<?php

declare(strict_types=1);

namespace Evomedien\Vitec\Command;

use Symfony\Component\Console\Attribute\AsCommand;
use Symfony\Component\Console\Command\Command;
use Symfony\Component\Console\Input\InputInterface;
use Symfony\Component\Console\Input\InputOption;
use Symfony\Component\Console\Output\OutputInterface;
use TYPO3\CMS\Core\Core\Bootstrap;
use TYPO3\CMS\Core\Database\ConnectionPool;
use TYPO3\CMS\Core\DataHandling\DataHandler;
use TYPO3\CMS\Core\Resource\Enum\DuplicationBehavior;
use TYPO3\CMS\Core\Resource\StorageRepository;
use TYPO3\CMS\Core\Utility\GeneralUtility;

/**
 * Imports the 22 Datapath success stories from the 2026-10-28 Dropbox
 * delivery (staged by the session into migrations/datapath_stories_in/
 * <slug>/{story.docx, story.pdf, images/*}) into the usecase model.
 *
 * The docx files share one structure (verified on Edinburgh / Woori):
 *   1st paragraph        -> title
 *   2nd paragraph        -> subtitle (one-liner)
 *   1st IntenseQuote     -> record quote, the two short lines after it
 *                           quote_person / quote_role
 *   known section names  -> text elements (Background, The Challenge,
 *                           The Solution, The Result, ... - plain
 *                           paragraphs, no heading styles in the files)
 *   ListParagraph runs   -> <ul> inside the section body
 *   further IntenseQuote -> vitec_quotation elements; attribution from
 *                           the preceding "... said:" line or the short
 *                           "Name, Role" line after the quote
 *
 * Per story: usecase record (pid/detail_page from the existing records),
 * hero_bgimage + card_image from the first staged image, remaining images
 * (max 4) as image elements, the staged Collateral pdf as a NEW download
 * record plus a vitec_downloadcard (layout story_download, header
 * Downloads) - the same shape the migrated vitec.com stories have after
 * the 2026-10-28 plugin conversion. Files land in
 * fileadmin/success_stories_import/datapath/<slug>/.
 *
 * Idempotent: a story whose normalised title already exists gets an
 * asset-only backfill (logo/hero/card/images where missing) instead.
 * Markets stay unassigned on purpose (editorial; reported).
 *
 *   vendor/bin/typo3 vitec:import-datapath-stories [--only=woori] [--apply]
 */
#[AsCommand(
    name: 'vitec:import-datapath-stories',
    description: 'Import the Datapath success-story delivery (docx + images + pdf) into the usecase model (dry-run by default)'
)]
final class ImportDatapathStoriesCommand extends Command
{
    private const TABLE = 'tx_vitec_domain_model_usecase';
    private const DOWNLOAD_TABLE = 'tx_vitec_domain_model_download';
    private const COLPOS = 999;
    private const MAX_IMAGE_ELEMENTS = 4;
    private const FILES_TARGET = 'success_stories_import/datapath';

    private const SECTION_NAMES = [
        'background', 'the challenge', 'challenge', 'the solution', 'solution',
        'the result', 'result', 'the results', 'results', 'future plans',
        'future expansion', 'the future', 'key features', 'benefits',
        'the benefits', 'conclusion', 'summary',
    ];

    private int $newCounter = 0;
    private bool $repairQuotes = false;

    protected function configure(): void
    {
        $this->addOption('dir', null, InputOption::VALUE_REQUIRED, 'Inbox directory', 'migrations/datapath_stories_in');
        $this->addOption('only', null, InputOption::VALUE_REQUIRED, 'Folder-name substring filter', '');
        $this->addOption('apply', null, InputOption::VALUE_NONE, 'Write changes (default: dry run)');
        $this->addOption('repair-quotes', null, InputOption::VALUE_NONE, 'Repair pass 0017-FE: move lead attribution into quote_person/quote_role, delete the bogus quotation elements, strip "Updated on" remnants');
    }

    protected function execute(InputInterface $input, OutputInterface $output): int
    {
        Bootstrap::initializeBackendAuthentication();
        $apply = (bool)$input->getOption('apply');
        $this->repairQuotes = (bool)$input->getOption('repair-quotes');
        $only = mb_strtolower((string)$input->getOption('only'));
        $dir = rtrim((string)$input->getOption('dir'), '/');
        $pool = GeneralUtility::makeInstance(ConnectionPool::class);

