|
1
|
|
|
<?php |
|
2
|
|
|
|
|
3
|
|
|
namespace Kitodo\Dlf\Common; |
|
4
|
|
|
|
|
5
|
|
|
/** |
|
6
|
|
|
* (c) Kitodo. Key to digital objects e.V. <[email protected]> |
|
7
|
|
|
* |
|
8
|
|
|
* This file is part of the Kitodo and TYPO3 projects. |
|
9
|
|
|
* |
|
10
|
|
|
* @license GNU General Public License version 3 or later. |
|
11
|
|
|
* For the full copyright and license information, please read the |
|
12
|
|
|
* LICENSE.txt file that was distributed with this source code. |
|
13
|
|
|
*/ |
|
14
|
|
|
|
|
15
|
|
|
use TYPO3\CMS\Core\Database\ConnectionPool; |
|
16
|
|
|
use TYPO3\CMS\Core\Utility\GeneralUtility; |
|
17
|
|
|
use TYPO3\CMS\Core\Database\Query\QueryBuilder; |
|
18
|
|
|
use Ubl\Iiif\Tools\IiifHelper; |
|
19
|
|
|
use Ubl\Iiif\Services\AbstractImageService; |
|
20
|
|
|
|
|
21
|
|
|
/** |
|
22
|
|
|
* MetsDocument class for the 'dlf' extension. |
|
23
|
|
|
* |
|
24
|
|
|
* @author Sebastian Meyer <[email protected]> |
|
25
|
|
|
* @author Henrik Lochmann <[email protected]> |
|
26
|
|
|
* @package TYPO3 |
|
27
|
|
|
* @subpackage tx_dlf |
|
28
|
|
|
* @access public |
|
29
|
|
|
* @property-write integer $cPid This holds the PID for the configuration |
|
30
|
|
|
* @property-read array $dmdSec This holds the XML file's dmdSec parts with their IDs as array key |
|
31
|
|
|
* @property-read array $fileGrps This holds the file ID -> USE concordance |
|
32
|
|
|
* @property-read boolean $hasFulltext Are there any fulltext files available? |
|
33
|
|
|
* @property-read string $location This holds the documents location |
|
34
|
|
|
* @property-read array $metadataArray This holds the documents' parsed metadata array |
|
35
|
|
|
* @property-read \SimpleXMLElement $mets This holds the XML file's METS part as \SimpleXMLElement object |
|
36
|
|
|
* @property-read integer $numPages The holds the total number of pages |
|
37
|
|
|
* @property-read integer $parentId This holds the UID of the parent document or zero if not multi-volumed |
|
38
|
|
|
* @property-read array $physicalStructure This holds the physical structure |
|
39
|
|
|
* @property-read array $physicalStructureInfo This holds the physical structure metadata |
|
40
|
|
|
* @property-read integer $pid This holds the PID of the document or zero if not in database |
|
41
|
|
|
* @property-read boolean $ready Is the document instantiated successfully? |
|
42
|
|
|
* @property-read string $recordId The METS file's / IIIF manifest's record identifier |
|
43
|
|
|
* @property-read integer $rootId This holds the UID of the root document or zero if not multi-volumed |
|
44
|
|
|
* @property-read array $smLinks This holds the smLinks between logical and physical structMap |
|
45
|
|
|
* @property-read array $tableOfContents This holds the logical structure |
|
46
|
|
|
* @property-read string $thumbnail This holds the document's thumbnail location |
|
47
|
|
|
* @property-read string $toplevelId This holds the toplevel structure's @ID (METS) or the manifest's @id (IIIF) |
|
48
|
|
|
* @property-read mixed $uid This holds the UID or the URL of the document |
|
49
|
|
|
*/ |
|
50
|
|
|
final class MetsDocument extends Document |
|
51
|
|
|
{ |
|
52
|
|
|
/** |
|
53
|
|
|
* This holds the whole XML file as string for serialization purposes |
|
54
|
|
|
* @see __sleep() / __wakeup() |
|
55
|
|
|
* |
|
56
|
|
|
* @var string |
|
57
|
|
|
* @access protected |
|
58
|
|
|
*/ |
|
59
|
|
|
protected $asXML = ''; |
|
60
|
|
|
|
|
61
|
|
|
/** |
|
62
|
|
|
* This holds the XML file's dmdSec parts with their IDs as array key |
|
63
|
|
|
* |
|
64
|
|
|
* @var array |
|
65
|
|
|
* @access protected |
|
66
|
|
|
*/ |
|
67
|
|
|
protected $dmdSec = []; |
|
68
|
|
|
|
|
69
|
|
|
/** |
|
70
|
|
|
* Are the METS file's dmdSecs loaded? |
|
71
|
|
|
* @see $dmdSec |
|
72
|
|
|
* |
|
73
|
|
|
* @var boolean |
|
74
|
|
|
* @access protected |
|
75
|
|
|
*/ |
|
76
|
|
|
protected $dmdSecLoaded = FALSE; |
|
77
|
|
|
|
|
78
|
|
|
/** |
|
79
|
|
|
* The extension key |
|
80
|
|
|
* |
|
81
|
|
|
* @var string |
|
82
|
|
|
* @access public |
|
83
|
|
|
*/ |
|
84
|
|
|
public static $extKey = 'dlf'; |
|
85
|
|
|
|
|
86
|
|
|
/** |
|
87
|
|
|
* This holds the file ID -> USE concordance |
|
88
|
|
|
* @see _getFileGrps() |
|
89
|
|
|
* |
|
90
|
|
|
* @var array |
|
91
|
|
|
* @access protected |
|
92
|
|
|
*/ |
|
93
|
|
|
protected $fileGrps = []; |
|
94
|
|
|
|
|
95
|
|
|
/** |
|
96
|
|
|
* Are the file groups loaded? |
|
97
|
|
|
* @see $fileGrps |
|
98
|
|
|
* |
|
99
|
|
|
* @var boolean |
|
100
|
|
|
* @access protected |
|
101
|
|
|
*/ |
|
102
|
|
|
protected $fileGrpsLoaded = FALSE; |
|
103
|
|
|
|
|
104
|
|
|
/** |
|
105
|
|
|
* This holds the XML file's METS part as \SimpleXMLElement object |
|
106
|
|
|
* |
|
107
|
|
|
* @var \SimpleXMLElement |
|
108
|
|
|
* @access protected |
|
109
|
|
|
*/ |
|
110
|
|
|
protected $mets; |
|
111
|
|
|
|
|
112
|
|
|
/** |
|
113
|
|
|
* This holds the whole XML file as \SimpleXMLElement object |
|
114
|
|
|
* |
|
115
|
|
|
* @var \SimpleXMLElement |
|
116
|
|
|
* @access protected |
|
117
|
|
|
*/ |
|
118
|
|
|
protected $xml; |
|
119
|
|
|
|
|
120
|
|
|
/** |
|
121
|
|
|
* This adds metadata from METS structural map to metadata array. |
|
122
|
|
|
* |
|
123
|
|
|
* @access public |
|
124
|
|
|
* |
|
125
|
|
|
* @param array &$metadata: The metadata array to extend |
|
126
|
|
|
* @param string $id: The @ID attribute of the logical structure node |
|
127
|
|
|
* |
|
128
|
|
|
* @return void |
|
129
|
|
|
*/ |
|
130
|
|
|
public function addMetadataFromMets(&$metadata, $id) |
|
131
|
|
|
{ |
|
132
|
|
|
$details = $this->getLogicalStructure($id); |
|
133
|
|
|
if (!empty($details)) { |
|
134
|
|
|
$metadata['mets_label'][0] = $details['label']; |
|
135
|
|
|
$metadata['mets_orderlabel'][0] = $details['orderlabel']; |
|
136
|
|
|
} |
|
137
|
|
|
} |
|
138
|
|
|
|
|
139
|
|
|
/** |
|
140
|
|
|
* |
|
141
|
|
|
* {@inheritDoc} |
|
142
|
|
|
* @see \Kitodo\Dlf\Common\Document::establishRecordId() |
|
143
|
|
|
*/ |
|
144
|
|
|
protected function establishRecordId($pid) |
|
145
|
|
|
{ |
|
146
|
|
|
// Check for METS object @ID. |
|
147
|
|
|
if (!empty($this->mets['OBJID'])) { |
|
148
|
|
|
$this->recordId = (string) $this->mets['OBJID']; |
