<?php

/**

 * WordPress eXtended RSS file parser implementations

 *

 * @package WordPress

 * @subpackage Importer

 */

/**

 * WordPress Importer class for managing parsing of WXR files.

 */

class WXR_Parser {

	function parse( $file ) {

		// Attempt to use proper XML parsers first

		if ( extension_loaded( 'simplexml' ) ) {

			$parser = new WXR_Parser_SimpleXML;

			$result = $parser->parse( $file );

			// If SimpleXML succeeds or this is an invalid WXR file then return the results

			if ( ! is_wp_error( $result ) || 'SimpleXML_parse_error' != $result->get_error_code() )

				return $result;

		} else if ( extension_loaded( 'xml' ) ) {

			$parser = new WXR_Parser_XML;

			$result = $parser->parse( $file );

			// If XMLParser succeeds or this is an invalid WXR file then return the results

			if ( ! is_wp_error( $result ) || 'XML_parse_error' != $result->get_error_code() )

				return $result;

		}

		// We have a malformed XML file, so display the error and fallthrough to regex

		if ( isset($result) && defined('IMPORT_DEBUG') && IMPORT_DEBUG ) {

			echo '<pre>';

			if ( 'SimpleXML_parse_error' == $result->get_error_code() ) {

				foreach  ( $result->get_error_data() as $error )

					echo $error->line . ':' . $error->column . ' ' . esc_html( $error->message ) . "\n";

			} else if ( 'XML_parse_error' == $result->get_error_code() ) {

				$error = $result->get_error_data();

				echo $error[0] . ':' . $error[1] . ' ' . esc_html( $error[2] );

			}

			echo '</pre>';

			echo '<p><strong>' . __( 'There was an error when reading this WXR file', 'wordpress-importer' ) . '</strong><br />';

			echo __( 'Details are shown above. The importer will now try again with a different parser...', 'wordpress-importer' ) . '</p>';

		}

		// use regular expressions if nothing else available or this is bad XML

		$parser = new WXR_Parser_Regex;

		return $parser->parse( $file );

	}

}

/**

 * WXR Parser that makes use of the SimpleXML PHP extension.

 */

class WXR_Parser_SimpleXML {

	function parse( $file ) {

		$authors = $posts = $categories = $tags = $terms = array();

		$internal_errors = libxml_use_internal_errors(true);

		$xml = simplexml_load_file( $file );

		// halt if loading produces an error

		if ( ! $xml )

			return new WP_Error( 'SimpleXML_parse_error', __( 'There was an error when reading this WXR file', 'wordpress-importer' ), libxml_get_errors() );

		$wxr_version = $xml->xpath('/rss/channel/wp:wxr_version');

		if ( ! $wxr_version )

			return new WP_Error( 'WXR_parse_error', __( 'This does not appear to be a WXR file, missing/invalid WXR version number', 'wordpress-importer' ) );

		$wxr_version = (string) trim( $wxr_version[0] );

		// confirm that we are dealing with the correct file format

		if ( ! preg_match( '/^\d+\.\d+$/', $wxr_version ) )

			return new WP_Error( 'WXR_parse_error', __( 'This does not appear to be a WXR file, missing/invalid WXR version number', 'wordpress-importer' ) );

		$base_url = $xml->xpath('/rss/channel/wp:base_site_url');

		$base_url = (string) trim( $base_url[0] );

		$namespaces = $xml->getDocNamespaces();

		if ( ! isset( $namespaces['wp'] ) )

			$namespaces['wp'] = 'http://wordpress.org/export/1.1/';

		if ( ! isset( $namespaces['excerpt'] ) )

			$namespaces['excerpt'] = 'http://wordpress.org/export/1.1/excerpt/';

		// grab authors

		foreach ( $xml->xpath('/rss/channel/wp:author') as $author_arr ) {

			$a = $author_arr->children( $namespaces['wp'] );

			$login = (string) $a->author_login;

			$authors[$login] = array(

				'author_id' => (int) $a->author_id,

				'author_login' => $login,

				'author_email' => (string) $a->author_email,

				'author_display_name' => (string) $a->author_display_name,

				'author_first_name' => (string) $a->author_first_name,

				'author_last_name' => (string) $a->author_last_name

			);

		}

		// grab cats, tags and terms

		foreach ( $xml->xpath('/rss/channel/wp:category') as $term_arr ) {

			$t = $term_arr->children( $namespaces['wp'] );

			$categories[] = array(

				'term_id' => (int) $t->term_id,

				'category_nicename' => (string) $t->category_nicename,

				'category_parent' => (string) $t->category_parent,

				'cat_name' => (string) $t->cat_name,

				'category_description' => (string) $t->category_description

			);

		}

		foreach ( $xml->xpath('/rss/channel/wp:tag') as $term_arr ) {

			$t = $term_arr->children( $namespaces['wp'] );

			$tags[] = array(

				'term_id' => (int) $t->term_id,

				'tag_slug' => (string) $t->tag_slug,

				'tag_name' => (string) $t->tag_name,

				'tag_description' => (string) $t->tag_description

			);