        $existing = [];
        $pidCounts = [];
        $detailCounts = [];
        foreach ($pool->getConnectionForTable(self::TABLE)->fetchAllAssociative(
            'SELECT uid, pid, title, detail_page FROM ' . self::TABLE . ' WHERE deleted = 0'
        ) as $row) {
            $existing[$this->norm((string)$row['title'])] = (int)$row['uid'];
            $pidCounts[(int)$row['pid']] = ($pidCounts[(int)$row['pid']] ?? 0) + 1;
            if ((int)$row['detail_page'] > 0) {
                $detailCounts[(int)$row['detail_page']] = ($detailCounts[(int)$row['detail_page']] ?? 0) + 1;
            }
        }
        arsort($pidCounts);
        arsort($detailCounts);
        $storagePid = (int)(array_key_first($pidCounts) ?? 0);
        $detailPage = (int)(array_key_first($detailCounts) ?? 0);

        $downloadPids = array_count_values(array_map('intval', $pool->getConnectionForTable(self::DOWNLOAD_TABLE)
            ->fetchFirstColumn('SELECT pid FROM ' . self::DOWNLOAD_TABLE . ' WHERE deleted = 0')));
        arsort($downloadPids);
        $downloadPid = (int)(array_key_first($downloadPids) ?? $storagePid);

        $folders = array_filter(glob($dir . '/*', GLOB_ONLYDIR) ?: [], static fn(string $f): bool =>
            is_file($f . '/story.docx') && ($only === '' || str_contains(mb_strtolower(basename($f)), $only)));
        $output->writeln(sprintf('%s · %d story folder(s) · storage pid %d · detail page %d · download pid %d',
            $apply ? 'APPLY' : 'DRY-RUN', count($folders), $storagePid, $detailPage, $downloadPid));

        $created = 0;
        foreach ($folders as $folder) {
            try {
                $created += $this->importOne($folder, $existing, $storagePid, $detailPage, $downloadPid, $apply, $output) ? 1 : 0;
            } catch (\Throwable $e) {
                $output->writeln('<error>  ' . basename($folder) . ': ' . get_class($e) . ' - ' . $e->getMessage() . '</error>');
            }
        }
        // --- repair pass: story downloadcards whose FlexForm still carries a
        // literal NEW id (first import wrote the placeholder unsubstituted) -
        // resolve via "<story title> Success Story" and write the real uid
        $content = $pool->getConnectionForTable('tt_content');
        $downloadByNorm = [];
        foreach ($pool->getConnectionForTable(self::DOWNLOAD_TABLE)->fetchAllAssociative(
            'SELECT uid, title FROM ' . self::DOWNLOAD_TABLE . ' WHERE deleted = 0'
        ) as $download) {
            $downloadByNorm[$this->norm((string)$download['title'])] = (int)$download['uid'];
        }
        $repairMap = [];
        foreach ($content->fetchAllAssociative(
            "SELECT c.uid, u.title FROM tt_content c
             JOIN " . self::TABLE . " u ON u.uid = c.tx_vitec_usecase_content AND u.deleted = 0
             WHERE c.deleted = 0 AND c.CType = 'vitec_downloadcard' AND c.pi_flexform LIKE '%NEWDS%'"
        ) as $card) {
            $target = $downloadByNorm[$this->norm((string)$card['title'] . ' success story')] ?? 0;
            if ($target > 0) {
                $repairMap['tt_content'][(string)(int)$card['uid']]['pi_flexform'] =
                    ['data' => ['sDEF' => ['lDEF' => ['settings.download' => ['vDEF' => (string)$target]]]]];
                $output->writeln('  ~ REPAIR card #' . $card['uid'] . ' ("' . $card['title'] . '") -> download #' . $target);
            } else {
                $output->writeln('  <comment>! REPAIR card #' . $card['uid'] . ' ("' . $card['title'] . '"): kein Download gefunden</comment>');
            }
        }
        if ($apply && $repairMap !== []) {
            $repairHandler = GeneralUtility::makeInstance(DataHandler::class);
            $repairHandler->start($repairMap, []);
            $repairHandler->process_datamap();
        }

        $output->writeln($apply
            ? "<info>$created story record(s) created. Flush the frontend cache; assign markets editorially; re-run the sub-solution import for the testimonials.</info>"
            : "\nDRY RUN - nothing written. Add --apply to write.");
        return Command::SUCCESS;
    }

