|
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 TYPO3\CMS\Core\Configuration\ExtensionConfiguration; |
|
16
|
|
|
use TYPO3\CMS\Core\Database\ConnectionPool; |
|
17
|
|
|
use TYPO3\CMS\Core\Messaging\FlashMessage; |
|
18
|
|
|
use TYPO3\CMS\Core\Utility\GeneralUtility; |
|
19
|
|
|
use TYPO3\CMS\Core\Utility\MathUtility; |
|
20
|
|
|
use Ubl\Iiif\Presentation\Common\Model\Resources\AnnotationContainerInterface; |
|
21
|
|
|
use Ubl\Iiif\Tools\IiifHelper; |
|
22
|
|
|
|
|
23
|
|
|
/** |
|
24
|
|
|
* Indexer class for the 'dlf' extension |
|
25
|
|
|
* |
|
26
|
|
|
* @author Sebastian Meyer <[email protected]> |
|
27
|
|
|
* @package TYPO3 |
|
28
|
|
|
* @subpackage dlf |
|
29
|
|
|
* @access public |
|
30
|
|
|
*/ |
|
31
|
|
|
class Indexer |
|
32
|
|
|
{ |
|
33
|
|
|
/** |
|
34
|
|
|
* The extension key |
|
35
|
|
|
* |
|
36
|
|
|
* @var string |
|
37
|
|
|
* @access public |
|
38
|
|
|
*/ |
|
39
|
|
|
public static $extKey = 'dlf'; |
|
40
|
|
|
|
|
41
|
|
|
/** |
|
42
|
|
|
* Array of metadata fields' configuration |
|
43
|
|
|
* @see loadIndexConf() |
|
44
|
|
|
* |
|
45
|
|
|
* @var array |
|
46
|
|
|
* @access protected |
|
47
|
|
|
*/ |
|
48
|
|
|
protected static $fields = [ |
|
49
|
|
|
'autocomplete' => [], |
|
50
|
|
|
'facets' => [], |
|
51
|
|
|
'sortables' => [], |
|
52
|
|
|
'indexed' => [], |
|
53
|
|
|
'stored' => [], |
|
54
|
|
|
'tokenized' => [], |
|
55
|
|
|
'fieldboost' => [] |
|
56
|
|
|
]; |
|
57
|
|
|
|
|
58
|
|
|
/** |
|
59
|
|
|
* Is the index configuration loaded? |
|
60
|
|
|
* @see $fields |
|
61
|
|
|
* |
|
62
|
|
|
* @var bool |
|
63
|
|
|
* @access protected |
|
64
|
|
|
*/ |
|
65
|
|
|
protected static $fieldsLoaded = false; |
|
66
|
|
|
|
|
67
|
|
|
/** |
|
68
|
|
|
* List of already processed documents |
|
69
|
|
|
* |
|
70
|
|
|
* @var array |
|
71
|
|
|
* @access protected |
|
72
|
|
|
*/ |
|
73
|
|
|
protected static $processedDocs = []; |
|
74
|
|
|
|
|
75
|
|
|
/** |
|
76
|
|
|
* Instance of \Kitodo\Dlf\Common\Solr class |
|
77
|
|
|
* |
|
78
|
|
|
* @var \Kitodo\Dlf\Common\Solr |
|
79
|
|
|
* @access protected |
|
80
|
|
|
*/ |
|
81
|
|
|
protected static $solr; |
|
82
|
|
|
|
|
83
|
|
|
/** |
|
84
|
|
|
* Insert given document into Solr index |
|
85
|
|
|
* |
|
86
|
|
|
* @access public |
|
87
|
|
|
* |
|
88
|
|
|
* @param \Kitodo\Dlf\Common\Document &$doc: The document to add |
|
89
|
|
|
* @param int $core: UID of the Solr core to use |
|
90
|
|
|
* |
|
91
|
|
|
* @return bool true on success or false on failure |
|
92
|
|
|
*/ |
|
93
|
|
|
public static function add(Document &$doc, $core = 0) |
|
94
|
|
|
{ |
|
95
|
|
|
if (in_array($doc->uid, self::$processedDocs)) { |
|
96
|
|
|
return true; |
|
97
|
|
|
} elseif (self::solrConnect($core, $doc->pid)) { |
|
98
|
|
|
$success = true; |
|
99
|
|
|
// Handle multi-volume documents. |
|
100
|
|
|
if ($doc->parentId) { |
|
101
|
|
|
$parent = Document::getInstance($doc->parentId, 0, true); |
|
102
