Merge branch 'master' into fix-page

kitodo · Jan 29, 2024 · 07f424d · 07f424d
2 parents 856c70d + 93a4b45
commit 07f424d
Show file tree

Hide file tree

Showing 22 changed files with 853 additions and 534 deletions.
diff --git a/.github/phpstan.neon b/.github/phpstan.neon
@@ -5,6 +5,7 @@ parameters:
  - '#Call to an undefined method Kitodo\\Dlf\\Domain\\Repository\\[a-zA-Z]+Repository::findByIsSortable\(\)\.#'
  - '#Call to an undefined method Kitodo\\Dlf\\Domain\\Repository\\[a-zA-Z]+Repository::findOneByFeUserId\(\)\.#'
  - '#Call to an undefined method Kitodo\\Dlf\\Domain\\Repository\\[a-zA-Z]+Repository::findOneByIndexName\(\)\.#'
+ - '#Call to an undefined method Kitodo\\Dlf\\Domain\\Repository\\[a-zA-Z]+Repository::findOneByLocation\(\)\.#'
  - '#Call to an undefined method Kitodo\\Dlf\\Domain\\Repository\\[a-zA-Z]+Repository::findOneByPid\(\)\.#'
  - '#Call to an undefined method Kitodo\\Dlf\\Domain\\Repository\\[a-zA-Z]+Repository::findOneByRecordId\(\)\.#'
  - '#Call to an undefined method Kitodo\\Dlf\\Domain\\Repository\\[a-zA-Z]+Repository::findOneByRoot\(\)\.#'

diff --git a/Classes/Command/HarvestCommand.php b/Classes/Command/HarvestCommand.php
@@ -112,7 +112,7 @@ protected function execute(InputInterface $input, OutputInterface $output): int
 
  if ($this->storagePid == 0) {
  $io->error('ERROR: No valid PID (' . $this->storagePid . ') given.');
- exit(1);
+ return BaseCommand::FAILURE;
  }
 
  if (
@@ -133,15 +133,15 @@ protected function execute(InputInterface $input, OutputInterface $output): int
  }
  if (empty($output_solrCores)) {
  $io->error('ERROR: No valid Solr core ("' . $input->getOption('solr') . '") given. No valid cores found on PID ' . $this->storagePid . ".\n");
- exit(1);
+ return BaseCommand::FAILURE;
  } else {
  $io->error('ERROR: No valid Solr core ("' . $input->getOption('solr') . '") given. ' . "Valid cores are (<uid>:<index_name>):\n" . implode("\n", $output_solrCores) . "\n");
- exit(1);
+ return BaseCommand::FAILURE;
  }
  }
  } else {
  $io->error('ERROR: Required parameter --solr|-s is missing or array.');
- exit(1);
+ return BaseCommand::FAILURE;
  }
 
  if (MathUtility::canBeInterpretedAsInteger($input->getOption('lib'))) {
@@ -152,11 +152,11 @@ protected function execute(InputInterface $input, OutputInterface $output): int
  $baseUrl = $this->owner->getOaiBase();
  } else {
  $io->error('ERROR: Required parameter --lib|-l is not a valid UID.');
- exit(1);
+ return BaseCommand::FAILURE;
  }
  if (!GeneralUtility::isValidUrl($baseUrl)) {
  $io->error('ERROR: No valid OAI Base URL set for library with given UID ("' . $input->getOption('lib') . '").');
- exit(1);
+ return BaseCommand::FAILURE;
  } else {
  try {
  $oai = Endpoint::build($baseUrl);
@@ -198,7 +198,7 @@ protected function execute(InputInterface $input, OutputInterface $output): int
  }
  if (empty($set)) {
  $io->error('ERROR: OAI interface does not provide a set with given setSpec ("' . $input->getOption('set') . '").');
- exit(1);
+ return BaseCommand::FAILURE;
  }
  }
 
@@ -261,7 +261,7 @@ protected function execute(InputInterface $input, OutputInterface $output): int
 
  $io->success('All done!');
 
