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 Flow\JSONPath\JSONPath; |
16
|
|
|
use TYPO3\CMS\Core\Configuration\ExtensionConfiguration; |
17
|
|
|
use TYPO3\CMS\Core\Database\ConnectionPool; |
18
|
|
|
use TYPO3\CMS\Core\Database\Query\Restriction\HiddenRestriction; |
19
|
|
|
use TYPO3\CMS\Core\Log\LogManager; |
20
|
|
|
use TYPO3\CMS\Core\Utility\GeneralUtility; |
21
|
|
|
use Ubl\Iiif\Presentation\Common\Model\Resources\AnnotationContainerInterface; |
22
|
|
|
use Ubl\Iiif\Presentation\Common\Model\Resources\AnnotationInterface; |
23
|
|
|
use Ubl\Iiif\Presentation\Common\Model\Resources\CanvasInterface; |
24
|
|
|
use Ubl\Iiif\Presentation\Common\Model\Resources\ContentResourceInterface; |
25
|
|
|
use Ubl\Iiif\Presentation\Common\Model\Resources\IiifResourceInterface; |
26
|
|
|
use Ubl\Iiif\Presentation\Common\Model\Resources\ManifestInterface; |
27
|
|
|
use Ubl\Iiif\Presentation\Common\Model\Resources\RangeInterface; |
28
|
|
|
use Ubl\Iiif\Presentation\Common\Vocabulary\Motivation; |
29
|
|
|
use Ubl\Iiif\Presentation\V1\Model\Resources\AbstractIiifResource1; |
30
|
|
|
use Ubl\Iiif\Presentation\V2\Model\Resources\AbstractIiifResource2; |
31
|
|
|
use Ubl\Iiif\Presentation\V3\Model\Resources\AbstractIiifResource3; |
32
|
|
|
use Ubl\Iiif\Services\AbstractImageService; |
33
|
|
|
use Ubl\Iiif\Services\Service; |
34
|
|
|
use Ubl\Iiif\Tools\IiifHelper; |
35
|
|
|
|
36
|
|
|
/** |
37
|
|
|
* IiifManifest class for the 'dlf' extension. |
38
|
|
|
* |
39
|
|
|
* @package TYPO3 |
40
|
|
|
* @subpackage dlf |
41
|
|
|
* |
42
|
|
|
* @access public |
43
|
|
|
* |
44
|
|
|
* @property int $cPid this holds the PID for the configuration |
45
|
|
|
* @property-read array $formats this holds the configuration for all supported metadata encodings |
46
|
|
|
* @property bool $formatsLoaded flag with information if the available metadata formats are loaded |
47
|
|
|
* @property-read bool $hasFulltext flag with information if there are any fulltext files available |
48
|
|
|
* @property array $lastSearchedPhysicalPage the last searched logical and physical page |
49
|
|
|
* @property array $logicalUnits this holds the logical units |
50
|
|
|
* @property-read array $metadataArray this holds the documents' parsed metadata array |
51
|
|
|
* @property bool $metadataArrayLoaded flag with information if the metadata array is loaded |
52
|
|
|
* @property-read int $numPages the holds the total number of pages |
53
|
|
|
* @property-read int $parentId this holds the UID of the parent document or zero if not multi-volumed |
54
|
|
|
* @property-read array $physicalStructure this holds the physical structure |
55
|
|
|
* @property-read array $physicalStructureInfo this holds the physical structure metadata |
56
|
|
|
* @property bool $physicalStructureLoaded flag with information if the physical structure is loaded |
57
|
|
|
* @property-read int $pid this holds the PID of the document or zero if not in database |
58
|
|
|
* @property array $rawTextArray this holds the documents' raw text pages with their corresponding structMap//div's ID (METS) or Range / Manifest / Sequence ID (IIIF) as array key |
59
|
|
|
* @property-read bool $ready Is the document instantiated successfully? |
60
|
|
|
* @property-read string $recordId the METS file's / IIIF manifest's record identifier |
61
|
|
|
* @property-read int $rootId this holds the UID of the root document or zero if not multi-volumed |
62
|
|
|
* @property-read array $smLinks this holds the smLinks between logical and physical structMap |
63
|
|
|
* @property bool $smLinksLoaded flag with information if the smLinks are loaded |
64
|
|
|
* @property-read array $tableOfContents this holds the logical structure |
65
|
|
|
* @property bool $tableOfContentsLoaded flag with information if the table of contents is loaded |
66
|
|
|
* @property-read string $thumbnail this holds the document's thumbnail location |
67
|
|
|
* @property bool $thumbnailLoaded flag with information if the thumbnail is loaded |
68
|
|
|
* @property-read string $toplevelId this holds the toplevel structure's "@ID" (METS) or the manifest's "@id" (IIIF) |
69
|
|
|
* @property \SimpleXMLElement $xml this holds the whole XML file as \SimpleXMLElement object |
70
|
|
|
* @property string $asJson this holds the manifest file as string for serialization purposes |
71
|
|
|
* @property ManifestInterface $iiif a PHP object representation of a IIIF manifest |
72
|
|
|
* @property string $iiifVersion 'IIIF1', 'IIIF2' or 'IIIF3', depending on the API $this->iiif conforms to |
73
|
|
|
* @property bool $hasFulltextSet flag if document has already been analyzed for presence of the fulltext for the Solr index |
74
|
|
|
* @property array $originalMetadataArray this holds the original manifest's parsed metadata array with their corresponding resource (Manifest / Sequence / Range) ID as array key |
75
|
|
|
* @property array $mimeTypes this holds the mime types of linked resources in the manifest (extracted during parsing) for later us |
76
|
|
|
* |
77
|
|
|
*/ |
78
|
|
|
final class IiifManifest extends AbstractDocument |
79
|
|
|
{ |
80
|
|
|
/** |
81
|
|
|
* @access protected |
82
|
|
|
* @var string This holds the manifest file as string for serialization purposes |
83
|
|
|
* |
84
|
|
|
* @see __sleep() / __wakeup() |
85
|
|
|
*/ |
86
|
|
|
protected string $asJson = ''; |
87
|
|
|
|
88
|
|
|
/** |
89
|
|
|
* @access protected |
90
|
|
|
* @var ManifestInterface|null A PHP object representation of a IIIF manifest |
91
|
|
|
*/ |
92
|
|
|
protected ?ManifestInterface $iiif; |
93
|
|
|
|
94
|
|
|
/** |
95
|
|
|
* @access protected |
96
|
|
|
* @var string 'IIIF1', 'IIIF2' or 'IIIF3', depending on the API $this->iiif conforms to: IIIF Metadata API 1, IIIF Presentation API 2 or 3 |
97
|
|
|
*/ |
98
|
|
|
protected string $iiifVersion; |
99
|
|
|
|
100
|
|
|
/** |