|
|
|
if ($parent->ready) { |
|
103
|
|
|
$success = self::add($parent, $core); |
|
104
|
|
|
} else { |
|
105
|
|
|
Helper::log('Could not load parent document with UID ' . $doc->parentId, LOG_SEVERITY_ERROR); |
|
106
|
|
|
return false; |
|
107
|
|
|
} |
|
108
|
|
|
} |
|
109
|
|
|
try { |
|
110
|
|
|
// Add document to list of processed documents. |
|
111
|
|
|
self::$processedDocs[] = $doc->uid; |
|
112
|
|
|
// Delete old Solr documents. |
|
113
|
|
|
$updateQuery = self::$solr->service->createUpdate(); |
|
114
|
|
|
$updateQuery->addDeleteQuery('uid:' . $doc->uid); |
|
115
|
|
|
self::$solr->service->update($updateQuery); |
|
116
|
|
|
|
|
117
|
|
|
// Index every logical unit as separate Solr document. |
|
118
|
|
|
foreach ($doc->tableOfContents as $logicalUnit) { |
|
119
|
|
|
if ($success) { |
|
120
|
|
|
$success = self::processLogical($doc, $logicalUnit); |
|
121
|
|
|
} else { |
|
122
|
|
|
break; |
|
123
|
|
|
} |
|
124
|
|
|
} |
|
125
|
|
|
// Index full text files if available. |
|
126
|
|
|
if ($doc->hasFulltext) { |
|
127
|
|
|
foreach ($doc->physicalStructure as $pageNumber => $xmlId) { |
|
128
|
|
|
if ($success) { |
|
129
|
|
|
$success = self::processPhysical($doc, $pageNumber, $doc->physicalStructureInfo[$xmlId]); |
|
130
|
|
|
} else { |
|
131
|
|
|
break; |
|
132
|
|
|
} |
|
133
|
|
|
} |
|
134
|
|
|
} |
|
135
|
|
|
// Commit all changes. |
|
136
|
|
|
$updateQuery = self::$solr->service->createUpdate(); |
|
137
|
|
|
$updateQuery->addCommit(); |
|
138
|
|
|
self::$solr->service->update($updateQuery); |
|
139
|
|
|
|
|
140
|
|
|
$queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class) |
|
141
|
|
|
->getQueryBuilderForTable('tx_dlf_documents'); |
|
142
|
|
|
|
|
143
|
|
|
// Get document title from database. |
|
144
|
|
|
$result = $queryBuilder |
|
145
|
|
|
->select('tx_dlf_documents.title AS title') |
|
146
|
|
|
->from('tx_dlf_documents') |
|
147
|
|
|
->where( |
|
148
|
|
|
$queryBuilder->expr()->eq('tx_dlf_documents.uid', intval($doc->uid)), |
|
149
|
|
|
Helper::whereExpression('tx_dlf_documents') |
|
150
|
|
|
) |
|
151
|
|
|
->setMaxResults(1) |
|
152
|
|
|
->execute(); |
|
153
|
|
|
|
|
154
|
|
|
$allResults = $result->fetchAll(); |
|
155
|
|
|
$resArray = $allResults[0]; |
|
156
|
|
|
if (!(\TYPO3_REQUESTTYPE & \TYPO3_REQUESTTYPE_CLI)) { |
|
157
|
|
|
if ($success) { |
|
158
|
|
|
Helper::addMessage( |
|
159
|
|
|
htmlspecialchars(sprintf(Helper::getMessage('flash.documentIndexed'), $resArray['title'], $doc->uid)), |
|
160
|
|
|
Helper::getMessage('flash.done', true), |
|
161
|
|
|
FlashMessage::OK, |
|
162
|
|
|
true, |
|
163
|
|
|
'core.template.flashMessages' |
|
164
|
|
|
); |
|
165
|
|
|
} else { |
|
166
|
|
|
Helper::addMessage( |
|
167
|
|
|
htmlspecialchars(sprintf(Helper::getMessage('flash.documentNotIndexed'), $resArray['title'], $doc->uid)), |
|
168
|
|
|
Helper::getMessage('flash.error', true), |
|
169
|
|
|
FlashMessage::ERROR, |
|
170
|
|
