1
|
|
|
<?php |
2
|
|
|
namespace EWW\Dpf\Services\ElasticSearch; |
3
|
|
|
|
4
|
|
|
/* |
5
|
|
|
* This file is part of the TYPO3 CMS project. |
6
|
|
|
* |
7
|
|
|
* It is free software; you can redistribute it and/or modify it under |
8
|
|
|
* the terms of the GNU General Public License, either version 2 |
9
|
|
|
* of the License, or any later version. |
10
|
|
|
* |
11
|
|
|
* For the full copyright and license information, please read the |
12
|
|
|
* LICENSE.txt file that was distributed with this source code. |
13
|
|
|
* |
14
|
|
|
* The TYPO3 project - inspiring people to share! |
15
|
|
|
*/ |
16
|
|
|
|
17
|
|
|
use Elasticsearch\ClientBuilder; |
18
|
|
|
use Elasticsearch\Common\Exceptions\Curl\CouldNotConnectToHost; |
19
|
|
|
use Elasticsearch\Common\Exceptions\Curl\CouldNotResolveHostException; |
20
|
|
|
use EWW\Dpf\Domain\Repository\FrontendUserRepository; |
21
|
|
|
use EWW\Dpf\Domain\Workflow\DocumentWorkflow; |
22
|
|
|
use EWW\Dpf\Exceptions\ElasticSearchConnectionErrorException; |
23
|
|
|
use EWW\Dpf\Exceptions\ElasticSearchMissingIndexNameException; |
24
|
|
|
use TYPO3\CMS\Extbase\Object\ObjectManager; |
25
|
|
|
use EWW\Dpf\Configuration\ClientConfigurationManager; |
26
|
|
|
use EWW\Dpf\Domain\Model\Document; |
27
|
|
|
use TYPO3\CMS\Core\Utility\GeneralUtility; |
28
|
|
|
use TYPO3\CMS\Core\Log\LogManager; |
29
|
|
|
use TYPO3\CMS\Extbase\Utility\LocalizationUtility; |
30
|
|
|
|
31
|
|
|
class ElasticSearch |
32
|
|
|
{ |
33
|
|
|
/** |
34
|
|
|
* @var \EWW\Dpf\Configuration\ClientConfigurationManager |
35
|
|
|
*/ |
36
|
|
|
protected $clientConfigurationManager; |
37
|
|
|
|
38
|
|
|
/** |
39
|
|
|
* @var \Elasticsearch\Client |
40
|
|
|
*/ |
41
|
|
|
protected $client; |
42
|
|
|
|
43
|
|
|
protected $server = 'host.docker.internal'; //127.0.0.1'; |
44
|
|
|
|
45
|
|
|
protected $port = '9200'; |
46
|
|
|
|
47
|
|
|
protected $indexName = 'kitodo_publication'; |
48
|
|
|
|
49
|
|
|
protected $results; |
50
|
|
|
|
51
|
|
|
|
52
|
|
|
protected $elasticsearchMapper; |
53
|
|
|
|
54
|
|
|
/** |
55
|
|
|
* @var int |
56
|
|
|
*/ |
57
|
|
|
protected $clientPid = 0; |
58
|
|
|
|
59
|
|
|
/** |
60
|
|
|
* elasticsearch client constructor |
61
|
|
|
* @param int|null $clientPid |
62
|
|
|
* @throws ElasticSearchMissingIndexNameException |
63
|
|
|
*/ |
64
|
|
|
public function __construct($clientPid = null) |
65
|
|
|
{ |
66
|
|
|
$objectManager = GeneralUtility::makeInstance(ObjectManager::class); |
67
|
|
|
|
68
|
|
|
$this->elasticsearchMapper = $objectManager->get(ElasticsearchMapper::class); |
69
|
|
|
|
70
|
|
|
$this->clientConfigurationManager = $objectManager->get(ClientConfigurationManager::class); |
71
|
|
|
|
72
|
|
|
if ($clientPid) { |
|
|
|
|
73
|
|
|
$this->clientConfigurationManager->setConfigurationPid($clientPid); |
74
|
|
|
$this->clientPid = $clientPid; |
75
|
|
|
} |
76
|
|
|
|
77
|
|
|
$this->server = $this->clientConfigurationManager->getElasticSearchHost(); |
78
|
|
|
$this->port = $this->clientConfigurationManager->getElasticSearchPort(); |
79
|
|
|
$this->indexName = $this->clientConfigurationManager->getElasticSearchIndexName(); |
