WordPress: GGA Joomla Content Importer - Imports articles from a Joomla database as WordPress posts

gga-import-joomla-articles.php

<?php
/*
Plugin Name: GGA Joomla Content Importer
Description: Imports articles from Joomla 1.5x into WordPress
Author: <a href="https://twitter.com/GunGeekATX" target="_blank">@GunGeekATX</a>
Version: 1.1
*/


class ggaJoomlaArticleImporter 
{

  /* 
	* This plugin was written to handle importing articles from a specific section in Joomla 1.5 into WordPress, specifically
	* for a Joomla 1.5 site that had several years of press releases.  It's only designed to get articles in one section from Joomla and 
	* assign one category in WordPress.  It's a good starting base if you need to import multiple articles from a Joomla 1.5 database.
	*/ 


	// Data for your Joomla DB
	var $joomla_db;
	var $joomla_host 				= 'localhost';
	var $joomla_username 			= 'your_username_here';
	var $joomla_password 			= 'your_password_here';
	var $joomla_dbname	 			= 'joomla_db_name';
	var $joomla_tbl_prefix 			= 'jos_';
	var $joomla_section_id			= 3;

	// Data for your WordPress install
	var $wp_target_category_slug	= 'press-announcements';
	var $wp_cleanup_category_slug	= 'press-announcements-cleanup';
	var $wp_post_author_id			= 2;

	var $actionPrefix 				= 'ggaJoomlaArticleImporter';
	var $version 					= '1.0';


	function __construct()
	{
		add_action( 'init', 		array(&$this, 'init'));
		add_action( 'admin_menu', 	array(&$this, 'admin_menu'));
		add_action( 'ggaJoomlaArticleImporter_run_cron_import_job', array(&$this, 'run_import_content_job'), 10, 1 );
	}

	function __destruct() {
    	if ($this->joomla_db)
    		mysql_close($this->joomla_db);
	}

	public function init() {
		add_action('wp_ajax_' . $this->actionPrefix . '_create_import_content_job', array(&$this, 'create_import_content_job'));
		add_action('wp_ajax_' . $this->actionPrefix . '_check_import_status', array(&$this, 'check_import_status'));
	}


	public function admin_menu() {
		add_menu_page( 'GGA Joomla Importer', 'Joomla Importer', 'import', 'gga-joomla-importer', array(&$this, 'admin_page'));
	}

	function connect_to_joomla_db() {

		$this->joomla_db = @mysql_connect($this->joomla_host, $this->joomla_username, $this->joomla_password);

		if (!$this->joomla_db)
			return new WP_Error('joomla_connect', 'Unable to connect to db');

		if (false === @mysql_select_db($this->joomla_dbname, $this->joomla_db))
			return new WP_Error('joomla_select_db', 'Unable to select db');

	}


	public function admin_page()
	{

		?>
		<div class="wrap">
			<h2>Joomla to WordPress Importer</h2>
			<p>
				This plugin will import articles from from your Joomla 1.5 database into this WordPress site as posts.
			</p>

			<p>
				<table>
					<tr>
						<td colspan="2">
							<h3>Source</h3>
						</td>
					</tr>
					<tr>
						<td>
							Joomla User@Host:
						</td>
						<td>
							<?php echo htmlspecialchars($this->joomla_username) ?>@<?php echo htmlspecialchars($this->joomla_host) ?>
						</td>
					</tr>
					<tr>
						<td>
							Joomla Database
						</td>
						<td>
							<?php echo htmlspecialchars($this->joomla_dbname) ?>
						</td>
					</tr>
					<tr>
						<td>
							Joomla Table Prefix:
						</td>
						<td>
							<?php echo htmlspecialchars($this->joomla_tbl_prefix) ?>
						</td>
					</tr>
					<tr>
						<td>
							Joomla Section Id:
						</td>
						<td>
							<?php echo htmlspecialchars($this->joomla_section_id) ?>
						</td>
					</tr>
					<tr>
						<td colspan="2">
							<h3>Target</h3>
						</td>
					</tr>
					<tr>
						<td>WordPress Category Slug:</td>
						<td>
							<?php echo htmlspecialchars($this->wp_target_category_slug) ?>
						</td>
					</tr>
					<tr>
						<td>WordPress Category Slug (cleanup):</td>
						<td>
							<?php echo htmlspecialchars($this->wp_cleanup_category_slug) ?>
						</td>
					</tr>
					<tr>
						<td>WordPress Post Author Id:</td>
						<td>
							<?php echo htmlspecialchars($this->wp_post_author_id) ?>
						</td>
					</tr>
				</table>
			</p>



