[BUGFIX] Allow OAI-PMH set names which are required for DINI

Signed-off-by: Stefan Weil <[email protected]>
kitodo · Feb 8, 2025 · cbac1b9 · cbac1b9
1 parent 9cc612a
commit cbac1b9
Show file tree

Hide file tree

Showing 2 changed files with 22 additions and 2 deletions.
diff --git a/Classes/Command/BaseCommand.php b/Classes/Command/BaseCommand.php
@@ -339,7 +339,26 @@ private function addCollections(Document &$document, array $collections): void
                 $documentCollection = GeneralUtility::makeInstance(Collection::class);
                 $documentCollection->setIndexName($collection);
                 $documentCollection->setLabel($collection);
-                $documentCollection->setOaiName((!empty($this->extConf['general']['publishNewCollections']) ? Helper::getCleanString($collection) : ''));
+                $setSpec = '';
+                if (!empty($this->extConf['general']['publishNewCollections'])) {
+                    // setSpec only allows unreserved characters (rfc2396).
+                    // alnum | "-" | "_" | "." | "!" | "~" | "*" | "'" | "(" | ")"
+                    // Here we are a little bit more restrictive because
+                    // some more unreserved characters are not allowed.
+                    // Convert whitespaces to dash.
+                    $setSpec = $collection;
+                    $setSpec = preg_replace('/[\s]/', '-', $setSpec);
+                    // Remove multiple dashes.
+                    $setSpec = preg_replace('/[-]{2,}/', '-', $setSpec);
+                    // Remove undesired characters.
+                    $setSpec = preg_replace('/[^\w:-]/', '', $setSpec);
+                    // A hierarchical setSpec consists of two or more
+                    // normal setSpec which are separated by colons.
+                    // Remove colons which don't separate hierarchical setSpec entries.
+                    $setSpec = trim($setSpec, ':');
+                    $setSpec = preg_replace('/:{2,}/', ':', $setSpec);
+                }
+                $documentCollection->setOaiName($setSpec);
                 $documentCollection->setIndexSearch('');
                 $documentCollection->setDescription('');
                 // add to CollectionRepository

diff --git a/Configuration/TCA/tx_dlf_collections.php b/Configuration/TCA/tx_dlf_collections.php
@@ -137,7 +137,8 @@
                 'type' => 'input',
                 'size' => 30,
                 'max' => 255,
-                'eval' => 'nospace,alphanum_x,uniqueInPid',
+                // TODO: Add own form evaluation (see OAI-PMH setSpec).
+                'eval' => 'nospace,uniqueInPid',
                 'default' => '',
             ],
         ],