/**
* @param string|SimpleXMLElement $file
* @param $num_posts split posts into segments
* @return array|WP_Error
*/
public function parse($file, $num_posts = 0, $page = 0)
{
$authors = $posts = $categories = $tags = $terms = $options = array();
//$skins_instances = array();
//parse paths
$paths = array();
if (is_string($file)) {
if (!$this->parser->data('import_path')) {
$paths['import_path'] = rtrim(HW_URL::get_path_url($file, true), '\\/') . '/';
}
if (!$this->parser->data('import_dir')) {
$paths['import_dir'] = dirname($file);
}
$this->parser->update_variables($paths);
}
$xml = self::read_simplexml_object($file);
$namespaces = $xml->namespaces;
$xml = $xml->xml;
$wxr_version = $xml->xpath('posts/wp:wxr_version');
#/rss/posts/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('posts/wp:base_site_url');
#/rss/posts/wp:base_site_url
$base_url = isset($base_url[0]) ? (string) trim($base_url[0]) : '';
//$x=simplexml_load_string($file->saveXML(), 'SimpleXMLElement');
//$p=$x->posts->item[0];__print((string)$p->children('http://wordpress.org/export/1.2/')->post_type);
//store namespace
$this->namespaces = $namespaces;
//gather skins instances
$this->gather_skins_data($xml, $namespaces);
//grab widgets data
$widgets = $this->grab_widgets($xml, $namespaces);
$sidebars = $this->grab_sidebars($xml, $namespaces);
// grab authors
if ($xml->xpath('posts/wp:author')) {
foreach ($xml->xpath('posts/wp:author') as $author_arr) {
#/rss/posts/wp:author
$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) $a->author_last_name);
}
}
// grab cats, tags and terms
if ($xml->xpath('posts/wp:category')) {
foreach ($xml->xpath('posts/wp:category') as $term_arr) {
#/rss/posts/wp:category
$atts = $term_arr->attributes();
$t = $term_arr->children($namespaces['wp']);
if (isset($atts['_id'])) {
$key = (string) $atts['_id'];
} else {
$key = (string) $t->category_nicename;
}
$args = array('_id' => $key, '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' => $this->parser->pre_shortcode_tags((string) $t->category_description));
if (isset($t->category_parent) && count($t->category_parent[0]->xpath('hw:params'))) {
$parent = $this->recursive_option_data($t->category_parent[0]->xpath('hw:params'));
//purpose of import result by hoangweb
/*if(!empty($parent->attributes['export']) && is_array($parent->option[0])) {
$export_path = (string) $parent->attributes['export'];
$export_path = hwArray::drwOfPath(explode('/', $export_path));
eval('$parent = $parent->option[0]'.$export_path.';');
}
elseif(!empty($parent->option[0])) {
$parent = $parent->option[0];
}*/
if (isset($parent->option[0])) {
$args['category_parent'] = $parent->option[0];
}
}
$categories[] = $args;
}
}
if ($xml->xpath('posts/wp:tag')) {
foreach ($xml->xpath('posts/wp:tag') as $term_arr) {
#/rss/posts/wp:tag
$atts = $term_arr->attributes();
$t = $term_arr->children($namespaces['wp']);
if (isset($atts['_id'])) {
$key = (string) $atts['_id'];
} else {
$key = (string) $t->tag_slug;
}
$tags[] = array('_id' => $key, 'term_id' => (int) $t->term_id, 'tag_slug' => (string) $t->tag_slug, 'tag_name' => (string) $t->tag_name, 'tag_description' => $this->parser->pre_shortcode_tags((string) $t->tag_description));
}
}
if ($xml->xpath('posts/wp:term')) {
foreach ($xml->xpath('posts/wp:term') as $term_arr) {
#/rss/posts/wp:term
$atts = $term_arr->attributes();
$t = $term_arr->children($namespaces['wp']);
if (isset($atts['_id'])) {
$key = (string) $atts['_id'];
} else {
$key = (string) $t->term_slug;
}
$args = array('_id' => $key, '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' => $this->parser->pre_shortcode_tags((string) $t->term_description));
if (isset($t->term_parent) && count($t->term_parent[0]->xpath('hw:params'))) {
$parent = $this->recursive_option_data($t->term_parent[0]->xpath('hw:params'));
//purpose of import result by hoangweb
if (!empty($parent->option[0])) {
$args['term_parent'] = $parent->option[0];
}
}
if ($args['term_taxonomy'] == 'nav_menu') {
$args['menu_location'] = (string) $t->menu_location;
}
//for nav menu
$terms[] = $args;
}
}
$start = $end = 0;
if ($num_posts != 0 && isset($xml->posts->item)) {
$segments = hwArray::split_loop_segments($num_posts, $xml->posts->item->count());
if (is_numeric($page) && isset($segments[$page])) {
$range = explode('-', $segments[$page]);
$start = $range[0];
$end = $range[1];
}
} elseif (isset($xml->posts->item)) {
$start = 0;
$end = $xml->posts->item->count();
}
$count = 0;