    /** @param array<string,int> $existing */
    private function importOne(string $folder, array $existing, int $storagePid, int $detailPage, int $downloadPid, bool $apply, OutputInterface $output): bool
    {
        $slug = basename($folder);
        $doc = $this->parseDocx($folder . '/story.docx');
        $title = $doc['title'];
        if ($title === '') {
            $output->writeln('<comment>  ' . $slug . ': kein Titel im docx - uebersprungen</comment>');
            return false;
        }
        if (isset($existing[$this->norm($title)])) {
            if ($this->repairQuotes) {
                $this->repairStoryQuotes($existing[$this->norm($title)], $slug, $doc, $apply, $output);
            } else {
                $this->backfillAssets($existing[$this->norm($title)], $slug, $folder, $storagePid, $apply, $output);
            }
            return false;
        }

        $images = array_values(array_filter(glob($folder . '/images/*') ?: [], 'is_file'));
        $pdf = is_file($folder . '/story.pdf') ? $folder . '/story.pdf' : null;
        $output->writeln(sprintf('+ "%s"  (%d Sektion(en), %d Zitat(e)%s, %d Bild(er), %s)',
            $title, count($doc['sections']), count($doc['quotes']),
            $doc['lead']['quote'] !== '' ? ' + Lead-Quote' : '',
            count($images), $pdf ? 'PDF' : 'KEIN PDF'));
        foreach ($doc['report'] as $line) {
            $output->writeln('      ' . $line);
        }
        if (!$apply) {
            return false;
        }

        $storage = GeneralUtility::makeInstance(StorageRepository::class)->getDefaultStorage();
        $targetFolder = $this->ensureFolder($storage, self::FILES_TARGET . '/' . $slug);

        $datamap = [];
        $elementIds = [];

        // --- sections as text elements -------------------------------------------------
        foreach ($doc['sections'] as $section) {
            if ($section['type'] === 'text') {
                $id = $this->newId();
                $datamap['tt_content'][$id] = [
                    'pid' => $storagePid, 'CType' => 'text', 'colPos' => self::COLPOS,
                    'header' => $section['header'], 'header_layout' => 2,
                    'bodytext' => $section['body'],
                ];
                $elementIds[] = $id;
            } else { // quotation
                $id = $this->newId();
                $datamap['tt_content'][$id] = [
                    'pid' => $storagePid, 'CType' => 'vitec_quotation', 'colPos' => self::COLPOS,
                    'vitec_quote' => $section['quote'],
                    'vitec_quote_name' => $section['name'],
                    'vitec_quote_position' => $section['position'],
                ];
                $elementIds[] = $id;
            }
        }

        // --- image elements (beyond hero/card) ------------------------------------------
        $heroRef = null;
        $cardRef = null;
        foreach ($images as $i => $imagePath) {
            $file = $storage->addFile($imagePath, $targetFolder, basename($imagePath), DuplicationBehavior::REPLACE, false);
            if ($i === 0) {
                $heroRef = $this->fileRef($datamap, $file->getUid(), $storagePid);
                $cardRef = $this->fileRef($datamap, $file->getUid(), $storagePid);
                continue;
            }
            if ($i > self::MAX_IMAGE_ELEMENTS) {
                continue;
            }
            $id = $this->newId();
            $datamap['tt_content'][$id] = [
                'pid' => $storagePid, 'CType' => 'image', 'colPos' => self::COLPOS,
                'image' => $this->fileRef($datamap, $file->getUid(), $storagePid),
            ];
            $elementIds[] = $id;
        }

        // --- download record (OWN DataHandler pass: NEW ids are NOT
        // substituted inside FlexForm values, found 2026-10-28 - the card
        // needs the REAL uid) + downloadcard --------------------------------------------
        if ($pdf !== null) {
            $pdfName = 'VITEC_' . $slug . '_USE-CASE.pdf';
            $pdfFile = $storage->addFile($pdf, $targetFolder, $pdfName, DuplicationBehavior::REPLACE, false);
            $downloadId = $this->newId();
            $downloadMap = [];
            $downloadMap[self::DOWNLOAD_TABLE][$downloadId] = [
                'pid' => $downloadPid,
                'title' => $title . ' Success Story',
                'slug' => $this->slugify($title . ' success story'),
                'file' => $this->fileRef($downloadMap, $pdfFile->getUid(), $downloadPid),
            ];
            $downloadHandler = GeneralUtility::makeInstance(DataHandler::class);
            $downloadHandler->start($downloadMap, []);
            $downloadHandler->process_datamap();
            $downloadUid = (int)($downloadHandler->substNEWwithIDs[$downloadId] ?? 0);
            if ($downloadUid > 0) {
                $cardId = $this->newId();
                $datamap['tt_content'][$cardId] = [
                    'pid' => $storagePid, 'CType' => 'vitec_downloadcard', 'colPos' => self::COLPOS,
                    'header' => 'Downloads',
                    'pi_flexform' => ['data' => ['sDEF' => ['lDEF' => [
                        'settings.download' => ['vDEF' => (string)$downloadUid],
                        'settings.layout' => ['vDEF' => 'story_download'],
                    ]]]],
                ];
                $elementIds[] = $cardId;
            }
        }