|
149
|
|
|
} |
|
150
|
|
|
// Get hook objects. |
|
151
|
|
|
$hookObjects = Helper::getHookObjects('Classes/Common/MetsDocument.php'); |
|
152
|
|
|
// Apply hooks. |
|
153
|
|
|
foreach ($hookObjects as $hookObj) { |
|
154
|
|
|
if (method_exists($hookObj, 'construct_postProcessRecordId')) { |
|
155
|
|
|
$hookObj->construct_postProcessRecordId($this->xml, $this->recordId); |
|
156
|
|
|
} |
|
157
|
|
|
} |
|
158
|
|
|
} |
|
159
|
|
|
|
|
160
|
|
|
/** |
|
161
|
|
|
* |
|
162
|
|
|
* {@inheritDoc} |
|
163
|
|
|
* @see \Kitodo\Dlf\Common\Document::getDownloadLocation() |
|
164
|
|
|
*/ |
|
165
|
|
|
public function getDownloadLocation($id) |
|
166
|
|
|
{ |
|
167
|
|
|
$fileMimeType = $this->getFileMimeType($id); |
|
168
|
|
|
$fileLocation = $this->getFileLocation($id); |
|
169
|
|
|
if ($fileMimeType == 'application/vnd.kitodo.iiif') { |
|
170
|
|
|
$fileLocation = strrpos($fileLocation, "info.json") == strlen($fileLocation) - 9 ? $fileLocation : strrpos($fileLocation, "/") == strlen($fileLocation) ? $fileLocation . "info.json" : $fileLocation . "/info.json"; |
|
171
|
|
|
$conf = unserialize($GLOBALS['TYPO3_CONF_VARS']['EXT']['extConf'][self::$extKey]); |
|
172
|
|
|
IiifHelper::setUrlReader(IiifUrlReader::getInstance()); |
|
173
|
|
|
IiifHelper::setMaxThumbnailHeight($conf['iiifThumbnailHeight']); |
|
174
|
|
|
IiifHelper::setMaxThumbnailWidth($conf['iiifThumbnailWidth']); |
|
175
|
|
|
$service = IiifHelper::loadIiifResource($fileLocation); |
|
176
|
|
|
if ($service != NULL && $service instanceof AbstractImageService) { |
|
177
|
|
|
return $service->getImageUrl(); |
|
178
|
|
|
} |
|
179
|
|
|
} elseif ($fileMimeType == 'application/vnd.netfpx') { |
|
180
|
|
|
$baseURL = $fileLocation . (strpos($fileLocation, "?") === FALSE ? "?" : ""); |
|
181
|
|
|
// TODO CVT is an optional IIP server capability; in theory, capabilities should be determined in the object request with '&obj=IIP-server' |
|
182
|
|
|
return $baseURL . "&CVT=jpeg"; |
|
183
|
|
|
} |
|
184
|
|
|
return $fileLocation; |
|
185
|
|
|
} |
|
186
|
|
|
|
|
187
|
|
|
/** |
|
188
|
|
|
* {@inheritDoc} |
|
189
|
|
|
* @see \Kitodo\Dlf\Common\Document::getFileLocation() |
|
190
|
|
|
*/ |
|
191
|
|
|
public function getFileLocation($id) |
|
192
|
|
|
{ |
|
193
|
|
|
$location = $this->mets->xpath('./mets:fileSec/mets:fileGrp/mets:file[@ID="' . $id . '"]/mets:FLocat[@LOCTYPE="URL"]'); |
|
194
|
|
|
if ( |
|
195
|
|
|
!empty($id) |
|
196
|
|
|
&& !empty($location) |
|
197
|
|
|
) { |
|
198
|
|
|
return (string) $location[0]->attributes('http://www.w3.org/1999/xlink')->href; |
|
199
|
|
|
} else { |
|
200
|
|
|
Helper::devLog('There is no file node with @ID "' . $id . '"', DEVLOG_SEVERITY_WARNING); |
|
201
|
|
|
return ''; |
|
202
|
|
|
} |
|
203
|
|
|
} |
|
204
|
|
|
|
|
205
|
|
|
/** |
|
206
|
|
|
* {@inheritDoc} |
|
207
|
|
|
* @see \Kitodo\Dlf\Common\Document::getFileMimeType() |
|
208
|
|
|
*/ |
|
209
|
|
|
public function getFileMimeType($id) |
|
210
|
|
|
{ |
|
211
|
|
|
$mimetype = $this->mets->xpath('./mets:fileSec/mets:fileGrp/mets:file[@ID="' . $id . '"]/@MIMETYPE'); |
|
212
|
|
|
if ( |
|
213
|
|
|
!empty($id) |
|
214
|
|
|
&& !empty($mimetype) |
|
215
|
|
|
) { |
|
216
|
|
|
return (string) $mimetype[0]; |
|
217
|
|
|
} else { |
|
218
|
|
|
Helper::devLog('There is no file node with @ID "' . $id . '" or no MIME type specified', DEVLOG_SEVERITY_WARNING); |
|
219
|
|
|
return ''; |
|
220
|
|
|
} |
|
221
|
|
|
} |
|
222
|
|
|
|
|
223
|
|
|
/** |
|
224
|
|
|
* {@inheritDoc} |
|
225
|
|
|
* @see \Kitodo\Dlf\Common\Document::getLogicalStructure() |
|
226
|
|
|
*/ |
|
227
|
|
|
public function getLogicalStructure($id, $recursive = FALSE) |
|
228
|
|
|
{ |
|
229
|
|
|
$details = []; |
|
230
|
|
|
// Is the requested logical unit already loaded? |
|
231
|
|
|
if ( |
|
232
|
|
|
!$recursive |
|
233
|
|
|
&& !empty($this->logicalUnits[$id]) |
|
234
|
|
|
) { |
|
235
|
|
|
// Yes. Return it. |
|
236
|
|
|
return $this->logicalUnits[$id]; |
|
237
|
|
|
} elseif (!empty($id)) { |
|
238
|
|
|
// Get specified logical unit. |
|
239
|
|
|
$divs = $this->mets->xpath('./mets:structMap[@TYPE="LOGICAL"]//mets:div[@ID="' . $id . '"]'); |
|
240
|
|
|
} else { |
|
241
|
|
|
// Get all logical units at top level. |
|
242
|
|
|
$divs = $this->mets->xpath('./mets:structMap[@TYPE="LOGICAL"]/mets:div'); |
|
243
|
|
|
} |
|
244
|
|
|
if (!empty($divs)) { |
|
245
|
|
|
if (!$recursive) { |
|
246
|
|
|
// Get the details for the first xpath hit. |
|
247
|
|
|
$details = $this->getLogicalStructureInfo($divs[0]); |
|
248
|
|
|
} else { |
|
249
|
|
|
// Walk the logical structure recursively and fill the whole table of contents. |
|
250
|
|
|
foreach ($divs as $div) { |
|
251
|
|
|
$this->tableOfContents[] = $this->getLogicalStructureInfo($div, TRUE); |
|
252
|
|
|
} |
|
253
|
|
|
} |
|
254
|
|
|
} |
|
255
|
|
|
return $details; |
|
256
|
|
|
} |
|
257
|
|
|
|
|
258
|
|
|
/** |
|
259
|
|
|
* This gets details about a logical structure element |
|
260
|
|
|
* |
|
261
|
|
|
* @access protected |
|
262
|
|
|
* |
|
263
|
|
|
* @param \SimpleXMLElement $structure: The logical structure node |
|
264
|
|
|
* @param boolean $recursive: Whether to include the child elements |
|
265
|
|
|
* |
|
266
|
|
|
* @return array Array of the element's id, label, type and physical page indexes/mptr link |
|
267
|
|
|
*/ |
|
268
|
|
|
protected function getLogicalStructureInfo(\SimpleXMLElement $structure, $recursive = FALSE) |
|
269
|
|
|
{ |
|
270
|
|
|
// Get attributes. |
|
271
|
|
|
foreach ($structure->attributes() as $attribute => $value) { |
|
272
|
|
|
$attributes[$attribute] = (string) $value; |
|
273
|
|
|
} |
|
274
|
|
|
// Load plugin configuration. |
|
275
|
|
|
$extConf = unserialize($GLOBALS['TYPO3_CONF_VARS']['EXT']['extConf'][self::$extKey]); |
|
276
|
|
|
// Extract identity information. |
|
277
|
|
|
$details = []; |
|
278
|
|
|
$details['id'] = $attributes['ID']; |
|
279
|
|
|
$details['dmdId'] = (isset($attributes['DMDID']) ? $attributes['DMDID'] : ''); |
|
280
|
|
|
$details['order'] = (isset($attributes['ORDER']) ? $attributes['ORDER'] : ''); |
|
281
|
|
|
$details['label'] = (isset($attributes['LABEL']) ? $attributes['LABEL'] : ''); |