80
|
|
|
|
81
|
|
|
if (empty($this->indexName)) { |
82
|
|
|
throw new ElasticSearchMissingIndexNameException('Missing search index name.'); |
83
|
|
|
} |
84
|
|
|
|
85
|
|
|
$hosts = array( |
86
|
|
|
$this->server . ':' . $this->port, |
87
|
|
|
); |
88
|
|
|
|
89
|
|
|
$clientBuilder = ClientBuilder::create(); |
90
|
|
|
$clientBuilder->setHosts($hosts); |
91
|
|
|
$this->client = $clientBuilder->build(); |
92
|
|
|
|
93
|
|
|
try { |
94
|
|
|
$this->initializeIndex($this->indexName); |
95
|
|
|
} catch (\Throwable $e) { |
96
|
|
|
$message = LocalizationUtility::translate( |
97
|
|
|
'elasticsearch.notRunning', 'dpf' |
98
|
|
|
); |
99
|
|
|
die($message); |
|
|
|
|
100
|
|
|
} |
101
|
|
|
} |
102
|
|
|
|
103
|
|
|
/** |
104
|
|
|
* @return string|null |
105
|
|
|
*/ |
106
|
|
|
protected function getIndexName() |
107
|
|
|
{ |
108
|
|
|
return $this->indexName; |
109
|
|
|
} |
110
|
|
|
|
111
|
|
|
/** |
112
|
|
|
* Creates an index named by $indexName if it doesn't exist. |
113
|
|
|
* |
114
|
|
|
* @param $indexName |
115
|
|
|
*/ |
116
|
|
|
protected function initializeIndex($indexName) |
117
|
|
|
{ |
118
|
|
|
$paramsIndex = [ |
119
|
|
|
'index' => $indexName, |
120
|
|
|
'body' => [ |
121
|
|
|
'settings' => [ |
122
|
|
|
//'index.requests.cache.enable' => false, |
123
|
|
|
'analysis' => [ |
124
|
|
|
'filter' => [ |
125
|
|
|
'ngram' => [ |
126
|
|
|
'type' => 'ngram', |
127
|
|
|
'min_gram' => 3, |
128
|
|
|
'max_gram' => 3, |
129
|
|
|
'token_chars' => [ |
130
|
|
|
'letter', |
131
|
|
|
'digit' |
132
|
|
|
], |
133
|
|
|
] |
134
|
|
|
], |
135
|
|
|
'analyzer' => [ |
136
|
|
|
'keyword_lowercase' => [ |
137
|
|
|
'tokenizer' => 'keyword', |
138
|
|
|
'filter' => ['lowercase'] |
139
|
|
|
] |
140
|
|
|
], |
141
|
|
|
'normalizer' => [ |
142
|
|
|
'lowercase_normalizer' => [ |
143
|
|
|
'type' => 'custom', |
144
|
|
|
'char_filter' => [], |
145
|
|
|
'filter' => [ |
146
|
|
|
'lowercase', |
147
|
|
|
'asciifolding' |
148
|
|
|
] |
149
|
|
|
] |
150
|
|
|
] |
151
|
|
|
] |
152
|
|
|
], |
153
|
|
|
'mappings' => [ |
154
|
|
|
'_source' => [ |
155
|
|
|
'enabled' => true |
156
|
|
|
], |
157
|
|
|
//'dynamic' => 'strict', |
158
|
|
|
'properties' => [ |
159
|
|
|
'title' => [ |
160
|
|
|
'type' => 'text', |
161
|
|
|
'fields' => [ |
162
|
|
|
'keyword' => [ |
163
|
|
|
'type' => 'keyword', |
164
|
|
|
'normalizer' => 'lowercase_normalizer' |
165
|
|
|
] |
166
|
|
|
] |
167
|
|
|
], |
168
|
|
|
'state' => [ |
169
|
|
|
'type' => 'keyword' |
170
|
|
|
], |
171
|
|
|
'aliasState' => [ |
172
|
|
|
'type' => 'keyword' |
173
|
|
|
], |
174
|
|
|
'year' => [ |
175
|
|
|
'type' => 'integer' |
176
|
|
|
], |
177
|
|
|
'persons' => [ |
178
|
|
|
'type' => 'keyword' |
179
|
|
|
], |
180
|
|
|
'personsSort' => [ |
181
|
|
|
'type' => 'text', |
182
|
|
|
'fields' => [ |
183
|
|
|
'keyword' => [ |
184
|
|
|
'type' => 'keyword', |
185
|
|
|
'normalizer' => 'lowercase_normalizer' |
186
|
|
|
] |
187
|
|
|
] |
188
|
|
|
], |
189
|
|
|
'doctype' => [ |
190
|
|
|
'type' => 'keyword' |
191
|
|
|
], |
192
|
|
|
'collections' => [ |
193
|
|
|
'type' => 'keyword' |
194
|
|
|
], |
195