|
true, |
|
171
|
|
|
'core.template.flashMessages' |
|
172
|
|
|
); |
|
173
|
|
|
} |
|
174
|
|
|
} |
|
175
|
|
|
return $success; |
|
176
|
|
|
} catch (\Exception $e) { |
|
177
|
|
|
if (!(\TYPO3_REQUESTTYPE & \TYPO3_REQUESTTYPE_CLI)) { |
|
178
|
|
|
Helper::addMessage( |
|
179
|
|
|
Helper::getMessage('flash.solrException', true) . '<br />' . htmlspecialchars($e->getMessage()), |
|
180
|
|
|
Helper::getMessage('flash.error', true), |
|
181
|
|
|
FlashMessage::ERROR, |
|
182
|
|
|
true, |
|
183
|
|
|
'core.template.flashMessages' |
|
184
|
|
|
); |
|
185
|
|
|
} |
|
186
|
|
|
Helper::log('Apache Solr threw exception: "' . $e->getMessage() . '"', LOG_SEVERITY_ERROR); |
|
187
|
|
|
return false; |
|
188
|
|
|
} |
|
189
|
|
|
} else { |
|
190
|
|
|
if (!(\TYPO3_REQUESTTYPE & \TYPO3_REQUESTTYPE_CLI)) { |
|
191
|
|
|
Helper::addMessage( |
|
192
|
|
|
Helper::getMessage('flash.solrNoConnection', true), |
|
193
|
|
|
Helper::getMessage('flash.warning', true), |
|
194
|
|
|
FlashMessage::WARNING, |
|
195
|
|
|
true, |
|
196
|
|
|
'core.template.flashMessages' |
|
197
|
|
|
); |
|
198
|
|
|
} |
|
199
|
|
|
Helper::log('Could not connect to Apache Solr server', LOG_SEVERITY_ERROR); |
|
200
|
|
|
return false; |
|
201
|
|
|
} |
|
202
|
|
|
} |
|
203
|
|
|
|
|
204
|
|
|
/** |
|
205
|
|
|
* Returns the dynamic index field name for the given metadata field. |
|
206
|
|
|
* |
|
207
|
|
|
* @access public |
|
208
|
|
|
* |
|
209
|
|
|
* @param string $index_name: The metadata field's name in database |
|
210
|
|
|
* @param int $pid: UID of the configuration page |
|
211
|
|
|
* |
|
212
|
|
|
* @return string The field's dynamic index name |
|
213
|
|
|
*/ |
|
214
|
|
|
public static function getIndexFieldName($index_name, $pid = 0) |
|
215
|
|
|
{ |
|
216
|
|
|
// Sanitize input. |
|
217
|
|
|
$pid = max(intval($pid), 0); |
|
218
|
|
|
if (!$pid) { |
|
219
|
|
|
Helper::log('Invalid PID ' . $pid . ' for metadata configuration', LOG_SEVERITY_ERROR); |
|
220
|
|
|
return ''; |
|
221
|
|
|
} |
|
222
|
|
|
// Load metadata configuration. |
|
223
|
|
|
self::loadIndexConf($pid); |
|
224
|
|
|
// Build field's suffix. |
|
225
|
|
|
$suffix = (in_array($index_name, self::$fields['tokenized']) ? 't' : 'u'); |
|
226
|
|
|
$suffix .= (in_array($index_name, self::$fields['stored']) ? 's' : 'u'); |
|
227
|
|
|
$suffix .= (in_array($index_name, self::$fields['indexed']) ? 'i' : 'u'); |
|
228
|
|
|
$index_name .= '_' . $suffix; |
|
229
|
|
|
return $index_name; |
|
230
|
|
|
} |
|
231
|
|
|
|
|
232
|
|
|
/** |
|
233
|
|
|
* Load indexing configuration |
|
234
|
|
|
* |
|
235
|
|
|
* @access protected |
|
236
|
|
|
* |
|
237
|
|
|
* @param int $pid: The configuration page's UID |
|
238
|
|
|
* |
|
239
|
|
|
* @return void |
|
240
|
|
|
*/ |
|
241
|
|
|
protected static function loadIndexConf($pid) |
|
242
|
|
|
{ |
|
243
|
|
|
if (!self::$fieldsLoaded) { |
|
244
|
|
|
$queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class) |
