventLog().write(event);
|
800
|
} catch (Exception ee) {
|
801
|
log.error("SolrIndex.insertToIndex - IndexEventLog can't log the index inserting event :"+ee.getMessage());
|
802
|
}
|
803
|
}
|
804
|
}
|