        // --- the record -------------------------------------------------------------------
        $recordId = $this->newId();
        $record = [
            'pid' => $storagePid,
            'title' => $title,
            'slug' => $this->slugify($title),
            'subtitle' => $doc['subtitle'],
            'teaser' => $doc['teaser'],
            'quote' => $doc['lead']['quote'],
            'quote_person' => $doc['lead']['name'],
            'quote_role' => $doc['lead']['position'],
            'seo_title' => $title,
            'seo_description' => $doc['subtitle'],
            'content_elements' => implode(',', $elementIds),
        ];
        if ($detailPage > 0) {
            $record['detail_page'] = $detailPage;
        }
        if ($heroRef !== null) {
            $record['hero_bgimage'] = $heroRef;
            $record['card_image'] = $cardRef;
        }
        $datamap[self::TABLE][$recordId] = $record;

        $dataHandler = GeneralUtility::makeInstance(DataHandler::class);
        $dataHandler->start($datamap, []);
        $dataHandler->process_datamap();
        if ($dataHandler->errorLog !== []) {
            throw new \RuntimeException('DataHandler: ' . implode(' | ', array_slice($dataHandler->errorLog, 0, 3)));
        }
        $uid = (int)($dataHandler->substNEWwithIDs[$recordId] ?? 0);
        $output->writeln('      -> Record #' . $uid . ' angelegt');
        return true;
    }

    /**
     * Asset backfill for an already imported story: the 2026-10-28 delivery
     * keeps the web images AND the customer logo in the Links/ folders,
     * which the first staging pass had excluded. Content-idempotent -
     * attaches customer_logo / hero_bgimage / card_image only when the
     * field has no reference yet and adds image elements only when the
     * story has none; existing references are never touched.
     */
    private function backfillAssets(int $uid, string $slug, string $folder, int $storagePid, bool $apply, OutputInterface $output): void
    {
        $pool = GeneralUtility::makeInstance(ConnectionPool::class);
        $refs = $pool->getConnectionForTable('sys_file_reference')->fetchFirstColumn(
            "SELECT DISTINCT fieldname FROM sys_file_reference
             WHERE deleted = 0 AND tablenames = ? AND uid_foreign = ?
               AND fieldname IN ('customer_logo', 'hero_bgimage', 'card_image')",
            [self::TABLE, $uid]
        );
        $imageElements = (int)$pool->getConnectionForTable('tt_content')->fetchOne(
            "SELECT COUNT(*) FROM tt_content WHERE deleted = 0 AND CType = 'image' AND tx_vitec_usecase_content = ?",
            [$uid]
        );

        $images = array_values(array_filter(glob($folder . '/images/*') ?: [], 'is_file'));
        $logos = array_values(array_filter(glob($folder . '/logo.*') ?: [], 'is_file'));

        $needLogo = $logos !== [] && !in_array('customer_logo', $refs, true);
        $needHero = $images !== [] && !in_array('hero_bgimage', $refs, true);
        $needCard = $images !== [] && !in_array('card_image', $refs, true);
        $newImageCount = ($imageElements === 0 && count($images) > 1)
            ? min(self::MAX_IMAGE_ELEMENTS, count($images) - 1)
            : 0;
        if (!$needLogo && !$needHero && !$needCard && $newImageCount === 0) {
            $output->writeln('  = #' . $uid . ' ' . $slug . ': Assets vollstaendig - nichts zu tun');
            return;
        }
        $output->writeln(sprintf('  ~ #%d %s: Backfill %s%s%s%s', $uid, $slug,
            $needLogo ? '[logo] ' : '', $needHero ? '[hero] ' : '', $needCard ? '[card] ' : '',
            $newImageCount > 0 ? '[' . $newImageCount . ' Bild-Element(e)]' : ''));
        if (!$apply) {
            return;
        }

