Import Articles Cli for Joomla 4, importing articles from form2content and Joomla Articles

importArticles.php

<?php
/*
Instructions for use:
- To skip the first 100 articles, use the argument "--skip 100".
- To manually define the limit of records to be imported, use the argument "-Limit" followed by the desired number, for example, "-Limit 50".
- To define the start date of import, use the argument "--start_date" followed by the date in "Yyyy-MM-DD" format, for example, "--start_date 2023-01-01".
- To define the end of importation, use the argument "--end_date" followed by the date in "Yyyy-MM-DD" format, for example, "--end_date 2023-12-31".
- To define the category of origin, use the argument "--source_category" followed by the desired ID, for example, "-source_category 10".
- To define the destination category, use the argument "--destination_category" followed by the desired category ID, for example, "--destination_category 5".
- To select Joomla Standard Articles from the Content table in the old database, use the argument "-standard_articles True".
  Example to import Joomla's standard articles:
  PHP ImportArticles.php --standard_Articles True --skip 100 --limit 50 --start_date 2023-01-01 --end_date 2023-12-31 --source_category 10 --destination_category 5
- To import articles from F2C (Form 2 Content), simply perform the script without the argument "-standard_articles":
  Example to import F2C articles:
  PHP ImportArticles.php --skip 100 --limit 50 --start_date 2023-01-01 --end_date 2023-12-31 --source_category 10 --destination_category 5
- To avoid cleaning HTML in imported articles, use the argument "--no_clean_html".
  Example to import articles without cleaning HTML:
  PHP ImportArticles.php --skip 100 --limit 50 --start_date 2023-01-01 --end_date 2023-12-31 --source_category 10 --destination_category 5 --no_clean_html
- The log file will be created in Administrator/Logs/Import_log.txt as defined in the $logfile variable and Joomla configuration.php.
*/

define('_JEXEC', 1);
use Joomla\CMS\Factory;
use Joomla\Console\Application;
use Joomla\CMS\HTML\HTMLHelper;
use Joomla\CMS\Language\Text;
use Joomla\CMS\Date\Date;

define('JPATH_BASE', dirname(dirname(__file__)));
define('DS', DIRECTORY_SEPARATOR);
require_once JPATH_BASE . DS . 'includes' . DS . 'defines.php';
require_once JPATH_BASE . DS . 'includes' . DS . 'framework.php';
define('SOURCE_CATEGORY', 90);
define('TARGET_CATEGORY', 90);
define('START_DATE', '0000-01-01');
define('END_DATE', '3500-12-31');
define('DEFAULT_LIMIT', 100);
// Set the user's time zone
date_default_timezone_set('America/Sao_Paulo');

class ImportArticles extends JApplicationCli
{
    public function slugify($text)
    {
        // replace non letter or digits by -
        $text = preg_replace('~[^\\pL\d]+~u', '-', $text);
        // trim
        $text = trim($text, '-');
        // transliterate
        $text = iconv('utf-8', 'us-ascii//TRANSLIT', $text);
        // lowercase
        $text = strtolower($text);
        // remove unwanted characters
        $text = preg_replace('~[^-\w]+~', '', $text);
        if (empty($text)) {
            return 'n-a';
        }
        // Check if slug already exists
        $db = Factory::getContainer()->get('DatabaseDriver');
        $baseSlug = $text;
        $counter = 1;
        while (true) {
            $slugToCheck = $counter > 1 ? $baseSlug . '-' . $counter : $baseSlug;
            $query = $db->getQuery(true)
                ->select($db->quoteName('alias'))
                ->from($db->quoteName('#__content'))
                ->where($db->quoteName('alias') . ' = ' . $db->quote($slugToCheck));
            $db->setQuery($query);
            $existingSlug = $db->loadResult();
            if (!$existingSlug) {
                return $slugToCheck;
            }
            $counter++;
        }
    }
    private function getOldDatabase()
    {
        $option = array(
            'driver' => 'mysql',
            'host' => 'yourhost.something',
            'user' => 'olddatabaseuser',
            'password' => 'olddatabasepassword',
            'database' => 'joomla_olddatabase',
            'prefix' => 'oldprefix_',
        );
        return JDatabaseDriver::getInstance($option);
    }
   
    private function checkOldIds($skipItems = 0, $limit = DEFAULT_LIMIT, $startDate = START_DATE, $endDate = END_DATE, $sourceCategory = SOURCE_CATEGORY, $destinationCategory = TARGET_CATEGORY, $standardArticles = false, $noCleanHtml = false)
   