			<input type="button" id="ggaStartImport" value="Start Import" class="button button-primary" />

			<br/>

			<div id="ggaImportMessageWrapper" style="display: none ">

				<h3>Import progress</h3>
				<div class="ggaImportMessage">
				</div>

			</div>


			<script type="text/javascript">

				jQuery(document).ready(function() {

					jQuery('#ggaStartImport').click(function() {

						jQuery(this).attr('disabled', 'disabled').val('Running Import...');
						jQuery('#ggaImportMessageWrapper').show();

						ggaSetImportMessage('Starting Import...<br/>')

						jQuery.post(ajaxurl, { 'action':'<?php echo $this->actionPrefix ?>_create_import_content_job' }, function(response) {
						
							if (response.errors) { 

								if (response.errors.joomla_connect)
									ggaSetImportMessage('Error: Unable to connect to Joomla SQL host');
								if (response.errors.joomla_select_db)
									ggaSetImportMessage('Error: Unable to select to Joomla DB');
								if (response.errors.joomla_query)
									ggaSetImportMessage('Error: Unable to query Joomla content table');
								
							} else {
								ggaAppendImportMessage('Transient job id ' + response.job_id + '<br/>');
								setTimeout(ggaCheckImportStatus, 1000, response.job_id);
							}



						}, 'json');	


					});


				});

				function ggaSetImportMessage(message) {
					jQuery('.ggaImportMessage').html(message);
				}

				function ggaAppendImportMessage(message) {
					jQuery('.ggaImportMessage').append(message);
				}

				function ggaCheckImportStatus(jobid) {

					jQuery.post(ajaxurl, { 'action':'<?php echo $this->actionPrefix ?>_check_import_status', 'job_id':jobid }, function(response) {
						
						ggaSetImportMessage('Imported ' + response.total_imported + '/' + response.total_to_import );
						
						if (response.total_imported < response.total_to_import )
							setTimeout(ggaCheckImportStatus, 1000, jobid);
						else 
							jQuery('#ggaStartImport').val('Import completed');

					}, 'json');
					
				}



			</script>

		</div>
		<?php 
	}


	public function create_import_content_job() {

		$id = '_gji_' . uniqid();  // short for _gga_joomla_import_

		$trans = new stdClass();
		$trans->job_id = $id;
		$trans->results = array();

		$ids = $this->get_content_ids_to_import();
		
		if (is_wp_error( $ids ))
			$trans->errors = $ids->errors;
		else {
			
			$trans->content_ids = $ids;

			set_transient( $id, $trans, 60*60*24 );
			
			// create a one-time cron job to run the import
			wp_schedule_single_event(time(), 'ggaJoomlaArticleImporter_run_cron_import_job', array($id));

		}

		echo json_encode($trans);
		die();

	}

	public function run_import_content_job($transient_id) {

		$trans = get_transient( $transient_id );

		$response = new stdClass();
		$response->job_id = '';

		if ($trans && $trans->content_ids && is_array($trans->content_ids)) {

			$response->job_id = $trans->job_id;

			for ($i=0; $i < count($trans->content_ids); $i++) 
				$this->import_content_id($trans->content_ids[$i], $trans->job_id);
			
		}

		echo json_encode($response);
		die();

	}

	function check_import_status() {

		$trans = get_transient( $_REQUEST['job_id'] );


		$results = new stdClass();
		$results->job_id = '';

		if ($trans) {
			$results->job_id = $trans->job_id;
			$results->total_to_import = count($trans->content_ids);
			$results->total_imported = 0;

			for ($i=0; $i < count($trans->results); $i++)  
				$results->total_imported += ( is_wp_error($trans->results[$i]->wp_post_error) || $trans->results[$i]->wp_post_id > 0) ? 1 : 0;

		}

		echo json_encode($results);
		die();

	}


	function get_content_ids_to_import() {

		$err = null;
		if (!$this->joomla_db) 
			$err = $this->connect_to_joomla_db();

		if (is_wp_error( $err ))
			return $err;

		$content_ids = array();

		$result = @mysql_query(
			sprintf('SELECT `%scontent`.`id` FROM  `%scontent`  WHERE `%scontent`.`sectionid` = %d and state = 1', 
				$this->joomla_tbl_prefix, $this->joomla_tbl_prefix, $this->joomla_tbl_prefix, $this->joomla_section_id
			), 
			$this->joomla_db);

		if (!$result)
			return new WP_Error('joomla_query', 'Unable to query db');