101
|
|
|
* @access protected |
102
|
|
|
* @var bool Document has already been analyzed if it contains fulltext for the Solr index |
103
|
|
|
*/ |
104
|
|
|
protected bool $hasFulltextSet = false; |
105
|
|
|
|
106
|
|
|
/** |
107
|
|
|
* @access protected |
108
|
|
|
* @var array This holds the original manifest's parsed metadata array with their corresponding resource (Manifest / Sequence / Range) ID as array key |
109
|
|
|
*/ |
110
|
|
|
protected array $originalMetadataArray = []; |
111
|
|
|
|
112
|
|
|
/** |
113
|
|
|
* @access protected |
114
|
|
|
* @var array Holds the mime types of linked resources in the manifest (extracted during parsing) for later use |
115
|
|
|
*/ |
116
|
|
|
protected array $mimeTypes = []; |
117
|
|
|
|
118
|
|
|
/** |
119
|
|
|
* @see AbstractDocument::establishRecordId() |
120
|
|
|
*/ |
121
|
|
|
protected function establishRecordId(int $pid): void |
122
|
|
|
{ |
123
|
|
|
if ($this->iiif !== null) { |
124
|
|
|
/* |
125
|
|
|
* FIXME This will not consistently work because we can not be sure to have the pid at hand. It may miss |
126
|
|
|
* if the plugin that actually loads the manifest allows content from other pages. |
127
|
|
|
* Up until now the cPid is only set after the document has been initialized. We need it before to |
128
|
|
|
* check the configuration. |
129
|
|
|
* TODO Saving / indexing should still work - check! |
130
|
|
|
*/ |
131
|
|
|
$queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class) |
132
|
|
|
->getQueryBuilderForTable('tx_dlf_metadata'); |
133
|
|
|
// Get hidden records, too. |
134
|
|
|
$queryBuilder |
135
|
|
|
->getRestrictions() |
136
|
|
|
->removeByType(HiddenRestriction::class); |
137
|
|
|
$result = $queryBuilder |
138
|
|
|
->select('tx_dlf_metadataformat.xpath AS querypath') |
139
|
|
|
->from('tx_dlf_metadata') |
140
|
|
|
->from('tx_dlf_metadataformat') |
141
|
|
|
->from('tx_dlf_formats') |
142
|
|
|
->where( |
143
|
|
|
$queryBuilder->expr()->eq('tx_dlf_metadata.pid', (int) $pid), |
144
|
|
|
$queryBuilder->expr()->eq('tx_dlf_metadataformat.pid', (int) $pid), |
145
|
|
|
$queryBuilder->expr()->orX( |
146
|
|
|
$queryBuilder->expr()->andX( |
147
|
|
|
$queryBuilder->expr()->eq('tx_dlf_metadata.uid', 'tx_dlf_metadataformat.parent_id'), |
148
|
|
|
$queryBuilder->expr()->eq('tx_dlf_metadataformat.encoded', 'tx_dlf_formats.uid'), |
149
|
|
|
$queryBuilder->expr()->eq('tx_dlf_metadata.index_name', $queryBuilder->createNamedParameter('record_id')), |
150
|
|
|
$queryBuilder->expr()->eq('tx_dlf_formats.type', $queryBuilder->createNamedParameter($this->getIiifVersion())) |
151
|
|
|
), |
152
|
|
|
$queryBuilder->expr()->eq('tx_dlf_metadata.format', 0) |
153
|
|
|
) |
154
|
|
|
) |
155
|
|
|
->execute(); |
156
|
|
|
while ($resArray = $result->fetchAssociative()) { |
157
|
|
|
$recordIdPath = $resArray['querypath']; |
158
|
|
|
if (!empty($recordIdPath)) { |
159
|
|
|
try { |
160
|
|
|
$this->recordId = $this->iiif->jsonPath($recordIdPath); |
|
|
|
|
161
|
|
|
} catch (\Exception $e) { |
162
|
|
|
$this->logger->warning('Could not evaluate JSONPath to get IIIF record ID'); |
163
|
|
|
} |
164
|
|
|
} |
165
|
|
|
} |
166
|
|
|
// For now, it's a hardcoded ID, not only as a fallback |
167
|
|
|
if (!isset($this->recordId)) { |
168
|
|
|
$this->recordId = $this->iiif->getId(); |
169
|
|
|
} |
170
|
|
|
} |
171
|
|
|
} |
172
|
|
|
|
173
|
|
|
/** |
174
|
|
|
* @see AbstractDocument::getDocument() |
175
|
|
|
*/ |
176
|
|
|
protected function getDocument(): IiifResourceInterface |
177
|
|
|
{ |
178
|
|
|
return $this->iiif; |
179
|
|
|
} |
180
|
|
|
|
181
|
|
|
/** |
182
|
|
|
* Returns a string representing the Metadata / Presentation API version which the IIIF resource |
183
|
|
|
* conforms to. This is used for example to extract metadata according to configured patterns. |
184
|
|
|
* |
185
|
|
|
* @access public |
186
|
|
|
* |
187
|
|
|
* @return string 'IIIF1' if the resource is a Metadata API 1 resource, 'IIIF2' / 'IIIF3' if |
188
|
|
|
* the resource is a Presentation API 2 / 3 resource |
189
|
|
|
*/ |
190
|
|
|
public function getIiifVersion(): string |
191
|
|
|
{ |
192
|
|
|
if (!isset($this->iiifVersion)) { |
193
|
|
|
if ($this->iiif instanceof AbstractIiifResource1) { |
194
|
|
|
$this->iiifVersion = 'IIIF1'; |
195
|
|
|
} elseif ($this->iiif instanceof AbstractIiifResource2) { |
196
|
|
|
$this->iiifVersion = 'IIIF2'; |
197
|
|
|
} elseif ($this->iiif instanceof AbstractIiifResource3) { |
198
|
|
|
$this->iiifVersion = 'IIIF3'; |
199
|
|
|
} |
200
|
|
|
} |
201
|
|
|
return $this->iiifVersion; |
202
|
|
|
} |
203
|
|
|
|
204
|
|
|
/** |
205
|
|
|
* True if getUseGroups() has been called and $this->useGrps is loaded |
206
|
|
|
* |
207
|
|
|
* @var bool |
208
|
|
|
* @access protected |
209
|
|
|
*/ |
210
|
|
|
protected bool $useGrpsLoaded = false; |
211
|
|
|
|
212
|
|
|
/** |
213
|
|
|
* Holds the configured useGrps as array. |
214
|
|
|
* |
215
|
|
|
* @var array |
216
|
|
|
* @access protected |
217
|
|
|
*/ |
218
|
|
|
protected array $useGrps = []; |
219
|
|
|
|
220
|
|
|
/** |
221
|
|
|
* IiifManifest also populates the physical structure array entries for matching |
222
|
|
|
* 'fileGrp's. To do that, the configuration has to be loaded; afterwards configured |
223
|
|
|
* 'fileGrp's for thumbnails, downloads, audio, fulltext and the 'fileGrp's for images |
224
|
|
|
* can be requested with this method. |
225
|
|
|
* |
226
|
|
|
* @access protected |
227
|
|
|
* |
228
|
|
|
* @param string $use |
229
|
|
|
* |
230
|
|
|
* @return array|string |
231
|
|
|
*/ |
232
|
|
|
protected function getUseGroups(string $use) |
233
|
|
|
{ |
234
|
|
|
if (!$this->useGrpsLoaded) { |
235
|
|
|
// Get configured USE attributes. |
236
|
|
|
$extConf = GeneralUtility::makeInstance(ExtensionConfiguration::class)->get(self::$extKey, 'files'); |
237
|
|
|