|
|
|
'hasFiles' => [ |
196
|
|
|
'type' => 'keyword' |
197
|
|
|
], |
198
|
|
|
'creator' => [ |
199
|
|
|
'type' => 'keyword' |
200
|
|
|
], |
201
|
|
|
'creatorRole' => [ |
202
|
|
|
'type' => 'keyword' |
203
|
|
|
], |
204
|
|
|
'source' => [ |
205
|
|
|
'type' => 'text' |
206
|
|
|
], |
207
|
|
|
'fobIdentifiers' => [ |
208
|
|
|
'type' => 'keyword' |
209
|
|
|
], |
210
|
|
|
'personData' => [ |
211
|
|
|
//'enabled' => false, |
212
|
|
|
'properties' => [ |
213
|
|
|
'name' => [ |
214
|
|
|
'type' => 'keyword' |
215
|
|
|
], |
216
|
|
|
'fobId' => [ |
217
|
|
|
//'type' => 'keyword' |
218
|
|
|
'enabled' => false |
219
|
|
|
], |
220
|
|
|
'index' => [ |
221
|
|
|
//'type' => 'integer' |
222
|
|
|
'enabled' => false |
223
|
|
|
] |
224
|
|
|
] |
225
|
|
|
], |
226
|
|
|
'affiliation' => [ |
227
|
|
|
'type' => 'keyword' |
228
|
|
|
], |
229
|
|
|
'process_number' => [ |
230
|
|
|
'type' => 'keyword' |
231
|
|
|
], |
232
|
|
|
'creationDate' => [ |
233
|
|
|
'type' => 'date', |
234
|
|
|
'format'=> "yyyy-MM-dd" |
235
|
|
|
], |
236
|
|
|
'embargoDate' => [ |
237
|
|
|
'type' => 'date', |
238
|
|
|
'format'=> "yyyy-MM-dd" |
239
|
|
|
] |
240
|
|
|
] |
241
|
|
|
] |
242
|
|
|
] |
243
|
|
|
]; |
244
|
|
|
|
245
|
|
|
if (!$this->client->indices()->exists(['index' => $indexName])) { |
246
|
|
|
$this->client->indices()->create($paramsIndex); |
247
|
|
|
} |
248
|
|
|
} |
249
|
|
|
|
250
|
|
|
/** |
251
|
|
|
* Adds an document to the index. |
252
|
|
|
* |
253
|
|
|
* @param Document $document |
254
|
|
|
*/ |
255
|
|
|
public function index($document) |
256
|
|
|
{ |
257
|
|
|
try { |
258
|
|
|
$data = json_decode($this->elasticsearchMapper->getElasticsearchJson($document)); |
259
|
|
|
} catch (\Throwable $throwable) { |
260
|
|
|
// Fixme: The solution via json_decode and the XSLT file needs to be replaced. |
261
|
|
|
} |
262
|
|
|
|
263
|
|
|
if (!$data) { |
264
|
|
|
$data->title[] = $document->getTitle(); |
265
|
|
|
$data->doctype = $document->getDocumentType()->getName(); |
266
|
|
|
} |
267
|
|
|
|
268
|
|
|
if ($data) { |
269
|
|
|
|
270
|
|
|
$data->state = $document->getState(); |
271
|
|
|
$data->aliasState = DocumentWorkflow::STATE_TO_ALIASSTATE_MAPPING[$document->getState()]; |
272
|
|
|
|
273
|
|
|
$data->objectIdentifier = $document->getObjectIdentifier(); |
274
|
|
|
|
275
|
|
|
if (!$data->identifier || !is_array($data->identifier)) { |
276
|
|
|
$data->identifier = []; |
277
|
|
|
} |
278
|
|
|
$data->identifier[] = $document->getObjectIdentifier(); |
279
|
|
|
$data->identifier[] = $document->getProcessNumber(); |
280
|
|
|
|
281
|
|
|
if ($document->getCreator()) { |
282
|
|
|
$data->creator = $document->getCreator(); |
283
|
|
|
} else { |
284
|
|
|
$data->creator = null; |
285
|
|
|
} |
286
|
|
|
|
287
|
|
|
if ($document->getCreator()) { |
288
|
|
|
$objectManager = GeneralUtility::makeInstance(ObjectManager::class); |
289
|
|
|
$frontendUserRepository = $objectManager->get(FrontendUserRepository::class); |
290
|
|
|
|
291
|
|
|
/** @var \EWW\Dpf\Domain\Model\FrontendUser $creatorFeUser */ |
292
|
|
|
$creatorFeUser = $frontendUserRepository->findByUid($document->getCreator()); |