|
282
|
|
|
$details['orderlabel'] = (isset($attributes['ORDERLABEL']) ? $attributes['ORDERLABEL'] : ''); |
|
283
|
|
|
$details['contentIds'] = (isset($attributes['CONTENTIDS']) ? $attributes['CONTENTIDS'] : ''); |
|
284
|
|
|
$details['volume'] = ''; |
|
285
|
|
|
// Set volume information only if no label is set and this is the toplevel structure element. |
|
286
|
|
|
if ( |
|
287
|
|
|
empty($details['label']) |
|
288
|
|
|
&& $details['id'] == $this->_getToplevelId() |
|
289
|
|
|
) { |
|
290
|
|
|
$metadata = $this->getMetadata($details['id']); |
|
291
|
|
|
if (!empty($metadata['volume'][0])) { |
|
292
|
|
|
$details['volume'] = $metadata['volume'][0]; |
|
293
|
|
|
} |
|
294
|
|
|
} |
|
295
|
|
|
$details['pagination'] = ''; |
|
296
|
|
|
$details['type'] = $attributes['TYPE']; |
|
297
|
|
|
$details['thumbnailId'] = ''; |
|
298
|
|
|
// Load smLinks. |
|
299
|
|
|
$this->_getSmLinks(); |
|
300
|
|
|
// Load physical structure. |
|
301
|
|
|
$this->_getPhysicalStructure(); |
|
302
|
|
|
// Get the physical page or external file this structure element is pointing at. |
|
303
|
|
|
$details['points'] = ''; |
|
304
|
|
|
// Is there a mptr node? |
|
305
|
|
|
if (count($structure->children('http://www.loc.gov/METS/')->mptr)) { |
|
306
|
|
|
// Yes. Get the file reference. |
|
307
|
|
|
$details['points'] = (string) $structure->children('http://www.loc.gov/METS/')->mptr[0]->attributes('http://www.w3.org/1999/xlink')->href; |
|
308
|
|
|
} elseif ( |
|
309
|
|
|
!empty($this->physicalStructure) |
|
310
|
|
|
&& array_key_exists($details['id'], $this->smLinks['l2p']) |
|
311
|
|
|
) { |
|
312
|
|
|
// Link logical structure to the first corresponding physical page/track. |
|
313
|
|
|
$details['points'] = max(intval(array_search($this->smLinks['l2p'][$details['id']][0], $this->physicalStructure, TRUE)), 1); |
|
314
|
|
|
if (!empty($this->physicalStructureInfo[$this->smLinks['l2p'][$details['id']][0]]['files'][$extConf['fileGrpThumbs']])) { |
|
315
|
|
|
$details['thumbnailId'] = $this->physicalStructureInfo[$this->smLinks['l2p'][$details['id']][0]]['files'][$extConf['fileGrpThumbs']]; |
|
316
|
|
|
} |
|
317
|
|
|
// Get page/track number of the first page/track related to this structure element. |
|
318
|
|
|
$details['pagination'] = $this->physicalStructureInfo[$this->smLinks['l2p'][$details['id']][0]]['orderlabel']; |
|
319
|
|
|
} elseif ($details['id'] == $this->_getToplevelId()) { |
|
320
|
|
|
// Point to self if this is the toplevel structure. |
|
321
|
|
|
$details['points'] = 1; |
|
322
|
|
|
if ( |
|
323
|
|
|
!empty($this->physicalStructure) |
|
324
|
|
|
&& !empty($this->physicalStructureInfo[$this->physicalStructure[1]]['files'][$extConf['fileGrpThumbs']]) |
|
325
|
|
|
) { |
|
326
|
|
|
$details['thumbnailId'] = $this->physicalStructureInfo[$this->physicalStructure[1]]['files'][$extConf['fileGrpThumbs']]; |
|
327
|
|
|
} |
|
328
|
|
|
} |
|
329
|
|
|
// Get the files this structure element is pointing at. |
|
330
|
|
|
$details['files'] = []; |
|
331
|
|
|
$fileUse = $this->_getFileGrps(); |
|
332
|
|
|
// Get the file representations from fileSec node. |
|
333
|
|
|
foreach ($structure->children('http://www.loc.gov/METS/')->fptr as $fptr) { |
|
334
|
|
|
// Check if file has valid @USE attribute. |
|
335
|
|
|
if (!empty($fileUse[(string) $fptr->attributes()->FILEID])) { |
|
336
|
|
|
$details['files'][$fileUse[(string) $fptr->attributes()->FILEID]] = (string) $fptr->attributes()->FILEID; |
|
337
|
|
|
} |
|
338
|
|
|
} |
|
339
|
|
|
// Keep for later usage. |
|
340
|
|
|
$this->logicalUnits[$details['id']] = $details; |
|
341
|
|
|
// Walk the structure recursively? And are there any children of the current element? |
|
342
|
|
|
if ( |
|
343
|
|
|
$recursive |
|
344
|
|
|
&& count($structure->children('http://www.loc.gov/METS/')->div) |
|
345
|
|
|
) { |
|
346
|
|
|
$details['children'] = []; |
|
347
|
|
|
foreach ($structure->children('http://www.loc.gov/METS/')->div as $child) { |
|
348
|
|
|
// Repeat for all children. |
|
349
|
|
|
$details['children'][] = $this->getLogicalStructureInfo($child, TRUE); |
|
350
|
|
|
} |
|
351
|
|
|
} |
|
352
|
|
|
return $details; |
|
353
|
|
|
} |
|
354
|
|
|
|
|
355
|
|
|
/** |
|
356
|
|
|
* {@inheritDoc} |
|
357
|
|
|
* @see \Kitodo\Dlf\Common\Document::getMetadata() |
|
358
|
|
|
*/ |
|
359
|
|
|
public function getMetadata($id, $cPid = 0) |
|
360
|
|
|
{ |
|
361
|
|
|
// Make sure $cPid is a non-negative integer. |
|
362
|
|
|
$cPid = max(intval($cPid), 0); |
|
363
|
|
|
// If $cPid is not given, try to get it elsewhere. |
|
364
|
|
|
if ( |
|
365
|
|
|
!$cPid |
|
366
|
|
|
&& ($this->cPid || $this->pid) |
|
367
|
|
|
) { |
|
368
|
|
|
// Retain current PID. |
|
369
|
|
|
$cPid = ($this->cPid ? $this->cPid : $this->pid); |
|
370
|
|
|
} elseif (!$cPid) { |
|
371
|
|
|
Helper::devLog('Invalid PID ' . $cPid . ' for metadata definitions', DEVLOG_SEVERITY_WARNING); |
|
372
|
|
|
return []; |
|
373
|
|
|
} |
|
374
|
|
|
// Get metadata from parsed metadata array if available. |
|
375
|
|
|
if ( |
|
376
|
|
|
!empty($this->metadataArray[$id]) |
|
377
|
|
|
&& $this->metadataArray[0] == $cPid |
|
378
|
|
|
) { |
|
379
|
|
|
return $this->metadataArray[$id]; |
|
380
|
|
|
} |
|
381
|
|
|
// Initialize metadata array with empty values. |
|
382
|
|
|
$metadata = [ |
|
383
|
|
|
'title' => [], |
|
384
|
|
|
'title_sorting' => [], |
|
385
|
|
|
'author' => [], |
|
386
|
|
|
'place' => [], |
|
387
|
|
|
'year' => [], |
|
388
|
|
|
'prod_id' => [], |
|
389
|
|
|
'record_id' => [], |
|
390
|
|
|
'opac_id' => [], |
|
391
|
|
|
'union_id' => [], |
|
392
|
|
|
'urn' => [], |
|
393
|
|
|
'purl' => [], |
|
394
|
|
|
'type' => [], |
|
395
|
|
|
'volume' => [], |
|
396
|
|
|
'volume_sorting' => [], |
|
397
|
|
|
'license' => [], |
|
398
|
|
|
'terms' => [], |
|
399
|
|
|
'restrictions' => [], |
|
400
|
|
|
'out_of_print' => [], |
|
401
|
|
|
'rights_info' => [], |
|
402
|
|
|
'collection' => [], |
|
403
|
|
|
'owner' => [], |
|
404
|
|
|
'mets_label' => [], |
|
405
|
|
|
'mets_orderlabel' => [], |
|
406
|
|
|
'document_format' => [], |
|
407
|
|
|
]; |
|
408
|
|
|
$metadata['document_format'][] = 'METS'; |
|
409
|
|
|
// Get the logical structure node's @DMDID. |
|
410
|
|
|
if (!empty($this->logicalUnits[$id])) { |
|
411
|
|
|
$dmdIds = $this->logicalUnits[$id]['dmdId']; |
|
412
|
|
|
} else { |
|
413
|
|
|
$dmdIds = $this->mets->xpath('./mets:structMap[@TYPE="LOGICAL"]//mets:div[@ID="' . $id . '"]/@DMDID'); |
|
414
|
|
|
$dmdIds = (string) $dmdIds[0]; |
|
415
|
|
|
} |
|
416
|
|
|
if (!empty($dmdIds)) { |
|
417
|
|
|
// Handle multiple DMDIDs separately. |
|
418
|
|
|
$dmdIds = explode(' ', $dmdIds); |
|
419
|
|
|
$hasSupportedMetadata = FALSE; |
|
420
|
|
|
} else { |
|
421
|
|
|
// There is no dmdSec for this structure node. |
|
422
|
|
|
return []; |
|
423
|
|
|
} |
|
424
|
|
|
// Load available metadata formats and dmdSecs. |
|
425
|
|
|
$this->loadFormats(); |
|
426
|
|
|
$this->_getDmdSec(); |
|
427
|
|
|
foreach ($dmdIds as $dmdId) { |
|
428
|
|
|
// Is this metadata format supported? |
|
429
|
|
|
if (!empty($this->formats[$this->dmdSec[$dmdId]['type']])) { |
|
430
|
|
|
if (!empty($this->formats[$this->dmdSec[$dmdId]['type']]['class'])) { |
|
431
|
|
|
$class = $this->formats[$this->dmdSec[$dmdId]['type']]['class']; |
|
432
|
|
|
// Get the metadata from class. |
|
433
|
|
|
if ( |
|
434
|
|
|
class_exists($class) |
|
435
|
|
|
&& ($obj = \TYPO3\CMS\Core\Utility\GeneralUtility::makeInstance($class)) instanceof MetadataInterface |
|
436
|
|
|
) { |
|
437
|
|
|
$obj->extractMetadata($this->dmdSec[$dmdId]['xml'], $metadata); |
|
438
|
|
|
} else { |
|
439
|
|
|
Helper::devLog('Invalid class/method "' . $class . '->extractMetadata()" for metadata format "' . $this->dmdSec[$dmdId]['type'] . '"', DEVLOG_SEVERITY_WARNING); |
|
440
|
|
|
} |
|
441
|
|
|
} |
|
442
|
|
|
} else { |
|
443
|
|
|
Helper::devLog('Unsupported metadata format "' . $this->dmdSec[$dmdId]['type'] . '" in dmdSec with @ID "' . $dmdId . '"', DEVLOG_SEVERITY_NOTICE); |
|
444
|
|
|
// Continue searching for supported metadata with next @DMDID. |
|
445
|
|
|
continue; |
|
446
|
|
|
} |
|
447
|
|
|
// Get the structure's type. |
|
448
|
|
|
if (!empty($this->logicalUnits[$id])) { |
|
449
|
|
|
$metadata['type'] = [$this->logicalUnits[$id]['type']]; |
|
450
|
|
|
} else { |
|
451
|
|
|
$struct = $this->mets->xpath('./mets:structMap[@TYPE="LOGICAL"]//mets:div[@ID="' . $id . '"]/@TYPE'); |
|
452
|
|
|
if (!empty($struct)) { |
|
453
|
|
|
$metadata['type'] = [(string) $struct[0]]; |
|
454
|
|
|
} |
|
455
|
|
|
} |
|
456
|
|
|
// Get the additional metadata from database. |
|
457
|
|
|
$result = $GLOBALS['TYPO3_DB']->exec_SELECTquery( |
|
458
|
|
|
'tx_dlf_metadata.index_name AS index_name,tx_dlf_metadataformat.xpath AS xpath,tx_dlf_metadataformat.xpath_sorting AS xpath_sorting,tx_dlf_metadata.is_sortable AS is_sortable,tx_dlf_metadata.default_value AS default_value,tx_dlf_metadata.format AS format', |
|
459
|
|
|
'tx_dlf_metadata,tx_dlf_metadataformat,tx_dlf_formats', |
|
460
|
|
|
'tx_dlf_metadata.pid=' . $cPid |
|
461
|
|
|
. ' AND tx_dlf_metadataformat.pid=' . $cPid |
|
462
|
|
|
. ' AND ((tx_dlf_metadata.uid=tx_dlf_metadataformat.parent_id AND tx_dlf_metadataformat.encoded=tx_dlf_formats.uid AND tx_dlf_formats.type=' . $GLOBALS['TYPO3_DB']->fullQuoteStr($this->dmdSec[$dmdId]['type'], 'tx_dlf_formats') . ') OR tx_dlf_metadata.format=0)' |
|
463
|
|
|
. Helper::whereClause('tx_dlf_metadata', TRUE) |
|
464
|
|
|
. Helper::whereClause('tx_dlf_metadataformat') |
|
465
|
|
|
. Helper::whereClause('tx_dlf_formats') |
|
466
|
|
|
); |
|
467
|
|
|
// We need a \DOMDocument here, because SimpleXML doesn't support XPath functions properly. |
|
468
|
|
|
$domNode = dom_import_simplexml($this->dmdSec[$dmdId]['xml']); |
|
469
|
|
|
$domXPath = new \DOMXPath($domNode->ownerDocument); |
|
470
|
|
|
$this->registerNamespaces($domXPath); |
|
471
|
|
|
// OK, now make the XPath queries. |
|
472
|
|
|
while ($resArray = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($result)) { |
|
473
|
|
|
// Set metadata field's value(s). |
|
474
|
|
|
if ( |
|
475
|
|
|
$resArray['format'] > 0 |
|
476
|
|
|
&& !empty($resArray['xpath']) |
|
477
|
|
|
&& ($values = $domXPath->evaluate($resArray['xpath'], $domNode)) |
|
478
|
|
|
) { |
|
479
|
|
|
if ( |
|
480
|
|
|
$values instanceof \DOMNodeList |
|
481
|
|
|
&& $values->length > 0 |
|
482
|
|
|
) { |
|
483
|
|
|
$metadata[$resArray['index_name']] = []; |
|
484
|
|
|
foreach ($values as $value) { |
|
485
|
|
|
$metadata[$resArray['index_name']][] = trim((string) $value->nodeValue); |
|
486
|
|
|
} |
|
487
|
|
|
} elseif (!($values instanceof \DOMNodeList)) { |
|
488
|
|
|
$metadata[$resArray['index_name']] = [trim((string) $values)]; |
|
489
|
|
|
} |
|
490
|
|
|
} |
|
491
|
|
|
// Set default value if applicable. |
|
492
|
|
|
if ( |
|
493
|
|
|
empty($metadata[$resArray['index_name']][0]) |
|
494
|
|
|
&& strlen($resArray['default_value']) > 0 |
|
495
|
|
|
) { |
|
496
|
|
|
$metadata[$resArray['index_name']] = [$resArray['default_value']]; |
|
497
|
|
|
} |
|
498
|
|
|
// Set sorting value if applicable. |
|
499
|
|
|
if ( |
|
500
|
|
|
!empty($metadata[$resArray['index_name']]) |
|
501
|
|
|
&& $resArray['is_sortable'] |
|
502
|
|
|
) { |
|
503
|
|
|
if ( |
|
504
|
|
|
$resArray['format'] > 0 |
|
505
|
|
|
&& !empty($resArray['xpath_sorting']) |
|
506
|
|
|
&& ($values = $domXPath->evaluate($resArray['xpath_sorting'], $domNode)) |
|
507
|
|
|
) { |
|
508
|
|
|
if ( |
|
509
|
|
|
$values instanceof \DOMNodeList |
|
510
|
|
|
&& $values->length > 0 |
|
511
|
|
|
) { |
|
512
|
|
|
$metadata[$resArray['index_name'] . '_sorting'][0] = trim((string) $values->item(0)->nodeValue); |
|
513
|
|
|
} elseif (!($values instanceof \DOMNodeList)) { |
|
514
|
|
|
$metadata[$resArray['index_name'] . '_sorting'][0] = trim((string) $values); |
|
515
|
|
|
} |
|
516
|
|
|
} |
|
517
|
|
|
if (empty($metadata[$resArray['index_name'] . '_sorting'][0])) { |
|
518
|
|
|
$metadata[$resArray['index_name'] . '_sorting'][0] = $metadata[$resArray['index_name']][0]; |
|
519
|
|
|
} |
|
520
|
|
|
} |
|
521
|
|
|
} |
|
522
|
|
|
// Set title to empty string if not present. |
|
523
|
|
|
if (empty($metadata['title'][0])) { |
|
524
|
|
|
$metadata['title'][0] = ''; |
|
525
|
|
|
$metadata['title_sorting'][0] = ''; |
|
526
|
|
|
} |
|
527
|
|
|
// Add collections from database to toplevel element if document is already saved. |