        $storage = GeneralUtility::makeInstance(StorageRepository::class)->getDefaultStorage();
        $targetFolder = $this->ensureFolder($storage, self::FILES_TARGET . '/' . $slug);
        $datamap = [];
        $recordFields = [];

        if ($needLogo) {
            $logoName = 'logo-' . $slug . '.' . mb_strtolower(pathinfo($logos[0], PATHINFO_EXTENSION));
            $logoFile = $storage->addFile($logos[0], $targetFolder, $logoName, DuplicationBehavior::REPLACE, false);
            $recordFields['customer_logo'] = $this->fileRefFull($datamap, $logoFile->getUid(), $storagePid, $uid, 'customer_logo');
        }
        if ($needHero || $needCard) {
            $first = $storage->addFile($images[0], $targetFolder, basename($images[0]), DuplicationBehavior::REPLACE, false);
            if ($needHero) {
                $recordFields['hero_bgimage'] = $this->fileRefFull($datamap, $first->getUid(), $storagePid, $uid, 'hero_bgimage');
            }
            if ($needCard) {
                $recordFields['card_image'] = $this->fileRefFull($datamap, $first->getUid(), $storagePid, $uid, 'card_image');
            }
        }
        if ($newImageCount > 0) {
            // append via the parent CSV (existing children in sorting order +
            // the new elements) so the inline count stays correct; new images
            // land BEFORE a trailing downloadcard
            $children = array_map('intval', $pool->getConnectionForTable('tt_content')->fetchFirstColumn(
                'SELECT uid FROM tt_content WHERE deleted = 0 AND tx_vitec_usecase_content = ? ORDER BY sorting',
                [$uid]
            ));
            $newIds = [];
            foreach (array_slice($images, 1, $newImageCount) as $imagePath) {
                $file = $storage->addFile($imagePath, $targetFolder, basename($imagePath), DuplicationBehavior::REPLACE, false);
                $id = $this->newId();
                $datamap['tt_content'][$id] = [
                    'pid' => $storagePid, 'CType' => 'image', 'colPos' => self::COLPOS,
                    'image' => $this->fileRef($datamap, $file->getUid(), $storagePid),
                ];
                $newIds[] = $id;
            }
            $last = $children !== [] ? (int)end($children) : 0;
            $lastType = $last > 0 ? (string)$pool->getConnectionForTable('tt_content')->fetchOne(
                'SELECT CType FROM tt_content WHERE uid = ?', [$last]) : '';
            if ($lastType === 'vitec_downloadcard') {
                array_pop($children);
                $csv = array_merge($children, $newIds, [$last]);
            } else {
                $csv = array_merge($children, $newIds);
            }
            $recordFields['content_elements'] = implode(',', array_map('strval', $csv));
        }

        $datamap[self::TABLE][(string)$uid] = $recordFields;
        $dataHandler = GeneralUtility::makeInstance(DataHandler::class);
        $dataHandler->start($datamap, []);
        $dataHandler->process_datamap();
        if ($dataHandler->errorLog !== []) {
            throw new \RuntimeException('DataHandler: ' . implode(' | ', array_slice($dataHandler->errorLog, 0, 3)));
        }
        $output->writeln('      -> Backfill geschrieben');
    }

    /** sys_file_reference NEW id for a field of an EXISTING parent record. */
    private function fileRefFull(array &$datamap, int $fileUid, int $pid, int $uidForeign, string $fieldname): string
    {
        $refId = $this->newId();
        $datamap['sys_file_reference'][$refId] = [
            'pid' => $pid,
            'table_local' => 'sys_file',
            'uid_local' => $fileUid,
            'tablenames' => self::TABLE,
            'uid_foreign' => $uidForeign,
            'fieldname' => $fieldname,
        ];
        return $refId;
    }
    /** Short, non-sentence line right after a quote = attribution (name/role). */
    private function isAttributionLine(array $paragraph, int $maxLen): bool
    {
        $text = (string)$paragraph['text'];
        if ($text === '' || mb_strlen($text) > $maxLen + 60) {
            return false;
        }
        $first = trim((string)($paragraph['lines'][0] ?? $text));
        if ($first === '' || mb_strlen($first) > $maxLen || str_ends_with($first, '.')) {
            return false;
        }
        if (count(preg_split('/\s+/u', $first) ?: []) > 8) {
            return false;
        }
        return !in_array(mb_strtolower(rtrim($text, ':')), self::SECTION_NAMES, true);
    }

    /**
     * Repair pass (0017-FE Punkt 3 + 5a) for an already imported story:
     * writes the docx lead attribution into quote_person/quote_role (only
     * when empty), deletes the quotation elements that carry just the name
     * or role (or the glued combination) as their quote, and strips the
     * "NN | YYYY Updated on NN | YYYY" footer from text elements.
     */
    private function repairStoryQuotes(int $uid, string $slug, array $doc, bool $apply, OutputInterface $output): void
    {
        $pool = GeneralUtility::makeInstance(ConnectionPool::class);
        $record = $pool->getConnectionForTable(self::TABLE)->fetchAssociative(
            'SELECT uid, quote_person, quote_role FROM ' . self::TABLE . ' WHERE uid = ?', [$uid]
        );
        if (!$record) {
            return;
        }
        $name = (string)$doc['lead']['name'];
        $position = (string)$doc['lead']['position'];