if (!empty($extConf['fileGrpImages'])) { |
238
|
|
|
$this->useGrps['fileGrpImages'] = GeneralUtility::trimExplode(',', $extConf['fileGrpImages']); |
239
|
|
|
} |
240
|
|
|
if (!empty($extConf['fileGrpThumbs'])) { |
241
|
|
|
$this->useGrps['fileGrpThumbs'] = GeneralUtility::trimExplode(',', $extConf['fileGrpThumbs']); |
242
|
|
|
} |
243
|
|
|
if (!empty($extConf['fileGrpDownload'])) { |
244
|
|
|
$this->useGrps['fileGrpDownload'] = GeneralUtility::trimExplode(',', $extConf['fileGrpDownload']); |
245
|
|
|
} |
246
|
|
|
if (!empty($extConf['fileGrpFulltext'])) { |
247
|
|
|
$this->useGrps['fileGrpFulltext'] = GeneralUtility::trimExplode(',', $extConf['fileGrpFulltext']); |
248
|
|
|
} |
249
|
|
|
if (!empty($extConf['fileGrpAudio'])) { |
250
|
|
|
$this->useGrps['fileGrpAudio'] = GeneralUtility::trimExplode(',', $extConf['fileGrpAudio']); |
251
|
|
|
} |
252
|
|
|
$this->useGrpsLoaded = true; |
253
|
|
|
} |
254
|
|
|
return array_key_exists($use, $this->useGrps) ? $this->useGrps[$use] : []; |
255
|
|
|
} |
256
|
|
|
|
257
|
|
|
/** |
258
|
|
|
* @see AbstractDocument::magicGetPhysicalStructure() |
259
|
|
|
*/ |
260
|
|
|
protected function magicGetPhysicalStructure(): array |
261
|
|
|
{ |
262
|
|
|
// Is there no physical structure array yet? |
263
|
|
|
if (!$this->physicalStructureLoaded) { |
264
|
|
|
if ($this->iiif == null || !($this->iiif instanceof ManifestInterface)) { |
265
|
|
|
return []; |
266
|
|
|
} |
267
|
|
|
$extConf = GeneralUtility::makeInstance(ExtensionConfiguration::class)->get(self::$extKey, 'iiif'); |
268
|
|
|
$iiifId = $this->iiif->getId(); |
269
|
|
|
$this->physicalStructureInfo[$iiifId]['id'] = $iiifId; |
|
|
|
|
270
|
|
|
$this->physicalStructureInfo[$iiifId]['dmdId'] = $iiifId; |
271
|
|
|
$this->physicalStructureInfo[$iiifId]['label'] = $this->iiif->getLabelForDisplay(); |
272
|
|
|
$this->physicalStructureInfo[$iiifId]['orderlabel'] = $this->iiif->getLabelForDisplay(); |
273
|
|
|
$this->physicalStructureInfo[$iiifId]['type'] = 'physSequence'; |
274
|
|
|
$this->physicalStructureInfo[$iiifId]['contentIds'] = null; |
275
|
|
|
|
276
|
|
|
$this->setFileUseDownload($iiifId, $this->iiif); |
277
|
|
|
$this->setFileUseFulltext($iiifId, $this->iiif); |
278
|
|
|
|
279
|
|
|
$fileUseThumbs = $this->getUseGroups('fileGrpThumbs'); |
280
|
|
|
$fileUses = $this->getUseGroups('fileGrpImages'); |
281
|
|
|
|
282
|
|
|
if (!empty($this->iiif->getDefaultCanvases())) { |
283
|
|
|
// canvases have not order property, but the context defines canveses as @list with a specific order, so we can provide an alternative |
284
|
|
|
$elements = []; |
285
|
|
|
$canvasOrder = 0; |
286
|
|
|
foreach ($this->iiif->getDefaultCanvases() as $canvas) { |
287
|
|
|
$canvasOrder++; |
288
|
|
|
$thumbnailUrl = $canvas->getThumbnailUrl(); |
289
|
|
|
// put thumbnails in thumbnail filegroup |
290
|
|
|
if ( |
291
|
|
|
!empty($thumbnailUrl) |
292
|
|
|
&& empty($this->physicalStructureInfo[$iiifId]['files'][$fileUseThumbs[0]]) |
293
|
|
|
) { |
294
|
|
|
$this->physicalStructureInfo[$iiifId]['files'][$fileUseThumbs[0]] = $thumbnailUrl; |
295
|
|
|
} |
296
|
|
|
// populate structural metadata info |
297
|
|
|
$elements[$canvasOrder] = $canvas->getId(); |
298
|
|
|
$this->physicalStructureInfo[$elements[$canvasOrder]]['id'] = $canvas->getId(); |
299
|
|
|
$this->physicalStructureInfo[$elements[$canvasOrder]]['dmdId'] = null; |
300
|
|
|
$this->physicalStructureInfo[$elements[$canvasOrder]]['label'] = $canvas->getLabelForDisplay(); |
301
|
|
|
$this->physicalStructureInfo[$elements[$canvasOrder]]['orderlabel'] = $canvas->getLabelForDisplay(); |
302
|
|
|
// assume that a canvas always represents a page |
303
|
|
|
$this->physicalStructureInfo[$elements[$canvasOrder]]['type'] = 'page'; |
304
|
|
|
$this->physicalStructureInfo[$elements[$canvasOrder]]['contentIds'] = null; |
305
|
|
|
$this->physicalStructureInfo[$elements[$canvasOrder]]['annotationContainers'] = null; |
306
|
|
|
if (!empty($canvas->getPossibleTextAnnotationContainers(Motivation::PAINTING))) { |
307
|
|
|
$this->physicalStructureInfo[$elements[$canvasOrder]]['annotationContainers'] = []; |
308
|
|
|
foreach ($canvas->getPossibleTextAnnotationContainers(Motivation::PAINTING) as $annotationContainer) { |
309
|
|
|
$this->physicalStructureInfo[$elements[$canvasOrder]]['annotationContainers'][] = $annotationContainer->getId(); |
310
|
|
|
if ($extConf['indexAnnotations']) { |
311
|
|
|
$this->hasFulltext = true; |
|
|
|
|
312
|
|
|
$this->hasFulltextSet = true; |
313
|
|
|
} |
314
|
|
|
} |
315
|
|
|
} |
316
|
|
|
|
317
|
|
|
$this->setFileUseFulltext($elements[$canvasOrder], $canvas); |
318
|
|
|
|
319
|
|
|
if (!empty($fileUses)) { |
320
|
|
|
$image = $canvas->getImageAnnotations()[0]; |
321
|
|
|
foreach ($fileUses as $fileUse) { |
322
|
|
|
if ($image->getBody() !== null && $image->getBody() instanceof ContentResourceInterface) { |
323
|
|
|
$this->physicalStructureInfo[$elements[$canvasOrder]]['files'][$fileUse] = $image->getBody()->getId(); |
324
|
|
|
} |
325
|
|
|
} |
326
|
|
|
} |
327
|
|
|
if (!empty($thumbnailUrl)) { |
328
|
|
|
$this->physicalStructureInfo[$elements[$canvasOrder]]['files'][$fileUseThumbs] = $thumbnailUrl; |
329
|
|
|
} |
330
|
|
|
|
331
|
|
|
$this->setFileUseDownload($elements[$canvasOrder], $canvas); |
332
|
|
|
} |
333
|
|
|
$this->numPages = $canvasOrder; |
|
|
|
|
334
|
|
|
// Merge and re-index the array to get nice numeric indexes. |
335
|
|
|
array_unshift($elements, $iiifId); |
336
|
|
|
$this->physicalStructure = $elements; |
|
|
|
|
337
|
|
|
} |
338
|
|
|
$this->physicalStructureLoaded = true; |
339
|
|
|
} |
340
|
|
|
return $this->physicalStructure; |
341
|
|
|
} |
342
|
|
|
|
343
|
|
|
/** |
344
|
|
|
* @see AbstractDocument::getDownloadLocation() |
345
|
|
|
*/ |
346
|
|
|
public function getDownloadLocation(string $id): string |
347
|
|
|
{ |
348
|
|
|