- return 0;
+ return BaseCommand::SUCCESS;
  }
 
  /**

diff --git a/Classes/Command/IndexCommand.php b/Classes/Command/IndexCommand.php
@@ -99,7 +99,7 @@ protected function execute(InputInterface $input, OutputInterface $output): int
 
  if ($this->storagePid == 0) {
  $io->error('ERROR: No valid PID (' . $this->storagePid . ') given.');
- exit(1);
+ return BaseCommand::FAILURE;
  }
 
  if (
@@ -117,15 +117,15 @@ protected function execute(InputInterface $input, OutputInterface $output): int
  }
  if (empty($output_solrCores)) {
  $io->error('ERROR: No valid Solr core ("' . $input->getOption('solr') . '") given. No valid cores found on PID ' . $this->storagePid . ".\n");
- exit(1);
+ return BaseCommand::FAILURE;
  } else {
  $io->error('ERROR: No valid Solr core ("' . $input->getOption('solr') . '") given. ' . "Valid cores are (<uid>:<index_name>):\n" . implode("\n", $output_solrCores) . "\n");
- exit(1);
+ return BaseCommand::FAILURE;
  }
  }
  } else {
  $io->error('ERROR: Required parameter --solr|-s is missing or array.');
- exit(1);
+ return BaseCommand::FAILURE;
  }
 
  if (
@@ -137,7 +137,7 @@ protected function execute(InputInterface $input, OutputInterface $output): int
  )
  ) {
  $io->error('ERROR: Required parameter --doc|-d is not a valid document UID or URL.');
- exit(1);
+ return BaseCommand::FAILURE;
  }
 
  if (!empty($input->getOption('owner'))) {
@@ -168,24 +168,12 @@ protected function execute(InputInterface $input, OutputInterface $output): int
  } else if (GeneralUtility::isValidUrl($input->getOption('doc'))) {
  $doc = AbstractDocument::getInstance($input->getOption('doc'), ['storagePid' => $this->storagePid], true);
 
- if ($doc->recordId) {
- $document = $this->documentRepository->findOneByRecordId($doc->recordId);
- }
-
- if ($document === null) {
- // create new Document object
- $document = GeneralUtility::makeInstance(Document::class);
- }
-
- // now there must exist a document object
- if ($document) {
- $document->setLocation($input->getOption('doc'));
- }
+ $document = $this->getDocumentFromUrl($doc, $input->getOption('doc'));
  }
 
  if ($doc === null) {
  $io->error('ERROR: Document "' . $input->getOption('doc') . '" could not be loaded.');
- exit(1);
+ return BaseCommand::FAILURE;
  }
 
  $document->setSolrcore($solrCoreUid);
@@ -205,7 +193,39 @@ protected function execute(InputInterface $input, OutputInterface $output): int
 
  $io->success('All done!');
 
- return 0;
+ return BaseCommand::SUCCESS;
  }
 
+ /**
+ * Get document from given URL. Find it in database, if not found create the new one.
+ *
+ * @access private
+ *
+ * @param AbstractDocument $doc
+ * @param string $url
+ *
+ * @return Document
+ */
+ private function getDocumentFromUrl($doc, string $url): Document
+ {
+ $document = null;
+
+ if ($doc->recordId) {
+ $document = $this->documentRepository->findOneByRecordId($doc->recordId);
+ } else {
+ $document = $this->documentRepository->findOneByLocation($url);
+ }
+
+ if ($document === null) {
+ // create new Document object
+ $document = GeneralUtility::makeInstance(Document::class);
+ }
+
+ // now there must exist a document object
+ if ($document) {
+ $document->setLocation($url);
+ }
+
+ return $document;
+ }
 }
diff --git a/Classes/Command/ReindexCommand.php b/Classes/Command/ReindexCommand.php
@@ -103,7 +103,7 @@ protected function execute(InputInterface $input, OutputInterface $output): int
 
