|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
/** |
|
4
|
|
|
* (c) Kitodo. Key to digital objects e.V. <[email protected]> |
|
5
|
|
|
* |
|
6
|
|
|
* This file is part of the Kitodo and TYPO3 projects. |
|
7
|
|
|
* |
|
8
|
|
|
* @license GNU General Public License version 3 or later. |
|
9
|
|
|
* For the full copyright and license information, please read the |
|
10
|
|
|
* LICENSE.txt file that was distributed with this source code. |
|
11
|
|
|
*/ |
|
12
|
|
|
|
|
13
|
|
|
namespace Kitodo\Dlf\Common; |
|
14
|
|
|
|
|
15
|
|
|
use Kitodo\Dlf\Common\Solr\Solr; |
|
16
|
|
|
use Kitodo\Dlf\Domain\Repository\DocumentRepository; |
|
17
|
|
|
use Kitodo\Dlf\Domain\Model\Document; |
|
18
|
|
|
use Kitodo\Dlf\Validation\DocumentValidator; |
|
19
|
|
|
use Solarium\Core\Query\DocumentInterface; |
|
20
|
|
|
use Solarium\QueryType\Update\Query\Query; |
|
21
|
|
|
use Symfony\Component\Console\Input\InputInterface; |
|
22
|
|
|
use TYPO3\CMS\Core\Configuration\ExtensionConfiguration; |
|
23
|
|
|
use TYPO3\CMS\Core\Database\ConnectionPool; |
|
24
|
|
|
use TYPO3\CMS\Core\Messaging\FlashMessage; |
|
25
|
|
|
use TYPO3\CMS\Core\Utility\GeneralUtility; |
|
26
|
|
|
use TYPO3\CMS\Core\Utility\MathUtility; |
|
27
|
|
|
use TYPO3\CMS\Core\Core\Environment; |
|
28
|
|
|
|
|
29
|
|
|
/** |
|
30
|
|
|
* Indexer class for the 'dlf' extension |
|
31
|
|
|
* |
|
32
|
|
|
* @package TYPO3 |
|
33
|
|
|
* @subpackage dlf |
|
34
|
|
|
* |
|
35
|
|
|
* @access public |
|
36
|
|
|
*/ |
|
37
|
|
|
class Indexer |
|
38
|
|
|
{ |
|
39
|
|
|
/** |
|
40
|
|
|
* @access public |
|
41
|
|
|
* @static |
|
42
|
|
|
* @var string The extension key |
|
43
|
|
|
*/ |
|
44
|
|
|
public static string $extKey = 'dlf'; |
|
45
|
|
|
|
|
46
|
|
|
/** |
|
47
|
|
|
* @access protected |
|
48
|
|
|
* @static |
|
49
|
|
|
* @var array Array of metadata fields' configuration |
|
50
|
|
|
* |
|
51
|
|
|
* @see loadIndexConf() |
|
52
|
|
|
*/ |
|
53
|
|
|
protected static array $fields = [ |
|
54
|
|
|
'autocomplete' => [], |
|
55
|
|
|
'facets' => [], |
|
56
|
|
|
'sortables' => [], |
|
57
|
|
|
'indexed' => [], |
|
58
|
|
|
'stored' => [], |
|
59
|
|
|
'tokenized' => [] |
|
60
|
|
|
]; |
|
61
|
|
|
|
|
62
|
|
|
/** |
|
63
|
|
|
* @access protected |
|
64
|
|
|
* @static |
|
65
|
|
|
* @var bool Is the index configuration loaded? |
|
66
|
|
|
* |
|
67
|
|
|
* @see $fields |
|
68
|
|
|
*/ |
|
69
|
|
|
protected static bool $fieldsLoaded = false; |
|
70
|
|
|
|
|
71
|
|
|
/** |
|
72
|
|
|
* @access protected |
|
73
|
|
|
* @static |
|
74
|
|
|
* @var array List of already processed documents |
|
75
|
|
|
*/ |
|
76
|
|
|
protected static array $processedDocs = []; |
|
77
|
|
|
|
|
78
|
|
|
/** |
|
79
|
|
|
* @access protected |
|
80
|
|
|
* @static |
|
81
|
|
|
* @var Solr Instance of Solr class |
|
82
|
|
|
*/ |
|
83
|
|
|
protected static Solr $solr; |
|
84
|
|
|
|
|
85
|
|
|
/** |
|
86
|
|
|
* Insert given document into Solr index |
|
87
|
|
|
* |
|
88
|
|
|
* @access public |
|
89
|
|
|
* |
|
90
|
|
|
* @static |
|
91
|
|
|
* |
|
92
|
|
|
* @param Document $document The document to add |
|
93
|
|
|
* @param DocumentRepository $documentRepository The document repository for search of parent |
|
94
|
|
|
* @param bool $softCommit If true, documents are just added by a soft commit to the index |
|
95
|
|
|
* |
|
96
|
|
|
* @return bool true on success or false on failure |
|
97
|
|
|
*/ |
|
98
|
|
|
public static function add(Document $document, DocumentRepository $documentRepository, bool $softCommit = false): bool |
|
99
|
|
|
{ |
|
100
|
|
|
if (in_array($document->getUid(), self::$processedDocs)) { |
|
101
|
|
|
return true; |
|
102
|
|
|
} elseif (self::solrConnect($document->getSolrcore(), $document->getPid())) { |
|
|
|
|
|
|
103
|
|
|
$success = true; |
|
104
|
|
|
Helper::getLanguageService()->includeLLFile('EXT:dlf/Resources/Private/Language/locallang_be.xlf'); |
|
105
|
|
|
// Handle multi-volume documents. |
|
106
|
|