|
245
|
|
|
->getQueryBuilderForTable('tx_dlf_metadata'); |
|
246
|
|
|
|
|
247
|
|
|
// Get the metadata indexing options. |
|
248
|
|
|
$result = $queryBuilder |
|
249
|
|
|
->select( |
|
250
|
|
|
'tx_dlf_metadata.index_name AS index_name', |
|
251
|
|
|
'tx_dlf_metadata.index_tokenized AS index_tokenized', |
|
252
|
|
|
'tx_dlf_metadata.index_stored AS index_stored', |
|
253
|
|
|
'tx_dlf_metadata.index_indexed AS index_indexed', |
|
254
|
|
|
'tx_dlf_metadata.is_sortable AS is_sortable', |
|
255
|
|
|
'tx_dlf_metadata.is_facet AS is_facet', |
|
256
|
|
|
'tx_dlf_metadata.is_listed AS is_listed', |
|
257
|
|
|
'tx_dlf_metadata.index_autocomplete AS index_autocomplete', |
|
258
|
|
|
'tx_dlf_metadata.index_boost AS index_boost' |
|
259
|
|
|
) |
|
260
|
|
|
->from('tx_dlf_metadata') |
|
261
|
|
|
->where( |
|
262
|
|
|
$queryBuilder->expr()->eq('tx_dlf_metadata.pid', intval($pid)), |
|
263
|
|
|
Helper::whereExpression('tx_dlf_metadata') |
|
264
|
|
|
) |
|
265
|
|
|
->execute(); |
|
266
|
|
|
|
|
267
|
|
|
while ($indexing = $result->fetch()) { |
|
268
|
|
|
if ($indexing['index_tokenized']) { |
|
269
|
|
|
self::$fields['tokenized'][] = $indexing['index_name']; |
|
270
|
|
|
} |
|
271
|
|
|
if ( |
|
272
|
|
|
$indexing['index_stored'] |
|
273
|
|
|
|| $indexing['is_listed'] |
|
274
|
|
|
) { |
|
275
|
|
|
self::$fields['stored'][] = $indexing['index_name']; |
|
276
|
|
|
} |
|
277
|
|
|
if ( |
|
278
|
|
|
$indexing['index_indexed'] |
|
279
|
|
|
|| $indexing['index_autocomplete'] |
|
280
|
|
|
) { |
|
281
|
|
|
self::$fields['indexed'][] = $indexing['index_name']; |
|
282
|
|
|
} |
|
283
|
|
|
if ($indexing['is_sortable']) { |
|
284
|
|
|
self::$fields['sortables'][] = $indexing['index_name']; |
|
285
|
|
|
} |
|
286
|
|
|
if ($indexing['is_facet']) { |
|
287
|
|
|
self::$fields['facets'][] = $indexing['index_name']; |
|
288
|
|
|
} |
|
289
|
|
|
if ($indexing['index_autocomplete']) { |
|
290
|
|
|
self::$fields['autocomplete'][] = $indexing['index_name']; |
|
291
|
|
|
} |
|
292
|
|
|
if ($indexing['index_boost'] > 0.0) { |
|
293
|
|
|
self::$fields['fieldboost'][$indexing['index_name']] = floatval($indexing['index_boost']); |
|
294
|
|
|
} else { |
|
295
|
|
|
self::$fields['fieldboost'][$indexing['index_name']] = false; |
|
296
|
|
|
} |
|
297
|
|
|
} |
|
298
|
|
|
self::$fieldsLoaded = true; |
|
299
|
|
|
} |
|
300
|
|
|
} |
|
301
|
|
|
|
|
302
|
|
|
/** |
|
303
|
|
|
* Processes a logical unit (and its children) for the Solr index |
|
304
|
|
|
* |
|
305
|
|
|
* @access protected |
|
306
|
|
|
* |
|
307
|
|
|
* @param \Kitodo\Dlf\Common\Document &$doc: The METS document |
|
308
|
|
|
* @param array $logicalUnit: Array of the logical unit to process |
|
309
|
|
|
* |
|
310
|
|
|
* @return bool true on success or false on failure |
|
311
|
|
|
*/ |
|
312
|
|
|
protected static function processLogical(Document &$doc, array $logicalUnit) |
|
313
|
|
|
{ |
|
314
|
|
|
$success = true; |
