| 1 |
<?php |
| 2 |
|
| 3 |
namespace ABlocks\import\XmlParsers; |
| 4 |
|
| 5 |
use SimpleXMLElement; |
| 6 |
use WP_Error; |
| 7 |
|
| 8 |
if ( ! defined( 'ABSPATH' ) ) { |
| 9 |
exit; |
| 10 |
} |
| 11 |
|
| 12 |
/** |
| 13 |
* WXR Parser that makes use of the SimpleXML PHP extension. |
| 14 |
*/ |
| 15 |
class WXR_Parser_SimpleXML { |
| 16 |
public function parse( $file ) { |
| 17 |
$authors = array(); |
| 18 |
$posts = array(); |
| 19 |
$categories = array(); |
| 20 |
$tags = array(); |
| 21 |
$terms = array(); |
| 22 |
|
| 23 |
$internal_errors = libxml_use_internal_errors( true ); |
| 24 |
|
| 25 |
$dom = new \DOMDocument(); |
| 26 |
$old_value = null; |
| 27 |
if ( function_exists( 'libxml_disable_entity_loader' ) && PHP_VERSION_ID < 80000 ) { |
| 28 |
// phpcs:ignore Generic.PHP.DeprecatedFunctions.Deprecated |
| 29 |
$old_value = libxml_disable_entity_loader( true ); |
| 30 |
} |
| 31 |
// phpcs:ignore WordPress.WP.AlternativeFunctions.file_get_contents_file_get_contents |
| 32 |
$success = $dom->loadXML( file_get_contents( $file ) ); |
| 33 |
if ( ! is_null( $old_value ) ) { |
| 34 |
// phpcs:ignore Generic.PHP.DeprecatedFunctions.Deprecated |
| 35 |
libxml_disable_entity_loader( $old_value ); |
| 36 |
} |
| 37 |
|
| 38 |
if ( ! $success || isset( $dom->doctype ) ) { |
| 39 |
return new WP_Error( 'SimpleXML_parse_error', __( 'There was an error when reading this WXR file', 'ablocks' ), libxml_get_errors() ); |
| 40 |
} |
| 41 |
|
| 42 |
$xml = simplexml_import_dom( $dom ); |
| 43 |
unset( $dom ); |
| 44 |
|
| 45 |
// halt if loading produces an error |
| 46 |
if ( ! $xml ) { |
| 47 |
return new WP_Error( 'SimpleXML_parse_error', __( 'There was an error when reading this WXR file', 'ablocks' ), libxml_get_errors() ); |
| 48 |
} |
| 49 |
|
| 50 |
$wxr_version = $xml->xpath( '/rss/channel/wp:wxr_version' ); |
| 51 |
if ( ! $wxr_version ) { |
| 52 |
return new WP_Error( 'WXR_parse_error', __( 'This does not appear to be a WXR file, missing/invalid WXR version number', 'ablocks' ) ); |
| 53 |
} |
| 54 |
|
| 55 |
$wxr_version = (string) trim( $wxr_version[0] ); |
| 56 |
// confirm that we are dealing with the correct file format |
| 57 |
if ( ! preg_match( '/^\d+\.\d+$/', $wxr_version ) ) { |
| 58 |
return new WP_Error( 'WXR_parse_error', __( 'This does not appear to be a WXR file, missing/invalid WXR version number', 'ablocks' ) ); |
| 59 |
} |
| 60 |
|
| 61 |
$base_url = $xml->xpath( '/rss/channel/wp:base_site_url' ); |
| 62 |
$base_url = trim( isset( $base_url[0] ) ? $base_url[0] : '' ); |
| 63 |
|
| 64 |
$base_blog_url = $xml->xpath( '/rss/channel/wp:base_blog_url' ); |
| 65 |
if ( $base_blog_url ) { |
| 66 |
$base_blog_url = trim( $base_blog_url[0] ); |
| 67 |
} else { |
| 68 |
$base_blog_url = $base_url; |
| 69 |
} |
| 70 |
|
| 71 |
$show_on_front = $xml->xpath( '/rss/channel/ablocks_options/show_on_front' ); |
| 72 |
if ( $show_on_front ) { |
| 73 |
$show_on_front = trim( $show_on_front[0] ); |