293
|
|
|
if ($creatorFeUser) { |
|
|
|
|
294
|
|
|
$data->creatorRole = $creatorFeUser->getUserRole(); |
295
|
|
|
} else { |
296
|
|
|
$data->creatorRole = ''; |
297
|
|
|
} |
298
|
|
|
} else { |
299
|
|
|
$data->creatorRole = ''; |
300
|
|
|
} |
301
|
|
|
|
302
|
|
|
$creationDate = new \DateTime($document->getCreationDate()); |
303
|
|
|
|
304
|
|
|
$data->creationDate = $creationDate->format('Y-m-d'); |
305
|
|
|
|
306
|
|
|
$data->year = $document->getPublicationYear(); |
307
|
|
|
|
308
|
|
|
$notes = $document->getNotes(); |
309
|
|
|
|
310
|
|
|
if ($notes && is_array($notes)) { |
311
|
|
|
$data->notes = $notes; |
312
|
|
|
} else { |
313
|
|
|
$data->notes = array(); |
314
|
|
|
} |
315
|
|
|
|
316
|
|
|
$files = $document->getFile(); |
317
|
|
|
if ($files->count() > 0) { |
318
|
|
|
$data->hasFiles = true; |
319
|
|
|
} else { |
320
|
|
|
$data->hasFiles = false; |
321
|
|
|
} |
322
|
|
|
|
323
|
|
|
$internalFormat = new \EWW\Dpf\Helper\InternalFormat($document->getXmlData(), $this->clientPid); |
324
|
|
|
|
325
|
|
|
//$persons = array_merge($internalFormat->getAuthors(), $internalFormat->getPublishers()); |
326
|
|
|
$persons = $internalFormat->getPersons(); |
327
|
|
|
|
328
|
|
|
$fobIdentifiers = []; |
329
|
|
|
$personData = []; |
330
|
|
|
foreach ($persons as $person) { |
331
|
|
|
$fobIdentifiers[] = $person['fobId']; |
332
|
|
|
$personData[] = $person; |
333
|
|
|
//$data->persons[] = $person['name']; |
334
|
|
|
$data->persons[] = $person['fobId']; |
335
|
|
|
|
336
|
|
|
foreach ($person['affiliations'] as $affiliation) { |
337
|
|
|
$data->affiliation[] = $affiliation; |
338
|
|
|
} |
339
|
|
|
|
340
|
|
|
foreach ($person['affiliationIdentifiers'] as $affiliationIdentifier) { |
341
|
|
|
$data->affiliation[] = $affiliationIdentifier; |
342
|
|
|
} |
343
|
|
|
} |
344
|
|
|
|
345
|
|
|
$data->fobIdentifiers = $fobIdentifiers; |
346
|
|
|
$data->personData = $personData; |
347
|
|
|
|
348
|
|
|
if (sizeof($persons) > 0) { |
349
|
|
|
if (array_key_exists('family', $persons[0])) { |
350
|
|
|
$data->personsSort = $persons[0]['family']; |
351
|
|
|
} |
352
|
|
|
} |
353
|
|
|
|
354
|
|
|
$data->source = $document->getSourceDetails(); |
355
|
|
|
|
356
|
|
|
$data->universityCollection = false; |
357
|
|
|
if ($data->collections && is_array($data->collections)) { |
358
|
|
|
foreach ($data->collections as $collection) { |
359
|
|
|
if ($collection == $this->clientConfigurationManager->getUniversityCollection()) { |
360
|
|
|
$data->universityCollection = true; |
361
|
|
|
break; |
362
|
|
|
} |
363
|
|
|
} |
364
|
|
|
} |
365
|
|
|
|
366
|
|
|
$embargoDate = $document->getEmbargoDate(); |
367
|
|
|
if ($embargoDate instanceof \DateTime) { |
|
|
|
|
368
|
|
|
$data->embargoDate = $embargoDate->format("Y-m-d"); |
369
|
|
|
} else { |
370
|
|
|
$data->embargoDate = null; |
371
|
|
|
} |
372
|
|
|
|
373
|
|
|
$data->originalSourceTitle = $internalFormat->getOriginalSourceTitle(); |
374
|
|
|
|
375
|
|
|
$data->fobIdentifiers = $internalFormat->getPersonFisIdentifiers(); |
376
|
|
|
|
377
|
|
|
$this->client->index([ |
378
|
|
|
