kitodo /
kitodo-presentation
We could not synchronize checks via GitHub's checks API since Scrutinizer's GitHub App is not installed for this repository.
| Total Complexity | 99 |
| Total Lines | 1298 |
| Duplicated Lines | 0 % |
| Changes | 1 | ||
| Bugs | 0 | Features | 0 |
Complex classes like Doc often do a lot of different things. To break such a class down, we need to identify a cohesive component within that class. A common approach to find such a component is to look for fields/methods that share the same prefixes, or suffixes.
Once you have determined the fields that belong together, you can apply the Extract Class refactoring. If the component makes sense as a sub-class, Extract Subclass is also a candidate, and is often faster.
While breaking up the class, it is a good idea to analyze how other classes use Doc, and based on these observations, apply Extract Interface, too.
| 1 | <?php |
||
| 53 | abstract class Doc |
||
| 54 | { |
||
| 55 | /** |
||
| 56 | * This holds the logger |
||
| 57 | * |
||
| 58 | * @var LogManager |
||
| 59 | * @access protected |
||
| 60 | */ |
||
| 61 | protected $logger; |
||
| 62 | |||
| 63 | /** |
||
| 64 | * This holds the PID for the configuration |
||
| 65 | * |
||
| 66 | * @var int |
||
| 67 | * @access protected |
||
| 68 | */ |
||
| 69 | protected $cPid = 0; |
||
| 70 | |||
| 71 | /** |
||
| 72 | * The extension key |
||
| 73 | * |
||
| 74 | * @var string |
||
| 75 | * @access public |
||
| 76 | */ |
||
| 77 | public static $extKey = 'dlf'; |
||
| 78 | |||
| 79 | /** |
||
| 80 | * This holds the configuration for all supported metadata encodings |
||
| 81 | * @see loadFormats() |
||
| 82 | * |
||
| 83 | * @var array |
||
| 84 | * @access protected |
||
| 85 | */ |
||
| 86 | protected $formats = [ |
||
| 87 | 'OAI' => [ |
||
| 88 | 'rootElement' => 'OAI-PMH', |
||
| 89 | 'namespaceURI' => 'http://www.openarchives.org/OAI/2.0/', |
||
| 90 | ], |
||
| 91 | 'METS' => [ |
||
| 92 | 'rootElement' => 'mets', |
||
| 93 | 'namespaceURI' => 'http://www.loc.gov/METS/', |
||
| 94 | ], |
||
| 95 | 'XLINK' => [ |
||
| 96 | 'rootElement' => 'xlink', |
||
| 97 | 'namespaceURI' => 'http://www.w3.org/1999/xlink', |
||
| 98 | ] |
||
| 99 | ]; |
||
| 100 | |||
| 101 | /** |
||
| 102 | * Are the available metadata formats loaded? |
||
| 103 | * @see $formats |
||
| 104 | * |
||
| 105 | * @var bool |
||
| 106 | * @access protected |
||
| 107 | */ |
||
| 108 | protected $formatsLoaded = false; |
||
| 109 | |||
| 110 | /** |
||
| 111 | * Are there any fulltext files available? This also includes IIIF text annotations |
||
| 112 | * with motivation 'painting' if Kitodo.Presentation is configured to store text |
||
| 113 | * annotations as fulltext. |
||
| 114 | * |
||
| 115 | * @var bool |
||
| 116 | * @access protected |
||
| 117 | */ |
||
| 118 | protected $hasFulltext = false; |
||
| 119 | |||
| 120 | /** |
||
| 121 | * Last searched logical and physical page |
||
| 122 | * |
||
| 123 | * @var array |
||
| 124 | * @access protected |
||
| 125 | */ |
||
| 126 | protected $lastSearchedPhysicalPage = ['logicalPage' => null, 'physicalPage' => null]; |
||
| 127 | |||
| 128 | /** |
||
| 129 | * This holds the documents location |
||
| 130 | * |
||
| 131 | * @var string |
||
| 132 | * @access protected |
||
| 133 | */ |
||
| 134 | protected $location = ''; |
||
| 135 | |||
| 136 | /** |
||
| 137 | * This holds the logical units |
||
| 138 | * |
||
| 139 | * @var array |
||
| 140 | * @access protected |
||
| 141 | */ |
||
| 142 | protected $logicalUnits = []; |
||
| 143 | |||
| 144 | /** |
||
| 145 | * This holds the documents' parsed metadata array with their corresponding |
||
| 146 | * structMap//div's ID (METS) or Range / Manifest / Sequence ID (IIIF) as array key |
||
| 147 | * |
||
| 148 | * @var array |
||
| 149 | * @access protected |
||
| 150 | */ |
||
| 151 | protected $metadataArray = []; |
||
| 152 | |||
| 153 | /** |
||
| 154 | * Is the metadata array loaded? |
||
| 155 | * @see $metadataArray |
||
| 156 | * |
||
| 157 | * @var bool |
||
| 158 | * @access protected |
||
| 159 | */ |
||
| 160 | protected $metadataArrayLoaded = false; |
||
| 161 | |||
| 162 | /** |
||
| 163 | * The holds the total number of pages |
||
| 164 | * |
||
| 165 | * @var int |
||
| 166 | * @access protected |
||
| 167 | */ |
||
| 168 | protected $numPages = 0; |
||
| 169 | |||
| 170 | /** |
||
| 171 | * This holds the UID of the parent document or zero if not multi-volumed |
||
| 172 | * |
||
| 173 | * @var int |
||
| 174 | * @access protected |
||
| 175 | */ |
||
| 176 | protected $parentId = 0; |
||
| 177 | |||
| 178 | /** |
||
| 179 | * This holds the physical structure |
||
| 180 | * |
||
| 181 | * @var array |
||
| 182 | * @access protected |
||
| 183 | */ |
||
| 184 | protected $physicalStructure = []; |
||
| 185 | |||
| 186 | /** |
||
| 187 | * This holds the physical structure metadata |
||
| 188 | * |
||
| 189 | * @var array |
||
| 190 | * @access protected |
||
| 191 | */ |
||
| 192 | protected $physicalStructureInfo = []; |
||
| 193 | |||
| 194 | /** |
||
| 195 | * Is the physical structure loaded? |
||
| 196 | * @see $physicalStructure |
||
| 197 | * |
||