    {   

      
        $olddatabase = $this->getOldDatabase();
        $db = Factory::getContainer()->get('DatabaseDriver');
        // Query for the rest of the content (excluding introtext and fulltext)
        if ($standardArticles) {
           // If the Argument --standard_srticles is defined as True
            // query to select articles from the Content table in the old database
            $queryfromOld = $olddatabase->getQuery(true)
                ->select('*')
                ->from('#__content')
                ->where('catid = ' . $olddatabase->quote($sourceCategory))
                ->where('created >= ' . $olddatabase->quote($startDate))
                ->where('created < ' . $olddatabase->quote($endDate))
                ->order('id ASC');
                if ($skipItems > 0) {
                    //  Add an OFFSET clause to skip the first X articles
                    $queryfromOld->setLimit($limit, $skipItems);
                } else {
                    // if there are no items to skip, use the normal limit
                    $queryfromOld->setLimit($limit);
                }
        } else {
        $queryfromOld = $olddatabase->getQuery(true)
            ->select('*')
            ->from('#__f2c_form')
            ->where('projectid = 5')
            ->where('state = 1')
            ->where('catid = ' . $olddatabase->quote($sourceCategory))
            ->where('created >= ' . $olddatabase->quote($startDate))
            ->where('created < ' . $olddatabase->quote($endDate))
            ->order('id ASC');
            if ($skipItems > 0) {
                // Add an OFFSET clause to skip the first X articles
                $queryfromOld->setLimit($limit, $skipItems);
            } else {
                // if there are no items to skip, use the normal limit
                $queryfromOld->setLimit($limit);
            }
        }
            
     
        $results = $olddatabase->setQuery($queryfromOld)->loadObjectList();
        echo "Total articles to be imported: " . count($results) . "\n";
        $this->logAction("Total articles to be imported: " . count($results) . "\n");
        $contador = $skipItems; // articles counter
        // Loop through the old articles
        foreach ($results as $oldarticle) {
            if ($standardArticles) {
                $introText = $oldarticle->introtext;
                $fullText = $oldarticle->fulltext;
            } else {
                $queryIntroText = $olddatabase->getQuery(true)
                    ->select('content')
                    ->from('#__f2c_fieldcontent')
                    ->where('formid = ' . $olddatabase->quote($oldarticle->id))
                    ->where('fieldid = 58');
                $introText = $olddatabase->setQuery($queryIntroText)->loadResult();
                $queryFullText = $olddatabase->getQuery(true)
                    ->select('content')
                    ->from('#__f2c_fieldcontent')
                    ->where('formid = ' . $olddatabase->quote($oldarticle->id))
                    ->where('fieldid = 59');
                $fullText = $olddatabase->setQuery($queryFullText)->loadResult();
            }
            
            $article = $db->getQuery(true)
                ->select('*')
                ->from($db->quoteName('#__content'))
                ->where($db->quoteName('id') . ' = ' . $db->quote($oldarticle->reference_id));
            $db->setQuery($article);
            $article = $db->loadObject();
          
            if (!$article) {
                $contador++;

                $oldDefaultArticleDataQuery = $olddatabase->getQuery(true)
                    ->select('*')
                    ->from('#__content')
                    ->where('id = ' . $olddatabase->quote($oldarticle->reference_id));
                $oldDefaultArticleData = $olddatabase->setQuery($oldDefaultArticleDataQuery)->loadObject();
                               
                $article = new stdClass();
                // $article->id = $oldarticle->id;
                $article->id = $oldarticle->reference_id;
                // get article->asset_id
                $queryAsset = $db->getQuery(true)
                    ->select('*')
                    ->from($db->quoteName('#__assets'))
                    ->where($db->quoteName('name') . ' = ' . $db->quote('com_content.article.' . $oldarticle->id));
                $db->setQuery($queryAsset);
                $asset = $db->loadObject();
                $article->asset_id = $asset->id;             
                $article->title = $oldarticle->title;
                $article->alias = $this->slugify($oldarticle->title);
                $article->introtext = $noCleanHtml ? $introText : $this->cleanTheHtml($introText);
                $article->fulltext = $noCleanHtml ? $fullText : $this->cleanTheHtml($fullText);
                $article->state = 1;
                $article->catid = $destinationCategory;
                $article->created = $oldarticle->created;
                $article->created_by = $oldarticle->created_by;
                $article->created_by_alias = $oldarticle->created_by_alias;
                $article->modified = $oldarticle->modified;
                $article->modified_by = $oldarticle->modified_by;
                $article->checked_out = '';
                $article->checked_out_time = '';
                $article->publish_up = $oldarticle->publish_up;
                $article->publish_down = '';
               
                $article->images = '{"image_intro":"","float_intro":"","image_intro_alt":"","image_intro_caption":"","image_fulltext":"","float_fulltext":"","image_fulltext_alt":"","image_fulltext_caption":""}';
                $article->urls = '{"urla":"","urlatext":"","targeta":"","urlb":"","urlbtext":"","targetb":"","urlc":"","urlctext":"","targetc":""}';
                $article->attribs = '{"article_layout":"","show_title":"","link_titles":"","show_tags":"","show_intro":"","info_block_position":"","info_block_show_title":"","show_category":"","link_category":"","show_parent_category":"","link_parent_category":"","show_author":"","link_author":"","show_create_date":"","show_modify_date":"","show_publish_date":"","show_item_navigation":"","show_hits":"","show_noauth":"","urls_position":"","alternative_readmore":"","article_page_title":"","show_publishing_options":"","show_article_options":"","show_urls_images_backend":"","show_urls_images_frontend":""}';
                $article->version = 1;
                $article->ordering = $oldDefaultArticleData->ordering ? $oldDefaultArticleData->ordering : '';
                $article->metakey = $oldDefaultArticleData->metakey ? $oldDefaultArticleData->metakey : '';
                $article->metadesc = $oldDefaultArticleData->metadesc ? $oldDefaultArticleData->metadesc : '';
                $article->access = 1;
                $article->hits = $oldDefaultArticleData->hits ? $oldDefaultArticleData->hits : 0;
                $article->metadata = '{"robots":"","author":"","rights":""}';
                $article->featured = 0;
                $article->language = '*';               
                $article->xreference = '';
                $article->note = '';
                $db->insertObject('#__content', $article, 'id');
                
                // plural
                if ($contador > 1) {
                    echo $contador . " completed tasks :::>> ";
                } else {
                    echo $contador . " completed tasks :::>> ";
                }
                
                echo "Article created: " . $article->id . ": " . $article->title . " - Created at: " . $article->created . " :::>> ";
        
                if ($contador > 1) {
                    $this->logAction($contador . " completed tasks :::>> ");
                } else {
                    $this->logAction($contador . " completed task :::>> ");
                }
        
            $this->logAction("Article created: " . $article->id . ": " . $article->title . " - Created at: " . $article->created . " :::>> ");
            $this->adicionarReferenciasAsset($article);
            $this->adicionarReferenciasWorkflow($article);
         
           
            } else {
                // pula artigo
                echo "The article already exists: " . $article->id . ": " . $article->title . " :::>> ";
                $this->logAction("The article already exists: " . $article->id . ": " . $article->title . " :::>> ");
                $this->adicionarReferenciasAsset($article);
                $this->adicionarReferenciasWorkflow($article);
            }
           
            
        }
    }
    private function adicionarReferenciasAsset($article)
    {
        $db = Factory::getContainer()->get('DatabaseDriver');
        $queryAsset = $db->getQuery(true)
            ->select('*')
            ->from($db->quoteName('#__assets'))
            ->where($db->quoteName('name') . ' = ' . $db->quote('com_content.article.' . $article->id));
        $db->setQuery($queryAsset);
        $asset = $db->loadObject();
        if (!$asset) {
            $asset = new stdClass();
            //  calculate the asset parent_id based on the category asset id
            $queryCategoryAsset = $db->getQuery(true)
                ->select('*')
                ->from($db->quoteName('#__assets'))
                ->where($db->quoteName('name') . ' = ' . $db->quote('com_content.category.' . $article->catid));
            $db->setQuery($queryCategoryAsset);
            $categoryAsset = $db->loadObject();
            $asset->parent_id = $categoryAsset->id;
            //  calculate the asset lft, rgt , lft is max rgt + 1, rgt is lft + 1
            $queryMaxRgt = $db->getQuery(true)
                ->select('MAX(rgt)')
                ->from($db->quoteName('#__assets'));
            $db->setQuery($queryMaxRgt);
            $maxRgt = $db->loadResult();
            $asset->lft = $maxRgt + 1;
            $asset->rgt = $maxRgt + 2;         
            //  asset level is the same as the category level
            $asset->level = $categoryAsset->level;
            $asset->name = 'com_content.article.' . $article->id;
            $asset->title = $article->title;
            $asset->rules = '{}';
            $db->insertObject('#__assets', $asset, 'id');
            echo "Asset created: " . $asset->id . ": " . $asset->name . " :::>> ";
            $this->logAction("Asset created: " . $asset->id . ": " . $asset->name . " :::>> ");
            $this->updateAssetId($article->id);
            
        } else {
            echo "The asset already exists: " . $asset->id . ": " . $asset->name . " :::>> ";
            $this->logAction("The asset already exists: " . $asset->id . ": " . $asset->name . " :::>> ");
        }
    }
    private function adicionarReferenciasWorkflow($article)
    {
        $db = Factory::getContainer()->get('DatabaseDriver');
        $queryWorkflow = $db->getQuery(true)
            ->select('*')
            ->from($db->quoteName('#__workflow_associations'))
            ->where($db->quoteName('item_id') . ' = ' . $db->quote($article->id));           
        $db->setQuery($queryWorkflow);
        $workflow = $db->loadObject();
        if (!$workflow) {
            $workflow = new stdClass();
            $workflow->item_id = $article->id;
            $workflow->stage_id = 1;
            $workflow->extension = "com_content.article";
            $db->insertObject('#__workflow_associations', $workflow, 'id');
            echo "The workflow was created: " . $workflow->item_id . ": " . $workflow->extension . "\n";
            $this->logAction("The workflow was created: " . $workflow->item_id . ": " . $workflow->extension . "\n");
        } else {
            echo "The workflow already exists: " . $workflow->item_id . ": " . $workflow->extension . "\n";
            $this->logAction("The workflow already exists: " . $workflow->item_id . ": " . $workflow->extension . "\n");
        }
    }
    // update asset_id in content table based on assets table and article id
    private function updateAssetId($articleId)
    {
        $db = Factory::getContainer()->get('DatabaseDriver');
        $queryAsset = $db->getQuery(true)
            ->select('*')
            ->from($db->quoteName('#__assets'))
            ->where($db->quoteName('name') . ' = ' . $db->quote('com_content.article.' . $articleId));
        $db->setQuery($queryAsset);
        $asset = $db->loadObject();
        if ($asset) {
            $queryContent = $db->getQuery(true)
                ->update($db->quoteName('#__content'))
                ->set($db->quoteName('asset_id') . ' = ' . $db->quote($asset->id))
                ->where($db->quoteName('id') . ' = ' . $db->quote($articleId));
            $db->setQuery($queryContent);
            $db->execute();
            echo "Asset updated: " . $asset->id . ": " . $asset->name . " :::>> ";
            $this->logAction("Asset updated: " . $asset->id . ": " . $asset->name . " :::>> ");
        } else {
            echo "Asset not found: " . $articleId . "\n";
            $this->logAction("Asset not found: " . $articleId . "\n");
        }
    }
   
    // Log actions
    private function logAction($message) {
       $logFile = Factory::getConfig()->get('log_path') . '/import_log.txt';
        $logEntry = date('Y-m-d H:i:s') . ' - ' . $message . PHP_EOL;
        file_put_contents($logFile, $logEntry, FILE_APPEND);
    }
  
    // Clean HTML
    private function cleanTheHtml($string, $limpaTudo = false)
    {
        $string = preg_replace('/(<[^>]+) class=".*?"/i', '$1', $string);
        $string = preg_replace('/(class|style)[^>]*/', '', $string);
        // replace <p > to <p> and other tags
        $string = preg_replace('/<p\s+>/', '<p>', $string);
        // replace • to -
        $string = preg_replace('/•/', '-', $string);
        
        return $limpaTudo ? strip_tags($string) : strip_tags($string, '<iframe><img><p><span><a><b><div><strong><em><br><br/><br />');
    }
    private function clearLog()
    {
        $logFile = Factory::getConfig()->get('log_path') . '/import_log.txt';
        if (file_exists($logFile)) {
            file_put_contents($logFile, ''); // Clear the log file
            echo "Log file cleared.\n";
        } else {
            echo "Log file does not exist.\n";
        }
    }

    // Execute the checkOldIds function
    public function doExecute()
    {
        $clearScreen = $this->input->get('clear_screen', false);
        if ($clearScreen) {
            $this->clearTerminalScreen();
        }

        $clearLog = $this->input->get('clear_log', false);
        if ($clearLog) {
            $this->clearLog();
        }

        $skipItems = $this->input->get('skip', 0);
        $limit = $this->input->get('limit', DEFAULT_LIMIT);
        $startDate = $this->input->get('start_date', START_DATE);
        $endDate = $this->input->get('end_date', END_DATE);
        $sourceCategory = $this->input->get('source_category', SOURCE_CATEGORY);
        $destinationCategory = $this->input->get('destination_category', TARGET_CATEGORY);
        $standardArticles = $this->input->get('standard_articles', false);
        $noCleanHtml = $this->input->get('no_clean_html', false);

        $this->checkOldIds($skipItems, $limit, $startDate, $endDate, $sourceCategory, $destinationCategory, $standardArticles, $noCleanHtml);
  


    }
    // Implement the getName method
    public function getName()
    {
        return 'ImportArticles';
    }
}
JApplicationCli::getInstance('ImportArticles')->execute();