$fileLocation = $this->getFileLocation($id); |
349
|
|
|
$resource = $this->iiif->getContainedResourceById($fileLocation); |
350
|
|
|
if ($resource instanceof AbstractImageService) { |
351
|
|
|
return $resource->getImageUrl(); |
352
|
|
|
} |
353
|
|
|
return $fileLocation; |
354
|
|
|
} |
355
|
|
|
|
356
|
|
|
/** |
357
|
|
|
* @see AbstractDocument::getFileInfo() |
358
|
|
|
*/ |
359
|
|
|
public function getFileInfo($id): ?array |
360
|
|
|
{ |
361
|
|
|
if (empty($this->fileInfos[$id]['location'])) { |
362
|
|
|
$this->fileInfos[$id]['location'] = $this->getFileLocation($id); |
363
|
|
|
} |
364
|
|
|
|
365
|
|
|
if (empty($this->fileInfos[$id]['mimeType'])) { |
366
|
|
|
$this->fileInfos[$id]['mimeType'] = $this->getFileMimeType($id); |
367
|
|
|
} |
368
|
|
|
|
369
|
|
|
return $this->fileInfos[$id] ?? null; |
370
|
|
|
} |
371
|
|
|
|
372
|
|
|
/** |
373
|
|
|
* @see AbstractDocument::getFileLocation() |
374
|
|
|
*/ |
375
|
|
|
public function getFileLocation(string $id): string |
376
|
|
|
{ |
377
|
|
|
if ($id == null) { |
378
|
|
|
return ''; |
379
|
|
|
} |
380
|
|
|
$resource = $this->iiif->getContainedResourceById($id); |
381
|
|
|
if (isset($resource)) { |
382
|
|
|
if ($resource instanceof CanvasInterface) { |
383
|
|
|
// TODO: Cannot call method getSingleService() on array<Ubl\Iiif\Presentation\Common\Model\Resources\AnnotationInterface>. |
384
|
|
|
// @phpstan-ignore-next-line |
385
|
|
|
return (!empty($resource->getImageAnnotations()) && $resource->getImageAnnotations()->getSingleService() != null) ? $resource->getImageAnnotations()[0]->getSingleService()->getId() : $id; |
386
|
|
|
} elseif ($resource instanceof ContentResourceInterface) { |
387
|
|
|
return $resource->getSingleService() instanceof Service ? $resource->getSingleService()->getId() : $id; |
|
|
|
|
388
|
|
|
} elseif ($resource instanceof AbstractImageService) { |
389
|
|
|
return $resource->getId(); |
390
|
|
|
} elseif ($resource instanceof AnnotationContainerInterface) { |
391
|
|
|
return $id; |
392
|
|
|
} |
393
|
|
|
} |
394
|
|
|
return $id; |
395
|
|
|
} |
396
|
|
|
|
397
|
|
|
/** |
398
|
|
|
* @see AbstractDocument::getFileMimeType() |
399
|
|
|
*/ |
400
|
|
|
public function getFileMimeType(string $id): string |
401
|
|
|
{ |
402
|
|
|
$fileResource = $this->iiif->getContainedResourceById($id); |
403
|
|
|
if ($fileResource instanceof CanvasInterface) { |
404
|
|
|
$format = "application/vnd.kitodo.iiif"; |
405
|
|
|
} elseif ($fileResource instanceof AnnotationInterface) { |
406
|
|
|
$format = "application/vnd.kitodo.iiif"; |
407
|
|
|
} elseif ($fileResource instanceof ContentResourceInterface) { |
408
|
|
|
if ($fileResource->isText() || $fileResource->isImage() && ($fileResource->getSingleService() == null || !($fileResource->getSingleService() instanceof AbstractImageService))) { |
409
|
|
|
// Support static images without an image service |
410
|
|
|
return $fileResource->getFormat(); |
411
|
|
|
} |
412
|
|
|
$format = "application/vnd.kitodo.iiif"; |
413
|
|
|
} elseif ($fileResource instanceof AbstractImageService) { |
414
|
|
|
$format = "application/vnd.kitodo.iiif"; |
415
|
|
|
} else { |
416
|
|
|
// Assumptions: this can only be the thumbnail and the thumbnail is a jpeg - TODO determine mimetype |
417
|
|
|
$format = "image/jpeg"; |
418
|
|
|
} |
419
|
|
|
return $format; |
420
|
|
|
} |
421
|
|
|
|
422
|
|
|
/** |
423
|
|
|
* @see AbstractDocument::getLogicalStructure() |
424
|
|
|
*/ |
425
|
|
|
public function getLogicalStructure(string $id, bool $recursive = false): array |
426
|
|
|
{ |
427
|
|
|
$details = []; |
428
|
|
|
if (!$recursive && !empty($this->logicalUnits[$id])) { |
429
|
|
|
return $this->logicalUnits[$id]; |
430
|
|
|
} elseif (!empty($id)) { |
431
|
|
|
$logUnits[] = $this->iiif->getContainedResourceById($id); |
|
|
|
|
432
|
|
|
} else { |
433
|
|
|
$logUnits[] = $this->iiif; |
434
|
|
|
} |
435
|
|
|
// TODO: Variable $logUnits in empty() always exists and is not falsy. |
436
|
|
|
// @phpstan-ignore-next-line |
437
|
|
|
if (!empty($logUnits)) { |
438
|
|
|
if (!$recursive) { |
439
|
|
|
$details = $this->getLogicalStructureInfo($logUnits[0]); |
440
|
|
|
} else { |
441
|
|
|
// cache the ranges - they might occur multiple times in the structures "tree" - with full data as well as referenced as id |
442
|
|
|
$processedStructures = []; |
443
|
|
|
foreach ($logUnits as $logUnit) { |
444
|
|
|
if (array_search($logUnit->getId(), $processedStructures) == false) { |
|
|
|
|
445
|
|
|
$this->tableOfContents[] = $this->getLogicalStructureInfo($logUnit, true, $processedStructures); |
|
|
|
|
446
|
|
|
} |
447
|
|
|
} |
448
|
|
|
} |
449
|
|
|
} |
450
|
|
|
return $details; |
451
|
|
|
} |
452
|
|
|
|
453
|
|
|
/** |
454
|
|
|
* Get the details about a IIIF resource (manifest or range) in the logical structure |
455
|
|
|
* |
456
|
|
|
* @access protected |
457
|
|
|
* |
458
|
|
|
* @param IiifResourceInterface $resource IIIF resource, either a manifest or range. |
459
|
|
|
* @param bool $recursive Whether to include the child elements |
460
|
|
|
* @param array $processedStructures IIIF resources that already have been processed |
461
|
|
|
* |
462
|
|
|
* @return array Logical structure array |
463
|
|
|
*/ |
464
|
|
|
protected function getLogicalStructureInfo(IiifResourceInterface $resource, bool $recursive = false, array &$processedStructures = []): array |
465
|
|
|
{ |
466
|
|
|
$details = []; |
467
|
|
|
$details['id'] = $resource->getId(); |
468
|
|
|
$details['dmdId'] = ''; |
469
|
|
|
$details['label'] = $resource->getLabelForDisplay() ?? ''; |
470
|
|
|
$details['orderlabel'] = $resource->getLabelForDisplay() ?? ''; |
471
|
|
|
$details['contentIds'] = ''; |
472
|
|
|
$details['volume'] = ''; |
473
|
|
|
$details['pagination'] = ''; |
474
|
|
|
$cPid = ($this->cPid ? $this->cPid : $this->pid); |
475
|
|
|