| 198 | * @var bool |
||
| 199 | * @access protected |
||
| 200 | */ |
||
| 201 | protected $physicalStructureLoaded = false; |
||
| 202 | |||
| 203 | /** |
||
| 204 | * This holds the PID of the document or zero if not in database |
||
| 205 | * |
||
| 206 | * @var int |
||
| 207 | * @access protected |
||
| 208 | */ |
||
| 209 | protected $pid = 0; |
||
| 210 | |||
| 211 | /** |
||
| 212 | * This holds the documents' raw text pages with their corresponding |
||
| 213 | * structMap//div's ID (METS) or Range / Manifest / Sequence ID (IIIF) as array key |
||
| 214 | * |
||
| 215 | * @var array |
||
| 216 | * @access protected |
||
| 217 | */ |
||
| 218 | protected $rawTextArray = []; |
||
| 219 | |||
| 220 | /** |
||
| 221 | * Is the document instantiated successfully? |
||
| 222 | * |
||
| 223 | * @var bool |
||
| 224 | * @access protected |
||
| 225 | */ |
||
| 226 | protected $ready = false; |
||
| 227 | |||
| 228 | /** |
||
| 229 | * The METS file's / IIIF manifest's record identifier |
||
| 230 | * |
||
| 231 | * @var string |
||
| 232 | * @access protected |
||
| 233 | */ |
||
| 234 | protected $recordId; |
||
| 235 | |||
| 236 | /** |
||
| 237 | * This holds the singleton object of the document |
||
| 238 | * |
||
| 239 | * @var array (\Kitodo\Dlf\Common\Doc) |
||
| 240 | * @static |
||
| 241 | * @access protected |
||
| 242 | */ |
||
| 243 | protected static $registry = []; |
||
| 244 | |||
| 245 | /** |
||
| 246 | * This holds the UID of the root document or zero if not multi-volumed |
||
| 247 | * |
||
| 248 | * @var int |
||
| 249 | * @access protected |
||
| 250 | */ |
||
| 251 | protected $rootId = 0; |
||
| 252 | |||
| 253 | /** |
||
| 254 | * Is the root id loaded? |
||
| 255 | * @see $rootId |
||
| 256 | * |
||
| 257 | * @var bool |
||
| 258 | * @access protected |
||
| 259 | */ |
||
| 260 | protected $rootIdLoaded = false; |
||
| 261 | |||
| 262 | /** |
||
| 263 | * This holds the smLinks between logical and physical structMap |
||
| 264 | * |
||
| 265 | * @var array |
||
| 266 | * @access protected |
||
| 267 | */ |
||
| 268 | protected $smLinks = ['l2p' => [], 'p2l' => []]; |
||
| 269 | |||
| 270 | /** |
||
| 271 | * Are the smLinks loaded? |
||
| 272 | * @see $smLinks |
||
| 273 | * |
||
| 274 | * @var bool |
||
| 275 | * @access protected |
||
| 276 | */ |
||
| 277 | protected $smLinksLoaded = false; |
||
| 278 | |||
| 279 | /** |
||
| 280 | * This holds the logical structure |
||
| 281 | * |
||
| 282 | * @var array |
||
| 283 | * @access protected |
||
| 284 | */ |
||
| 285 | protected $tableOfContents = []; |
||
| 286 | |||
| 287 | /** |
||
| 288 | * Is the table of contents loaded? |
||
| 289 | * @see $tableOfContents |
||
| 290 | * |
||
| 291 | * @var bool |
||
| 292 | * @access protected |
||
| 293 | */ |
||
| 294 | protected $tableOfContentsLoaded = false; |
||
| 295 | |||
| 296 | /** |
||
| 297 | * This holds the document's thumbnail location |
||
| 298 | * |
||
| 299 | * @var string |
||
| 300 | * @access protected |
||
| 301 | */ |
||
| 302 | protected $thumbnail = ''; |
||
| 303 | |||
| 304 | /** |
||
| 305 | * Is the document's thumbnail location loaded? |
||
| 306 | * @see $thumbnail |
||
| 307 | * |
||
| 308 | * @var bool |
||
| 309 | * @access protected |
||
| 310 | */ |
||
| 311 | protected $thumbnailLoaded = false; |
||
| 312 | |||
| 313 | /** |
||
| 314 | * This holds the toplevel structure's @ID (METS) or the manifest's @id (IIIF) |
||
| 315 | * |
||
| 316 | * @var string |
||
| 317 | * @access protected |
||
| 318 | */ |
||
| 319 | protected $toplevelId = ''; |
||
| 320 | |||
| 321 | /** |
||
| 322 | * This holds the UID or the URL of the document |
||
| 323 | * |
||
| 324 | * @var mixed |
||
| 325 | * @access protected |
||
| 326 | */ |
||
| 327 | protected $uid = 0; |
||
| 328 | |||
| 329 | /** |
||
| 330 | * This holds the whole XML file as \SimpleXMLElement object |
||
| 331 | * |
||
| 332 | * @var \SimpleXMLElement |
||
| 333 | * @access protected |
||
| 334 | */ |
||
| 335 | protected $xml; |
||
| 336 | |||
| 337 | /** |
||
| 338 | * This clears the static registry to prevent memory exhaustion |
||
| 339 | * |
||
| 340 | * @access public |
||
| 341 | * |
||
| 342 | * @static |
||
| 343 | * |
||
| 344 | * @return void |
||
| 345 | */ |
||
| 346 | public static function clearRegistry() |
||
| 347 | { |
||
| 348 | // Reset registry array. |
||
| 349 | self::$registry = []; |
||
| 350 | } |
||
| 351 | |||
| 352 | /** |
||
| 353 | * This ensures that the recordId, if existent, is retrieved from the document |
||
| 354 | * |
||
| 355 | * @access protected |
||
| 356 | * |
||
| 357 | * @abstract |
||
| 358 | * |
||
| 359 | * @param int $pid: ID of the configuration page with the recordId config |
||
| 360 | * |
||
| 361 | */ |
||
| 362 | protected abstract function establishRecordId($pid); |
||
| 363 | |||
| 364 | /** |
||
| 365 | * Source document PHP object which is represented by a Document instance |
||
| 366 | * |
||
| 367 | * @access protected |
||
| 368 | * |
||
| 369 | * @abstract |
||
| 370 | * |
||
| 371 | * @return \SimpleXMLElement|IiifResourceInterface An PHP object representation of |
||