|
528
|
|
|
if ( |
|
529
|
|
|
\TYPO3\CMS\Core\Utility\MathUtility::canBeInterpretedAsInteger($this->uid) |
|
530
|
|
|
&& $id == $this->_getToplevelId() |
|
531
|
|
|
) { |
|
532
|
|
|
$result = $GLOBALS['TYPO3_DB']->exec_SELECT_mm_query( |
|
533
|
|
|
'tx_dlf_collections.index_name AS index_name', |
|
534
|
|
|
'tx_dlf_documents', |
|
535
|
|
|
'tx_dlf_relations', |
|
536
|
|
|
'tx_dlf_collections', |
|
537
|
|
|
'AND tx_dlf_collections.pid=' . intval($cPid) |
|
538
|
|
|
. ' AND tx_dlf_documents.uid=' . intval($this->uid) |
|
539
|
|
|
. ' AND tx_dlf_relations.ident=' . $GLOBALS['TYPO3_DB']->fullQuoteStr('docs_colls', 'tx_dlf_relations') |
|
540
|
|
|
. ' AND tx_dlf_collections.sys_language_uid IN (-1,0)' |
|
541
|
|
|
. Helper::whereClause('tx_dlf_documents') |
|
542
|
|
|
. Helper::whereClause('tx_dlf_collections'), |
|
543
|
|
|
'tx_dlf_collections.index_name' |
|
544
|
|
|
); |
|
545
|
|
|
while ($resArray = $GLOBALS['TYPO3_DB']->sql_fetch_assoc($result)) { |
|
546
|
|
|
if (!in_array($resArray['index_name'], $metadata['collection'])) { |
|
547
|
|
|
$metadata['collection'][] = $resArray['index_name']; |
|
548
|
|
|
} |
|
549
|
|
|
} |
|
550
|
|
|
} |
|
551
|
|
|
// Extract metadata only from first supported dmdSec. |
|
552
|
|
|
$hasSupportedMetadata = TRUE; |
|
553
|
|
|
break; |
|
554
|
|
|
} |
|
555
|
|
|
if ($hasSupportedMetadata) { |
|
556
|
|
|
return $metadata; |
|
557
|
|
|
} else { |
|
558
|
|
|
Helper::devLog('No supported metadata found for logical structure with @ID "' . $id . '"', DEVLOG_SEVERITY_WARNING); |
|
559
|
|
|
return []; |
|
560
|
|
|
} |
|
561
|
|
|
} |
|
562
|
|
|
|
|
563
|
|
|
/** |
|
564
|
|
|
* {@inheritDoc} |
|
565
|
|
|
* @see \Kitodo\Dlf\Common\Document::getRawText() |
|
566
|
|
|
*/ |
|
567
|
|
|
public function getRawText($id) |
|
568
|
|
|
{ |
|
569
|
|
|
$rawText = ''; |
|
570
|
|
|
// Get text from raw text array if available. |
|
571
|
|
|
if (!empty($this->rawTextArray[$id])) { |
|
572
|
|
|
return $this->rawTextArray[$id]; |
|
573
|
|
|
} |
|
574
|
|
|
// Load fileGrps and check for fulltext files. |
|
575
|
|
|
$this->_getFileGrps(); |
|
576
|
|
|
if ($this->hasFulltext) { |
|
577
|
|
|
$rawText = $this->getRawTextFromXml($id); |
|
578
|
|
|
} |
|
579
|
|
|
return $rawText; |
|
580
|
|
|
} |
|
581
|
|
|
|
|
582
|
|
|
/** |
|
583
|
|
|
* {@inheritDoc} |
|
584
|
|
|
* @see Document::getStructureDepth() |
|
585
|
|
|
*/ |
|
586
|
|
|
public function getStructureDepth($logId) |
|
587
|
|
|
{ |
|
588
|
|
|
$ancestors = $this->mets->xpath('./mets:structMap[@TYPE="LOGICAL"]//mets:div[@ID="' . $logId . '"]/ancestor::*'); |
|
589
|
|
|
if (!empty($ancestors)) { |
|
590
|
|
|
return count($ancestors); |
|
591
|
|
|
} else { |
|
592
|
|
|
return 0; |
|
593
|
|
|
} |
|
594
|
|
|
} |
|
595
|
|
|
|
|
596
|
|
|
/** |
|
597
|
|
|
* {@inheritDoc} |
|
598
|
|
|
* @see \Kitodo\Dlf\Common\Document::init() |
|
599
|
|
|
*/ |
|
600
|
|
|
protected function init() |
|
601
|
|
|
{ |
|
602
|
|
|
// Get METS node from XML file. |
|
603
|
|
|
$this->registerNamespaces($this->xml); |
|
604
|
|
|
$mets = $this->xml->xpath('//mets:mets'); |
|
605
|
|
|
if (!empty($mets)) { |
|
606
|
|
|
$this->mets = $mets[0]; |
|
607
|
|
|
// Register namespaces. |
|
608
|
|
|
$this->registerNamespaces($this->mets); |
|
609
|
|
|
} else { |
|
610
|
|
|
Helper::devLog('No METS part found in document with UID ' . $this->uid, DEVLOG_SEVERITY_ERROR); |
|
611
|
|
|
} |
|
612
|
|
|
} |
|
613
|
|
|
|
|
614
|
|
|
/** |
|
615
|
|
|
* {@inheritDoc} |
|
616
|
|
|
* @see \Kitodo\Dlf\Common\Document::loadLocation() |
|
617
|
|
|
*/ |
|
618
|
|
|
protected function loadLocation($location) |
|
619
|
|
|
{ |
|
620
|
|
|
$fileResource = GeneralUtility::getUrl($location); |
|
621
|
|
|
if ($fileResource !== FALSE) { |
|
622
|
|
|
// Turn off libxml's error logging. |
|
623
|
|
|
$libxmlErrors = libxml_use_internal_errors(TRUE); |
|
624
|
|
|
// Disables the functionality to allow external entities to be loaded when parsing the XML, must be kept |
|
625
|
|
|
$previousValueOfEntityLoader = libxml_disable_entity_loader(TRUE); |
|
626
|
|
|
// Load XML from file. |
|
627
|
|
|
$xml = simplexml_load_string($fileResource); |
|
628
|
|
|
// reset entity loader setting |
|
629
|
|
|
libxml_disable_entity_loader($previousValueOfEntityLoader); |
|
630
|
|
|
// Reset libxml's error logging. |
|
631
|
|
|
libxml_use_internal_errors($libxmlErrors); |
|
632
|
|
|
// Set some basic properties. |
|
633
|
|
|
if ($xml !== FALSE) { |
|
634
|
|
|
$this->xml = $xml; |
|
635
|
|
|
return TRUE; |
|
636
|
|
|
} |
|
637
|
|
|
} |
|
638
|
|
|
Helper::devLog('Could not load XML file from "' . $location . '"', DEVLOG_SEVERITY_ERROR); |
|
639
|
|
|
return FALSE; |
|
640
|
|
|
} |
|
641
|
|
|
|
|
642
|
|
|
/** |
|
643
|
|
|
* {@inheritDoc} |
|
644
|
|
|
* @see \Kitodo\Dlf\Common\Document::ensureHasFulltextIsSet() |
|
645
|
|
|
*/ |
|
646
|
|
|
protected function ensureHasFulltextIsSet() |
|
647
|
|
|
{ |
|
648
|
|
|
// Are the fileGrps already loaded? |
|
649
|
|
|
if (!$this->fileGrpsLoaded) { |
|
650
|
|
|
$this->_getFileGrps(); |
|
651
|
|
|
} |
|
652
|
|
|
} |
|
653
|
|
|
|
|
654
|
|
|
/** |
|
655
|
|
|
* {@inheritDoc} |
|
656
|
|
|
* @see Document::getParentDocumentUid() |
|
657
|
|
|
*/ |
|
658
|
|
|
protected function getParentDocumentUidForSaving($pid, $core) |
|
659
|
|
|
{ |
|
660
|
|
|
$partof = 0; |
|
661
|
|
|
// Get the closest ancestor of the current document which has a MPTR child. |
|
662
|
|
|
$parentMptr = $this->mets->xpath('./mets:structMap[@TYPE="LOGICAL"]//mets:div[@ID="' . $this->_getToplevelId() . '"]/ancestor::mets:div[./mets:mptr][1]/mets:mptr'); |
|
663
|
|
|
if (!empty($parentMptr)) { |
|
664
|
|
|
$parentLocation = (string) $parentMptr[0]->attributes('http://www.w3.org/1999/xlink')->href; |
|
665
|
|
|
if ($parentLocation != $this->location) { |
|
666
|
|
|
$parentDoc = self::getInstance($parentLocation, $pid); |
|
667
|
|
|
if ($parentDoc->ready) { |
|
668
|
|
|
if ($parentDoc->pid != $pid) { |
|
669
|
|
|
$parentDoc->save($pid, $core); |
|
670
|
|
|
} |
|
671
|
|
|
$partof = $parentDoc->uid; |
|
672
|
|
|
} |
|
673
|
|
|
} |
|
674
|
|
|
} |
|
675
|
|
|
return $partof; |
|
676
|
|
|
} |
|
677
|
|
|
|
|
678
|
|
|
/** |
|
679
|
|
|
* {@inheritDoc} |
|
680
|
|
|
* @see Document::setPreloadedDocument() |
|
681
|
|
|
*/ |
|
682
|
|
|
protected function setPreloadedDocument($preloadedDocument) |
|
683
|
|
|
{ |
|
684
|
|
|
|
|
685
|
|
|
if ($preloadedDocument instanceof \SimpleXMLElement) { |
|
686
|
|
|
$this->xml = $preloadedDocument; |
|
687
|
|
|
return TRUE; |
|
688
|
|
|
} |
|
689
|
|
|
return FALSE; |
|
690
|
|
|
} |
|
691
|
|
|
|
|
692
|
|
|
/** |
|
693
|
|
|
* {@inheritDoc} |
|
694
|
|
|
* @see Document::getDocument() |
|
695
|
|
|
*/ |
|
696
|
|
|
protected function getDocument() |
|
697
|
|
|
{ |
|
698
|
|
|
return $this->mets; |
|
699
|
|
|
} |
|
700
|
|
|
|
|
701
|
|
|
/** |
|
702
|
|
|
* This returns $this->cPid via __get() |
|
703
|
|
|
* |
|
704
|
|
|
* @access protected |
|
705
|
|
|
* |
|
706
|
|
|
* @return integer The PID of the metadata definitions |
|
707
|
|
|
*/ |
|
708
|
|
|
protected function _getCPid() |
|
709
|
|
|
{ |
|
710
|
|
|
return $this->cPid; |
|
711
|
|
|
} |
|
712
|
|
|
|
|
713
|
|
|
/** |
|
714
|
|
|
* This builds an array of the document's dmdSecs |
|
715
|
|
|
* |
|
716
|
|
|
* @access protected |
|
717
|
|
|
* |
|
718
|
|
|
* @return array Array of dmdSecs with their IDs as array key |
|
719
|
|
|
*/ |
|
720
|
|
|
protected function _getDmdSec() |
|
721
|
|
|
{ |
|
722
|
|
|
if (!$this->dmdSecLoaded) { |
|
723
|
|
|
// Get available data formats. |
|
724
|
|
|
$this->loadFormats(); |
|
725
|
|
|
// Get dmdSec nodes from METS. |
|
726
|
|
|
$dmdIds = $this->mets->xpath('./mets:dmdSec/@ID'); |
|
727
|
|
|
if (!empty($dmdIds)) { |
|
728
|
|
|
foreach ($dmdIds as $dmdId) { |
|
729
|
|
|
if ($type = $this->mets->xpath('./mets:dmdSec[@ID="' . (string) $dmdId . '"]/mets:mdWrap[not(@MDTYPE="OTHER")]/@MDTYPE')) { |
|
730
|
|
|
if (!empty($this->formats[(string) $type[0]])) { |
|
731
|
|
|
$type = (string) $type[0]; |
|
732
|
|
|
$xml = $this->mets->xpath('./mets:dmdSec[@ID="' . (string) $dmdId . '"]/mets:mdWrap[@MDTYPE="' . $type . '"]/mets:xmlData/' . strtolower($type) . ':' . $this->formats[$type]['rootElement']); |
|
733
|
|
|
} |
|
734
|
|
|
} elseif ($type = $this->mets->xpath('./mets:dmdSec[@ID="' . (string) $dmdId . '"]/mets:mdWrap[@MDTYPE="OTHER"]/@OTHERMDTYPE')) { |
|
735
|
|
|
if (!empty($this->formats[(string) $type[0]])) { |
|
736
|
|
|
$type = (string) $type[0]; |
|
737
|
|
|
$xml = $this->mets->xpath('./mets:dmdSec[@ID="' . (string) $dmdId . '"]/mets:mdWrap[@MDTYPE="OTHER"][@OTHERMDTYPE="' . $type . '"]/mets:xmlData/' . strtolower($type) . ':' . $this->formats[$type]['rootElement']); |
|
738
|
|
|
} |
|
739
|
|
|
} |
|
740
|
|
|
if (!empty($xml)) { |
|
741
|
|
|
$this->dmdSec[(string) $dmdId]['type'] = $type; |
|
742
|
|
|
$this->dmdSec[(string) $dmdId]['xml'] = $xml[0]; |
|
743
|
|
|
$this->registerNamespaces($this->dmdSec[(string) $dmdId]['xml']); |
|
744
|
|
|
} |
|
745
|
|
|
} |
|
746
|
|
|
} |
|
747
|
|
|
$this->dmdSecLoaded = TRUE; |
|
748
|
|
|
} |
|
749
|
|
|
return $this->dmdSec; |
|
750
|
|
|
} |
|
751
|
|
|
|
|
752
|
|
|
/** |
|
753
|
|
|
* This builds the file ID -> USE concordance |
|
754
|
|
|
* |
|
755
|
|
|
* @access protected |
|
756
|
|
|
* |
|
757
|
|
|
* @return array Array of file use groups with file IDs |
|
758
|
|
|
*/ |
|
759
|
|
|
protected function _getFileGrps() |
|
760
|
|
|
{ |
|
761
|
|
|
if (!$this->fileGrpsLoaded) { |
|
762
|
|
|
// Get configured USE attributes. |
|
763
|
|
|
$extConf = unserialize($GLOBALS['TYPO3_CONF_VARS']['EXT']['extConf'][self::$extKey]); |
|
764
|
|
|
$useGrps = \TYPO3\CMS\Core\Utility\GeneralUtility::trimExplode(',', $extConf['fileGrps']); |
|
765
|
|
|
if (!empty($extConf['fileGrpThumbs'])) { |
|
766
|
|
|
$useGrps[] = $extConf['fileGrpThumbs']; |
|
767
|
|
|
} |
|
768
|
|
|
if (!empty($extConf['fileGrpDownload'])) { |
|
769
|
|
|
$useGrps[] = $extConf['fileGrpDownload']; |
|
770
|
|
|
} |
|
771
|
|
|
if (!empty($extConf['fileGrpFulltext'])) { |
|
772
|
|
|
$useGrps[] = $extConf['fileGrpFulltext']; |
|
773
|
|
|
} |
|
774
|
|
|
if (!empty($extConf['fileGrpAudio'])) { |
|
775
|
|
|
$useGrps[] = $extConf['fileGrpAudio']; |
|
776
|
|
|
} |
|
777
|
|
|
// Get all file groups. |
|
778
|
|
|
$fileGrps = $this->mets->xpath('./mets:fileSec/mets:fileGrp'); |
|
779
|
|
|
if (!empty($fileGrps)) { |
|
780
|
|
|
// Build concordance for configured USE attributes. |
|
781
|
|
|
foreach ($fileGrps as $fileGrp) { |
|
782
|
|
|
if (in_array((string) $fileGrp['USE'], $useGrps)) { |
|
783
|
|
|
foreach ($fileGrp->children('http://www.loc.gov/METS/')->file as $file) { |
|
784
|
|
|
$this->fileGrps[(string) $file->attributes()->ID] = (string) $fileGrp['USE']; |
|
785
|
|
|
} |
|
786
|
|
|
} |
|
787
|
|
|
} |
|
788
|
|
|
} |
|
789
|
|
|
// Are there any fulltext files available? |
|
790
|
|
|
if ( |
|
791
|
|
|
!empty($extConf['fileGrpFulltext']) |
|
792
|
|
|
&& in_array($extConf['fileGrpFulltext'], $this->fileGrps) |
|
793
|
|
|
) { |
|
794
|
|
|
$this->hasFulltext = TRUE; |
|
795
|
|
|
} |
|
796
|
|
|
$this->fileGrpsLoaded = TRUE; |
|
797
|
|
|
} |
|
798
|
|
|
return $this->fileGrps; |
|
799
|
|
|
} |
|
800
|
|
|
|
|
801
|
|
|
/** |
|
802
|
|
|
* {@inheritDoc} |
|
803
|
|
|
* @see \Kitodo\Dlf\Common\Document::prepareMetadataArray() |
|
804
|
|
|
*/ |
|
805
|
|
|
protected function prepareMetadataArray($cPid) |
|
806
|
|
|
{ |
|
807
|
|
|
$ids = $this->mets->xpath('./mets:structMap[@TYPE="LOGICAL"]//mets:div[@DMDID]/@ID'); |
|
808
|
|
|
// Get all logical structure nodes with metadata. |
|
809
|
|
|
if (!empty($ids)) { |
|
810
|
|
|
foreach ($ids as $id) { |
|
811
|
|
|
$this->metadataArray[(string) $id] = $this->getMetadata((string) $id, $cPid); |
|
812
|
|
|
} |
|
813
|
|
|
} |
|
814
|
|
|
// Set current PID for metadata definitions. |
|
815
|
|
|
} |
|
816
|
|
|
|
|
817
|
|
|
/** |
|
818
|
|
|
* This returns $this->mets via __get() |
|
819
|
|
|
* |
|
820
|
|
|
* @access protected |
|
821
|
|
|
* |
|
822
|
|
|
* @return \SimpleXMLElement The XML's METS part as \SimpleXMLElement object |
|
823
|
|
|
*/ |
|
824
|
|
|
protected function _getMets() |
|
825
|
|
|
{ |
|
826
|
|
|
return $this->mets; |
|
827
|
|
|
} |
|
828
|
|
|
|
|
829
|
|
|
/** |
|
830
|
|
|
* {@inheritDoc} |
|
831
|
|
|