|
$parentId = $document->getPartof(); |
|
107
|
|
|
if ($parentId) { |
|
108
|
|
|
// get parent document |
|
109
|
|
|
$parent = $documentRepository->findByUid($parentId); |
|
110
|
|
|
if ($parent) { |
|
111
|
|
|
// get XML document of parent |
|
112
|
|
|
$doc = AbstractDocument::getInstance($parent->getLocation(), ['storagePid' => $parent->getPid()], true); |
|
113
|
|
|
if ($doc !== null) { |
|
114
|
|
|
$parent->setCurrentDocument($doc); |
|
115
|
|
|
$success = self::add($parent, $documentRepository); |
|
116
|
|
|
} else { |
|
117
|
|
|
Helper::log('Could not load parent document with UID ' . $document->getCurrentDocument()->parentId, LOG_SEVERITY_ERROR); |
|
118
|
|
|
return false; |
|
119
|
|
|
} |
|
120
|
|
|
} |
|
121
|
|
|
} |
|
122
|
|
|
try { |
|
123
|
|
|
// Add document to list of processed documents. |
|
124
|
|
|
self::$processedDocs[] = $document->getUid(); |
|
125
|
|
|
// Delete old Solr documents. |
|
126
|
|
|
self::deleteDocument('uid', (string) $document->getUid()); |
|
127
|
|
|
|
|
128
|
|
|
// Index every logical unit as separate Solr document. |
|
129
|
|
|
foreach ($document->getCurrentDocument()->tableOfContents as $logicalUnit) { |
|
130
|
|
|
if ($success) { |
|
131
|
|
|
$success = self::processLogical($document, $logicalUnit); |
|
132
|
|
|
} else { |
|
133
|
|
|
break; |
|
134
|
|
|
} |
|
135
|
|
|
} |
|
136
|
|
|
// Index full text files if available. |
|
137
|
|
|
if ($document->getCurrentDocument()->hasFulltext) { |
|
138
|
|
|
foreach ($document->getCurrentDocument()->physicalStructure as $pageNumber => $xmlId) { |
|
139
|
|
|
if ($success) { |
|
140
|
|
|
$success = self::processPhysical($document, $pageNumber, $document->getCurrentDocument()->physicalStructureInfo[$xmlId]); |
|
141
|
|
|
} else { |
|
142
|
|
|
break; |
|
143
|
|
|
} |
|
144
|
|
|
} |
|
145
|
|
|
} |
|
146
|
|
|
// Commit all changes. |
|
147
|
|
|
$updateQuery = self::$solr->service->createUpdate(); |
|
148
|
|
|
$updateQuery->addCommit($softCommit); |
|
149
|
|
|
self::$solr->service->update($updateQuery); |
|
150
|
|
|
|
|
151
|
|
|
if (!(Environment::isCli())) { |
|
152
|
|
|
if ($success) { |
|
153
|
|
|
self::addMessage( |
|
154
|
|
|
sprintf(Helper::getLanguageService()->getLL('flash.documentIndexed'), $document->getTitle(), $document->getUid()), |
|
155
|
|
|
'flash.done', |
|
156
|
|
|
FlashMessage::OK |
|
157
|
|
|
); |
|
158
|
|
|
} else { |
|
159
|
|
|
self::addErrorMessage(sprintf(Helper::getLanguageService()->getLL('flash.documentNotIndexed'), $document->getTitle(), $document->getUid())); |
|
160
|
|
|
} |
|
161
|
|
|
} |
|
162
|
|
|
return $success; |
|
163
|
|
|
} catch (\Exception $e) { |
|
164
|
|
|
self::handleException($e->getMessage()); |
|
165
|
|
|
return false; |
|
166
|
|
|
} |
|
167
|
|
|
} else { |
|
168
|
|
|
if (!(Environment::isCli())) { |
|
169
|
|
|
self::addMessage( |
|
170
|
|
|
Helper::getLanguageService()->getLL('flash.solrNoConnection'), |
|
171
|
|
|
'flash.warning', |
|
172
|
|
|
FlashMessage::WARNING |
|
173
|
|
|
); |
|
174
|
|
|
} |
|
175
|
|
|
Helper::log('Could not connect to Apache Solr server', LOG_SEVERITY_ERROR); |
|
176
|
|
|
return false; |
|
177
|
|
|
} |
|
178
|
|
|
} |
|
179
|
|
|
|
|
180
|
|
|
/** |
|
181
|
|
|
* Delete document from Solr index |
|
182
|
|
|
* |
|
183
|
|
|
* @access public |
|
184
|
|
|
* |
|
185
|
|
|
* @static |
|
186
|
|
|
* |
|
187
|
|
|
* @param InputInterface $input The input parameters |
|
188
|
|
|
* @param string $field by which document should be removed |
|
189
|
|
|
* @param int $solrCoreUid UID of the SolrCore |
|
190
|
|
|
* @param bool $softCommit If true, documents are just deleted from the index by a soft commit |
|
191
|
|
|
* |
|
192
|
|
|
* @return bool true on success or false on failure |
|
193
|
|
|
*/ |
|
194
|
|
|
public static function delete(InputInterface $input, string $field, int $solrCoreUid, bool $softCommit = false): bool |
|
195
|
|
|
{ |
|
196
|
|
|