if ($details['id'] == $this->magicGetToplevelId()) { |
476
|
|
|
$metadata = $this->getMetadata($details['id'], $cPid); |
477
|
|
|
if (!empty($metadata['type'][0])) { |
478
|
|
|
$details['type'] = $metadata['type'][0]; |
479
|
|
|
} |
480
|
|
|
} |
481
|
|
|
$details['thumbnailId'] = $resource->getThumbnailUrl(); |
482
|
|
|
$details['points'] = ''; |
483
|
|
|
// Load structural mapping |
484
|
|
|
$this->magicGetSmLinks(); |
485
|
|
|
// Load physical structure. |
486
|
|
|
$this->magicGetPhysicalStructure(); |
487
|
|
|
|
488
|
|
|
if ($resource instanceof ManifestInterface || $resource instanceof RangeInterface) { |
489
|
|
|
$startCanvas = $resource->getStartCanvasOrFirstCanvas(); |
490
|
|
|
} |
491
|
|
|
if (isset($startCanvas)) { |
492
|
|
|
$details['pagination'] = $startCanvas->getLabel(); |
493
|
|
|
$startCanvasIndex = array_search($startCanvas, $this->iiif->getDefaultCanvases()); |
494
|
|
|
if ($startCanvasIndex !== false) { |
495
|
|
|
$details['points'] = $startCanvasIndex + 1; |
496
|
|
|
} |
497
|
|
|
} |
498
|
|
|
$useGroups = $this->getUseGroups('fileGrpImages'); |
499
|
|
|
if (is_string($useGroups)) { |
500
|
|
|
$useGroups = [$useGroups]; |
|
|
|
|
501
|
|
|
} |
502
|
|
|
// Keep for later usage. |
503
|
|
|
$this->logicalUnits[$details['id']] = $details; |
504
|
|
|
// Walk the structure recursively? And are there any children of the current element? |
505
|
|
|
if ($recursive) { |
506
|
|
|
$processedStructures[] = $resource->getId(); |
507
|
|
|
$details['children'] = []; |
508
|
|
|
if ($resource instanceof ManifestInterface && $resource->getRootRanges() != null) { |
509
|
|
|
$rangesToAdd = []; |
510
|
|
|
$rootRanges = []; |
511
|
|
|
if (count($this->iiif->getRootRanges()) == 1 && $this->iiif->getRootRanges()[0]->isTopRange()) { |
512
|
|
|
$rangesToAdd = $this->iiif->getRootRanges()[0]->getMemberRangesAndRanges(); |
|
|
|
|
513
|
|
|
} else { |
514
|
|
|
$rangesToAdd = $this->iiif->getRootRanges(); |
515
|
|
|
} |
516
|
|
|
foreach ($rangesToAdd as $range) { |
517
|
|
|
$rootRanges[] = $range; |
518
|
|
|
} |
519
|
|
|
foreach ($rootRanges as $range) { |
520
|
|
|
if ((array_search($range->getId(), $processedStructures) == false)) { |
|
|
|
|
521
|
|
|
$details['children'][] = $this->getLogicalStructureInfo($range, true, $processedStructures); |
522
|
|
|
} |
523
|
|
|
} |
524
|
|
|
} elseif ($resource instanceof RangeInterface) { |
525
|
|
|
if (!empty($resource->getAllRanges())) { |
526
|
|
|
foreach ($resource->getAllRanges() as $range) { |
527
|
|
|
if ((array_search($range->getId(), $processedStructures) == false)) { |
|
|
|
|
528
|
|
|
$details['children'][] = $this->getLogicalStructureInfo($range, true, $processedStructures); |
529
|
|
|
} |
530
|
|
|
} |
531
|
|
|
} |
532
|
|
|
} |
533
|
|
|
} |
534
|
|
|
return $details; |
535
|
|
|
} |
536
|
|
|
|
537
|
|
|
/** |
538
|
|
|
* Returns metadata for IIIF resources with the ID $id in there original form in |
539
|
|
|
* the manifest, but prepared for display to the user. |
540
|
|
|
* |
541
|
|
|
* @access public |
542
|
|
|
* |
543
|
|
|
* @param string $id the ID of the IIIF resource |
544
|
|
|
* @param bool $withDescription add description / summary to the return value |
545
|
|
|
* @param bool $withRights add attribution and license / rights and requiredStatement to the return value |
546
|
|
|
* @param bool $withRelated add related links / homepage to the return value |
547
|
|
|
* |
548
|
|
|
* @return array |
549
|
|
|
* |
550
|
|
|
* @todo This method is still in experimental; the method signature may change. |
551
|
|
|
*/ |
552
|
|
|
public function getManifestMetadata(string $id, bool $withDescription = true, bool $withRights = true, bool $withRelated = true): array |
553
|
|
|
{ |
554
|
|
|
if (!empty($this->originalMetadataArray[$id])) { |
555
|
|
|
return $this->originalMetadataArray[$id]; |
556
|
|
|
} |
557
|
|
|
$iiifResource = $this->iiif->getContainedResourceById($id); |
558
|
|
|
$result = []; |
559
|
|
|
if ($iiifResource != null) { |
560
|
|
|
if (!empty($iiifResource->getLabel())) { |
561
|
|
|
$result['label'] = $iiifResource->getLabel(); |
562
|
|
|
} |
563
|
|
|
if (!empty($iiifResource->getMetadata())) { |
564
|
|
|
$result['metadata'] = []; |
565
|
|
|
foreach ($iiifResource->getMetadataForDisplay() as $metadata) { |
566
|
|
|
$result['metadata'][$metadata['label']] = $metadata['value']; |
567
|
|
|
} |
568
|
|
|
} |
569
|
|
|
if ($withDescription && !empty($iiifResource->getSummary())) { |
570
|
|
|
$result["description"] = $iiifResource->getSummaryForDisplay(); |
571
|
|
|
} |
572
|
|
|
if ($withRights) { |
573
|
|
|
if (!empty($iiifResource->getRights())) { |
574
|
|
|
$result["rights"] = $iiifResource->getRights(); |
575
|
|
|
} |
576
|
|
|
if (!empty($iiifResource->getRequiredStatement())) { |
577
|
|
|
$result["requiredStatement"] = $iiifResource->getRequiredStatementForDisplay(); |
578
|
|
|
} |
579
|
|
|
} |
580
|
|
|
if ($withRelated && !empty($iiifResource->getWeblinksForDisplay())) { |
581
|
|
|
$result["weblinks"] = []; |
582
|
|
|
foreach ($iiifResource->getWeblinksForDisplay() as $link) { |
583
|
|
|
$key = array_key_exists("label", $link) ? $link["label"] : $link["@id"]; |
584
|
|
|
$result["weblinks"][$key] = $link["@id"]; |
585
|
|
|
} |
586
|
|
|
} |
587
|
|
|
} |
588
|
|
|
return $result; |
589
|
|
|
} |
590
|
|
|
|
591
|
|
|
/** |
592
|
|
|
* @see AbstractDocument::getMetadata() |
593
|
|
|
*/ |
594
|
|
|
public function getMetadata(string $id, int $cPid = 0): array |
595
|
|
|
{ |
596
|
|
|
if (!empty($this->metadataArray[$id]) && $this->metadataArray[0] == $cPid) { |
597
|
|
|
return $this->metadataArray[$id]; |
598
|
|
|
} |
599
|
|
|
|
600
|
|
|
$metadata = $this->initializeMetadata('IIIF'); |
601
|
|
|
|
602
|
|
|
$queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class) |
603
|
|
|
->getQueryBuilderForTable('tx_dlf_metadata'); |
604
|
|
|