  if ($this->storagePid == 0) {
  $io->error('ERROR: No valid PID (' . $this->storagePid . ') given.');
- exit(1);
+ return BaseCommand::FAILURE;
  }
 
  if (
@@ -121,15 +121,15 @@ protected function execute(InputInterface $input, OutputInterface $output): int
  }
  if (empty($output_solrCores)) {
  $io->error('ERROR: No valid Solr core ("' . $input->getOption('solr') . '") given. No valid cores found on PID ' . $this->storagePid . ".\n");
- exit(1);
+ return BaseCommand::FAILURE;
  } else {
  $io->error('ERROR: No valid Solr core ("' . $input->getOption('solr') . '") given. ' . "Valid cores are (<uid>:<index_name>):\n" . implode("\n", $output_solrCores) . "\n");
- exit(1);
+ return BaseCommand::FAILURE;
  }
  }
  } else {
  $io->error('ERROR: Required parameter --solr|-s is missing or array.');
- exit(1);
+ return BaseCommand::FAILURE;
  }
 
  if (!empty($input->getOption('owner'))) {
@@ -152,13 +152,13 @@ protected function execute(InputInterface $input, OutputInterface $output): int
  // "coll" may be a single integer or a comma-separated list of integers.
  if (empty(array_filter(GeneralUtility::intExplode(',', $input->getOption('coll'), true)))) {
  $io->error('ERROR: Parameter --coll|-c is not a valid comma-separated list of collection UIDs.');
- exit(1);
+ return BaseCommand::FAILURE;
  }
  // Get all documents of given collections.
  $documents = $this->documentRepository->findAllByCollectionsLimited(GeneralUtility::intExplode(',', $input->getOption('coll'), true), 0);
  } else {
  $io->error('ERROR: One of parameters --all|-a or --coll|-c must be given.');
- exit(1);
+ return BaseCommand::FAILURE;
  }
 
  foreach ($documents as $id => $document) {
@@ -187,6 +187,6 @@ protected function execute(InputInterface $input, OutputInterface $output): int
 
  $io->success('All done!');
 
- return 0;
+ return BaseCommand::SUCCESS;
  }
 }
diff --git a/Classes/Common/AbstractDocument.php b/Classes/Common/AbstractDocument.php
@@ -309,7 +309,7 @@ abstract public function getDownloadLocation(string $id): string;
  * @abstract
  *
  * @param string $id The "@ID" attribute of the file node (METS) or the "@id" property of the IIIF resource
- * 
+ *
  * @return array|null The set of file information
  */
  abstract public function getFileInfo($id): ?array;
@@ -581,7 +581,7 @@ public static function &getInstance(string $location, array $settings = [], bool
  }
 
  // Sanitize input.
- $pid = max(intval($settings['storagePid']), 0);
+ $pid = max((int) $settings['storagePid'], 0);
  if ($documentFormat == 'METS') {
  $instance = new MetsDocument($pid, $location, $xml, $settings);
  } elseif ($documentFormat == 'IIIF') {
@@ -590,7 +590,7 @@ public static function &getInstance(string $location, array $settings = [], bool
  $instance = new IiifManifest($pid, $location, $iiif);
  }
 
- if (!is_null($instance)) {
+ if ($instance !== null) {
  self::setDocumentCache($location, $instance);
  }
 
@@ -686,19 +686,7 @@ protected function getFullTextFromXml(string $id): string
  if (!empty($fileContent) && !empty($this->formats[$textFormat])) {
  $textMiniOcr = '';
  if (!empty($this->formats[$textFormat]['class'])) {
- $class = $this->formats[$textFormat]['class'];
- // Get the raw text from class.
- if (
- class_exists($class)
- && ($obj = GeneralUtility::makeInstance($class)) instanceof FulltextInterface
- ) {
- // Load XML from file.
- $ocrTextXml = Helper::getXmlFileAsString($fileContent);
- $textMiniOcr = $obj->getTextAsMiniOcr($ocrTextXml);
- $this->rawTextArray[$id] = $textMiniOcr;
- } else {
- $this->logger->warning('Invalid class/method "' . $class . '->getRawText()" for text format "' . $textFormat . '"');
- }
+ $textMiniOcr = $this->getRawTextFromClass($id, $fileContent, $textFormat);
  }
  $fullText = $textMiniOcr;
  } else {
@@ -707,6 +695,38 @@ class_exists($class)
  return $fullText;
  }
 