if (self::solrConnect($solrCoreUid, $input->getOption('pid'))) { |
|
197
|
|
|
try { |
|
198
|
|
|
self::deleteDocument($field, $input->getOption('doc'), $softCommit); |
|
199
|
|
|
return true; |
|
200
|
|
|
} catch (\Exception $e) { |
|
201
|
|
|
if (!(Environment::isCli())) { |
|
202
|
|
|
Helper::addMessage( |
|
203
|
|
|
Helper::getLanguageService()->getLL('flash.solrException') . ' ' . htmlspecialchars($e->getMessage()), |
|
204
|
|
|
Helper::getLanguageService()->getLL('flash.error'), |
|
205
|
|
|
FlashMessage::ERROR, |
|
206
|
|
|
true, |
|
207
|
|
|
'core.template.flashMessages' |
|
208
|
|
|
); |
|
209
|
|
|
} |
|
210
|
|
|
Helper::log('Apache Solr threw exception: "' . $e->getMessage() . '"', LOG_SEVERITY_ERROR); |
|
211
|
|
|
return false; |
|
212
|
|
|
} |
|
213
|
|
|
} |
|
214
|
|
|
|
|
215
|
|
|
Helper::log('Document not deleted from SOLR - problem with the connection to the SOLR core ' . $solrCoreUid, LOG_SEVERITY_ERROR); |
|
216
|
|
|
return false; |
|
217
|
|
|
} |
|
218
|
|
|
|
|
219
|
|
|
/** |
|
220
|
|
|
* Returns the dynamic index field name for the given metadata field. |
|
221
|
|
|
* |
|
222
|
|
|
* @access public |
|
223
|
|
|
* |
|
224
|
|
|
* @static |
|
225
|
|
|
* |
|
226
|
|
|
* @param string $indexName The metadata field's name in database |
|
227
|
|
|
* @param int $pid UID of the configuration page |
|
228
|
|
|
* |
|
229
|
|
|
* @return string The field's dynamic index name |
|
230
|
|
|
*/ |
|
231
|
|
|
public static function getIndexFieldName(string $indexName, int $pid = 0): string |
|
232
|
|
|
{ |
|
233
|
|
|
// Sanitize input. |
|
234
|
|
|
$pid = max((int) $pid, 0); |
|
235
|
|
|
if (!$pid) { |
|
236
|
|
|
Helper::log('Invalid PID ' . $pid . ' for metadata configuration', LOG_SEVERITY_ERROR); |
|
237
|
|
|
return ''; |
|
238
|
|
|
} |
|
239
|
|
|
// Load metadata configuration. |
|
240
|
|
|
self::loadIndexConf($pid); |
|
241
|
|
|
// Build field's suffix. |
|
242
|
|
|
$suffix = (in_array($indexName, self::$fields['tokenized']) ? 't' : 'u'); |
|
243
|
|
|
$suffix .= (in_array($indexName, self::$fields['stored']) ? 's' : 'u'); |
|
244
|
|
|
$suffix .= (in_array($indexName, self::$fields['indexed']) ? 'i' : 'u'); |
|
245
|
|
|
$indexName .= '_' . $suffix; |
|
246
|
|
|
return $indexName; |
|
247
|
|
|
} |
|
248
|
|
|
|
|
249
|
|
|
/** |
|
250
|
|
|
* Load indexing configuration |
|
251
|
|
|
* |
|
252
|
|
|
* @access protected |
|
253
|
|
|
* |
|
254
|
|
|
* @static |
|
255
|
|
|
* |
|
256
|
|
|
* @param int $pid The configuration page's UID |
|
257
|
|
|
* |
|
258
|
|
|
* @return void |
|
259
|
|
|
*/ |
|
260
|
|
|
protected static function loadIndexConf(int $pid): void |
|
261
|
|
|
{ |
|
262
|
|
|
if (!self::$fieldsLoaded) { |
|
263
|
|
|
$queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class) |
|
264
|
|
|
->getQueryBuilderForTable('tx_dlf_metadata'); |
|
265
|
|
|
|
|
266
|
|
|
// Get the metadata indexing options. |
|
267
|
|
|
$result = $queryBuilder |
|
268
|
|
|
->select( |
|
269
|
|
|
'tx_dlf_metadata.index_name AS index_name', |
|
270
|
|
|
'tx_dlf_metadata.index_tokenized AS index_tokenized', |
|
271
|
|
|
'tx_dlf_metadata.index_stored AS index_stored', |
|
272
|
|
|
'tx_dlf_metadata.index_indexed AS index_indexed', |
|
273
|
|
|
'tx_dlf_metadata.is_sortable AS is_sortable', |
|
274
|
|
|
'tx_dlf_metadata.is_facet AS is_facet', |
|
275
|
|
|
'tx_dlf_metadata.is_listed AS is_listed', |
|
276
|
|
|
'tx_dlf_metadata.index_autocomplete AS index_autocomplete', |
|
277
|
|
|
'tx_dlf_metadata.index_boost AS index_boost' |
|
278
|
|
|
) |
|
279
|
|
|
->from('tx_dlf_metadata') |
|
280
|
|
|
->where( |
|
281
|
|
|
$queryBuilder->expr()->eq('tx_dlf_metadata.pid', (int) $pid), |
|
282
|
|
|
Helper::whereExpression('tx_dlf_metadata') |
|
283
|
|
|
) |
|
284
|
|
|
->execute(); |
|
285
|
|
|
|
|
286
|
|
|
while ($indexing = $result->fetchAssociative()) { |
|
287
|
|
|
if ($indexing['index_tokenized']) { |
|
288
|
|
|
self::$fields['tokenized'][] = $indexing['index_name']; |
