importer-parsers-archive-2022/parsers/FDH-Konvolute.php

<?PHP
/**
 * Parser for XML generated through CSVXML.
 *
 * @author Joshua Ramon Enslin <joshua@museum-digital.de>
 * @link   https://imports.museum-digital.org/importer/parsers/csvxml.php
 */
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)) {
        throw new Exception("You need to select a data folder to run image-only imports");
    }
    else $importImages = true;

    // Set up writers

    $outputHandler = new MDOutputHandler;
    $outputHandler->setVerbosity(2);

    $objectWriter = new MDObjectWriter($version['mainDB'], $version['nodaDB'], $version['link'], $version['filepath'], $version['dataFolderLink']);

    $objectWriter->disableUpdateBaseData = true;
    $objectWriter->disableImportAdditionalData = true;
    # $objectWriter->disableImportImagesResources = true;
    $objectWriter->disableImportTags = true;
    $objectWriter->disableImportLiterature = true;
    $objectWriter->disableImportHyperlinks = true;
    $objectWriter->disableImportSeries = true;
    $objectWriter->disableImportCollections = true;
    $objectWriter->disableImportObjectRecords = true;
    $objectWriter->disableImportTranscriptions = true;
    $objectWriter->disableImportMarkings = true;
    $objectWriter->disableImportExhibitions = true;
    $objectWriter->disableImportReception = true;

    $i = 0;
    $startAtCounter = 0;

    foreach (MD_STD::scandir(MD_IMPORTER_CONF::$import_dir_xml . "{$XMLFolder}") as $xmlFile) {

        if (pathinfo($xmlFile, PATHINFO_EXTENSION) !== "txt") continue;

        ++$i;
        if ($i < $startAtCounter) {
            continue;
        }
        $outputHandler->toLog("Starting to process file #{$i}", 2);

        // File name is inventory number
        $inventory_number = pathinfo($xmlFile, PATHINFO_FILENAME);

        $all_images_raw = MD_STD::file_get_contents(MD_IMPORTER_CONF::$import_dir_xml . "{$XMLFolder}" . '/' . $xmlFile);

        // Clean away spaces after newlines
        while (strpos($all_images_raw, PHP_EOL . " ") !== false) {
            $all_images_raw = str_replace(PHP_EOL . " ", PHP_EOL, $all_images_raw);
        }
        // Clean away tabs after newlines
        while (strpos($all_images_raw, PHP_EOL . "\t") !== false) {
            $all_images_raw = str_replace(PHP_EOL . "\t", PHP_EOL, $all_images_raw);
        }
        while (strpos($all_images_raw, "\u{feff}") !== false) {
            $all_images_raw = str_replace("\u{feff}", "", $all_images_raw);
        }

        // All single images are separated by two newlines
        $all_images = explode(PHP_EOL . PHP_EOL . '[', $all_images_raw);

        $object = new MDObject($version['mainDB'], $version['nodaDB'], $version['language'], $institution_id, $inventory_number, "ignored_anyway", "ignored_anyway", "ignored_anyway", $outputHandler);
        $images = [];

        foreach ($all_images as $cur_img) {

            $img_info = array_diff(explode(PHP_EOL, $cur_img), ['', ' ']);

            $filename_base = trim($img_info[0], "[] \t[");
            $filename_base = strtr($filename_base, ["[" => ""]);
            if (file_exists(MD_IMPORTER_CONF::$import_dir_files . $dataFolder . "/" . $filename_base . '.jpg')) {
                $filename = $filename_base . '.jpg';
            }
            else if (file_exists(MD_IMPORTER_CONF::$import_dir_files . $dataFolder . "/" . $filename_base . '.JPG')) {
                $filename = $filename_base . '.JPG';
            }
            else {
                echo MD_IMPORTER_CONF::$import_dir_files . $dataFolder . "/" . $filename_base . '.jpg';
                throw new MDFileDoesNotExist("There is no corresponding file for " . $filename_base);
            }

            // Image title is the first line after the file name. Everything after is the description.
            $image_title = $img_info[1];
            unset($img_info[0], $img_info[1]);
            $img_description = implode(PHP_EOL, $img_info);

            $image = new MDImage($version['mainDB'], $image_title, MD_IMPORTER_CONF::$import_dir_files . $dataFolder . "/" . $filename);
            $image->set_image_beschreibung($img_description);
            /*
            if (isset($objectData["image_rights" . $suffix])) $image->set_image_rights($objectData["image_rights" . $suffix]);
            if (isset($objectData["image_owner" . $suffix])) $image->set_image_owner($objectData["image_owner" . $suffix]);
            if (isset($objectData["image_creator" . $suffix])) $image->set_image_creator($objectData["image_creator" . $suffix]);
             */
            $image->set_visible(true);
            # if (isset($objectData["image_master_filename" . $suffix])) $image->set_image_master_filename($objectData["image_master_filename" . $suffix]);
            $images[] = $image;

            usleep(50);

        }

        $images[0]->set_main_image(true);
        foreach ($images as $image) $object->appendImage($image);

        $newObjectID = $objectWriter->writeObject($object, true, $insertOnly, $outputHandler);
        usleep(IMPORTER_DELAY_PER_OBJECT);

    }

}