|
315
|
|
|
// Get metadata for logical unit. |
|
316
|
|
|
$metadata = $doc->metadataArray[$logicalUnit['id']]; |
|
317
|
|
|
if (!empty($metadata)) { |
|
318
|
|
|
$metadata['author'] = self::removeAppendsFromAuthor($metadata['author']); |
|
319
|
|
|
// Replace owner UID with index_name. |
|
320
|
|
|
if ( |
|
321
|
|
|
!empty($metadata['owner'][0]) |
|
322
|
|
|
&& MathUtility::canBeInterpretedAsInteger($metadata['owner'][0]) |
|
323
|
|
|
) { |
|
324
|
|
|
$metadata['owner'][0] = Helper::getIndexNameFromUid($metadata['owner'][0], 'tx_dlf_libraries', $doc->cPid); |
|
325
|
|
|
} |
|
326
|
|
|
// Create new Solr document. |
|
327
|
|
|
$updateQuery = self::getUpdateQuery($doc, $logicalUnit); |
|
328
|
|
|
if (MathUtility::canBeInterpretedAsInteger($logicalUnit['points'])) { |
|
329
|
|
|
$solrDoc->setField('page', $logicalUnit['points']); |
|
|
|
|
|
|
330
|
|
|
} |
|
331
|
|
|
if ($logicalUnit['id'] == $doc->toplevelId) { |
|
332
|
|
|
$solrDoc->setField('thumbnail', $doc->thumbnail); |
|
333
|
|
|
} elseif (!empty($logicalUnit['thumbnailId'])) { |
|
334
|
|
|
$solrDoc->setField('thumbnail', $doc->getFileLocation($logicalUnit['thumbnailId'])); |
|
335
|
|
|
} |
|
336
|
|
|
// There can be only one toplevel unit per UID, independently of backend configuration |
|
337
|
|
|
$solrDoc->setField('toplevel', $logicalUnit['id'] == $doc->toplevelId ? true : false); |
|
338
|
|
|
$solrDoc->setField('title', $metadata['title'][0], self::$fields['fieldboost']['title']); |
|
339
|
|
|
$solrDoc->setField('volume', $metadata['volume'][0], self::$fields['fieldboost']['volume']); |
|
340
|
|
|
$solrDoc->setField('record_id', $metadata['record_id'][0]); |
|
341
|
|
|
$solrDoc->setField('purl', $metadata['purl'][0]); |
|
342
|
|
|
$solrDoc->setField('location', $doc->location); |
|
343
|
|
|
$solrDoc->setField('urn', $metadata['urn']); |
|
344
|
|
|
$solrDoc->setField('license', $metadata['license']); |
|
345
|
|
|
$solrDoc->setField('terms', $metadata['terms']); |
|
346
|
|
|
$solrDoc->setField('restrictions', $metadata['restrictions']); |
|
347
|
|
|
$coordinates = json_decode($metadata['coordinates'][0]); |
|
348
|
|
|
if (is_object($coordinates)) { |
|
349
|
|
|
$solrDoc->setField('geom', json_encode($coordinates->features[0])); |
|
350
|
|
|
} |
|
351
|
|
|
$autocomplete = []; |
|
352
|
|
|
foreach ($metadata as $index_name => $data) { |
|
353
|
|
|
if ( |
|
354
|
|
|
!empty($data) |
|
355
|
|
|
&& substr($index_name, -8) !== '_sorting' |
|
356
|
|
|
) { |
|
357
|
|
|
$solrDoc->setField(self::getIndexFieldName($index_name, $doc->pid), $data, self::$fields['fieldboost'][$index_name]); |
|
358
|
|
|
if (in_array($index_name, self::$fields['sortables'])) { |
|
359
|
|
|
// Add sortable fields to index. |
|
360
|
|
|
$solrDoc->setField($index_name . '_sorting', $metadata[$index_name . '_sorting'][0]); |
|
361
|
|
|
} |
|
362
|
|
|
if (in_array($index_name, self::$fields['facets'])) { |
|
363
|
|
|