|
289
|
|
|
} |
|
290
|
|
|
if ( |
|
291
|
|
|
$indexing['index_stored'] |
|
292
|
|
|
|| $indexing['is_listed'] |
|
293
|
|
|
) { |
|
294
|
|
|
self::$fields['stored'][] = $indexing['index_name']; |
|
295
|
|
|
} |
|
296
|
|
|
if ( |
|
297
|
|
|
$indexing['index_indexed'] |
|
298
|
|
|
|| $indexing['index_autocomplete'] |
|
299
|
|
|
) { |
|
300
|
|
|
self::$fields['indexed'][] = $indexing['index_name']; |
|
301
|
|
|
} |
|
302
|
|
|
if ($indexing['is_sortable']) { |
|
303
|
|
|
self::$fields['sortables'][] = $indexing['index_name']; |
|
304
|
|
|
} |
|
305
|
|
|
if ($indexing['is_facet']) { |
|
306
|
|
|
self::$fields['facets'][] = $indexing['index_name']; |
|
307
|
|
|
} |
|
308
|
|
|
if ($indexing['index_autocomplete']) { |
|
309
|
|
|
self::$fields['autocomplete'][] = $indexing['index_name']; |
|
310
|
|
|
} |
|
311
|
|
|
} |
|
312
|
|
|
self::$fieldsLoaded = true; |
|
313
|
|
|
} |
|
314
|
|
|
} |
|
315
|
|
|
|
|
316
|
|
|
/** |
|
317
|
|
|
* Processes a logical unit (and its children) for the Solr index |
|
318
|
|
|
* |
|
319
|
|
|
* @access protected |
|
320
|
|
|
* |
|
321
|
|
|
* @static |
|
322
|
|
|
* |
|
323
|
|
|
* @param Document $document The METS document |
|
324
|
|
|
* @param array $logicalUnit Array of the logical unit to process |
|
325
|
|
|
* |
|
326
|
|
|
* @return bool true on success or false on failure |
|
327
|
|
|
*/ |
|
328
|
|
|
protected static function processLogical(Document $document, array $logicalUnit): bool |
|
329
|
|
|
{ |
|
330
|
|
|
$success = true; |
|
331
|
|
|
$doc = $document->getCurrentDocument(); |
|
332
|
|
|
$doc->cPid = $document->getPid(); |
|
333
|
|
|
// Get metadata for logical unit. |
|
334
|
|
|
$metadata = $doc->metadataArray[$logicalUnit['id']]; |
|
335
|
|
|
if (!empty($metadata)) { |
|
336
|
|
|
$extConf = GeneralUtility::makeInstance(ExtensionConfiguration::class)->get(self::$extKey, 'general'); |
|
337
|
|
|
$validator = new DocumentValidator($metadata, explode(',', $extConf['requiredMetadataFields'])); |
|
338
|
|
|
|
|
339
|
|
|
if ($validator->hasAllMandatoryMetadataFields()) { |
|
340
|
|
|
$metadata['author'] = self::removeAppendsFromAuthor($metadata['author']); |
|
341
|
|
|
// set Owner if available |
|
342
|
|
|
if ($document->getOwner()) { |
|
343
|
|
|
$metadata['owner'][0] = $document->getOwner()->getIndexName(); |
|
344
|
|
|
} |
|
345
|
|
|
// Create new Solr document. |
|
346
|
|
|
$updateQuery = self::$solr->service->createUpdate(); |
|
347
|
|
|
$solrDoc = self::getSolrDocument($updateQuery, $document, $logicalUnit); |
|
348
|
|
|
if (MathUtility::canBeInterpretedAsInteger($logicalUnit['points'])) { |
|
349
|
|
|
$solrDoc->setField('page', $logicalUnit['points']); |
|
|
|
|
|
|
350
|
|
|
} |
|
351
|
|
|
if ($logicalUnit['id'] == $doc->toplevelId) { |
|
352
|
|
|
$solrDoc->setField('thumbnail', $doc->thumbnail); |
|
353
|
|
|
} elseif (!empty($logicalUnit['thumbnailId'])) { |
|
354
|
|
|
$solrDoc->setField('thumbnail', $doc->getFileLocation($logicalUnit['thumbnailId'])); |
|
355
|
|
|
} |
|
356
|
|
|
// There can be only one toplevel unit per UID, independently of backend configuration |
|
357
|
|
|
$solrDoc->setField('toplevel', $logicalUnit['id'] == $doc->toplevelId ? true : false); |
|
358
|
|
|
$solrDoc->setField('title', $metadata['title'][0]); |
|
359
|
|
|
$solrDoc->setField('volume', $metadata['volume'][0]); |
|
360
|
|
|
// verify date formatting |
|
361
|
|
|
if(strtotime($metadata['date'][0])) { |
|
362
|
|
|
$solrDoc->setField('date', self::getFormattedDate($metadata['date'][0])); |
|
363
|
|
|
} |
|
364
|
|
|
$solrDoc->setField('record_id', $metadata['record_id'][0]); |
|
365
|
|
|
$solrDoc->setField('purl', $metadata['purl'][0]); |
|
366
|
|
|
$solrDoc->setField('location', $document->getLocation()); |
|
367
|
|
|
$solrDoc->setField('urn', $metadata['urn']); |
|
368
|
|
|
$solrDoc->setField('license', $metadata['license']); |
|
369
|
|
|
$solrDoc->setField('terms', $metadata['terms']); |
|
370
|
|
|