| 74 |
} |
| 75 |
|
| 76 |
$page_on_front = $xml->xpath( '/rss/channel/ablocks_options/page_on_front' ); |
| 77 |
if ( $page_on_front ) { |
| 78 |
$page_on_front = (int) $page_on_front[0]; |
| 79 |
} |
| 80 |
|
| 81 |
$page_for_posts = $xml->xpath( '/rss/channel/ablocks_options/page_for_posts' ); |
| 82 |
if ( $page_for_posts ) { |
| 83 |
$page_for_posts = (int) $page_for_posts[0]; |
| 84 |
} |
| 85 |
|
| 86 |
$namespaces = $xml->getDocNamespaces(); |
| 87 |
if ( ! isset( $namespaces['wp'] ) ) { |
| 88 |
$namespaces['wp'] = 'http://wordpress.org/export/1.1/'; |
| 89 |
} |
| 90 |
if ( ! isset( $namespaces['excerpt'] ) ) { |
| 91 |
$namespaces['excerpt'] = 'http://wordpress.org/export/1.1/excerpt/'; |
| 92 |
} |
| 93 |
|
| 94 |
// grab authors |
| 95 |
foreach ( $xml->xpath( '/rss/channel/wp:author' ) as $author_arr ) { |
| 96 |
$a = $author_arr->children( $namespaces['wp'] ); |
| 97 |
$login = (string) $a->author_login; |
| 98 |
$authors[ $login ] = array( |
| 99 |
'author_id' => (int) $a->author_id, |
| 100 |
'author_login' => $login, |
| 101 |
'author_email' => (string) $a->author_email, |
| 102 |
'author_display_name' => (string) $a->author_display_name, |
| 103 |
'author_first_name' => (string) $a->author_first_name, |
| 104 |
'author_last_name' => (string) $a->author_last_name, |
| 105 |
); |
| 106 |
} |
| 107 |
|
| 108 |
// grab cats, tags and terms |
| 109 |
foreach ( $xml->xpath( '/rss/channel/wp:category' ) as $term_arr ) { |
| 110 |
$t = $term_arr->children( $namespaces['wp'] ); |
| 111 |
$category = array( |
| 112 |
'term_id' => (int) $t->term_id, |
| 113 |
'category_nicename' => (string) $t->category_nicename, |
| 114 |
'category_parent' => (string) $t->category_parent, |
| 115 |
'cat_name' => (string) $t->cat_name, |
| 116 |
'category_description' => (string) $t->category_description, |
| 117 |
); |
| 118 |
|
| 119 |
foreach ( $t->termmeta as $meta ) { |
| 120 |
$category['termmeta'][] = array( |
| 121 |
'key' => (string) $meta->meta_key, |
| 122 |
'value' => (string) $meta->meta_value, |
| 123 |
); |
| 124 |
} |
| 125 |
|
| 126 |
$categories[] = $category; |
| 127 |
} |
| 128 |
|
| 129 |
foreach ( $xml->xpath( '/rss/channel/wp:tag' ) as $term_arr ) { |
| 130 |
$t = $term_arr->children( $namespaces['wp'] ); |
| 131 |
$tag = array( |
| 132 |
'term_id' => (int) $t->term_id, |
| 133 |
'tag_slug' => (string) $t->tag_slug, |
| 134 |
'tag_name' => (string) $t->tag_name, |
| 135 |
'tag_description' => (string) $t->tag_description, |
| 136 |
); |
| 137 |
|
| 138 |
foreach ( $t->termmeta as $meta ) { |
| 139 |
$tag['termmeta'][] = array( |
| 140 |
'key' => (string) $meta->meta_key, |
| 141 |
'value' => (string) $meta->meta_value, |
| 142 |
); |
| 143 |
} |
| 144 |
|
| 145 |
$tags[] = $tag; |
| 146 |
} |
| 147 |
|
| 148 |
foreach ( $xml->xpath( '/rss/channel/wp:term' ) as $term_arr ) { |
| 149 |
$t = $term_arr->children( $namespaces['wp'] ); |
| 150 |
$term = array( |