		}

		foreach ( $xml->xpath('/rss/channel/wp:term') as $term_arr ) {

			$t = $term_arr->children( $namespaces['wp'] );

			$terms[] = array(

				'term_id' => (int) $t->term_id,

				'term_taxonomy' => (string) $t->term_taxonomy,

				'slug' => (string) $t->term_slug,

				'term_parent' => (string) $t->term_parent,

				'term_name' => (string) $t->term_name,

				'term_description' => (string) $t->term_description

			);

		}

		// grab posts

		foreach ( $xml->channel->item as $item ) {

			$post = array(

				'post_title' => (string) $item->title,

				'guid' => (string) $item->guid,

			);

			$dc = $item->children( 'http://purl.org/dc/elements/1.1/' );

			$post['post_author'] = (string) $dc->creator;

			$content = $item->children( 'http://purl.org/rss/1.0/modules/content/' );

			$excerpt = $item->children( $namespaces['excerpt'] );

			$post['post_content'] = (string) $content->encoded;

			$post['post_excerpt'] = (string) $excerpt->encoded;

			$wp = $item->children( $namespaces['wp'] );

			$post['post_id'] = (int) $wp->post_id;

			$post['post_date'] = (string) $wp->post_date;

			$post['post_date_gmt'] = (string) $wp->post_date_gmt;

			$post['comment_status'] = (string) $wp->comment_status;

			$post['ping_status'] = (string) $wp->ping_status;

			$post['post_name'] = (string) $wp->post_name;

			$post['status'] = (string) $wp->status;

			$post['post_parent'] = (int) $wp->post_parent;

			$post['menu_order'] = (int) $wp->menu_order;

			$post['post_type'] = (string) $wp->post_type;

			$post['post_password'] = (string) $wp->post_password;

			$post['is_sticky'] = (int) $wp->is_sticky;

			if ( isset($wp->attachment_url) )

				$post['attachment_url'] = (string) $wp->attachment_url;

			foreach ( $item->category as $c ) {

				$att = $c->attributes();

				if ( isset( $att['nicename'] ) )

					$post['terms'][] = array(

						'name' => (string) $c,

						'slug' => (string) $att['nicename'],

						'domain' => (string) $att['domain']

					);

			}

			foreach ( $wp->postmeta as $meta ) {

				$post['postmeta'][] = array(

					'key' => (string) $meta->meta_key,

					'value' => (string) $meta->meta_value

				);

			}

			foreach ( $wp->comment as $comment ) {

				$meta = array();

				if ( isset( $comment->commentmeta ) ) {

					foreach ( $comment->commentmeta as $m ) {

						$meta[] = array(

							'key' => (string) $m->meta_key,

							'value' => (string) $m->meta_value

						);

					}

				}

				$post['comments'][] = array(

					'comment_id' => (int) $comment->comment_id,

					'comment_author' => (string) $comment->comment_author,

					'comment_author_email' => (string) $comment->comment_author_email,

					'comment_author_IP' => (string) $comment->comment_author_IP,

					'comment_author_url' => (string) $comment->comment_author_url,

					'comment_date' => (string) $comment->comment_date,

					'comment_date_gmt' => (string) $comment->comment_date_gmt,

					'comment_content' => (string) $comment->comment_content,

					'comment_approved' => (string) $comment->comment_approved,

					'comment_type' => (string) $comment->comment_type,

					'comment_parent' => (string) $comment->comment_parent,

					'comment_user_id' => (int) $comment->comment_user_id,

					'commentmeta' => $meta,

				);

			}

			$posts[] = $post;

		}

		return array(

			'authors' => $authors,

			'posts' => $posts,

			'categories' => $categories,

			'tags' => $tags,

			'terms' => $terms,

			'base_url' => $base_url,

			'version' => $wxr_version

		);

	}

}

/**

 * WXR Parser that makes use of the XML Parser PHP extension.

 */

class WXR_Parser_XML {

	var $wp_tags = array(

		'wp:post_id', 'wp:post_date', 'wp:post_date_gmt', 'wp:comment_status', 'wp:ping_status', 'wp:attachment_url',

		'wp:status', 'wp:post_name', 'wp:post_parent', 'wp:menu_order', 'wp:post_type', 'wp:post_password',

		'wp:is_sticky', 'wp:term_id', 'wp:category_nicename', 'wp:category_parent', 'wp:cat_name', 'wp:category_description',

		'wp:tag_slug', 'wp:tag_name', 'wp:tag_description', 'wp:term_taxonomy', 'wp:term_parent',

		'wp:term_name', 'wp:term_description', 'wp:author_id', 'wp:author_login', 'wp:author_email', 'wp:author_display_name',

		'wp:author_first_name', 'wp:author_last_name',

	);

	var $wp_sub_tags = array(

		'wp:comment_id', 'wp:comment_author', 'wp:comment_author_email', 'wp:comment_author_url',

		'wp:comment_author_IP',	'wp:comment_date', 'wp:comment_date_gmt', 'wp:comment_content',

		'wp:comment_approved', 'wp:comment_type', 'wp:comment_parent', 'wp:comment_user_id',