| 372 | * the current document. SimpleXMLElement for METS, IiifResourceInterface for IIIF |
||
| 373 | */ |
||
| 374 | protected abstract function getDocument(); |
||
| 375 | |||
| 376 | /** |
||
| 377 | * This gets the location of a downloadable file for a physical page or track |
||
| 378 | * |
||
| 379 | * @access public |
||
| 380 | * |
||
| 381 | * @abstract |
||
| 382 | * |
||
| 383 | * @param string $id: The @ID attribute of the file node (METS) or the @id property of the IIIF resource |
||
| 384 | * |
||
| 385 | * @return string The file's location as URL |
||
| 386 | */ |
||
| 387 | public abstract function getDownloadLocation($id); |
||
| 388 | |||
| 389 | /** |
||
| 390 | * This gets the location of a file representing a physical page or track |
||
| 391 | * |
||
| 392 | * @access public |
||
| 393 | * |
||
| 394 | * @abstract |
||
| 395 | * |
||
| 396 | * @param string $id: The @ID attribute of the file node (METS) or the @id property of the IIIF resource |
||
| 397 | * |
||
| 398 | * @return string The file's location as URL |
||
| 399 | */ |
||
| 400 | public abstract function getFileLocation($id); |
||
| 401 | |||
| 402 | /** |
||
| 403 | * This gets the MIME type of a file representing a physical page or track |
||
| 404 | * |
||
| 405 | * @access public |
||
| 406 | * |
||
| 407 | * @abstract |
||
| 408 | * |
||
| 409 | * @param string $id: The @ID attribute of the file node |
||
| 410 | * |
||
| 411 | * @return string The file's MIME type |
||
| 412 | */ |
||
| 413 | public abstract function getFileMimeType($id); |
||
| 414 | |||
| 415 | /** |
||
| 416 | * This is a singleton class, thus an instance must be created by this method |
||
| 417 | * |
||
| 418 | * @access public |
||
| 419 | * |
||
| 420 | * @static |
||
| 421 | * |
||
| 422 | * @param string $location: The URL of XML file or the IRI of the IIIF resource |
||
| 423 | * @param array $settings |
||
| 424 | * @param bool $forceReload: Force reloading the document instead of returning the cached instance |
||
| 425 | * |
||
| 426 | * @return \Kitodo\Dlf\Common\Doc Instance of this class, either MetsDocument or IiifManifest |
||
| 427 | */ |
||
| 428 | public static function &getInstance($location, $settings = [], $forceReload = false) |
||
| 429 | { |
||
| 430 | // Create new instance depending on format (METS or IIIF) ... |
||
| 431 | $instance = null; |
||
| 432 | $documentFormat = null; |
||
| 433 | $xml = null; |
||
| 434 | $iiif = null; |
||
| 435 | |||
| 436 | // Try to load a file from the url |
||
| 437 | if (GeneralUtility::isValidUrl($location)) { |
||
| 438 | // Load extension configuration |
||
| 439 | $extConf = GeneralUtility::makeInstance(ExtensionConfiguration::class)->get(self::$extKey); |
||
| 440 | // Set user-agent to identify self when fetching XML data. |
||
| 441 | if (!empty($extConf['useragent'])) { |
||
| 442 | @ini_set('user_agent', $extConf['useragent']); |
||
| 443 | } |
||
| 444 | $content = GeneralUtility::getUrl($location); |
||
| 445 | if ($content !== false) { |
||
| 446 | $xml = Helper::getXmlFileAsString($content); |
||
| 447 | if ($xml !== false) { |
||
| 448 | /* @var $xml \SimpleXMLElement */ |
||
| 449 | $xml->registerXPathNamespace('mets', 'http://www.loc.gov/METS/'); |
||
| 450 | $xpathResult = $xml->xpath('//mets:mets'); |
||
| 451 | $documentFormat = !empty($xpathResult) ? 'METS' : null; |
||
| 452 | } else { |
||
| 453 | // Try to load file as IIIF resource instead. |
||
| 454 | $contentAsJsonArray = json_decode($content, true); |
||
| 455 | if ($contentAsJsonArray !== null) { |
||
| 456 | // Load plugin configuration. |
||
| 457 | $conf = GeneralUtility::makeInstance(ExtensionConfiguration::class)->get(self::$extKey); |
||
| 458 | IiifHelper::setUrlReader(IiifUrlReader::getInstance()); |
||
| 459 | IiifHelper::setMaxThumbnailHeight($conf['iiifThumbnailHeight']); |
||
| 460 | IiifHelper::setMaxThumbnailWidth($conf['iiifThumbnailWidth']); |
||
| 461 | $iiif = IiifHelper::loadIiifResource($contentAsJsonArray); |
||
| 462 | if ($iiif instanceof IiifResourceInterface) { |
||
| 463 | $documentFormat = 'IIIF'; |
||
| 464 | } |
||
| 465 | } |
||
| 466 | } |
||
| 467 | } |
||
| 468 | } |
||
| 469 | |||
| 470 | // Sanitize input. |
||
| 471 | $pid = max(intval($settings['storagePid']), 0); |
||
| 472 | if ($documentFormat == 'METS') { |
||
| 473 | $instance = new MetsDocument($location, $pid, $xml); |
||
| 474 | } elseif ($documentFormat == 'IIIF') { |
||
| 475 | $instance = new IiifManifest($location, $pid, $iiif); |
||
| 476 | } |
||
| 477 | |||
| 478 | return $instance; |
||
| 479 | } |
||
| 480 | |||
| 481 | /** |
||
| 482 | * This gets details about a logical structure element |
||
| 483 | * |
||
| 484 | * @access public |
||
| 485 | * |
||
| 486 | * @abstract |
||
| 487 | * |
||
| 488 | * @param string $id: The @ID attribute of the logical structure node (METS) or |
||
| 489 | * the @id property of the Manifest / Range (IIIF) |
||
| 490 | * @param bool $recursive: Whether to include the child elements / resources |
||
| 491 | * |
||
| 492 | * @return array Array of the element's id, label, type and physical page indexes/mptr link |
||
| 493 | */ |
||
| 494 | public abstract function getLogicalStructure($id, $recursive = false); |
