Atomic_Wind_Meta_Handler.php
3 months ago
Beaver_Data_Fix.php
5 years ago
Beaver_ParserXML.php
5 years ago
Elementor_Meta_Handler.php
3 months ago
WP_Import.php
3 months ago
WXR_Parser.php
5 years ago
WXR_Parser_SimpleXML.php
4 years ago
WXR_Parser_XML.php
5 years ago
WXR_Parser_SimpleXML.php
208 lines
| 1 | <?php |
| 2 | |
| 3 | namespace TIOB\Importers\WP; |
| 4 | use WP_Error; |
| 5 | |
| 6 | /** |
| 7 | * WXR Parser that makes use of the SimpleXML PHP extension. |
| 8 | */ |
| 9 | class WXR_Parser_SimpleXML { |
| 10 | /** |
| 11 | * @param string $file file path. |
| 12 | * |
| 13 | * @return array | WP_Error |
| 14 | */ |
| 15 | public function parse( $file ) { |
| 16 | global $wp_filesystem; |
| 17 | WP_Filesystem(); |
| 18 | $authors = $posts = $categories = $tags = $terms = array(); |
| 19 | $internal_errors = libxml_use_internal_errors( true ); |
| 20 | $dom = new \DOMDocument; |
| 21 | $old_value = null; |
| 22 | if ( \PHP_VERSION_ID < 80000 && function_exists( 'libxml_disable_entity_loader' ) ) { |
| 23 | $old_value = libxml_disable_entity_loader( true ); |
| 24 | } |
| 25 | $success = $dom->loadXML( $wp_filesystem->get_contents( $file ) ); |
| 26 | if ( ! is_null( $old_value ) ) { |
| 27 | libxml_disable_entity_loader( $old_value ); |
| 28 | } |
| 29 | if ( ! $success || isset( $dom->doctype ) ) { |
| 30 | return new WP_Error( 'SimpleXML_parse_error', 'There was an error when reading this WXR file', libxml_get_errors() ); |
| 31 | } |
| 32 | $xml = simplexml_import_dom( $dom ); |
| 33 | unset( $dom ); |
| 34 | // halt if loading produces an error |
| 35 | if ( ! $xml ) { |
| 36 | return new WP_Error( 'SimpleXML_parse_error', 'There was an error when reading this WXR file', libxml_get_errors() ); |
| 37 | } |
| 38 | $wxr_version = $xml->xpath( '/rss/channel/wp:wxr_version' ); |
| 39 | if ( ! $wxr_version ) { |
| 40 | return new WP_Error( 'WXR_parse_error', 'This does not appear to be a WXR file, missing/invalid WXR version number' ); |
| 41 | } |
| 42 | $wxr_version = (string) trim( $wxr_version[0] ); |
| 43 | // confirm that we are dealing with the correct file format |
| 44 | if ( ! preg_match( '/^\d+\.\d+$/', $wxr_version ) ) { |
| 45 | return new WP_Error( 'WXR_parse_error', 'This does not appear to be a WXR file, missing/invalid WXR version number' ); |
| 46 | } |
| 47 | $base_url = $xml->xpath( '/rss/channel/wp:base_site_url' ); |
| 48 | $base_url = (string) trim( $base_url[0] ); |
| 49 | $base_blog_url = $xml->xpath( '/rss/channel/wp:base_blog_url' ); |
| 50 | $base_blog_url = (string) trim( $base_blog_url[0] ); |
| 51 | $namespaces = $xml->getDocNamespaces(); |
| 52 | if ( ! isset( $namespaces['wp'] ) ) { |
| 53 | $namespaces['wp'] = 'http://wordpress.org/export/1.1/'; |
| 54 | } |
| 55 | if ( ! isset( $namespaces['excerpt'] ) ) { |
| 56 | $namespaces['excerpt'] = 'http://wordpress.org/export/1.1/excerpt/'; |
| 57 | } |
| 58 | // grab authors |
| 59 | foreach ( $xml->xpath( '/rss/channel/wp:author' ) as $author_arr ) { |
| 60 | $a = $author_arr->children( $namespaces['wp'] ); |