		$rows = mysql_num_rows($result);

		for ($i=0; $i < $rows; $i++) 
			$content_ids[] = mysql_result($result, $i, 'id') or die(mysql_error());

		return $content_ids;

	}

	function import_content_id($id, $transient_id) {

		if (!$this->joomla_db) 
			$this->connect_to_joomla_db();

		$trans = get_transient( $transient_id);


		$results = new stdClass();
		$results->wp_post_id = 0;
		$results->wp_post_error = null;
		$results->id = $content_ids[$i];
		$results->title =  '';
		$results->existed = false;


		$cat = get_category_by_slug( $this->wp_target_category_slug );


		$content_ids = array();

		$result = mysql_query(sprintf("SELECT `id`, `title`, `alias`, `fulltext`, `publish_up`, `alias` FROM  `%scontent`  WHERE `%scontent`.`id` = %d", 
				$this->joomla_tbl_prefix, $this->joomla_tbl_prefix, $id
			), $this->joomla_db) or die(mysql_error());
		
		$rows = mysql_num_rows($result);
		
		if ($rows > 0) {


			$alias = mysql_result($result, 0, 'alias');
			$title = $this->word_cleanup(mysql_result($result, 0, 'title'));
			$fulltext = $this->word_cleanup(mysql_result($result, 0, 'fulltext'));
			
			$results = new stdClass();
			$results->id = mysql_result($result, 0, 'id');
			$results->title =  $title;

			$slug = wp_unique_post_slug(
				sanitize_title_with_dashes($results->id . '-' . $alias),
				0, 
				'publish',
				'post',
				0);

			$post_data = array(
				'post_title' => $title,
				'post_content' => $fulltext,
				'post_status' => 'publish',
				'post_name' => $slug,
				'post_date' => mysql_result($result, 0, 'publish_up'),
				'comment_status' => 'closed',
				'ping_status' => 'closed',
				'post_type' => 'post',
				'post_author' => $this->wp_post_author_id
			);

			// query by post_meta to see if this has already been imported
			$query = new WP_Query( array( 'meta_key' => '_gga_joomla_content_id', 'meta_value' => $results->id ) );

			if ($query->have_posts() ) {
				$query->the_post();
				$results->existed = true;
				$results->existing_id = get_the_id();
				$results->wp_post_id = $results->existing_id;
				
				wp_update_post( $post_data );

			}
			else 
			{

				$insert_results = wp_insert_post( $post_data, true );
				
				if (is_wp_error( $insert_results ) ) {
					$results->wp_post_error = $insert_results;
				}
				else  {
					$results->wp_post_id = $insert_results;

					// flag this with the content ID from Joomla
					update_post_meta( $results->wp_post_id, '_gga_joomla_content_id', $results->id );  
					
					// flag that this was converted using this plugin, just in case you need to query for it later
					update_post_meta( $results->wp_post_id, '_gga_imported_from_joomla', true );  

				}




			}

			wp_set_post_categories($results->wp_post_id, array($cat->cat_ID));

			// check the post content to see if anything needs cleanup, like img tags, links to index.php, etc
			if ($this->post_needs_cleanup($fulltext))
				wp_set_post_categories($results->wp_post_id, array($cat->cat_ID, $cat_cleanup->cat_ID));  // need to pass all categories, this is an overwrite function

			wp_reset_query();
			wp_reset_postdata();

		}


		$trans->results[] = $results;

		set_transient( $transient_id, $trans, 60*60*24 );
		


	}


	function post_needs_cleanup($content) {

		// images linked to images/stories, etc
		// linkes to the default index.php file in Joomla vs permalinks

		return 
			false !== stripos($content, '<img') ||
			false !== stripos($content, 'index.php')
		;

	}


	function word_cleanup ($str)
	{

		// this cleans up some stuff copy/pasted from MS Word

		$str = mb_convert_encoding($str, 'HTML-ENTITIES', 'UTF-8');

		$str = str_replace('&#2013266080;', ' ', $str); // space
		$str = str_replace('&#2013266094;', '®', $str); // reg trademark
		$str = str_replace('&#2013266073;', '™', $str); // non-reg trademark
		$str = str_replace('&#2013266070;', '-', $str); 
		$str = str_replace('&#2013266067;', '"', $str); 
		$str = str_replace('&#2013266068;', '"', $str); 
		$str = str_replace('&#2013266065;', "'", $str); 
		$str = str_replace('&#2013266066;', "'", $str); 

		return $str;

	}


}

new ggaJoomlaArticleImporter();