||
| 495 | |||
| 496 | /** |
||
| 497 | * This extracts all the metadata for a logical structure node |
||
| 498 | * |
||
| 499 | * @access public |
||
| 500 | * |
||
| 501 | * @abstract |
||
| 502 | * |
||
| 503 | * @param string $id: The @ID attribute of the logical structure node (METS) or the @id property |
||
| 504 | * of the Manifest / Range (IIIF) |
||
| 505 | * @param int $cPid: The PID for the metadata definitions |
||
| 506 | * (defaults to $this->cPid or $this->pid) |
||
| 507 | * |
||
| 508 | * @return array The logical structure node's / the IIIF resource's parsed metadata array |
||
| 509 | */ |
||
| 510 | public abstract function getMetadata($id, $cPid = 0); |
||
| 511 | |||
| 512 | /** |
||
| 513 | * This returns the first corresponding physical page number of a given logical page label |
||
| 514 | * |
||
| 515 | * @access public |
||
| 516 | * |
||
| 517 | * @param string $logicalPage: The label (or a part of the label) of the logical page |
||
| 518 | * |
||
| 519 | * @return int The physical page number |
||
| 520 | */ |
||
| 521 | public function getPhysicalPage($logicalPage) |
||
| 522 | { |
||
| 523 | if ( |
||
| 524 | !empty($this->lastSearchedPhysicalPage['logicalPage']) |
||
| 525 | && $this->lastSearchedPhysicalPage['logicalPage'] == $logicalPage |
||
| 526 | ) { |
||
| 527 | return $this->lastSearchedPhysicalPage['physicalPage']; |
||
| 528 | } else { |
||
| 529 | $physicalPage = 0; |
||
| 530 | foreach ($this->physicalStructureInfo as $page) { |
||
| 531 | if (strpos($page['orderlabel'], $logicalPage) !== false) { |
||
| 532 | $this->lastSearchedPhysicalPage['logicalPage'] = $logicalPage; |
||
| 533 | $this->lastSearchedPhysicalPage['physicalPage'] = $physicalPage; |
||
| 534 | return $physicalPage; |
||
| 535 | } |
||
| 536 | $physicalPage++; |
||
| 537 | } |
||
| 538 | } |
||
| 539 | return 1; |
||
| 540 | } |
||
| 541 | |||
| 542 | /** |
||
| 543 | * This extracts the OCR full text for a physical structure node / IIIF Manifest / Canvas. Text might be |
||
| 544 | * given as ALTO for METS or as annotations or ALTO for IIIF resources. |
||
| 545 | * |
||
| 546 | * @access public |
||
| 547 | * |
||
| 548 | * @abstract |
||
| 549 | * |
||
| 550 | * @param string $id: The @ID attribute of the physical structure node (METS) or the @id property |
||
| 551 | * of the Manifest / Range (IIIF) |
||
| 552 | * |
||
| 553 | * @return string The OCR full text |
||
| 554 | */ |
||
| 555 | public abstract function getFullText($id); |
||
| 556 | |||
| 557 | /** |
||
| 558 | * This extracts the OCR full text for a physical structure node / IIIF Manifest / Canvas from an |
||
| 559 | * XML full text representation (currently only ALTO). For IIIF manifests, ALTO documents have |
||
| 560 | * to be given in the Canvas' / Manifest's "seeAlso" property. |
||
| 561 | * |
||
| 562 | * @param string $id: The @ID attribute of the physical structure node (METS) or the @id property |
||
| 563 | * of the Manifest / Range (IIIF) |
||
| 564 | * |
||
| 565 | * @return string The OCR full text |
||
| 566 | */ |
||
| 567 | protected function getFullTextFromXml($id) |
||
| 568 | { |
||
| 569 | $fullText = ''; |
||
| 570 | // Load available text formats, ... |
||
| 571 | $this->loadFormats(); |
||
| 572 | // ... physical structure ... |
||
| 573 | $this->_getPhysicalStructure(); |
||
| 574 | // ... and extension configuration. |
||
| 575 | $extConf = GeneralUtility::makeInstance(ExtensionConfiguration::class)->get(self::$extKey); |
||
| 576 | $fileGrpsFulltext = GeneralUtility::trimExplode(',', $extConf['fileGrpFulltext']); |
||
| 577 | if (!empty($this->physicalStructureInfo[$id])) { |
||
| 578 | while ($fileGrpFulltext = array_shift($fileGrpsFulltext)) { |
||
| 579 | if (!empty($this->physicalStructureInfo[$id]['files'][$fileGrpFulltext])) { |
||
| 580 | // Get full text file. |
||
| 581 | $fileContent = GeneralUtility::getUrl($this->getFileLocation($this->physicalStructureInfo[$id]['files'][$fileGrpFulltext])); |
||
| 582 | if ($fileContent !== false) { |
||
| 583 | $textFormat = $this->getTextFormat($fileContent); |
||
| 584 | } else { |
||
| 585 | $this->logger->warning('Couldn\'t load full text file for structure node @ID "' . $id . '"'); |
||
| 586 | return $fullText; |
||
| 587 | } |
||
| 588 | break; |
||
| 589 | } |
||
| 590 | } |
||
| 591 | } else { |
||
| 592 | $this->logger->warning('Invalid structure node @ID "' . $id . '"'); |
||
| 593 | return $fullText; |
||
| 594 | } |
||
| 595 | // Is this text format supported? |
||
| 596 | // This part actually differs from previous version of indexed OCR |
||
| 597 | if (!empty($fileContent) && !empty($this->formats[$textFormat])) { |
||
|
Comprehensibility
Best Practice
introduced
by
|
|||
| 598 | $textMiniOcr = ''; |
||
| 599 | if (!empty($this->formats[$textFormat]['class'])) { |
||
| 600 | $class = $this->formats[$textFormat]['class']; |
||
| 601 | // Get the raw text from class. |
||
| 602 | if ( |
||
| 603 | class_exists($class) |
||
| 604 | && ($obj = GeneralUtility::makeInstance($class)) instanceof FulltextInterface |
||
| 605 | ) { |
||
| 606 | // Load XML from file. |
||