| 151 |
'term_id' => (int) $t->term_id, |
| 152 |
'term_taxonomy' => (string) $t->term_taxonomy, |
| 153 |
'slug' => (string) $t->term_slug, |
| 154 |
'term_parent' => (string) $t->term_parent, |
| 155 |
'term_name' => (string) $t->term_name, |
| 156 |
'term_description' => (string) $t->term_description, |
| 157 |
); |
| 158 |
|
| 159 |
foreach ( $t->termmeta as $meta ) { |
| 160 |
$term['termmeta'][] = array( |
| 161 |
'key' => (string) $meta->meta_key, |
| 162 |
'value' => (string) $meta->meta_value, |
| 163 |
); |
| 164 |
} |
| 165 |
|
| 166 |
$terms[] = $term; |
| 167 |
} |
| 168 |
|
| 169 |
// grab posts |
| 170 |
foreach ( $xml->channel->item as $item ) { |
| 171 |
$post = array( |
| 172 |
'post_title' => (string) $item->title, |
| 173 |
'guid' => (string) $item->guid, |
| 174 |
); |
| 175 |
|
| 176 |
$dc = $item->children( 'http://purl.org/dc/elements/1.1/' ); |
| 177 |
$post['post_author'] = (string) $dc->creator; |
| 178 |
|
| 179 |
$wp = $item->children( $namespaces['wp'] ); |
| 180 |
$content = $item->children( 'http://purl.org/rss/1.0/modules/content/' ); |
| 181 |
$excerpt = $item->children( $namespaces['excerpt'] ); |
| 182 |
$content = (string) $content->encoded; |
| 183 |
if ( in_array( (string) $wp->post_type, array( 'wp_template', 'wp_template_part' ), true ) ) { |
| 184 |
$content = preg_replace( |
| 185 |
'/\s*,?\s*"theme"\s*:\s*"[^"]+"/', |
| 186 |
'', |
| 187 |
$content |
| 188 |
); |
| 189 |
} |
| 190 |
$post['post_content'] = $content; |
| 191 |
$post['post_excerpt'] = (string) $excerpt->encoded; |
| 192 |
|
| 193 |
$post['post_id'] = (int) $wp->post_id; |
| 194 |
$post['post_date'] = (string) $wp->post_date; |
| 195 |
$post['post_date_gmt'] = (string) $wp->post_date_gmt; |
| 196 |
$post['comment_status'] = (string) $wp->comment_status; |
| 197 |
$post['ping_status'] = (string) $wp->ping_status; |
| 198 |
$post['post_name'] = (string) $wp->post_name; |
| 199 |
$post['status'] = (string) $wp->status; |
| 200 |
$post['post_parent'] = (int) $wp->post_parent; |
| 201 |
$post['menu_order'] = (int) $wp->menu_order; |
| 202 |
$post['post_type'] = (string) $wp->post_type; |
| 203 |
$post['post_password'] = (string) $wp->post_password; |
| 204 |
$post['is_sticky'] = (int) $wp->is_sticky; |
| 205 |
|
| 206 |
if ( isset( $wp->attachment_url ) ) { |
| 207 |
$post['attachment_url'] = (string) $wp->attachment_url; |
| 208 |
} |
| 209 |
|
| 210 |
foreach ( $item->category as $c ) { |
| 211 |
$att = $c->attributes(); |
| 212 |
if ( isset( $att['nicename'] ) ) { |
| 213 |
$post['terms'][] = array( |
| 214 |
'name' => (string) $c, |
| 215 |
'slug' => (string) $att['nicename'], |
| 216 |
'domain' => (string) $att['domain'], |
| 217 |
); |
| 218 |
} |
| 219 |
} |
| 220 |
|
| 221 |
foreach ( $wp->postmeta as $meta ) { |
| 222 |
$post['postmeta'][] = array( |
| 223 |
'key' => (string) $meta->meta_key, |
| 224 |
'value' => (string) $meta->meta_value, |
| 225 |
); |
| 226 |
} |
| 227 |
|
| 228 |