$solrDoc->setField('restrictions', $metadata['restrictions']); |
|
371
|
|
|
$coordinates = json_decode($metadata['coordinates'][0]); |
|
372
|
|
|
if (is_object($coordinates)) { |
|
373
|
|
|
$solrDoc->setField('geom', json_encode($coordinates->features[0])); |
|
374
|
|
|
} |
|
375
|
|
|
$autocomplete = self::processMetadata($document, $metadata, $solrDoc); |
|
376
|
|
|
// Add autocomplete values to index. |
|
377
|
|
|
if (!empty($autocomplete)) { |
|
378
|
|
|
$solrDoc->setField('autocomplete', $autocomplete); |
|
379
|
|
|
} |
|
380
|
|
|
// Add collection information to logical sub-elements if applicable. |
|
381
|
|
|
if ( |
|
382
|
|
|
in_array('collection', self::$fields['facets']) |
|
383
|
|
|
&& empty($metadata['collection']) |
|
384
|
|
|
&& !empty($doc->metadataArray[$doc->toplevelId]['collection']) |
|
385
|
|
|
) { |
|
386
|
|
|
$solrDoc->setField('collection_faceting', $doc->metadataArray[$doc->toplevelId]['collection']); |
|
387
|
|
|
} |
|
388
|
|
|
try { |
|
389
|
|
|
$updateQuery->addDocument($solrDoc); |
|
390
|
|
|
self::$solr->service->update($updateQuery); |
|
391
|
|
|
} catch (\Exception $e) { |
|
392
|
|
|
self::handleException($e->getMessage()); |
|
393
|
|
|
return false; |
|
394
|
|
|
} |
|
395
|
|
|
} else { |
|
396
|
|
|
Helper::log('Tip: If "record_id" field is missing then there is possibility that METS file still contains it but with the wrong source type attribute in "recordIdentifier" element', LOG_SEVERITY_NOTICE); |
|
397
|
|
|
return false; |
|
398
|
|
|
} |
|
399
|
|
|
} |
|
400
|
|
|
// Check for child elements... |
|
401
|
|
|
if (!empty($logicalUnit['children'])) { |
|
402
|
|
|
foreach ($logicalUnit['children'] as $child) { |
|
403
|
|
|
if ($success) { |
|
404
|
|
|
// ...and process them, too. |
|
405
|
|
|
$success = self::processLogical($document, $child); |
|
406
|
|
|
} else { |
|
407
|
|
|
break; |
|
408
|
|
|
} |
|
409
|
|
|
} |
|
410
|
|
|
} |
|
411
|
|
|
return $success; |
|
412
|
|
|
} |
|
413
|
|
|
|
|
414
|
|
|
/** |
|
415
|
|
|
* Processes a physical unit for the Solr index |
|
416
|
|
|
* |
|
417
|
|
|
* @access protected |
|
418
|
|
|
* |
|
419
|
|
|
* @static |
|
420
|
|
|
* |
|
421
|
|
|
* @param Document $document The METS document |
|
422
|
|
|
* @param int $page The page number |
|
423
|
|
|
* @param array $physicalUnit Array of the physical unit to process |
|
424
|
|
|
* |
|
425
|
|
|
* @return bool true on success or false on failure |
|
426
|
|
|
*/ |
|
427
|
|
|
protected static function processPhysical(Document $document, int $page, array $physicalUnit): bool |
|
428
|
|
|
{ |
|
429
|
|
|
$doc = $document->getCurrentDocument(); |
|
430
|
|
|
$doc->cPid = $document->getPid(); |
|
431
|
|
|
if ($doc->hasFulltext && $fullText = $doc->getFullText($physicalUnit['id'])) { |
|
432
|
|
|
// Read extension configuration. |
|
433
|
|
|
$extConf = GeneralUtility::makeInstance(ExtensionConfiguration::class)->get(self::$extKey, 'files'); |
|
434
|
|
|
// Create new Solr document. |
|
435
|
|
|
$updateQuery = self::$solr->service->createUpdate(); |
|
436
|
|
|
$solrDoc = self::getSolrDocument($updateQuery, $document, $physicalUnit, $fullText); |
|
437
|
|
|
$solrDoc->setField('page', $page); |
|
438
|
|
|
$fileGrpsThumb = GeneralUtility::trimExplode(',', $extConf['fileGrpThumbs']); |
|
439
|
|
|
while ($fileGrpThumb = array_shift($fileGrpsThumb)) { |
|
440
|
|
|
if (!empty($physicalUnit['files'][$fileGrpThumb])) { |
|
441
|
|
|
$solrDoc->setField('thumbnail', $doc->getFileLocation($physicalUnit['files'][$fileGrpThumb])); |
|
442
|
|
|
break; |
|
443
|
|
|
} |
|
444
|
|
|
} |
|
445
|
|
|
$solrDoc->setField('toplevel', false); |
|
446
|
|
|
$solrDoc->setField('type', $physicalUnit['type']); |
|
447
|
|
|
$solrDoc->setField('collection', $doc->metadataArray[$doc->toplevelId]['collection']); |
|
448
|
|
|
$solrDoc->setField('location', $document->getLocation()); |
|
449
|
|
|
|
|
450
|
|
|
$solrDoc->setField('fulltext', $fullText); |
|