| 607 | $ocrTextXml = Helper::getXmlFileAsString($fileContent); |
||
| 608 | $textMiniOcr = $obj->getTextAsMiniOcr($ocrTextXml); |
||
| 609 | $this->rawTextArray[$id] = $textMiniOcr; |
||
| 610 | } else { |
||
| 611 | $this->logger->warning('Invalid class/method "' . $class . '->getRawText()" for text format "' . $textFormat . '"'); |
||
| 612 | } |
||
| 613 | } |
||
| 614 | $fullText = $textMiniOcr; |
||
| 615 | } else { |
||
| 616 | $this->logger->warning('Unsupported text format "' . $textFormat . '" in physical node with @ID "' . $id . '"'); |
||
| 617 | } |
||
| 618 | return $fullText; |
||
| 619 | } |
||
| 620 | |||
| 621 | /** |
||
| 622 | * Get format of the OCR full text |
||
| 623 | * |
||
| 624 | * @access private |
||
| 625 | * |
||
| 626 | * @param string $fileContent: content of the XML file |
||
| 627 | * |
||
| 628 | * @return string The format of the OCR full text |
||
| 629 | */ |
||
| 630 | private function getTextFormat($fileContent) |
||
| 631 | { |
||
| 632 | // Get the root element's name as text format. |
||
| 633 | return strtoupper(Helper::getXmlFileAsString($fileContent)->getName()); |
||
| 634 | } |
||
| 635 | |||
| 636 | /** |
||
| 637 | * This determines a title for the given document |
||
| 638 | * |
||
| 639 | * @access public |
||
| 640 | * |
||
| 641 | * @static |
||
| 642 | * |
||
| 643 | * @param int $uid: The UID of the document |
||
| 644 | * @param bool $recursive: Search superior documents for a title, too? |
||
| 645 | * |
||
| 646 | * @return string The title of the document itself or a parent document |
||
| 647 | */ |
||
| 648 | public static function getTitle($uid, $recursive = false) |
||
| 649 | { |
||
| 650 | $title = ''; |
||
| 651 | // Sanitize input. |
||
| 652 | $uid = max(intval($uid), 0); |
||
| 653 | if ($uid) { |
||
| 654 | $queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class) |
||
| 655 | ->getQueryBuilderForTable('tx_dlf_documents'); |
||
| 656 | |||
| 657 | $result = $queryBuilder |
||
| 658 | ->select( |
||
| 659 | 'tx_dlf_documents.title', |
||
| 660 | 'tx_dlf_documents.partof' |
||
| 661 | ) |
||
| 662 | ->from('tx_dlf_documents') |
||
| 663 | ->where( |
||
| 664 | $queryBuilder->expr()->eq('tx_dlf_documents.uid', $uid), |
||
| 665 | Helper::whereExpression('tx_dlf_documents') |
||
| 666 | ) |
||
| 667 | ->setMaxResults(1) |
||
| 668 | ->execute(); |
||
| 669 | |||
| 670 | if ($resArray = $result->fetch()) { |
||
| 671 | // Get title information. |
||
| 672 | $title = $resArray['title']; |
||
| 673 | $partof = $resArray['partof']; |
||
| 674 | // Search parent documents recursively for a title? |
||
| 675 | if ( |
||
| 676 | $recursive |
||
| 677 | && empty($title) |
||
| 678 | && intval($partof) |
||
| 679 | && $partof != $uid |
||
| 680 | ) { |
||
| 681 | $title = self::getTitle($partof, true); |
||
| 682 | } |
||
| 683 | } else { |
||
| 684 | Helper::log('No document with UID ' . $uid . ' found or document not accessible', LOG_SEVERITY_WARNING); |
||
| 685 | } |
||
| 686 | } else { |
||
| 687 | Helper::log('Invalid UID ' . $uid . ' for document', LOG_SEVERITY_ERROR); |
||
| 688 | } |
||
| 689 | return $title; |
||
| 690 | } |
||
| 691 | |||
| 692 | /** |
||
| 693 | * This extracts all the metadata for the toplevel logical structure node / resource |
||
| 694 | * |
||
| 695 | * @access public |
||
| 696 | * |
||
| 697 | * @param int $cPid: The PID for the metadata definitions |
||
| 698 | * |
||
| 699 | * @return array The logical structure node's / resource's parsed metadata array |
||
| 700 | */ |
||
| 701 | public function getTitledata($cPid = 0) |
||
| 702 | { |
||
| 703 | $titledata = $this->getMetadata($this->_getToplevelId(), $cPid); |
||
| 704 | // Add information from METS structural map to titledata array. |
||
| 705 | if ($this instanceof MetsDocument) { |
||
| 706 | $this->addMetadataFromMets($titledata, $this->_getToplevelId()); |
||
| 707 | } |
||
| 708 | // Set record identifier for METS file / IIIF manifest if not present. |
||
| 709 | if ( |
||
| 710 | is_array($titledata) |
||
| 711 | && array_key_exists('record_id', $titledata) |
||
| 712 | ) { |
||
| 713 | if ( |
||
| 714 | !empty($this->recordId) |
||
| 715 | && !in_array($this->recordId, $titledata['record_id']) |
||
| 716 | ) { |
||
| 717 | array_unshift($titledata['record_id'], $this->recordId); |
||
| 718 | } |
||
| 719 | } |
||
| 720 | return $titledata; |
||
| 721 | } |
||
| 722 | |||
| 723 | /** |
||
| 724 | * Traverse a logical (sub-) structure tree to find the structure with the requested logical id and return it's depth. |
||
| 725 | * |
||
| 726 | * @access protected |
||
| 727 | * |
||
| 728 | * @param array $structure: logical structure array |
||
| 729 | * @param int $depth: current tree depth |
||
| 730 | * @param string $logId: ID of the logical structure whose depth is requested |
||
| 731 | * |
||
| 732 | * @return int|bool: false if structure with $logId is not a child of this substructure, |
||
| 733 | * or the actual depth. |
||
| 734 | */ |
||
| 735 | protected function getTreeDepth($structure, $depth, $logId) |
||
| 748 | } |
||
| 749 | |||
| 750 | /** |
||
| 751 | * Get the tree depth of a logical structure element within the table of content |