// Get hidden records, too. |
605
|
|
|
$queryBuilder |
606
|
|
|
->getRestrictions() |
607
|
|
|
->removeByType(HiddenRestriction::class); |
608
|
|
|
$result = $queryBuilder |
609
|
|
|
->select( |
610
|
|
|
'tx_dlf_metadata.index_name AS index_name', |
611
|
|
|
'tx_dlf_metadataformat.xpath AS xpath', |
612
|
|
|
'tx_dlf_metadataformat.xpath_sorting AS xpath_sorting', |
613
|
|
|
'tx_dlf_metadata.is_sortable AS is_sortable', |
614
|
|
|
'tx_dlf_metadata.default_value AS default_value', |
615
|
|
|
'tx_dlf_metadata.format AS format' |
616
|
|
|
) |
617
|
|
|
->from('tx_dlf_metadata') |
618
|
|
|
->from('tx_dlf_metadataformat') |
619
|
|
|
->from('tx_dlf_formats') |
620
|
|
|
->where( |
621
|
|
|
$queryBuilder->expr()->eq('tx_dlf_metadata.pid', (int) $cPid), |
622
|
|
|
$queryBuilder->expr()->eq('tx_dlf_metadataformat.pid', (int) $cPid), |
623
|
|
|
$queryBuilder->expr()->orX( |
624
|
|
|
$queryBuilder->expr()->andX( |
625
|
|
|
$queryBuilder->expr()->eq('tx_dlf_metadata.uid', 'tx_dlf_metadataformat.parent_id'), |
626
|
|
|
$queryBuilder->expr()->eq('tx_dlf_metadataformat.encoded', 'tx_dlf_formats.uid'), |
627
|
|
|
$queryBuilder->expr()->eq('tx_dlf_formats.type', $queryBuilder->createNamedParameter($this->getIiifVersion())) |
628
|
|
|
), |
629
|
|
|
$queryBuilder->expr()->eq('tx_dlf_metadata.format', 0) |
630
|
|
|
) |
631
|
|
|
) |
632
|
|
|
->execute(); |
633
|
|
|
$iiifResource = $this->iiif->getContainedResourceById($id); |
634
|
|
|
while ($resArray = $result->fetchAssociative()) { |
635
|
|
|
// Set metadata field's value(s). |
636
|
|
|
if ($resArray['format'] > 0 && !empty($resArray['xpath'])) { |
637
|
|
|
$values = $iiifResource->jsonPath($resArray['xpath']); |
638
|
|
|
if (is_string($values)) { |
639
|
|
|
$metadata[$resArray['index_name']] = [trim((string) $values)]; |
640
|
|
|
} elseif ($values instanceof JSONPath && is_array($values->data()) && count($values->data()) > 1) { |
641
|
|
|
$metadata[$resArray['index_name']] = []; |
642
|
|
|
foreach ($values->data() as $value) { |
643
|
|
|
$metadata[$resArray['index_name']][] = trim((string) $value); |
644
|
|
|
} |
645
|
|
|
} |
646
|
|
|
} |
647
|
|
|
// Set default value if applicable. |
648
|
|
|
if (empty($metadata[$resArray['index_name']][0]) && strlen($resArray['default_value']) > 0) { |
649
|
|
|
$metadata[$resArray['index_name']] = [$resArray['default_value']]; |
650
|
|
|
} |
651
|
|
|
// Set sorting value if applicable. |
652
|
|
|
if (!empty($metadata[$resArray['index_name']]) && $resArray['is_sortable']) { |
653
|
|
|
if ($resArray['format'] > 0 && !empty($resArray['xpath_sorting'])) { |
654
|
|
|
$values = $iiifResource->jsonPath($resArray['xpath_sorting']); |
655
|
|
|
if (is_string($values)) { |
656
|
|
|
$metadata[$resArray['index_name'] . '_sorting'][0] = [trim((string) $values)]; |
657
|
|
|
} elseif ($values instanceof JSONPath && is_array($values->data()) && count($values->data()) > 1) { |
658
|
|
|
$metadata[$resArray['index_name']] = []; |
659
|
|
|
foreach ($values->data() as $value) { |
660
|
|
|
$metadata[$resArray['index_name'] . '_sorting'][0] = trim((string) $value); |
661
|
|
|
} |
662
|
|
|
} |
663
|
|
|
} |
664
|
|
|
if (empty($metadata[$resArray['index_name'] . '_sorting'][0])) { |
665
|
|
|
$metadata[$resArray['index_name'] . '_sorting'][0] = $metadata[$resArray['index_name']][0]; |
666
|
|
|
} |
667
|
|
|
} |
668
|
|
|
} |
669
|
|
|
// Set date to empty string if not present. |
670
|
|
|
if (empty($metadata['date'][0])) { |
671
|
|
|
$metadata['date'][0] = ''; |
672
|
|
|
} |
673
|
|
|
return $metadata; |
674
|
|
|
} |
675
|
|
|
|
676
|
|
|
/** |
677
|
|
|
* @see AbstractDocument::magicGetSmLinks() |
678
|
|
|
*/ |
679
|
|
|
protected function magicGetSmLinks(): array |
680
|
|
|
{ |
681
|
|
|
if (!$this->smLinksLoaded && isset($this->iiif) && $this->iiif instanceof ManifestInterface) { |
682
|
|
|
if (!empty($this->iiif->getDefaultCanvases())) { |
683
|
|
|
foreach ($this->iiif->getDefaultCanvases() as $canvas) { |
684
|
|
|
$this->smLinkCanvasToResource($canvas, $this->iiif); |
685
|
|
|
} |
686
|
|
|
} |
687
|
|
|
if (!empty($this->iiif->getStructures())) { |
688
|
|
|
foreach ($this->iiif->getStructures() as $range) { |
689
|
|
|
$this->smLinkRangeCanvasesRecursively($range); |
690
|
|
|
} |
691
|
|
|
} |
692
|
|
|
$this->smLinksLoaded = true; |
693
|
|
|
} |
694
|
|
|
return $this->smLinks; |
695
|
|
|
} |
696
|
|
|
|
697
|
|
|
/** |
698
|
|
|
* Construct a link between a range and it's sub ranges and all contained canvases. |
699
|
|
|
* |
700
|
|
|
* @access private |
701
|
|
|
* |
702
|
|
|
* @param RangeInterface $range Current range whose canvases shall be linked |
703
|
|
|
* |
704
|
|
|
* @return void |
705
|
|
|
*/ |
706
|
|
|
private function smLinkRangeCanvasesRecursively(RangeInterface $range): void |
707
|
|
|
{ |
708
|
|
|
// map range's canvases including all child ranges' canvases |
709
|
|
|
if (!$range->isTopRange()) { |
710
|
|
|
foreach ($range->getAllCanvasesRecursively() as $canvas) { |
711
|
|
|
$this->smLinkCanvasToResource($canvas, $range); |
712
|
|
|
} |
713
|
|
|
} |
714
|
|
|
// recursive call for all ranges |
715
|
|
|
if (!empty($range->getAllRanges())) { |
716
|
|
|
foreach ($range->getAllRanges() as $childRange) { |
717
|
|
|
$this->smLinkRangeCanvasesRecursively($childRange); |
718
|
|
|
} |
719
|
|
|
} |
720
|
|
|
} |
721
|
|
|
|
722
|
|
|
/** |
723
|
|
|
* Link a single canvas to a containing range |
724
|
|
|
* |
725
|
|
|
* @access private |
726
|
|
|
* |
727
|
|
|
* @param CanvasInterface $canvas |
728
|
|
|
* @param IiifResourceInterface $resource |
729
|
|
|
* |
730
|
|
|
* @return void |
731
|
|
|
*/ |
732
|
|
|
private function smLinkCanvasToResource(CanvasInterface $canvas, IiifResourceInterface $resource): void |
733
|
|
|
{ |
734
|
|
|
$this->smLinks['l2p'][$resource->getId()][] = $canvas->getId(); |
|
|
|
|
735