451
|
|
|
if (is_array($doc->metadataArray[$doc->toplevelId])) { |
|
452
|
|
|
self::addFaceting($doc, $solrDoc); |
|
453
|
|
|
} |
|
454
|
|
|
// Add collection information to physical sub-elements if applicable. |
|
455
|
|
|
if ( |
|
456
|
|
|
in_array('collection', self::$fields['facets']) |
|
457
|
|
|
&& !empty($doc->metadataArray[$doc->toplevelId]['collection']) |
|
458
|
|
|
) { |
|
459
|
|
|
$solrDoc->setField('collection_faceting', $doc->metadataArray[$doc->toplevelId]['collection']); |
|
460
|
|
|
} |
|
461
|
|
|
try { |
|
462
|
|
|
$updateQuery->addDocument($solrDoc); |
|
463
|
|
|
self::$solr->service->update($updateQuery); |
|
464
|
|
|
} catch (\Exception $e) { |
|
465
|
|
|
self::handleException($e->getMessage()); |
|
466
|
|
|
return false; |
|
467
|
|
|
} |
|
468
|
|
|
} |
|
469
|
|
|
return true; |
|
470
|
|
|
} |
|
471
|
|
|
|
|
472
|
|
|
/** |
|
473
|
|
|
* Connects to Solr server. |
|
474
|
|
|
* |
|
475
|
|
|
* @access protected |
|
476
|
|
|
* |
|
477
|
|
|
* @static |
|
478
|
|
|
* |
|
479
|
|
|
* @param int $core UID of the Solr core |
|
480
|
|
|
* @param int $pid UID of the configuration page |
|
481
|
|
|
* |
|
482
|
|
|
* @return bool true on success or false on failure |
|
483
|
|
|
*/ |
|
484
|
|
|
protected static function solrConnect(int $core, int $pid = 0): bool |
|
485
|
|
|
{ |
|
486
|
|
|
// Get Solr instance. |
|
487
|
|
|
$solr = Solr::getInstance($core); |
|
488
|
|
|
// Connect to Solr server. |
|
489
|
|
|
if ($solr->ready) { |
|
490
|
|
|
self::$solr = $solr; |
|
491
|
|
|
// Load indexing configuration if needed. |
|
492
|
|
|
if ($pid) { |
|
493
|
|
|
self::loadIndexConf($pid); |
|
494
|
|
|
} |
|
495
|
|
|
return true; |
|
496
|
|
|
} |
|
497
|
|
|
return false; |
|
498
|
|
|
} |
|
499
|
|
|
|
|
500
|
|
|
/** |
|
501
|
|
|
* Process metadata: add facets, sortable fields and create autocomplete array. |
|
502
|
|
|
* |
|
503
|
|
|
* @static |
|
504
|
|
|
* |
|
505
|
|
|
* @access private |
|
506
|
|
|
* |
|
507
|
|
|
* @param Document $document |
|
508
|
|
|
* @param array $metadata |
|
509
|
|
|
* @param DocumentInterface &$solrDoc |
|
510
|
|
|
* |
|
511
|
|
|
* @return array empty array or autocomplete values |
|
512
|
|
|
*/ |
|
513
|
|
|
private static function processMetadata($document, $metadata, &$solrDoc): array |
|
514
|
|
|
{ |
|
515
|
|
|
$autocomplete = []; |
|
516
|
|
|
foreach ($metadata as $indexName => $data) { |
|
517
|
|
|
// TODO: Include also subentries if available. |
|
518
|
|
|
if ( |
|
519
|
|
|
!empty($data) |
|
520
|
|
|
&& substr($indexName, -8) !== '_sorting' |
|
521
|
|
|
) { |
|
522
|
|
|
$solrDoc->setField(self::getIndexFieldName($indexName, $document->getPid()), $data); |
|
|
|
|
|
|
523
|
|
|
if (in_array($indexName, self::$fields['sortables'])) { |
|
524
|
|
|
// Add sortable fields to index. |
|
525
|
|
|
$solrDoc->setField($indexName . '_sorting', $metadata[$indexName . '_sorting'][0]); |
|
526
|
|
|
} |
|
527
|
|
|
if (in_array($indexName, self::$fields['facets'])) { |
|
528
|
|
|
// Add facets to index. |
|
529
|
|
|
$solrDoc->setField($indexName . '_faceting', $data); |
|
530
|
|
|
} |
|
531
|
|
|
if (in_array($indexName, self::$fields['autocomplete'])) { |
|
532
|
|
|
$autocomplete = array_merge($autocomplete, $data); |
|
533
|
|
|
} |
|
534
|
|
|
} |
|
535
|
|
|
} |
|
536
|
|
|
return $autocomplete; |
|
537
|
|
|
} |
|
538
|
|
|
|
|
539
|
|
|
/** |
|
540
|
|
|
* Add faceting information to physical sub-elements if applicable. |
|
541
|
|
|
* |
|
542
|
|
|
* @static |
|
543
|
|
|
* |
|
544
|
|
|
* @access private |
|
545
|
|
|
* |
|
546
|
|
|
* @param AbstractDocument $doc |
|
547
|
|
|
* @param DocumentInterface &$solrDoc |
|
548
|
|
|
* |
|
549
|
|
|
* @return void |
|
550
|
|
|
*/ |
|
551
|
|
|
private static function addFaceting($doc, &$solrDoc): void |
|
552
|
|
|
{ |
|
553
|
|
|
// TODO: Include also subentries if available. |
|
554
|
|
|
foreach ($doc->metadataArray[$doc->toplevelId] as $indexName => $data) { |