foreach ( $wp->comment as $comment ) { |
| 229 |
$meta = array(); |
| 230 |
if ( isset( $comment->commentmeta ) ) { |
| 231 |
foreach ( $comment->commentmeta as $m ) { |
| 232 |
$meta[] = array( |
| 233 |
'key' => (string) $m->meta_key, |
| 234 |
'value' => (string) $m->meta_value, |
| 235 |
); |
| 236 |
} |
| 237 |
} |
| 238 |
|
| 239 |
$post['comments'][] = array( |
| 240 |
'comment_id' => (int) $comment->comment_id, |
| 241 |
'comment_author' => (string) $comment->comment_author, |
| 242 |
'comment_author_email' => (string) $comment->comment_author_email, |
| 243 |
'comment_author_IP' => (string) $comment->comment_author_IP, |
| 244 |
'comment_author_url' => (string) $comment->comment_author_url, |
| 245 |
'comment_date' => (string) $comment->comment_date, |
| 246 |
'comment_date_gmt' => (string) $comment->comment_date_gmt, |
| 247 |
'comment_content' => (string) $comment->comment_content, |
| 248 |
'comment_approved' => (string) $comment->comment_approved, |
| 249 |
'comment_type' => (string) $comment->comment_type, |
| 250 |
'comment_parent' => (string) $comment->comment_parent, |
| 251 |
'comment_user_id' => (int) $comment->comment_user_id, |
| 252 |
'commentmeta' => $meta, |
| 253 |
); |
| 254 |
}//end foreach |
| 255 |
|
| 256 |
$posts[] = $post; |
| 257 |
}//end foreach |
| 258 |
|
| 259 |
return array( |
| 260 |
'authors' => $authors, |
| 261 |
'posts' => $posts, |
| 262 |
'categories' => $categories, |
| 263 |
'tags' => $tags, |
| 264 |
'terms' => $terms, |
| 265 |
'base_url' => $base_url, |
| 266 |
'base_blog_url' => $base_blog_url, |
| 267 |
'version' => $wxr_version, |
| 268 |
'show_on_front' => $show_on_front, |
| 269 |
'page_on_front' => $page_on_front, |
| 270 |
'page_for_posts' => $page_for_posts, |
| 271 |
'storeengine_pages' => $this->get_storeengine_pages( $xml ), |
| 272 |
'academy_pages' => $this->get_academy_pages( $xml ), |
| 273 |
'ablocks_pages' => $this->get_ablocks_pages( $xml ), |
| 274 |
'ablocks_options' => $this->get_ablocks_options( $xml ), |
| 275 |
); |
| 276 |
} |
| 277 |
|
| 278 |
private function get_storeengine_pages( SimpleXMLElement $xml ): array { |
| 279 |
$storeengine_pages = [ |
| 280 |
'shop_page' => 0, |
| 281 |
'cart_page' => 0, |
| 282 |
'checkout_page' => 0, |
| 283 |
'thankyou_page' => 0, |
| 284 |
'dashboard_page' => 0, |
| 285 |
'membership_pricing_page' => 0, |
| 286 |
'affiliate_registration_page' => 0, |
| 287 |
]; |
| 288 |
|
| 289 |
foreach ( array_keys( $storeengine_pages ) as $page_key ) { |
| 290 |
$value = $xml->xpath( '/rss/channel/ablocks_options/storeengine_' . $page_key ); |
| 291 |
if ( $value ) { |
| 292 |
$storeengine_pages[ $page_key ] = (int) $value[0]; |
| 293 |
} |
| 294 |
} |
| 295 |
|
| 296 |
return $storeengine_pages; |
| 297 |
} |
| 298 |
|
| 299 |
private function get_academy_pages( SimpleXMLElement $xml ): array { |
| 300 |
$academy_pages = [ |
| 301 |
'frontend_dashboard_page' => 0, |
| 302 |