|
|
|
if (!is_array($this->smLinks['p2l'][$canvas->getId()]) || !in_array($resource->getId(), $this->smLinks['p2l'][$canvas->getId()])) { |
736
|
|
|
$this->smLinks['p2l'][$canvas->getId()][] = $resource->getId(); |
737
|
|
|
} |
738
|
|
|
} |
739
|
|
|
|
740
|
|
|
/** |
741
|
|
|
* @see AbstractDocument::getFullText() |
742
|
|
|
*/ |
743
|
|
|
//TODO: rewrite it to get full OCR |
744
|
|
|
public function getFullText(string $id): string |
745
|
|
|
{ |
746
|
|
|
$rawText = ''; |
747
|
|
|
// Get text from raw text array if available. |
748
|
|
|
if (!empty($this->rawTextArray[$id])) { |
749
|
|
|
return $this->rawTextArray[$id]; |
750
|
|
|
} |
751
|
|
|
$this->ensureHasFulltextIsSet(); |
752
|
|
|
if ($this->hasFulltext) { |
753
|
|
|
// Load physical structure ... |
754
|
|
|
$this->magicGetPhysicalStructure(); |
755
|
|
|
// ... and extension configuration. |
756
|
|
|
$extConf = GeneralUtility::makeInstance(ExtensionConfiguration::class)->get(self::$extKey); |
757
|
|
|
$fileGrpsFulltext = GeneralUtility::trimExplode(',', $extConf['files']['fileGrpFulltext']); |
758
|
|
|
if (!empty($this->physicalStructureInfo[$id])) { |
759
|
|
|
while ($fileGrpFulltext = array_shift($fileGrpsFulltext)) { |
760
|
|
|
if (!empty($this->physicalStructureInfo[$id]['files'][$fileGrpFulltext])) { |
761
|
|
|
$rawText = parent::getFullTextFromXml($id); |
762
|
|
|
break; |
763
|
|
|
} |
764
|
|
|
} |
765
|
|
|
if ($extConf['iiif']['indexAnnotations'] == 1) { |
766
|
|
|
$iiifResource = $this->iiif->getContainedResourceById($id); |
767
|
|
|
// Get annotation containers |
768
|
|
|
$annotationContainerIds = $this->physicalStructureInfo[$id]['annotationContainers']; |
769
|
|
|
if (!empty($annotationContainerIds)) { |
770
|
|
|
$annotationTexts = $this->getAnnotationTexts($annotationContainerIds, $iiifResource->getId()); |
771
|
|
|
$rawText .= implode(' ', $annotationTexts); |
772
|
|
|
} |
773
|
|
|
} |
774
|
|
|
} else { |
775
|
|
|
$this->logger->warning('Invalid structure resource @id "' . $id . '"'); |
776
|
|
|
return $rawText; |
777
|
|
|
} |
778
|
|
|
$this->rawTextArray[$id] = $rawText; |
779
|
|
|
} |
780
|
|
|
return $rawText; |
781
|
|
|
} |
782
|
|
|
|
783
|
|
|
/** |
784
|
|
|
* Returns the underlying IiifResourceInterface. |
785
|
|
|
* |
786
|
|
|
* @access public |
787
|
|
|
* |
788
|
|
|
* @return IiifResourceInterface |
789
|
|
|
*/ |
790
|
|
|
public function getIiif(): IiifResourceInterface |
791
|
|
|
{ |
792
|
|
|
return $this->iiif; |
793
|
|
|
} |
794
|
|
|
|
795
|
|
|
/** |
796
|
|
|
* @see AbstractDocument::init() |
797
|
|
|
*/ |
798
|
|
|
protected function init(string $location, array $settings = []): void |
799
|
|
|
{ |
800
|
|
|
$this->logger = GeneralUtility::makeInstance(LogManager::class)->getLogger(static::class); |
801
|
|
|
} |
802
|
|
|
|
803
|
|
|
/** |
804
|
|
|
* @see AbstractDocument::loadLocation() |
805
|
|
|
*/ |
806
|
|
|
protected function loadLocation(string $location): bool |
807
|
|
|
{ |
808
|
|
|
$fileResource = GeneralUtility::getUrl($location); |
809
|
|
|
if ($fileResource !== false) { |
810
|
|
|
$conf = GeneralUtility::makeInstance(ExtensionConfiguration::class)->get(self::$extKey, 'iiif'); |
811
|
|
|
IiifHelper::setUrlReader(IiifUrlReader::getInstance()); |
812
|
|
|
IiifHelper::setMaxThumbnailHeight($conf['thumbnailHeight']); |
813
|
|
|
IiifHelper::setMaxThumbnailWidth($conf['thumbnailWidth']); |
814
|
|
|
$resource = IiifHelper::loadIiifResource($fileResource); |
815
|
|
|
if ($resource instanceof ManifestInterface) { |
816
|
|
|
$this->iiif = $resource; |
817
|
|
|
return true; |
818
|
|
|
} |
819
|
|
|
} |
820
|
|
|
$this->logger->error('Could not load IIIF manifest from "' . $location . '"'); |
821
|
|
|
return false; |
822
|
|
|
} |
823
|
|
|
|
824
|
|
|
/** |
825
|
|
|
* @see AbstractDocument::prepareMetadataArray() |
826
|
|
|
*/ |
827
|
|
|
protected function prepareMetadataArray(int $cPid): void |
828
|
|
|
{ |
829
|
|
|
$id = $this->iiif->getId(); |
830
|
|
|
$this->metadataArray[(string) $id] = $this->getMetadata((string) $id, $cPid); |
|
|
|
|
831
|
|
|
} |
832
|
|
|
|
833
|
|
|
/** |
834
|
|
|
* @see AbstractDocument::setPreloadedDocument() |
835
|
|
|
*/ |
836
|
|
|
protected function setPreloadedDocument($preloadedDocument): bool |
837
|
|
|
{ |
838
|
|
|
if ($preloadedDocument instanceof ManifestInterface) { |
839
|
|
|
$this->iiif = $preloadedDocument; |
840
|
|
|
return true; |
841
|
|
|
} |
842
|
|
|
return false; |
843
|
|
|
} |
844
|
|
|
|
845
|
|
|
/** |
846
|
|
|
* @see AbstractDocument::ensureHasFulltextIsSet() |
847
|
|
|
*/ |
848
|
|
|
protected function ensureHasFulltextIsSet(): void |
849
|
|
|
{ |
850
|
|
|
/* |
851
|
|
|
* TODO Check annotations and annotation lists of canvas for ALTO documents. |
852
|
|
|
* Example: |
853
|
|
|
* https://digi.ub.uni-heidelberg.de/diglit/iiif/hirsch_hamburg1933_04_25/manifest.json links |
854
|
|
|
* https://digi.ub.uni-heidelberg.de/diglit/iiif/hirsch_hamburg1933_04_25/list/0001.json |
855
|
|
|
*/ |
856
|
|
|
if (!$this->hasFulltextSet && $this->iiif instanceof ManifestInterface) { |
857
|
|
|
$manifest = $this->iiif; |
858
|
|
|
$canvases = $manifest->getDefaultCanvases(); |
859
|
|
|
foreach ($canvases as $canvas) { |
860
|
|
|
if ( |
861
|
|
|
!empty($canvas->getSeeAlsoUrlsForFormat("application/alto+xml")) || |
862
|
|
|
!empty($canvas->getSeeAlsoUrlsForProfile("http://www.loc.gov/standards/alto/")) |
863
|
|
|
) { |
864
|
|
|
$this->hasFulltextSet = true; |
865
|
|
|
$this->hasFulltext = true; |
|
|
|
|
866
|
|
|
return; |
867
|
|
|
} |
868
|
|
|
$extConf = GeneralUtility::makeInstance(ExtensionConfiguration::class)->get(self::$extKey, 'iiif'); |
869
|
|
|
if ($extConf['indexAnnotations'] == 1 && !empty($canvas->getPossibleTextAnnotationContainers(Motivation::PAINTING))) { |
870
|
|
|
foreach ($canvas->getPossibleTextAnnotationContainers(Motivation::PAINTING) as $annotationContainer) { |