'refresh' => 'wait_for', |
379
|
|
|
'index' => $this->getIndexName(), |
380
|
|
|
'id' => $document->getDocumentIdentifier(), |
381
|
|
|
'body' => $data |
382
|
|
|
]); |
383
|
|
|
|
384
|
|
|
} |
385
|
|
|
|
386
|
|
|
} |
387
|
|
|
|
388
|
|
|
|
389
|
|
|
/** |
390
|
|
|
* Deletes a document from the index |
391
|
|
|
* |
392
|
|
|
* @param string $identifier |
393
|
|
|
*/ |
394
|
|
|
public function delete($identifier) |
395
|
|
|
{ |
396
|
|
|
try { |
397
|
|
|
|
398
|
|
|
$params = [ |
399
|
|
|
'refresh' => 'wait_for', |
400
|
|
|
'index' => $this->getIndexName(), |
401
|
|
|
'id' => $identifier |
402
|
|
|
]; |
403
|
|
|
|
404
|
|
|
$this->client->delete($params); |
405
|
|
|
|
406
|
|
|
} catch (\Exception $e) { |
407
|
|
|
/** @var $logger \TYPO3\CMS\Core\Log\Logger */ |
408
|
|
|
$logger = GeneralUtility::makeInstance(LogManager::class)->getLogger(__CLASS__); |
409
|
|
|
$logger->warning('Document could not be deleted from the index.', |
410
|
|
|
[ |
411
|
|
|
'Document identifier' => $identifier |
412
|
|
|
] |
413
|
|
|
); |
414
|
|
|
} |
415
|
|
|
} |
416
|
|
|
|
417
|
|
|
|
418
|
|
|
/** |
419
|
|
|
* @param $identifier |
420
|
|
|
*/ |
421
|
|
|
public function getDocument($identifier) |
422
|
|
|
{ |
423
|
|
|
$params = [ |
424
|
|
|
'index' => $this->getIndexName(), |
425
|
|
|
'id' => $identifier |
426
|
|
|
]; |
427
|
|
|
|
428
|
|
|
return $this->client->get($params); |
429
|
|
|
} |
430
|
|
|
|
431
|
|
|
|
432
|
|
|
/** |
433
|
|
|
* performs the |
434
|
|
|
* @param array $query search query |
435
|
|
|
* @return array result list |
436
|
|
|
*/ |
437
|
|
|
public function search($query, $type = null) |
438
|
|
|
{ |
439
|
|
|
try { |
440
|
|
|
// define type and index |
441
|
|
|
if (empty($query['index'])) { |
442
|
|
|
$query['index'] = $this->getIndexName(); |
443
|
|
|
} |
444
|
|
|
if (!empty($type)) { |
445
|
|
|
//$query['type'] = $type; |
446
|
|
|
// $query['type'] = $this->type; |
447
|
|
|
} |
448
|
|
|
|
449
|
|
|
// Search request |
450
|
|
|
$results = $this->client->search($query); |
451
|
|
|
|
452
|
|
|
//$this->hits = $results['hits']['total']; |
453
|
|
|
|
454
|
|
|
//$this->resultList = $results['hits']; |
455
|
|
|
|
456
|
|
|
$this->results = $results; |
457
|
|
|
|
458
|
|
|
return $this->results; |
459
|
|
|
} catch (CouldNotConnectToHost $exception) { |
460
|
|
|
throw new ElasticSearchConnectionErrorException("Could not connect to repository server."); |
461
|
|
|
} catch (CouldNotResolveHostException $exception) { |
462
|
|
|
throw new ElasticSearchConnectionErrorException("Could not connect to repository server."); |
463
|
|
|
} |
464
|
|
|
} |
465
|
|
|
|
466
|
|
|
/** |
467
|
|
|
* Get the results |
468
|
|
|
* @return mixed |
469
|
|
|
*/ |
470
|
|
|
public function getResults() |
471
|
|
|
{ |
472
|
|
|
// return results from the last search request |
473
|
|
|
return $this->results; |
474
|
|
|
} |
475
|
|
|
} |
476
|
|
|
|
In PHP, under loose comparison (like
==
, or!=
, orswitch
conditions), values of different types might be equal.For
integer
values, zero is a special case, in particular the following results might be unexpected: