MokoSuiteOpenGraph/source/packages/com_mokoog/src/Controller/BatchController.php

<?php

/**
 * @package     MokoSuiteOpenGraph
 * @subpackage  com_mokoog
 * @author      Moko Consulting <hello@mokoconsulting.tech>
 * @copyright   Copyright (C) 2026 Moko Consulting. All rights reserved.
 * @license     GNU General Public License version 3 or later; see LICENSE
 */

namespace Joomla\Component\MokoOG\Administrator\Controller;

defined('_JEXEC') or die;

use Joomla\CMS\Factory;
use Joomla\CMS\Language\Text;
use Joomla\CMS\MVC\Controller\BaseController;
use Joomla\CMS\Response\JsonResponse;
use Joomla\CMS\Session\Session;

class BatchController extends BaseController
{
    /**
     * Count the total articles eligible for batch generation.
     *
     * @return  void
     */
    public function count(): void
    {
        Session::checkToken('get') || throw new \RuntimeException(Text::_('JINVALID_TOKEN'), 403);

        $identity = Factory::getApplication()->getIdentity();

        if (!$identity->authorise('mokoog.batch', 'com_mokoog')
            && !$identity->authorise('core.create', 'com_mokoog')) {
            throw new \RuntimeException(Text::_('JLIB_APPLICATION_ERROR_ACCESS_FORBIDDEN'), 403);
        }

        $db    = Factory::getContainer()->get(\Joomla\Database\DatabaseInterface::class);
        $query = $db->getQuery(true)
            ->select('COUNT(*)')
            ->from($db->quoteName('#__content', 'c'))
            ->leftJoin(
                $db->quoteName('#__mokoog_tags', 't')
                . ' ON ' . $db->quoteName('t.content_type') . ' = ' . $db->quote('com_content')
                . ' AND ' . $db->quoteName('t.content_id') . ' = ' . $db->quoteName('c.id')
            )
            ->where($db->quoteName('c.state') . ' = 1')
            ->where($db->quoteName('t.id') . ' IS NULL');

        $db->setQuery($query);
        $total = (int) $db->loadResult();

        echo new JsonResponse(['total' => $total]);

        Factory::getApplication()->close();
    }

    /**
     * Process a chunk of articles for batch OG generation.
     *
     * @return  void
     */
    public function process(): void
    {
        Session::checkToken('get') || throw new \RuntimeException(Text::_('JINVALID_TOKEN'), 403);

        $identity = Factory::getApplication()->getIdentity();

        if (!$identity->authorise('mokoog.batch', 'com_mokoog')
            && !$identity->authorise('core.create', 'com_mokoog')) {
            throw new \RuntimeException(Text::_('JLIB_APPLICATION_ERROR_ACCESS_FORBIDDEN'), 403);
        }

        $app    = Factory::getApplication();
        $input  = $app->getInput();
        $limit  = min($input->getInt('limit', 50), 200);
        $lastId = max(0, $input->getInt('lastid', 0));

        $db    = Factory::getContainer()->get(\Joomla\Database\DatabaseInterface::class);
        $query = $db->getQuery(true)
            ->select($db->quoteName([
                'c.id', 'c.title', 'c.metadesc', 'c.introtext', 'c.fulltext', 'c.images',
            ]))
            ->from($db->quoteName('#__content', 'c'))
            ->leftJoin(
                $db->quoteName('#__mokoog_tags', 't')
                . ' ON ' . $db->quoteName('t.content_type') . ' = ' . $db->quote('com_content')
                . ' AND ' . $db->quoteName('t.content_id') . ' = ' . $db->quoteName('c.id')
            )
            ->where($db->quoteName('c.state') . ' = 1')
            ->where($db->quoteName('t.id') . ' IS NULL')
            ->where($db->quoteName('c.id') . ' > ' . $lastId)
            ->order($db->quoteName('c.id') . ' ASC');

        // Cursor-based pagination by id: each chunk fetches the next articles whose
        // id is greater than the previous chunk's highest id. A row that fails to
        // insert is passed over on the next chunk (its id is already behind the
        // cursor) instead of being re-fetched forever, so the batch always reaches
        // the end. The client stops when a chunk examines 0 rows.
        $db->setQuery($query, 0, $limit);
        $articles = $db->loadObjectList();

        $created         = 0;
        $skipped         = 0;
        $lastProcessedId = $lastId;
        $now             = Factory::getDate()->toSql();

        foreach ($articles as $article) {
            $lastProcessedId = (int) $article->id;

            $ogTitle       = $article->title;
            $ogDescription = $this->extractDescription($article);
            $ogImage       = $this->extractImage($article);

            $record = (object) [
                'content_type'     => 'com_content',
                'content_id'       => (int) $article->id,
                'og_title'         => $ogTitle,
                'og_description'   => $ogDescription,
                'og_image'         => $ogImage,
                'og_type'          => 'article',
                'seo_title'        => '',
                'meta_description' => $article->metadesc ?: '',
                'robots'           => '',
                'canonical_url'    => '',
                'language'         => '*',
                'published'        => 1,
                'created'          => $now,
                'modified'         => $now,
            ];

            try {
                $db->insertObject('#__mokoog_tags', $record);
                $created++;
            } catch (\RuntimeException $e) {
                $skipped++;
                \Joomla\CMS\Log\Log::add('Batch insert failed for article ' . $article->id . ': ' . $e->getMessage(), \Joomla\CMS\Log\Log::WARNING, 'mokoog');
            }
        }

        echo new JsonResponse([
            'created'  => $created,
            'skipped'  => $skipped,
            'examined' => \count($articles),
            'last_id'  => $lastProcessedId,
        ]);

        $app->close();
    }

    /**
     * Extract a description from article content.
     *
     * @param   object  $article  Article record
     *
     * @return  string
     */
    private function extractDescription(object $article): string
    {
        // Prefer meta description if set
        if (!empty($article->metadesc)) {
            return $article->metadesc;
        }

        // Fall back to intro text
        $text = $article->introtext ?: $article->fulltext;
        $text = strip_tags($text);
        $text = trim(preg_replace('/\s+/', ' ', $text));

        if (mb_strlen($text) > 160) {
            $text = mb_substr($text, 0, 157) . '...';
        }

        return $text;
    }

    /**
     * Extract the best image from article data.
     *
     * @param   object  $article  Article record
     *
     * @return  string
     */
    private function extractImage(object $article): string
    {
        if (!empty($article->images)) {
            $images = json_decode($article->images, true);

            if (!empty($images['image_fulltext'])) {
                return $images['image_fulltext'];
            }

            if (!empty($images['image_intro'])) {
                return $images['image_intro'];
            }
        }

        return '';
    }
}