* @see \Kitodo\Dlf\Common\Document::_getPhysicalStructure() |
|
832
|
|
|
*/ |
|
833
|
|
|
protected function _getPhysicalStructure() |
|
834
|
|
|
{ |
|
835
|
|
|
// Is there no physical structure array yet? |
|
836
|
|
|
if (!$this->physicalStructureLoaded) { |
|
837
|
|
|
// Does the document have a structMap node of type "PHYSICAL"? |
|
838
|
|
|
$elementNodes = $this->mets->xpath('./mets:structMap[@TYPE="PHYSICAL"]/mets:div[@TYPE="physSequence"]/mets:div'); |
|
839
|
|
|
if (!empty($elementNodes)) { |
|
840
|
|
|
// Get file groups. |
|
841
|
|
|
$fileUse = $this->_getFileGrps(); |
|
842
|
|
|
// Get the physical sequence's metadata. |
|
843
|
|
|
$physNode = $this->mets->xpath('./mets:structMap[@TYPE="PHYSICAL"]/mets:div[@TYPE="physSequence"]'); |
|
844
|
|
|
$physSeq[0] = (string) $physNode[0]['ID']; |
|
845
|
|
|
$this->physicalStructureInfo[$physSeq[0]]['id'] = (string) $physNode[0]['ID']; |
|
846
|
|
|
$this->physicalStructureInfo[$physSeq[0]]['dmdId'] = (isset($physNode[0]['DMDID']) ? (string) $physNode[0]['DMDID'] : ''); |
|
847
|
|
|
$this->physicalStructureInfo[$physSeq[0]]['order'] = (isset($physNode[0]['ORDER']) ? (string) $physNode[0]['ORDER'] : ''); |
|
848
|
|
|
$this->physicalStructureInfo[$physSeq[0]]['label'] = (isset($physNode[0]['LABEL']) ? (string) $physNode[0]['LABEL'] : ''); |
|
849
|
|
|
$this->physicalStructureInfo[$physSeq[0]]['orderlabel'] = (isset($physNode[0]['ORDERLABEL']) ? (string) $physNode[0]['ORDERLABEL'] : ''); |
|
850
|
|
|
$this->physicalStructureInfo[$physSeq[0]]['type'] = (string) $physNode[0]['TYPE']; |
|
851
|
|
|
$this->physicalStructureInfo[$physSeq[0]]['contentIds'] = (isset($physNode[0]['CONTENTIDS']) ? (string) $physNode[0]['CONTENTIDS'] : ''); |
|
852
|
|
|
// Get the file representations from fileSec node. |
|
853
|
|
|
foreach ($physNode[0]->children('http://www.loc.gov/METS/')->fptr as $fptr) { |
|
854
|
|
|
// Check if file has valid @USE attribute. |
|
855
|
|
|
if (!empty($fileUse[(string) $fptr->attributes()->FILEID])) { |
|
856
|
|
|
$this->physicalStructureInfo[$physSeq[0]]['files'][$fileUse[(string) $fptr->attributes()->FILEID]] = (string) $fptr->attributes()->FILEID; |
|
857
|
|
|
} |
|
858
|
|
|
} |
|
859
|
|
|
// Build the physical elements' array from the physical structMap node. |
|
860
|
|
|
foreach ($elementNodes as $elementNode) { |
|
861
|
|
|
$elements[(int) $elementNode['ORDER']] = (string) $elementNode['ID']; |
|
862
|
|
|
$this->physicalStructureInfo[$elements[(int) $elementNode['ORDER']]]['id'] = (string) $elementNode['ID']; |
|
863
|
|
|
$this->physicalStructureInfo[$elements[(int) $elementNode['ORDER']]]['dmdId'] = (isset($elementNode['DMDID']) ? (string) $elementNode['DMDID'] : ''); |
|
864
|
|
|
$this->physicalStructureInfo[$elements[(int) $elementNode['ORDER']]]['order'] = (isset($elementNode['ORDER']) ? (string) $elementNode['ORDER'] : ''); |
|
865
|
|
|
$this->physicalStructureInfo[$elements[(int) $elementNode['ORDER']]]['label'] = (isset($elementNode['LABEL']) ? (string) $elementNode['LABEL'] : ''); |
|
866
|
|
|
$this->physicalStructureInfo[$elements[(int) $elementNode['ORDER']]]['orderlabel'] = (isset($elementNode['ORDERLABEL']) ? (string) $elementNode['ORDERLABEL'] : ''); |
|
867
|
|
|
$this->physicalStructureInfo[$elements[(int) $elementNode['ORDER']]]['type'] = (string) $elementNode['TYPE']; |
|
868
|
|
|
$this->physicalStructureInfo[$elements[(int) $elementNode['ORDER']]]['contentIds'] = (isset($elementNode['CONTENTIDS']) ? (string) $elementNode['CONTENTIDS'] : ''); |
|
869
|
|
|
// Get the file representations from fileSec node. |
|
870
|
|
|
foreach ($elementNode->children('http://www.loc.gov/METS/')->fptr as $fptr) { |
|
871
|
|
|
// Check if file has valid @USE attribute. |
|
872
|
|
|
if (!empty($fileUse[(string) $fptr->attributes()->FILEID])) { |
|
873
|
|
|
$this->physicalStructureInfo[$elements[(int) $elementNode['ORDER']]]['files'][$fileUse[(string) $fptr->attributes()->FILEID]] = (string) $fptr->attributes()->FILEID; |
|
874
|
|
|
} |
|
875
|
|
|
} |
|
876
|
|
|
} |
|
877
|
|
|
// Sort array by keys (= @ORDER). |
|
878
|
|
|
if (ksort($elements)) { |
|
879
|
|
|
// Set total number of pages/tracks. |
|
880
|
|
|
$this->numPages = count($elements); |
|
881
|
|
|
// Merge and re-index the array to get nice numeric indexes. |
|
882
|
|
|
$this->physicalStructure = array_merge($physSeq, $elements); |
|
883
|
|
|
} |
|
884
|
|
|
} |
|
885
|
|
|
$this->physicalStructureLoaded = TRUE; |
|
886
|
|
|
} |
|
887
|
|
|
return $this->physicalStructure; |
|
888
|
|
|
} |
|
889
|
|
|
|
|
890
|
|
|
/** |
|
891
|
|
|
* {@inheritDoc} |
|
892
|
|
|
* @see \Kitodo\Dlf\Common\Document::_getSmLinks() |
|
893
|
|
|
*/ |
|
894
|
|
|
protected function _getSmLinks() |
|
895
|
|
|
{ |
|
896
|
|
|
if (!$this->smLinksLoaded) { |
|
897
|
|
|
$smLinks = $this->mets->xpath('./mets:structLink/mets:smLink'); |
|
898
|
|
|
if (!empty($smLinks)) { |
|
899
|
|
|
foreach ($smLinks as $smLink) { |
|
900
|
|
|
$this->smLinks['l2p'][(string) $smLink->attributes('http://www.w3.org/1999/xlink')->from][] = (string) $smLink->attributes('http://www.w3.org/1999/xlink')->to; |
|
901
|
|
|
$this->smLinks['p2l'][(string) $smLink->attributes('http://www.w3.org/1999/xlink')->to][] = (string) $smLink->attributes('http://www.w3.org/1999/xlink')->from; |
|
902
|
|
|
} |
|
903
|
|
|
} |
|
904
|
|
|
$this->smLinksLoaded = TRUE; |
|
905
|
|
|
} |
|
906
|
|
|
return $this->smLinks; |
|
907
|
|
|
} |
|
908
|
|
|
|
|
909
|
|
|
/** |
|
910
|
|
|
* {@inheritDoc} |
|
911
|
|
|
* @see \Kitodo\Dlf\Common\Document::_getThumbnail() |
|
912
|
|
|
*/ |
|
913
|
|
|
protected function _getThumbnail($forceReload = FALSE) |
|
914
|
|
|
{ |
|
915
|
|
|
if ( |
|
916
|
|
|
!$this->thumbnailLoaded |
|
917
|
|
|
|| $forceReload |
|
918
|
|
|
) { |
|
919
|
|
|
// Retain current PID. |
|
920
|
|
|
$cPid = ($this->cPid ? $this->cPid : $this->pid); |
|
921
|
|
|
if (!$cPid) { |
|
922
|
|
|
Helper::devLog('Invalid PID ' . $cPid . ' for structure definitions', DEVLOG_SEVERITY_ERROR); |
|
923
|
|
|
$this->thumbnailLoaded = TRUE; |
|
924
|
|
|
return $this->thumbnail; |
|
925
|
|
|
} |
|
926
|
|
|
// Load extension configuration. |
|
927
|
|
|