// Add facets to index. |
|
364
|
|
|
$solrDoc->setField($index_name . '_faceting', $data); |
|
365
|
|
|
} |
|
366
|
|
|
if (in_array($index_name, self::$fields['autocomplete'])) { |
|
367
|
|
|
$autocomplete = array_merge($autocomplete, $data); |
|
368
|
|
|
} |
|
369
|
|
|
} |
|
370
|
|
|
} |
|
371
|
|
|
// Add autocomplete values to index. |
|
372
|
|
|
if (!empty($autocomplete)) { |
|
373
|
|
|
$solrDoc->setField('autocomplete', $autocomplete); |
|
374
|
|
|
} |
|
375
|
|
|
// Add collection information to logical sub-elements if applicable. |
|
376
|
|
|
if ( |
|
377
|
|
|
in_array('collection', self::$fields['facets']) |
|
378
|
|
|
&& empty($metadata['collection']) |
|
379
|
|
|
&& !empty($doc->metadataArray[$doc->toplevelId]['collection']) |
|
380
|
|
|
) { |
|
381
|
|
|
$solrDoc->setField('collection_faceting', $doc->metadataArray[$doc->toplevelId]['collection']); |
|
382
|
|
|
} |
|
383
|
|
|
try { |
|
384
|
|
|
$updateQuery->addDocument($solrDoc); |
|
385
|
|
|
self::$solr->service->update($updateQuery); |
|
386
|
|
|
} catch (\Exception $e) { |
|
387
|
|
|
if (!(\TYPO3_REQUESTTYPE & \TYPO3_REQUESTTYPE_CLI)) { |
|
388
|
|
|
Helper::addMessage( |
|
389
|
|
|
Helper::getMessage('flash.solrException', true) . '<br />' . htmlspecialchars($e->getMessage()), |
|
390
|
|
|
Helper::getMessage('flash.error', true), |
|
391
|
|
|
FlashMessage::ERROR, |
|
392
|
|
|
true, |
|
393
|
|
|
'core.template.flashMessages' |
|
394
|
|
|
); |
|
395
|
|
|
} |
|
396
|
|
|
Helper::log('Apache Solr threw exception: "' . $e->getMessage() . '"', LOG_SEVERITY_ERROR); |
|
397
|
|
|
return false; |
|
398
|
|
|
} |
|
399
|
|
|
} |
|
400
|
|
|
// Check for child elements... |
|
401
|
|
|
if (!empty($logicalUnit['children'])) { |
|
402
|
|
|
foreach ($logicalUnit['children'] as $child) { |
|
403
|
|
|
if ($success) { |
|
404
|
|
|
// ...and process them, too. |
|
405
|
|
|
$success = self::processLogical($doc, $child); |
|
406
|
|
|
} else { |
|
407
|
|
|
break; |
|
408
|
|
|
} |
|
409
|
|
|
} |
|
410
|
|
|
} |
|
411
|
|
|
return $success; |
|
412
|
|
|
} |
|
413
|
|
|
|
|
414
|
|
|
/** |
|
415
|
|
|
* Processes a physical unit for the Solr index |
|
416
|
|
|
* |
|
417
|
|
|
* @access protected |
|
418
|
|
|
* |
|
419
|
|
|
* @param \Kitodo\Dlf\Common\Document &$doc: The METS document |
|
420
|
|
|
* @param int $page: The page number |
|
421
|
|
|
* @param array $physicalUnit: Array of the physical unit to process |
|
422
|
|
|
* |
|
423
|
|
|
* @return bool true on success or false on failure |
|
424
|
|
|
*/ |
|
425
|
|
|
protected static function processPhysical(Document &$doc, $page, array $physicalUnit) |
|
426
|
|
|
{ |
|
427
|
|
|
if ($doc->hasFulltext && $fullText = $doc->getFullText($physicalUnit['id'])) { |
|
428
|
|
|
// Read extension configuration. |
|
429
|
|
|
$extConf = GeneralUtility::makeInstance(ExtensionConfiguration::class)->get(self::$extKey); |
|
430
|
|
|
// Create new Solr document. |
|
431
|
|