        // normalise for the record (element matching below uses the RAW
        // parsed strings): trailing commas, "Name, Role" in ONE line, and a
        // single line that is clearly a role ("FIA Head of ...") -> role only
        $person = trim($name, " \t,");
        $role = trim($position, " \t,");
        if ($role === '' && str_contains($person, ',')) {
            [$person, $role] = $this->splitAttribution($person);
        }
        if ($role === '' && $person !== ''
            && count(preg_split('/\s+/u', $person) ?: []) >= 3
            && preg_match('/\b(Head|Manager|Director|President|CEO|CTO|Chief|Officer|Engineer|Architect|Strategy)\b/i', $person)) {
            $role = $person;
            $person = '';
        }

        $datamap = [];
        $cmdmap = [];
        $actions = [];

        if ($person !== '' && (string)$record['quote_person'] === '') {
            $datamap[self::TABLE][(string)$uid]['quote_person'] = $person;
            $actions[] = 'person="' . $person . '"';
        }
        if ($role !== '' && (string)$record['quote_role'] === '') {
            $datamap[self::TABLE][(string)$uid]['quote_role'] = $role;
            $actions[] = 'role="' . $role . '"';
        }

        // bogus quotation elements: quote == name / role / glued combination
        $targets = array_filter([
            $this->norm($name),
            $this->norm($position),
            $this->norm($name . $position),
        ], static fn(string $v): bool => $v !== '');
        if ($targets !== []) {
            foreach ($pool->getConnectionForTable('tt_content')->fetchAllAssociative(
                "SELECT uid, vitec_quote FROM tt_content
                 WHERE deleted = 0 AND CType = 'vitec_quotation' AND tx_vitec_usecase_content = ?",
                [$uid]
            ) as $element) {
                if (in_array($this->norm((string)$element['vitec_quote']), $targets, true)) {
                    $cmdmap['tt_content'][(int)$element['uid']]['delete'] = 1;
                    $actions[] = 'delete quotation #' . $element['uid'] . ' ("' . mb_substr((string)$element['vitec_quote'], 0, 40) . '")';
                }
            }
        }

        // "Updated on" remnants inside text elements (incl. glued into <li>)
        foreach ($pool->getConnectionForTable('tt_content')->fetchAllAssociative(
            "SELECT uid, bodytext FROM tt_content
             WHERE deleted = 0 AND CType = 'text' AND tx_vitec_usecase_content = ?
               AND bodytext LIKE '%Updated on%'",
            [$uid]
        ) as $element) {
            $body = (string)$element['bodytext'];
            $clean = (string)preg_replace('/\s*\d{1,2}\s*\|\s*\d{4}\s*Updated on\s*\d{1,2}\s*\|\s*\d{4}\.?/iu', '', $body);
            $clean = (string)preg_replace('/<(p|li)>\s*<\/\1>/u', '', $clean);
            if ($clean !== $body) {
                $datamap['tt_content'][(string)(int)$element['uid']]['bodytext'] = $clean;
                $actions[] = 'strip "Updated on" in #' . $element['uid'];
            }
        }

        if ($actions === []) {
            $output->writeln('  = #' . $uid . ' ' . $slug . ': nichts zu reparieren');
            return;
        }
        $output->writeln('  ~ #' . $uid . ' ' . $slug . ':');
        foreach ($actions as $action) {
            $output->writeln('      - ' . $action);
        }
        if (!$apply) {
            return;
        }
        $dataHandler = GeneralUtility::makeInstance(DataHandler::class);
        $dataHandler->start($datamap, $cmdmap);
        if ($datamap !== []) {
            $dataHandler->process_datamap();
        }
        if ($cmdmap !== []) {
            $dataHandler->process_cmdmap();
        }
        if ($dataHandler->errorLog !== []) {
            throw new \RuntimeException('DataHandler: ' . implode(' | ', array_slice($dataHandler->errorLog, 0, 3)));
        }
        $output->writeln('      -> repariert');
    }
    /** sys_file_reference NEW id for use in a parent FAL field. */
    private function fileRef(array &$datamap, int $fileUid, int $pid): string
    {
        $refId = $this->newId();
        $datamap['sys_file_reference'][$refId] = [
            'pid' => $pid,
            'table_local' => 'sys_file',
            'uid_local' => $fileUid,
        ];
        return $refId;
    }

    private function ensureFolder(\TYPO3\CMS\Core\Resource\ResourceStorage $storage, string $path): \TYPO3\CMS\Core\Resource\Folder
    {
        $folder = $storage->getRootLevelFolder();
        foreach (explode('/', trim($path, '/')) as $segment) {
            $folder = $folder->hasFolder($segment) ? $folder->getSubfolder($segment) : $folder->createFolder($segment);
        }
        return $folder;
    }