$extConf = unserialize($GLOBALS['TYPO3_CONF_VARS']['EXT']['extConf'][self::$extKey]); |
|
928
|
|
|
if (empty($extConf['fileGrpThumbs'])) { |
|
929
|
|
|
Helper::devLog('No fileGrp for thumbnails specified', DEVLOG_SEVERITY_WARNING); |
|
930
|
|
|
$this->thumbnailLoaded = TRUE; |
|
931
|
|
|
return $this->thumbnail; |
|
932
|
|
|
} |
|
933
|
|
|
$strctId = $this->_getToplevelId(); |
|
934
|
|
|
$metadata = $this->getTitledata($cPid); |
|
935
|
|
|
|
|
936
|
|
|
$queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class) |
|
937
|
|
|
->getQueryBuilderForTable('tx_dlf_structures'); |
|
938
|
|
|
|
|
939
|
|
|
// Get structure element to get thumbnail from. |
|
940
|
|
|
$result = $queryBuilder |
|
941
|
|
|
->select('tx_dlf_structures.thumbnail AS thumbnail') |
|
942
|
|
|
->from('tx_dlf_structures') |
|
943
|
|
|
->where( |
|
944
|
|
|
$queryBuilder->expr()->eq('tx_dlf_structures.pid', intval($cPid)), |
|
945
|
|
|
$queryBuilder->expr()->eq('tx_dlf_structures.index_name', $queryBuilder->expr()->literal($metadata['type'][0])), |
|
946
|
|
|
Helper::whereExpression('tx_dlf_structures') |
|
947
|
|
|
) |
|
948
|
|
|
->setMaxResults(1) |
|
949
|
|
|
->execute(); |
|
950
|
|
|
|
|
951
|
|
|
$allResults = $result->fetchAll(); |
|
952
|
|
|
|
|
953
|
|
|
if (count($allResults) == 1) { |
|
954
|
|
|
$resArray = $allResults[0]; |
|
955
|
|
|
// Get desired thumbnail structure if not the toplevel structure itself. |
|
956
|
|
|
if (!empty($resArray['thumbnail'])) { |
|
957
|
|
|
$strctType = Helper::getIndexNameFromUid($resArray['thumbnail'], 'tx_dlf_structures', $cPid); |
|
958
|
|
|
// Check if this document has a structure element of the desired type. |
|
959
|
|
|
$strctIds = $this->mets->xpath('./mets:structMap[@TYPE="LOGICAL"]//mets:div[@TYPE="' . $strctType . '"]/@ID'); |
|
960
|
|
|
if (!empty($strctIds)) { |
|
961
|
|
|
$strctId = (string) $strctIds[0]; |
|
962
|
|
|
} |
|
963
|
|
|
} |
|
964
|
|
|
// Load smLinks. |
|
965
|
|
|
$this->_getSmLinks(); |
|
966
|
|
|
// Get thumbnail location. |
|
967
|
|
|
if ( |
|
968
|
|
|
$this->_getPhysicalStructure() |
|
969
|
|
|
&& !empty($this->smLinks['l2p'][$strctId]) |
|
970
|
|
|
) { |
|
971
|
|
|
$this->thumbnail = $this->getFileLocation($this->physicalStructureInfo[$this->smLinks['l2p'][$strctId][0]]['files'][$extConf['fileGrpThumbs']]); |
|
972
|
|
|
} else { |
|
973
|
|
|
$this->thumbnail = $this->getFileLocation($this->physicalStructureInfo[$this->physicalStructure[1]]['files'][$extConf['fileGrpThumbs']]); |
|
974
|
|
|
} |
|
975
|
|
|
} else { |
|
976
|
|
|
Helper::devLog('No structure of type "' . $metadata['type'][0] . '" found in database', DEVLOG_SEVERITY_ERROR); |
|
977
|
|
|
} |
|
978
|
|
|
$this->thumbnailLoaded = TRUE; |
|
979
|
|
|
} |
|
980
|
|
|
return $this->thumbnail; |
|
981
|
|
|
} |
|
982
|
|
|
|
|
983
|
|
|
/** |
|
984
|
|
|
* {@inheritDoc} |
|
985
|
|
|
* @see \Kitodo\Dlf\Common\Document::_getToplevelId() |
|
986
|
|
|
*/ |
|
987
|
|
|
protected function _getToplevelId() |
|
988
|
|
|
{ |
|
989
|
|
|
if (empty($this->toplevelId)) { |
|
990
|
|
|
// Get all logical structure nodes with metadata, but without associated METS-Pointers. |
|
991
|
|
|
$divs = $this->mets->xpath('./mets:structMap[@TYPE="LOGICAL"]//mets:div[@DMDID and not(./mets:mptr)]'); |
|
992
|
|
|
if (!empty($divs)) { |
|
993
|
|
|
// Load smLinks. |
|
994
|
|
|
$this->_getSmLinks(); |
|
995
|
|
|
foreach ($divs as $div) { |
|
996
|
|
|
$id = (string) $div['ID']; |
|
997
|
|
|
// Are there physical structure nodes for this logical structure? |
|
998
|
|
|
if (array_key_exists($id, $this->smLinks['l2p'])) { |
|
999
|
|
|
// Yes. That's what we're looking for. |
|
1000
|
|
|
$this->toplevelId = $id; |
|
1001
|
|
|
break; |
|
1002
|
|
|
} elseif (empty($this->toplevelId)) { |
|
1003
|
|
|
// No. Remember this anyway, but keep looking for a better one. |
|
1004
|
|
|
$this->toplevelId = $id; |
|
1005
|
|
|
} |
|
1006
|
|
|
} |
|
1007
|
|
|
} |
|
1008
|
|
|
} |
|
1009
|
|
|
return $this->toplevelId; |
|
1010
|
|
|
} |
|
1011
|
|
|
|
|
1012
|
|
|
/** |
|
1013
|
|
|
* This magic method is executed prior to any serialization of the object |
|
1014
|
|
|
* @see __wakeup() |
|
1015
|
|
|
* |
|
1016
|
|
|
* @access public |
|
1017
|
|
|
* |
|
1018
|
|
|
* @return array Properties to be serialized |
|
1019
|
|
|
*/ |
|
1020
|
|
|
public function __sleep() |
|
1021
|
|
|
{ |
|
1022
|
|
|
// \SimpleXMLElement objects can't be serialized, thus save the XML as string for serialization |
|
1023
|
|
|
$this->asXML = $this->xml->asXML(); |
|
1024
|
|
|
return ['uid', 'pid', 'recordId', 'parentId', 'asXML']; |
|
1025
|
|
|
} |
|
1026
|
|
|
|
|
1027
|
|
|
/** |
|
1028
|
|
|
* This magic method is used for setting a string value for the object |
|
1029
|
|
|
* |
|
1030
|
|
|
* @access public |
|
1031
|
|
|
* |
|
1032
|
|
|
* @return string String representing the METS object |
|
1033
|
|
|
*/ |
|
1034
|
|
|
public function __toString() |
|
1035
|
|
|
{ |
|
1036
|
|
|
$xml = new \DOMDocument('1.0', 'utf-8'); |
|
1037
|
|
|
$xml->appendChild($xml->importNode(dom_import_simplexml($this->mets), TRUE)); |
|
1038
|
|
|
$xml->formatOutput = TRUE; |
|
1039
|
|
|
return $xml->saveXML(); |
|
1040
|
|
|
} |
|
1041
|
|
|
|
|
1042
|
|
|
/** |
|
1043
|
|
|
* This magic method is executed after the object is deserialized |
|
1044
|
|
|
* @see __sleep() |
|
1045
|
|
|
* |
|
1046
|
|
|
* @access public |
|
1047
|
|
|
* |
|
1048
|
|
|
* @return void |
|
1049
|
|
|
*/ |
|
1050
|
|
|
public function __wakeup() |
|
1051
|
|
|
{ |
|
1052
|
|
|
// Turn off libxml's error logging. |
|
1053
|
|
|
$libxmlErrors = libxml_use_internal_errors(TRUE); |
|
1054
|
|
|
// Reload XML from string. |
|
1055
|
|
|
$xml = @simplexml_load_string($this->asXML); |
|
1056
|
|
|
// Reset libxml's error logging. |
|
1057
|
|
|
libxml_use_internal_errors($libxmlErrors); |
|
1058
|
|
|
if ($xml !== FALSE) { |
|
1059
|
|
|
$this->asXML = ''; |
|
1060
|
|
|
$this->xml = $xml; |
|
1061
|
|
|
// Rebuild the unserializable properties. |
|
1062
|
|
|
$this->init(); |
|
1063
|
|
|
} else { |
|
1064
|
|
|
Helper::devLog('Could not load XML after deserialization', DEVLOG_SEVERITY_ERROR); |
|
1065
|
|
|
} |
|
1066
|
|
|
} |
|
1067
|
|
|
} |
|
1068
|
|
|
|
kitodo /
kitodo-presentation