|
$updateQuery = self::getUpdateQuery($doc, $physicalUnit, $fullText); |
|
432
|
|
|
$solrDoc->setField('page', $page); |
|
|
|
|
|
|
433
|
|
|
$fileGrpsThumb = GeneralUtility::trimExplode(',', $extConf['fileGrpThumbs']); |
|
434
|
|
|
while ($fileGrpThumb = array_shift($fileGrpsThumb)) { |
|
435
|
|
|
if (!empty($physicalUnit['files'][$fileGrpThumb])) { |
|
436
|
|
|
$solrDoc->setField('thumbnail', $doc->getFileLocation($physicalUnit['files'][$fileGrpThumb])); |
|
437
|
|
|
break; |
|
438
|
|
|
} |
|
439
|
|
|
} |
|
440
|
|
|
$solrDoc->setField('toplevel', false); |
|
441
|
|
|
// Add faceting information to physical sub-elements if applicable. |
|
442
|
|
|
foreach ($doc->metadataArray[$doc->toplevelId] as $index_name => $data) { |
|
443
|
|
|
if ( |
|
444
|
|
|
!empty($data) |
|
445
|
|
|
&& substr($index_name, -8) !== '_sorting' |
|
446
|
|
|
) { |
|
447
|
|
|
|
|
448
|
|
|
if (in_array($index_name, self::$fields['facets'])) { |
|
449
|
|
|
if ($index_name == 'author') { |
|
450
|
|
|
$data = self::removeAppendsFromAuthor($data); |
|
451
|
|
|
} |
|
452
|
|
|
// Add facets to index. |
|
453
|
|
|
$solrDoc->setField($index_name . '_faceting', $data); |
|
454
|
|
|
} |
|
455
|
|
|
} |
|
456
|
|
|
} |
|
457
|
|
|
// Add collection information to physical sub-elements if applicable. |
|
458
|
|
|
if ( |
|
459
|
|
|
in_array('collection', self::$fields['facets']) |
|
460
|
|
|
&& !empty($doc->metadataArray[$doc->toplevelId]['collection']) |
|
461
|
|
|
) { |
|
462
|
|
|
$solrDoc->setField('collection_faceting', $doc->metadataArray[$doc->toplevelId]['collection']); |
|
463
|
|
|
} |
|
464
|
|
|
try { |
|
465
|
|
|
$updateQuery->addDocument($solrDoc); |
|
466
|
|
|
self::$solr->service->update($updateQuery); |
|
467
|
|
|
} catch (\Exception $e) { |
|
468
|
|
|
if (!(\TYPO3_REQUESTTYPE & \TYPO3_REQUESTTYPE_CLI)) { |
|
469
|
|
|
Helper::addMessage( |
|
470
|
|
|
Helper::getMessage('flash.solrException', true) . '<br />' . htmlspecialchars($e->getMessage()), |
|
471
|
|
|
Helper::getMessage('flash.error', true), |
|
472
|
|
|
FlashMessage::ERROR, |
|
473
|
|
|
true, |
|
474
|
|
|
'core.template.flashMessages' |
|
475
|
|
|
); |
|
476
|
|
|
} |
|
477
|
|
|
Helper::log('Apache Solr threw exception: "' . $e->getMessage() . '"', LOG_SEVERITY_ERROR); |
|
478
|
|
|
return false; |
|
479
|
|
|
} |
|
480
|
|
|
} |
|
481
|
|
|
return true; |
|
482
|
|
|
} |
|
483
|
|
|
|
|
484
|
|
|
/** |
|
485
|
|
|
* Connects to Solr server. |
|
486
|
|
|
* |
|
487
|
|
|
* @access protected |
|
488
|
|
|
* |
|
489
|
|
|
* @param int $core: UID of the Solr core |
|
490
|
|
|
* @param int $pid: UID of the configuration page |
|
491
|
|
|
* |
|
492
|
|
|
* @return bool true on success or false on failure |
|
493
|
|
|
*/ |
|
494
|
|
|
protected static function solrConnect($core, $pid = 0) |
|
495
|
|
|
{ |
|
496
|
|
|
// Get Solr instance. |
|
497
|
|
|
if (!self::$solr) { |
|
498
|
|
|
// Connect to Solr server. |
|
499
|
|