    // ---- docx parsing ------------------------------------------------------------------

    /**
     * @return array{title:string,subtitle:string,teaser:string,
     *   lead:array{quote:string,name:string,position:string},
     *   sections:array<int,array<string,string>>,quotes:array<int,int>,report:string[]}
     */
    private function parseDocx(string $file): array
    {
        $zip = new \ZipArchive();
        if ($zip->open($file) !== true) {
            throw new \RuntimeException('docx laesst sich nicht oeffnen');
        }
        $xml = (string)$zip->getFromName('word/document.xml');
        $zip->close();
        $dom = new \DOMDocument();
        $dom->loadXML($xml);
        $xpath = new \DOMXPath($dom);
        $w = 'http://schemas.openxmlformats.org/wordprocessingml/2006/main';
        $xpath->registerNamespace('w', $w);

        $paragraphs = [];
        foreach ($xpath->query('//w:body/w:p') as $p) {
            $text = '';
            foreach ($xpath->query('.//w:t|.//w:br', $p) as $node) {
                // w:br as line break - attribution lines come as "Name<br>Role"
                // in some files and glued without it the names stick together
                // ("Shawn HanssonCEO, ...", ticket 0017-FE)
                $text .= $node->localName === 'br' ? "\n" : $node->textContent;
            }
            $text = trim((string)preg_replace('/[ \t]+/u', ' ', $text));
            $text = (string)preg_replace('/\s*\n\s*/u', "\n", $text);
            // import footer "04 | 2025 Updated on 10 | 2026" (0017-FE 5a)
            $text = trim((string)preg_replace('/\s*\d{1,2}\s*\|\s*\d{4}\s*Updated on\s*\d{1,2}\s*\|\s*\d{4}\.?\s*$/iu', '', $text));
            if ($text === '') {
                continue;
            }
            $styleNode = $xpath->query('./w:pPr/w:pStyle/@w:val', $p)->item(0);
            $paragraphs[] = [
                'text' => str_replace("\n", ' ', $text),
                'lines' => explode("\n", $text),
                'style' => $styleNode ? (string)$styleNode->nodeValue : '',
            ];
        }

        $report = [];
        $title = $paragraphs[0]['text'] ?? '';
        $subtitle = $paragraphs[1]['text'] ?? '';
        $lead = ['quote' => '', 'name' => '', 'position' => ''];
        $index = 2;
        if (isset($paragraphs[$index]) && $paragraphs[$index]['style'] === 'IntenseQuote') {
            $lead['quote'] = $this->unquote($paragraphs[$index]['text']);
            $index++;
            // attribution lines are often ALSO IntenseQuote-styled in these
            // files (0017-FE: they became bogus quotation elements) and may
            // be glued into ONE paragraph via w:br
            if (isset($paragraphs[$index]) && $this->isAttributionLine($paragraphs[$index], 60)) {
                $lines = $paragraphs[$index]['lines'];
                $lead['name'] = trim((string)$lines[0]);
                if (count($lines) > 1) {
                    $lead['position'] = trim(implode(', ', array_map('trim', array_slice($lines, 1))));
                }
                $index++;
                if ($lead['position'] === '' && isset($paragraphs[$index]) && $this->isAttributionLine($paragraphs[$index], 100)) {
                    $lead['position'] = $paragraphs[$index]['text'];
                    $index++;
                }
            }
        } else {
            $report[] = 'HINWEIS: kein Lead-Quote an Position 3';
        }