871
|
|
|
$textAnnotations = $annotationContainer->getTextAnnotations(Motivation::PAINTING); |
872
|
|
|
if ($textAnnotations != null) { |
873
|
|
|
foreach ($textAnnotations as $annotation) { |
874
|
|
|
if ( |
875
|
|
|
$annotation->getBody() != null && |
876
|
|
|
$annotation->getBody()->getFormat() == "text/plain" && |
877
|
|
|
$annotation->getBody()->getChars() != null |
878
|
|
|
) { |
879
|
|
|
$this->hasFulltextSet = true; |
880
|
|
|
$this->hasFulltext = true; |
881
|
|
|
return; |
882
|
|
|
} |
883
|
|
|
} |
884
|
|
|
} |
885
|
|
|
} |
886
|
|
|
} |
887
|
|
|
} |
888
|
|
|
$this->hasFulltextSet = true; |
889
|
|
|
} |
890
|
|
|
} |
891
|
|
|
|
892
|
|
|
/** |
893
|
|
|
* @see AbstractDocument::magicGetThumbnail() |
894
|
|
|
*/ |
895
|
|
|
protected function magicGetThumbnail(bool $forceReload = false): string |
896
|
|
|
{ |
897
|
|
|
return $this->iiif->getThumbnailUrl(); |
898
|
|
|
} |
899
|
|
|
|
900
|
|
|
/** |
901
|
|
|
* @see AbstractDocument::magicGetToplevelId() |
902
|
|
|
*/ |
903
|
|
|
protected function magicGetToplevelId(): string |
904
|
|
|
{ |
905
|
|
|
if (empty($this->toplevelId)) { |
906
|
|
|
if (isset($this->iiif)) { |
907
|
|
|
$this->toplevelId = $this->iiif->getId(); |
|
|
|
|
908
|
|
|
} |
909
|
|
|
} |
910
|
|
|
return $this->toplevelId; |
911
|
|
|
} |
912
|
|
|
|
913
|
|
|
/** |
914
|
|
|
* Get annotation texts. |
915
|
|
|
* |
916
|
|
|
* @access private |
917
|
|
|
* |
918
|
|
|
* @param array $annotationContainerIds |
919
|
|
|
* @param string $iiifId |
920
|
|
|
* |
921
|
|
|
* @return array |
922
|
|
|
*/ |
923
|
|
|
private function getAnnotationTexts($annotationContainerIds, $iiifId): array |
924
|
|
|
{ |
925
|
|
|
$annotationTexts = []; |
926
|
|
|
foreach ($annotationContainerIds as $annotationListId) { |
927
|
|
|
$annotationContainer = $this->iiif->getContainedResourceById($annotationListId); |
928
|
|
|
/* @var $annotationContainer \Ubl\Iiif\Presentation\Common\Model\Resources\AnnotationContainerInterface */ |
929
|
|
|
foreach ($annotationContainer->getTextAnnotations(Motivation::PAINTING) as $annotation) { |
|
|
|
|
930
|
|
|
if ( |
931
|
|
|
$annotation->getTargetResourceId() == $iiifId && |
932
|
|
|
$annotation->getBody() != null && $annotation->getBody()->getChars() != null |
933
|
|
|
) { |
934
|
|
|
$annotationTexts[] = $annotation->getBody()->getChars(); |
935
|
|
|
} |
936
|
|
|
} |
937
|
|
|
} |
938
|
|
|
return $annotationTexts; |
939
|
|
|
} |
940
|
|
|
|
941
|
|
|
/** |
942
|
|
|
* Set files used for download (PDF). |
943
|
|
|
* |
944
|
|
|
* @access private |
945
|
|
|
* |
946
|
|
|
* @param string $iiifId |
947
|
|
|
* @param IiifResourceInterface $iiif |
948
|
|
|
* |
949
|
|
|
* @return void |
950
|
|
|
*/ |
951
|
|
|
private function setFileUseDownload(string $iiifId, $iiif): void |
952
|
|
|
{ |
953
|
|
|
$fileUseDownload = $this->getUseGroups('fileGrpDownload'); |
954
|
|
|
|
955
|
|
|
if (!empty($fileUseDownload)) { |
956
|
|
|
$docPdfRendering = $iiif->getRenderingUrlsForFormat('application/pdf'); |
957
|
|
|
if (!empty($docPdfRendering)) { |
958
|
|
|
$this->physicalStructureInfo[$iiifId]['files'][$fileUseDownload[0]] = $docPdfRendering[0]; |
|
|
|
|
959
|
|
|
} |
960
|
|
|
} |
961
|
|
|
} |
962
|
|
|
|
963
|
|
|
/** |
964
|
|
|
* Set files used for full text (ALTO). |
965
|
|
|
* |
966
|
|
|
* @access private |
967
|
|
|
* |
968
|
|
|
* @param string $iiifId |
969
|
|
|
* @param IiifResourceInterface $iiif |
970
|
|
|
* |
971
|
|
|
* @return void |
972
|
|
|
*/ |
973
|
|
|
private function setFileUseFulltext(string $iiifId, $iiif): void |
974
|
|
|
{ |
975
|
|
|
$fileUseFulltext = $this->getUseGroups('fileGrpFulltext'); |
976
|
|
|
|
977
|
|
|
if (!empty($fileUseFulltext)) { |
978
|
|
|
$alto = $iiif->getSeeAlsoUrlsForFormat('application/alto+xml'); |
979
|
|
|
if (empty($alto)) { |
980
|
|
|
$alto = $iiif->getSeeAlsoUrlsForProfile('http://www.loc.gov/standards/alto/', true); |
981
|
|
|
} |
982
|
|
|
if (!empty($alto)) { |
983
|
|
|
$this->mimeTypes[$alto[0]] = 'application/alto+xml'; |
984
|
|
|
$this->physicalStructureInfo[$iiifId]['files'][$fileUseFulltext[0]] = $alto[0]; |
|
|
|
|
985
|
|
|
$this->hasFulltext = true; |
|
|
|
|
986
|
|
|
$this->hasFulltextSet = true; |
987
|
|
|
} |
988
|
|
|
} |
989
|
|
|
} |
990
|
|
|
|
991
|
|
|
/** |
992
|
|
|
* This magic method is executed after the object is deserialized |
993
|
|
|
* @see __sleep() |
994
|
|
|
* |
995
|
|
|
* @access public |
996
|
|
|
* |
997
|
|
|
* @return void |
998
|
|
|
*/ |
999
|
|
|
public function __wakeup(): void |
1000
|
|
|
{ |
1001
|
|
|
$conf = GeneralUtility::makeInstance(ExtensionConfiguration::class)->get(self::$extKey, 'iiif'); |
1002
|
|
|
IiifHelper::setUrlReader(IiifUrlReader::getInstance()); |
1003
|
|
|
IiifHelper::setMaxThumbnailHeight($conf['thumbnailHeight']); |
1004
|
|
|
IiifHelper::setMaxThumbnailWidth($conf['thumbnailWidth']); |
1005
|
|
|
$resource = IiifHelper::loadIiifResource($this->asJson); |
1006
|
|
|
if ($resource instanceof ManifestInterface) { |
1007
|
|
|
$this->asJson = ''; |
1008
|
|
|
$this->iiif = $resource; |
1009
|
|
|
$this->init(''); |
1010
|
|
|
} else { |
1011
|
|
|
$this->logger = GeneralUtility::makeInstance(LogManager::class)->getLogger(static::class); |
1012
|
|
|
$this->logger->error('Could not load IIIF after deserialization'); |
1013
|
|
|
} |
1014
|
|
|
} |
1015
|
|
|
|
1016
|
|
|
/** |
1017
|
|
|
* @access public |
1018
|
|
|
* |
1019
|
|
|
* @return string[] |
1020
|
|
|
*/ |
1021
|
|
|
public function __sleep(): array |
1022
|
|
|
{ |
1023
|
|
|
// TODO implement serialization in IIIF library |
1024
|
|
|
$jsonArray = $this->iiif->getOriginalJsonArray(); |
1025
|
|
|
$this->asJson = json_encode($jsonArray); |
1026
|
|
|
return ['pid', 'recordId', 'parentId', 'asJson']; |
1027
|
|
|
} |
1028
|
|
|
} |
1029
|
|
|
|