|
$solr = Solr::getInstance($core); |
|
500
|
|
|
if ($solr->ready) { |
|
501
|
|
|
self::$solr = $solr; |
|
502
|
|
|
// Load indexing configuration if needed. |
|
503
|
|
|
if ($pid) { |
|
504
|
|
|
self::loadIndexConf($pid); |
|
505
|
|
|
} |
|
506
|
|
|
} else { |
|
507
|
|
|
return false; |
|
508
|
|
|
} |
|
509
|
|
|
} |
|
510
|
|
|
return true; |
|
511
|
|
|
} |
|
512
|
|
|
|
|
513
|
|
|
/** |
|
514
|
|
|
* Get update query with set standard fields (identical for logical and physical unit) |
|
515
|
|
|
* |
|
516
|
|
|
* @access private |
|
517
|
|
|
* |
|
518
|
|
|
* @param \Kitodo\Dlf\Common\Document &$doc: The METS document |
|
519
|
|
|
* @param array $unit: Array of the logical or physical unit to process |
|
520
|
|
|
* @param string $fullText: Text containing full text for indexing |
|
521
|
|
|
* |
|
522
|
|
|
* @return \Solarium\Core\Query\AbstractQuery|\Solarium\QueryType\Update\Query\Query |
|
523
|
|
|
*/ |
|
524
|
|
|
private static function getUpdateQuery($doc, $unit, $fullText = '') { |
|
525
|
|
|
$updateQuery = self::$solr->service->createUpdate(); |
|
526
|
|
|
$solrDoc = $updateQuery->createDocument(); |
|
527
|
|
|
// Create unique identifier from document's UID and unit's XML ID. |
|
528
|
|
|
$solrDoc->setField('id', $doc->uid . $unit['id']); |
|
|
|
|
|
|
529
|
|
|
$solrDoc->setField('uid', $doc->uid); |
|
530
|
|
|
$solrDoc->setField('pid', $doc->pid); |
|
531
|
|
|
$solrDoc->setField('partof', $doc->parentId); |
|
532
|
|
|
$solrDoc->setField('root', $doc->rootId); |
|
533
|
|
|
$solrDoc->setField('sid', $unit['id']); |
|
534
|
|
|
$solrDoc->setField('type', $unit['type'], self::$fields['fieldboost']['type']); |
|
535
|
|
|
$solrDoc->setField('collection', $doc->metadataArray[$doc->toplevelId]['collection']); |
|
536
|
|
|
$solrDoc->setField('fulltext', $fullText); |
|
537
|
|
|
return $updateQuery; |
|
538
|
|
|
} |
|
539
|
|
|
|
|
540
|
|
|
/** |
|
541
|
|
|
* Remove appended "valueURI" from authors' names for indexing. |
|
542
|
|
|
* |
|
543
|
|
|
* @access private |
|
544
|
|
|
* |
|
545
|
|
|
* @param array|string $authors: Array or string containing author/authors |
|
546
|
|
|
* |
|
547
|
|
|
* @return array|string |
|
548
|
|
|
*/ |
|
549
|
|
|
private static function removeAppendsFromAuthor($authors) { |
|
550
|
|
|
if (is_array($authors)) { |
|
551
|
|
|
foreach ($authors as $i => $author) { |
|
552
|
|
|
$splitName = explode(chr(31), $author); |
|
553
|
|
|
$authors[$i] = $splitName[0]; |
|
554
|
|
|
} |
|
555
|
|
|
} |
|
556
|
|
|
return $authors; |
|
557
|
|
|
} |
|
558
|
|
|
|
|
559
|
|
|
/** |
|
560
|
|
|
* Prevent instantiation by hiding the constructor |
|
561
|
|
|
* |
|
562
|
|
|
* @access private |
|
563
|
|
|
*/ |
|
564
|
|
|
private function __construct() |
|
565
|
|
|
{ |
|
566
|
|
|
// This is a static class, thus no instances should be created. |
|
567
|
|
|
} |
|
568
|
|
|
} |
|
569
|
|
|
|
kitodo /
kitodo-presentation
Loading history...