+ /**
+ * Get raw text from class for given format.
+ *
+ * @access private
+ *
+ * @param $id
+ * @param $fileContent
+ * @param $textFormat
+ *
+ * @return string
+ */
+ private function getRawTextFromClass($id, $fileContent, $textFormat): string
+ {
+ $textMiniOcr = '';
+ $class = $this->formats[$textFormat]['class'];
+ // Get the raw text from class.
+ if (class_exists($class)) {
+ $obj = GeneralUtility::makeInstance($class);
+ if ($obj instanceof FulltextInterface) {
+ // Load XML from file.
+ $ocrTextXml = Helper::getXmlFileAsString($fileContent);
+ $textMiniOcr = $obj->getTextAsMiniOcr($ocrTextXml);
+ $this->rawTextArray[$id] = $textMiniOcr;
+ } else {
+ $this->logger->warning('Invalid class/method "' . $class . '->getRawText()" for text format "' . $textFormat . '"');
+ }
+ } else {
+ $this->logger->warning('Class "' . $class . ' does not exists for "' . $textFormat . ' text format"');
+ }
+ return $textMiniOcr;
+ }
+
  /**
  * Get format of the OCR full text
  *
@@ -744,7 +764,7 @@ public static function getTitle(int $uid, bool $recursive = false): string
  {
  $title = '';
  // Sanitize input.
- $uid = max(intval($uid), 0);
+ $uid = max($uid, 0);
  if ($uid) {
  $queryBuilder = GeneralUtility::makeInstance(ConnectionPool::class)
  ->getQueryBuilderForTable('tx_dlf_documents');
@@ -762,15 +782,16 @@ public static function getTitle(int $uid, bool $recursive = false): string
  ->setMaxResults(1)
  ->execute();
 
- if ($resArray = $result->fetchAssociative()) {
+ $resArray = $result->fetchAssociative();
+ if ($resArray) {
  // Get title information.
  $title = $resArray['title'];
  $partof = $resArray['partof'];
  // Search parent documents recursively for a title?
  if (
  $recursive
  && empty($title)
- && intval($partof)
+ && (int) $partof
  && $partof != $uid
  ) {
  $title = self::getTitle($partof, true);
@@ -950,7 +971,8 @@ public function registerNamespaces(&$obj): void
  *
  * @return array
  */
- protected function initializeMetadata(string $format): array {
+ protected function initializeMetadata(string $format): array
+ {
  return [
  'title' => [],
  'title_sorting' => [],
@@ -1161,7 +1183,7 @@ protected function magicGetTableOfContents(): array
  */
  protected function _setCPid(int $value): void
  {
- $this->cPid = max(intval($value), 0);
+ $this->cPid = max($value, 0);
  }
 
  /**
@@ -1258,7 +1280,7 @@ public function __set(string $var, $value): void
  */
  private static function getDocumentCache(string $location)
  {
- $cacheIdentifier = md5($location);
+ $cacheIdentifier = hash('md5', $location);
  $cache = GeneralUtility::makeInstance(CacheManager::class)->getCache('tx_dlf_doc');
  $cacheHit = $cache->get($cacheIdentifier);
 
@@ -1279,11 +1301,10 @@ private static function getDocumentCache(string $location)
  */
  private static function setDocumentCache(string $location, AbstractDocument $currentDocument): void
  {
- $cacheIdentifier = md5($location);
+ $cacheIdentifier = hash('md5', $location);
  $cache = GeneralUtility::makeInstance(CacheManager::class)->getCache('tx_dlf_doc');
 
  // Save value in cache
  $cache->set($cacheIdentifier, $currentDocument);
  }
-
 }