|
555
|
|
|
if ( |
|
556
|
|
|
!empty($data) |
|
557
|
|
|
&& substr($indexName, -8) !== '_sorting' |
|
558
|
|
|
) { |
|
559
|
|
|
|
|
560
|
|
|
if (in_array($indexName, self::$fields['facets'])) { |
|
561
|
|
|
// Remove appended "valueURI" from authors' names for indexing. |
|
562
|
|
|
if ($indexName == 'author') { |
|
563
|
|
|
$data = self::removeAppendsFromAuthor($data); |
|
564
|
|
|
} |
|
565
|
|
|
// Add facets to index. |
|
566
|
|
|
$solrDoc->setField($indexName . '_faceting', $data); |
|
567
|
|
|
} |
|
568
|
|
|
} |
|
569
|
|
|
// Add sorting information to physical sub-elements if applicable. |
|
570
|
|
|
if ( |
|
571
|
|
|
!empty($data) |
|
572
|
|
|
&& substr($indexName, -8) == '_sorting' |
|
573
|
|
|
) { |
|
574
|
|
|
$solrDoc->setField($indexName, $doc->metadataArray[$doc->toplevelId][$indexName]); |
|
575
|
|
|
} |
|
576
|
|
|
} |
|
577
|
|
|
} |
|
578
|
|
|
|
|
579
|
|
|
/** |
|
580
|
|
|
* Delete document from SOLR by given field and value. |
|
581
|
|
|
* |
|
582
|
|
|
* @access private |
|
583
|
|
|
* |
|
584
|
|
|
* @static |
|
585
|
|
|
* |
|
586
|
|
|
* @param string $field by which document should be removed |
|
587
|
|
|
* @param string $value of the field by which document should be removed |
|
588
|
|
|
* @param bool $softCommit If true, documents are just deleted from the index by a soft commit |
|
589
|
|
|
* |
|
590
|
|
|
* @return void |
|
591
|
|
|
*/ |
|
592
|
|
|
private static function deleteDocument(string $field, string $value, bool $softCommit = false): void |
|
593
|
|
|
{ |
|
594
|
|
|
$update = self::$solr->service->createUpdate(); |
|
595
|
|
|
$query = ""; |
|
596
|
|
|
if ($field == 'uid' || $field == 'partof') { |
|
597
|
|
|
$query = $field . ':' . $value; |
|
598
|
|
|
} else { |
|
599
|
|
|
$query = $field . ':"' . $value . '"'; |
|
600
|
|
|
} |
|
601
|
|
|
$update->addDeleteQuery($query); |
|
602
|
|
|
$update->addCommit($softCommit); |
|
603
|
|
|
self::$solr->service->update($update); |
|
604
|
|
|
} |
|
605
|
|
|
|
|
606
|
|
|
/** |
|
607
|
|
|
* Get SOLR document with set standard fields (identical for logical and physical unit) |
|
608
|
|
|
* |
|
609
|
|
|
* @access private |
|
610
|
|
|
* |
|
611
|
|
|
* @static |
|
612
|
|
|
* |
|
613
|
|
|
* @param Query $updateQuery solarium query |
|
614
|
|
|
* @param Document $document The METS document |
|
615
|
|
|
* @param array $unit Array of the logical or physical unit to process |
|
616
|
|
|
* @param string $fullText Text containing full text for indexing |
|
617
|
|
|
* |
|
618
|
|
|
* @return DocumentInterface |
|
619
|
|
|
*/ |
|
620
|
|
|
private static function getSolrDocument(Query $updateQuery, Document $document, array $unit, string $fullText = ''): DocumentInterface |
|
621
|
|
|
{ |
|
622
|
|
|
$solrDoc = $updateQuery->createDocument(); |
|
623
|
|
|
// Create unique identifier from document's UID and unit's XML ID. |
|
624
|
|
|
$solrDoc->setField('id', $document->getUid() . $unit['id']); |
|
625
|
|
|
$solrDoc->setField('uid', $document->getUid()); |
|
626
|
|
|
$solrDoc->setField('pid', $document->getPid()); |
|
627
|
|
|
$solrDoc->setField('partof', $document->getPartof()); |
|
628
|
|
|
$solrDoc->setField('root', $document->getCurrentDocument()->rootId); |
|
629
|
|
|
$solrDoc->setField('sid', $unit['id']); |
|
630
|
|
|
$solrDoc->setField('type', $unit['type']); |
|
631
|
|
|
$solrDoc->setField('collection', $document->getCurrentDocument()->metadataArray[$document->getCurrentDocument()->toplevelId]['collection']); |
|
632
|
|
|
$solrDoc->setField('fulltext', $fullText); |
|
633
|
|
|
return $solrDoc; |
|
634
|
|
|
} |
|
635
|
|
|
|
|
636
|
|
|
/** |
|
637
|
|
|
* Get formatted date without alteration. |
|
638
|
|
|
* Possible formats: YYYY or YYYY-MM or YYYY-MM-DD. |
|
639
|
|
|
* |
|
640
|
|
|
* @static |
|
641
|
|
|
* |
|
642
|
|
|
* @access private |
|
643
|
|
|
* |
|
644
|
|
|
* @param string $date |
|
645
|
|
|
* |
|
646
|
|
|