||
| 752 | * |
||
| 753 | * @access public |
||
| 754 | * |
||
| 755 | * @param string $logId: The id of the logical structure element whose depth is requested |
||
| 756 | * @return int|bool tree depth as integer or false if no element with $logId exists within the TOC. |
||
| 757 | */ |
||
| 758 | public function getStructureDepth($logId) |
||
| 759 | { |
||
| 760 | return $this->getTreeDepth($this->_getTableOfContents(), 1, $logId); |
||
| 761 | } |
||
| 762 | |||
| 763 | /** |
||
| 764 | * This sets some basic class properties |
||
| 765 | * |
||
| 766 | * @access protected |
||
| 767 | * |
||
| 768 | * @abstract |
||
| 769 | * |
||
| 770 | * @return void |
||
| 771 | */ |
||
| 772 | protected abstract function init(); |
||
| 773 | |||
| 774 | /** |
||
| 775 | * Reuse any document object that might have been already loaded to determine wether document is METS or IIIF |
||
| 776 | * |
||
| 777 | * @access protected |
||
| 778 | * |
||
| 779 | * @abstract |
||
| 780 | * |
||
| 781 | * @param \SimpleXMLElement|IiifResourceInterface $preloadedDocument: any instance that has already been loaded |
||
| 782 | * |
||
| 783 | * @return bool true if $preloadedDocument can actually be reused, false if it has to be loaded again |
||
| 784 | */ |
||
| 785 | protected abstract function setPreloadedDocument($preloadedDocument); |
||
| 786 | |||
| 787 | /** |
||
| 788 | * METS/IIIF specific part of loading a location |
||
| 789 | * |
||
| 790 | * @access protected |
||
| 791 | * |
||
| 792 | * @abstract |
||
| 793 | * |
||
| 794 | * @param string $location: The URL of the file to load |
||
| 795 | * |
||
| 796 | * @return bool true on success or false on failure |
||
| 797 | */ |
||
| 798 | protected abstract function loadLocation($location); |
||
| 799 | |||
| 800 | /** |
||
| 801 | * Load XML file / IIIF resource from URL |
||
| 802 | * |
||
| 803 | * @access protected |
||
| 804 | * |
||
| 805 | * @param string $location: The URL of the file to load |
||
| 806 | * |
||
| 807 | * @return bool true on success or false on failure |
||
| 808 | */ |
||
| 809 | protected function load($location) |
||
| 810 | { |
||
| 811 | // Load XML / JSON-LD file. |
||
| 812 | if (GeneralUtility::isValidUrl($location)) { |
||
| 813 | // Load extension configuration |
||
| 814 | $extConf = GeneralUtility::makeInstance(ExtensionConfiguration::class)->get(self::$extKey); |
||
| 815 | // Set user-agent to identify self when fetching XML / JSON-LD data. |
||
| 816 | if (!empty($extConf['useragent'])) { |
||
| 817 | @ini_set('user_agent', $extConf['useragent']); |
||
| 818 | } |
||
| 819 | // the actual loading is format specific |
||
| 820 | return $this->loadLocation($location); |
||
| 821 | } else { |
||
| 822 | $this->logger->error('Invalid file location "' . $location . '" for document loading'); |
||
| 823 | } |
||
| 824 | return false; |
||
| 825 | } |
||
| 826 | |||
| 827 | /** |
||
| 828 | * Analyze the document if it contains any fulltext that needs to be indexed. |
||
| 829 | * |
||
| 830 | * @access protected |
||
| 831 | * |
||
| 832 | * @abstract |
||
| 833 | */ |
||
| 834 | protected abstract function ensureHasFulltextIsSet(); |
||
| 835 | |||
| 836 | /** |
||
| 837 | * Register all available data formats |
||
| 838 | * |
||
| 839 | * @access protected |
||
| 840 | * |
||
| 841 | * @return void |
||
| 842 | */ |
||
| 843 | protected function loadFormats() |
||
| 872 | } |
||
| 873 | } |
||
| 874 | |||
| 875 | /** |
||
| 876 | * Register all available namespaces for a \SimpleXMLElement object |
||
| 877 | * |
||
| 878 | * @access public |
||
| 879 | * |
||
| 880 | * @param \SimpleXMLElement|\DOMXPath &$obj: \SimpleXMLElement or \DOMXPath object |
||
| 881 | * |
||
| 882 | * @return void |
||
| 883 | */ |
||
| 884 | public function registerNamespaces(&$obj) |
||
| 900 | } |
||
| 901 | } |
||
| 902 | |||
| 903 | /** |
||
| 904 | * Get the ID of the parent document if the current document has one. Also save a parent document |
||
| 905 | * to the database and the Solr index if their $pid and the current $pid differ. |
||
| 906 | * Currently only applies to METS documents. |
||
| 907 | * |
||
| 908 | * @access protected |
||
| 909 | * |
||
| 910 | * @abstract |
||
| 911 | * |
||
| 912 | * @return int The parent document's id. |
||
| 913 | */ |
||
| 914 | protected abstract function getParentDocumentUidForSaving($pid, $core, $owner); |
||
| 915 | |||
| 916 | /** |
||
| 917 | * This returns $this->cPid via __get() |
||
| 918 | * |
||
| 919 | * @access protected |
||
| 920 | * |
||
| 921 | * @return int The PID of the metadata definitions |
||
| 922 | */ |
||
| 923 | protected function _getCPid() |
||
| 924 | { |
||
| 925 | return $this->cPid; |
||
| 926 | } |
||
| 927 | |||
| 928 | /** |
||
| 929 | * This returns $this->hasFulltext via __get() |
||
| 930 | * |
||
| 931 | * @access protected |
||
| 932 | * |
||
| 933 | * @return bool Are there any fulltext files available? |
||
| 934 | */ |
||
| 935 | protected function _getHasFulltext() |
||
| 936 | { |
||
| 937 | $this->ensureHasFulltextIsSet(); |
||
| 938 | return $this->hasFulltext; |
||
| 939 | } |
||
| 940 | |||
| 941 | /** |