| 61 | $login = (string) $a->author_login; |
| 62 | $authors[ $login ] = array( |
| 63 | 'author_id' => (int) $a->author_id, |
| 64 | 'author_login' => $login, |
| 65 | 'author_email' => (string) $a->author_email, |
| 66 | 'author_display_name' => (string) $a->author_display_name, |
| 67 | 'author_first_name' => (string) $a->author_first_name, |
| 68 | 'author_last_name' => (string) $a->author_last_name, |
| 69 | ); |
| 70 | } |
| 71 | // grab cats, tags and terms |
| 72 | foreach ( $xml->xpath( '/rss/channel/wp:category' ) as $term_arr ) { |
| 73 | $t = $term_arr->children( $namespaces['wp'] ); |
| 74 | $category = array( |
| 75 | 'term_id' => (int) $t->term_id, |
| 76 | 'category_nicename' => (string) $t->category_nicename, |
| 77 | 'category_parent' => (string) $t->category_parent, |
| 78 | 'cat_name' => (string) $t->cat_name, |
| 79 | 'category_description' => (string) $t->category_description, |
| 80 | ); |
| 81 | foreach ( $t->termmeta as $meta ) { |
| 82 | $category['termmeta'][] = array( |
| 83 | 'key' => (string) $meta->meta_key, |
| 84 | 'value' => (string) $meta->meta_value, |
| 85 | ); |
| 86 | } |
| 87 | $categories[] = $category; |
| 88 | } |
| 89 | foreach ( $xml->xpath( '/rss/channel/wp:tag' ) as $term_arr ) { |
| 90 | $t = $term_arr->children( $namespaces['wp'] ); |
| 91 | $tag = array( |
| 92 | 'term_id' => (int) $t->term_id, |
| 93 | 'tag_slug' => (string) $t->tag_slug, |
| 94 | 'tag_name' => (string) $t->tag_name, |
| 95 | 'tag_description' => (string) $t->tag_description, |
| 96 | ); |
| 97 | foreach ( $t->termmeta as $meta ) { |
| 98 | $tag['termmeta'][] = array( |
| 99 | 'key' => (string) $meta->meta_key, |
| 100 | 'value' => (string) $meta->meta_value, |
| 101 | ); |
| 102 | } |
| 103 | $tags[] = $tag; |
| 104 | } |
| 105 | foreach ( $xml->xpath( '/rss/channel/wp:term' ) as $term_arr ) { |
| 106 | $t = $term_arr->children( $namespaces['wp'] ); |
| 107 | $term = array( |
| 108 | 'term_id' => (int) $t->term_id, |
| 109 | 'term_taxonomy' => (string) $t->term_taxonomy, |
| 110 | 'slug' => (string) $t->term_slug, |
| 111 | 'term_parent' => (string) $t->term_parent, |
| 112 | 'term_name' => (string) $t->term_name, |
| 113 | 'term_description' => (string) $t->term_description, |
| 114 | ); |
| 115 | foreach ( $t->termmeta as $meta ) { |
| 116 | $term['termmeta'][] = array( |
| 117 | 'key' => (string) $meta->meta_key, |
| 118 | 'value' => (string) $meta->meta_value, |
| 119 | ); |
| 120 | } |
| 121 | $terms[] = $term; |
| 122 | } |
| 123 | // grab posts |
| 124 | foreach ( $xml->channel->item as $item ) { |
| 125 | $post = array( |
| 126 | 'post_title' => (string) $item->title, |
| 127 | 'guid' => (string) $item->guid, |
| 128 | ); |
| 129 | $dc = $item->children( 'http://purl.org/dc/elements/1.1/' ); |
| 130 | $post['post_author'] = (string) $dc->creator; |
| 131 | $content = $item->children( 'http://purl.org/rss/1.0/modules/content/' ); |
| 132 | $excerpt = $item->children( $namespaces['excerpt'] ); |
| 133 | $post['post_content'] = (string) $content->encoded; |