* @return string formatted date YYYY or YYYY-MM or YYYY-MM-DD or empty string |
|
647
|
|
|
*/ |
|
648
|
|
|
private static function getFormattedDate($date): string |
|
649
|
|
|
{ |
|
650
|
|
|
if ( |
|
651
|
|
|
preg_match("/^[\d]{4}$/", $date) |
|
652
|
|
|
|| preg_match("/^[\d]{4}-[\d]{2}$/", $date) |
|
653
|
|
|
|| preg_match("/^[\d]{4}-[\d]{2}-[\d]{2}$/", $date) |
|
654
|
|
|
) { |
|
655
|
|
|
return $date; |
|
656
|
|
|
// change date YYYYMMDD to YYYY-MM-DD |
|
657
|
|
|
} elseif (preg_match("/^[\d]{8}$/", $date)) { |
|
658
|
|
|
return date("Y-m-d", strtotime($date)); |
|
659
|
|
|
// convert any datetime to proper ISO extended datetime format and timezone for SOLR |
|
660
|
|
|
} elseif (preg_match("/^[0-9]{4}-[0-9]{2}-[0-9]{2}T.*$/", $date)) { |
|
661
|
|
|
return date('Y-m-d\TH:i:s\Z', strtotime($date)); |
|
662
|
|
|
} |
|
663
|
|
|
// date doesn't match any standard |
|
664
|
|
|
return ''; |
|
665
|
|
|
} |
|
666
|
|
|
|
|
667
|
|
|
/** |
|
668
|
|
|
* Remove appended "valueURI" from authors' names for indexing. |
|
669
|
|
|
* |
|
670
|
|
|
* @access private |
|
671
|
|
|
* |
|
672
|
|
|
* @static |
|
673
|
|
|
* |
|
674
|
|
|
* @param array|string $authors Array or string containing author/authors |
|
675
|
|
|
* |
|
676
|
|
|
* @return array|string |
|
677
|
|
|
*/ |
|
678
|
|
|
private static function removeAppendsFromAuthor($authors) |
|
679
|
|
|
{ |
|
680
|
|
|
if (is_array($authors)) { |
|
681
|
|
|
foreach ($authors as $i => $author) { |
|
682
|
|
|
$splitName = explode(pack('C', 31), $author); |
|
683
|
|
|
$authors[$i] = $splitName[0]; |
|
684
|
|
|
} |
|
685
|
|
|
} |
|
686
|
|
|
return $authors; |
|
687
|
|
|
} |
|
688
|
|
|
|
|
689
|
|
|
/** |
|
690
|
|
|
* Handle exception. |
|
691
|
|
|
* |
|
692
|
|
|
* @static |
|
693
|
|
|
* |
|
694
|
|
|
* @access private |
|
695
|
|
|
* |
|
696
|
|
|
* @param string $errorMessage |
|
697
|
|
|
* |
|
698
|
|
|
* @return void |
|
699
|
|
|
*/ |
|
700
|
|
|
private static function handleException(string $errorMessage): void |
|
701
|
|
|
{ |
|
702
|
|
|
if (!(Environment::isCli())) { |
|
703
|
|
|
self::addErrorMessage(Helper::getLanguageService()->getLL('flash.solrException') . '<br />' . htmlspecialchars($errorMessage)); |
|
704
|
|
|
} |
|
705
|
|
|
Helper::log('Apache Solr threw exception: "' . $errorMessage . '"', LOG_SEVERITY_ERROR); |
|
706
|
|
|
} |
|
707
|
|
|
|
|
708
|
|
|
/** |
|
709
|
|
|
* Add error message only with message content. |
|
710
|
|
|
* |
|
711
|
|
|
* @static |
|
712
|
|
|
* |
|
713
|
|
|
* @access private |
|
714
|
|
|
* |
|
715
|
|
|
* @param string $message |
|
716
|
|
|
* |
|
717
|
|
|
* @return void |
|
718
|
|
|
*/ |
|
719
|
|
|
private static function addErrorMessage(string $message): void |
|
720
|
|
|
{ |
|
721
|
|
|
self::addMessage( |
|
722
|
|
|
$message, |
|
723
|
|
|
'flash.error', |
|
724
|
|
|
FlashMessage::ERROR |
|
725
|
|
|
); |
|
726
|
|
|
} |
|
727
|
|
|
|
|
728
|
|
|
/** |
|
729
|
|
|
* Add message only with changeable parameters. |
|
730
|
|
|
* |
|
731
|
|
|
* @static |
|
732
|
|
|
* |
|
733
|
|
|
* @access private |
|
734
|
|
|
* |
|
735
|
|
|
* @param string $message |
|
736
|
|
|
* @param string $type |
|
737
|
|
|
* @param int $status |
|
738
|
|
|
* |
|
739
|
|
|
* @return void |
|
740
|
|
|
*/ |
|
741
|
|
|
private static function addMessage(string $message, string $type, int $status): void |
|
742
|
|
|
{ |
|
743
|
|
|
Helper::addMessage( |
|
744
|
|
|
$message, |
|
745
|
|
|
Helper::getLanguageService()->getLL($type), |
|
746
|
|
|
$status, |
|
747
|
|
|
true, |
|
748
|
|
|
'core.template.flashMessages' |
|
749
|
|
|
); |
|
750
|
|
|
} |
|
751
|
|
|
|
|
752
|
|
|
/** |
|
753
|
|
|
* Prevent instantiation by hiding the constructor |
|
754
|
|
|
* |
|
755
|
|
|
* @access private |
|
756
|
|
|
* |
|
757
|
|
|
* @return void |
|
758
|
|
|
*/ |
|
759
|
|
|
private function __construct() |
|
760
|
|
|
{ |
|
761
|
|
|
} |
|
762
|
|
|
} |
|
763
|
|
|
|
kitodo /
kitodo-presentation
Loading history...