'frontend_student_reg_page' => 0, |
| 303 |
'password_reset_page' => 0, |
| 304 |
'lessons_page' => 0, |
| 305 |
'course_page' => 0, |
| 306 |
'frontend_instructor_reg_page' => 0, |
| 307 |
'tutor_booking_page' => 0, |
| 308 |
]; |
| 309 |
|
| 310 |
foreach ( array_keys( $academy_pages ) as $page_key ) { |
| 311 |
$value = $xml->xpath( '/rss/channel/ablocks_options/academy_' . $page_key ); |
| 312 |
if ( $value ) { |
| 313 |
$academy_pages[ $page_key ] = (int) $value[0]; |
| 314 |
} |
| 315 |
} |
| 316 |
|
| 317 |
return $academy_pages; |
| 318 |
} |
| 319 |
|
| 320 |
private function get_ablocks_pages( SimpleXMLElement $xml ): array { |
| 321 |
$ablocks_pages = [ |
| 322 |
'login_page' => 0, |
| 323 |
'registration_page' => 0, |
| 324 |
'forget_password_page' => 0, |
| 325 |
]; |
| 326 |
|
| 327 |
foreach ( array_keys( $ablocks_pages ) as $page_key ) { |
| 328 |
$value = $xml->xpath( '/rss/channel/ablocks_options/ablocks_' . $page_key ); |
| 329 |
if ( $value ) { |
| 330 |
$ablocks_pages[ $page_key ] = (int) $value[0]; |
| 331 |
} |
| 332 |
} |
| 333 |
|
| 334 |
return $ablocks_pages; |
| 335 |
} |
| 336 |
|
| 337 |
private function get_ablocks_options( SimpleXMLElement $xml ): array { |
| 338 |
$ablocks_options = [ |
| 339 |
'default_container_width' => null, |
| 340 |
'container_padding' => null, |
| 341 |
'container_element_gap' => null, |
| 342 |
'global_color' => 'json', |
| 343 |
'global_typography' => 'json', |
| 344 |
'global_font_family_fallback' => null, |
| 345 |
'global_body_text_color' => null, |
| 346 |
'global_body_typography' => 'json', |
| 347 |
'global_body_paragraph_space' => 'json', |
| 348 |
'global_link_color' => null, |
| 349 |
'global_link_hover_color' => null, |
| 350 |
'global_link_typography' => 'json', |
| 351 |
'global_link_hover_typography' => 'json', |
| 352 |
'global_h1_color' => null, |
| 353 |
'global_h1_typography' => 'json', |
| 354 |
'global_h2_color' => null, |
| 355 |
'global_h2_typography' => 'json', |
| 356 |
'global_h3_color' => null, |
| 357 |
'global_h3_typography' => 'json', |
| 358 |
'global_h4_color' => null, |
| 359 |
'global_h4_typography' => 'json', |
| 360 |
'global_h5_color' => null, |
| 361 |
'global_h5_typography' => 'json', |
| 362 |
'global_h6_color' => null, |
| 363 |
'global_h6_typography' => 'json', |
| 364 |
'frontend_dashboard_page' => null, |
| 365 |
'frontend_dashboard_sub_pages' => 'json', |
| 366 |
]; |
| 367 |
|
| 368 |
foreach ( array_keys( $ablocks_options ) as $page_key ) { |
| 369 |
$value = $xml->xpath( '/rss/channel/ablocks_options/ablocks_' . $page_key ); |
| 370 |
if ( $value ) { |
| 371 |
$ablocks_options[ $page_key ] = 'json' === $ablocks_options[ $page_key ] ? json_decode( |
| 372 |
base64_decode( (string) $value[0] ), true |
| 373 |
) : (string) $value[0]; |
| 374 |
} else { |
| 375 |
unset( $ablocks_options[ $page_key ] ); |
| 376 |
} |
| 377 |
} |
| 378 |
|
| 379 |
return $ablocks_options; |
| 380 |
} |
| 381 |
} |
| 382 |
|