        // sections
        $sections = [];
        $quotes = [];
        $current = null;
        $body = [];     // items: ['p', html] | ['ul', [li,...]]
        $flushBody = function () use (&$sections, &$current, &$body): void {
            if ($current === null && $body === []) {
                return;
            }
            $html = '';
            foreach ($body as [$kind, $value]) {
                $html .= $kind === 'ul'
                    ? '<ul>' . implode('', array_map(static fn(string $li): string => '<li>' . htmlspecialchars($li, ENT_NOQUOTES | ENT_HTML5) . '</li>', $value)) . '</ul>'
                    : '<p>' . htmlspecialchars($value, ENT_NOQUOTES | ENT_HTML5) . '</p>';
            }
            if ($html !== '' || $current !== null) {
                $sections[] = ['type' => 'text', 'header' => (string)$current, 'body' => $html];
            }
            $body = [];
        };

        for ($i = $index; $i < count($paragraphs); $i++) {
            $paragraph = $paragraphs[$i];
            $lower = mb_strtolower(rtrim($paragraph['text'], ':'));
            if ($paragraph['style'] !== 'IntenseQuote' && in_array($lower, self::SECTION_NAMES, true)) {
                $flushBody();
                $current = $paragraph['text'];
                continue;
            }
            if ($paragraph['style'] === 'IntenseQuote') {
                // attribution BEFORE the quote ("... said:" / "... continues:")
                $name = '';
                $position = '';
                $last = $body !== [] ? $body[count($body) - 1] : null;
                if ($last !== null && $last[0] === 'p' && preg_match('/^(.{3,90}?)\s+(said|says|continues|adds|explains)\s*:?\s*$/iu', $last[1], $m)) {
                    [$name, $position] = $this->splitAttribution($m[1]);
                    array_pop($body);
                } elseif (isset($paragraphs[$i + 1]) && $paragraphs[$i + 1]['style'] !== 'IntenseQuote') {
                    $next = $paragraphs[$i + 1]['text'];
                    $nextLower = mb_strtolower(rtrim($next, ':'));
                    if (mb_strlen($next) <= 90 && str_contains($next, ',') && !in_array($nextLower, self::SECTION_NAMES, true)) {
                        [$name, $position] = $this->splitAttribution($next);
                        $i++;
                    }
                }
                $flushBody();
                $current = null;
                $sections[] = ['type' => 'quote', 'quote' => $this->unquote($paragraph['text']), 'name' => $name, 'position' => $position];
                $quotes[] = 1;
                continue;
            }
            if ($paragraph['style'] === 'ListParagraph') {
                $last = $body !== [] ? count($body) - 1 : -1;
                if ($last >= 0 && $body[$last][0] === 'ul') {
                    $body[$last][1][] = $paragraph['text'];
                } else {
                    $body[] = ['ul', [$paragraph['text']]];
                }
                continue;
            }
            if ($current === null && $sections === []) {
                $report[] = 'HINWEIS: Absatz vor erster Sektion: "' . mb_substr($paragraph['text'], 0, 50) . '..."';
            }
            $body[] = ['p', $paragraph['text']];
        }
        $flushBody();

        // teaser from the first text section
        $teaser = '';
        foreach ($sections as $section) {
            if ($section['type'] === 'text' && preg_match('/<p>(.*?)<\/p>/s', $section['body'], $m)) {
                $teaser = html_entity_decode($m[1], ENT_NOQUOTES | ENT_HTML5);
                if (mb_strlen($teaser) > 300) {
                    $teaser = mb_substr($teaser, 0, 300);
                    $teaser = mb_substr($teaser, 0, (int)mb_strrpos($teaser, ' ')) . ' ...';
                }
                break;
            }
        }

        return ['title' => $title, 'subtitle' => $subtitle, 'teaser' => $teaser,
            'lead' => $lead, 'sections' => $sections, 'quotes' => $quotes, 'report' => $report];
    }

    /** @return array{0:string,1:string} name, position */
    private function splitAttribution(string $value): array
    {
        $parts = explode(',', $value, 2);
        return [trim($parts[0]), trim($parts[1] ?? '')];
    }

    private function unquote(string $value): string
    {
        // multibyte-safe strip of straight and curly quotes at both ends
        return trim((string)preg_replace('/^[\s"\'\x{201C}\x{201D}\x{2018}\x{2019}]+|[\s"\'\x{201C}\x{201D}\x{2018}\x{2019}]+$/u', '', $value));
    }

    private function slugify(string $value): string
    {
        $value = str_replace(['+', '&'], ['plus', 'and'], mb_strtolower(trim($value)));
        return trim((string)preg_replace('/[^a-z0-9]+/', '-', $value), '-');
    }

    private function norm(string $value): string
    {
        return (string)preg_replace('/[^a-z0-9]+/', '', str_replace(['+', '&'], ['plus', 'and'], mb_strtolower($value)));
    }

    private function newId(): string
    {
        return 'NEWDS' . (++$this->newCounter);
    }
}