| 134 | $post['post_excerpt'] = (string) $excerpt->encoded; |
| 135 | $wp = $item->children( $namespaces['wp'] ); |
| 136 | $post['post_id'] = (int) $wp->post_id; |
| 137 | $post['post_date'] = (string) $wp->post_date; |
| 138 | $post['post_date_gmt'] = (string) $wp->post_date_gmt; |
| 139 | $post['comment_status'] = (string) $wp->comment_status; |
| 140 | $post['ping_status'] = (string) $wp->ping_status; |
| 141 | $post['post_name'] = (string) $wp->post_name; |
| 142 | $post['status'] = (string) $wp->status; |
| 143 | $post['post_parent'] = (int) $wp->post_parent; |
| 144 | $post['menu_order'] = (int) $wp->menu_order; |
| 145 | $post['post_type'] = (string) $wp->post_type; |
| 146 | $post['post_password'] = (string) $wp->post_password; |
| 147 | $post['is_sticky'] = (int) $wp->is_sticky; |
| 148 | if ( isset( $wp->attachment_url ) ) { |
| 149 | $post['attachment_url'] = (string) $wp->attachment_url; |
| 150 | } |
| 151 | foreach ( $item->category as $c ) { |
| 152 | $att = $c->attributes(); |
| 153 | if ( isset( $att['nicename'] ) ) { |
| 154 | $post['terms'][] = array( |
| 155 | 'name' => (string) $c, |
| 156 | 'slug' => (string) $att['nicename'], |
| 157 | 'domain' => (string) $att['domain'], |
| 158 | ); |
| 159 | } |
| 160 | } |
| 161 | foreach ( $wp->postmeta as $meta ) { |
| 162 | $post['postmeta'][] = array( |
| 163 | 'key' => (string) $meta->meta_key, |
| 164 | 'value' => (string) $meta->meta_value, |
| 165 | ); |
| 166 | } |
| 167 | foreach ( $wp->comment as $comment ) { |
| 168 | $meta = array(); |
| 169 | if ( isset( $comment->commentmeta ) ) { |
| 170 | foreach ( $comment->commentmeta as $m ) { |
| 171 | $meta[] = array( |
| 172 | 'key' => (string) $m->meta_key, |
| 173 | 'value' => (string) $m->meta_value, |
| 174 | ); |
| 175 | } |
| 176 | } |
| 177 | $post['comments'][] = array( |
| 178 | 'comment_id' => (int) $comment->comment_id, |
| 179 | 'comment_author' => (string) $comment->comment_author, |
| 180 | 'comment_author_email' => (string) $comment->comment_author_email, |
| 181 | 'comment_author_IP' => (string) $comment->comment_author_IP, |
| 182 | 'comment_author_url' => (string) $comment->comment_author_url, |
| 183 | 'comment_date' => (string) $comment->comment_date, |
| 184 | 'comment_date_gmt' => (string) $comment->comment_date_gmt, |
| 185 | 'comment_content' => (string) $comment->comment_content, |
| 186 | 'comment_approved' => (string) $comment->comment_approved, |
| 187 | 'comment_type' => (string) $comment->comment_type, |
| 188 | 'comment_parent' => (string) $comment->comment_parent, |
| 189 | 'comment_user_id' => (int) $comment->comment_user_id, |
| 190 | 'commentmeta' => $meta, |
| 191 | ); |
| 192 | } |
| 193 | $posts[] = $post; |
| 194 | } |
| 195 | |
| 196 | return array( |
| 197 | 'authors' => $authors, |
| 198 | 'posts' => $posts, |
| 199 | 'categories' => $categories, |
| 200 | 'tags' => $tags, |
| 201 | 'terms' => $terms, |
| 202 | 'base_url' => $base_url, |
| 203 | 'base_blog_url' => $base_blog_url, |
| 204 | 'version' => $wxr_version, |
| 205 | ); |
| 206 | } |
| 207 | } |
| 208 |