||
| 942 | * This returns $this->location via __get() |
||
| 943 | * |
||
| 944 | * @access protected |
||
| 945 | * |
||
| 946 | * @return string The location of the document |
||
| 947 | */ |
||
| 948 | protected function _getLocation() |
||
| 949 | { |
||
| 950 | return $this->location; |
||
| 951 | } |
||
| 952 | |||
| 953 | /** |
||
| 954 | * Format specific part of building the document's metadata array |
||
| 955 | * |
||
| 956 | * @access protected |
||
| 957 | * |
||
| 958 | * @abstract |
||
| 959 | * |
||
| 960 | * @param int $cPid |
||
| 961 | */ |
||
| 962 | protected abstract function prepareMetadataArray($cPid); |
||
| 963 | |||
| 964 | /** |
||
| 965 | * This builds an array of the document's metadata |
||
| 966 | * |
||
| 967 | * @access protected |
||
| 968 | * |
||
| 969 | * @return array Array of metadata with their corresponding logical structure node ID as key |
||
| 970 | */ |
||
| 971 | protected function _getMetadataArray() |
||
| 972 | { |
||
| 973 | // Set metadata definitions' PID. |
||
| 974 | $cPid = ($this->cPid ? $this->cPid : $this->pid); |
||
| 975 | if (!$cPid) { |
||
| 976 | $this->logger->error('Invalid PID ' . $cPid . ' for metadata definitions'); |
||
| 977 | return []; |
||
| 978 | } |
||
| 979 | if ( |
||
| 980 | !$this->metadataArrayLoaded |
||
| 981 | || $this->metadataArray[0] != $cPid |
||
| 982 | ) { |
||
| 983 | $this->prepareMetadataArray($cPid); |
||
| 984 | $this->metadataArray[0] = $cPid; |
||
| 985 | $this->metadataArrayLoaded = true; |
||
| 986 | } |
||
| 987 | return $this->metadataArray; |
||
| 988 | } |
||
| 989 | |||
| 990 | /** |
||
| 991 | * This returns $this->numPages via __get() |
||
| 992 | * |
||
| 993 | * @access protected |
||
| 994 | * |
||
| 995 | * @return int The total number of pages and/or tracks |
||
| 996 | */ |
||
| 997 | protected function _getNumPages() |
||
| 998 | { |
||
| 999 | $this->_getPhysicalStructure(); |
||
| 1000 | return $this->numPages; |
||
| 1001 | } |
||
| 1002 | |||
| 1003 | /** |
||
| 1004 | * This returns $this->parentId via __get() |
||
| 1005 | * |
||
| 1006 | * @access protected |
||
| 1007 | * |
||
| 1008 | * @return int The UID of the parent document or zero if not applicable |
||
| 1009 | */ |
||
| 1010 | protected function _getParentId() |
||
| 1011 | { |
||
| 1012 | return $this->parentId; |
||
| 1013 | } |
||
| 1014 | |||
| 1015 | /** |
||
| 1016 | * This builds an array of the document's physical structure |
||
| 1017 | * |
||
| 1018 | * @access protected |
||
| 1019 | * |
||
| 1020 | * @abstract |
||
| 1021 | * |
||
| 1022 | * @return array Array of physical elements' id, type, label and file representations ordered |
||
| 1023 | * by @ORDER attribute / IIIF Sequence's Canvases |
||
| 1024 | */ |
||
| 1025 | protected abstract function _getPhysicalStructure(); |
||
| 1026 | |||
| 1027 | /** |
||
| 1028 | * This gives an array of the document's physical structure metadata |
||
| 1029 | * |
||
| 1030 | * @access protected |
||
| 1031 | * |
||
| 1032 | * @return array Array of elements' type, label and file representations ordered by @ID attribute / Canvas order |
||
| 1033 | */ |
||
| 1034 | protected function _getPhysicalStructureInfo() |
||
| 1035 | { |
||
| 1036 | // Is there no physical structure array yet? |
||
| 1037 | if (!$this->physicalStructureLoaded) { |
||
| 1038 | // Build physical structure array. |
||
| 1039 | $this->_getPhysicalStructure(); |
||
| 1040 | } |
||
| 1041 | return $this->physicalStructureInfo; |
||
| 1042 | } |
||
| 1043 | |||
| 1044 | /** |
||
| 1045 | * This returns $this->pid via __get() |
||
| 1046 | * |
||
| 1047 | * @access protected |
||
| 1048 | * |
||
| 1049 | * @return int The PID of the document or zero if not in database |
||
| 1050 | */ |
||
| 1051 | protected function _getPid() |
||
| 1052 | { |
||
| 1053 | return $this->pid; |
||
| 1054 | } |
||
| 1055 | |||
| 1056 | /** |
||
| 1057 | * This returns $this->ready via __get() |
||
| 1058 | * |
||
| 1059 | * @access protected |
||
| 1060 | * |
||
| 1061 | * @return bool Is the document instantiated successfully? |
||
| 1062 | */ |
||
| 1063 | protected function _getReady() |
||
| 1064 | { |
||
| 1065 | return $this->ready; |
||
| 1066 | } |
||
| 1067 | |||
| 1068 | /** |
||
| 1069 | * This returns $this->recordId via __get() |
||
| 1070 | * |
||
| 1071 | * @access protected |
||
| 1072 | * |
||
| 1073 | * @return mixed The METS file's / IIIF manifest's record identifier |
||
| 1074 | */ |
||
| 1075 | protected function _getRecordId() |
||
| 1076 | { |
||
| 1077 | return $this->recordId; |
||
| 1078 | } |
||
| 1079 | |||
| 1080 | /** |
||
| 1081 | * This returns $this->rootId via __get() |
||
| 1082 | * |
||
| 1083 | * @access protected |
||
| 1084 | * |
||
| 1085 | * @return int The UID of the root document or zero if not applicable |
||
| 1086 | */ |
||
| 1087 | protected function _getRootId() |
||
| 1088 | { |
||
| 1089 | if (!$this->rootIdLoaded) { |
||
| 1090 | if ($this->parentId) { |
||
| 1091 | $parent = self::getInstance($this->parentId, ['storagePid' => $this->pid]); |
||
| 1092 | $this->rootId = $parent->rootId; |
||
| 1093 | } |
||
| 1094 | $this->rootIdLoaded = true; |
||
