Initial

2023-01-23 15:21:35 +01:00
commit 43d8b878d1
15 changed files with 4224 additions and 0 deletions
--- a/parsers/gos.php
+++ b/parsers/gos.php
@ -0,0 +1,94 @@
+<?PHP
+/**
+ * Parser for GOS XML as provided by the MBK (Berlin).
+ *
+ * @author Joshua Ramon Enslin <joshua@museum-digital.de>
+ */
+declare(strict_types = 1);
+
+/**
+ * Parse function.
+ *
+ * @param array<mixed>     $version        Instance to import into.
+ * @param integer          $institution_id Institution to import to.
+ * @param non-empty-string $XMLFolder      Folder of the XML files to import.
+ * @param string           $dataFolder     Data folder.
+ * @param integer          $sammlung_id    Collection ID. Optional.
+ * @param boolean          $visibility     Import objects to be directly visible?.
+ * @param boolean          $insertOnly     If set to true, only new objects are added,
+ *                                         old are not updated.
+ *
+ * @return void
+ */
+function parseImportXML(array $version, int $institution_id, string $XMLFolder, string $dataFolder = "", int $sammlung_id = 0, bool $visibility = false, bool $insertOnly = false) {
+
+    if (!is_dir(MD_IMPORTER_CONF::$import_dir_xml . "{$XMLFolder}")) throw new MDFileDoesNotExist("The folder to import from ($XMLFolder) does not exist.");
+
+    if (empty($dataFolder)) {
+        $importImages = false;
+    }
+    else $importImages = true;
+
+    $collectionWriter = new MDCollectionWriter($version['mainDB']);
+    $seriesWriter = new MDSeriesWriter($version['mainDB']);
+    $literatureWriter = new MDLiteratureWriter($version['mainDB']);
+    $tagWriter = new MDTagWriter($version['nodaDB']);
+
+    // Set up writers
+    $outputHandler = new MDOutputHandler;
+    $outputHandler->setVerbosity(2);
+
+    $objectWriter = new MDObjectWriter($version['mainDB'], $version['nodaDB'], $version['link'], $version['filepath'], $version['dataFolderLink']);
+
+    $startFile = 0;
+    $counter = 0;
+    foreach (MD_STD::scandir(MD_IMPORTER_CONF::$import_dir_xml . "{$XMLFolder}") as $xmlFile) {
+
+        ++$counter;
+        if (pathinfo($xmlFile, PATHINFO_EXTENSION) !== "xml") {
+            continue;
+        }
+        if ($counter < $startFile) {
+            continue;
+        }
+
+        // LIDO contains : in tag names. SimpleXML can't handle those.
+
+        $rawData = MD_STD::file_get_contents(MD_IMPORTER_CONF::$import_dir_xml . "{$XMLFolder}/{$xmlFile}");
+        $rawData = strtr(str_replace(":", "___", str_replace("lido:", "", $rawData)), [
+            "http___//" => "http://", "xmlns___" => "xmlns:", "xsi___" => "xsi:"
+        ]);
+
+        if (($xmlData = simplexml_load_string($rawData)) === false) {
+            throw new Exception("XML couldn't be loaded");
+        }
+
+        //echo "HI";
+        $lidoObject = new GosParser($version['mainDB'], $version['nodaDB'],
+            $version['filepath'],
+            $version['language'], $institution_id,
+            $xmlData, $dataFolder, $outputHandler,
+            $collectionWriter,
+            $seriesWriter,
+            $literatureWriter,
+            $tagWriter,
+            $sammlung_id, $visibility,
+        );
+        $insertOnly = false;
+
+        $newObjectID = $objectWriter->writeObject($lidoObject->get_mdobject(), true, $insertOnly, $outputHandler);
+
+        // Sleep for a millisecond
+        $outputHandler->toLog("Done with object $counter", 2);
+        usleep(IMPORTER_DELAY_PER_OBJECT);
+
+        /*
+        if ($counter === 100) {
+            break;
+        }
+        */
+
+    }
+
+}
+