	);

	function parse( $file ) {

		$this->wxr_version = $this->in_post = $this->cdata = $this->data = $this->sub_data = $this->in_tag = $this->in_sub_tag = false;

		$this->authors = $this->posts = $this->term = $this->category = $this->tag = array();

		$xml = xml_parser_create( 'UTF-8' );

		xml_parser_set_option( $xml, XML_OPTION_SKIP_WHITE, 1 );

		xml_parser_set_option( $xml, XML_OPTION_CASE_FOLDING, 0 );

		xml_set_object( $xml, $this );

		xml_set_character_data_handler( $xml, 'cdata' );

		xml_set_element_handler( $xml, 'tag_open', 'tag_close' );

		if ( ! xml_parse( $xml, file_get_contents( $file ), true ) ) {

			$current_line = xml_get_current_line_number( $xml );

			$current_column = xml_get_current_column_number( $xml );

			$error_code = xml_get_error_code( $xml );

			$error_string = xml_error_string( $error_code );

			return new WP_Error( 'XML_parse_error', 'There was an error when reading this WXR file', array( $current_line, $current_column, $error_string ) );

		}

		xml_parser_free( $xml );

		if ( ! preg_match( '/^\d+\.\d+$/', $this->wxr_version ) )

			return new WP_Error( 'WXR_parse_error', __( 'This does not appear to be a WXR file, missing/invalid WXR version number', 'wordpress-importer' ) );

		return array(

			'authors' => $this->authors,

			'posts' => $this->posts,

			'categories' => $this->category,

			'tags' => $this->tag,

			'terms' => $this->term,

			'base_url' => $this->base_url,

			'version' => $this->wxr_version

		);

	}

	function tag_open( $parse, $tag, $attr ) {

		if ( in_array( $tag, $this->wp_tags ) ) {

			$this->in_tag = substr( $tag, 3 );

			return;

		}

		if ( in_array( $tag, $this->wp_sub_tags ) ) {

			$this->in_sub_tag = substr( $tag, 3 );

			return;

		}

		switch ( $tag ) {

			case 'category':

				if ( isset($attr['domain'], $attr['nicename']) ) {

					$this->sub_data['domain'] = $attr['domain'];

					$this->sub_data['slug'] = $attr['nicename'];

				}

				break;

			case 'item': $this->in_post = true;

			case 'title': if ( $this->in_post ) $this->in_tag = 'post_title'; break;

			case 'guid': $this->in_tag = 'guid'; break;

			case 'dc:creator': $this->in_tag = 'post_author'; break;

			case 'content:encoded': $this->in_tag = 'post_content'; break;

			case 'excerpt:encoded': $this->in_tag = 'post_excerpt'; break;

			case 'wp:term_slug': $this->in_tag = 'slug'; break;

			case 'wp:meta_key': $this->in_sub_tag = 'key'; break;

			case 'wp:meta_value': $this->in_sub_tag = 'value'; break;

		}

	}

	function cdata( $parser, $cdata ) {

		if ( ! trim( $cdata ) )

			return;

		$this->cdata .= trim( $cdata );

	}

	function tag_close( $parser, $tag ) {

		switch ( $tag ) {

			case 'wp:comment':

				unset( $this->sub_data['key'], $this->sub_data['value'] ); // remove meta sub_data

				if ( ! empty( $this->sub_data ) )

					$this->data['comments'][] = $this->sub_data;

				$this->sub_data = false;

				break;

			case 'wp:commentmeta':

				$this->sub_data['commentmeta'][] = array(

					'key' => $this->sub_data['key'],

					'value' => $this->sub_data['value']

				);

				break;

			case 'category':

				if ( ! empty( $this->sub_data ) ) {

					$this->sub_data['name'] = $this->cdata;

					$this->data['terms'][] = $this->sub_data;

				}

				$this->sub_data = false;

				break;

			case 'wp:postmeta':

				if ( ! empty( $this->sub_data ) )

					$this->data['postmeta'][] = $this->sub_data;

				$this->sub_data = false;

				break;

			case 'item':

				$this->posts[] = $this->data;

				$this->data = false;

				break;

			case 'wp:category':

			case 'wp:tag':

			case 'wp:term':

				$n = substr( $tag, 3 );

				array_push( $this->$n, $this->data );

				$this->data = false;

				break;

			case 'wp:author':

				if ( ! empty($this->data['author_login']) )

					$this->authors[$this->data['author_login']] = $this->data;

				$this->data = false;

				break;

			case 'wp:base_site_url':

				$this->base_url = $this->cdata;

				break;

			case 'wp:wxr_version':

				$this->wxr_version = $this->cdata;

				break;

			default:

				if ( $this->in_sub_tag ) {

					$this->sub_data[$this->in_sub_tag] = ! empty( $this->cdata ) ? $this->cdata : '';

					$this->in_sub_tag = false;

				} else if ( $this->in_tag ) {

					$this->data[$this->in_tag] = ! empty( $this->cdata ) ? $this->cdata : '';

					$this->in_tag = false;

				}

		}

		$this->cdata = false;

	}

}

/**

 * WXR Parser that uses regular expressions. Fallback for installs without an XML parser.