//$end = isset($xml->posts->item)? $xml->posts->item->count() : 0;
// grab posts
if (isset($xml->posts->item)) {
foreach ($xml->posts->item as $item) {
if ($start > ++$count) {
continue;
}
if ($count && $end + 1 < $count) {
break;
}
//by hoang
$atts = (array) $item->attributes();
$atts = isset($atts['@attributes']) ? $atts['@attributes'] : array();
//cast to array
$post = array('post_title' => (string) $item->title, 'guid' => (string) $item->guid, 'hw_attributes' => !empty($atts) ? $atts : array());
$dc = $item->children($namespaces['dc']);
$post['post_author'] = (string) $dc->creator;
$content = $item->children($namespaces['content']);
$excerpt = $item->children($namespaces['excerpt']);
if (isset($content->encoded) && count($content->encoded[0]->xpath('hw:params'))) {
$content_encoded = $this->recursive_option_data($content->encoded[0]->xpath('hw:params'));
//purpose of import result by hoangweb
if (!empty($content_encoded->option[0])) {
$content_encoded = $content_encoded->option[0];
}
} else {
$content_encoded = (string) $content->encoded;
}
//i decided to use import result in content encoded for post
if (!is_object($content_encoded)) {
$post['post_content'] = $this->parser->pre_shortcode_tags($content_encoded);
} else {
$post['post_content'] = $content_encoded;
}
$post['post_excerpt'] = (string) $excerpt->encoded;
$wp = $item->children($namespaces['wp']);
$hw = $item->children($namespaces['hw']);
//import id or save post id
$post['post_id'] = (int) $wp->post_id;
if (!empty($atts['_id'])) {
$post['_id'] = $atts['_id'];
} elseif (!empty($hw->_id)) {
$post['_id'] = (string) $hw->_id;
} elseif (isset($post['title'])) {
$post['_id'] = sanitize_title($post['title']);
}
$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;
$post['status'] = !empty($wp->status) ? (string) $wp->status : 'publish';
$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)) {
//for attachment of post type
$post['attachment_url'] = (string) $wp->attachment_url;
}
foreach ($hw->attachment as $a) {
//fetch attachments in other post type
//$_post = hwArray::cloneArray($post);
$url = (string) $a->url;
if ($this->parser->data('demo')) {
$url = $this->parser->data('demo')->get_file_url($url);
} else {
if (!HW_URL::valid_url($url)) {
$url = $this->parser->data('import_path') . '/' . $url;
}
}
//$post['_id'] = (string) $a->_id;
if ($post['post_type'] !== 'attachment') {
$_post = hwArray::cloneArray($post);
$_post['post_content'] = HW_String::limit($_post['post_content'], 50);
$_post['post_excerpt'] = HW_String::limit($_post['post_excerpt'], 50);
$_post['attachment_url'] = $url;
$_post['post_type'] = 'attachment';
$_post['_id'] = (string) $a->_id;
$posts[] = $_post;
if ((string) $a->thumbnail) {
$post['hw_thumbnail_id'] = (string) $a->_id;
}
}
//continue;
}
foreach ($item->category as $c) {
$att = $c->attributes();
if (isset($att['nicename'])) {
$post['terms'][] = array('name' => (string) $c, 'slug' => (string) $att['nicename'], 'domain' => (string) $att['domain']);
}
}
//post meta
foreach ($wp->postmeta as $meta) {
//by hoangweb
$hw_params = $meta->meta_value->xpath('hw:params');
if (!empty($hw_params)) {
//$meta_value = $this->grab_options($meta->meta_value->xpath('hw:params'), $namespaces);
$meta_value = $this->recursive_option_data($meta->meta_value->xpath('hw:params'));
if (!empty($meta_value->attributes['export']) && is_array($meta_value->option[0])) {
$export_path = (string) $meta_value->attributes['export'];
$export_path = hwArray::drwOfPath(explode('/', $export_path));
eval('$meta_value = $meta_value->option[0]' . $export_path . ';');
} elseif (!empty($meta_value->option[0])) {
$meta_value = $meta_value->option[0];
} else {
$meta_value = array();
}
//HW_Logger::log_file($meta_value);
if (!is_string($meta_value) && !is_object($meta_value)) {
$meta_value = serialize($meta_value);
}
} else {
$meta_value = (string) $meta->meta_value;
}
$post['postmeta'][] = array('key' => (string) $meta->meta_key, 'value' => $meta_value);
}
//for comment
foreach ($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[] = $post;
}
}
//grab options
if (isset($xml->options->option)) {
$options = $this->grab_options($xml->options->option, $namespaces);
}
return array('authors' => $authors, 'posts' => $posts, 'categories' => $categories, 'tags' => $tags, 'terms' => $terms, 'options' => $options, 'widgets' => $widgets, 'sidebars' => $sidebars, 'skins_data' => $this->skins_data, 'base_url' => $base_url, 'version' => $wxr_version);
}