| 1095 | } |
||
| 1096 | return $this->rootId; |
||
| 1097 | } |
||
| 1098 | |||
| 1099 | /** |
||
| 1100 | * This returns the smLinks between logical and physical structMap (METS) and models the |
||
| 1101 | * relation between IIIF Canvases and Manifests / Ranges in the same way |
||
| 1102 | * |
||
| 1103 | * @access protected |
||
| 1104 | * |
||
| 1105 | * @abstract |
||
| 1106 | * |
||
| 1107 | * @return array The links between logical and physical nodes / Range, Manifest and Canvas |
||
| 1108 | */ |
||
| 1109 | protected abstract function _getSmLinks(); |
||
| 1110 | |||
| 1111 | /** |
||
| 1112 | * This builds an array of the document's logical structure |
||
| 1113 | * |
||
| 1114 | * @access protected |
||
| 1115 | * |
||
| 1116 | * @return array Array of structure nodes' id, label, type and physical page indexes/mptr / Canvas link with original hierarchy preserved |
||
| 1117 | */ |
||
| 1118 | protected function _getTableOfContents() |
||
| 1127 | } |
||
| 1128 | |||
| 1129 | /** |
||
| 1130 | * This returns the document's thumbnail location |
||
| 1131 | * |
||
| 1132 | * @access protected |
||
| 1133 | * |
||
| 1134 | * @abstract |
||
| 1135 | * |
||
| 1136 | * @param bool $forceReload: Force reloading the thumbnail instead of returning the cached value |
||
| 1137 | * |
||
| 1138 | * @return string The document's thumbnail location |
||
| 1139 | */ |
||
| 1140 | protected abstract function _getThumbnail($forceReload = false); |
||
| 1141 | |||
| 1142 | /** |
||
| 1143 | * This returns the ID of the toplevel logical structure node |
||
| 1144 | * |
||
| 1145 | * @access protected |
||
| 1146 | * |
||
| 1147 | * @abstract |
||
| 1148 | * |
||
| 1149 | * @return string The logical structure node's ID |
||
| 1150 | */ |
||
| 1151 | protected abstract function _getToplevelId(); |
||
| 1152 | |||
| 1153 | /** |
||
| 1154 | * This returns $this->uid via __get() |
||
| 1155 | * |
||
| 1156 | * @access protected |
||
| 1157 | * |
||
| 1158 | * @return mixed The UID or the URL of the document |
||
| 1159 | */ |
||
| 1160 | protected function _getUid() |
||
| 1161 | { |
||
| 1162 | return $this->uid; |
||
| 1163 | } |
||
| 1164 | |||
| 1165 | /** |
||
| 1166 | * This sets $this->cPid via __set() |
||
| 1167 | * |
||
| 1168 | * @access protected |
||
| 1169 | * |
||
| 1170 | * @param int $value: The new PID for the metadata definitions |
||
| 1171 | * |
||
| 1172 | * @return void |
||
| 1173 | */ |
||
| 1174 | protected function _setCPid($value) |
||
| 1177 | } |
||
| 1178 | |||
| 1179 | /** |
||
| 1180 | * This is a singleton class, thus the constructor should be private/protected |
||
| 1181 | * (Get an instance of this class by calling \Kitodo\Dlf\Common\Doc::getInstance()) |
||
| 1182 | * |
||
| 1183 | * @access protected |
||
| 1184 | * |
||
| 1185 | * @param int $location: The location URL of the XML file to parse |
||
| 1186 | * @param int $pid: If > 0, then only document with this PID gets loaded |
||
| 1187 | * @param \SimpleXMLElement|IiifResourceInterface $preloadedDocument: Either null or the \SimpleXMLElement |
||
| 1188 | * or IiifResourceInterface that has been loaded to determine the basic document format. |
||
| 1189 | * |
||
| 1190 | * @return void |
||
| 1191 | */ |
||
| 1192 | protected function __construct($location, $pid, $preloadedDocument) |
||
| 1291 | } |
||
| 1292 | } |
||
| 1293 | |||
| 1294 | /** |
||
| 1295 | * This magic method is called each time an invisible property is referenced from the object |
||
| 1296 | * |
||
| 1297 | * @access public |
||
| 1298 | * |
||
| 1299 | * @param string $var: Name of variable to get |
||
| 1300 | * |
||
| 1301 | * @return mixed Value of $this->$var |
||
| 1302 | */ |
||
| 1303 | public function __get($var) |
||
| 1304 | { |
||
| 1305 | $method = '_get' . ucfirst($var); |
||
| 1306 | if ( |
||
| 1307 | !property_exists($this, $var) |
||
| 1308 | || !method_exists($this, $method) |
||
| 1309 | ) { |
||
| 1310 | $this->logger->warning('There is no getter function for property "' . $var . '"'); |
||
| 1311 | return; |
||
| 1312 | } else { |
||
| 1313 | return $this->$method(); |
||
| 1314 | } |
||
| 1315 | } |
||
| 1316 | |||
| 1317 | /** |
||
| 1318 | * This magic method is called each time an invisible property is checked for isset() or empty() |
||
| 1319 | * |
||
| 1320 | * @access public |
||
| 1321 | * |
||
| 1322 | * @param string $var: Name of variable to check |
||
| 1323 | * |
||
| 1324 | * @return bool true if variable is set and not empty, false otherwise |
||
| 1325 | */ |
||
| 1326 | public function __isset($var) |
||
| 1327 | { |
||
| 1328 | return !empty($this->__get($var)); |
||
| 1329 | } |
||
| 1330 | |||
| 1331 | /** |
||
| 1332 | * This magic method is called each time an invisible property is referenced from the object |
||
| 1333 | * |
||
| 1334 | * @access public |
||
| 1335 | * |
||
| 1336 | * @param string $var: Name of variable to set |
||
| 1337 | * @param mixed $value: New value of variable |
||
| 1338 | * |
||
| 1339 | * @return void |
||
| 1340 | */ |
||
| 1341 | public function __set($var, $value) |
||
| 1351 | } |
||
| 1352 | } |
||
| 1353 | } |
||
| 1354 |
Loading history...
